An evolutionarily conserved mode of modulation of Shaw-like K⁺ channels

Abstract

Voltage-gated K⁺ channels of the Shaw family (also known as the KCNC or Kv3 family) play pivotal roles in mammalian brains, and genetic or pharmacological disruption of their activities in mice results in a spectrum of behavioral defects. We have used the model system of Caenorhabditis elegans to elucidate conserved molecular mechanisms that regulate these channels. We have now found that the C. elegans Shaw channel KHT-1, and its mammalian homologue, murine Kv3.1b, are both modulated by acid phosphatases. Thus, the C. elegans phosphatase ACP-2 is stably associated with KHT-1, while its mammalian homolog, prostatic acid phosphatase (PAP; also known as ACPP-201) stably associates with murine Kv3.1b K⁺ channels in vitro and in vivo. In biochemical experiments both phosphatases were able to reverse phosphorylation of their associated channel. The effect of phosphorylation on both channels is to produce a decrease in current amplitude and electrophysiological analyses demonstrated that dephosphorylation reversed the effects of phosphorylation on the magnitude of the macroscopic currents. ACP-2 and KHT-1 were colocalized in the nervous system of C. elegans and, in the mouse nervous system, PAP and Kv3.1b were colocalized in subsets of neurons, including in the brain stem and the ventricular zone. Taken together, this body of evidence suggests that acid phosphatases are general regulatory partners of Shaw-like K⁺ channels.—Cotella, D., Hernandez-Enriquez, B., Duan, Z., Wu, X., Gazula, V.-R., Brown, M. R., Kaczmarek, L. K., and Sesti, F. An evolutionarily conserved mode of modulation of Shaw-like K⁺ channels.

In response to constant sensory inputs, neurons can decrease their responsiveness through a process known as neuronal adaptation. This type of neuronal plasticity underlies several behaviors, including habituation, a common form of nonassociative learning, through which living organisms are able to distinguish meaningful information—and thus potential threats—from the background. Neuronal excitability is mediated by ion channels, integral membrane proteins that generate electrical signals by allowing passive diffusion of ions across the plasma membrane. Potassium (K⁺) channels facilitate neuronal repolarization, thereby shaping the duration and magnitude of the cell's electrical responses. K⁺ channels play a crucial role in the molecular mechanisms underlying neuronal adaptation. For example, members of four different K⁺ channel subfamilies have been involved in the signaling pathways causing habituation in Drosophila (1). Large-conductance calcium-activated K⁺ (BK) channels mediate neuronal adaptation to ethanol through mechanisms that are conserved to a significant extent from C. elegans to rats (2–5). In rodents, small-conductance calcium-activated K⁺ (SK) channels have been related to both passive avoidance and habituation of exploratory activity (6). Moreover, mice lacking the inwardly rectifying K⁺ channel (GIRK/Kir3) exhibit delayed habituation in the open-field test (7).

The worm Caenorhabditis elegans responds to a tap on the Petri dish by moving backward (or rarely, forward; ref. 8). With repeated taps, the animal habituates. Autophosphorylation of a voltage-gated K⁺-channel complex, containing a pore-forming subunit, K⁺ channel for habituation to tap isoform 1 (KHT-1), and an accessory subunit protein kinase, MiRP potassium channel accessory subunit 1 (MPS-1), represents a key step in this form of habituation (9, 10). In particular, evidence suggests that the decrease in current that follows phosphorylation of KHT-1 by MPS-1 alters the excitability of touch neurons to produce the temporary desensitization to tap. KHT-1 belongs to the Shaw-related family of K⁺ channels, and its mammalian homologue is neuronal Kv3.1b (KCNC1b; ref. 10). Shaw-like K⁺ channels shape the action potentials of fast-spiking neurons and interneurons and are expressed in many other neurons within the mammalian brain (reviewed in ref 11). Accordingly, genetic disruption of Kv3 genes is associated with severe phenotypes in mice (12–18). Just as KHT-1 forms a complex with MPS-1, so the mammalian Kv3.1b channel assembles a homologue of MPS-1, the mammalian accessory subunit KCNE2 in vitro (although KCNE2 does not possess kinase activity; 19–21). The conservation between the C. elegans and the mammalian channels goes beyond their structure and stoichiometry. Thus, in rat brain, Kv3.1b plays a key role in determining adaptation to auditory stimulation. Specifically, in the auditory brain stem of rats, regulation of Kv3.1b by protein kinase C (PKC) enables the neurons to adapt to changes in the ambient acoustic environment (15). In rat brain, Kv3.1b is phosphorylated under basal conditions and becomes rapidly dephosphorylated in response to physiological increases in auditory stimulation. Thus, in both C. elegans and mammalian neurons, enzymatic modulation of a conserved K⁺ channel induces adaptation. Nevertheless, while the protein kinases that modulate these channels are known (MPS-1 and PKC), the molecular identities of the phosphatases, which presumably act to reverse the effects of the kinases, are as yet unknown.

The remarkable level of similarity between KHT-1 and Kv3.1b argues that C. elegans can provide an attractive system for unveiling and later investigating conserved regulatory mechanisms in Shaw-like K⁺ channels. As a first step toward this effort, here we report on the identification and characterization of acid phosphatase isoform 2 (ACP-2), which modulates KHT-1-MPS-1. Acid phosphatases constitute a superfamily of enzymes present in both prokaryote and eukaryote genomes. Accordingly, we further show that the mammalian homologue of ACP-2, prostatic acid phosphatase (PAP), stably interacts with Kv3.1b in vitro and in mouse brain. PAP reverses Kv3.1b phosphorylation by PKC, and the two proteins colocalize in regions of the mouse nervous system that include the brain stem and the ventricular zone.

MATERIALS AND METHODS

Molecular biology

The C. elegans acp-2 gene (wild-type-acp-2), including a 1074-bp promoter region (P_acp-2), was amplified from genomic DNA, and the resulting PCR product (P_acp-2::wild-type-acp-2, 2937 bp) was inserted in the Fire vector pPD95.75. A construct bearing a t to g mutation in the acp-2 gene leading to a His to Gln replacement at position 34 in the catalytic site of ACP-2 (P_acp-2::h34q–acp-2) was constructed by site-directed mutagenesis using the wild-type P_acp-2::wild-type-acp-2 construct as template. A transcriptional reporter P_acp-2::gfp was constructed by inverse PCR using the wild-type construct as a template.

Cloning of ACP-2 was performed with a Smart RACE kit (Clontech, Palo, Alto, CA, USA) using poly(A)⁺ mRNA extracted from total C. elegans RNA with the Oligotex kit (Qiagen, Valencia, CA, USA). A canonical Kozak sequence (ccacc) was added before the start codon of ACP-2. This construct was inserted into a modified pCI-neo vector (Promega, Madison, WI, USA) containing a V5 tag sequence (GKPIPNPLLGLD) for C-terminal epitope tagging (pV5-CIneo).

A Basic Local Alignment Search Tool (BLAST)-verified construct containing the longer isoform (isoform b) of the human prostatic acid phosphatase (hPAP) was purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) with the IMAGE clone identification no. 3951204 (GenBank no. BC007460). The cDNA of the hPAP shorter isoform (isoform a) was obtained from the cDNA of the longer isoform by inverse PCR. The coding sequence of hPAP was amplified by PCR from the library clone and inserted in the pV5-CIneo vector for functional expression in mammalian cells. Rat Kv3.1b cDNA was subcloned in a pCI-neo vector (Promega) modified by adding a hemagglutinin (HA) tag (YPYDVPDYA) for C-terminal epitope tagging (pHA-CIneo).

All sequences were confirmed by automated DNA sequencing.

Strains and transgenic animals

C. elegans strains

Background strains were Bristol (N2) and RB1704 [acp-2 knockout (KO); ok2129 allele]. We constructed FDX(ses40): ok2129(P_acp-2::gfp); FDX(ses41): ok2129(P_acp-2::wild-type-acp-2)(myo-2), and FDX(ses42): ok2129(P_acp-2::h34q-acp-2)(myo-2).

The constructs were injected into the syncytial gonads of adult hermaphrodites. Transformant lines were stabilized by a mutagenesis-induced integration into a chromosome by irradiating 40 animals with γ-rays at 4000 rad for 40 min. The progeny were checked for 100% transmission of the marker and also for the presence of the transgene by PCR amplification. Transgenic worms were outcrossed 4 times.

Mice

Adult C57BL6/J mice of either sex were used in this study. All experimental protocols involving animals were approved by the Yale University Animal Use and Care Committee.

Biochemistry

To reduce antibody background (ACP-2, PAP, and IgG bands all have molecular masses ranging between 50 and 60 kDa and, therefore, overlap on a Western blot), antibodies were covalently cross-linked to protein G-agarose beads (Roche, Mannheim, Germany) by exposure to glutaraldehyde (1% in PBS) for 15 min at room temperature on a rotary platform. An example of IgG depletion is shown in Fig. 7B.

Figure 7.

Kv3.1b and PAP coimmunoprecipitate in mouse brain. A) Kv3.1 and PAP immunoprecipitates and controls [nonimmune rabbit serum (NIRS)] were run on a 10% polyacrylamide gel and transferred onto nitrocellulose. The blotting membrane was cut in half, and the two pieces were stained with either anti-Kv3.1 (top panel) or anti-PAP (bottom panel). Arrows indicate Kv3.1 and PAP. Kv3.1 is detected in 2 forms running with apparent molecular masses of roughly 80 and 100 kDa. B) IgG depletion following cross-linking with the G protein was obtained by adding 1% glutaraldehyde for 15 min. Western blot visualization was done with mouse anti-IgG. Images were acquired with a Versadoc Imaging System (Bio-Rad) and analyzed with the Quantity One software (Bio-Rad).

In vitro coimmunoprecipitation and Western blot analysis

Chinese hamster ovary K1 (CHO-K1) cells were grown in HAM-F12 medium and transfected in 6-cm dishes with Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). For each transfection, 10 μl of transfection reagent and a total of 4 μg of plasmid DNA were used. At 24 h following transfection, cells were lysed with 1 ml of Nonidet P-40 buffer (50 mM Tris-Cl, pH 7.4; 150 mM NaCl; 1 mM EDTA; 0.1% IGEPAL CA-630; and protease inhibitors) and subjected to a brief sonication pulse. Cell lysates were centrifuged for 20 min at 4°C, precleared with protein G-agarose, and then immunoprecipitated with the cross-linked antibody/agarose complexes for 2 h at 4°C. Samples were extensively washed with Nonidet P-40 buffer, and bound proteins were recovered by incubating in SDS sample buffer at 95°C for 5 min. Proteins were resolved by SDS-PAGE and transferred onto a PVDF membrane. After blocking in a 5% solution of nonfat milk in 0.1% Tween 20-PBS (PBST), membranes were probed for 2 h with a rat monoclonal anti-HA antibody conjugated to HRP (1:1000; Roche), washed with PBST for 20 min, and incubated with POD chemiluminescence substrates (Roche) and exposed. Membranes were stripped with a mild stripping buffer (0.2 M glycine-Cl, pH 2.2; 0.1% SDS; 1% Tween-20; and 0.02% sodium azide) and reprobed with an anti-V5 antibody (1:2000).

In vivo coimmunoprecipitation

C57BL6/J mouse brains of either sex were homogenized in sucrose buffer (5 mM HEPES, pH 7.4; 320 mM sucrose; and protease inhibitors), and debris was removed by centrifugation (2000 rpm, 10 min). To stabilize protein complexes, protein samples were cross-linked with the thiol-cleavable dimethyl 3,3′-dithiopropionimidate (DTDP) added to 0.5 mg/ml, and incubated on a rotary platform for 1 h at 4°C. Cell membranes were collected by centrifugation (12,000 rpm, 30 min) and solubilized in Nonidet P-40 buffer. After preclearing with protein G-agarose, protein samples were immunoprecipitated with the cross-linked antibody/agarose complexes (3 μg of either rabbit anti-Kv3.1 polyclonal antibody (cat. AB5188; Millipore, Bedford, MA, USA), mouse anti-PAP polyclonal antibody (H00000055-B01P; Abnova, Taipei, Taiwan) or nonimmune rabbit IgGs) for 2 h at 4°C. Following extensive washes, bound proteins were resuspended in Laemmli buffer and analyzed by Western blot analysis as follows. Samples were run on a 10% SDS-PAGE, and then proteins were transferred onto nitrocellulose membranes, blocked with 5% milk-PBST, and incubated overnight with either the anti-Kv3.1 or anti-PAP antibodies (diluted 1:500). After incubation with the proper HRP-conjugated secondary antibody, membranes were developed by chemiluminescence, and images were acquired with a Versadoc Imaging System (Bio-Rad, Hercules, CA, USA) and analyzed with the Quantity One software (Bio-Rad).

Metabolic labeling of phosphorylated proteins

CHO-K1 cells were transfected in 6-cm dishes, as described above. At 6 h following incubation with the DNA/Lipofectamine complex, culture medium was removed and replaced with phosphate-free Dulbecco's modified Eagle medium (DMEM; 11971-025; Life Technologies, Carlsbad, CA, USA) containing 0.5 mCi/ml of [³²P]orthophosphate (Perkin Elmer, Wellesley, MA, USA), and cells were incubated for an additional 16 h. To induce phosphorylation of Kv3.1b, cells were stimulated with 100 nM phorbol 12-myristate 13-acetate (PMA) for 15 min before harvesting (phosphorylation of KHT-1 was obtained cotransfecting the accessory subunit MPS-1). Cells were washed with ice-cold TBS and lysed in 1 ml RIPA buffer (50 mM Tris, pH 7.4; 150 mM NaCl; 1 mM EDTA; 1% IGEPAL CA-630; 0.5% deoxycholate; 0.1% SDS; and protease inhibitors) for 30 min at 4°C. Cell lysates were centrifuged for 20 min at 4°C, and the supernatant was mixed with HA-conjugated beads (Roche) and rocked at 4°C for 3 h. Beads were extensively washed with RIPA buffer, and bound proteins were recovered by incubating in SDS sample buffer at 95°C for 5 min.

Samples were subjected to SDS-PAGE on an 8% gel. Gels were fixed with 10% acetic acid and 50% methanol for 1 h, washed, and dried; bands were visualized by autoradiography.

Immunocytochemistry

The detailed immunohistochemistry procedure was previously described (22). Briefly, 5 adult C57BL6/J mice were anesthetized with sodium pentobarbital and perfused through the left ventricle with PBS solutions. Brains were removed, postfixed in 4% paraformaldehyde overnight at 4°C, and sliced sagittally on a vibratome (Leica VT1000S; Leica Microsystems, Wetzlar, Germany) at 20 μm. After postfixing and permeabilized preparation, free-floating slice sections were processed for double labeling with chicken anti-PAP (20 μg/ml, 1:500; Neuromics; Edina, MN, USA) and mouse anti-Kv3.1b (5.10 μg/ml, 1:200; NeuroMab, Davis, CA, USA). After 48 h of incubation at 4°C in primary antibody, sections were washed 3 times for 10 min each. Fluorescent secondary antibodies were used to identify the double labeling: mouse Alexa Fluor 488 (Invitrogen) for Kv3.1b staining and chicken Cy 3 (Jackson ImmunoResearch, Bar Harbor, ME, USA) for PAP staining. In the negative control experiments, sections were processed through the same procedure without primary or secondary antibody. Sections were examined with a Zeiss laser scanning microscope (LSM 510 Meta; Carl Zeiss, Oberkochen, Germany) with Zeiss image acquisition software. Images were acquired in multitrack mode, because this function permits several tracks to be defined as one configuration for the scan procedure, and this does not allow fluorescence to bleed into the other channel. Images were acquired using Confocal–Apochromat ×40/1.2 water-correction or ×63/1.2 water-correction objective for brain slices, and optical thickness was constant for both tracks. Alexa Fluor 488 has excitation/emission of 496/519, and Cy3 has excitation/emission of 550/570. Double-immunofluorescence images were displayed as dual-color merged images. The Zeiss C-Apochromat objective was developed to match the requirements of diffraction limited optics and combines high numerical apertures and perfect chromatic correction. The C-Apochromat series water-immersion objectives are corrected for an extended range of wavelengths.

Electrophysiology

CHO cells were plated and transfected using the Qiagen SuperFect kit (which has a lower transfection efficiency than Lipofectamine but is more gentle to the cells) and used 24 to 36 h post-transfection. Data were recorded with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA) a PC (Dell) and Clampex software (Molecular Devices). Data were filtered at f_c = 1 kHz and sampled at 2.5 kHz. Bath solution contained (in mM): 4 KCl, 100 NaCl, 10 HEPES (pH 7.5 with NaOH), 1.8 CaCl₂, and 1.0 MgCl₂. Pipette solution contained (in mM): 100 KCl, 10 HEPES (pH 7.5 with KOH), 1.0 MgCl₂, 1.0 CaCl₂, and 10 EGTA (pH 7.5 with KOH). For recordings in cells preincubated with PMA, the standard pipette solution was replaced with one not containing the Ca²⁺ chelator EGTA and ATP: 140 mM KCl, 10 mM HEPES (pH 7.5 with KOH), 1.0 mM MgCl₂, and 3 mM ATP. Currents were elicited by voltage steps from −80 to +120 mV in 20-mV intervals. In some recordings, the voltage stimulation was repeated 3 to 5 times, and the currents were averaged. Offset potentials due to series resistances (≤5 mV) were not compensated for, when generating current-voltage relationships.

Behavioral assays

Age synchronization

Nematodes were grown in standard 10-cm nematode growth medium (NGM) plates with OP50 E. coli until a large population of gravid adults was reached (3–5 d). The animals were collected in 50-ml Falcon tubes (BD Biosciences, San Jose, CA, USA) washed in M9 buffer (22 mM KH₂PO₄, 22 mM NaH₂PO₄, 85 mM NaCl, and 1 mM MgSO₄) and treated with 10 vol of basic hypochlorite solution (0.25 M NaOH and 1% hypochlorite, freshly mixed). Worms were incubated at room temperature for 10 min; the eggs (and carcasses) were collected by centrifugation at 400 g for 5 min at 4°C, incubated overnight in M9 buffer, and seeded on standard NGM plates.

Behavioral tests were performed without knowledge of the worms' genotype.

Gentle body touch

Experiments were performed as described previously (23). Briefly, single worms were picked to individual plates and tested 6 times for response to light touch to the head and tail with an eyelash. Responses to head/tail touch were recorded as backward/forward movement. The overall response to touch of each group of worms was expressed as the average percentage of times the worms responded.

Habituation to tap

Testing was performed by recording individual worms under a stereomicroscope (Olympus SZX7; Olympus, Tokyo, Japan) equipped with a digital camera (Infinity 2; Lumenera Corp., Ottawa, ON, Canada) and dedicated software. Taps (impulse ∼5 mN·s) were delivered manually by raising the Petri dish and dropping it. The elevation of the dish was controlled by a glass rod, L-shaped, placed vertically at the base of the microscope. The habituation coefficient, H, was defined as the distance (backward or forward) traveled in response to tap and was normalized to 1 for 0.5 of the worm's length. Distances ≤0.25 of the animal's length scored 0. Animals can also pause in response to tap, and if pause lasted ≥1 s, H also scored 0.

To evaluate recovery from habituation, the worms were habituated by a train of 25 taps at 5-s intervals, and the last 3 responses (corresponding to taps 23–25) were averaged (steady-state response). Then, at 5, 7, 15, and 30 min after habituation training, 3 taps were delivered at 5-s intervals and averaged (recovered response). Recovery was expressed as the percentage increase of the recovered response vs. the steady-state response.

A standard experiment consisted of independent tests with groups of ≥10 animals, and then the groups were averaged.

Statistical analysis

Quantitative data are presented as means ± se. The level of significance was calculated using Student's t test for single comparisons and ANOVA for multiple comparisons. Statistical significance was assumed at the 95% confidence limit or greater (P<0.05).

RESULTS

Acid phosphatases are an evolutionarily conserved family

The C. elegans genome contains many phosphatase genes that could be potential interacting partners of the KHT-1 and MPS-1. We focused on the acp-2 gene, which putatively encodes an acid phosphatase, because this gene has been shown to be expressed in the touch neurons (24, 25). We confirmed acp-2 expression by gene reporter analysis (Fig. 1). Thus, a transcriptional reporter expressing GFP, driven by the acp-2 promoter, yielded expression in the ALM touch neurons (Fig. 1A, B. All cells were identified by eye) and the PLM touch neurons (Fig. 1A, C), the same cells where KHT-1 and MPS-1 form a functional complex. Expression was also observed in the spermatheca, intestine, and in several other neurons that we did not attempt to identify (Fig. 1A).

Figure 1.

acp-2 is expressed in touch neurons. A) Fluorescence microscopy and bright-field images taken from a transgenic worm expressing GFP under the acp-2 promoter (P_acp-2::gfp). P_acp-2 drives GFP signals in several neurons, intestine, and spermatheca (large fluorescent spots). B) Fluorescence microscopy and bright light images of the anterior part of the body. Arrow indicates an ALM neuron. C) Fluorescence microscopy and bright light images of the posterior part of the body. Inset: GFP fluorescence in the two PLM neurons. Worms were analyzed and photographed with an Olympus BX61 microscope equipped with a digital camera.

We confirmed the ACP-2 expression and sequence by analysis of RT-PCR products (RACE). The ACP-2 sequence turned out to be as predicted (by Wormbase, http://www.wormbase.org). Acid phosphatases constitute a large family of enzymes, found in prokaryote and eukaryote kingdoms, that hydrolyze phosphate esters. A phylogenetic tree of this family is shown in Fig. 2A. Sequence alignment of ACP-2 with its rat and human homologues (PAP isoform a) is illustrated in Fig. 2B. The homology between ACP-2 and hPAP is 61%, whereas the homology between hPAP and rPAP is 92%. Acid phosphatases exhibit a catalytic site centered around a conserved histidine in the N-terminal, which forms a phosphohistidine intermediate. A second histidine in the C terminus possibly acts as proton donor (ref. 26 and Fig. 2B, asterisks).

Figure 2.

Acid phosphatases are evolutionarily conserved. A) Phylogenetic tree of acid phosphatase genes present in the genome of Homo sapiens (cPAP), rats (rPAP), fish (Danio rerio lysosomal acid phosphatase precursor), frogs (Xenopus laevis ACPT acid phosphatase), C. elegans (ACP-2), bacteria (Escherichia coli pH 2.5 acid phosphatase APPA), yeast (Saccharomyces cerevisiae constitutive and repressible acid phosphatase, PHO3), and fungi (Aspergillus awamori phytase, phyA). Scale bar = 20% estimated substitutions. Diagram was calculated using Seaview software (41). B) Amino acid sequence alignments of C. elegans ACP-2, H. sapiens and Rattus norvegicus PAP.a isoform. Homology between ACP-2 and human PAP (hPAP) is 61%. Homology between hPAP and rat PAP (rPAP) is 92% (red, identical residues; green, strongly similar residues; blue, weakly similar residues). Asterisks indicate histidine acid phosphatase signatures. Alignment was calculated using ClustalW software (42).

ACP-2 forms a stable complex with KHT-1 and MPS-1

We carried out coimmunoprecipitation experiments to determine whether ACP-2 can associate with the KHT-1-MPS-1 complex. To this end, KHT-1 and MPS-1 tagged to the HA epitope tag in the C terminus and ACP-2 to the V5 tag in the C terminus (ACP-2-V5) were transiently expressed in CHO cells, immunoprecipitated with V5 antibodies (αV5), and subsequently immunoblotted with HA (to visualize KHT-1 and MPS-1) or V5 (to visualize ACP-2) antibodies. A representative Western blot of these experiments is shown in Fig. 3A. The first 4 lanes of the blot show antibody staining of total lysates. KHT-1 and MPS-1 were detected into a single, ∼60-kDa band and two separate bands, ∼31 and ∼35 kDa, respectively, as previously shown (the mature MPS-1 protein corresponds to the 35-kDa band; refs. 10, 27). ACP-2 yielded a single, ∼55-kDa band, consistent with its theoretical molecular mass (49 kDa). Acid phosphatases are generally glycosylated (28). Coimmunoprecipitations are shown in last 4 lanes in the blot in Fig. 3A. Thus, in cells coexpressing all three proteins, ACP-2 pulled down both KHT-1 and MPS-1. Further, ACP-2 singularly coimmunoprecipitated with either KHT-1 or MPS-1. Together, these data indicated that KHT-1, MPS-1, and ACP-2 form a stable, ternary complex in vitro. Neither KHT-1 nor MPS-1 seems to be essential for this association, because ACP-2 can individually interact with each.

Figure 3.

Stable association of acid phosphatases with Shaw-like channels. A) CHO cells were cotransfected with cDNAs encoding the indicated tagged subunits. On the first 4 lanes of the immunoblot, total lysates were stained with anti-HA or anti-V5 antibodies. KHT-1 runs in a single band with an apparent molecular mass of ∼60 kDa. MPS-1 yields 2 bands running with apparent masses of ∼31 and ∼35 kDa. ACP-2 runs with a single band with apparent molecular mass of ∼55 kDa. On the right, ACP-2 immunoprecipitates (αV5) were visualized with anti-HA antibodies, to detect KHT-1 and MPS-1, or anti-V5 to detect ACP-2. B) On the left part of the immunoblot, total lysates were stained with anti-HA to detect Kv3.1b channels. In total lysates, Kv3.1b exists in 2 bands running with apparent molecular masses of roughly 80 and 100 kDa. On the right, PAP immunoprecipitates (αV5) were visualized with anti-HA antibodies, to detect Kv3.1b or anti-V5 to detect PAP isoforms, which migrated with apparent molecular masses of ∼50 kDa.

PAP forms a stable complex with Kv3.1b

We carried out parallel coimmunoprecipitation experiments to determine whether PAP associates with Kv31.b. Mammalian PAP genes generally encode two spliced variants. The PAP homologue of ACP-2 is isoform a (PAP.a), also known as the short isoform. This is the most abundantly expressed isoform and is predicted to be soluble. By contrast, isoform b (PAP.b), also known as the long isoform or TM-PAP, is a type I membrane protein with phosphatase activity predicted to be extracellular (29). The accessory subunit KCNE2 is the mammalian homologue of MPS-1 (23). Kv3.1b and KCNE2 interact in vitro (19–21), but KCNE2 does not possess kinase activity. Therefore, KCNE2 was not included in this study. Thus, Kv3.1b channels, epitope tagged with an HA tag in the C terminus, were individually cotransfected with PAP.a or PAP.b tagged with a V5 tag in the C terminus and immunoprecipitated using V5 antibodies (αV5) to pull down the complexes. In total lysates (first 3 lanes of the blot in Fig. 3B), Kv3.1b was detected in two forms, migrating with molecular masses of ∼80 and ∼100 kDa, respectively, which may reflect post-translational modified forms of the protein (the predicted molecular mass of Kv3.1b is 65.9 kDa). Native channels expressed in mouse brain also migrated with similar molecular masses (see Fig. 7A). By contrast, both PAP isoforms were detected in a single, ∼50-kDa band, consistent with their predicted molecular masses (43.9 and 47.9 kDa respectively). Coimmunoprecipitations are shown in the last 3 lanes of the blot in Fig. 3B. Both isoforms coimmunoprecipitated with the 80-kDa form of Kv3.1b, probably a consequence of the fact that this was the most abundant form.

KHT-1 is a substrate for ACP-2

MPS-1 constitutively phosphorylates the KHT-1 channel (10). To determine whether ACP-2 has enzymatic activity and whether it is capable of dephosphorylating KHT-1 in the presence of MPS-1, CHO cells transiently expressing these proteins were radiolabeled to equilibrium with ³²P[orthophosphate]. Incorporation of ³²P into KHT-1 was then evaluated after immunoprecipitation and quantified by densitometry analysis. As reported previously (10), there is substantial incorporation of ³²P into the channel, a 2-fold increase (band ratio 1.94±0.1; n=4, P<0.031), within binary complexes of KHT-1 and MPS-1 (Fig. 4A). When ACP-2 was added to the complex, ³²P incorporation was markedly inhibited (band ratio 1.1±0.06; n=3). Furthermore, a putatively enzymatically inactive ACP-2 variant, produced by replacing a key histidine in the N-terminal catalytic site with a glutamine (H34Q) failed to inhibit phosphorylation of KHT-1 (band ratio 1.27±0.1; n=3), even though the mutant phosphatase coimmunoprecipitated with the channel (Fig. 4C). We conclude that ACP-2 is an enzymatically active acid phosphatase and that it is capable of dephosphorylating the voltage-gated K⁺ channel KHT-1.

Figure 4.

Shaw-like channels are substrates of acid phosphatases. A) Representative autoradiograms showing ³²P incorporation into immunoprecipitated KHT-1-HA channels transfected alone or with the indicated subunits. Arrow indicates band corresponding to KHT-1. B) As in A, for Kv3.1b-HA channels expressed alone or together with the indicated PAP isoforms and variants. Prior to lysis, cells were incubated with 100 nM PMA for 20 min. C) Coimmunoprecipitations of KHT-1-HA channels with wild-type or H34Q ACP-2. D) As in C, for Kv3.1b-HA channel with wild-type or H44Q PAP.a isoform.

Kv3.1b is a substrate for PAP.a but not PAP.b

Kv3.1b is a substrate for PKC in vitro and in vivo (15, 30). To determine whether PAP could dephosphorylate Kv3.1b, CHO cells transiently expressing the channel with either one of the two PAP isoforms were, therefore, radiolabeled to equilibrium with ³²P and were then incubated with PMA to stimulate PKC phosphorylation. Immunoprecipitation revealed incorporation of ³²P into the channel, as reported previously (ref. 30 and Fig. 4B). In cells coexpressing the PAP.a isoform, ³²P incorporation into Kv3.1b was reversed (band ratio 0.54±0.03; n=3, P<0.017). In contrast, the PAP.b isoform or a putatively inactive PAP.a variant, H44Q, did not dephosphorylate Kv3.1b, even though they were coimmunoprecipitated with the channel (band ratio 0.85±0.09, n=2, Fig. 4B; and 0.81±0.08, n=2, Fig. 4D for PAP.b and H44Q, respectively). It should be noted that the failure of the two phosphatase proteins to dephosphorylate Kv3.1b had different causes. PAP.b was probably active, but its activity is extracellular (29). In contrast, H44Q is cytosolic, but its catalytic activity was impaired due to the His to Gln replacement. We conclude that PAP.a can specifically reverse PKC-mediated phosphorylation of Kv3.1b.

ACP-2 functionally modifies KHT-1 current

To characterize the effects of ACP-2 on the functional properties of their respective channel partners, we recorded whole-cell currents in CHO cells transfected with KHT-1 alone or with MPS-1, ACP-2, or both, using the patch-clamp technique. Representative recordings are shown in Fig. 5A, and quantitative analysis of the currents is summarized in Table 1. In response to voltage steps ranging from −80 to +120 mV, KHT-1 channels alone conducted robust, rapidly activating outward currents. Coexpression with MPS-1 significantly reduced current amplitude—an effect specifically due to its constitutive phosphorylation of KHT-1—without altering the voltage dependence of activation or the channel's surface expression (10).

Figure 5.

Acid phosphatases rescue the currents of Shaw-like channels. A) Representative whole-cell currents recorded in CHO cells expressing the indicated C. elegans subunits were elicited with 1-s voltage steps from −80 mV (holding potential) to +120 mV in 20-mV increments. B) As in A for the mammalian subunits. Where indicated, cells were preincubated with 100 nM PMA for 10 min prior to recording.

Table 1.

Electrophysiological properties of KHT-1 and Kv3.1b channels

Channel	σ120 (pA/pF)	V1/2 (mV)	Vs (mV)	n
C. elegans
KHT-1	279 ± 76	85 ± 4	20 ± 1	31
KHT-1+MPS-1	61 ± 7; P < 0.03	91 ± 7	25 ± 2	14
KHT-1+ACP-2	194 ± 42	73 ± 10; P < 0.07	37 ± 3; P < 0.008	21
KHT-1+MPS-1+ACP-2	186 ± 38	70 ± 7 P < 0.04	36 ± 4; P < 10−5	19
KHT-1+MPS-1+H33Q	56 ± 14; P < 0.007	69 ± 7 P < 0.02	34 ± 5; P < 0.01	23
Mammalian
Kv3.1b	4984 ± 302	22 ± 3	19 ± 1	12
Kv3.1b	3912 ± 271	25 ± 2	19 ± 1	32
Kv3.1b+PAP.a	5023 ± 234; P < 0.02	26 ± 2	17 ± 1	24
Kv3.1b+PAP.b	3725 ± 357	24 ± 3	18 ± 1	21
Kv3.1b+H44Q	3845 ± 301	22 ± 5	18 ± 1	15

Currents were elicited by voltage jumps from −80 to +120 mV in 20-mV increments. Mammalian cells, except for first row, were preincubated with 100 nM PMA for 10 min prior current recording. Current density, σ₁₂₀, was calculated by normalizing the steady-state current at +120 mV to the cell capacitance. Half-maximal voltage, V_1/2, and slope factor, V_s, were calculated by fitting macroscopic conductance-V relationships (G) to the Boltzmann equation, G(V) = I/(V − V_rev) = G_max/[1 + e^{(V_1/2 − V)/V_s}], where I is the macroscopic steady-state current, V is the applied voltage, and V_rev is the computed Nernst potential for K⁺ at the experimental concentrations (−87 mV). Data are indicated as means ± se. Number of observations, n, is reported in the last column. Statistically significant differences from KHT-1 or Kv3.1b alone were estimated using the Student's t test.

ACP-2 alone was capable of modifying the characteristics of KHT-1 currents. Coexpression of KHT-1 with ACP-2 produced a marked shift in the half-maximal voltage for activation (V_1/2,) and slope factor (V_s) and only a moderate decrease in the magnitude of the steady-state current. Under these conditions, however, association of the channel with MPS-1 had no further effect. Ternary complexes formed with KHT-1, MPS-1, and ACP-2 retained all the attributes of KHT-1-ACP-2 binary complexes. In particular, the amounts of current carried by these channels corresponded to the dephosphorylated state of KHT-1. Accordingly, current amplitudes in channels formed with the H34Q mutant phosphatase were markedly diminished (as a result of MPS-1-mediated phosphorylation), while they retained the shifts in the V_1/2 and V_s that result from the stable association of the phosphatase with the channel complex. In summary, ACP-2 induces multiple modifications in the KHT-1 current. Its activity acts to maintain KHT-1 current at normal levels, and it is dominant over the activity of MPS-1.

PAP phosphatase reverses the effects of PKC phosphorylation on Kv3.1b currents

We next characterized Kv3.1b currents alone or with the PAP phosphatases. Kv3.1b channels expressed in CHO cells conducted robust, rapidly activating currents that were decreased by ∼30% on treatment with PMA (to stimulate PKC activity) in agreement with previous studies (refs. 31, 32; Fig. 5B; and Table 1). Coexpression with PAP.a following PMA treatment (PMA preincubation was implemented in all experiments involving the phosphatase) restored the magnitude of the current to control values, with no significant shifts in voltage dependence or slope of activation. Consistent with the biochemical data, neither PAP.b nor the H44Q mutant induced modifications in the Kv3.1b current. Biotinylation experiments showed that the fraction of Kv3.1b protein at the plasma membrane, normalized to the total Kv3.1b protein, was 0.17 ± 0.08 and 0.15 ± 0.09 in CHO cells expressing Kv3.1 alone or with PAP (n=2 experiments, data not shown), respectively. This ruled out the possibility that changes in current magnitude were due to a corresponding increase in the number of channels at the plasma membrane rather than PAP-mediated dephosphorylation of Kv3.1b. Thus, the fact that PAP.a acted to rescue the amplitude of the macroscopic current without apparently altering the gating characteristics of the channel suggests that the phosphatase may act to specifically reverse PKC-mediated phosphorylation of Kv3.1b (31, 32).

ACP-2 is required for touch neuron function

To probe the physiological role of ACP-2 in mechanosensation, we compared the behavior of worms lacking ACP-2 (acp-2-KO) with that of parental worms (N2 genotype). We also tested transgenic worms in which either the wild-type acp-2 gene (including its promoter sequence, Pacp-2::wild-type-acp-2 transgenic worm, labeled WT in Fig. 6) or the acp-2 gene containing the t to g nucleotide mutation that encodes the H34Q variant (Pacp-2::h34q-acp-2 transgenic worm, labeled H34Q in Fig. 6) were introduced into the acp-2-KO background.

Figure 6.

ACP-2 is necessary for touch sensitivity. A) Forward and reverse responses to gentle touch in N2, acp-2-KO, P_acp-2::wild-type-acp-2 [WT; strain FDX(ses41)] and P_acp-2::h34q-acp-2 [H34Q; strain FDX(ses42)] transgenic worms. The mec-4(d) mutant, which induces necrotic touch-neuron death and, therefore, is touch insensitive, was used as control strain. n ≥ 34 worms/genotype. B) Time course of habituation to tap and recovery from habituation in N2 (solid squares) and acp-2-KO (open squares) worms. Note that the responses of acp-2-KO worms to the first taps were diminished compared to that of normal worms as if they were already habituated. n = 4–5 experiments/genotype. Interstimulus interval = 5 s. C) As in B for P_acp-2::wild-type-acp-2 (solid circles) and P_acp-2::h34q-acp-2 (open circles). n = 3 experiments/genotype. Worms were analyzed and photographed with an Olympus SZX7 microscope equipped with a digital camera. Ten or more worms/genotype were tested in a single experiment. Statistically significant differences between pairs of means were calculated with the Student's t test. Statistically significant differences between time course of habituation to tap in N2 and acp-2-KO (B) and between P_acp-2::wild-type-acp-2 and P_acp-2::h34q-acp-2 (C) were calculated with ANOVA, yielding P < 0.01 in both cases. *P < 0.05, **P < 0.01.

We measured the sensitivity of the animals to gentle touch (body touch phenotype) and their ability to adapt to repeated mechanical stimulation (habituation to tap phenotype). Figure 6A shows that parental worms (N2 genotype) responded nearly 100% of the time to gentle touches delivered on the head and tail. By contrast, the responses of acp-2-KO worms were significantly diminished, but introduction of the acp-2 gene (wild type) into the KO genotype restored susceptibility to gentle touch. The enzymatically inactive H34Q variant, however, failed to rescue touch sensitivity.

ACP-2 controls habituation to tap

Parental N2 worms quickly habituated to taps on the Petri dish when the taps were repeated every 5 s (Fig. 6B, left panel). When acp-2-KO worms were subjected to the same stimulus protocol, they responded weakly (as if they were already partially desensitized to taps; Fig. 6B, arrow) and did not completely habituate. Habituation behavior was rescued in transgenic worms expressing the wild-type ACP-2 but not in the worms expressing H34Q mutation (Fig. 6C, left panel). It must be noted that the inability of acp-2-KO and H34Q worms to habituate to tap simply reflected their defective touch sensitivity (Fig. 6A). This did not necessarily imply that ACP-2 has a direct role in the habituation pathway. Moreover the acp-2-KO and H34Q strains recovered from habituation partially and more slowly than parental or wild-type ACP-2 worms (Fig. 6B, C; right panels) a fact that presumably reflected incomplete KHT-1 dephosphorylation (10). In summary, the activity of ACP-2 is essential for the function of the touch neurons and is consistent with an active role in modulating the KHT-1-MPS-1 complex.

PAP and Kv3.1b coimmunoprecipitate in mouse brain

We carried out coimmunoprecipitation experiments to determine whether PAP and Kv3.1b interact in vivo. A representative Western blot is shown in Fig. 7. Thus, native proteins from cell membranes purified from mice brains were immunoprecipitated with either anti-Kv3.1b or anti-PAP and transferred on the same blotting membrane along with controls [nonimmune rabbit serum (NIRS)]. For Western visualization, the membrane was cut in half. One half was stained with anti-Kv3.1 (Fig. 7A, top blot), and the other half was stained with anti-PAP (Fig. 7A, bottom blot). Notably, Kv3.1 immunoprecipitates pulled down PAP. Conversely, PAP immunoprecipitates did not pull down Kv3.1 channels. This was probably a consequence of the relatively low abundance of PAP in our preparations (in the input lane, corresponding to 5% of the immunoprecipitated sample, it was not possible to see any band corresponding to PAP). Figure 7B shows the depletion of the IgG band obtained by cross-linking the antibodies with the agarose beads (see Materials and Methods). PAP and IgG run with similar molecular masses (∼55 kDa) and overlap in a Western blot. We conclude that Kv3.1b and PAP interact in mouse brain.

PAP and Kv3.1b colocalize in specific parts of mouse brain

Both PAP and Kv3.1b are widely distributed in the mouse brain. To determine the colocalization of PAP and Kv3.1b, we incubated sagittal sections of 2-mo-old mouse brain and spinal cord with PAP and Kv3.1b antibodies and imaged immunofluorescence using laser-scanning confocal microscopy. Strongest colocalization of PAP and Kv3.1b signals were detected in the periventricular cells of the ventricular region (Fig. 8). The periventricular cells are progenitor stem cells from a neurogenic lineage of endocrine cells (33) known to express Kv3.1b. We also found colocalization in sections of brain stem. In particular, prominent colocalization of PAP and Kv3.1b was detected in the gigantocellular reticular nucleus or Gi region. Sparse puncta of colocalization of these two proteins was found on axon fibers in the spinal cord. In contrast, we did not observe colocalization in hippocampus, where separate populations of neurons were positive for either PAP or Kv3.1b.

Figure 8.

Kv3.1b and PAP colocalize in distinct regions of mouse brain. Kv3.1b (green) and PAP (red) were strongly colocalized in periventricular cells in ventricular zone and in portions of the brainstem in particular the gigantocellular reticular nucleus. Some fiber tracts of the spinal cord also had colocalization. However, there was no overlap in Kv3.1b and PAP immunostaining in hippocampal regions. Scale bars = 20 μm.

DISCUSSION

Voltage-gated K⁺ channels of the Shaw-related family are broadly expressed in mammalian brains (11). Accordingly, the large spectrum of behavioral defects associated with disruption of Shaw-like channel's activities in mice underscores the criticality of these proteins to health issues and, therefore, the importance of elucidating the signaling mechanisms that regulate their function (12–18). Here, we describe an evolutionarily conserved mode of regulation of Shaw-like K⁺ channels by acid phosphatases.

In heterologous expression systems, C. elegans ACP-2 and its mammalian homologue, PAP.a, coimmunoprecipitated with KHT-1 and Kv3.1b. In particular, ACP-2 immunoprecipitates pulled down both KHT-1 and MPS-1, suggesting that ACP-2, KHT-1, and MPS-1 formed a stable, ternary complex. Both phosphatases reversed channel's phosphorylation mediated by, respectively, the accessory subunit MPS-1 and PKC. By contrast, PAP.b or inactive phosphatase variants (H34Q-ACP-2 and H44Q-PAP.a) failed to reverse phosphate incorporation. In the mammalian case, while both the PAP.b isoform and the H44Q variant coimmunoprecipitated with Kv3.1b, the causes for their failure to dephosphorylate the protein were different. The activity of PAP.b is probably extracellular. In contrast, H44Q truly lacked catalytic activity. Electrophysiological analysis recapitulated biochemical data. They showed that wild-type ACP-2 and PAP.a but not PAP.b or the inactive mutants could reverse the effect of phosphorylation—a marked macroscopic current decrease—in their respective channel partners. In addition, ACP-2 shifted the V_1/2 and V_s of KHT-1, in a fashion consistent with the formation of a stable complex, whereas PAP had no effects on the gating of Kv3.1b. These findings prompted the question as to whether these interactions exist in vivo and what are their physiological implications.

In C. elegans mechanosensory neurons, KHT-1 is dephosphorylated under basal conditions and becomes phosphorylated by MPS-1 (autophosphorylation of the channel complex) in response to repetitive stimuli (10). KHT-1 autophosphorylation constitutes a central step in the neuron's adaptive response to touch. This catalytic process leads to a robust decrease in KHT-1 current, which causes temporary desensitization to the mechanical stimulus. To prevent the channel from autophosphorylating during basal conditions, a third protein must stably interact with it. The evidence presented in this study argues that ACP-2 is the candidate protein. ACP-2 was detected in the touch neurons, the same that express KHT-1 and MPS-1. Acp-2-null or transgenic worms expressing the inactive H34Q ACP-2 mutant were touch defective and unable to habituate. Hence, ACP-2 is crucial for the function of the touch sensory neurons, specifically through its enzymatic activity. Behavioral observations are consistent with the notion that KHT-1 is constitutively phosphorylated in acp-2-KO or H34Q animals. This implies that ACP-2 may maintain KHT-1 dephosphorylated under basal conditions and also reset the channel during the recovery from habituation (without the need of another, unknown protein). In conclusion, we propose a working model of neuronal adaptation to touch in C. elegans that predicts that ACP-2 is an integral component of the KHT-1-MPS-1 complex. Accordingly, repetitive tapping triggers the disengagement of ACP-2 from the channel complex enabling autophosphorylation. The modifications in the current that follow alter touch neuron's excitability causing temporary desensitization to the stimulus. Preliminary support to this model comes from in vitro assays showing that KVS-1, MPS-1, and ACP-2 can individually interact with each and form stable ternary complexes. Future studies will further test this model, including the existence, in vivo, of KVS-1, MPS-1, and ACP-2 ternary complexes by the means of coimmunoprecipitation and colocalization, which were not attempted here for lack of specific antibodies against ACP-2.

PAP and Kv3.1b coimmunoprecipitated in mouse brain. They colocalized in specific regions of brain, including periventricular cells, sections of brain stem and on axon fibers in the spinal cord. Such modulation is likely to be region-specific because there are multiple brain regions that express Kv3.1b and in which we could not detect PAP. Thus, PAP is an interacting partner of Kv3.1b in vivo. The physiological significance of PAP modulation of Kv3.1b remains to be elucidated. However, considering the similarity with C. elegans ACP-2, it is tempting to speculate that PAP may also be involved in mechanisms of neuronal adaptation. Moreover, Kv3.1b has been already implicated in this type of process. In rat auditory brain stem, Kv3.1b is phosphorylated by PKC under basal conditions, and it becomes dephosphorylated in response to changes in the ambient acoustic environment (15). Pharmacological studies have suggested that activity-dependent dephosphorylation of Kv3.1b can occur through activation of phosphatases PP1/PP2A (15). We found that PAP strongly colocalized with Kv3.1b in gigantocellular reticular nucleus. This region of the brain controls a variety of behaviors ranging from arousal to sensory responsiveness. The activity of mGi neurons has been associated with cortical desynchronization (34) motor modulation (35), pain modulation (36), and orientation responses (37), functions that are subject to habituation. Thus, it is possible that PAP may underlie adaptation responses in mGi neurons by modulating Kv3.1b current.

The short PAP isoform is abundantly secreted by epithelial cells of the prostate gland, and its levels are elevated in patients with prostatic carcinoma. In fact, the protein has been used for years to monitor the progress of prostate cancer and the efficacy of the therapy (reviewed in ref. 38). For these reasons, PAP.a was considered to be prostate specific. Only recently, Graddis et al. (39) and, in an independent study, Quintero et al. (29) reported wide expression of both PAP isoforms in a variety of nonprostatic tissues, including brain. Taylor-Blake and Zylka (40) detected PAP.a expression in peptidergic and nonpeptidergic neurons of mice and rats. Notably, in the axon terminals of these neurons, there was significant overlap of PAP with PKC-γ. This investigation corroborates the emerging notion that PAP plays a general role in mammalian nervous systems.

This study provides an example of how information inferred from a simple system, such as C. elegans, can find fertile ground in higher organisms. It is likely that this worm can teach us many other things about mammalian acid phosphatases and their signaling pathways. For example, the mechanisms that regulate the recruitment and disengagement of ACP-2 from the KHT-1-MPS-1 complex awaits elucidation, and we argue that they are conserved to a considerable extent in mammalian brains. The same logic argues that C. elegans could be exploited to study key aspects of acid phosphatase function that go beyond neuronal adaptation, also in lieu of the fact that the protein exhibits broad expression in the worm. In summary, it is easy to predict that the elucidation of the role of acid phosphatases in nervous system will advance our understanding of basic neuronal processes.