Large-conductance Ca2+-activated potassium (BK) channels are composed of pore-forming α-subunits and auxiliary β-subunits. The α-subunits are widely expressed in many cell types, whereas the β-subunits are more tissue specific and influence diverse aspects of channel function. In the current study, we identified the presence of the smooth muscle-specific β1-subunit in murine colonic tissue using Western blotting. The native β1-subunits migrated in SDS-PAGE as two molecular mass bands. Enzymatic removal of N-linked glycosylations from the β1-subunit resulted in a single band that migrated at a lower molecular mass than the native β1-subunit bands, suggesting that the native β1-subunit exists in either a core glycosylated or highly glycosylated form. We investigated the functional consequence of deglycosylating the β1-subunit during inside-out single-channel recordings. During inside-out single-channel recordings, with N-glycosidase F in the pipette solution, the open probability (Po) and mean open time of BK channels increased in a time-dependent manner. Deglycosylation of BK channels did not affect the conductance but shifted the steady-state voltage of activation toward more positive potentials without affecting slope when Ca2+ concentration was <1 μM. Treatment of myocytes lacking the β1-subunits of the BK channel with N-glycosidase F had no effect. These data suggest that glycosylations on the β1-subunit in smooth muscle cells can modify the biophysical properties of BK channels.
- peptide N-glycosidase F
- large-conductance Ca2+-activated K+ channels
- N-linked glycosylation
- single-channel recording
- auxiliary subunit
extracellular domains of many plasma membrane proteins, including ion channels, are N-glycosylated. This results in transfer of an oligosaccharide to a nascent polypeptide at the amino acid sequon (Asn-X-Ser/Thr) in the endoplasmic reticulum (ER). Glycoproteins are further processed and new sugars are added in the Golgi complex, resulting in a mature glycoprotein. Carbohydrates on membrane proteins affect the efficiency of protein folding, trafficking, targeting, and stability (10, 25).
Recently, glycosylation has been shown to affect ion channel gating (8, 20, 24). For example, mammalian Kv1 subfamily members (Kv1.1–1.5) are N-glycosylated in the first extracellular loop between transmembrane domains S1 and S2. Expression of Kv1.1 in glycosylation-deficient Chinese hamster ovary (CHO) cells (Lec-1) resulted in channels with a positive shift in the voltage dependence of half activation (V1/2) and slower activation kinetics compared with wild-type Kv1.1 channels expressed in control CHO cells (Pro-5). The change in kinetics was attributed to loss of negative charges and a shift in the surface potential that influences the local electric field detected by the voltage sensor (20). Kv1.1N207Q channels, which lack the amino acid sequon for glycosylation, also display slower activation kinetics, a positive shift in V1/2, slower C-type inactivation, and a shallower slope in the conductance vs. voltage (G-V) relationship compared with wildtype channels. Modeling suggested that the changes in kinetics are due to modification of surface potential and cooperative subunit interactions (24). Glycosylation of accessory proteins can also affect channel kinetics. KvLQT1 and the accessory protein minK form heterooctomeric delayed rectifier K+ channels (IsK) that have one and two potential N-glycosylation sites, respectively. When the IsK channels are expressed in Lec-1 cells, channel kinetics are altered with respect to channels expressed in Pro-5 cells. N-glycosylation appears to contribute more to the functional surface charge of IsK than to homotetrameric KvLQT1 channels (8).
The pore-forming α-subunits of the large-conductance Ca2+-activated K+ (BK) channels are expressed in a variety of tissues, and accessory β-subunits modify Ca2+-sensitivity and pharmacological and biophysical characteristics of the channel (6, 14, 15). In smooth muscle cells, BK channels are comprised of four α-subunits and four β1-subunits forming a heterooctomer. Murine BK channel α-subunits have extracellular domains containing one possible N-glycosylation site, and the 120-amino acid extracellular loop of β1-subunits has two putative glycosylation sites (21). Enzymatic removal of the sialic acid associated with N-linked oligosaccharides did not affect the functional properties of BK channels in cultured rat hippocampal neurons (23); however, the subunit composition of the BK channels in these cells was not characterized.
In the present study, we used the peptide N-glycosidase F (PNGase F) to enzymatically deglycosylate membrane proteins in murine colonic smooth muscle cells and recorded BK channel activity during this treatment. Comparative studies were performed on cells lacking β1-subunits (β−/−) to confirm that effects were due to removal of glycosylations from the accessory protein and not due to deglycosylation of random membrane proteins. Our data suggest that glycosylations on the β1-subunit affect BK channel activity by a nonsurface potential mechanism.
BALB/c and β1−/− mice (generously provided by Drs. Rich Aldrich and Robert Brenner, Stanford University) (60–90 days old) of either sex were anesthetized with isoflurane and killed by decapitation. Mice were maintained and experiments performed in accordance with the National Research Council's Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the Institutional Animal Use and Care Committee at the University of Nevada, Reno.
After death, colons were removed from the mice and opened along the mesenteric border. Luminal contents were removed by washing with Krebs-Ringer bicarbonate buffer. Tissues were pinned to the base of a Sylgard-coated dish, and the mucosa and submucosa were dissected away. The tissues were equilibrated in Hanks' Ca2+-free solution for 60 min and then digested at 37°C for 16 min without agitation in a solution containing collagenase F (Sigma Chemical, St. Louis, MO) (3). At the end of the digestion period, the tissues were washed four times with Ca2+-free Hanks' solution to remove the enzyme and then triturated with blunt pipettes of decreasing tip diameter to free single smooth muscle cells.
Unitary currents resulting from openings of single channels were measured in inside-out and outside-out patches. Experimental (bath and pipette) solutions contained the following: 140 mM KCl, 10 mM HEPES, 5 mM Tris, and 1 mM EGTA (pCa2+ 7.3–6.1) or HETDA (pCa2+ 5.6–5) and various amounts of CaCl2 (for inside-out experiments, pipette solution had a pCa2+ of 7.3). Solutions with various concentrations of free Ca2+ were calculated using Max Chelator, provided by Chris Patton (http://www.stanford.edu/∼cpatton/maxc.html), and verified with Ca2+ electrodes (2) calibrated with Ca2+ standards from Molecular Probes (Invitrogen, Carlsbad, CA). High-resistance seals (>10 GΩ) were obtained using borosilate electrodes (7–12 MΩ). The data were digitized at 5 kHz and filtered at 2 kHz, using pClamp Software (version 9.2; Molecular Devices, Sunnyvale, CA). G-V curves were created by stepping patches from 0 mV to various test potentials for 500 ms and repeating at least three times. Open probability (Po) of BK channels at +50 mV was measured during a 60-s step. All single-channel recordings were made with 4-aminopyridine (5 mM) and nicardipine (1 μM) in the pipette and bath solutions to decrease contamination from delayed rectifier K+ and voltage-dependent Ca2+ channels.
Western blot analysis.
Frozen colons were homogenized on ice in a glass tissue grinder containing the following: 10% sucrose, 10 mM HEPES, 1 mM EDTA, 1 mM PMSF, and protease inhibitors. Large tissue debris and nuclear fragments were removed by low-speed centrifugation (1,000 g, 10 min). The membrane fraction protein was isolated by a high-speed spin (100,000 g, 1 h). Samples were read for total protein concentration using the Bio-Rad method, with bovine γ-globulin as a standard. Proteins from the membrane fractions, some treated with PNGase F (New England Biolabs, Ipswich, MA) as the manufacturer described, were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Anti-KCa1.1(1098–1196) from Alomone Labs (Jerusalem, Israel) was used for the 1′ antibody for the α-subunit blots, and anti-Maxi K+ Beta(90–103) (Affinity BioReagents, Deerfield, IL) was used for β1-subunit blots. The Odyssey infrared imaging system controlled by Odyssey program (version 1.2.15; Li-Cor, Lincoln, NE) was used to detect the 2′ antibody, Alexafluor 680 goat anti-rabbit IgG (Molecular Probes, Eugene, OR).
β−/− mice were confirmed by genotyping each animal studied. Genomic DNA was isolated from previously frozen mouse ear punches, ∼2-mm round pieces, using the Qiagen DNeasy tissue kit (Qiagen, Valencia, CA). PCR was completed on a Gene Amp PCR System 2700 (Applied Biosystems, Foster City, CA). Primers used in PCR were chosen to target GadPH, a housekeeping gene, β1Slo-wild-type, and/or β1Slo-knockout genes. Primers targeted to the β1Slo-knockout and β1Slo-wild-type gene were designed from information provided by Drs. Rick Aldrich and Olga Sprina (Stanford University). PCR products were separated on a 2% agarose gel running at 70 V for 45 min and then visualized using an ethidium bromide stain.
Mean open time and NPo were determined, using the Windows Analysis of Single Channel Data program (WinASCD; Dr. Guy Droogmans, Laboratorium voor Fysiologie, Leuven, Belgium). Po was converted from NPo by calculating the number of channels in each patch. The number of channels per patch was calculated by dividing the current at the end of a 200-mV step (top of the 100 nM Ca2+ steady-state activation curve) by the calculated single-channel amplitude for 200 mV (calculated single-channel conductance times 0.2 V). The conductance of individual channels was calculated by dividing the measured single-channel amplitude by holding potential. At least three trace recordings were averaged and analyzed, using Clampfit from pCLAMP 9.0 (Molecular Devices), to create individual current-voltage relationship data. Current-voltage relationships were converted and plotted to normalized conductance-voltage relationships, using Prism 3.02 (GraphPad Software, San Diego, CA). Data are expressed as the means ± SE of n no. of cells. Statistical analyses were made with SigmaStat version 2.03 (San Rafael, CA) using Student's t-test or paired t-test. The threshold for statistical significance was set as P < 0.05.
Treatment of protein samples with PNGase F.
Membrane-enriched fractions from murine colonic muscles were probed with antibodies specific for BK α- and β1-subunits. A dark ∼29-kDa band and lighter bands (∼26 to ∼24 kDa) were observed when probing for β1-subunits (see Fig. 1B, PNGase F “−” lane). Treatment of extracts with PNGase F (750 units) reduced the multiple bands of the β1-subunit to a single band with a molecular mass of ∼22 kDa (see Fig. 1B, PNGase F “+” lane). No change was observed in BK α-subunit molecular mass after digestion with PNGase F (Fig. 1A). These results suggest that both α- and β1-subunits of BK channels are found in murine colonic muscles. In colonic muscle cells, BK channels are composed of nonglycosylated α- and double-glycosylated β1-subunits.
Effect of PNGase F on single-channel openings.
Inside-out excised patch recordings were performed in symmetrical 140 nM K+ and 100 nM Ca2+ solutions to investigate the effects of PNGase F on openings of single BK channels. Patches were held at +50 mV, and PNGase F (1,000 units) was included in the pipette solution. Inclusion of PNGase F caused a time-dependent increase in the Po of BK channels (Fig. 2A) that was not observed in control recordings of BK channel currents. The time-dependent increase in Po of BK channels started within 5–10 min after establishing the inside-out patch configuration and peaked at about minute 30 of recording (Fig. 2C). Po of BK channels increased from 0.027 ± 0.001 at initiation of the inside-out configuration to 0.155 ± 0.020 after 30 min of exposure to PNGase F (P < 0.005, n = 5). The increase in Po was associated with an increase in mean open time from a tau (τ) of 10.97 ± 1.41 ms to 13.97 ± 1.2 ms (P < 0.01, n = 5). There was no significant change in the conductance of BK channels after PNGase F treatment (i.e., 264.1 ± 3.4 pS before vs. 259.7 ± 4.8 pS after PNGase F; P > 0.1, n = 5) (Fig. 2D). Boiling PNGase F for 10 min before adding it to the pipette solution abolished its effects on BK channels (Fig. 2B). Po of BK channels was 0.0357 ± 0.0025 at time 0 and 0.0266 ± 0.0063 after 30 min of exposure to boiled PNGase F (P < 0.1, n = 4). Boiled PNGase F also had no effect on mean open time (τ = 10.84 ± 0.27 vs. 8.61 ± 0.88 ms; P > 0.1, n = 5) or conductance (254.0 ± 2.9 vs. 260.6 ± 6.7 pS; P > 0.5, n = 5). These results suggest that deglycosylation affects the Po and kinetics of native smooth muscle BK channels.
PNGase F treatment on steady-state activation.
The pore-forming α-subunits of the BK channel are voltage and Ca2+ sensitive. If deglycosylation affects steady-state Po of BK channels, it is possible that glycosylation also affects the steady-state activation of BK channels at various Ca2+ concentrations. We tested this hypothesis by performing inside-out recordings of BK channels in a high-flow, low-volume bath that allowed rapid exchange of solutions. Patch potentials were held at 0 mV and stepped to different potentials to determine steady-state activation with and without PNGase F in the pipette solution. Potentials were stepped positive until normalized BK conductance was maximal, and normalized G-V curves were plotted. PNGase F did not affect the population mean amplitude or activation kinetics of BK channels (Fig. 3). The time constant of activation at +150 mV was 27 ± 9 and 27 ± 3 ms in 50 nM free Ca2+ with or without PNGase F, respectively (n = 5). Similar recordings were made from pCa 7.3 to 5. PNGase F shifted V1/2 to more negative values at lower free Ca2+ concentrations (P < 0.05, n = 5) (Fig. 4A). However, at higher free Ca2+ concentrations, there was no shift in V1/2 in response to PNGase F (P > 0.5, n = 4) (Fig. 4B). PNGase F treatment only affected V1/2 at free Ca2+ concentrations <1 μM (Fig. 4C, Table 1). Summation of data from all free Ca2+ concentrations tested shows that the slopes of the G-V curves were not affected by PNGase F (P > 0.5, Table 1).
Effect of increased divalent cations on BK channels.
We tested the effects of screening membrane negative charges on the voltage dependence of activation of BK channels, because deglycosylation would be expected to produce a similar effect of reducing membrane negative charge. Outside-out recordings were made in symmetrical 140 mM K+ and 250 nM free Ca2+ solutions with an agar bridge to eliminate changes in reference potential. After holding at 0 mV, the patches were stepped from −40 to +160 mV, and normalized G-V curves were plotted. Addition of Mn2+ (20 mM) caused a small decrease in BK current amplitude and a rightward shift in the G-V curve toward more positive potentials (Fig. 5, A and B). V1/2 was shifted from 70.1 ± 5.5 to 105.1 ± 5.4 mV (P = 0.001, n = 4) without a change in slope (P > 0.1, Fig. 5C). These data suggest that screening of negative charges on colonic myocyte membranes causes a rightward shift in the voltage of activation of BK channels.
PNGase F treatment did not alter BK channels from the β1-deficient mice.
Our data suggest that glycosylation influences the voltage of activation of BK channels. BK channels in colonic myocytes are composed of pore-forming α-subunits that do not appear to be glycosylated and glycosylated accessory β1-subunits (see Fig. 1). The extracellular loops of β1-subunit may lie near the pore of the BK channels (9, 12, 13). Thus it is possible that the sites of glycosylation in β1-subunits may also lie close to the voltage sensor of the α-subunits of BK channels, and interactions between glycosylated β1-subunits and the voltage sensor might explain our results. If deglycosylation of β1-subunits is responsible for the changes in voltage dependence of BK channels we noted, then loss of the β1-subunits should eliminate effects of PNGase F. Therefore, we conducted experiments on colonic myocytes from β1-deficient (β1−/−) mice.
Patches from β1−/− colonic myocytes displayed BK channels that had a single-channel conductance of 254.5 ± 4.1 pS and were sensitive to charybdotoxin (ChTX). Inside-out patches were held at +50 mV in 100 nM free Ca2+, and PNGase F (1,000 units) was included in the pipette solution. There was no time-dependent increase in the Po of BK channels in patches from β1−/− colonic myocytes (i.e., Po was 0.0114 ± 0.0021 at the initiation of the inside-out configuration and 0.0152 ± 0.0058 after 30 min of exposure to PNGase F; P < 0.5, n = 6) (Fig. 6, A–E). It should be noted that BK channels from β1−/− myocytes displayed a lower mean open time (1.92 ± 0.30 ms) compared with channels of wild-type myocytes, and this is consistent with previous single-channel studies demonstrating the effects of β1-subunits on the BK channel Po (18). PNGase F treatment did not significantly affect the mean open time (Fig. 6F, P > 0.5). The time constant of activation was much faster in myocytes from β1−/− mice compared with wild-type cells, and this parameter was also unaffected by PNGase F (Fig. 7, A and B, τ of 4.02 ± 1.79 vs. 4.61 ± 1.41 ms; P > 0.1, n = 5). Similarly, steady-state activation was not shifted with PNGase F treatment of patches from the β1−/− myocytes (Fig. 7C, V1/2 109.6 ± 1.10 vs. 114.6 ± 3.17 mV; P > 0.1, n = 3). These data suggest that deglycosylation of β1-subunits of BK channels is responsible for the effects of PNGase F treatment.
Our data suggest that glycosylation can alter BK channel activity in smooth muscle cells. We found that β1-subunits of BK channels are glycosylated in colonic smooth muscle cells, but no evidence was found to suggest that the pore-forming α-subunits are glycosylated in native cells. When BK channels were enzymatically deglycosylated in situ, an increase in the Po of BK channels and a leftward (negative) shift in the voltage dependence of activation were observed. The effects of deglycosylation were not observed when BK channels lacked β1-subunits.
Studies of deglycosylation of K+ channels have mainly focused on glycosylation of pore-forming subunits, as with voltage-gated K+ channels Kv1.1 and Kv1.2. Glycosylation sites in these channels are located between the S1 and S2 linker (24, 27). The pore-forming α-subunit of BK channel is unlike other Kv channels because it consists of seven transmembrane domains (S0–S6). The hydrophobic segment (S0) creates an exoplasmic NH2 terminus (16), and in some species (e.g., Drosophila) the extracellular NH2 terminus is glycosylated. Interestingly, the Drosophila α-subunits are not modified by co-expression with the mammalian accessory β1-subunit (21). α-Subunits of the human and mouse BK channel lack the extracellular NH2 terminus sequon. There is a consensus N-glycosylation site located in the short loop between S3 and S4 of α-subunits of murine BK channels (position 200, sequon NRSW); however, this site is very close to the predicted S3 transmembrane domain and only slightly over the predicted threshold for glycosylation. [N-glycosylation sites were predicted by use of the NetNGly 1.0 server (Technical University of Denmark, http://www.cbs.dtu.dk), using the amino acid sequence of BK α- and β1-proteins that can be accessed through the National Center for Biotechnology Information (NCBI) Protein Database under NCBI accession numbers Q08460 and NP_112446, respectively.] No studies have shown that this site can be enzymatically deglycosylated; this is consistent with the lack of effect of PNGase F we observed on the molecular mass of α-subunit and the lack of effects of PNGase F on channel function of BK channels lacking β1-subunits.
Glycosylation of β1-subunits of BK channel has been documented previously (11, 12). Western blots of in vitro translated human β1-subunits (KCNMB1) demonstrated a nonglycosylated form with a mass of 22 kDa, a fully glycosylated form with an apparent mass of 29 kDa, and a partially glycosylated β-subunit with an apparent mass of 26 kDa. The higher molecular mass forms of human KCNMB1 were reduced to a single 22-kDa band by digestion with PNGase F (11). We performed Western blots on extracts of colonic smooth muscle cells and obtained evidence for similar glycosylated forms of β1-subunits of BK channels in native colonic smooth muscles. We found no evidence of glycosylation on the α-subunit with the same techniques. Thus our data suggest that β-subunits of BK channel exist in a glycosylated state in colonic smooth muscle cells.
Negative charges on and near ion channels affect the voltage dependence of activation, and screening of these negative charges with divalent cations shifts the voltage of activation of ion channels toward more positive potentials, or rightward on the G-V curve (7, 17). We observed this phenomenon in the present study when surface charges were screened with Mn2+. Ion channels lacking glycosylation sites or expressed in glycosylation-deficient cell lines also experience reduction in the local electrical field, and previous studies have demonstrated shifts in voltage dependence of activation consistent with a reduction in negative surface charges. Studies on K+ channels have reported rightward shifts in the V1/2 with loss of glycosylation (20, 24). For example, KvLQT1 co-expressed with accessory subunit minK displayed a rightward shift in V1/2 when expressed in glycosylation-deficient CHO cells (8). A leftward shift in the steady-state voltage of activation and increase in steady-state Po with loss of glycosylation have been described for BK channels in a study in which asparagine residues in the glycosylation sequon located in the S3-S4 linkers of α-subunits were replaced with alanine. This mutation did not affect trafficking or sorting of the mutant channels but caused a leftward shift in the voltage of activation (4). However, it is unclear from these studies whether the effects were due to the mutation or to the loss of glycosylation per se. We observed a leftward shift in the voltage dependence of activation of native BK channels on enzymatic deglycosylation. Thus our data are consistent with the interpretation that modulation of BK channel activity by glycosylation of β1-subunits involves a mechanism other than simple surface charge effects.
We found that deglycosylation of β1-subunits increased Po and shifted the steady-state activation curve of BK channels at Ca2+ concentrations <1 μM. At low Ca2+, glycosylation appears to destabilize an open state of BK channels or to stabilize channels in a closed state. Loss of β1-subunits eliminated this effect, suggesting that β1 glycosylations are situated in a manner that affects gating. The data also suggest that the fundamental interactions of glycosylated β1-subunits with α-subunits inhibit channel opening. Although this seems contradictory to the well-known enhancement in the Ca2+ sensitivity of BK channel that is conveyed by β1-subunits, effects of β1-subunits on Ca2+ sensitivity occur at Ca2+ concentrations above ∼300 nM (6, 15). In fact, it was shown that the V1/2 for BK channels was shifted toward more positive potentials by co-expression with β1-subunits at Ca2+ concentrations below ∼660 nM (6). These results may be explained by our data showing that glycosylation on the β1-subunits shifts the V1/2 of BK channel toward positive potentials at low Ca2+ concentrations. Above the [Ca2+] necessary for β1-subunits to enhance V1/2, termed [Ca2+]critical, the effectiveness of glycosylation to change V1/2 is reduced.
In summary, the activities of ion channels, including BK channels, are influenced by the co-expression of accessory β-subunits and other signaling/structural proteins (5, 19, 22, 26). Accessory subunits can modulate subcellular localization, stability, and channel activity. Smooth muscle BK channels are heavily dependent on β1-subunits for voltage and Ca2+ dependence. In the present study, we have demonstrated another means by which BK channels can be affected by β1-subunits. Our data suggest that glycosylation of β1-subunits modifies voltage sensing by BK channels. This effect is mainly seen at lower Ca2+ concentrations but within the physiological window. Although there have been no pathological abnormalities associated with loss or defects in glycosylation of ion channels, there are disorders that are associated with increased nonenzymatic glycation (e.g., type 1 and type 2 diabetes) (1). Thus the effects of glycosylation on ion channel function are important to understand.
This project was supported by National Institute of Diabetes and Digestive and Kidney Diseases Program Project Grant DK-41315.
We thank Drs. Rick Aldrich, Robert Brenner, and Olga Sprina of Stanford University for providing the β1−/− mice and technical information on genotyping these animals. We also thank Dr. Brian Perrino for technical assistance on Western blotting techniques and Dr. James Kenyon for helpful discussion and expertise on Ca2+ electrode measurements.
Present address of B. M. Hagen: University of Maryland Biotechnology Institute, Medical Biotechnology Center, 725 W. Lombard St., Suite 340, Baltimore, MD 21201.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society