Abstract
Voltage-gated K+ channels of the Kv2 family are highly expressed in brain and play dual roles in regulating neuronal excitability and in organizing endoplasmic reticulum-plasma membrane (ER-PM) junctions. Studies in heterologous cells suggest that Kv2.1 and Kv2.2 co-assemble with “electrically silent” KvS subunits to form heterotetrameric channels with distinct biophysical properties, but the prevalence and localization of these channels in native neurons are unknown. Here, using mass spectrometry-based proteomics, we identified five KvS subunits as components of native Kv2.1 channels immunopurified from mouse brain of both sexes, the most abundant being Kv5.1. We found that Kv5.1 co-immunoprecipitates with Kv2.1 and to a lesser extent with Kv2.2 from brain lysates and that Kv5.1 protein levels are decreased by 70% in Kv2.1 knock-out mice and 95% in Kv2.1/Kv2.2 double knock-out mice. RNAscope and immunolabeling revealed that Kv5.1 is prominently expressed in neocortex, where it is detected in a substantial fraction of Kv2.1/Kv2.2 positive neurons in layers 2/3, 5, and 6. At the subcellular level, Kv5.1 protein is coclustered with Kv2.1 and Kv2.2 at presumptive ER-PM junctions on the soma and proximal dendrites of cortical neurons. Moreover, in addition to modifying channel conductance, we found that Kv2/Kv5.1 channels are less phosphorylated and insensitive to RY785, a potent and selective Kv2 channel inhibitor. Together, these findings demonstrate that KvS subunits create multiple Kv2 channel subtypes in brain. Most notably, Kv2/Kv5.1 channels are highly expressed in cortical neurons, where their unique properties likely modulate the critical conducting and nonconducting roles of Kv2 channels.
Significance Statement
Voltage-gated Kv2 potassium channels play important roles in regulating neuronal excitability and organizing endoplasmic reticulum-plasma membrane (ER-PM) junctions in brain neurons. Here, we use mass spectrometry to identify five KvS channel subunits as components of native Kv2 channels in brain. The most abundant of these subunits, Kv5.1, is prominently expressed in the cortex, where it clusters at ER-PM junctions with Kv2 subunits in a subpopulation of cortical neurons. Kv5.1 expression depends on Kv2 subunits, and Kv2/Kv5.1 heteromeric channels differ in their biophysical and pharmacological properties. We propose that differential expression of Kv5.1 in subclasses of cortical neurons diversifies Kv2 channel function.
Introduction
The Kv2 family of voltage-gated K+ channels are highly expressed in brain and play important canonical roles as voltage-activated K+ -selective channels that regulate neuronal action potentials and membrane excitability (Guan et al., 2007; McKeown et al., 2008; Liu and Bean, 2014; Trimmer, 2015). Indeed, Kv2.1 KO mice are hyperactive, are susceptible to seizures, and have enhanced neuronal activity (Speca et al., 2014), and patients with de novo mutations in Kv2.1 have epileptic encephalopathy (Allen et al., 2020). In addition, Kv2 channels are highly clustered on neuronal cell bodies and proximal dendrites and have a separate structural role in organizing endoplasmic reticulum (ER)-plasma membrane (PM) junctions (Fox et al., 2015; Kirmiz et al., 2018b). These sites are formed by Kv2 channels in the PM binding to VAP proteins in the underlying ER (Johnson et al., 2018; Kirmiz et al., 2018a) in a phosphorylation-dependent manner (Lim et al., 2000; Kirmiz et al., 2018a), making these junctions dynamic and responsive to several types of stimuli (Misonou et al., 2004, 2005). Numerous signaling proteins are recruited to Kv2-containing ER-PM junctions, including L-type Ca2+ channels (LTCCs) and an array of Ca2+-regulated signaling proteins (Vierra et al., 2019), lipid handling proteins (Kirmiz et al., 2019; Sun et al., 2019), and protein kinase A (PKA) signaling proteins (Vierra et al., 2023). Thus, these microdomains are thought to serve as important hubs for somatodendritic Ca2+, lipid, and PKA signaling in brain neurons (Kirmiz et al., 2019; Vierra et al., 2019, 2021, 2023).
The Kv2 family consists of only two voltage-sensing and pore-forming α subunits (Kv2.1 and Kv2.2; McKeown et al., 2008; Trimmer, 2015). Kv2.1 and Kv2.2 are robustly expressed in multiple neuron types throughout brain including the neocortex and hippocampus (Trimmer, 1991; Maletic-Savatic et al., 1995; Bishop et al., 2015) and form both homo- and heterotetrameric channels, which have similar functional and structural properties (Kihira et al., 2010). Another potential source of Kv2 channel diversity, however, comes from the 10 modulatory or “electrically silent” K+ channel (KvS) subunit genes (Bocksteins et al., 2009; Bocksteins and Snyders, 2012; Bocksteins, 2016). The KvS subunits (Kv5.1, Kv6.1−6.4, Kv8.1–8.2, and Kv9.1–9.3) have sequence similarity to Kv2 subunits and are also pore-forming alpha subunits, but intriguingly they do not form functional channels by themselves. Rather, studies in heterologous cells suggest that KvS subunits co-assemble selectively with Kv2 subunits to form heterotetrameric channels which have distinct biophysical properties (Bocksteins and Snyders, 2012). Consistent with this, KvS mRNAs are expressed in tissue- and cell-specific manners that partially overlap with Kv2.1 and Kv2.2 expression (Castellano et al., 1997; Salinas et al., 1997a; Kramer et al., 1998; Bocksteins and Snyders, 2012; Bocksteins, 2016). Moreover, genetic mutations and gene targeting studies have linked disruptions in the function of KvS-containing channels to epilepsy (Jorge et al., 2011), neuropathic pain sensitivity (Tsantoulas et al., 2018), labor pain (Lee et al., 2020), and retinal cone dystrophy (Wu et al., 2006; Hart et al., 2019; Inamdar et al., 2022), stressing their functional importance in specific cell types. A lack of immunological and pharmacological tools has limited the study of KvS subunits in native tissues, however, and the prevalence, subunit composition, and localization of Kv2/KvS channels in brain are primarily based on mRNA expression.
In this study, we used mass spectrometry-based proteomics analyses to identify several KvS subunits as components of native Kv2 channels in brain, with Kv5.1 being the most abundant. We find that Kv5.1 is a relatively common subunit of brain Kv2 channels and that Kv5.1 protein is differentially expressed in select layers of cortex. Moreover, Kv2/Kv5.1 heteromeric channels localize at ER-PM junctions on neuronal somata and proximal dendrites. This provides direct evidence that neuron-specific expression of KvS subunits creates diversity in Kv2 channels in brain, which likely impacts the role(s) of Kv2 channels in both regulating neuronal excitability and in organizing ER-PM junctions in native brain neurons.
Materials and Methods
Antibodies
All primary antibodies used in this study are listed in Table 1. Anti-Kv5.1 rabbit polyclonal (pAb) antibodies were generated for this study. In brief, rabbits were immunized with a recombinant fragment corresponding to the C-terminal 76 amino acids (419–494 aa) of human Kv5.1 (Uniprot Q9H3M0) produced in E. coli (GenScript). This C-terminal sequence is unrelated to the C-termini of Kv2.1 and Kv2.2 and is also highly divergent between KvS family members. Rabbit antisera were generated at Pocono Rabbit Farm and Laboratory. Anti-Kv5.1 pAbs were affinity purified from antisera by strip purification against nitrocellulose membranes onto which 2 mg of the recombinant Kv5.1 protein had been electrophoretically transferred from an SDS curtain gel (Olmsted, 1981). The anti-Kv5.1C pAb (RRID:AB_3076240) was validated by immunocytochemistry and immunoblotting of HEK cells transfected with recombinant Kv5.1, and its specificity confirmed by a lack of immunolabeling of parallel samples of HEK cells transfected with Kv2.1, Kv2.2, and other KvS family members.
Antibody information
The anti-Kv5.1 mouse mAb L134/44 was generated for this study. In brief, hybridomas producing mAbs were generated using standard methods (Trimmer et al., 1985; Bekele-Arcuri et al., 1996) from BALB/C mice immunized with a bacterially expressed protein fragment containing a C-terminal fragment (424–493 aa) of mouse Kv5.1 (accession number Q7TSH7). The anti-Kv5.1 mouse mAb L134/44 (IgG2b, RRID:AB_3510121) was selected by a multistep screening and validation procedure (Bekele-Arcuri et al., 1996; Gong et al., 2016). Briefly, 3,000 candidates were initially screened by ELISA assays against both the immunogen and fixed heterologous cells expressing full-length mouse Kv5.1 protein. A set of 92 ELISA-positive candidates was then evaluated for specificity on immunoblots against rat brain membrane proteins, by immunofluorescent immunocytochemistry against transiently transfected COS-1 cells expressing full-length mouse Kv5.1, and HRP-DAB immunohistochemistry against adult rat brain sagittal sections. Hybridoma cultures producing L134/44 were subcloned to monoclonality by limiting dilution.
DNA constructs
Plasmids encoding untagged and GFP-, DsRed-, or HA-tagged rat Kv2.1 and Kv2.2 have been previously described (Lim et al., 2000; Bishop et al., 2018; Kirmiz et al., 2018a). A plasmid encoding human Kv5.1 was generously provided by Dr. Dirk Snyders (Stas et al., 2015), and a plasmid encoding human Kv5.1 with a C-terminal GFP tag was generated by GenScript. The Kv5.1BBS construct was generated at GenScript by inserting a bungarotoxin binding site (GGWRYYESSLLPYPDGG) at amino acid 215 in the external S1–S2 loop of Kv5.1, as well as an HA tag at the C terminus. DsRed-Kv2.1BBS was generated at GenScript using an analogous approach with the BBS inserted at amino acid 218 in the S1–S2 loop of rat Kv2.1 (Uniprot A0A0H2UI34). The DsRed-Kv2.1 S586A plasmid has been described previously (Kirmiz et al., 2018a). Plasmids encoding mouse KvS subunits with C-terminal myc-DDK tags were obtained from Origene [Kv6.1 (MR223857); Kv6.4 (MR224440); Kv9.1 (MR218800) and Kv9.2 (MR207651)]. Plasmid encoding rat Kv8.1 with a C-terminal HA tag was generated by Twist Bioscience. Human Navβ2 plasmid was a kind gift from Dr. Alfred George (Lossin et al., 2002).
Animals
All procedures using mice and rats were approved by the University of California, Davis Institutional Animal Care and Use Committee and performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Animals were maintained under standard light/dark cycles and allowed to feed and drink ad libitum. Adult C57BL/6J mice 12–16 weeks old of both sexes were used in immunohistochemistry and proteomic experiments. Kv2.1-KO mice (RRID:IMSR_MGI:3806050) have been described previously (Jacobson et al., 2007; Speca et al., 2014) and were generated from breeding of Kv2.1+/− mice that had been backcrossed on the C57/BL6J background (RRID:IMSR_JAX:000664). All experiments with Kv2.1-KO mice used wild-type (WT) littermates as controls. Kv2.2-KO mice were obtained from Drs. Tracey Hermanstyne and Jeanne Nerbonne and have been described previously (Hermanstyne et al., 2010, 2013). Kv2.1 and Kv2.2 double-KO (Kv2 dKO) mice (Kv2.1−/−/Kv2.2−/−) were generated by breeding Kv2.1+/− mice with Kv2.2−/− mice (Bishop et al., 2018).
Proteomics
Three pairs of WT and Kv2.1 KO mouse littermates were used to prepare DSP-cross-linked mouse brain samples for immunopurifications. Mice were acutely decapitated in the absence of anesthesia and their brains were rapidly removed. Excised brains were homogenized over ice in a Dounce homogenizer containing 5 ml homogenization and cross-linking buffer (in mM): 320 sucrose, 5 NaPO4, pH 7.4, supplemented with 100 NaF, 1 PMSF, protease inhibitors, and 1 DSP (Lomant's reagent, Thermo Fisher Scientific, catalog #22585). Following a 1 h incubation on ice, DSP was quenched with 20 mM Tris, pH 7.4 (JT Baker, catalog #4109-01 (Tris base); and 4103-01 (Tris-HCl)]. Two milliliters of this homogenate were then added to an equal volume of ice-cold 2× radioimmunoprecipitation assay (RIPA) buffer (final concentrations): 1% (vol/vol) TX-100, 0.5% (wt/vol) deoxycholate, 0.1% (wt/vol) SDS, 150 NaCl, 50 Tris, pH 8.0, and incubated for 30 min at 4°C on a tube rotator. Insoluble material was then pelleted by centrifugation at 12,000 × g for 10 min at 4°C. The supernatant was incubated overnight at 4°C with the anti-Kv2.1 rabbit polyclonal antibody KC (Trimmer, 1991) and then incubated with 100 μl of magnetic protein G beads (Thermo Fisher Scientific, catalog #10004D) for 1 h at 4°C on a tube rotator. Beads were then washed six times following capture on a magnet in ice-cold 1× RIPA buffer, followed by four washes in 50 mM ammonium bicarbonate (Sigma, catalog #A6141). Proteins captured on magnetic beads were digested with 1.5 mg/ml trypsin (Promega, catalog #V5111) in 50 mM ammonium bicarbonate overnight at 37°C. Non-cross-linked mouse brain samples were prepared using similar methodology, except that DSP was omitted from the homogenization buffer. Immunoprecipitations were then performed using anti-Kv2.1 (KC) and anti-Kv5.1 antibodies (Kv5.1C), as described above. Proteomic profiling was performed at the University of California, Davis Proteomics Facility. Tryptic peptide fragments were analyzed by LC-MS/MS on a Thermo Scientific Q Exactive Plus Orbitrap Mass spectrometer as described previously (Kirmiz et al., 2018a; Vierra et al., 2019, 2023). The entire Kv2.1 proteomics dataset that was the source of the values shown in Figure 1A has been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD044574 (Vierra et al., 2023).
Kv5.1 co-assembles with Kv2 subunits to form heteromeric channels. A, Mass spectrometry-based proteomics analyses of Kv2.1 complexes immunopurified from cross-linked adult mouse brain samples. The identified proteins are listed by their mean spectral abundance, expressed as a percentage of Kv2.1 spectral counts (mean ± SEM, n = 3). Several KvS subunits copurified with Kv2 channels, with Kv5.1 being the most abundant. B, HEK cells expressing Kv2.1 plus Kv5.1 or other myc-tagged KvS subunits were solubilized, size fractionated on SDS gels and immunoblotted for Kv5.1 with polyclonal antibody 5.1C (top) or monoclonal antibody L134/44 (middle) and for myc and Kv2.1 (bottom). Both Kv5.1 antibodies detected Kv5.1 but no other KvS subunit proteins. Numbers to the left are molecular weights standards in kD. C, Kv5.1 was immunoprecipitated from rat brain lysates with 5.1C antibody and immunoblotted with polyclonal antibody 5.1C (left) or monoclonal antibody L134/44 (right). Both antibodies recognized only the characteristic Kv5.1 doublet bands at ∼52kD. D, HEK cells expressing GFP-tagged Kv5.1 and Kv2.1 or Kv2.2, as designated in the lane labels (c, untransfected cells), were solubilized in RIPA buffer and immunoprecipitations (IPs) performed using affinity-purified Kv5.1C antibody. The IP reactions were size fractionated on SDS gels and immunoblotted for GFP. Kv2.1 and Kv2.2 were both co-IPed with Kv5.1, consistent with their co-assembly into heteromeric channels. Arrows to the right denote positions of the target proteins. E, Live HEK cells expressing Kv2.1 and/or Kv5.1 with an extracellular bungarotoxin binding site (BBS) were cell surface labeled with AF647-Btx (blue), and then fixed, permeabilized, and immunolabeled with anti-Kv2.1 (magenta) and Kv5.1 (green) antibodies. Kv5.1BBS was detected on the cell surface with Btx and anti-Kv5.1 antibody only when coexpressed with Kv2.1. Scale bar, 10 μm. F, HEK cells expressing HA-Kv5.1BBS alone or HA-Kv2.1 + HA-Kv5.1BBS were either labeled live with biotin-Btx (surface) or extracted and then labeled with biotin-Btx (total). The streptavidin precipitation reactions were size fractionated on SDS gels and immunoblotted for the HA epitope tag. Kv5.1 was detected on the cell surface only when coexpressed with Kv2.1. DsRed-HA-Kv2.1BBS is shown as a positive control for surface expression. Arrows to the right denote positions of the IPed proteins.
Immunoprecipitations and immunoblotting
For immunoblot analyses of Kv2 and Kv5.1 channels in brain, mouse brains were removed rapidly and homogenized in ice-cold homogenization buffer (described above). Samples of homogenate were extracted by adding an equal volume of 2× RIPA buffer for 15 min, and insoluble material was pelleted by centrifugation at 12,000 × g for 10 min at 4°C. For IPs, antibodies were added to the supernatant and incubated for 2–3 h at 4°C on a rotator or rocking platform, and then magnetic protein G beads (Thermo Fisher Scientific, catalog #10004D) were added for an additional 1 h. After washing three times in 1× RIPA buffer, the IPed proteins were eluted in 2× SDS LB at 70°C for 5 min. Where appropriate, samples of the lysate were taken before and after each IP reaction to monitor its efficiency. All samples were separated on 8% SDS-PAGE gels and transferred to PVDF membrane. Immunoblots were blocked in Li-Cor Intercept PBS blocking buffer (Li-Cor Biosciences) and probed for Kv2.1 (K89/34 mAb or KC pAb), Kv2.2 (N372B/60 mAb or Kv2.2C pAb), and Kv5.1 (Kv5.1C pAb). Grp75/Mortalin (mAb N52A/42) was used as a loading control. Dye-conjugated fluorescent secondary antibodies (Li-Cor Biosciences) were used to detect bound primary antibodies and imaged using an Odyssey DLx scanner (Li-Cor Biosciences). A dye-conjugated light chain-specific anti-rabbit secondary antibody (Jackson Labs) was used to detect bound Kv5.1 antibody, to avoid recognition of rabbit antibody heavy chain, whose molecular weight is similar to Kv5.1. Immunoblots were analyzed using Image Studio software (Li-Cor Biosciences), and statistical analysis was performed using GraphPad Prism. Immunoblot analyses of Kv2 and Kv5.1 channels expressed in HEK cells was performed in a similar manner.
HEK293T cell culture and transfection
HEK293T cells (ATCC, catalog #CRL-3216) were maintained in Dulbecco's modified Eagle's medium (Invitrogen, catalog #11995065) supplemented with 10% Fetal Clone III (HyClone, catalog #SH30109.03), 1% penicillin/streptomycin, and 1× GlutaMAX (Thermo Fisher Scientific, catalog #35050061) in a humidified incubator at 37°C and 5% CO2. Cells were transiently transfected using Lipofectamine 3000 (Life Technologies, catalog #L3000008) following the manufacturer's protocol, then rescued in fresh growth medium, and used for experiments 36–48 h after transfection. For immunolabeling, cells were plated and transfected on poly-ʟ-lysine (Sigma, catalog #P1524)-coated microscope cover glasses (VWR, catalog #48366-227). For immunoblotting, cells were grown and transfected on 60 mm tissue culture dishes.
CHO-K1 cell culture and transfection
The CHO-K1 stable cell line expressing a tetracycline-inducible rat Kv2.1 construct (Kv2.1-CHO; Trapani and Korn, 2003) was cultured as described previously (Tilley et al., 2014). Cells were transiently transfected using Lipofectamine 3000 for 6 h, rescued in fresh media, and then Kv2.1 expression was induced in Kv2.1-CHO cells with 1 μg/ml minocycline (Enzo Life Sciences, catalog #ALX-380-109-M050), prepared in 70% ethanol at 2 mg/ml. Voltage-clamp recordings were performed 12–24 h later. During recordings, the experimenter was blinded to which cells had been transfected with Kv5.1 or Navβ2.
Electrophysiology
Voltage clamp was achieved with a dPatch amplifier (Sutter Instrument) run by SutterPatch (Sutter Instrument). Solutions for Kv2.1-CHO cell voltage-clamp recordings were as follows: CHO-internal (in mM) 120 K-methylsulfonate, 10 KCl, 10 NaCl, 5 EGTA, 0.5 CaCl2, 10 HEPES, 2.5 MgATP pH adjusted to 7.2 with KOH, 289 mOsm. CHO-external were as follows (in mM): 145 NaCl, 5 KCl, 2 CaCl2, 2 MgCl2, 10 HEPES pH adjusted to 7.3 with NaOH, 298 mOsm. Osmolality was measured with a vapor pressure osmometer (Wescor #5520). The calculated liquid junction potential for the internal and external recording solutions was 9.7 mV and not accounted for. For voltage-clamp recordings, Kv2.1-CHO cells were detached in a PBS-EDTA solution (Invitrogen, catalog #15040-066), spun at 500 × g for 2 min and then resuspended in 50% cell culture media and 50% CHO-external recording solution. Cells were then added to a recording chamber (Warner, catalog #64–0381) and were rinsed with the CHO-external patching solution after adhering to the bottom of the recording chamber. Transfected Kv2.1-CHO cells were identified by GFP fluorescence and were selected for whole-cell voltage clamp. Thin-wall borosilicate glass recording pipettes (Sutter, catalog #BF150-110-10) were pulled with blunt tips, coated with silicone elastomer (Sylgard 184, Dow Corning), heat cured, and tip fire-polished to resistances <3 MΩ. Series resistance of 2–14 MΩ was estimated from the whole-cell parameters circuit. Series resistance compensation between 13 and 90% was used to constrain voltage error to <15 mV, lag was 6 μs. Capacitance and Ohmic leak were subtracted using a P/4 protocol. Output was low-pass filtered at 5 kHz using the amplifier's built-in Bessel and digitized at 25 kHz. Experiments were performed on Kv2.1-CHO cells with membrane resistance >1 GΩ assessed prior to running voltage-clamp protocols while neurons were held at a membrane potential of −80 mV. A stock of 1 mM RY785 (Cayman, catalog #19813) in 10% DMSO which had been stored at −20°C was kept on ice and diluted in room temperature (20–22°C) CHO-external solution just prior to application. Solutions were flushed over the voltage clamped cell at a rate of ∼1 ml/min, and then flow was stopped for recording. Kv2.1-CHO cells were given voltage steps from −80 to 0 mV for 200 ms every 6 s during application of RY785 until currents stabilized. When vehicle control was applied to cells, 0 mV steps were given for a similar duration. DMSO concentration in RY785 and vehicle control was 0.01%. Perfusion lines were cleaned with 70% ethanol and then Milli-Q water.
Conductance values were determined from tail current levels at 0 mV after 200 ms steps to the indicated voltage. Tail currents were mean current amplitude from 1 to 5 ms into the 0 mV step. Conductance–voltage relations were fit with a Boltzmann function:
Immunolabeling of cells
HEK293T cells were fixed for 15 min at 4°C either in 4% formaldehyde prepared fresh from paraformaldehyde in PBS buffer, pH 7.4, or in 2% formaldehyde prepared fresh from paraformaldehyde in Na acetate buffer, pH 6.0. All subsequent steps were performed at RT. Cells were then washed three times for 5 min in PBS, followed by blocking in blotto-T [Tris-buffered saline (10 mM Tris, 150 mM NaCl, pH 7.4) supplemented with 4% (w/v) nonfat milk powder and 0.1% (v/v) Triton X-100 (Roche, catalog #10789704001)] for 1 h. Cells were immunolabeled for 1–2 h with primary antibodies diluted in blotto-T and subsequently washed three times for 5 min in blotto-T. They were then incubated with mouse IgG subclass- and/or species-specific Alexa-conjugated fluorescent secondary antibodies (Invitrogen) diluted in blotto-T for 45 min and washed three times for 5 min in PBS. Cover glasses were mounted on microscope slides with ProLong Gold mounting medium (Thermo Fisher Scientific, catalog #P36930) according to the manufacturer's instructions.
For cell surface immunolabeling of HEK293T cells, live cells were incubated for 30 min in 2 μg/ml (250 nM) AF647-alpha-Btx diluted in normal growth medium. Cells were then washed two times in PBS without Triton X-100, followed by fixation with 2% formaldehyde prepared in PBS for 15 min. Cells were then washed three times for 5 min in PBS and processed for labeling of intracellular proteins as described above.
Immunolabeling of brain sections
Following administration of sodium pentobarbital (Nembutal, 60 mg/kg) to induce deep anesthesia, animals were transcardially perfused with either 4% formaldehyde (freshly prepared from paraformaldehyde) in 0.1 M sodium phosphate buffer, pH 7.4, 2% formaldehyde (freshly prepared from paraformaldehyde) in 0.05 M Na acetate buffer, pH 6.0, or 9% glyoxal/8% acetic acid (Richter et al., 2018; Konno et al., 2023). Sagittal brain sections (30–50 μm thick) were prepared and immunolabeled using free-floating methods as detailed previously (Rhodes et al., 2004; Speca et al., 2014; Bishop et al., 2015; Palacio et al., 2017). Sections were permeabilized and blocked in 0.1 M PB containing 10% goat serum and 0.3% Triton X-100 (vehicle) for 1 h at RT and then incubated overnight at 4°C in primary antibodies (Table 1) diluted in vehicle. All subsequent steps were performed at RT. After washing four times for 5 min in 0.1 M PB, sections were incubated with mouse IgG subclass- and/or species-specific Alexa-conjugated fluorescent secondary antibodies (Invitrogen) and Hoechst 33258 DNA stain diluted in vehicle for 1 h. After washing two times for 5 min in 0.1 M PB followed by a single 5 min wash in 0.05 M PB, sections were mounted and air-dried onto gelatin-coated microscope slides, treated with 0.05% Sudan Black (EM Sciences) in 70% ethanol for 2 min (Schnell et al., 1999). Samples were then washed extensively in water and mounted with ProLong Gold (Thermo Fisher Scientific, catalog #P36930).
RNAscope in situ hybridization
Coronal sections (15 µm) were cut from brains of adult mice which had been perfusion fixed with 4% formaldehyde as described above. Sections were processed for RNA in situ detection using a slightly modified version (Griffith et al., 2019) of the manufacturer's RNAscope protocol (Advanced Cell Diagnostics). Following in situ hybridization with Kcnf1 (Kv5.1) probe (catalog #508731), sections were blocked with 5% normal goat serum in PBS plus 0.1% Triton X-100) and incubated in primary antibodies overnight at 4°C (see Table 1 for details). Sections were washed in PBS, incubated with secondary antibodies for 1 h at RT, washed again in PBS, and mounted with ProLong Gold.
Image acquisition and analysis
Wide-field fluorescence images were acquired with an AxioCam MRm digital camera installed on a Zeiss AxioImager M2 microscope or with an AxioCam HRm digital camera installed on a Zeiss AxioObserver Z1 microscope with a 63×/1.40 NA Plan-Apochromat oil immersion objective and an ApoTome coupled to AxioVision software version 4.8.2.0 (Zeiss). High magnification confocal images of brain sections were acquired using a Zeiss LSM880 confocal laser scanning microscope equipped with an Airyscan detection unit and a Plan-Apochromat 63×/1.40 NA oil immersion DIC M27 objective.
Morphological and colocalization analyses of fluorescent and immunolabeled proteins were performed using Fiji software. For analysis of Kv2.1, Kv2.2, and Kv5.1 colocalization, images were subjected to “rolling ball” background subtraction and an ROI was drawn around the cell body. Pearson's correlation coefficient (PCC) values were then collected using the Coloc2 plugin. Relative clustering of Kv2 and Kv5.1 was measured by quantifying the coefficient of variation (CV) of the fluorescence intensity (fluorescence SD/mean) in each cell, with a greater CV indicative of a higher degree of protein clustering (Cobb et al., 2015). For comparison of WT and Kv2.1 KOs, analysis was performed on images of brain sections taken using the same exposure times. For presentation, images were linearly scaled for min/max intensity in Fiji and saved as RGB TIFFs.
Statistical analysis
Measurements derived from immunoblots and image analysis were imported into GraphPad Prism for statistical analysis and presentation. Reported values are mean ± SEM. We used a one-way ANOVA to compare multiple experimental groups, with post hoc Tukey's or Sidak's multiple-comparisons tests to determine which individual means differed. A one-sample t and Wilcoxon test was used to compare experimental groups to a normalized, control group. For all tests, p values <0.05 were considered significantly different, and exact p values are reported in each figure or figure legend. At least two sets of animals or three independent cultures were used for all experiments; the number of samples analyzed is stated in each figure or figure legend. Electrophysiology statistical tests were performed in Igor Pro software version 8 (WaveMetrics). Independent replicates are individual cells.
Results
KvS subunits are components of Kv2.1 channel complexes in brain
To characterize the composition of Kv2 channels and proteins associated with Kv2-containing ER-PM junctions in mammalian brain, we performed mass spectrometry-based proteomics analyses of immunopurified Kv2.1 channel complexes. The Kv2.1 channels were immunoprecipitated (IPed) from mouse brain homogenates that were chemically cross-linked with DSP during preparation (as in Kirmiz et al., 2018a), which creates intra- and intermolecular links between lysine residues in close spatial proximity (12 angstroms) of one another. Parallel IPs were performed from wild-type (WT) and Kv2.1 knock-out (KO) adult mouse brain samples to identify only those proteins that were isolated specifically with Kv2.1 channels. Protein components of Kv2.1 channel complexes were identified by mass spectrometry in data-dependent acquisition mode as described previously (Kirmiz et al., 2018a; Vierra et al., 2023). Notably, in addition to Kv2.1, Kv2.2, VAPs (Johnson et al., 2018; Kirmiz et al., 2018a), phosphatidylinositol transfer proteins Nir2/3 (Kirmiz et al., 2019), Ca2+ signaling proteins (Vierra et al., 2019), and PKA signaling machinery (Vierra et al., 2023), we identified several members of the electrically silent or KvS subunit family. These are listed in Figure 1A by the total unique mass spectra returned, a method known as spectral counting, and are expressed as a percentage of total Kv2.1 spectra. We found that the spectral abundance of Kv5.1 was ≈18% of the levels of Kv2.1 and ≈50% of the levels of Kv2.2, indicating that this KvS subunit is a relatively common component of these native brain Kv2 channel complexes. Other KvS subunits were less abundant, with their spectral abundance relative to Kv2.1 ranging from ≈5% for Kv8.1 to 2% for Kv9.1. No KvS subunits were detected in parallel Kv2.1 immunoprecipitations (IPs) from Kv2.1 KO brain homogenates. These findings indicate that KvS subunits contribute, to varying degrees, to native Kv2 channel complexes in mouse brain.
Kv5.1 and Kv2 subunits form heteromeric channels in heterologous cells
The simplest interpretation of these proteomics data is that KvS subunits co-assemble with Kv2 subunits to form heteromeric channels, although the possibility of indirect cross-linking of independent KvS subunits to Kv2.1 channels cannot be excluded. To better define the role of KvS subunits in brain, we next investigated the composition and localization of putative Kv2/KvS channels focusing on Kv5.1 as an exemplar due to its abundance in our proteomics analyses. For these experiments we utilized rabbit polyclonal (Kv5.1C) and mouse monoclonal (L134/44) antibodies that we generated against the Kv5.1 cytoplasmic C-terminal tail, whose sequence is highly divergent between different KvS subunits. The antibodies were validated by ELISA and immunofluorescent labeling, immunoblotting, and immunoprecipitation from Kv5.1-transfected heterologous cells. Moreover, we confirmed that the Kv5.1C polyclonal and L134/44 monoclonal antibodies did not cross-react to Kv2.1 and Kv2.2 subunits (Fig. 1) or other KvS subunits by immunoblotting (Fig. 1B) and in the case of the Kv5.1C polyclonal antibody by proteomic analysis of immunoprecipitations from mouse brain (Fig. 2). Finally, the Kv5.1C and L134/44 antibodies did not recognize any bands in addition to Kv5.1 (∼52kD) in immunoprecipitations from rodent brain lysates (Fig. 1C).
Kv5.1 is a common component of native Kv2 channels in brain. A, Rat brain membranes (RBM) were solubilized with RIPA buffer and IPs performed using polyclonal antibodies against Kv2.1, Kv1.2, or Kv5.1 (s, serum; p, affinity-purified). The RBM starting material and the reaction products from IPs performed with the various antibodies as designated in the lane labels were size fractionated on SDS gels and immunoblotted for Kv2.2 (green), and Kv2.1 (red). The panel below shows a section of the blot that was probed for Kv1.2. Both Kv2.1 and Kv2.2 were robustly co-IPed together with Kv5.1, but not with Kv1.2. Labels to the right denote positions of the target proteins. Numbers to the left are molecular weights standards in kD. B, Complementary IPs were performed from mouse brain lysates using Kv2.1 or Kv2.2 subunit antibodies. The input lysate (I), the post-IP depleted lysates (top panels), and the IP fractions (bottom panels) performed with the various antibodies as designated in the lane labels were size fractionated on SDS gels. Immunoblotting of pre- and post-IP lysates (top panel) for Kv2.2 (green) and Kv2.1 (red) confirmed that both IPs were highly efficient. The panel below shows a section of the blot that was probed for GRP75/Mortalin as a loading control. Immunoblotting of the IP fractions (bottom panel) for Kv5.1 shows that Kv5.1 co-IPed together with Kv2.1 and to a lesser extent with Kv2.2, but not with Kv1.2. Labels to the right denote positions of the target proteins. Numbers to the left are molecular weights standards in kD. Asterisk denotes the light chain of the rabbit antibodies used in the IPs. C, Top graph, Quantification of pre- and post-IP brain lysates showed that 87% of Kv2.2 and 79% of Kv2.1 protein were depleted from brain lysates with Kv2.2 and Kv2.1 IPs, respectively (n = 3–5). Bottom graph, Quantification of Kv2.2, Kv2.1 and Kv5.1 subunits in each IP, normalized to the amount IPed by each subunit-specific antibody (n = 4–6, one-way ANOVA and Sidak's multiple-comparisons test, ***p < 0.001, **p < 0.01, *p < 0.05). Approximately 21–22% of Kv2.1 and Kv2.2 were co-IPed with the other Kv2 subunit, and 16% of Kv2.1 and 7% of Kv2.2 were co-IPed together with Kv5.1. Twice as much Kv5.1 was co-IPed together with Kv2.1 as compared with Kv2.2. D, Mass spectrometry-based proteomics analyses of Kv2 channels immunopurified from non-cross-linked brain lysates using Kv2.1- and Kv5.1-specific antibodies. Mean spectral abundance is expressed as a percentage of Kv2.1 spectral counts (mean ± SEM, n = 3). Several KvS subunits were detected in Kv2.1 IPs, with Kv5.1 being the most abundant (18.6% of Kv2.1 levels). In contrast, no other KvS subunits were detected in Kv5.1 IPs.
First, we tested whether Kv5.1 forms heteromeric channels with Kv2.1 and/or Kv2.2 subunits in heterologous cells. HEK293T (HEK) cells expressing Kv5.1 together with Kv2.1 or Kv2.2 were solubilized in RIPA buffer and IPs performed using the Kv5.1C antibody. Notably, we found that both Kv2.1 and Kv2.2 co-IPed together with Kv5.1 from transfected cell lysates (Fig. 1D). To confirm that Kv5.1 co-assembles with Kv2 subunits to form heteromeric channels expressed on the cell surface, we utilized a Kv5.1 construct with a bungarotoxin binding site (BBS) inserted into the extracellular S1–S2 loop. The BBS binds alpha-bungarotoxin (Btx) with high affinity (Sekine-Aizawa and Huganir, 2004) and live cell labeling with Btx thereby allows for specific detection of cell surface-expressed Kv5.1. Indeed, in HEK cells expressing Kv2.1BBS as a positive control, surface Kv2.1 channels were robustly detected with AF647-Btx. In contrast, in HEK cells expressing Kv5.1BBS alone, Kv5.1 was undetectable after surface labeling with AF647-Btx, and immunolabeling with Kv5.1 antibody showed that it was retained intracellularly in an ER-like pattern (Fig. 1E). When Kv5.1BBS was coexpressed with Kv2.1, however, Kv5.1 was readily detected on the cell surface with AF647-Btx, where it colocalized precisely with Kv2.1 (Fig. 1E). Similarly, in biochemical experiments following live cell labeling with biotin-Btx, Kv5.1BBS was isolated with streptavidin beads from cell lysates only when it was coexpressed with Kv2.1 (Fig. 1F). Together, these findings provide strong evidence that Kv5.1 co-assembles with Kv2.1 and Kv2.2 subunits to form heteromeric channels expressed on the cell surface of heterologous cells.
Kv5.1 is a common component of native Kv2 channels in brain
Next, we investigated whether Kv5.1 is a component of native Kv2 channels in brain. Non-cross-linked rat brain membranes or mouse whole brain homogenates were solubilized with RIPA buffer and Kv2/KvS channels were IPed with specific antibodies and their component subunits analyzed by immunoblotting. In IPs performed using Kv5.1-specific antibody (Kv5.1C), we readily detected both Kv2.1 and Kv2.2 subunits by immunoblotting (Fig. 2A). Neither of these subunits were detected in IPs using a control antibody against the axonal Kv1.2 channel (Fig. 2A). Similarly, in reverse IPs using either Kv2 subunit antibody, we found that Kv5.1 was co-immunoprecipitated with Kv2.1 and to a lesser extent with Kv2.2 subunits (Fig. 2B). In contrast, Kv5.1 was not detected in control IPs using anti-Kv1.2 antibodies (Fig. 2B). As all IPs were performed under stringent detergent extraction conditions (i.e., in RIPA buffer), these findings provide compelling evidence that Kv5.1 is an intrinsic component (i.e., subunit) of native Kv2 channels.
To estimate the abundance of Kv2/Kv5.1 channels, we compared the relative amounts of Kv2.1, Kv2.2, and Kv5.1 subunits isolated in each IP reaction. Immunoblotting of the pre- and post-IP lysates showed that both the Kv2.2 and Kv2.1 IPs are highly efficient, depleting 87% of Kv2.2 and 79% of Kv2.1 from the lysate, respectively (Fig. 2C). Consequently, we compared the amount of Kv2.2, Kv2.1, and Kv5.1 subunits in each co-IP to that in each subunit-specific IP. Approximately 21% as much Kv2.2 was isolated in Kv2.1 IPs as compared with Kv2.2 IPs and 22% as much Kv2.1 was isolated in Kv2.2 IPs as compared with Kv2.1 IPs (Fig. 2C). This suggests that ∼20% of Kv2.1 and Kv2.2 subunits in brain occur in Kv2.1/Kv2.2 heteromeric channels. By comparison, we found that 7.4% of Kv2.2 and 16% of Kv2.1 were isolated in Kv5.1 IPs, and conversely, 31% of Kv5.1 was isolated in Kv2.2 IPs and 81% in Kv2.1 IPs. Thus, Kv2.1/Kv5.1 heteromeric channels predominate over Kv2.2/Kv5.1 channels in brain.
In a complementary approach, we analyzed Kv2/KvS channel composition using mass spectrometry-based proteomics. For this, IPs were performed from non-cross-linked mouse brain lysates using Kv2.1 or Kv5.1 antibodies and stringent detergent extraction and wash conditions. The identity and spectral abundance of subunits in the Kv2.1- and Kv5.1-containing channels is shown in Figure 2D, expressed as a percentage of total Kv2.1 spectra. In Kv2.1 IPs, Kv2.2 and Kv5.1 subunit spectra were detected at 25 and 19% of Kv2.1 levels, respectively. Consistent with our experiments on cross-linked brain samples, several other KvS subunits (Kv9.2, Kv6.3, Kv8.1, and Kv9.1) were also identified, although with significantly lower spectral counts. These findings suggest that a substantial subpopulation of native Kv2 channels in brain are heteromeric, containing both Kv2.1 and Kv2.2 subunits and/or Kv2 and KvS subunits. Most notably, the spectral abundance of KvS subunits combined is ∼30% of that of Kv2.1, indicating that KvS subunits are relatively common components of brain Kv2 channels, and create considerable diversity in this class of Kv channel.
In analogous mass spectrometry analysis of Kv5.1 subunit-containing channels isolated in Kv5.1 IPs, Kv2.1 subunit spectral abundance was over twice that of Kv2.2, supporting our findings from standard IP and IB experiments (Fig. 2D). Moreover, we found that the ratio of spectral counts for Kv5.1 compared with those for Kv2.1 plus Kv2.2 subunits combined (39:100 + 41) was ≈1:3.5, although this is likely an underestimate of Kv5.1's abundance due to its smaller molecular mass compared with Kv2.1 and Kv2.2. Indeed, when normalized for either the number of observed unique peptides or expected tryptic peptides per subunit, the ratio of spectral counts of Kv5.1 to Kv2 subunits is ∼1:2.3. Interestingly, we did not detect any other KvS subunits in the Kv5.1 IPs, suggesting that native brain Kv2/Kv5.1 heterotetrameric channels do not commonly contain multiple KvS subunit types. Moreover, we did not detect any other subclasses of potassium channel subunits in Kv5.1 IPs, consistent with specific co-assembly with Kv2 family subunits in brain. Finally, this finding provides additional evidence that our Kv5.1 polyclonal antibody does not cross-react with other KvS subunits in brain.
Kv5.1 protein expression is significantly reduced in Kv2 KOs
As Kv5.1 forms heteromeric channels with both Kv2.1 and Kv2.2 subunits in brain, we tested how Kv5.1 expression is impacted in mice with a genetic deletion of one or both Kv2 subunits. We previously found that protein abundance of the Kv2 channel auxiliary subunit AMIGO-1 is significantly lower in brains from Kv2 KO mice, suggesting that expression of AMIGO-1 in brain depends on its assembly with Kv2 channels (Bishop et al., 2018). Similar to AMIGO-1, we found by immunoblotting of brain lysates and Kv5.1 IPs that Kv5.1 protein levels are reduced by 70% in Kv2.1 KO mouse brains and by 95% in Kv2.1/Kv2.2 DKO mouse brains as compared with brains of wild-type (WT) mice (Fig. 3B). The magnitude of these decreases in Kv5.1 expression approximate its relative association with Kv2.1 and Kv2.2 in WT brain (Fig. 2) and provide further evidence that Kv2.1/Kv5.1 heterotetrameric channels predominate over Kv2.2/5.1 channels. Moreover, we found no indication of any compensatory increase in Kv5.1 association with Kv2.2 in Kv2.1 KO brains. Thus, Kv5.1 protein expression depends on Kv2.1 and Kv2.2 subunits and most likely their co-assembly into heteromeric channels.
Kv5.1 protein is significantly reduced in Kv2 KO brain. A, Immunoblots of brain lysates from wild-type (WT), Kv2.1 KO, and Kv2.1/Kv2.2 DKO mice show that Kv5.1 protein levels are severely reduced in Kv2 KO brain. Labels to the right denote positions of the target proteins. Numbers to the left are molecular weights standards in kD. B, Immunoblots of Kv5.1 IPs from brain lysates of wild-type (WT), Kv2.1 KO, and Kv2.1/Kv2.2 DKO mice show that Kv5.1 protein levels are severely reduced in Kv2.1 KO and Kv2.1/Kv2.2 DKO brain. Asterisk denotes the light chain of the rabbit 5.1C antibody used in the IPs. Graph shows quantitation of IP reaction products. Compared with WT mouse brain, the amount of Kv5.1 IPed is decreased by 70% in Kv2.1 KO and 95% in Kv2.1/Kv2.2 DKO brain samples (one-sample t and Wilcoxon test, n = 4–5, **p = 0.0018, ****p < 0.0001). Thus, Kv5.1 expression is dependent on Kv2 subunits and primarily on Kv2.1. C, HEK cells transfected with Kv5.1 alone, Kv5.1 + Kv2.1, or Kv2.1 alone were solubilized, size fractionated on SDS gels, and immunoblotted with the designated antibodies. The amount of Kv5.1 protein was significantly higher (∼180%) in cells expressing Kv5.1 + Kv2.1 as compared with Kv5.1 alone (one-sample t and Wilcoxon test, n = 6, **p < 0.01). The amount of Kv5.1 plasmid was kept constant for both transfections. D, Kv5.1 and Kv2.1 were IPed from HEK cell lysates transfected as in C, and IBed with antibodies to Kv5.1 (left gel) and ubiquitin (P4D1, right gel). In addition to the major Kv5.1 band at 52kD, high molecular weight bands were detected for Kv5.1 that overlapped with bands for ubiquitin (bracket). Similar laddering was not evident for Kv2.1. The amount of Kv5.1 and ubiquitin laddering were lower in cells expressing Kv5.1 + Kv2.1 as compared with Kv5.1 alone, when normalized to the amount of Kv5.1 IPed (one-sample t and Wilcoxon test, n = 4–5, **p < 0.01).
To test this further, we transfected HEK cells with Kv5.1 alone or together with Kv2.1 and compared Kv5.1 protein amounts in cell lysates by immunoblotting. Notably, we found that Kv5.1 protein amounts were increased by ∼180% when coexpressed with Kv2.1 (Fig. 3C). In addition, in Kv5.1 immunoprecipitations from HEK cell lysates, we observed higher molecular weight forms or “laddering” of Kv5.1, suggestive of ubiquitination. Indeed, immunoblotting with an anti-ubiquitin antibody showed ubiquitin bands that overlapped with Kv5.1 bands (Fig. 3D, bracket) and which were not observed in Kv2.1 immunoprecipitations. Moreover, the amount of high molecular weight Kv5.1 and ubiquitin bands were significantly higher for Kv5.1 expressed alone compared with Kv5.1 coexpressed with Kv2.1. Thus, Kv2.1 subunits enhance Kv5.1 protein expression and decrease Kv5.1 ubiquitination, suggesting that their co-assembly increases the stability of Kv5.1 protein.
Kv5.1 subunit is highly expressed in rodent cortex
Gene expression studies have shown the KvS family subunit mRNAs are differentially expressed in rodent brain, but their protein expression and localization has remained unknown. Consequently, we defined the regional and cellular patterns of Kv5.1 protein expression in brain by multiplex IF labeling using polyclonal (5.1C) and monoclonal (L134/44) antibodies that we generated against Kv5.1 protein. Immunofluorescent labeling of brain sections using the Kv5.1C polyclonal antibody required specialized fixation conditions in which animals were perfusion fixed with 2% formaldehyde in pH 6 buffer (Berod et al., 1981), whereas labeling using the L134/44 monoclonal antibody was detected using standard fixation conditions (4% formaldehyde, pH 7.4) but enhanced using 2% formaldehyde in pH 6 buffer or 9% glyoxal/8% acetic acid (GAA; Richter et al., 2018; Konno et al., 2023). Each antibody was carefully validated (see Materials and Methods and Fig. 1), and similar patterns of Kv5.1 immunolabeling were obtained using both antibodies. Importantly, we found substantial regional differences in Kv5.1 protein immunolabeling in brain sections. Highest levels of expression were observed in neocortex and subiculum (Fig. 4A), where Kv5.1 immunolabeling was detected in a large fraction of neurons that immunolabeled for Kv2.1, and their labeling overlapped extensively at the cellular level (Fig. 4B). Kv5.1 immunolabeling was weak (e.g., thalamus) or undetectable in other brain regions (e.g., CA1, striatum, cerebellum), and in the thalamus some diffuse labeling was apparent that did not correspond to cellular Kv2.1 immunolabeling (Fig. 4B). These findings are consistent with in situ hybridization and transcriptomic analysis of Kv5.1 mRNA expression in mouse brain (e.g., Allen Mouse Brain Atlas, mouse.brain-map.org/experiment/show/68798944). Based on these findings, we focused our detailed analysis of Kv5.1 protein expression and localization in the somatosensory region of neocortex.
Kv5.1 protein is highly expressed in cortex. A, A sagittal section of mouse brain perfusion fixed with GAA and immunolabeled for Kv2.1 (magenta) and Kv5.1 (green). Kv5.1 immunolabeling is prominent in the neocortex and subiculum, where it overlaps substantially with Kv2.1 immunolabeling (white indicates magenta + green overlap). Abbreviations: CTX, cortex; Sub, subiculum; HC, hippocampus; ST, striatum; TH, thalamus; MB, midbrain; CB, cerebellum. Scale bar, 1,000 µm. B, Higher magnification images of Kv2.1/Kv5.1 immunofluorescent labeling in different brain regions. In the neocortex, Kv5.1 (green) immunolabeling is detected in a large fraction of neurons that express Kv2.1 (magenta) and their immunolabeling is extensively colocalized at the cellular level (arrow). Robust Kv5.1 immunolabeling that colocalizes with Kv2.1 is also detected in many neurons in subiculum (arrow) but is lacking in adjacent CA1 pyramidal neurons in hippocampus. In most other brain regions, cellular Kv5.1 immunolabeling is weak [e.g., thalamus (arrow)] or undetectable (e.g., striatum). Some diffuse Kv5.1 labeling is also apparent throughout neuropil in thalamus that does not correspond to cellular Kv2.1 labeling (arrowhead). Images were acquired with the same exposure times and were subjected to identical linear adjustments of min/max signals for display purposes. Scale bar, 20 µm.
In mouse somatosensory neocortex, multiplex IF labeling of sagittal and coronal brain sections revealed that Kv5.1 immunolabeling varies substantially across cortical layers (Fig. 5A,B). Kv5.1 immunolabeling was most prominent in superficial (laminae 2/3) and deep (laminae 5/6) layers of the cortex, where robust immunolabeling was detected in a substantial subset of neurons that also immunolabeled for Kv2.1 and/or Kv2.2 (Fig. 5A,C). Many Kv2/Kv5.1-positive neurons appeared to be pyramidal neurons, based on morphological criteria. By comparison, low Kv5.1 immunolabeling was detected in layer 4 cortical neurons.
Kv5.1 protein is differentially expressed across cortical layers. A, Multiplex immunofluorescent labeling of a sagittal section of mouse somatosensory cortex. Kv5.1 (green) immunolabeling varies across cortical layers and is most prominent in layers 2/3, 5 and 6, where it is detected in a significant subpopulation of neurons that express Kv2.1 (blue) and/or Kv2.2 (magenta). Kv5.1 immunolabeling is low in L4. Scale bar, 50 µm. B, Line scan (of 100 µm band) shows the relative immunolabeling intensity for Kv2.1, Kv2.2, and Kv5.1 across cortical layers. C, Higher magnification images showing that Kv5.1 immunolabeling is restricted to neurons expressing Kv2.1 and/or Kv2.2, depending on the cortical layer. Scale bar, 10 µm.
To further substantiate the differential expression of Kv5.1 in neocortex, we performed RNAscope experiments to localize Kv5.1 transcript at the cellular level. Notably, we found that labeling for Kv5.1 transcript also varied across different cortical layers and was strongest in layers 2/3, 5, and 6 and weak in layer 4 (Fig. 6A). In all layers, Kv5.1 mRNA labeling was almost exclusively restricted to neurons that immunolabeled for Kv2.1, and Kv5.1 mRNA labeling intensity varied between neurons over a broad range (Fig. 6B). L2/3 neurons exhibited the highest Kv5.1 mRNA expression, both in terms of the proportion of neurons with Kv5.1 mRNA labeling and the mean intensity of their labeling. In contrast, only a small proportion of L4 neurons exhibited Kv5.1 mRNA labeling and the mean intensity of their labeling was low (Fig. 6B).
Kv5.1 transcript is differentially expressed across cortical layers. A, Coronal cryostat sections of mouse cortex were hybridized using RNAscope with probes targeting Kcnf1 (Kv5.1) and then immunolabeled for Kv2.1. Prominent labeling for Kv5.1 transcript was evident in cortical layers 2/3, 5, and 6 and was confined to neurons that immunolabeled for Kv2.1 protein. Scale bar, 50 µm. B, Mean intensity of Kcnf1 RNAscope probe labeling in Kv2.1-positive neurons in each cortical layer (n = 30–43 neurons per layer). In layers 2/3, 5, and 6, a large fraction of Kv2.1-positive neurons exhibited Kcnf1 probe labeling and the intensity of probe labeling varied between neurons across a broad range. The dashed line (at 2× mean background labeling) represents the approximate division between Kv5.1-positive and Kv5.1-negative neurons.
Taken together, we find that Kv5.1 transcript and protein are expressed at varying levels and in varying proportions of Kv2 positive neurons in each cortical layer. Kv5.1 is most enriched in layers 2/3, where it is expressed in most neurons.
Kv5.1 is colocalized with Kv2 subunits in a subset of cortical neurons
Focusing on L2/3 of the cortex, a more detailed analysis revealed several important features regarding Kv5.1 protein expression. First, using both polyclonal and monoclonal Kv5.1 antibodies, we found that Kv5.1 immunolabeling is extensively colocalized at the cellular level with both Kv2.1 and Kv2.2 and that like Kv2 subunits it is prominently expressed on the neuronal somata and proximal dendrites (Fig. 7A). As Kv5.1 subunits do not contain the VAP-interacting PRC motif required for somatodendritic targeting and clustering (Lim et al., 2000), this immunolabeling pattern is consistent with Kv5.1 incorporation into heteromeric channels containing one or more obligate PRC domain containing Kv2 α subunits. This is reflected in similar Pearson correlation coefficient (PCC) values for Kv2.1 versus Kv5.1 and Kv2.2 versus Kv5.1 immunolabeling, which are only slightly lower than the PCC values for Kv2.1 versus Kv2.2 immunolabeling (Fig. 7B). Second, although Kv5.1 immunolabeling colocalizes with that of Kv2.1 and/or Kv2.2, we find that the mean intensities of Kv2.1, Kv2.2, and Kv5.1 immunolabeling vary between neurons even in the same cortical layer. This is shown qualitatively by the differing hues of neighboring L2/3 neurons immunolabeled for Kv2.1, Kv2.2, and Kv5.1 (Fig. 7A) and quantitatively by the variable intensity of Kv2.1, Kv2.2, and Kv5.1 immunolabeling between individual neurons (Fig. 6C). On average, however, the highest Kv5.1 immunolabeling occurred in neurons with high Kv2.1 and/or Kv2.2 immunolabeling, consistent with its obligate co-assembly with Kv2.1 and/or Kv2.2 subunits into channels (Fig. 7C). Thus, Kv2.1, Kv2.2, and Kv5.1 protein expression vary, implying that Kv2 channel composition is heterogenous in L2/3 neurons. Third, we find that Kv5.1 is coclustered together with Kv2 channels on neuronal soma and proximal dendrites (see below).
Kv5.1 protein is differentially expressed in L2/3 neurons and colocalizes with Kv2 subunits. A, Confocal image of multiplex immunofluorescent labeling of L2/3 neurons shows that Kv5.1 (green) colocalizes extensively with both Kv2.1 (blue) and Kv2.2 (magenta) on somata and proximal dendrites. However, the ratios of Kv2.1, Kv2.2, and Kv5.1 immunolabeling vary widely between neighboring neurons (note varying hues of immunolabeling). Similar Kv5.1 immunolabeling was obtained with 5.1C polyclonal (top) and L134/44 monoclonal (bottom) antibodies. Scale bar, 10 µm. B, In L2/3 neurons, Pearson's correlation coefficient (PCC) values are similar for Kv2.1 versus Kv5.1 and Kv2.2 versus Kv5.1, indicating that Kv5.1 colocalizes with both Kv2 subunits (one-way ANOVA and Tukey's multiple-comparisons test, ****p < 0.0001, **p = 0.0026, n = 33 neurons, 3 WT brains). In contrast, PCC values for Kv2.1 or Kv5.1 versus GAD67-labeled inhibitory terminals on the neuronal soma are much lower, reflecting little colocalization (n = 13 neurons). C, In L2/3 neurons, simple linear regression shows that the mean intensity of Kv5.1 immunolabeling correlates with that of Kv2.1 (**p < 0.001, PCC = 0.41), and to a slightly lesser extent with Kv2.2 (p < 0.01, PCC = 0.33, n = 68 neurons). D, Top row, Combined RNAscope using probes for Kcnf1 (Kv5.1) transcript and Kv2.1 immunolabeling of L2/3 neurons. Kv5.1 transcript is detected at varying levels in most Kv2.1-positive neurons (∼84%). Bottom row, Combined RNAscope using probes for Kcnf1 (Kv5.1) transcript and Kv5.1 (L134/44) immunolabeling of L2/3 neurons. Kv5.1 mRNA labeling and protein immunolabeling are detected in the same neurons. Scale bar, 10 µm.
In complementary RNAscope experiments, we found that a high percentage of Kv2.1-positive neurons in L2/3 express Kv5.1 transcript (∼84%), again at varying levels (Figs. 7D, top row; Fig. 6B). In similar experiments combining RNAscope with Kv5.1 immunolabeling, we consistently observed colabeling for Kv5.1 protein and transcript in individual neurons (Fig. 7D, bottom row). These experiments required milder protease treatment conditions to successfully detect both Kv5.1 protein and transcript, and the lower Kv5.1 mRNA labeling precluded a reliable comparison of Kv5.1 transcript and protein levels in neurons. These findings show, however, that Kv5.1 mRNA and protein are both expressed in a large proportion of L2/3 neurons and that their expression levels vary between neurons.
The results obtained for Kv5.1 expression in L2/3 neurons broadly applies to neurons in other cortical layers. In layers 5 and 6, we found that Kv5.1 also colocalized extensively with Kv2.1 and/ Kv2.2 and that it was coclustered with Kv2 channels on neuronal soma and proximal dendrites (Fig. 5). In addition, Kv5.1 protein levels varied considerably between neurons in both layers 5 and 6 (Fig. 5).
Kv5.1 immunolabeling is significantly reduced in Kv2.1 KO brain sections
Our biochemical analysis of native Kv2/Kv5.1 channels indicates that Kv5.1 is predominantly associated with Kv2.1 subunits and that Kv5.1 protein is reduced by 70% in Kv2.1 KO brain samples (Fig. 3). Consistent with this, we found that Kv5.1 immunolabeling is significantly reduced in cortical neurons in Kv2.1 KO mice, as compared with wild-type controls. In L2/3 neurons, both plasma membrane localization and clustering of Kv5.1 were greatly reduced or often undetectable in Kv2.1 KO brains (Fig. 8A). In neurons where residual Kv5.1 immunolabeling was detected, it co-colocalized with that of Kv2.2 (Fig. 8B). This is shown in intensity profile plots of L2/3 neurons in Kv2.1 KO brains, where Kv5.1 immunolabeling is much reduced but still evident at Kv2.2 clusters on some neuronal soma (Fig. 8B). This is also reflected in the PCC values for Kv2.2 versus Kv5.1, which were decreased significantly in Kv2.1 KO neurons as compared with WT neurons (Fig. 8C,D). Similarly, the mean intensity of Kv5.1 immunolabeling was significantly reduced in L2/3 neurons of Kv2.1 KO compared with WT brains (Fig. 8E). Together with our biochemical analysis (Fig. 3), these findings suggest that Kv5.1 protein expression largely depends on Kv2.1 subunits and presumably their co-assembly into heteromeric channels.
Kv5.1 immunolabeling is significantly reduced in Kv2.1 KO brain. A, Immunolabeling of L2/3 neurons for Kv5.1 (green), Kv2.1 (blue), and Kv2.2 (magenta) in WT (top) and Kv2.1 KO (bottom) mouse brain sections. Immunolabeling shows that the intensity of Kv5.1 immunolabeling associated with the somatic PM is undetectable or severely reduced in L2/3 neurons in Kv2.1 KO brain as compared with wild type (WT). Images were acquired with the same exposure times and were subjected to identical linear adjustments of min/max signals for display purposes. Scale bar, 20 µm. B, The reduced levels of Kv5.1 immunolabeling persisting in certain Kv2.1 KO neurons colocalize with Kv2.2 immunolabeling. Intensity profile plots show that this residual Kv5.1 is coclustered with Kv2.2 on the plasma membrane. Scale bar, 10 µm. C, D, PCC values for Kv2.1, Kv2.2, and Kv5.1 immunolabeling in L2/3 neurons in WT and Kv2.1 KO brain. Using both polyclonal (C) and monoclonal (L134/44; D) Kv5.1 antibodies, the PCCs for Kv2.2 versus Kv5.1 immunolabeling are significantly reduced in Kv2.1 KO brain as compared with WT (one-way ANOVA and Sidak's multiple-comparisons test, ****p < 0.0001, n = 33 WT and 18 KO neurons for C, and 33 WT and 38 KO neurons for D, n = 3 pairs of WT and Kv2.1 KO brains). E, Mean intensity of Kv5.1 immunolabeling in Kv2.2 positive L2/3 neurons in a representative pair of WT and Kv2.1 KO brains. Kv5.1 immunolabeling is significantly reduced in L2/3 neurons in Kv2.1 KO as compared with WT brain (one-way ANOVA and Sidak's multiple-comparisons test, ****p < 0.0001, ***p < 0.001, points represent individual neurons).
Kv2/Kv5.1 channels are clustered at ER-PM junctions
Kv2 channels cluster at ER-PM junctions by interacting with VAP proteins via the phospho-FFAT/PRC motif in their C-terminal domain (Johnson et al., 2018; Kirmiz et al., 2018a; Di Mattia et al., 2020). Although KvS subunits lack a PRC motif, we found that Kv5.1 coclusters with Kv2.1 or Kv2.2 on the soma and proximal dendrites of cortical neurons in brain (Figs. 5C, 9A) at presumptive ER-PM junctions. This identifies Kv5.1 as a novel component of Kv2-containing junctions in many cortical neurons and indicates that clustering does not require PRC motifs in all four channel subunits. To assess the possibility that Kv2/Kv5.1 channels might cluster less efficiently, we compared CV values for Kv2.1, Kv2.2, and Kv5.1 immunolabeling in L2/3 neurons. The coefficient of variation (CV: SD/mean) is used as a measure of nonuniformity of subcellular distribution, with clustered distributions having high CV values and uniform or diffuse signals having low CV values (Bishop et al., 2015; Kirmiz et al., 2018a, b). Using both Kv5.1 polyclonal and monoclonal antibodies, we found no significant difference between the relative clustering of Kv2.1, Kv2.2, and Kv5.1 proteins (Fig. 9C). Moreover, while CV values (i.e., extent of clustering) varied between neurons, the CV values for Kv2.1 and Kv5.1 were highly correlated in individual neurons (Fig. 9D) and were not negatively correlated with Kv5.1 immunolabeling intensity (Fig. 9E), as might be expected if Kv5.1 subunits reduce steady-state Kv2 channel clustering. Together, these findings indicate that Kv2 homomers and Kv2/Kv5.1 heteromers cluster similarly in native brain neurons, at least under basal conditions.
Kv5.1/Kv2 channels are clustered at ER-PM junctions in brain neurons. A, In cortical L2/3 neurons, immunolabeling shows that Kv5.1 (green) is coclustered with Kv2.1 (blue) and Kv2.2 (magenta) on neuronal soma at presumptive ER-PM junctions. This is evident in the nonuniformity and coincidence of the three labels in the intensity profile (B). Scale bar, 2.5 µm. C, Clustering of Kv2.1, Kv2.2 and Kv5.1 was compared in L2/3 neurons by measuring the coefficient of variation of labeling intensity (CV: SD/mean of pixel intensity). CV values were similar for Kv2.1, Kv2.2, and Kv5.1, suggesting that there is no significant difference in the clustering of Kv2- and Kv5.1-containing channels (one-way ANOVA and Tukey's multiple-comparisons test, p > 0.05, n = 59 neurons, 3 WT brains, black and orange points indicate immunolabeling experiments using Kv5.1 polyclonal and monoclonal antibodies, respectively). D, In L2/3 neurons, the CV values of Kv2.1 and Kv5.1 immunolabeling are correlated (Pearson's correlation coefficient r = 0.61, p < 0.001, n = 35 neurons) and unrelated to Kv5.1 immunofluorescent labeling intensity (5.1IF). E, In L2/3 neurons, the CV values for Kv2.1 and Kv2.2 immunolabeling are not correlated with Kv5.1 immunolabeling intensity (r = 0.04 and −0.26, n = 35 neurons). F, Immunolabeling shows that Kv5.1 clusters in L2/3 neurons (magenta) commonly overlap with or are juxtaposed to SPHKAP puncta (green), which localize to stacked ER cisternae at ER-PM junctions. Scale bar, 5 µm.
Kv2 channel clusters on the soma and proximal dendrites of brain neurons have previously been shown to correspond to ER-PM junctions, as identified by immunoelectron microscopy (Du et al., 1998; Mandikian et al., 2014; Bishop et al., 2015; Vierra et al., 2023) and the coclustering of ER-associated proteins including VAPs (Johnson et al., 2018; Kirmiz et al., 2018a), RyRs (Vierra et al., 2019), and SPHKAP (Vierra et al., 2023). Indeed, in L2/3 neurons, we find that Kv5.1 clusters overlap extensively with SPHKAP puncta (Fig. 9F), which localize to stacked ER cisternae at ER-PM junctions (Vierra et al., 2023). Similarly, in L6 neurons, Kv5.1 clusters colocalize with the prominent RyR2 clusters in those neurons (data not shown). Thus, Kv2/Kv5.1 channels cluster at ER-PM junctions where they could modulate Kv2 channel function.
Kv5.1 impacts Kv2 channel phosphorylation
Next, we tested whether Kv5.1 subunits impact the phosphorylation of Kv2 channels, as Kv2.1 phosphorylation has been shown to regulate multiple aspects of channel function including its biophysical properties, clustering, and protein interactions (Misonou et al., 2004; Park et al., 2006; Kirmiz et al., 2018a). For this, we performed a mass spectrometry-based analysis of Kv2.1 subunit phosphorylation in channels IPed from brain lysates using Kv2.1 or Kv5.1 antibodies. Interestingly, we found that the fraction of Kv2.1 spectra that contained phosphorylated serine (S) and/or threonine (T) residues was ∼40% lower in Kv5.1-containing channels present in the Kv5.1 IPs as compared with the Kv2.1 present in IPs performed with anti-Kv2.1 antibodies (Fig. 10A). Reduced phosphorylation was evident at multiple S/T sites in the Kv2.1 C-terminal domain; however, we were unable to assay phosphorylation within the PRC domain itself as those tryptic peptides were not detected in our MS analysis. In addition, we also detected phosphorylation of five S/T residues in the Kv5.1 C-terminal domain, with phosphorylation of S470 and S472 being the most common (Fig. 10B). These results suggest that Kv5.1 subunits have the capacity to modulate multiple aspects of Kv2 subunit and channel function.
Kv5.1 subunits alter Kv2 channel phosphorylation. A, Kv2.1 C-terminal domain phosphorylation was assessed by proteomic analysis of Kv2.1 protein immunopurified from brain samples with Kv2.1 or Kv5.1 antibodies. Left panel, The fraction of Kv2.1 spectra containing phosphorylated S/T residues was lower in Kv5.1-containing channels compared with total Kv2.1 channels (paired t test, *p = 0.03, n = 3). Right panel, Kv2.1 phosphorylation was decreased by ∼40% in Kv2/Kv5.1 heteromeric channels compared with all Kv2.1 channels (one-sample t test, *p = 0.03, n = 3). B, Kv5.1 C-terminal domain phosphorylation was assessed by proteomic analysis of Kv5.1 protein immunopurified from brain samples. Five S/T residues were phosphorylated to varying degrees. Notably, phosphorylation of either S470 or S472 was observed in a large fraction of those spectra (mean ± SEM = 0.82 ± 0.07), but phosphorylation of both sites was never observed.
A potent and selective Kv2 inhibitor is ineffective against Kv5.1-containing channels
The voltage-gated K+ currents from Kv2 channels are selectively blocked by the drug RY785 (Herrington et al., 2011; Marquis and Sack, 2022). We found that transient transfection of Kv5.1 into CHO cells stably transfected with inducible Kv2.1 resulted in RY785-resistant voltage-gated currents not observed in transfection controls (Fig. 11A,B), and this drug resistance generalizes to all families of KvS/Kv2.1 heteromers (Stewart et al., 2024). The RY785-resistant currents of Kv5.1-transfected cells have a conductance–voltage activation relation shifted to more positive voltages (Fig. 11C), and slower deactivation (Fig. 11D,E), similar to conductances reported after coinjection of Kv2.1 and Kv5.1 into Xenopus oocytes (Kramer et al., 1998). This finding that channels containing Kv5.1 subunits can be pharmacologically distinguished from Kv2-only channels suggests that methods of distinguishing KvS/Kv2 currents (Stewart et al., 2024) could reveal unique electrophysiological roles played by the extraordinary variegated prevalence of Kv5.1 in cortical neurons.
Kv5.1/Kv2.1 channels have distinct biophysical properties and pharmacology. A, Exemplar current traces from Kv2.1-CHO cells before (black) and after (red) application of 1 µM RY785. Left panel, Transfection control. Right panel, Kv2.1-CHO cells transfected with Kv5.1. B, Current remaining after application of 1 μM RY785 as in panel A. Black bars represent mean. Each point is current from one cell at the end of a 200 ms, 0 mV voltage step. Currents >100% remaining indicates current increase after addition of RY785. t test ***p < 0.001. C, Voltage dependence of activation normalized to maximum conductance of initial tail currents at 0 mV. Mean ± SEM. Kv2.1/control (black) n = 6 cells, Kv2.1/Kv5.1 (green) n = 7 cells. Lines are Boltzmann fits (Eq. 1) Kv2.1/control: V1/2 = 2.6 ± 1 mV; z = 1.6 ± 0.1 e0, Kv5.1/Kv2.1: V1/2 = 15 ± 2 mV; z = 1.4 ± 0.1 e0. D, Exemplar traces of channel deactivation at −40 mV after a 50 ms step to +20 mV. Traces are normalized to max current during −40 mV step. E, The faster time constant of double exponential fits to channel deactivation. t test p = 0.001.
Discussion
Kv2 subunits co-assemble with KvS subunits to form heterotetrameric channels with distinct biophysical properties, but the prevalence and localization of these channels in brain neurons has remained unclear. Here, we identify several KvS subunits as components of native Kv2 channels in brain and show that Kv5.1 is prominently expressed in subpopulations of cortical neurons. Thus, KvS subunits create diverse Kv2 channel subtypes in brain, which likely tune Kv2 channel function in a neuron-specific manner.
Gene expression studies show that KvS subunit family members have restricted regional and cellular expression patterns in mouse brain, which partially overlap with that of Kv2.1 and Kv2.2 (e.g., Allen Brain Atlas, Mousebrain.org; Bocksteins, 2016). For example, Kv5.1 (KCNF1) mRNA is broadly expressed in the cortex (Drewe et al., 1992), Kv8.1 is expressed in the cortex and hippocampus in a subset of excitatory neurons (Castellano et al., 1997; Allen Brain Transcriptomics), Kv9.2 is expressed in the hippocampus in CA1-3 and dentate gyrus (Salinas et al., 1997b; Allen Brain Atlas), and Kv6.3 is expressed in the hippocampus only in dentate gyrus (Allen Brain Atlas). Consistent with their differing mRNA expression patterns, our mass spectrometry analysis of Kv2 channel protein complexes IPed from mouse brain identified multiple KvS subunits that varied significantly in their relative abundance. In native channels IPed with a Kv2.1 antibody, we found that spectral counts for Kv5.1 were ≈18% of that for Kv2.1, indicating that Kv2.1/Kv5.1 heteromers are somewhat common in brain. Other KvS subunits were detected at lower abundance in immunopurified Kv2.1 complexes (Figs. 1A, 2D). These counts may be an underestimate, as KvS subunits (∼50–55 kD) are approximately half the size of Kv2.1 (∼100 kD), which could result in fewer spectra being detected per KvS molecule compared with Kv2.1. Moreover, when taken in aggregate, the spectral counts for all five KvS subunits combined are ∼30% of that for Kv2.1, suggesting that a substantial subpopulation of native Kv2 channels in brain contain a KvS subunit.
Our proteomic and biochemical analyses reveal several additional key insights into Kv2 and Kv5.1 channel composition in brain. First, our experiments allow for some estimates on the frequency of Kv2.1/Kv2.2 heteromers in brain. Proteomic analysis of Kv2.1-containing channels identified Kv2.2 at ∼25% of the spectral abundance of Kv2.1. Similarly, IP and IB analysis showed that Kv2.1 IPs pull down ∼21% of total Kv2.2 subunit and Kv2.2 IPs pull down ∼22% of total Kv2.1. Thus, we estimate that ∼20% of Kv2.1 and Kv2.2 subunits occur as Kv2.1/Kv2.2 heteromers. Second, immunoblotting of Kv2.1 and Kv2.2 IPs showed that more Kv5.1 is associated with Kv2.1 than Kv2.2 subunits and mass spectrometry analysis of Kv5.1 IPs identified twofold higher spectral counts of Kv2.1 than Kv2.2. Thus, Kv2.1/Kv5.1 heteromers predominate over Kv5.1/Kv2.2 heteromers in brain. Third, while multiple KvS subunits were detected in Kv2.1 IPs, no additional KvS subunits were identified in Kv5.1 IPs. This indicates that Kv2/Kv5.1 channels typically do not contain more than one type of KvS subunit. Potentially, this could be due either to nonoverlapping expression patterns of KvS subunits, molecular constraints that prohibit co-assembly of distinct KvS subunits, or only one KvS subunit being permitted in each tetrameric channel. Fourth, there is currently considerable debate over the stoichiometry of Kv2/KvS channels with evidence from heterologous cell experiments for KvS to Kv2 subunit ratios of either 2:2 (Moller et al., 2020) or 1:3 (Kerschensteiner et al., 2005; Pisupati et al., 2018). In Kv5.1 IPs from brain, we find that the spectral abundance of Kv5.1 to Kv2 subunits combined (Kv2.1 + Kv2.2) is ∼1:3.5 and ∼1:2.3 when normalized for either the number of observed unique peptides or expected tryptic peptides per subunit. These findings support a subunit stoichiometry between 1:3 and 2:2 but given the semiquantitative nature of spectral counting and other caveats, they cannot resolve the predominant stoichiometry of KvS/Kv2 channels in brain.
Additional evidence that Kv5.1 and Kv2 subunits form heteromeric channels is that Kv5.1 protein levels are decreased in Kv2.1 KO mice and even more so in Kv2.1/Kv2.2 DKO mice. Similarly, Kv5.1 immunolabeling was greatly reduced or often undetectable in Kv2.1 KO neurons, and any residual Kv5.1 colocalized with remaining Kv2.2 subunit. Thus, similar to the AMIGO-1 auxiliary subunit (Bishop et al., 2018), Kv5.1 protein expression in brain neurons depends on its obligate Kv2 subunit partners and likely their co-assembly into heteromeric channels.
At the regional and cellular level, our immunolabeling of mouse brain sections for Kv5.1 showed that it is predominantly expressed in neocortex with differential labeling across cortical layers. Kv5.1 immunolabeling is highest in layers 2/3, 5, and 6 where it is present in a large subset of Kv2 positive neurons, and this pattern closely resembles Kv5.1 mRNA expression in cortex (Drewe et al., 1992, Fig. 6). In individual neurons, we found that Kv5.1 immunolabeling colocalizes precisely with Kv2.1 and Kv2.2, providing further evidence that they form heteromeric channels. Interestingly, however, the relative levels of both Kv5.1 mRNA hybridization and protein immunolabeling signals vary widely between cells even in the same cortical layer. The reason for this variability remains unclear, but it could reflect differential Kv5.1 expression in specific neuronal subtypes, or regulated Kv5.1 expression linked to electrical activity or signaling in individual neurons. Together, these findings suggest that the relative proportion of Kv2 homomers and Kv2/Kv5.1 heteromers varies between neurons and that Kv2/Kv5.1 heteromers may constitute a substantial proportion of Kv2 channels in cortical neurons with high Kv5.1 expression. Indeed, we recently found that Kv2/KvS channels constitute >50% of the total Kv2 channel conductance in nociceptor sensory neurons in mouse dorsal root ganglia (Stewart et al., 2024).
The co-assembly of Kv5.1 with Kv2 subunits to form heteromeric channels has the potential to modify multiple aspects of Kv2 channel function. Previous studies have shown that Kv2 channels regulate neuronal excitability in a complex manner that depends on patterns of excitation (Liu and Bean, 2014; Speca et al., 2014; Honigsperger et al., 2017; Romer et al., 2019; Newkirk et al., 2022) and that KvS subunits modify the biophysical properties of Kv2 channels in a subunit-specific manner (Bocksteins, 2016). Consistent with prior studies (Kramer et al., 1998), we find that Kv5.1 shifts the voltage dependence of Kv2 channel activation to more positive membrane potentials and slows channel deactivation. As Kv5.1 also shifts the voltage dependence of channel inactivation to more negative potentials (Kramer et al., 1998), Kv2/Kv5.1 channels may either increase or decrease neuronal excitability depending on the pattern of electrical activity and resultant level of channel inactivation. Interestingly, our mass spectroscopy analysis revealed that Kv2.1/Kv5.1 heteromers had lower levels of Kv2.1 phosphorylation than seen for the overall population of Kv2.1, although the underlying mechanism remains unclear. Potentially, Kv2.1 phosphorylation could be altered by its co-assembly with Kv5.1 or be a secondary effect of Kv2/Kv5.1 channels altering neuronal excitability (Misonou et al., 2004). Finally, we found that Kv2.1/Kv5.1 heteromers are insensitive to RY785, a potent and selective Kv2 channel blocker. Quaternary ammonium compounds, 4-aminopyridine, and other broad-spectrum K+ channel blockers also have different potencies against certain KvS-containing channels as compared with Kv2 channels (Post et al., 1996; Thorneloe and Nelson, 2003; Stas et al., 2015; Gayet-Primo et al., 2018). Thus, Kv2/Kv5.1 channels have several distinct properties and likely make a unique contribution to the canonical, conducting function of Kv2 channels in regulating excitability and firing of brain neurons.
Another key, nonconducting function of Kv2 channels is to organize ER-PM junctions on neuronal somata and proximal dendrites (Johnson et al., 2019; Vierra and Trimmer, 2022). This structural role is mediated by phosphorylation-dependent binding of the Kv2.1 PRC domain to ER-localized VAP proteins (Johnson et al., 2018; Kirmiz et al., 2018a). Although all KvS subunits lack a PRC domain, we found that Kv5.1 coclusters with Kv2.1 and Kv2.2 at presumptive ER-PM junctions in cortical neurons in brain, indicating that Kv2/Kv5.1 channel clustering does not require a PRC domain in all four channel subunits. It remains to be determined, however, whether Kv2/Kv5.1 channels cluster less stably at ER-PM junctions or otherwise impact their composition and function. Indeed, Kv2 channels help assemble a macromolecular complex at ER-PM junctions that includes LTCCs, Ca2+ signaling and scaffolding proteins (Vierra et al., 2019), lipid handling proteins (Kirmiz et al., 2019), and PKA signaling proteins (Vierra et al., 2023), thereby creating a specialized Ca2+, lipid, and PKA signaling microdomain. As many of these interactions are mediated directly or indirectly by the large cytoplasmic C-termini of Kv2 subunits, replacement of Kv2 with Kv5.1 in channel tetramers could impact the channel interactome and composition of the associated ER-PM junctions. Most obviously, Kv2/Kv5.1 channels could potentially interact with fewer of these junctional components, reducing their localization at ER-PM junctions, but it is also possible that Kv5.1 might recruit select proteins though direct binding. Thus, important questions for future studies are whether Kv5.1 and other KvS subunits modulate the ability of Kv2 channels to organize ER-PM junctions, and if they impact Ca2+ and lipid signaling functions at these specialized junctional complexes in brain neurons.
Footnotes
This work was supported by National Institutes of Health research grants R21NS123417 to M.F, R03TR004200 to M.F. and J.T.S, and R01NS114210 and R21NS101648 to J.S.T. We thank members of the Ferns, Sack, and Trimmer laboratories for their support and useful discussions of the data.
The authors declare no competing financial interests.
- Correspondence should be addressed to Michael Ferns at mjferns{at}ucdavis.edu.
