The expression system influences stability, maturation efficiency, and oligomeric properties of the potassium-chloride co-transporter KCC2

Abstract

The neuron-specific K⁺/Cl⁻ co-transporter-2, KCC2, which is critical for brain development, regulates γ-aminobutyric acid-dependent inhibitory neurotransmission. Consistent with its function, mutations in KCC2 are linked to neurodevelopmental disorders, including epilepsy, schizophrenia, and autism. KCC2 possesses 12 transmembrane spans and forms an intertwined dimer. Based on its complex architecture and function, reduced cell surface expression and/or activity have been reported when select disease-associated mutations are present in the gene encoding the protein, SLC12A5. These data suggest that KCC2 might be inherently unstable, as seen for other complex polytopic ion channels, thus making it susceptible to cellular quality control pathways that degrade misfolded proteins. To test these hypotheses, we examined KCC2 stability and/or maturation in five model systems: yeast, HEK293 cells, primary rat neurons, and rat and human brain synaptosomes. Although studies in yeast revealed that KCC2 is selected for endoplasmic reticulum-associated degradation (ERAD), experiments in HEK293 cells supported a more subtle role for ERAD in maintaining steady-state levels of KCC2. Nevertheless, this system allowed for an analysis of KCC2 glycosylation in the ER and Golgi, which serves as a read-out for transport through the secretory pathway. In turn, KCC2 was remarkably stable in primary rat neurons, suggesting that KCC2 folds efficiently in more native systems. Consistent with these data, the mature glycosylated form of KCC2 was abundant in primary rat neurons as well as in rat and human brain. Together, this work details the first insights into the influence that the cellular and membrane environments have on several fundamental KCC2 properties, acknowledges the advantages and disadvantages of each system, and helps set the stage for future experiments to assess KCC2 in a normal or disease setting.

Graphical Abstract

Introduction

The construction and maintenance of neural networks is a critical and particularly vulnerable step during brain development. Notably, the synchronization of inhibitory and excitatory neurotransmission is a delicate process in which ion transporters on the neuronal plasma membrane must effectively regulate ion flux, which, in turn, controls the resting membrane potential and generation of action potentials [1]. For example, γ-aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the adult central nervous system [2]. However, hyperexcitability—due to defects in inhibitory neurotransmission—is observed when GABA signaling is disrupted and has been associated with multiple neurodevelopmental disorders [3, 4]. Given that GABA signaling undergoes a pivotal shift from excitatory neurotransmission in very early brain development to inhibitory neurotransmission, the transfer of ions across the plasma membrane is crucial and must be properly controlled by ion transporters [5, 6]. This transition in GABA signaling from excitatory to inhibitory as the brain develops is essential for proper neuronal development and is determined by the neuron-specific K⁺/Cl⁻ co-transporter 2, KCC2 [7–9].

KCC2 modulates GABA signaling by controlling intracellular chloride ([Cl−]_i) in somatodendritic regions of developing neurons [10–12]. Therefore, it is perhaps unsurprising that approximately a dozen mutations in KCC2 have been identified in patients with neurodevelopmental disorders such as epilepsy, autism spectrum disorder, and schizophrenia [13]. For some of these disease-associated mutations, KCC2 cell surface expression and/or function appears to be compromised [14–19]. Additionally, postmortem analyses of patients with schizophrenia revealed reduced KCC2 expression in the dorsolateral prefrontal cortex (DLPC) and auditory cortex (AC), two regions of the brain that demonstrate severe functional deficits in the setting of disease [20, 21]. Nevertheless, the molecular defects associated with many disease-associated KCC2 mutations remain unknown.

KCC2 is a member of the solute carrier 12 (SLC12) family of electroneutral cation-chloride cotransporters (CCCs), and its expression within the central nervous system is widespread in the brain, including the cortex, brainstem, and hippocampus [8, 22–24]. Like other members of the CCC family, KCC2 is predicted to form a tightly wrapped dimer, as indicated by recent cryo-EM structures [25–27]. Previous reports have explored the biological significance of the KCC2 dimer, with some studies claiming its localization is brain-region specific, and that this form may increase in an age-dependent manner [28, 29]. For example, native gel electrophoresis experiments in mouse cortical lysates show a high KCC2 monomer:dimer ratio in early brain development (E18), and relatively equal amounts of monomer:dimer in late development (P30) [28]. This was not consistent across all regions in the mouse brain, with areas such as the olfactory bulb and cerebellum showing comparable amounts of the monomer and dimer in both early and late development. Another study found a less dramatic difference in KCC2 abundance across brain regions, and no age-dependence was noted [29]. KCC2 has also been expressed exogenously in cell systems such as HEK293 cells and N2a cells [30–33]. In each case, dimeric KCC2 was abundant, and attempts to dissociate the dimer were unsuccessful, even under strong denaturing conditions. Discrepancies between studies may be explained by levels of protein expression (i.e., overexpression versus endogenous levels) and/or differences in methodology that affect detergent sensitivity [34]. Given some of the inconsistencies that have persisted in the field regarding the appearance of dimeric KCC2, an analysis of KCC2 expression across cell systems using comparable methodology is warranted.

In order to ensure that large and complex membrane proteins—such as KCC2—assemble and fold properly, cells harbor quality control checkpoints to maintain protein homeostasis, or “proteostasis” [35, 36]. The first of these checkpoints takes place in the endoplasmic reticulum (ER). Upon entry into the ER, newly synthesized membrane proteins begin to fold into their native conformation and receive post-translational modifications. Most are glycosylated via the acquisition of an ~3 kDa N-linked core glycan on the asparagine in the N-X-S/T consensus site [37]. In the case of KCC2, each monomer has 12 transmembrane domains with intracellular N- and C-termini and six N-linked glycan moieties on a large extracellular loop. Often, core glycosylation is required before a protein is encapsulated into COPII vesicles and transported to the Golgi apparatus. In the Golgi, this ER-derived glycan core is elaborated with more complex glycans [38]. Previous experiments with a KCC2 relative, KCC4, found that mutations in the N-linked glycan-acceptor sequence leads to defects in KCC4 localization to the cell surface [39]. While these mutations have yet to be introduced in the context of KCC2, it is worth noting that the glycosylation of select disease-associated KCC2 mutants is impaired when studied in HEK293 cells [16, 34]. Therefore, a better assessment of KCC2 modification and maturation through the secretory pathway may reveal insights into the pathogenesis of KCC2-associated diseases.

In addition to compromising ER-to-Golgi transport, defects in protein folding or modification in the ER can target aberrant proteins for endoplasmic reticulum-associated degradation (ERAD) [40–42]. During ERAD, substrates are recognized by ER-associated chaperones, ubiquitinated by ER-associated enzymes, “retrotranslocated” or extracted from the ER, and degraded by the 26S proteasome. Although the targeting of KCC2 to ERAD has not been investigated, other CCCs within the same family as KCC2, such as the Na⁺/Cl⁻ co-transporter (NCC) and the Na⁺/K⁺/Cl⁻ co-transporter 2 (NKCC2), are ERAD substrates, even if they lack mutations [43–45]. These data are consistent with the fact that most wild-type membrane proteins are targeted to some extent for ERAD, at least in cell-based models, due to their complex folding pathways. Other studies established factors that influence KCC2 steady-state expression and turnover, such as calpain, which truncates the KCC2 C-terminus and triggers degradation in response to calcium [46, 47], and amyloid precursor protein (APP), which limits KCC2 ubiquitination and prevents degradation [48]. Whether KCC2 is targeted for ERAD has yet to be determined. Given the complexity of KCC2 assembly, regulation, and post-translational modification, it is anticipated that known—and likely many other unknown—mutations in KCC2 that lead to disease compromise protein folding and/or modification, thus targeting the protein for ERAD.

As proteins are synthesized, modified, and assembled, they exist in a dynamic state that is dependent upon cellular homeostasis, which can be altered in disease states or if the system in which they are studied is non-native [49, 50]. To begin to examine whether KCC2 biogenesis may be similarly affected—and more specifically whether the transporter is an ERAD substrate—we conducted an analysis of KCC2 stability, maturation, and oligomeric properties in a variety of systems. Although KCC2 was stable and mature in primary rodent neurons, studies in a new yeast expression system and in a HEK293 cell model allowed us to observe KCC2 maturation intermediates and establish KCC2 as an ERAD substrate. We also examined select properties of KCC2 in human brain lysates to further highlight the advantages of each system and the ways in which they can be used to study KCC2.

Materials and Methods

Yeast strains and transformation

To express KCC2 in the baker’s yeast, S. cerevisiae, we developed a plasmid containing the human KCC2a coding sequence preceded by an N-terminal HA tag and under the control of a promoter for constitutive expression in yeast (Figure 1A). To this end, the KCC2a coding sequence was excised from pcDNA3.1-KCC2a (obtained from the MRC Protein Phosphorylation and Ubiquitylation Unit, University of Dundee, Scotland, UK) with the XbaI and SalI restriction enzymes, and ligated the insert into the leucine-selectable plasmid, pRS415TEF1 [51]. DNA sequence was confirmed by Plasmidsaurus (Eugene, Oregon). To create a yeast expression vector containing the KCC2b isoform, the KCC2b coding sequence and HA tag were excised from the mammalian KCC2-HA-mCherry expression plasmid (see below). The insert was cloned into the yeast pRS415TEF vector using the Gibson Assembly kit and protocol (New England Biolabs, Ipswich, MA, USA). The DNA sequence was confirmed by Plasmidsaurus (Eugene, Oregon).

Figure 1. KCC2 is an ERAD substrate in S. cerevisiae.

(A) Plasmid map of the KCC2 expression vector in S. cerevisiae. Created with Biorender.com. (B) Yeast lysates were treated with endoglycosidase H (endoH) where indicated. Anti-HA antibody was used to detect KCC2 protein in western blot and the filled triangles represent glycosylated (upper) and unglycosylated (lower) KCC2. Detection of G6PD provided a loading control. (C) Cycloheximide (CHX) chase analysis of KCC2 in S. cerevisiae. Cells were treated with MG132 or vehicle (DMSO) (left). Graph of quantitative immunoblot analysis showing total levels of KCC2 (right). N=6. (D) Cycloheximide chase analysis of KCC2 in S. cerevisiae containing or lacking E3 ligases Hrd1 and Doa10 (hrd1Δdoa10Δ). Graph of quantitative immunoblot analysis showing total levels of KCC2 (right). N=5. (Please note: the 90 min. timepoint was analyzed with a Mann-Whitney statistical test as described in the Methods). (E) Cycloheximide chase analysis of KCC2 in S. cerevisiae containing or lacking the Pep4 vacuolar protease (pep4Δ). Graph of quantitative immunoblot analysis showing total levels of KCC2 (right). N=6. All western blots were normalized to G6PD loading control to ensure consistent loading. For all experiments, data is the mean ± SEM. N= number of individual clones from at least two independent transformations.

** = p < 0.01, *** = p < 0.001

Plasmid transformation into yeast was conducted using the lithium acetate procedure [52]. In brief, yeast cells were grown at 30°C in complete growth medium to mid-log phase. After isolation by centrifugation, the cell pellet was resuspended in TE/LiAC, and combined with plasmid DNA, salmon sperm DNA, polyethylene glycol, and TE/LiAC. Cells were rotated at 37°C for 2 hr and then heat shocked at 42°C for 15 min. Centrifugation was followed by resuspension of the cell pellet in selection media, and the resuspended solution was plated onto synthetic complete media lacking leucine and incubated at 37°C for 2 days to isolate transformants. Supplemental Table 1 lists the yeast strains used in this study [53, 54]. No ethics approval number was required for yeast experiments.

HEK293 cell culture

HEK293H cells were grown in 6-well poly-L-lysine coated plates to a density of ~600,000 cells/mL in DMEM containing 10% FBS and at 37°C in 5% CO₂. Transfection with plasmids engineered to express untagged human KCC2b (originally obtained from Dr. Eric Delpire, Vanderbilt University) and HA- and mCherry-tagged rat KCC2b in pcDNA3.1(+) were previously described [15], and sequences were confirmed by Plasmidsaurus (Eugene, Oregon). Transfection was conducted with cells from passage 1–2 at 70–90% confluency using Lipofectamine 2000 (Invitrogen) in Gibco Opti-MEM media, which were first incubated together at room temperature for 5 min prior to the addition of 4 μg plasmid DNA. The media was then aspirated and replaced with fresh culture medium and plates were returned to 37°C. All assays and analyses were performed 24 hr post-transfection. Per the manufacturer (Thermo Fisher), the HEK293H cell line was cloned from the original HEK293 cell line. HEK293H cells contain sheared human adenovirus type 5 DNA and express the E1A adenovirus gene. The HEK293H cell line is not listed as a commonly misidentified cell line by the International Cell Line Authentication Committee, and the lab stocks of HEK293H cells were last authenticated by the University of Arizona Genetics Core in December, 2016. No ethics approval number was required for HEK293H cell experiments. Hereafter, HEK293H cells will be referred to as “HEK293 cells” for simplicity.

Cerebrocortical cultures

All animal procedures were approved by the IACUC of the University of Pittsburgh (Protocol # 21039053). Primary cortical cultures were prepared from embryonic day 17 (E17) Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA, USA) as previously described [55]. Briefly, pregnant rats were housed alone in a standard cage in the University central animal facility for a maximum of three days with free access to food and water and were sacrificed humanely and painlessly by CO₂ inhalation and further exsanguinated via severing of the jugular vein to assure completion of euthanasia. The animals showed no sign of distress during this procedure and perished quickly and painlessly. Embryos of either sex were removed and quickly decapitated. Six-8 embryonic cortices from a single litter were pooled and dissociated with trypsin and cells were plated on 12 mm, poly-L-ornithine (PLO) coated glass coverslips in six well plates at a density of 680,000 cells per well. We consider a single dissociation, from one pregnant animal, a single biological replicate. Non-neuronal proliferation was inhibited after 14 days in vitro (DIV) with 2 μM cytosine arabinoside. Cultures were utilized at 1–4 weeks in vitro. We calculate that approximately 15 pregnant rats (approximately 90–120 embryos) were utilized for these studies. This is both the initial and the total number of animals used since none were excluded and no exclusion criteria was predetermined. No sample size calculation was performed and sample size was based on previous related work [56–58].

Rat brain synaptosome preparation

Heterogenous Sprague Dawley/Wistar adult rats of both sexes were sacrificed by euthanasia (see above) and brains were excised, snap frozen in liquid nitrogen, and stored at −80°C. To prepare synaptosomes, as published [59], rat brain matter was homogenized in 10 vol of cold homogenization buffer (320 mM sucrose, 5 mM Tris-Cl, pH 7.4) using a Teflon-glass homogenizer at 4°C. Next, the homogenate was centrifuged at 1,200g for 5 min, the supernatant was removed and centrifuged at 23,000g for 20 min, and the resulting pellet was resuspended in cold homogenization buffer and applied to a discontinuous 1.2, 1.0, and 0.8 M sucrose gradient. The mixture was centrifuged at 83,000g for 1 hr and the synaptosomes, at the 1.2 M – 1.0 M interface, were removed using a glass pipette. Synaptosomes in this interface were diluted 1:1 in 200 mM Tris-Cl, pH 7.4, isolated by centrifugation, and stored at −80°C until use.

Human brain sample collection and dissection

In accordance with University of Pittsburgh’s Committee for Oversight in Research Involving Decedents and the International Review Board for Biomedical Research, postmortem brain tissue was obtained during autopsies conducted at the Allegheny County Medical Examiner’s Office in Pittsburgh, PA. Informed consent from next of kin was obtained, and an independent committee of experienced clinicians obtained consensus diagnoses [60] using data obtained from medical records, structured interviews with relatives and, if available, toxicology reports. No patients from whom samples were used received a diagnosis of a neurocognitive or psychiatric disorder, and no subject included was deemed to have sufficient evidence for a neuropathologic diagnosis of a neurodegenerative disease. The precuneus from the right hemibrain was identified using the marginal branch of the cingulate sulcus as the anterior boarder, the subparietal sulcus as the inferior boarder, and the parietooccipital sulcus as the posterior boarder. This area was then isolated by dissection and frozen at −80°C.

Human brain homogenate and synaptosome preparation

As previously reported [21], gray matter was isolated by microdissection, weighed, and homogenized with a Dounce homogenizer on ice in 1 mL of Syn-Per reagent (Thermo Scientific) with protease and phosphatase inhibitors. A total of 100 μL of this homogenate was added to an equal volume of 2X S-trap buffer (100 mM triethylammonium bicarbonate (TEAB) and 10% sodium dodecyl sulphate (SDS)), to generate the total homogenate fraction, which was then stored at −80°C. To obtain synaptosomes, the remaining homogenate was centrifuged at 1,200g at 4°C for 10 min, and the supernatant from this step was centrifuged at 15,000g for 20 min. The pellet was resuspended in 0.1 mM CaCl₂ in the presence of protease and phosphatase inhibitors and a final centrifugation at 15,000g for 20 min at 4°C was used to obtain a pellet fraction, which was resuspended in 1X S-trap buffer to generate an enriched synaptosome fraction.

Western blot analysis

Yeast cells were grown to mid-log phase in selection media at 30°C. Yeast samples were lysed in 0.3 M NaOH, 1% 2-mercaptoethanol (BME), and protease inhibitors (1 μg/mL leupeptin, 0.5 μg/mL Pepstatin A, 1 mM phenylmethylsulfonyl fluoride (PMSF)). The lysate was agitated on a Vortex mixer, and incubated on ice for 10 min. To each sample, tricholoracetic acid (TCA) to a final concentration of 10% was added, followed by incubation on ice for 20 min. After centrifugation, cell pellets were diluted and homogenized in TCA sample buffer (80 mM Tris-HCl pH 8, 8mM EDTA, 3.5% SDS, 10% glycerol, 1% BME, 0.5 % Tris pH 9.4, 0.16% Tris base, 0.01% bromophenol blue) and loaded onto either a 10% polyacrylamide gel or a 7.5% gel (for endoH experiments; see below) for SDS-PAGE. After transfer to nitrocellulose, KCC2 was detected with 1:10000 HRP-conjugated rat monoclonal anti-HA antibody (Roche Applied Science) and washed with the Prosignal PICO kit (Prometheus). For a loading control, anti-glucose-6-phosphate dehydrogenase (G6PD) was detected with rabbit anti-G6PD antibody (Sigma-Aldrich) and incubated with anti-rabbit antibody (Cell Signaling Technologies) and treated as noted above. The signals corresponding to KCC2 and G6PD were detected on a BioRAD Universal Hood II Imager. A Pageruler Plus Prestained Protein ladder (10 kDa to 250 kDa) was used to assess protein molecular weight in all analyses.

Transfected HEK293 cells were lysed in TNT buffer (50 mM Tris, pH 7.2, 150 mM NaCl, 1% Triton X-100) and a protease inhibitor cocktail (PIC) tablet (Roche Diagnostics) followed by centrifugation. All cleared lysates were diluted in SDS sample buffer (1% BME, 2% SDS, 0.05 mg/mL bromophenol blue, 65 mM Tris, pH 6.8), heated at 37°C for 10 min, and analyzed by 5% SDS-PAGE. Primary rat neurons were instead lysed with Cell Extraction Buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 20 mM Na₄P₂O₇, 2 mM Na₃VO₄, 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate, 1 mM PMSF; Life Technologies, FNN0011) supplemented with a PIC tablet. Cell Extraction Buffer has been routinely used in our lab with neurons (see e.g., [61]), as it favors isolation of the membrane fraction and cytosol and excludes nuclei. After centrifugation to remove soluble debris, samples were diluted in SDS sample buffer and heated at 37°C for 10 min. Analysis by 5% SDS-PAGE was used for all primary neuron protein experiments, with the exception of the DIV time course which used 4–20% Tris-glycine gradient gels (Invitrogen). In both HEK293 and primary neuron lysates, KCC2 was detected with 1:1000 rabbit anti-KCC2 polyclonal primary antibody (Sigma-Aldrich) followed by washing and then treatment with 1:5000 goat anti-rabbit IRDye-labeled secondary antibody (LI-COR). For a loading control, REVERT Total Protein Stain (LI-COR) was used. Signals were detected using a LI-COR Odyssey CLX imaging system.

To assess KCC2 in rat brain synaptosomes, the final brain synaptosome pellets (see above) were immediately lysed with TNT buffer and a PIC tablet (Roche Diagnostics). Samples were briefly centrifuged, and the supernatant was diluted in SDS sample buffer, followed by incubation at 37°C for 10 min and analysis by 5% SDS-PAGE. Human brain synaptosomes were also lysed and processed using these methods for the deglycosylation experiments (see below). For the comparative analysis of the 10 human samples, synaptosome pellets were lysed in PBS with 5% SDS. Samples were then diluted in SDS sample buffer, incubated at 37°C for 10 min, and analyzed by 5% SDS-PAGE.

Overall, for these samples as well as each of the lysates prepared above, the choice of solubilization conditions as well as methods for gel electrophoresis were based on the examined protein and/or the system in which that protein was expressed [44, 62–64]

Measurements of protein glycosylation

After yeast cells were grown, lysed, and diluted in TCA sample buffer as above, they were treated with endoglycosidase H (endoH; Sigma-Aldrich) or the equivalent volume of water in 0.1 M sodium citrate, pH 5. All tubes were incubated at 37°C for 2 hr and loaded onto 7.5% polyacrylamide gels for SDS-PAGE and western blot analysis, as above.

HEK293 and neuronal cell lysates were grown and lysed as stated above. The lysates were next combined with NEB glycoprotein denaturing buffer (0.5% SDS, 40 mM DTT), and the samples were incubated at 37°C for 15 min. EndoH treated samples were processes as above, and PNGaseF treated samples contained lysate, glycobuffer 2, NP-40, and PNGaseF (NEB) or an equivalent volume of water. All reactions were incubated at 37°C for 1 hr, after which SDS sample buffer, DTT to a final concentration of 150 mM, and Tris, pH 9.4, to a final concentration of 2 mM were added to each tube. Samples were incubated at 37°C for 10 min and, after a brief agitation, analyzed by 5% SDS-PAGE.

Measurements of protein using different lysis conditions

To detect KCC2 using different lysis conditions in HEK293 cells, cells were transfected with the untagged human KCC2b expression plasmid, as described above. Cells were lysed directly into SDS sample buffer (2% SDS, 0.05 mg/mL bromophenol blue, 65 mM Tris, pH 6.8). Sample buffer contained a PIC and either lacked reductant or contained final concentrations of 1% BME, 100 mM DTT, or 5 mM tris(2-carboxyethyl)phosphine (TCEP) [65]. Since TCEP is acidic in solution, samples containing TCEP were neutralized with 1 M Tris, pH 9.4, where indicated. In all cases, the treated lysate was passed through a 0.7mm × 40mm sterile needle to shear DNA. Samples were then incubated for 10 min at room temperature (RT), 10 min at 37°C, 5 min at 75°C, or 5 min at 100°C prior to analysis by 5% SDS-PAGE.

Alternatively, cells were lysed in TNT buffer (50 mM Tris, pH 7.2, 150 mM NaCl, 1% Triton X-100) containing a PIC and either no reductant or at a final concentration again contained 1% BME, 100 mM DTT, or 5 mM TCEP. Samples containing TCEP were not neutralized for this analysis. After centrifugation, SDS sample buffer containing each of the above reductants was added to the cleared lysate. Samples were incubated at RT for 10 min, 37°C for 10 min, 75°C for 5 min, or 100°C for 5 min. All samples were analyzed by 5% SDS-PAGE as described above.

Measurements of protein stability

Cycloheximide chase analyses in yeast and HEK293 cells were performed as described previously [43]. In brief, yeast cells were grown at 30°C in selection media to mid-log phase, and equal numbers of cells were used for all time courses. For experiments requiring drug treatments, yeast lacking Pdr5 (pdr5Δ) were incubated in MG132 to a final concentration of 25 μM or in an equal volume of DMSO. Aliquots were taken at the indicated time points and flash frozen in liquid nitrogen. Cycloheximide at a final concentration of 50 μg/mL was added to each sample immediately after the 0-min timepoint and collected. Samples were stored at −80°C until lysed and processed for western blot analysis as indicated above.

HEK293 cells and primary rat neurons were pre-incubated at 37°C for 1 hr with MG132 or leupeptin at final concentrations of 25 μM and 50 μM, respectively. After pre-incubation, cycloheximide (50 μg/mL) was added, and cells were incubated for the indicated times. Due to its lability, cycloheximide was spiked in at each timepoint to ensure that protein synthesis remained inhibited for the duration of the timecourse.

Live cell imaging

HEK293 cells were seeded in 35 mm glass bottom Mat Tek dishes and transfected with the mammalian HA-mCherry-KCC2 plasmid, as described above. At 24 hr post-transfection, cells were incubated with 15 μg/mL DAPI (Thermo Scientific) for 20 min and 15 μg/mL Alexa Fluor 647 conjugated Concanavalin A dye (Thermo Scientific) for 5 min in 2 mL culture medium at 37°C in 5% CO₂. Medium-containing dye was aspirated, and cells were washed and incubated in DMEM supplemented with 10% FBS that lacked phenol red (Sigma-Aldrich) to reduce background fluorescence. The cells were then immediately taken for live cell imaging and visualized on a Nikon A1R Confocal microscope equipped with a live cell incubator using a 100X oil immersion lens. Images were taken using the NIS Elements software and image analysis was conducted using ImageJ. All images were taken at the University of Pittsburgh Microscopy and Imaging Suite (RRID# SCR_022084).

RT-qPCR analysis

Total cellular RNA was extracted from primary rat neurons using the RNeasy mini kit (Qiagen) at each time point. Extraction included a DNaseI digestion step per protocol. The qScript cDNA Superscript Mix (Quantabio) was used for subsequent cDNA synthesis and conducted under the following PCR conditions: 5 min at 25°C, 30 min at 42°C, and 5 min at 85°C, and then storage at 4°C until further analysis. Each real-time (RT) qPCR reaction contained 80 ng cDNA, forward and reverse primer mix, and SybrGreen I Mastermix (ThermoFisher Scientific). The reaction was conducted using the Applied Biosystems 7300 real-time PCR system at the following thermal cycling conditions: 10 min at 95°C, 40 cycles of three steps including 15 sec at 95°C, 1 min at 60°C, and 30 sec at 95°C. Relative expression of KCC2, NKCC1, and GluN2B was determined using the indicated primers in Supplemental Table 2. C_T levels of each were normalized to actin. The Δ-C_T was calculated as C_{T KCC2/NKCC1/GluN2B} – C_{T actin} for each DIV and normalized to DIV 0.

Statistical analysis

Western blots were quantified using densitometry on ImageJ (NIH). Raw integrated densities were exported and analyzed in Microsoft Excel. To determine parametricity, all data were subjected to the Shapiro-Wilk normality test. Data that satisfied the conditions of this test were considered normally distributed. For experiments in this category, the data shown were analyzed by Student’s two-tailed T-tests to differentiate between experimental vs. control protein levels at each timepoint. For data that did not pass the Shapiro-Wilk test, the nonparametric Mann-Whitney test was performed. Where indicated in the figure legends, a one-way ANOVA was also used to verify significance. All graphs and statistical analyses were conducted using Prism version 10.1.2. No blinding was performed for any of the experiments, and no test for outliers was conducted. At least three biological replicates were conducted for each experiment, and the N for each experiment is indicated in each figure legend. Specifically, data for studies in yeast represent 5–6 individual clones from at least two independent transformations; data for studies in transfected HEK293 cells represent independent preparations from three biological replicates (each examined as technical duplicates or triplicates); for rat neurons, at every time point, each independent experiment was performed using the pooled embryos from pregnant rats (also see above). In all figures, *p < 0.05, **p < 0.01, ***p < 0.001.

Results

Development of a new KCC2 expression system in S. cerevisiae

Despite its critical function in brain development and disease relevance, many questions remain regarding the stability of KCC2 and whether quality control mechanisms regulate KCC2 biogenesis. To address these questions, we first developed a novel KCC2 expression system in S. cerevisiae. The power of using yeast is that conserved cellular processes can be quickly tested and elucidated to reveal functional information regarding their homologous human counterparts [66–71]. While yeast have been previously used to express truncated KCC2 in two-hybrid assays [72], here we sought to analyze full-length KCC2 expression and degradation in yeast. Notably, our laboratory has used yeast for >25 years to identify factors that mediate protein degradation, facilitate protein folding, and traffic proteins to and from the cell surface in this system, and we have established that some disease-causing alleles in ion channels and transporters target misfolded proteins for degradation [44, 62, 73–75]. For example, select mutations in NCC, a CCC family member, destabilize the protein so it is routed to the ERAD pathway more readily than WT NCC, which is also folding-compromised and turned over to a lower degree via ERAD [43, 44]. In the case of NCC–and in each publication cited above—the key data in yeast were recapitulated in human cell lines.

To examine if KCC2 exhibits similar characteristics, we utilized an HA-tagged KCC2 plasmid to examine the expression, stability, and post-translational modification of KCC2 in yeast (Figure 1A). Western blot analysis revealed the detection of both monomeric and dimeric KCC2 (Figure 1B). This is consistent with published data in other overexpression systems, indicating the presence of monomers and dimers after SDS-PAGE, as well as the predicted structure of KCC2, which shows two monomers interwoven to form a highly interconnected dimer [25–28, 30, 76] (Supplemental Figure 1). To determine whether the yeast system faithfully recapitulates the entry of nascent KCC2 into the ER, where core N-linked glycans are added [30], we treated the cellular lysate with endoglycosidase H (endoH), which removes both ER and mature glycans from proteins. As shown in Figure 1B, this resulted in a downward shift in the molecular weight of the monomeric KCC2 species, and based on the size shift (~18 kDa), this represents the cleavage of the six N-linked glycosylation sites located on the large extracellular/ER lumenal loop of the protein [30]. The lack of a shift observed for dimeric KCC2 may be due to the formation of detergent-resistant aggregates, or simply that the change in mobility cannot be resolved in this gel system or others (e.g., gradient gels) since the relative change in the size of the deglycosylated dimer is too small to observe.

KCC2 is degraded by the ERAD pathway in S. cerevisiae

We next examined KCC2 stability and the possible mechanisms regulating its quality control in the secretory pathway using cycloheximide chase assays. Given that other proteins in the same family as KCC2 – such as NCC and NKCC1 – are targeted for ERAD [43, 45], we hypothesized that KCC2 is also an ERAD substrate. Yeast expressing HA-KCC2 were treated with cycloheximide (CHX) to inhibit protein synthesis, and the degradation of the cotransporter was observed over time. Our results show that KCC2 is highly unstable in yeast to the extent that protein levels were almost undetectable after 90 minutes (Figure 1C). To assess whether KCC2 degradation was proteasome-dependent, a key indicator of ERAD, the KCC2 expression vector was transformed into a pdr5Δ strain, which lacks a multidrug efflux pump, and cells were treated with the proteasome-specific inhibitor MG132 or in the presence of the solvent [77]. Consistent with ERAD targeting, MG132 treatment significantly increased KCC2 stability (Figure 1C). Quantitative analysis of monomeric as well as the dimeric form of KCC2, respectively, were also significantly stabilized in the presence of MG132 (Supplemental Figure 2A).

To further support the hypothesis that KCC2 is an ERAD substrate, cycloheximide chase assays were conducted in strains lacking Hrd1 and Doa10, the two primary ERAD-requiring E3 ubiquitin ligases in yeast [78]. As displayed in Figure 1D, KCC2 was more stable over time in a hrd1Δ doa10Δ yeast strain relative to yeast in which the E3 ligases were intact. Independent quantification of monomeric and dimeric KCC2 again showed significant stabilization (Supplemental Figure 2B), providing additional evidence that KCC2 is an ERAD substrate in this model.

Yet another avenue for the degradation of membrane proteins is the vacuole, the lysosome equivalent in fungi and plants [79]. Notably, some ion channels and transporters in both yeast and human cells are targeted for both ERAD and post-ER degradation in the vacuole/lysosome [74, 80–82]. Therefore, a yeast strain lacking Pep4, which is required for vacuole-dependent proteolysis [83], was used. Interestingly, the expression of KCC2 in pep4Δ cells failed to stabilize the protein at any timepoint (Figure 1E), indicating that ERAD is the primary mechanism by which KCC2 is degraded in yeast.

It is worth noting that two KCC2 isoforms, known as KCC2a and KCC2b, exist in neurons and differ in the first 40 amino acids at the N-terminus [84]. During early brain development, the isoforms are present at comparable levels, with KCC2b protein levels increasing and becoming higher than KCC2a in the maturing brain [85, 86]. KCC2 knockout mice lacking both isoforms exhibit respiratory failure at birth [9], and knockout of the KCC2b isoform leads to frequent seizure activity and death shortly after birth [7]. Because our analyses in yeast, above, utilized the KCC2a isoform, we asked if differences in protein stability are evident between the two isoforms. Therefore, we also created a yeast expression plasmid encoding KCC2b and conducted cycloheximide chase assays in the pdr5Δ strain. No differences in growth were observed between KCC2a- or KCC2b-expressing yeast (data not shown), and as shown in Supplemental Figure 3, side-by-side analysis of the isoforms revealed comparable stabilities, with little starting material remaining after 90 minutes. In addition, treatment with MG132 stabilized the isoforms, suggesting that both are targeted for degradation by ERAD, despite differences at the N-termini.

KCC2 maturation can be visualized in HEK293 cells

As indicated in the preceding section, experiments in yeast are suggestive of cellular processes that occur in mammalian cells, and results are commonly recapitulated in higher model systems. To determine whether this is also the case for KCC2, we expressed KCC2b in HEK293 cells using two different plasmids. One plasmid encodes an HA- and mCherry-co-tagged KCC2 that has been used extensively to observe KCC2 function at the cell-surface and localization in mouse neuroblastoma (N2a) cells [15], while the other encodes untagged KCC2. We first examined the pattern of KCC2 expression utilizing live cell imaging of KCC2-HA-mCherry in HEK293 cells. The transporter was not only prominent in cytosolic regions of the cell but was also present at the plasma membrane (note “merge” in Figure 2). Moreover, the reticular staining pattern around the nucleus was strongly suggestive of KCC2 residence in the ER.

Figure 2. KCC2-HA-mCherry residence in HEK293 cells.

Live cell imaging of HEK293 cells expressing KCC2-HA-mCherry. Nucleus is labeled with DAPI (blue) and plasma membrane is labeled with Concanavalin A (ConA, magenta).

Consistent with this supposition, western blot analysis of KCC2 uncovered the presence of multiple monomeric KCC2 bands (labeled “A”, “B”, and “C” in Figure 3A). Studies using specific deglycosylases were then used to determine the extent of KCC2 protein maturation through the secretory pathway. Importantly, endoH and PNGaseF differ in their ability to cleave glycans from glycoproteins [87]. In mammalian cells, endoH trims mannose from N-linked glycans located on immature proteins that have not yet left the ER (i.e., band B). More complex glycans, such as those on mature proteins that leave the ER and enter the Golgi or traffic to later steps in the secretory pathway (e.g., the plasma membrane), are endoH-resistant (indicated by band C) but can be cleaved by PNGaseF. As shown in Figure 3A, endoH triggers the collapse of immature glycosylated KCC2 B into the unglycosylated KCC2 A protein, while leaving the more mature glycosylated KCC2 C intact. These data are consistent with live cell imaging (see above), indicating that a KCC2 pool resided at the plasma membrane. Indeed, in contrast to endoH, PNGaseF trims all glycans, so both the B and C proteins are deglycosylated (Figure 3A and the model in Figure 3B). We further noted that the KCC2 A state was absent in cells expressing KCC2-HA-mCherry (Figure 3A). One explanation for this result is that KCC2, when containing the mCherry motif, simply has more time to access the ER resident oligosaccharyltransferases that act during and after protein translocation into the ER [88].

Figure 3. KCC2 maturation is observed in HEK293 cells.

(A) Analysis of KCC2 glycosylation states with endoH and PNGaseF enzymes. Western blot analysis of HEK293 cells expressing KCC2 (left) or KCC2-HA-mCherry (right). EndoH cleaves glycans from immature core glycosylated KCC2 B into the unglycosylated A state. PNGaseF trims glcyans from both immature core B and mature C glycosylated KCC2 into the unglycosylated A state. (B) Representative diagram of KCC2 protein maturation in mammalian cells. KCC2 (purple) progresses from its unglycosylated state A to its immature ER/core glycosylated state B in the endoplasmic reticulum (indicated by the attachment of green core glycans). Once it leaves the ER, KCC2 is transported to the Golgi and receives more complex glycans to be converted into its mature glycosylated C state (indicated with blue, yellow, and red glycans). From here, KCC2 leaves the Golgi and resides on the plasma membrane in order to function. Created with Biorender.com. (C) Western blot analysis of HEK293 cells expressing untagged KCC2 show maturation of the monomeric A and B forms into the mature glycosylated C form over time (left). Quantitative immunoblot analysis reflects the observed maturation from A and B into C (right). N=3. (D) Western blot analysis of HEK293 cells expressing KCC2-HA-mCherry show maturation of the monomeric B form into the mature glycosylated C form over time (left). Quantitative immunoblot analysis reflects the observed maturation from B to C (right). N=3. Data represents the mean ± SEM. N= number of independent cell culture preparations (each examined in technical duplicates or triplicates).

Due to the ability to visualize distinct KCC2 glycosylated states in HEK293 cells that represent different steps in the maturation process, we next wondered whether a cycloheximide chase assay could be used to monitor KCC2 biogenesis in HEK293 cells. In cells expressing untagged KCC2, maturation from the ER-resident KCC2 A and B proteins into the post-ER C protein was evident; the monomeric protein was converted into its mature state after 8 hrs (Figure 3C). This was also observed in cells expressing KCC2-HA-mCherry, in which the ER-resident KCC2 B protein was almost entirely converted into the post-ER C protein after 8 hrs (Figure 3D). These results indicate that the HEK293 model system can be used to study KCC2 maturation through the secretory pathway, and, as will be discussed later, this system may provide a platform to measure the effects of KCC2 mutant alleles on protein trafficking and ER quality control/retention in future studies.

Finally, we again note that the detergent-insoluble KCC2 dimer was mostly resistant to enzyme treatment (Figure 3A), consistent with data in the yeast model, and which may once again be due to overexpression/partial aggregation. To explore whether the dimer can be disrupted under different lysis conditions, cells were lysed into an SDS buffer containing one of the following reducing agents: BME, DTT, or TCEP. The temperature was also taken into consideration, so the lysates were incubated at room temperature (RT), 37°C, 75°C, or 100°C prior to western blot analysis. As shown in Supplemental Figure 4A, none of the reducing agents disrupted the appearance of the KCC2 dimer under any condition, consistent with previous observations with NKCC1 [65]. In fact, increasing the temperature seemed to have the greatest effect on the amount of KCC2 dimer and high molecular weight (HMW) bands, which were noticeably increased when samples were incubated at 100°C. This further suggests that at least some of dimeric KCC2 in HEK293 cells is aggregating and resistant to SDS, even under various reducing conditions.

We also lysed the cells into a Triton-based “TNT” buffer (see Materials and Methods) containing each of the reducing agents (Supplemental Figure 4B). Cells were then diluted in SDS buffer containing the respective reductant and examined by 5% SDS-PAGE. The results for this experiment were similar to those above, i.e., incubation at high temperature (100°C) increased protein aggregation to the extent that almost all of KCC2 remained at the separating-stacker interface. In addition, while BME and DTT did not affect the dimer, TCEP globally reduced the amount of total protein and KCC2 detected, possibly due to low pH activation of denaturing-resistant proteases (note the magnification of lower molecular weight species in the “Total Protein” blot in the presence of TCEP). These results are perhaps also unsurprising given that previous efforts to disrupt the SDS-resistant KCC2 dimer in HEK293 cells were unsuccessful [31]. These data suggest that the accumulation of KCC2 in overexpression systems leads to the aggregation of oligomeric complexes that cannot be disassociated under standard–or even more extreme–denaturing conditions.

It is also relevant to mention that when analyzed side-by-side by western blot, monomeric KCC2a in yeast migrates slightly slower (~4 kDa) than its counterpart in HEK293 cells (Supplemental Figure 5). The change in apparent molecular weight is consistent with the fact that the yeast plasmid contains the N-terminally extended KCC2a isoform as well as an HA tag. Interestingly, dimeric KCC2 in yeast also migrates higher than would be expected after taking these considerations into account; it is noteworthy that the acquisition of additional post-translational modifications, such as O-glycosylation, is common for ER-retained proteins in yeast [89].

A subpopulation of KCC2 is targeted for ERAD in HEK293 cells

Since our experiments in yeast identified ERAD as a quality control mechanism that regulates KCC2 stability (Figure 1), we hypothesized that the protein would similarly be targeted to this pathway in HEK293 cells. Therefore, cycloheximide chase assays were conducted in HEK293 cells expressing untagged KCC2 (Figure 4A–B) or KCC2-HA-mCherry (Figure 4C–D). Measurements of protein stability show that, in both cases, the total amount of KCC2 modestly declined, with ~50% of the starting material remaining after 8 hrs. This result suggests that the more native-like environment in human cell-derived lines is increasingly conducive to proper protein folding. Consistent with this view, upon treatment with MG132, there was modest but statistically insignificant stabilization (see Statistical analysis in the Methods section and the figure legend). In contrast, leupeptin, which inhibits lysosomal proteases, was without effect.

Figure 4. A portion of KCC2 is targeted for ERAD in HEK293 cells.

(A) Cycloheximide (CHX) chase analysis of HEK293 cells expressing KCC2. Cells were treated with MG132 or vehicle (DMSO) (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=4. (B) Cycloheximide chase analysis of HEK293 cells expressing KCC2 treated with leupeptin or vehicle (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=3. (C) Cycloheximide chase analysis of KCC2-HA-mCherry in HEK293 cells. Cells were treated with MG132 or vehicle (DMSO) (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=3. (D) Cycloheximide chase analysis of KCC2-HA-mCherry in HEK293 cells. Cells were treated with leupeptin or vehicle (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=3. Please note: For vehicle control experiments, graphs include experiments that are shown in Figure 3. For all experiments data is the mean ± SEM. N= number of independent cell culture preparations (each examined in technical duplicates or triplicates).

* = p < 0.05

Notably, quantification of each monomeric KCC2 band showed differences in stability in response to MG132 treatment: The ER-resident KCC2 A and B proteins were modestly stabilized by MG132 treatment, while the post-ER KCC2 C was not (Supplemental Figure 6A). This result is consistent with the notion that ER-resident KCC2 may be targeted for ERAD to some degree, and perhaps other proteolytic pathways that operate at the ER [90, 91], whereas the pool of KCC2 that has exited the ER is more resistant. Similar observations were made for both the tagged and untagged protein (Supplemental Figure 6B). It is also important to recognize that treatment with MG132 increased the amount of dimeric KCC2 over time versus the vehicle control. (As stated above, dimeric KCC2 has been seen in other studies using HEK293 cells; also see Discussion). One model to explain these data—which is again consistent with the hypothesis that the dimeric species represents aggregation-prone KCC2—is that protein polyubiquitination, which should increase with MG132 treatment, leads to further aggregation [92]. Furthermore, enhanced ubiquitination exacerbates the pathology of misfolded disease-causing mutations in cells by augmenting protein aggregation [93, 94]. Regardless, consistent with our KCC2 data in the yeast versus the HEK293 cell models, the degree to which other membrane proteins are targeted for ERAD is higher in yeast cells [44, 62, 95, 96].

Recapitulating KCC2 expression during neurodevelopment

While yeast and HEK293 cells provided distinct advantages, i.e., easily measuring specific ERAD requirements versus exploring its maturation through the secretory pathway, respectively, KCC2 is overexpressed in both systems, and neither has necessarily evolved to accommodate this complex membrane protein. Therefore, we next sought to evaluate KCC2 in neurons. As such, we examined KCC2 in primary rat cortical neurons, which represent a preferred model to study developmentally regulated neuronal proteins, such as KCC2, because they are harvested during embryogenesis [97].

We first wanted to establish the timeline in which KCC2 expression is induced during development in this system. Primary rat cortical neurons were harvested at embryonic day 17 (E17) from pregnant Sprague Dawley rats, and the neurons were cultured for up to 20 days in vitro (DIV) (Figure 5A). For each timepoint, KCC2 mRNA and protein levels were examined by RT-qPCR and western blot analysis. As shown in Figure 5B, KCC2 mRNA levels elevate gradually between DIV0 and DIV10, and then increase from DIV10 to DIV20. For controls, we also examined the mRNAs corresponding to two other neuronal proteins, GluN2B and NKCC1. As expected, GluN2B message levels modestly increased while NKCC1 mRNA remained relatively low [98, 99]. Compatible with the qPCR data, western blot analysis at each developmental timepoint reflected a corresponding increase in the KCC2 protein (Figure 5C).

Figure 5. KCC2 expression and stability in primary rat neurons.

(A) Experimental design. Primary neurons were harvested from pregnant Sprague-Dawley rats at embryonic day 17 (E17) and cultured for 0, 3, 5, 10, 15 and 20 days in vitro (DIV). At each timepoint, neurons were lysed and subjected to RT-qPCR and Western blot analysis. Image created with Biorender.com. (B) RT-qPCR of rat primary neurons detected RNA levels of KCC2, NKCC1 and GluN2B. Fold change was normalized to actin reference gene. A one-way ANOVA yielded p = 0.030 for increased KCC2 RNA levels during development. N=3. (C) Western blot analysis identified endogenous KCC2 protein levels in rat primary neurons throughout development. Antibody specific for KCC2 detected primarily monomeric (~130 kDa) KCC2. N=3. (D) Western blot analysis of KCC2 glycosylation states with endoH and PNGaseF enzymes in DIV19 primary rat neurons indicate fully mature KCC2. (E) Cycloheximide (CHX) chase analysis of endogenous KCC2 in DIV20 primary rat neurons with MG132 or vehicle (DMSO) treatment (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=3. (F) Cycloheximide chase analysis of endogenous KCC2 in DIV20 primary rat neurons with leupeptin or vehicle treatment (left). Graph of quantitative immunoblot analysis of total KCC2 protein levels (right). N=3. All quantified western blots were normalized to total protein loading control to ensure consistent loading. Data is the mean ± SEM. N= number of independent neuronal dissociations from one pregnant rat.

The levels of dimeric KCC2 were substantially lower when compared to monomeric KCC2, with 82% of the total KCC2 on average detected as the monomeric species at DIV20. This is consistent with previous reports that endogenous KCC2 resolves as a monomer on SDS-PAGE in standard SDS-denaturing conditions, while dimeric KCC2 is SDS-sensitive and less abundant [29, 31, 46]. This differs from experiments in our yeast and HEK293 cell overexpression systems, in which dimeric KCC2 is abundant and SDS-resistant. To ensure that differences in the KCC2 monomer:dimer ratio between HEK293 cells and neuronal cells were not due to discrepancies in lysis buffers or conditions, cellular extracts for both model systems were lysed in TNT versus Cell Extraction Buffer and analyzed by SDS-PAGE (Supplemental Figure 7). The results of this study did not yield any notable differences in KCC2 migration by SDS-PAGE when samples from each cell system were assessed under the alternative lysis condition.

Although these findings demonstrate the pronounced and anticipated increase in KCC2 message and protein during development in vitro, it was unclear whether the KCC2 protein being detected represented the immature or mature protein. Therefore, the glycosylation status of KCC2 was again investigated using endoH and PNGaseF. Treatment of DIV19 neuronal lysates with endoH showed no shift in monomeric KCC2, suggesting that this represents mature protein that has exited the ER (designated C) (Figure 5D). This suggestion was supported by the collapse of mature KCC2 into the unglycosylated KCC2 A protein by PNGaseF. We conclude that primary neurons, which endogenously synthesize KCC2, support robust KCC2 maturation and lack appreciable amounts of the detergent-insoluble dimer.

Based on these data, primary neurons should provide an optimal environment to support KCC2 biogenesis, one in which the protein will fold efficiently and exhibit high stability. To begin to explore this hypothesis, cycloheximide chases were implemented with DIV20 primary rat neurons and treated in the presence or absence of MG132 or leupeptin (Figure 5E and F). DIV20 neurons were selected for analysis to best detect changes in KCC2 levels. As anticipated, no degradation of KCC2 was observed over time.

KCC2 expression in rat and human brain synaptosomes

Thus far, we have assessed endogenous KCC2 in primary cortical neurons that are comprised of whole cell neuronal cultures. To analyze KCC2 maturation in a more specialized region of the neuron, we isolated neuronal synaptosomes from the brain of an adult heterogenous Sprague Dawley/Wistar rat. Notably, the results from adult rat brain synaptosomes were similar to those seen with DIV19 primary neurons (Figure 6A). Specifically, monomeric KCC2 was detectable while dimeric KCC2 levels were negligible. Also, similar to DIV19 neurons, only the post-ER KCC2 C protein was present since this band collapsed into the unglycosylated KCC2 A protein in the presence of PNGaseF but not endoH. Our results indicate that monomeric KCC2 is detectable with SDS-PAGE in the mature glycosylated state while dimeric KCC2 remains SDS-sensitive in adult rat synaptosomes.

Figure 6. KCC2 maturation is consistent between rat and human synaptosomes.

Western blot analysis of KCC2 glycosylation states with endoH and PNGaseF enzymes in (A) adult rat brain synaptsomes and (B) adult human brain synaptosomes. (C) Western blot analysis of KCC2 protein levels in male and female human brain lysates. Replicates represent five different female and male brain samples selected for analysis. Total protein was used as a loading control.

To expand upon this finding, we asked if the results of these experiments in rat primary neurons and rat synaptosomes were representative of what occurs within the adult human brain. This was accomplished by generating synaptosomes from post-mortem precuneus human brain tissue—a region of the cortex critical to consciousness and memory [100]—and treating samples with endoH or PNGaseF (Figure 6B). Results from these experiments were identical to those obtained using primary neurons and from rat synaptosomes and demonstrated the presence of mature glycosylated monomeric KCC2 and the absence of dimeric KCC2.

Since synaptosomes only contain the synaptic terminals of neurons, we also sought a more comprehensive approach to detect KCC2 in human brain homogenates. Precuneus brain samples from five females and five males were selected to control to some extent for age and the cause of death (Supplemental Table 3), e.g., all individuals were between the ages of 22 and 45. In each case, the cause of death was not attributed to a head injury or psychiatric condition, as individuals that met this criterion were excluded. While the material used for the analysis in Figure 6B was from an individual with a post-mortem index of 25.5 hrs, the samples used for this next analysis of ten individuals reflected a PMI from 4.8 to 12.9 hrs. The results from SDS-PAGE again showed that the KCC2 dimer was nearly absent, comprising <4% of the total on average for the 10 samples (Figure 6C). Moreover, there was no correlation between the relative amount of the KCC2 monomer or dimer with age or PMI. These data further support the importance of matching the cellular environment (i.e., the proteostasis network) to endogenously expressed protein, and underscores the value of using primary rat neurons to study KCC2 since these assays yielded similar results to those seen when samples from human brain were examined.

Discussion

In this study, we implemented multiple model systems, each with unique attributes, to gain insight into the biogenesis of KCC2, specifically highlighting differences and similarities in its stability, maturation, and oligomeric properties in each system. For example, our newly developed yeast KCC2 expression system was ideal for observing robust degradation via the ERAD pathway, and as a proof of principle, we demonstrated the contributions of two ER-resident ubiquitin ligases (Hrd1 and Doa10) that play a role in this process. In the future, comparing the effects of the myriad other ERAD-associated factors, many of which exhibit substrate-specific effects [101–103], will be undertaken, along with an analysis of KCC2 mutants that we predict will strongly destabilize the protein and heighten the ERAD requirement. We then explored KCC2 stability in HEK293 cells and showed that ERAD played a more modest role in KCC2 degradation, consistent with a hyperactive ERAD pathway in yeast [43, 62, 95, 96]. Nonetheless, the transfected human cells afforded the previously unexplored advantage of observing KCC2 maturation through the secretory pathway in real time. Here, too, an analysis of KCC2 mutants that compromise maturation efficiency, perhaps in contrast to destabilization, will be pursued in future studies. In turn, experiments in primary rat cortical neurons confirmed the developmental upregulation of both KCC2 mRNA and protein in a tandem analysis, and biochemical assays allowed us to assess KCC2 maturation in primary cultures and specifically in lysates prepared from synaptosomes. This system also showed that endogenously expressed dimeric KCC2 is SDS-sensitive in western blot analyses, as opposed to the SDS-resistance that we observed in overexpression systems. Ultimately, these experiments led us to explore KCC2 maturation and dimer solubility in human brain cortical neurons and synaptosomes. These results were not only consistent with those obtained in primary rat neurons, but were also consistent across multiple human individuals of each sex, at least in our relatively small sample set. Thus, disease-associated mutations that compromise protein trafficking might be uncovered in post-mortem tissue via future efforts.

Historically, diverse model systems have been employed to gain a more complete understanding of protein biogenesis in both native and disease states [62, 74, 75, 104]. As outlined in the Results, yeast have been used to characterize the quality control mechanisms of other proteins within the same family of KCC2, such as NCC and NKCC2 [43–45]. For NCC, a new yeast model was coopted to similarly identify the protein as an ERAD substrate, but then significantly extended to determine the requirements for specific molecular chaperones, the E3 ligases, and also the “retrotranslocation machinery” that liberates the protein from the ER membrane prior to degradation [43]. Because the cycloheximide chase assays performed with NCC in yeast displayed a similar degradation profile as those observed here for KCC2, analogous experiments can now be pursued for KCC2. Importantly, the results for NCC in the yeast system were then recapitulated in mammalian cells and then extended in the setting of disease mutants [44]. To further explore the mechanisms that regulate NKCC2 biogenesis, a yeast two-hybrid system was used to identify factors that contribute to this event [45], an effort that led to the discovery that an ER lectin, OS9, which binds and mediates the degradation of NKCC2 in HEK293 cells. These results were then validated in renal cells. Therefore, it will be enlightening to not only better characterize the mechanisms of KCC2 protein quality control in yeast, e.g., investigate the effects of the cytoplasmic Hsp70 chaperone and the ER-resident lectin OS9, since these were discovered to be involved in the degradation of NCC and NKCC2 [43, 45], but to dissect the mechanisms that oversee the quality control of other members of the CCC family that, to date, remain unexplored. Furthermore, functional assays in yeast to determine whether the KCC2 protein we detected is active—and how this activity may be impacted under disease states and/or destabilizing conditions—represents a promising avenue for future work.

To date, primary neuronal cultures have been repeatedly implemented to study the function, localization, and regulation of neuron-specific proteins, with KCC2 being no exception [99, 105–109]. Here, we utilized primary rat neurons to effectively detect KCC2 developmental expression and assess its maturation in an endogenous expression system. Experiments investigating KCC2 degradation were challenging given the inherent stability of KCC2, indicating that expression systems in which KCC2 is inherently less stable—such as in yeast and HEK293 cells—may be advantageous to assess the factors that regulate KCC2 stability. In addition, the presence of an SDS-resistant KCC2 dimer in yeast and HEK293 systems, and the absence of the dimer in rat and human brain, is likely explained as a result of protein overexpression and/or differences compared to the native lipid environment, either of which could affect detergent solubility. Several previous reports have similarly shown prominent levels of dimeric KCC2 species in overexpression systems [8, 30–33], and despite our efforts (Supplemental Figure 4) and those in other labs [29, 31] the use of various denaturing conditions failed to prevent the appearance of the KCC2 dimer in HEK293 cells. To address the SDS-sensitivity of dimeric KCC2 in endogenous systems, previous studies have utilized native gel electrophoresis to successfully preserve and detect the KCC2 dimer in mouse brain lysates [29, 110, 111]. Therefore, while our studies consistently used SDS-PAGE to compare KCC2 across model systems, future endeavors using native-PAGE should be explored in the event that subtle differences in the KCC2 dimer exist that could not be observed here.

It is worth nothing that while KCC2 likely functions as an oligomer [25–27], questions remain regarding the mechanism of its assembly and how dimerization/multimerization might be regulated at the cell surface to support ion transport. Interestingly, experiments in a rat nerve injury model showed a reduction in dimeric KCC2 when calpain is activated by NMDA to cleave and target KCC2 for degradation [46, 47]. Therefore, future efforts to characterize not only the assembly and function of dimeric KCC2, but also its breakdown and how this process is regulated may reveal new insights into KCC2-associated disease.

To take our analysis of endogenous KCC2 biogenesis one step further, we also assessed KCC2 maturation and dimer solubility in post-mortem human brain samples. These results were consistent with those seen in rat neurons and allowed us to examine sex-type differences, albeit in a small sample size. This is relevant for studies on KCC2 given that previous results have found that differences in KCC2 expression between the sexes affect the duration of GABA_AR-mediated excitation, potentially predisposing females to seizures during early brain development [112–114]. While future experiments that investigate KCC2 biogenesis in humans of both sexes are warranted, here we provide a glimpse into KCC2 characteristics in human cortices under standard denaturing conditions.

HEK293 cells have proved vital to studying KCC2 using electrophysiology and in cell surface expression experiments and drug discovery efforts [16, 17, 115–117]. We now present HEK293 cells as a new tool to measure time-dependent KCC2 maturation through the secretory pathway. In fact, our observation of KCC2 maturation is reminiscent of what has been seen with other disease-relevant ion channels, such as the cystic fibrosis transmembrane conductance regulator (CFTR). Indeed, when expressed in HEK293 cells, pulse chase analyses of CFTR showed a clear transition from a core glycosylated B band to the mature complex glycosylated C state [118]. This was especially useful when mutations that lead to cystic fibrosis, such as the F508del allele, were shown to be retained in the ER and then targeted for ERAD, which results in CF [119]. Therefore, it is plausible that some KCC2 mutations identified in patients with epilepsy, schizophrenia, and/or autism spectrum disorders are maturation-defective [13], especially those that reside in regions (e.g., at or near a transmembrane domain) that are likely essential for KCC2 to attain its final native state. As noted earlier, the HEK293 cell system is consequently a powerful model for future efforts to characterize these and other disease-associated KCC2 mutations. Furthermore, drugs that promote the folding, trafficking, and function of F508del-CFTR were discovered as a result of experiments initially conducted in tissue culture cells that report on protein trafficking to the cell surface [120–122]. One such drug combination that includes a protein folding corrector is Trikafta, a highly effective FDA-approved pharmaceutical for individuals containing at least one copy of the F508del mutation [123]. With this in mind, future ambitions to study the maturation of disease-associated KCC2 mutations may similarly help characterize the molecular defects underlying neurodevelopmental disorders, and open the door to new therapeutic treatments.

Conclusion:

We developed a novel yeast expression system in which KCC2 is rapidly turned over and degraded by ERAD. HEK293 cells were implemented to study KCC2 stability and revealed a new method to assess time-dependent KCC2 maturation through the secretory pathway. In addition, we used primary rat neurons and human cortical samples to assess KCC2 maturation efficiency. In each system, the acquisition of dimeric KCC2 was also measured under standard denaturing conditions. These experiments allowed us to highlight ways in which diverse cellular environments impact KCC2 biogenesis and provide a framework for future efforts to characterize KCC2 in normal and disease states.

Highlights:

• KCC2 is targeted for ERAD in a novel yeast expression system

• The maturation of KCC2 is observed in HEK293 cells in a time-dependent manner

• In rat neurons and human brain, SDS-PAGE shows KCC2 in its monomeric state

Abbreviations List

AC

auditory cortex

BME

2-mercaptoethanol

CCCs

cation-chloride cotransporters

CFTR

cystic fibrosis transmembrane conductance regulator

DIV

day in vitro

DLPC

dorsolateral prefrontal cortex

E17

embryonic day 17

endoH

endoglycosidase H

ER

endoplasmic reticulum

ERAD

endoplasmic reticulum-associated degradation

G6PD

glucose-6-phosphate dehydrogenase

GABA

γ-aminobutyric acid

GABA_AR

GABA_A receptor

HMW

high molecular weight

NCC

Na⁺/Cl⁻ co-transporter

NKCC1

Na⁺/K⁺/Cl⁻ co-transporter 1

NKCC2

Na⁺/K⁺/Cl⁻ co-transporter 2

PIC

protease inhibitor cocktail

PMI

post-mortem interval

PMSF

phenylmethylsulfonyl fluoride

RRIDs

Research Resource Identifiers

SLC12

solute carrier 12

SDS

sodium dodecyl sulphate

TCA

tricholoracetic acid

TCEP

tris(2-carboxyethyl)phosphine