Abstract
Brain damage and disease involve activation of microglia and production of potentially neurotoxic molecules, but there are no treatments that effectively target their harmful properties. We present evidence that the small-conductance Ca2+/calmodulin-activated K+ channel KCNN4/ KCa3.1/SK4/IK1 is highly expressed in rat microglia and is a potential therapeutic target for acute brain damage. Using a Transwell cell-culture system that allows separate treatment of the microglia or neurons, we show that activated microglia killed neurons, and this was markedly reduced by treating only the microglia with a selective inhibitor of KCa3.1 channels, triarylmethane-34 (TRAM-34). To assess the role of KCa3.1 channels in microglia activation and key signaling pathways involved, we exploited several fluorescence plate-reader-based assays. KCa3.1 channels contributed to microglia activation, inducible nitric oxide synthase upregulation, production of nitric oxide and peroxynitrite, and to consequent neurotoxicity, protein tyrosine nitration, and caspase 3 activation in the target neurons. Microglia activation involved the signaling pathways p38 mitogen-activated protein kinase (MAPK) and nuclear factor κB (NF-κB), which are important for upregulation of numerous proinflammatory molecules, and the KCa3.1 channels were functionally linked to activation of p38 MAPK but not NF-κB. These in vitro findings translated into in vivo neuroprotection, because we found that degeneration of retinal ganglion cells after optic nerve transection was reduced by intraocular injection of TRAM-34. This study provides evidence that KCa3.1 channels constitute a therapeutic target in the CNS and that inhibiting this K+ channel might benefit acute and chronic neurodegenerative disorders that are caused by or exacerbated by inflammation.
- microglia activation
- neurodegeneration
- neuroinflammation
- Ca2+-activated K+ channel
- KCNN4/KCa3.1/IK1/SK4
- optic nerve
- retinal ganglion cell
Introduction
Under pathological conditions, microglia (resident CNS immune cells) activate and can produce reactive oxygen and nitrogen species and proinflammatory cytokines: molecules that can contribute to axon demyelination and neuron death. Because some microglia functions can exacerbate CNS disorders, including stroke, traumatic brain injury, multiple sclerosis, and several retinal diseases (for review, see Minagar et al., 2002; Nelson et al., 2002; Pocock et al., 2002), controlling their activation might ameliorate immune-mediated CNS disorders. We present evidence that the Ca2+-dependent K+ channel KCa3.1 (KCNN4/IK1/SK4) is a potential therapeutic target for inflammation-mediated neurotoxicity in the CNS. KCa3.1 channels are well suited for roles in immune cells, requiring only a small elevation in Ca2+ (Kd of ∼300 nm), then remaining active at physiologically relevant voltages (Mahaut-Smith and Schlichter, 1989; Ishii et al., 1997; Joiner et al., 1997; Khanna et al., 1999). KCa3.1/KCNN4 current and expression in human T cells increase with activation and regulates proliferation of specific T cell subsets (Mahaut-Smith and Schlichter, 1989; Grissmer et al., 1993; Logsdon et al., 1997; Khanna et al., 1999; Ghanshani et al., 2000; Wulff et al., 2003); thus, it is considered a therapeutic target for disorders mediated by T lymphocytes (e.g., transplant rejection and multiple sclerosis). This channel may also directly affect CNS inflammation, because KCa3.1 currents have been observed in murine and rat microglia (Eder et al., 1997; Khanna et al., 2001), and nonselective KCa3.1 blockers implicate them in production of superoxide free radicals, a potentially neurotoxic outcome of microglia activation (Khanna et al., 2001) (for review, see Schlichter and Khanna, 2002). Here, we used lipopolysaccharide (LPS) and the KCa3.1-selective blocker triarylmethane-34 (TRAM-34) (Wulff et al., 2000) to assess whether KCa3.1 channels contribute to microglia activation and neurotoxic properties and then to elucidate the mechanisms involved.
LPS is well known to upregulate inducible nitric oxide synthase (iNOS) and increases nitric oxide (NO) production in microglia (Possel et al., 2000; Fordyce et al., 2005). The resulting neurotoxicity has been attributed to production of NO (Golde et al., 2003), which inhibits mitochondrial respiration, and to peroxynitrite, a highly reactive molecule that nitrates tyrosine residues on membranes and triggers apoptosis of target cells (for review, see Bolanos et al., 1997). There is no information linking KCa3.1 channels to these processes; therefore, we used TRAM-34 to examine iNOS induction and NO production and processes underlying their ability to kill healthy neurons. Next, to determine whether KCa3.1 channels are involved in signaling pathways by which LPS activates microglia, we examined the mitogen-activated protein kinase (p38 MAPK) and the transcription factor nuclear-factor-κB (NF-κB). After LPS binds to the Toll-like receptor 4, these pathways are important for upregulation of numerous inflammatory molecules (for review, see Zielasek and Hartung, 1996). Finally, we addressed whether KCa3.1 channels are likely to be a good therapeutic target for rescuing CNS neurons from inflammation-mediated degeneration in vivo. Because we found robust KCNN4 mRNA expression in both the healthy and damaged retina, we used intraocular TRAM-34 injections to examine the role of KCa3.1 channels in the apoptotic death of retinal ganglion cells (RGCs) after optic nerve transection.
Parts of this work have been published previously in abstract form (Schlichter et al., 2005).
Materials and Methods
Cell cultures and treatments.
Animal handling followed guidelines from the Canadian Council on Animal Care. Microglia were purified from whole brains of 1- to 2-d-old Wistar rat pups (Charles River, St. Constant, Quebec, Canada) as described previously (Khanna et al., 2001; Fordyce et al., 2005). Briefly, after 10–12 d of culturing, microglia were harvested by shaking the flasks on an orbital shaker (80 rpm, 4–6 h, 37°C) and seeded in serum-free Neurobasal A medium with 2% B27 supplement, 0.05 mg/ml gentamycin, and 0.5 mm l-glutamine (all from Invitrogen, Carlsbad, CA). Astrocyte purity was increased by culturing for 2–3 d, shaking, and removing the nonadherent microglia. Neuron cultures were prepared as described previously (Fordyce et al., 2005), except that embryonic day 18 rat embryos were used, and the cultures were grown for 7–10 d to increase the proportion of mature neurons. Cells were seeded on poly-l-ornithine (Sigma, St. Louis, MO)-treated German coverslips (Bellco Glass, Vineland, NJ) at 3 × 104 cells per well in the medium described above.
When desired, microglia were activated with 100 ng/ml LPS (Sigma), which is commonly used to activate these cells (for review, see Zielasek and Hartung, 1996) without cytotoxicity (Xie et al., 2002; present study). The selective KCa3.1 channel blocker TRAM-34 (gift from Dr. H. Wulff, University of California, Davis, Davis, CA; or synthesized by Toronto Research Chemicals, North York, Toronto, Canada) was dissolved in DMSO and used at 1 μm. As expected from its Kd of 20–25 nm (Wulff et al., 2000; Chandy et al., 2004), we found that this concentration blocked >95% of the KCa3.1 current in transfected CHO cells (data not shown). Even much higher concentrations do not affect other microglial channels, including SK1–SK3, Kv1.3, or Kir2.1 (Wulff et al., 2000). All statistical comparisons were made between TRAM-34 and the solvent (DMSO) control, and we also confirmed that DMSO had no effect. When exposing neuron cultures to microglia, experiments were divided into two phases, as follows. (1) Microglia (106 cells per well) were grown on porous upper inserts of Transwell chambers (BD Biosciences, Franklin Lakes, NJ) in 24-well plates and incubated overnight before stimulation. After treatment with LPS for 24 h, with or without TRAM-34 or the iNOS inhibitor S-methylisothiourea (SMT) (Calbiochem, La Jolla, CA), the inserts were thoroughly washed; thus, target neuron cultures were never exposed to LPS, TRAM-34, or SMT. (2) The washed insert (with 3 μm diameter pores) bearing microglia was placed above naive cortical neurons growing on a coverslip in the bottom well of the Transwell chamber, allowing diffusion of soluble molecules. The chamber was incubated for 24 or 48 h, with or without the peroxynitrite scavenger [5,10,15,20-tetrakis(N-methyl-4′-pyridyl)porphinato iron (III) chloride] (FeTmPyP) (Calbiochem).
KCa3.1 immunohistochemistry.
Microglia mounted on coverslips were washed with PBS (three times for 5 min each), fixed for 30 min in 4% paraformaldehyde, washed (three times for 5 min each), permeabilized for 2 min on ice with 0.01% Triton X-100, and washed again (three times for 5 min each). The cells were labeled with a rabbit polyclonal KCa3.1/SK4 antibody (1:200; Alomone Labs, Jerusalem, Israel) (18 h, 4°C) and then washed with PBS (three times for 5 min each) and labeled (2 h, room temperature) with a cyanine 3 (Cy3)-conjugated secondary antibody (1:500; Jackson ImmunoResearch, West Grove, PA). After a final wash with PBS (three times for 5 min each), the coverslips were mounted on glass slides with 1:1 glycerol/PBS for viewing.
Whole-cell patch-clamp recording.
Microglia on coverslips were mounted in a perfusion chamber (model RC-25; Warner Instruments, Hamden, CT), and the tissue culture medium was replaced with an extracellular (bath) solution containing the following (in mm): 125 NaMeSO4, 5 KMeSO4, 1 MgCl2, 1 CaCl2, 5 glucose, and 10 HEPES, pH 7.4. Whole-cell currents were recorded at room temperature (22 ± 1°C) with an Axopatch-200A amplifier and pClamp version 9 software (Molecular Devices, Palo Alto, CA), and compensated on-line for series resistance and capacitance (for recording details, see Newell and Schlichter, 2005). Pipettes (3–4 MΩ resistance) were filled with a solution buffered to 1.0 μm free Ca2+, which contained the following (in mm): 133 KMeSO4, 2 KCl, 2 K2ATP, 0.9 CaCl2, 1 EGTA, and 10 HEPES, pH 7.2. These low-Cl− solutions eliminated the swelling-activated Cl− current.
Plate-reader assays.
Microglia on Transwell inserts or neurons on coverslips were washed with PBS (three times for 5 min each), fixed for 30 min in 4% paraformaldehyde, washed, permeabilized for 2 min on ice with 0.01% Triton X-100, and placed in 24-well black-walled plates (PerkinElmer, Woodbridge, Ontario, Canada). Rabbit polyclonal antibodies were used to monitor total p38 MAPK (1:750; Cell Signaling Technology, Beverley, MA), phospho-p38 MAPK (1:50 for plate reader; 1:750 for Western blots; Cell Signaling Technology), inhibitory κB-α (IκB-α) (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), and iNOS was monitored with a mouse monoclonal (1:200; Cell Signaling Technology). Neurons were fixed (2% glutaraldehyde and 2% paraformaldehyde) and labeled with a rabbit polyclonal nitrotyrosine antibody (1:200; Cell Signaling Technology). Primary antibody labeling (18 h, 4°C) was followed (2 h, room temperature) by the appropriate Cy3-conjugated secondary antibody (1:500; Jackson ImmunoResearch). A fluorescence plate reader (SPECTRAmax Gemini EM; Molecular Devices, Sunnyvale, CA) was used to measure the fluorescence intensity of each well. Background subtraction was done using control wells without primary antibody. Protein concentrations were measured with a Bio-Rad (Hercules, CA) colorimetric protein assay and BSA standards (Bio-Rad), using an ELISA plate reader (model EL311SX; Bio-Tek Instruments, Winooski, VT). Plate-reader-based signals were standardized as relative fluorescence units (RFU) per milligram of protein in each well. For each treatment, an average RFU per milligram of protein was obtained from three coverslips of cells cultured from one animal, and multiple n values were obtained using cultures from different animals. Nitric oxide production by microglia was quantified with the Griess assay according to the protocol of the manufacturer (Invitrogen), as absorbance at 450 nm from the ELISA plate reader.
Assessing neuron damage.
DNA damage was determined by terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling (TUNEL), according to the protocol of the manufacturer (Roche Applied Science, Laval, Quebec, Canada), and fluorescence was detected using FITC-conjugated streptavidin (1:500; Invitrogen). Control wells with no terminal transferase were included to determine background. To validate the new plate-reader method (above), results were compared with conventional cell counting, as follows. After fixation and washing, coverslips were stained for TUNEL and 4′,6′-diamidino-2-phenylindole (DAPI) (1:3000; Sigma) for 5 min, washed, and mounted on glass slides with 1:1 glycerol/PBS. Cells were imaged at room temperature with a Zeiss (Oberkochen, Germany) Axioplan 2 epifluorescence microscope and 20× quartz objective, photographed with a Zeiss digital camera (Axiocam HRm, black and white), and imaged with Zeiss Axiovision release 4.4 software. For each cell culture (from n separate animals) and each experimental condition, TUNEL-positive and DAPI-positive cells were counted in five microscope fields (150–200 cells per field) on two coverslips and averaged. To further examine whether neuron death was apoptotic, caspase 3 activity was measured after neuron cultures were exposed to microglia, treated with or without LPS or TRAM-34. The cells in the lower Transwell chamber were scraped and harvested and solubilized, and the supernatant was mixed with the fluorescent synthetic substrate Ac-DEVD-AMC (N-acetyl-l-aspartyl-l-glutamyl-l-valyl-l-aspartic acid amide 7-amino-4-methylcoumarin) according to the protocol of the manufacturer (Calbiochem). The fluorescence intensity (excitation, 360 nm; emission, 460 nm) was determined with the plate reader and standardized to protein content, as described above.
Optic nerve transection, drug injection, and cell counting.
The left optic nerve of adult female Sprague Dawley rats (225–250 g; Charles River) was transected within 1.5 mm of the eye under chloral hydrate anesthesia (Koeberle and Ball, 1999). A small piece of Gelfoam soaked in 3% Fluorogold was immediately placed on the cut optic nerve stump to retrograde label RGCs. The vitreous chamber of the eye was injected with 3 μl of TRAM-34 in DMSO, immediately after surgery and at 4 d after axotomy, the time at which RGC death is first observed. Injections were posterior to the limbus of the eye, using a pulled glass micropipette, to ensure that the anterior ocular structures were not damaged. Control intraocular injections consisted of sterile PBS with or without the vehicle, DMSO. At 14 d after optic nerve transection, rats were killed using 7% chloral hydrate anesthetic. Eyes were enucleated, the cornea and lens were removed, and the eye cup was fixed (1.5 h, room temperature) in 4% paraformaldehyde in PBS containing 2% sucrose and then washed with PBS for 15 min. The neural retina was dissected, flat mounted in 1:1 glycerol/PBS, viewed at room temperature with a Zeiss LSM 510 confocal microscope using a 20× quartz objective, and analyzed with Zeiss LSM 510 software version 3.2 SP2 software provided with the microscope. Surviving Fluorogold-labeled RGCs were counted in 70,000 μm2 areas, and densities are expressed as number per square millimeter.
Real-time quantitative reverse transcription (qRT)-PCR was used to monitor gene transcript levels, using primers (Table 1) designed with the “Primer3Output” program (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). RNeasy mini kits (Qiagen, Mississauga, Ontario, Canada) were used to isolate RNA after degrading any contaminating DNA with DNaseI (0.1 U/ml, 15 min, 37°C; Amersham Biosciences, Baie d'Urfe, Quebec, Canada). A two-step reaction was performed according to the instructions of the manufacturer (Invitrogen): total RNA (2 μg) was reverse transcribed using 200 U of SuperScriptII RNase H-reverse transcriptase, with 0.5 mm dNTPs (Invitrogen) and 0.5 μm oligo-dT (Sigma). Amplification was performed on an ABI PRISM 7700 Sequence Detection System (PE Biosystems, Foster City, CA) at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s, 55°C for 15 s, and 72°C for 30 s. “No-template” and “no-amplification” controls (Bustin and Nolan, 2004) were included for each gene. Relative input RNA amounts were determined from a relative standard curve for each gene of interest and the housekeeping gene TATA-box binding protein. The efficiency of each reaction was calculated from a standard curve made by diluting the input mRNA, and data are presented as (1 + efficiency)−ΔCT (Pfaffl, 2001) (ABI Prism 7700 User Bulletin 2, 2001), unless otherwise indicated. Of note, correcting for the slightly different reaction efficiencies made no qualitative difference to the results.
Results
KCNN4 mRNA expression
mRNA expression for KCNN4 (KCa3.1/SK4/IK1) was compared in neuron, astrocyte, and microglia cell cultures whose purity was assessed by immunohistochemistry and qRT-PCR (Fig. 1). Based on cell counting (Fig. 1A), after 7 d of culturing, the neuron cultures contained >70% MAP-2-positive cells (mature neurons), the remaining cells being astrocytes that labeled with glial fibrillary acidic protein (GFAP). Microglia cultures were >99% pure, as judged by staining with tomato lectin, which binds to cell surface N-acetyl-glucosamine and to N-acetyl-lactosamine in activated lysosome membranes (Acarin et al., 1994; Bass et al., 1998). The same results were obtained with qRT-PCR, using complement receptor 3 for microglia, synaptophysin for neurons, and GFAP for astrocytes. In Figure 1B, relative mRNA expression was calculated with respect to the housekeeping gene TATA-box binding protein (see Materials and Methods). To facilitate comparisons with cell counts, the proportional signal contributed by each cell-specific marker was calculated after setting the total signal in each cell culture to 100%. The cortical neuron cultures contained ∼70% neurons and ∼30% astrocytes, and both microglia and astrocyte cultures were 97–99% pure.
Next (Fig. 1C), the KCNN4 mRNA expression in each cell culture was compared. Low KCNN4 mRNA levels were present in the astrocyte and neuron cultures, with some in the neuron cultures deriving from the contaminating astrocytes. Most importantly, microglia expressed much more KCNN4 mRNA than astrocyte or neuron cultures, and activating microglia with LPS did not change this expression. Cultured microglia labeled with an anti-KCa3.1 antibody (Fig. 1D), and, in accord with the lack of change in mRNA level, there was no apparent change in staining intensity after LPS. Because the microglia morphology changes profoundly after LPS treatment (processes retract, cell bodies enlarge), it is not feasible to make quantitative comparisons of staining intensity.
Blocking KCa3.1 channels in microglia reduces their neurotoxic behavior
We next asked whether inhibiting SK4 channels in microglia reduces their ability to activate and damage cultured neurons. It was necessary to rely on a KCa3.1-selective drug, TRAM-34, rather than small interfering RNA-mediated knockdown because, despite considerable testing (>15 different transfection reagents, several retroviral, and lentiviral constructs), none yielded effective transfection or infection, and most treatments were toxic to the microglia or activated them. As shown in Figure 1E, TRAM-34 was used to isolate the KCa3.1 current in cultured rat microglia, because it was extremely small compared with the prevalent inward-rectifier (Kir2.1) and outward rectifier (Kv1.3) currents. The TRAM-34-sensitive component was isolated by a point-by-point subtraction of the current before and after adding the drug. This KCa3.1 current was small and noisy, reversed near the K+ Nernst potential (approximately −85 mV) was active over the entire voltage range examined, and showed the slight relaxation at very positive potentials that was described previously (Khanna et al., 1999).
To test the hypothesis that KCa3.1 channels are involved in the ability of activated microglia to kill neurons, microglia were activated with 100 ng/ml LPS. Most importantly, the target neuron cultures in the lower chamber were never exposed to LPS or TRAM-34. After cell fixation, DNA damage was assessed by counting TUNEL-positive nuclei as a percentage of total DAPI-stained cell nuclei and by measuring changes in TUNEL fluorescence intensity with a plate reader. We confirmed that the TUNEL-positive cells were neurons, not contaminating astrocytes, because almost no cells colabeled with GFAP, and astrocyte cultures exposed to LPS-treated microglia did not become TUNEL-positive during the relevant 48 h time period. After 48 h incubation with unstimulated microglia, 12 ± 0.1% of the target neurons were TUNEL positive (Fig. 2A) compared with <1% in the absence of microglia. TUNEL-positive neurons increased to 22 ± 1% (p < 0.01) after exposure to LPS-activated microglia, whereas the vehicle, DMSO, had no effect. Because TUNEL expression represents a snapshot in time, some dead neurons could have detached when the wells were washed; however, the same result was seen with the caspase 3 assay (below), which did not require washing. When TRAM-34 was present during the LPS treatment, the microglia-mediated killing of target neurons was decreased (p < 0.01) (Fig. 2A) to the control level. Moreover, neither TRAM-34 nor DMSO affected microglia viability, as judged by lactate dehydrogenase release or by Alamar blue staining (data not shown). A similar result was obtained with the more efficient spectrofluorometric analysis (Fig. 2B), wherein mean TUNEL fluorescence intensity was measured from thousands of neurons on each coverslip and reported as RFU, standardized to the protein content for each well (see Materials and Methods). LPS-activated microglia increased TUNEL staining (p < 0.05), whether the vehicle was saline or DMSO, whereas, 1 μm TRAM-34 fully abrogated this excess killing (p < 0.001). Then, because TUNEL can represent both apoptosis and necrosis (Labat-Moleur et al., 1998), caspase 3 activity was used as an additional indicator of apoptotic cell death (Gorman et al., 1998). After 24 h exposure to LPS-treated microglia (Fig. 2C), caspase 3 activity in the target neuron cultures increased by ∼70% (p < 0.05), and TRAM-34 reduced this activation by ∼64% (p < 0.05). Together, our results show that the neurotoxic activity of LPS-activated microglia was dramatically reduced by blocking their KCa3.1 channels.
Mechanism of neuroprotection by KCa3.1 channel blockade
Microglia activation is often accompanied by increased iNOS and NO production (Dheen et al., 2005). Neurotoxicity by NO can be direct, by inhibiting mitochondrial respiration (Baud et al., 2004), or indirect, when peroxynitrite oxidizes cellular proteins, lipids, and nucleic acids (Zhu et al., 2004). Based on the known role of peroxynitrite in neuron killing by microglia (Xie et al., 2002; Fordyce et al., 2005), we used the same experimental approach as above to assess whether iNOS and peroxynitrite contribute to the neurotoxicity. The increased TUNEL signal in the target neurons (Fig. 3A) was substantially reduced by adding the iNOS inhibitor SMT (Calbiochem) at the same time as the LPS (p < 0.05) or by adding the peroxynitrite scavenger FeTmPyP (Calbiochem) during the incubation of microglia with target neurons (p < 0.05).
Although KCa3.1 channels are present in microglia (Eder et al., 1997; Khanna et al., 2001; Schlichter and Khanna, 2002), nothing is known about their involvement in iNOS induction and NO production. We found that LPS stimulation (100 ng/ml, 24 h) greatly increased iNOS protein levels in microglia (measured by spectrofluorometry; p < 0.05) (Fig. 3B), and this induction was strongly inhibited by blocking KCa3.1 channels (1 μm TRAM-34 vs DMSO control; p < 0.05). Although lower iNOS levels should result in less NO production, it is important to directly monitor NO, because iNOS activity might also be affected. LPS stimulation increased NO production more than threefold (p < 0.025) (Fig. 3C), and this was completely abrogated by TRAM-34 (p < 0.01). Finally, because high NO levels and peroxynitrite can affect neuron viability through protein nitration (Alvarez and Radi, 2003), we monitored tyrosine nitration levels of proteins in the target neuron cultures. As shown in Figure 3Di, there was some nitrotyrosine staining in control neuron cultures exposed to unstimulated microglia (conditions under which ∼12% of neurons died), and neuronal staining increased after exposure to LPS-treated microglia. These changes, quantified with the fluorescence plate reader, showed that exposure to LPS-treated microglia increased neuronal tyrosine nitration by ∼1.6-fold (p < 0.05) (Fig. 3Dii), whereas, nitration was reduced to below control levels when the microglia were pretreated with TRAM-34 (p < 0.01). This result is consistent with the effects of TRAM-34 on iNOS and NO production by microglia.
Blocking KCa3.1 channels inhibits activation of p38 MAPK, but not NF-κB, in microglia
Signal transduction pathways involving p38 MAPK and NF-κB contribute to microglia activation and production of proinflammatory molecules (Pawate et al., 2004) that can lead to neurotoxicity in vivo (for review, see Zhang and Stanimirovic, 2002). Therefore, we examined whether activation of either pathway is affected by KCa3.1 channel blockade. p38 MAPK activation was measured using a standard technique, wherein its phosphorylated active form is monitored with a phospho-p38-specific antibody. The plate-reader assay was validated by comparison with Western blotting, which we described previously (Fordyce et al., 2005). Spectrofluorometric monitoring of phospho-p38 MAPK (Fig. 4) showed that treating microglia with 100 ng/ml LPS transiently activated p38 MAPK (Fig. 4Ai), and, based on the robust increase at ∼30 min, this time was chosen for all additional experiments. A representative Western blot (Fig. 4Aii) confirmed the increase in phospho-p38 protein, with no change in total p38 MAPK. Our finding that TRAM-34 reduced phospho-p38 by ∼73% compared with the DMSO control after LPS stimulation (p < 0.05) (Fig. 4B) provides the first evidence that KCa3.1 channels are involved in p38 MAPK phosphorylation/activation in any cell type. We next examined NF-κB signaling, which begins with phosphorylation and degradation of IκB-α, a key component of the cytoplasmic NF-κB complex (Viatour et al., 2005). This releases the p50 and p65 subunits that translocate to the nucleus and promote transcription of proinflammatory genes (Chan and Murphy, 2003). In Figure 4C, a standard approach was used to assess NF-κB activation, which monitors degradation of IκB-α (Nikodemova et al., 2006). As expected, LPS treatment activated NF-κB in microglia, which is seen as a ∼57% reduction in IκB-α levels compared with untreated microglia (p < 0.05). Of note, TRAM-34 did not affect this NF-κB activation; thus, KCa3.1 blockade discriminated between these two important signaling pathways. Of course, it is possible that KCa3.1 blockade also affects other microglia functions or signaling molecules that were not examined in this study.
KCa3.1 blockade in vivo reduces death of retinal ganglion cells after optic nerve transection
To determine whether these in vitro results translate to neuroprotection in the CNS in vivo, we used an optic nerve transection model. Microglia activate and migrate after retinal damage (Thanos and Richter, 1993) (for review, see Vilhardt, 2005), and inflammation has been implicated in the apoptotic degeneration of RGCs, which reaches 80–90% by 2 weeks after axotomy (Koeberle and Ball, 1999; Bahr, 2000; Chen et al., 2002; Koeberle et al., 2004). First (Fig. 5A), we conducted real-time qRT-PCR in whole retinas at 7 d after axotomy, when the rate of RGC apoptosis is maximal (for review, see Bahr, 2000; Chen et al., 2002). KCNN4 mRNA was expressed and unaltered after axotomy, whereas major histocompatibility complex (MHC) class II, a molecule that increases when antigen-presenting cells (including microglia) are activated, was greatly upregulated (approximately eightfold; p < 0.01). Figure 5B shows key changes in the location (Bi) and morphology of retinal microglia after axotomy. In the healthy retina, only a few OX-42-labeled cells were present in the nerve fiber layer (NFL) (Fig. 5Bii), in which they had oval cell bodies and short processes, characteristic of perivascular retinal microglia (Chen et al., 2002). In the adjacent inner plexiform layer (IPL) of the healthy retina (Fig. 5Biii), microglia were densely packed and extensively ramified. By 14 d after axotomy (Fig. 5Biv), large numbers of microglia had migrated to the NFL, in which they aligned with the fascicles of degenerating RGC axons. Thus, our four lines of evidence for microglia activation after axotomy are migration, realignment along the axons, a less ramified morphology, and upregulation of MHC class II mRNA.
To assess the role of KCa3.1 channels in degeneration of RGCs, intraocular injections of saline, TRAM-34, or the DMSO solvent were given at the time of axotomy and again at day 4, the time at which RGC apoptosis begins (Villegas-Perez et al., 1988). Intraocular injection was favored over intravenous or intraperitoneal routes for two main reasons: (1) to restrict the site of drug action, avoiding effects on the peripheral immune system and other tissues that express KCa3.1 (e.g., epithelia); and (2) to ensure entry into retinal tissue and control of the initial (maximal) drug concentration within the orbit. A range of TRAM-34 concentrations was tested because diffusion out of the retina into the brain parenchyma is likely to be slow and incomplete (Ghate and Edelhauser, 2006). Care was taken to avoid puncturing the lens, which evokes release of soluble growth factors (Leon et al., 2000). At 1 d after axotomy (Fig. 5Ci), large numbers of Fluorogold-labeled RGCs were seen, along with more faintly labeled axon bundles. In contrast, by 14 d after axotomy, most of the RGCs had degenerated (Fig. 5Bii), and many microglia contained neuronal debris (Fig. 5Cv). Intraocular TRAM-34 injections increased the numbers of surviving RGCs (Fig. 5Ciii,Civ) but did not affect the microglia relocation, realignment, or morphological changes seen after axotomy (Fig. 5, compare Cvi with Biv). To compare RGC densities at 14 d after axotomy (Fig. 5D), Fluorogold-labeled RGC cell bodies were counted in flat-mounted retinas, from four fields for each retinal region (inner, midperiphery, and outer). TRAM-34 (5 or 50 μm) produced a 2- to 2.5-fold increase in RGC density compared with either saline- or DMSO-injected controls (p < 0.05). Together, our results provide evidence that blocking KCa3.1 channels reduces microglia activation and neuron killing in vitro and that a decrease in production of neurotoxic factors renders TRAM-34 neuroprotective after traumatic CNS injury in vivo.
Discussion
Significance of findings
Among the KCNN family of Ca2+-activated K+ channels, KCNN4/KCa3.1 is thought to be restricted to non-excitable cells (Cahalan et al., 2001; Jensen et al., 2002; Schlichter and Khanna, 2002; Chandy et al., 2004). There are several key findings in the present study. Using real-time qRT-PCR, we show that (1) the KCNN4 mRNA level in microglia does not change when they are activated in vitro (by LPS); (2) KCNN4 mRNA is present in the healthy CNS (the retina); (3) the KCNN4 mRNA level in the retina does not change 7 d after optic nerve transection, the time at which the inflammatory response is maximal; and (4) astrocytes express KCNN4 mRNA but at a much lower level than microglia. Next, by monitoring several cellular functions, we provide the first evidence that microglial KCa3.1 channels (5) contribute to iNOS induction and nitric oxide production and (6) are functionally linked to a key signaling molecule, p38 MAPK, in a gene transcription pathway that upregulates numerous proinflammatory molecules. Then, by examining the outcome of exposing naive neurons to activated microglia, we show that microglia KCa3.1 channels contribute to (7) their neurotoxic capacity and (8) to the consequent increases in tyrosine nitration of proteins and caspase 3 activity in the target neurons. Finally, we show that (9) intraocular injection of the KCa3.1 blocker significantly reduces the degeneration of retinal ganglion cells in vivo. This work also demonstrates the utility of fluorescence plate-reader assays for studies of microglia activation and their effects on neurons.
KCa3.1 and cell functions
Expression of KCNN channels (mainly mRNA) has been examined in numerous tissues, but their roles in non-excitable cells are only beginning to be elucidated. KCa3.1 is involved in microglia functions (Khanna et al., 2001; present study), lymphocyte activation (for review, see Jensen et al., 2002; Schlichter and Khanna, 2002), volume regulation in red blood cells and lymphocytes (Brugnara et al., 1996; Khanna et al., 1999), and migration of renal epithelial cells and mast cells (Schwab et al., 1999; Cruse et al., 2006). Because SK channels are not voltage gated, they are well designed to contribute to the resting potential of non-excitable cells, including immune cells (e.g., microglia). They require only a small elevation in intracellular Ca2+ to activate and then are open over a wide voltage range, including very negative membrane potentials (Kohler et al., 1996; Ishii et al., 1997; Joiner et al., 1997; Khanna et al., 1999). A general role for K+ channels is to provide a pathway for K+ efflux, which drives the membrane potential toward the K+ Nernst potential (approximately −85 mV in mammalian cells) and thus can counteract the depolarizing effect of Ca2+ influx. Maintaining a large driving force for Ca2+ entry is thought to be the mechanism whereby KCa3.1 channels contribute to T-cell functions (Logsdon et al., 1997; Khanna et al., 1999) (for review, see Jensen et al., 2002; Chandy et al., 2004). We found that the KCa3.1 current is very small in primary rat microglia and extremely difficult to separate from several other endogenous currents (especially Cl−, inward rectifier and ERG K+ channels, and multiple TRP channels). This small amplitude is not surprising, because we found previously that the membrane resistance of rat microglia is very high (∼8 GΩ) around the resting potential (−40 to −70 mV) (Newell and Schlichter, 2005). Thus, the activity of only a few KCa3.1 channels could have a large impact on the membrane potential and microglia functions. We showed previously evidence that KCa3.1 channels contribute to the NADPH-mediated respiratory burst, which involves Ca2+ entry (Khanna et al., 2001; Fordyce et al., 2005) (for review, see Schlichter and Khanna, 2002). Those studies used charybdotoxin, which also blocks Kv1.3 channels, or clotrimazole, which has other cellular effects in addition to blocking KCa3.1 channels (Chandy et al., 2004). Here, we exploit the KCa3.1 blocker TRAM-34, which has excellent selectivity for KCa3.1 over other K+ channels at the concentrations used (Wulff et al., 2000).
KCa3.1 and neurotoxicity
LPS upregulates iNOS expression and NO production in microglia (Possel et al., 2000; Fordyce et al., 2005). NO can combine with superoxide, which is produced by the respiratory burst in microglia (Colton and Gilbert, 1993), to produce highly reactive peroxynitrite (Torreilles, 2001), a molecule that can exacerbate CNS damage (for review, see Bolanos et al., 1997). iNOS induction and peroxynitrite formation are thought to be the main toxic mediators when LPS-stimulated microglia are cocultured with neurons (Xie et al., 2002; Fordyce et al., 2005), and, in agreement, we found that either an iNOS inhibitor or a peroxynitrite scavenger reduced neurotoxicity. Microglia were apparently the main NO source in this model, because blocking their KCa3.1 channels with TRAM-34 prevented LPS-stimulated iNOS upregulation and NO production and was sufficient to abrogate tyrosine nitration and death of target neurons in vitro. These roles of KCa3.1 channels in vitro correspond well with the neuroprotection seen in the optic nerve transection model in vivo. In this model of traumatic neurodegeneration, it is known that microglia recognize and eliminate severed RGCs (Chen et al., 2002) and that iNOS inhibition reduces RGC death (Koeberle and Ball, 1999). We found that intraocular injections of TRAM-34 reduced this RGC death, providing evidence that KCa3.1 channels may be a therapeutic target for brain pathologies involving inflammation and microglia activation. Importantly, there was no evidence of drug toxicity with even the highest doses of TRAM-34, i.e., there were no seizures or changes in feeding, activity, or behavior. The outcome of microglia activation can be complex and determined primarily by the factors they produce (for review, see Zielasek and Hartung, 1996; Nelson et al., 2002; Vilhardt, 2005). Together with our in vitro results showing an involvement of KCa3.1 channels in microglia iNOS induction, nitric oxide production, and p38 MAPK activation, the in vivo results suggest that KCa3.1 blockade likely acts by reducing production and/or secretion of soluble neurotoxic molecules in the retina. Our results suggest that KCa3.1 channels are not essential for microglial shape changes, migration, or phagocytosis in the retinal damage model in vivo, because TRAM-34 treatment did not restore the resting microglia morphology (many remained less ramified than in the healthy retina), did not prevent their migration and realignment along axons in the ganglion cell layer, and did not prevent them from phagocytosing damaged neurons.
The proinflammatory responses of immune cells, including microglia, involve p38 MAPK and the transcription factor NF-κB (Madrid et al., 2001). There is considerable interest in using p38 MAPK inhibitors to reduce neuron death in vivo, for instance, after stroke (for review, see Barone et al., 2001; Zhang and Stanimirovic, 2002). LPS activates p38 MAPK and NF-κB, promoting the transcription of proinflammatory molecules, including tumor necrosis factor-α, interleukin-1β, and iNOS (Laflamme and Rivest, 1999). Although crosstalk can occur between these two pathways (Madrid et al., 2001), there is also evidence that they can mediate production of different molecules. In particular, p38 MAPK phosphorylation has been linked to iNOS induction and NO production (Barone et al., 2001; Zhang and Stanimirovic, 2002). Our studies appear to provide the first evidence that these two signaling pathways are affected by different K+ channels. That is, blocking KCa3.1 channels in microglia inhibited p38 MAPK (but not NF-κB) activation, iNOS induction, and NO production (present study), whereas, blocking voltage-gated Kv1.3 channels inhibited NF-κB (but not p38 MAPK) activation and reduced superoxide but not nitric oxide production (Fordyce et al., 2005).
Broader implications
Potential therapeutic uses of KCa3.1 blockers were first examined for disorders involving T-cell activation, sickle cell anemia (for review, see Jensen et al., 2002; Chandy et al., 2004), and arterial restenosis (Kohler et al., 2003), and this channel is also involved in endothelial cell proliferation and angiogenesis (Grgic et al., 2005), growth of pancreatic cancer cells (Jager et al., 2004), and phenotypic modulation of coronary smooth muscle (Tharp et al., 2006). Use of KCa3.1 channel blockers to treat CNS disorders is beginning to be addressed, and the results are encouraging. Bayer (Wuppertal, Germany) developed several KCa3.1 blockers, tested them in a rat model of traumatic brain injury (subdural hematoma), and found that several hours of intravenous administration of either a triazole or a cyclohexadiene compound significantly reduced the ensuing edema, intracranial pressure, and infarct volume (Mauler et al., 2004). KCa3.1 blockade reduced inflammatory cytokine production and damage in the spinal cord in a murine model of multiple sclerosis (Reich et al., 2005), but potential contributions of CNS cells were not determined. Blocking KCa3.1 channels in endothelia and smooth muscle cells in the rat middle cerebral artery inhibited hyperpolarization and vessel relaxation (McNeish et al., 2006). Together with the roles of KCa3.1 in microglia-mediated neurodegeneration, this suggests a possible target for stroke. The present in vitro studies and improved survival of neurons after optic nerve transection in vivo provide additional evidence that blocking KCa3.1 channels can reduce microglia-induced neuroinflammation and consequent apoptotic neurodegeneration. Moreover, because microglia and peripheral macrophages share many mechanisms for proinflammatory molecule production, these results have broader implications for treating other inflammatory pathologies that involve the innate immune system.
Footnotes
-
L.C.S. was supported by Canadian Institutes for Health Research Grant MT-13657 and Heart and Stroke Foundation, Ontario Chapter, Grants T4670 and T5546. P.D.K. was supported by a fellowship from the Heart and Stroke Foundation. We thank Xiaoping Zhu and Chris Fordyce for technical assistance.
- Correspondence should be addressed to Lyanne C. Schlichter, Toronto Western Hospital, MC9-417, 399 Bathurst Street, Toronto, Ontario, Canada M5T 2S8. schlicht{at}uhnres.utoronto.ca