Microglia Kv1.3 channels contribute to their ability to kill neurons

Abstract

Many CNS disorders involve an inflammatory response that is orchestrated by cells of the innate immune system: macrophages, neutrophils, and microglia (the endogenous CNS immune cell). Hence, there is considerable interest in anti-inflammatory strategies that target these cells. Microglia express Kv1.3 (KCNA3) channels, which we showed previously are important for their proliferation and the NADPH-mediated respiratory burst. Here, we demonstrate the potential for targeting Kv1.3 channels to control CNS inflammation. Rat microglia express Kv1.2, Kv1.3, and Kv1.5 transcripts and protein, but only a Kv1.3 current was detected. When microglia were activated with lipopolysaccharide or a phorbol ester, only the Kv1.3 transcript (but not protein) expression changed. Using a Transwell cell-culture system that allows separate drug treatment of microglia or neurons, we found that activated microglia killed postnatal hippocampal neurons through a process that requires Kv1.3 channel activity in microglia but not in neurons. A major neurotoxic molecule in this model was peroxynitrite, which is formed from superoxide and nitric oxide; thus, it is significant that Kv1.3 channel blockers reduced the respiratory burst, but not nitric oxide production, by the activated microglia. In addressing the biochemical pathway affected by Kv1.3 channel activity, we found that Kv1.3 acts via a different cellular mechanism from the broad-spectrum drug minocycline, which is often used in animal models of neuroinflammation. That is, the dose-dependent reduction in neuron killing by minocycline corresponded with a reduction in p38 mitogen-activated protein kinase activation in microglia; however, none of the Kv1.3 blockers affected p38 activation.

Figure 1.

Kv channel transcript expression after microglia activation. A-C, Essentially pure rat microglia were left untreated [control (ctl)] or treated for 16 h with PMA (100 nm) or LPS (10 ng/ml). Left panels, Representative gels showing RT-PCR products for each Kv channel (A, Kv1.2; B, Kv1.3; C, Kv1.5) and an internal control in the same sample (β-actin or GAPDH). Right, Each band density (in arbitrary units) was normalized to its internal control, and the mean ± SEM is shown for the number of rat litters indicated on each bar. ^**p < 0.01.

Figure 2.

Protein expression of Kv channels after microglia activation. Microglia were treated as in Figure 1. A-C, The left panels show representative Western blots (A, Kv1.2; B, Kv1.3;C, Kv1.5). The anti-Kv1.2 antibody recognized a strong band of the expected size (67 kDa) and a weaker lower-molecular-weight band. The anti-Kv1.3 antibody recognized the expected 65 kDa band, and the anti-Kv1.5 antibody recognized the expected 60 kDa band. The blot was simultaneously (Kv1.2) probed for the smaller housekeeping protein GAPDH or was cut and the lower portion separately probed for GAPDH (Kv1.3, Kv1.5). Right, Each band density (in arbitrary units) was normalized to its internal GAPDH control, and the mean ± SEM is shown for the number of independent samples indicated. ctl, Control.

Figure 3.

Microglia activation and Kv1.3 current. Conventional whole-cell recordings were made from microglia. The membrane potential was held at -100 mV, and voltage steps were applied between -80 and +40 mV in 20 mV increments. Representative currents are shown for an untreated microglia cell (A) and one treated overnight with 100 ng/ml LPS (B). Top, Recordings in normal bath solution; middle, after addition of 5 nm agitoxin-2; bottom, difference currents to isolate the AgTx-2-sensitive component. C, Summary of untreated [control (ctl)] cells or cells treated with LPS or PMA. Mean amplitude (±SEM) of AgTx-2-sensitive (Kv1.3) current measured during a step to +40 mV and normalized to the cell capacitance (number of cells indicated on each bar). ^*p < 0.05.

Figure 4.

Activated microglia kill neurons. A-C, Properties of postnatal hippocampal neurons grown in Neurobasal A/B27 medium. Immunofluorescence image with differential interference contrast image overlaid (A) of fixed and permeabilized postnatal neurons after 5 DIC in serum-free Neurobasal A/B27 medium. All cell nuclei were labeled with DAPI, and the somato-dendritic compartments of mature neurons were labeled with an anti-MAP-2 monoclonal antibody. Scale bar, 20 μm. Representative whole-cell voltage-clamp recordings (B) from a neuron after 9 DIC in Neurobasal A/B27. From a holding potential of -100 mV, voltage steps were applied in 20 mV increments from -80 to +80 mV. C, D, Diffusible factors from activated microglia kill neurons. C, A representative experiment. Top, Postnatal hippocampal neurons growing in the bottom chamber of the Transwell apparatus, showing negative (no terminal transferase added) and positive controls (DNase I added) for the TUNEL assay. Bottom left, An upper chamber containing a porous insert with microglia was pretreated with LPS (100 ng/ml; 18-24 h), washed three times, and then placed above the lower chamber containing neurons. After 48 h, cells in the lower chamber were labeled with DAPI and TUNEL. Bottom right, Differential interference contrast image of same field of cells. D, Summary of neuron death (percentage TUNEL-positive cells); mean ± SEM, number of independent experiments indicated on each bar. ^**p < 0.01. -ve, Negative; +ve, positive.

Figure 5.

Kv1.3 blockers reduce neuron killing by microglia. Microglia were grown on membranes in the upper Transwell chamber, then treated with LPS (as in Fig. 4), with or without a K⁺ channel blocker, and then washed three times before placing them above neurons in the lower Transwell chamber. The percentage channel block was calculated as 1/[1 + (K_d/[D])], where [D] is the concentration of blocker used in the experiment. From the K_d values in Table 2, ChTx (50 nm) blocks 78% of Kv1.2, 95% of Kv1.3, and 96% of SK4 channels; AgTx (5 nm) blocks ∼100% of Kv1.3 and Kv1.6 channels; α-DTx (50 nm) blocks 75% of Kv1.2 and 71% of Kv1.6 channels. Neuron killing was assessed by TUNEL as in Figure 4 and is expressed as the mean ± SEM for the number of independent experiments indicated on each bar. ^*p<0.05; ^**p<0.01.

Figure 6.

Mechanism of Kv1.3-mediated neuron killing. A, Microglia-dependent killing is caused by peroxynitrite. Microglia growing in porous upper chambers were treated with LPS, washed three times, and then placed above neurons in the lower Transwell chambers, as in Figures 4 and 5. The peroxynitrite scavenger FeTmPyP (2 μm) when added to neuron cocultures at the same time as LPS-stimulated microglia, dramatically reduced the neuron killing assessed at 48 h (^***p < 0.001). B, The respiratory burst in microglia. In separate wells, microglia alone were untreated [control (ctl)] or treated with PMA (150 nm), an inactive analog (4α-PMA; 150 nm), or the NADPH-oxidase inhibitor DPI (1 μm). Each compound was added to cells in medium containing 2 μm DHR-123, and the fluorescence was monitored by flow cytometry after 60 min. For each experiment, 5000 cells were counted, with three replicates from each microglia culture isolated from four different rat litters. The summarized data are presented as mean fluorescence intensity (±SEM) in arbitrary units (^**p < 0.01; n = 4). C, Kv1.3 blockade inhibits the respiratory burst by microglia. Microglia, with or without a Kv channel blocker, were subjected to flow cytometry after PMA treatment as in B. Data are presented as the percentage increase in respiratory burst (mean ± SEM; n = 4 separate experiments) compared with unstimulated microglia. The PMA-induced respiratory burst was inhibited by both compounds that block Kv1.3 (5 mm 4-aminopyridine, 5 nm agitoxin-2; ^#p < 0.05) but not by the Kv1.2 and Kv1.6 blocker (50 nm α-dendrotoxin). D, Nitric oxide production is not inhibited by the Kv channel blockers. Microglia were untreated (ctl) or treated with 10 ng/ml LPS for 12 h, with or without a Kv blocker. Then, the concentration of nitrite (micromolar) released from 50,000 cells per well was determined using the Griess assay. Four replicates for each blocker were used from each of three microglia cultures. Data are presented as the relative percentage increase compared with unstimulated microglia (±SEM; n = 4 separate experiments).

Figure 7.

p38 MAPK activation. Effect of Kv blockers compared with minocycline. A, Minocycline inhibits neuron killing by activated microglia. As in Figures 4 and 5, microglia growing in porous upper chambers of the Transwell apparatus were treated with LPS, with or without minocycline. The microglia were washed three times and then placed above postnatal hippocampal neurons growing in the bottom Transwell chamber. Neuron death was assessed by TUNEL after 48 h, and expressed as the percentage increase in TUNEL-positive (+ve) cells beyond that with unstimulated microglia (mean ± SEM; number of independent cell cultures tested). ^**p < 0.01; ^***p < 0.001. B, LPS induces time-dependent activation of p38 MAPK. Cultured pure microglia were treated with LPS (100 ng/ml) and then lysed at the indicated times and subjected to Western blotting using an antibody that recognizes the activated, phosphorylated form of p38 MAPK. Then, the gel was stripped and reprobed with an antibody that recognizes total p38 MAPK. This gel, which is representative of three independent experiments, shows a transient p38 MAPK phosphorylation with a peak at 30 min. C-E, Minocycline (mino), but not Kv channel blockers, reduces LPS-induced p38 MAPK activation. Cultured pure microglia were treated with LPS (100 ng/ml) for 30 min with or without 200 nm minocycline (C) or one of the Kv channels blockers (D) that was found to be neuroprotective (see Fig. 5): 4-AP (5 mm), AgTx (5 nm), ChTx (50 nm). ctl, Control. Densitometric analysis (E) was used to assess p38 MAPK activation as in B.