KCa3.1/IK1 Channel Regulation by cGMP-Dependent Protein Kinase (PKG) via Reactive Oxygen Species and CaMKII in Microglia: An Immune Modulating Feedback System?

The intermediate conductance Ca2+-activated K+ channel, KCa3.1 (IK1/SK4/KCNN4) is widely expressed in the innate and adaptive immune system. KCa3.1 contributes to proliferation of activated T lymphocytes, and in CNS-resident microglia, it contributes to Ca2+ signaling, migration, and production of pro-inflammatory mediators (e.g., reactive oxygen species, ROS). KCa3.1 is under investigation as a therapeutic target for CNS disorders that involve microglial activation and T cells. However, KCa3.1 is post-translationally regulated, and this will determine when and how much it can contribute to cell functions. We previously found that KCa3.1 trafficking and gating require calmodulin (CaM) binding, and this is inhibited by cAMP kinase (PKA) acting at a single phosphorylation site. The same site is potentially phosphorylated by cGMP kinase (PKG), and in some cells, PKG can increase Ca2+, CaM activation, and ROS. Here, we addressed KCa3.1 regulation through PKG-dependent pathways in primary rat microglia and the MLS-9 microglia cell line, using perforated-patch recordings to preserve intracellular signaling. Elevating cGMP increased both the KCa3.1 current and intracellular ROS production, and both were prevented by the selective PKG inhibitor, KT5823. The cGMP/PKG-evoked increase in KCa3.1 current in intact MLS-9 microglia was mediated by ROS, mimicked by applying hydrogen peroxide (H2O2), inhibited by a ROS scavenger (MGP), and prevented by a selective CaMKII inhibitor (mAIP). Similar results were seen in alternative-activated primary rat microglia; their KCa3.1 current required PKG, ROS, and CaMKII, and they had increased ROS production that required KCa3.1 activity. The increase in current apparently did not result from direct effects on the channel open probability (Po) or Ca2+ dependence because, in inside-out patches from transfected HEK293 cells, single-channel activity was not affected by cGMP, PKG, H2O2 at normal or elevated intracellular Ca2+. The regulation pathway we have identified in intact microglia and MLS-9 cells is expected to have broad implications because KCa3.1 plays important roles in numerous cells and tissues.

KCa3.1 expression, activity, and contributions to T cell functions are governed by the cells' activation state but until recently, little was known about this aspect for microglia. In rat microglia, we found that KCa3.1 contributes to production of reactive oxygen species (ROS) (14), to p38 MAPK activation, nitric oxide, and peroxynitrite production in classical-activated microglia ("M1," by analogy with macrophages polarized by T cells), and to their capacity to kill neurons in vitro and in vivo (15). We recently discovered that when rat microglia are skewed to the anti-inflammatory"alternative" (M2) activation state using interleukin-4, both KCa3.1 expression and current are highly upregulated through the type I IL-4 receptor and subsequent signaling through JAK3, Ras/MEK/ERK, and the transcription factor, AP-1 (20). While KCa3.1 is involved in microglial ROS production, it is not known if its activity is regulated by ROS. There is some indirect evidence that this might occur. ROS can evoke Ca 2+ release from internal stores in Jurkat T cells (21) and pancreatic β-cells (22). We found that, in microglia, the KCa3.1 current is functionally coupled to Ca 2+ -release activated Ca 2+ (CRAC) channels. When CRAC was activated through P 2 Y 2 metabotropic purinergic receptors, this activated KCa3.1 channels, and they then contributed to the microglia migratory phenotype (23). Migration is also increased in alternative-activated microglia, and this depends on KCa3.1 (20,24).
KCa3.1 activity is also post-translationally regulated. Most fundamental is its absolute requirement for Ca 2+ and calmodulin (CaM) in order for the channels to open (8,25) and to traffic to the cell surface (26). CaM is bound to the C terminus of the channel but there is evidence that this interaction can be modulated. We found that the KCa3.1 current is inhibited by cAMP kinase (PKA) through a single phosphorylation consensus site (S334 in human; S332 in rodent, which is two amino acids shorter), and a consequent decrease in CaM binding to the channel (27). Many years ago, we observed that the KCa3.1 current in human T lymphoblasts was reduced by the CaM kinase inhibitor, KN-62, but only at 37°C (not room temperature) and not in KCa3.1 heterologously expressed in CHO cells (8). Both observations suggested an indirect modulation by CaM kinase, but this was not investigated further.
More recent studies suggested a possible link between KCa3.1 and CaMKII, ROS and cGMP-protein kinase (PKG). In cardiac and neuronal cells, the PKG pathway can stimulate ROS production and activate CaM/CaMKII signaling (28)(29)(30). Numerous stimuli can activate PKG in immune cells, and we noticed that the Ser334/S332 site in KCa3.1 that is regulated by PKA (27) is also a potential consensus site for phosphorylation by PKG. Thus, the present study was designed to test whether KCa3.1 is directly inhibited by cGMP/PKG, and if not, whether it is indirectly regulated and whether this involves crosstalk between Ca 2+ , ROS, and CaM/CaMKII. First, we analyzed the endogenous KCa3.1 current in a rat microglial cell line because we found that every MLS-9 cell expresses a robust KCa3.1 current that can be easily isolated (23,31). Then, we corroborated the salient findings on native KCa3.1 channels in primary rat microglia. Importantly, regulation of native channels was studied using the perforated-patch recording configuration to maintain intracellular soluble mediators and biochemical signaling pathways, and to allow cytoplasmic Ca 2+ to remain at physiological levels and change with treatments.

CELLS
Primary cultured rat microglia and the MLS-9 microglia cell line were used to study native KCa3.1 channels, and transfected HEK293 cells were used to facilitate single-channel analysis.

Single-channel recordings
Inside-out patches were excised from HEK293 cells that had been transfected with hKCa3.1. Recordings were made at room temperature, sampled at 5 kHz, and low-pass filtered at 1 kHz (−3 dB cut-off frequency). Pipettes were pulled from thin-walled borosilicate glass (6)(7)(8) and filled with an extracellular solution containing (in mM) 145 KCl, 1 MgCl 2 , and 1 CaCl 2 , 5 HEPES; pH 7.4 (adjusted with KOH), adjusted to~300 mOsm with sucrose. The bath solution contained (in mM) 145 KCl, 1 MgCl 2 , 1 CaCl 2 , 0.1 MgATP, 5 HEPES, and 5 glucose and 1.2, 1.5 or 2.8 EGTA, to yield intracellular free Ca 2+ concentrations of 1 µM, 500 nM, and 120 nM respectively; pH 7.2 (adjusted with KOH),~300 mOsm. NP o , the product of the apparent number of active channels in the patch (N ) and the channel open probability (P o ) was calculated in pClamp by dividing the mean total current (I ) by the single-channel current amplitude (i), where NP o = I/i. The singlechannel current was determined from the best Gaussian fit to the single-channel event amplitude histogram, and this also indicated the apparent number of active channels in the patch. At the end of each recording, 1 µM TRAM-34 was added to block the channels and identify them as KCa3.1.

MEASURING INTRACELLULAR ROS
Primary rat microglia and MLS-9 cells in 96-well tissue culture plates were incubated with 5 µM of CM-H 2 DCFDA (5-(and-6)-chloromethyl-2 ,7 -dichlorodihydrofluorescein diacetate acetyl ester) at 37°C for 1 h. After this cell-permeant reagent diffuses into cells, the acetyl group is removed by intracellular esterases, and it can then be oxidized by the intracellular reactive species, hydrogen peroxide (H 2 O 2 ), hydroxyl radical (OH • ), and peroxynitrite (ONOO -) (42). The resulting fluorescent adduct, dichlorofluorescein (DCF), is produced in proportion to the reactive species and remains trapped inside the cell. Before plate-reader analysis, cells were washed with the same extracellular solution used for perforated-patch recordings. Intracellular DCF fluorescence was excited at 480 nm, and measured at 530 nm with a fluorescence plate reader (Victor3 1420 Multilabel Plate Counter; Perkin Elmer Inc, Waltham, MA, USA). For each treatment, the fluorescence was averaged from two separate wells of cells cultured from one animal (for primary microglia) or from one passage (for MLS-9 cells), and then multiple n values were obtained using cultures from different animals or cell passages. Background subtraction was performed using control wells without CM-H 2 DCFDA. The fluorescence intensity for each treatment group was normalized to the corresponding control value.

INTRACELLULAR FREE Ca 2+
The Fura-2 imaging methods were the same as recently described (23,27,31). In brief, cells growing on glass coverslips (~7 × 10 4 cells/15 mm diameter coverslip) were incubated at room temperature with 3.5 µg/mL Fura-2AM (Invitrogen) for 40 min in the dark. For recording, a coverslip was mounted in a 300 µL volume perfusion chamber (Model RC-25, Warner Instruments, Hamden, CT, USA) that contained the same bath solution as for perforated-patch recording. Responses to db-cGMP and H 2 O 2 were assessed on different batches of cells from separate coverslips. Images were acquired at room temperature using a Nikon Diaphot inverted microscope, Retiga-EX camera (Q-Imaging, Burnaby, BC, Canada), and Northern Eclipse image acquisition software (Empix Imaging, Mississauga, ON, Canada). A Lambda DG-4 Ultra High Speed Wavelength Switcher (Sutter Instruments, Novato, CA, USA) was used to alternately acquire images at 340 and 380 nm excitation wavelengths. Images were acquired every 4 s, and the excitation shutter was closed between acquisitions to prevent photobleaching. The intracellular free Ca 2+ concentration was calculated from the standard equation (43).

CHEMICALS
Working solutions were prepared just before use by diluting fresh aliquots of stock solutions of dibutyryl-cGMP (db-cGMP), cGMP, KT5823, N -(2-mercaptopropionyl)glycine (MPG), myristolated autocamtide-2 related inhibitory peptide for CaMKII (mAIP), TRAM-34, apamin, riluzole, and rat recombinant IL-4. Stock solutions were prepared as follows: KT5823, TRAM-34, apamin, and riluzole were prepared in DMSO and stored at -20°C; db-cGMP, cGMP, and mAIP were prepared in distilled H 2 O and stored at -20°C; MPG was prepared in distilled H 2 O and stored at 4°C; IL-4 was prepared in sterile PBS containing 0.1% BSA and stored at -20°C. H 2 O 2 was prepared fresh daily from a 30% w/w liquid stock (9.8 M). Aliquots of the original stock of PKG Iα holoenzyme were stored at -80°C, and diluted immediately before use in the bath solution that was used for single-channel recordings. PKG Iα holoenzyme and mAIP were obtained from Calbiochem (EMD Biosciences; San Diego, CA, USA), rat recombinant IL-4 was from R&D Systems Inc. (Minneapolis, MN, USA), and all other reagents were from Sigma Aldrich (Oakville, ON, Canada).

STATISTICAL ANALYSIS
Data are expressed as mean ± SEM. For single and multiple comparisons to assess treatment effects on currents, either an unpaired Student's t -test or 1-way ANOVA with Tukey's post hoc test was used. Changes in intracellular Ca 2+ following treatment with db-cGMP or H 2 O 2 were analyzed using a paired Student's t -test. For analysis of intracellular ROS production, a 2-way ANOVA followed by Bonferroni's post hoc test was used, with stimulation (untreated, 24 h IL-4, 48 h IL-4) and inhibitors (untreated, KT5823, TRAM-34) as the two independent variables. Analyses were conducted www.frontiersin.org FIGURE 1 | The endogenous KCa3.1 current in MLS-9 microglial cells is increased by cGMP, which requires cGMP-protein kinase. For all traces, the voltage protocol was a holding potential of -70 mV, and repeated ramps from -100 to +80 mV. Recordings were conducted at room temperature in the perforated-patch configuration, and riluzole was used simply to activate the KCa3.1 current at the normal low intracellular Ca 2+ concentration. The bath always contained 100 nM apamin, a KCa2.1-2.3 channel blocker. Summarized data from a population study using the treatments in panels A-C. For each cell, the KCa3.1 current amplitude was measured at +80 mV, as the component of the riluzole-activated current that was blocked by TRAM-34 (1 µM). The current was always normalized to the cell capacitance (in pF) to account for any differences in cell size and expressed as current density. The TRAM-34-sensitive KCa3.1 current is expressed as mean ± SEM for the number of cells indicated on each bar, and data were compared using one-way ANOVA, with Tukey's post hoc test. **p < 0.01, indicates a difference from both controls and KT5823-treated cells. † † p < 0.01, for the comparison indicated. There was no difference between the control and KT5823-treated cells.
using GraphPad Prism ver 6.01 (GraphPad Software, San Diego, CA, USA), and statistical significance was taken as p < 0.05.
To study the native channels in MLS-9 and primary microglial cells, we used riluzole because it reliably activated a KCa3.1 current in perforated-patch recordings. The current was also stable enough to add TRAM-34 to confirm the channel identity and quantify the current density (20,31). As expected, the KCa3.1 current was not activated at resting levels of intracellular Ca 2+ in MLS-9 cells. However, a stable KCa3.1 current was activated by riluzole in all cells tested ( Figure 1A). As expected for KCa3.1, current activation was independent of voltage, and it reversed close to the Nernst potential for K + (-84 mV with the solutions used). The current was entirely KCa3.1 (in the presence of apamin), as demonstrated by full inhibition by the selective KCa3.1 blocker, 1 µM TRAM-34 ( Figure 1A). In all subsequent experiments, the KCa3.1 current was quantified as the TRAM-34-sensitive component.
Effects of elevating intracellular cGMP were analyzed in MLS-9 cells exposed for 20 min to the membrane-permeant cyclic GMP analog, db-cGMP, before establishing a perforated-patch recording. The KCa3.1 current was much larger after db-cGMP treatment ( Figure 1B) than the control current. Two observations provided the initial evidence for an indirect action of cGMP that requires intact cells (perhaps a diffusible mediator is lost when the cell integrity is disrupted). cGMP alone did not activate the channels (riluzole was required), and adding db-cGMP during whole-cell recordings did not increase the current (not shown).

Frontiers in Immunology | Inflammation
Therefore, we focused on perforated-patch recordings, and tested 1 µM KT5823, a membrane-permeant compound that selectively inhibits PKG (cGMP-protein kinase) with no effect on PKA (29). After treatment with db-cGMP in the presence of KT5823, only a small KCa3.1 current was activated ( Figure 1C). As summarized in Figure 1D, the current density was 13.9 ± 2.1 pA/pF (n = 9) in control cells, more than twofold larger after db-cGMP (29.0 ± 2.6 pA/pF; n = 12; p < 0.01), and 11.3 ± 1.4 pA/pF (n = 5; not different from control) when the PKG inhibitor was also added. These results indicate that the current enhancement by db-cGMP required PKG.

LACK OF DIRECT KCa3.1 ACTIVATION BY cGMP-PROTEIN KINASE (PKG) IN EXCISED PATCHES
The next question, whether PKG directly affects KCa3.1 channel activity, was prompted by our recent study in which KCa3.1 was assessed in excised inside-out patches from transfected HEK293 cells (27). PKA decreased the channel open probability (P o ), and this was abolished by mutating the S334 site to S334A, which cannot be phosphorylated (27). Here, the rationale was that S334 is also the only putative PKG phosphorylation site, and the channel could not distinguish which kinase has phosphorylated it. We hypothesized that PKG would decrease P o if it acts directly on the channel. For direct comparison with our earlier PKA study, we used the same experimental system: HEK293 cells transfected with human KCNN4 (hKCa3.1). This system offered several advantages. HEK293 cells lack endogenous KCa3.1 current but after transfection, there were more active channels than in microglia (larger whole-cell currents), which made it easier to find channels in a patch, and the current was readily activated by 1 µM Ca 2+ without requiring a gating modifier such as riluzole (27). That is, for human KCNN4 (hKCa3.1), the threshold for current activation is 100 nM Ca 2+ , the EC 50 is~270 nM, and the current is essentially fully activated at 1 µM Ca 2+ (6,7).
Inside-out patches were excised into a bath (intracellular) solution containing 100 µM ATP and free Ca 2+ concentrations of 120 nM, 500 nM, or 1 µM (Figure 2). The intracellular and extracellular solutions contained symmetrical high K + (140 mM) to set the Nernst potential to 0 mV, increase the unitary inward current amplitude, and expose the innate inward rectification of the single-channel current. One to three channels were usually active in each patch, their activity was stable for several minutes, and channel activity was recorded at -100 mV. Channel activity was quantified as NP o : the number of active channels, N, times the open probability, P o . Using representative 2 min-long segments of each recording, thresholds were set for the closed level and each open level (based on amplitude histograms; see Methods).
These results on isolated channels show that cGMP and PKG did not change the number or activity of the channels or their Ca 2+ dependence, and amplitude histograms (not illustrated) showed that the unitary current amplitude was unaffected. This is in contrast to our recent finding that activated PKA directly reduced P o by~45%, and required the channel's PKA phosphorylation site (27). The present results provide strong evidence against direct channel phosphorylation by PKG.

PKG INCREASES ROS PRODUCTION, WHICH ACTIVATES KCa3.1 CURRENT THROUGH A CaMKII-MEDIATED PATHWAY
The lack of effect of cGMP and PKG on excised patches (Figure 2) suggested that intracellular signaling was required for the current enhancement in MLS-9 cells (Figure 1). Therefore, we used perforated-patch recordings to maintain intracellular signaling when conducting experiments on native channels in MLS-9 cells and primary microglia. We first considered ROS because PKG increases ROS production in cardiac cells and neurons (28)(29)(30). Treating MLS-9 cells with db-cGMP increased ROS production by 38 ± 5% (Figure 3A; n = 5; p < 0.001), an effect that was prevented by the PKG inhibitor, KT5823. ROS can activate CaMKII (21,47) and, as described in the Introduction, CaMK regulates native KCa3.1 channels in T lymphocytes (8). Intriguingly, CaMKII can be activated by ROS without elevated intracellular Ca 2+ (21) or in a long-lasting Ca 2+ -independent manner after a transient Ca 2+ elevation (47).
Together, these data provide evidence that the enhancement in KCa3.1 currents by PKG requires both ROS and CaMKII.

APPLICATION OF H 2 O 2 INCREASES KCa3.1 CURRENT THROUGH A CaMKII-DEPENDENT PATHWAY, NOT THROUGH DIRECT ACTIVATION OF THE CHANNEL
To further analyze the ROS-mediated increase in KCa3.  (Figure 4A) that peaked at 456 ± 33 nM (n = 18) by 11.6 ± 0.1 min after treatment. The peak elevation in Ca 2+ evoked by H 2 O 2 was higher than for db-cGMP and occurred~5 min sooner. Again, it was insufficient to activate the KCa3.1 current in MLS-9 cells (n = 6 cells tested; example in Figure 4A), which requires supramicromolar concentrations (described above). In contrast, in perforated-patch recordings from cells that were pre-incubated with 1 mM H 2 O 2 , the KCa3.1 current was more than twofold larger (21.5 ± 1.6 pA/pF; n = 6; p < 0.0001) than in control cells (10.3 ± 0.5 pA/pF; n = 4; Figures 4B,C). The potentiation of the current was prevented if cells were simultaneously pre-treated with the ROS scavenger (500 µM MGP) or the CaMKII inhibitor (1 µM mAIP). The current density remained at 9.5 ± 1.4 pA/pF in MGPtreated cells and 9.2 ± 1.8 pA/pF in mAIP-treated cells. These data show that KCa3.1 function can be enhanced by this identified, stable ROS species through a similar pathway involving CaMKII. Next, inside-out patches from transfected HEK293 cells were exploited to ask whether H 2 O 2 can directly affect channel activity (NP o ; calculated as in Figure 2). There were 1-3 active channels in each patch, channel activity increased with increasing Ca 2+ (Figure 5), and the NP o values were the same as in Figure 2. Perfusing 1 mM H 2 O 2 into the bath did not affect channel activity at any of the Ca 2+ concentrations (Figure 5A, summarized in Figure 5B). At 120 nM Ca 2+ , NP o was 0.21 ± 0.05 (n = 4) and 0. 23

IN ALTERNATIVE-ACTIVATED RAT MICROGLIA, INCREASED ROS PRODUCTION POTENTIATES THE KCa3.1 CURRENT THROUGH A PKG-AND CaMKII-DEPENDENT PATHWAY
The above results on MLS-9 cells show that ROS and cGMP increase KCa3.1 current through a pathway requiring PKG and CaMKII. It was important to determine whether the same pathway regulates KCa3.1 in primary cultured microglia. For these experiments, we used alternative-activated (IL-4 treated) rat microglia, which have upregulated KCNN4 expression and a much larger KCa3.1 current than resting microglia (20), and in which KCa3.1 contributes to Ca 2+ signaling (27) and migration (20). Here, we found that at 24 and 48 h after IL-4 treatment, ROS production was increased by 93 ± 31% (p < 0.01) and 78 ± 21% (p < 0.01), respectively ( Figure 6A). The PKG inhibitor, KT5823, decreased this induced ROS production by 53% at 24 h and 73% at 48 h after IL-4 treatment, but did not significantly affect resting microglia. The enhanced ROS production in alternative-activated microglia was moderately dependent on KCa3.1 channels, as TRAM-34 reduced it by 29% at 24 h and 31% at 48 h after IL-4 treatment. Perforated-patch recordings from alternative-activated rat microglia were used to assess KCa3.1 regulation. For all recordings, the bath contained 100 nM apamin to block the KCa2.3 channels (41), KCa3.1 was activated by riluzole, and 1 µM TRAM-34 was added at the end of each recording to quantify the TRAM-34sensitive KCa3.1 component. This subtraction procedure was used because primary rat microglia have a voltage-dependent Kv1.3 current, which can be seen in control traces before riluzole was added to activate KCa3.1. Note, however, that riluzole reduces the microglial Kv1.3 current (20). As for MLS-9 cells, riluzoleactivated a KCa3.1 current in every microglia cell that was examined. Again, the current activation was not voltage-dependent; it reversed near the K + Nernst potential, and was fully blocked by 1 µM TRAM-34 ( Figure 6B). The only differences from MLS-9 cells were that the current was several-fold larger in IL-4-treated Frontiers in Immunology | Inflammation primary microglia (compare with Figure 1) and pre-treatment with db-cGMP did not further increase it. The current density was 43.7 ± 1.4 pA/pF in control microglia and 39.9 ± 1.9 pA/pF in cells pre-treated with db-cGMP, which raises the possibility that the current is already maximal in alternative-activated microglia. The key finding was that the KCa3.1 current in primary microglia was also regulated by ROS, PKG and CaMKII, as it was in MLS-9 cells. The current density was reduced from 43.7 ± 1.4 pA/pF in control microglia to 29.8 ± 2.1 pA/pF by the ROS scavenger (MPG), to 32.6 ± 2.2 pA/pF by the PKG inhibitor (KT5823), and to 23.8 ± 2.7 pA/pF by the CaMKII inhibitor (mAIP) (Figures 6C,D). Evidence for a common regulatory pathway was that combining all three inhibitors did not further reduce the current, as it would have for separate, additive pathways. Together, these results suggest that in alternativeactivated microglia, the KCa3.1 current is maximally activated through a regulation pathway involving endogenous PKG, ROS, and CaMKII. Figure 7 summarizes our results and presents a model of KCa3.1 post-translational regulation based on our observations and the literature. Elevating cGMP activates PKG, which increases ROS production, evokes Ca 2+ release from intracellular stores, which binds to CaM, and opens the KCa3.1 channel. CaM also activates CaMKII, which enhances the KCa3.1 current through an unknown mechanism. This is the first report of KCa3.1 regulation by cGMP/PKG and ROS. In comparing the present results with the literature, it is important to note several experimental considerations.

DISCUSSION
One major consideration is the patch-clamp configuration. When studying signaling pathways that regulate ion channel function, diffusible regulators can be lost or compromised in whole-cell recordings or excised patches. Therefore, we performed all recordings of native channels in the perforated-patch configuration to maintain the cytoplasmic integrity, allow intracellular Ca 2+ to "free-run," and prevent loss of soluble mediators. We applied all modulators by pre-incubation (10-60 min) before recordings were established; i.e., db-cGMP with or without KT5823 (PKG inhibitor), MPG (ROS scavenger), or mAIP (CaMKII inhibitor); H 2 O 2 with or without MPG or mAIP; or in alternative-activated (IL-4 treated) primary microglia with or without KT5823, MPG or mAIP. Results can differ in excised patches (e.g., our observed lack of effect of PKG in inside-out patches) or in whole-cell recordings. For instance, we showed that cAMP/PKA increased activity of Kv1.3 channels in intact human T lymphocytes but this regulation was lost in whole-cell recordings (49). In a pilot study using whole-cell recordings from MLS-9 cells, acute application of db-cGMP (data not shown) did not increase the KCa3.1 current, and this provided the first evidence for loss of a soluble mediator.
Another concern is whether the channels have been convincingly identified as KCa3.1. Following cloning of KCNN4 [the gene coding for KCa3.1 (5-7)], KCa3.1 currents were identified by their requirement for Ca 2+ (EC 50 200-700 nM); voltageindependent gating; reversal near the Nernst potential for K + (-85 mV with physiological internal and external K + ), an inwardrectifying single-channel conductance in symmetrical high K + solutions (~10 pS at positive potentials and 25-35 pS at very negative potentials); block by TRAM-34 (selective at ≤1 µM) and the less selective blockers, charybdotoxin (ChTx; IC 50~5 nM), and clotrimazole (IC 50 < 70 nM). Here, and in several recent papers on microglia, MLS-9 cells (20,27,31), we identified the KCa3.1 current by several of these criteria: (i) voltage-independent gating www.frontiersin.org Four papers have implicated cGMP and/or PKG in activating Ca 2+ -dependent K + channels; two published before the channel was cloned and two afterward. In the early study of vascular smooth muscle cells (50), the type of K + channel activated by Frontiers in Immunology | Inflammation Elevating intracellular cGMP activates PKG, which can then phosphorylate numerous downstream cellular targets, one of which triggers mitochondrial production of ROS (proposed to be via the "5-hydroxydecanoate-sensitive factor" that is likely the mitoK ATP channel). Intracellular ROS can then contributes to KCa3.1 regulation through its role as a signaling intermediate; e.g., by evoking Ca 2+ release from intracellular stores on the ER, leading to CaM-dependent activation of CaMKII, which then increases KCa3.1 activity (by an unknown mechanism). Activator used : 100 µM db-cGMP (membrane-permeant cGMP analog) to activate PKG. cGMP was not identified, but is unlikely to be KCa3.1. That is, the whole-cell current was depolarization-activated and blocked by 10 mM tetraethylammonium (TEA) as well as 200 nM ChTx, which is more consistent with large-conductance Ca 2+ -dependent K + (BK) channels. It is now well known that BK channels are activated by the NO-cGMP-PKG pathway [reviewed in Ref. (51)(52)(53)]. In the study on rat cortical collecting duct epithelial cells, insideout patches showed that two channel types (28 and 85 pS) were activated by cGMP and this was prevented by the PKG inhibitor, KT5823 (54). However, these recordings were made in the absence of intracellular Ca 2+ or a gating modifier, and thus neither is likely to be KCa3.1. The two later papers reported that cGMP regulates a clotrimazole-sensitive channel, and one concluded that it was KCa3.1. We reported that the IC 50 for KCa3.1 block by clotrimazole was 40 and 56 nM for native channels in human T lymphoblasts and hKCa3.1 expressed in CHO cells, respectively (8) and we subsequently used 200 nM in functional studies (14). However, clotrimazole also inhibits Ca 2+ channels (55), and at micromolar concentrations, it inhibits NMDA channels (56), TRPM2 channels (57), and several K + channels; i.e., transient outward, ultra-rapid delayed-rectifier, hERG and KCNQ1/KCNE1 channels (58). The third paper on potential KCa channel regulation by PKG showed that a K + current in interstitial cells of Cajal was activated by Ca 2+ and by the nitric oxide donor, sodium nitroprusside (SNP), blocked by 1 µM clotrimazole, and had a single-channel conductance of~38 pS (59), which are consistent with KCa3.1. However, it exhibited a steeply voltage-dependent activation at -40 mV, which is not consistent with KCa3.1. Finally, a study of singlechannel activity in human dermal fibroblasts showed that activity was increased by the NO donor, S-nitroso-N -acetylpenicillamine (SNAP) in cell-attached patches, and by applying cGMP + PKG to the cytoplasmic face of inside-out patches (60). The channel was blocked by 10 µM clotrimazole, was depolarization activated at about -40 mV, and had a large single-channel conductance (116 pS reported, >150 pS in some recordings); properties that are inconsistent with KCa3.1.
We began this study by noting that KCa3.1 has a single consensus motif for phosphorylation by PKG ( 331 RKES 334 in the human gene), but our results rule out this mechanism. First, applying PKG (with cGMP and ATP) to the cytoplasmic membrane face had no effect on channel activity and second, when S334 is phosphorylated by PKA the current is decreased, not increased (27). Failure of a kinase to phosphorylate a potential consensus motif is not unusual. For instance, PKG might have a low affinity due to conformational considerations; activated PKG (2 PKG monomers + 4 cGMP molecules) is~148 kDa compared with the catalytic subunit of PKA (43.5 kDa). Our results instead support a linear mechanism for enhancing the native KCa3.1 current: elevating cGMP (by adding membrane-permeant db-cGMP) activates PKG (inhibited by KT5823), evokes ROS production (mimicked by H 2 O 2 and inhibited by MPG), elevates intracellular Ca 2+ (but not high enough to directly activate the channel), and activates CaMKII (inhibited by mAIP). Activation of KCa3.1 by CaMKII is consistent with our earlier study in which this current in human T lymphoblasts was inhibited by the CaMK antagonist, .
Several aspects would be worth considering in future studies. (i) We do not know the mechanism by which CaMKII increases the KCa3.1 current. One possibility is that it promotes channel trafficking/insertion into the surface membrane or, conversely, reduces endocytosis. CaM is involved in assembly and trafficking of KCa3.1 (26) but there is no evidence that CaMKII is involved. In expression systems, KCa3.1 turns over rapidly through clathrin-mediated endocytosis and is then targeting for lysosomal degradation (61). (ii) CaMK-dependent changes in gene transcription can produce long-lasting cellular outcomes, such as learning and memory [reviewed in Ref. (62,63)] but we think altered gene expression is unlikely. For instance, the CaMKII www.frontiersin.org inhibitor, mAIP, was only applied for 1 h to alternative-activated microglia. (iii) In alternative-activated microglia, ROS production was increased, and dependent on both PKG signaling and KCa3.1 channels. We previously found that KCa3.1 was necessary for efficient ROS production in classical-activated microglia. Because ROS increased the KCa3.1 current this might be a positive feedback mechanism to increase KCa3.1 contributions to microglia functions under conditions of oxidative stress; e.g., after acute injuries, such as stroke. (iv) When exposed to classicalactivation stimuli (e.g., LPS) or in conjunction with phagocytosis, microglia produce ROS through an NADPH oxidase (NOX)mediated respiratory burst (64,65). While ROS play anti-microbial roles, endogenously generated ROS can potentially feed back onto physiological functions of microglia; e.g., activating NFκB and synthesis of TNFα (66). In contrast, ROS can reduce classical activation of peritoneal macrophages (67), suggesting complex roles in regulating immune cell functions. One question is whether the source matters; i.e., ROS produced by NOX versus mitochondria. We think the long-lasting ROS production in IL-4treated microglia (elevated for at least 2 days) is likely mediated by mitochondria, because PKG is known to stimulate mitochondrial ROS production in cardiomyocytes and a neuroblastoma cell line (28)(29)(30). However, the IL-4 signaling pathway increased ROS production through PI3K-dependent activation of NOX enzymes in an epithelial cell line (68); thus, we cannot rule out NOX contributions.

BROADER IMPLICATIONS
KCa3.1 is expressed in numerous cell types, including red blood cells, some immune cells [see Introduction, and reviewed in Ref. (69)], neurons (15)(16)(17), epithelia (70), vascular smooth muscle and endothelial cells (71,72), fibroblasts (73,74), stem cells (75,76), and cancer cells (77,78). Because KCa3.1 plays diverse roles in these cells (e.g., proliferation, volume regulation, migration, cytokine production, and others) its post-translational regulation by cGMP/PKG, ROS, and Ca 2+ /CaM/CaMKII is likely to have broad consequences. There are many opportunities for KCa3.1 to be regulated by cGMP/PKG signaling because this pathway can be activated through NO produced by nNOS (neurons), eNOS (endothelial cells) and iNOS (innate immune cells), and acting on soluble guanylate cyclase. In microglia, iNOS is upregulated and NO is produced in the classical-activated state. Proinflammatory mediators are induced by LPS, and there is evidence that atrial natriuretic peptide (ANP)-induced cGMP/PKG signaling can reduce this (79). In microglia, the roles of PKG are not well known but we found that PKG is involved in ROS production in alternative-activated microglia. Further evidence for involvement of PKG and ROS in alternative activation is that PKG increased IL-4 production by Th2 lymphocytes (80), and ROS increased IL-4 release by macrophages (81,82). The widespread coincidence of this channel and signaling pathway is reinforces the importance of future studies addressing its regulation in other cell types.

AUTHOR CONTRIBUTIONS
LS, RF, and RW contributed to the conception, design of this study, and wrote the manuscript. RF and RW performed the experiments. LS, RF, and RW agree to be accountable for all aspects of the work.