(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



The NKCC1 ion transporter modulates microglial phenotype and inflammatory response to brain injury in a cell-autonomous manner

['Krisztina Tóth', 'Momentum Laboratory Of Neuroimmunology', 'Institute Of Experimental Medicine', 'Budapest', 'János Szentágothai Doctoral School Of Neurosciences', 'Semmelweis University', 'Nikolett Lénárt', 'Péter Berki', 'Laboratory Of Cerebral Cortex Research', 'Rebeka Fekete']

Date: 2022-02 

The NKCC1 ion transporter contributes to the pathophysiology of common neurological disorders, but its function in microglia, the main inflammatory cells of the brain, has remained unclear to date. Therefore, we generated a novel transgenic mouse line in which microglial NKCC1 was deleted. We show that microglial NKCC1 shapes both baseline and reactive microglia morphology, process recruitment to the site of injury, and adaptation to changes in cellular volume in a cell-autonomous manner via regulating membrane conductance. In addition, microglial NKCC1 deficiency results in NLRP3 inflammasome priming and increased production of interleukin-1β (IL-1β), rendering microglia prone to exaggerated inflammatory responses. In line with this, central (intracortical) administration of the NKCC1 blocker, bumetanide, potentiated intracortical lipopolysaccharide (LPS)-induced cytokine levels. In contrast, systemic bumetanide application decreased inflammation in the brain. Microglial NKCC1 KO animals exposed to experimental stroke showed significantly increased brain injury, inflammation, cerebral edema, and, worse, neurological outcome. Thus, NKCC1 emerges as an important player in controlling microglial ion homeostasis and inflammatory responses through which microglia modulate brain injury. The contribution of microglia to central NKCC1 actions is likely to be relevant for common neurological disorders.

Funding: This work was supported by „Momentum” research grant from the Hungarian Academy of Sciences (LP2016-4/2016 to A.D.; https://mta.hu/lendulet ) and ERC-CoG 724994 (to A.D.; https://erc.europa.eu/ ), with contribution from 2019-2.1.7-ERA-NET-2020-00004 ( https://nkfih.gov.hu ). Additionally, this work was funded by Hungarian National Scientific Research Fund (NKFIH-OTKA Grant No. K131844 to S.B.), the János Bolyai Research Scholarship of the Hungarian Academy of Sciences (to C.C. and N.L., BO/00558/19/5; https://mta.hu/bolyai-osztondij/bolyai-janos-kutatasi-osztondij-105319 ), ÚNKP-20-3-II (to B.P.) and ÚNKP-21-5 (to C.C. and N.L.; http://www.unkp.gov.hu/unkp-rol ) of the New National Excellence Program of the Ministry for Innovation and Technology, Hungary; German Research Foundation (SPP 1665; https://www.dfg.de/en/ ) and the Federal Ministry of Education and Research (NEURON ACRoBAT 01EW1706) to C.A.H.; and the Academy of Finland and Sigrid Jusélius Foundation to K.K.. Prepared with the professional support of the Doctoral Scholarship Program of the Cooperative Doctoral Program of the Ministry of Innovation and Technology Financed from The National Research, Development and Innovation Fund (to B.P.). A.A. holds a Stipendium Hungaricum Scholarship from the Government of Hungary ( https://stipendiumhungaricum.hu/ ). K.K.’s work was supported by the Academy of Finland and by the Sigrid Jusélius Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

In the healthy brain, microglia regulate a wide variety of neuronal functions, whereas altered microglial activity is linked with the pathophysiology of most common brain disorders, such as neurodegenerative and psychiatric diseases, stroke, and epilepsy [ 27 – 33 ]. Extracellular accumulation of potassium is related to diverse neuropathological alterations [ 32 , 34 ], and the contribution of microglial potassium channels and transporters to microglial activity is widely recognized. These include membrane-expressed potassium channels (Kv1.3, THIK-1, Kir2.1) that regulate microglial motility, immune surveillance, and cytokine release, among others [ 35 , 36 ]. Moreover, changes in intracellular potassium and chloride levels are associated with inflammasome activation contributing to the regulation of interleukin-1β (IL-1β) release [ 37 – 39 ]. Interestingly, recent transcriptomic data verified high-level NKCC1 expression in microglia [ 21 , 40 ]. However, currently no experimental data are available on the function of microglial NKCC1 under physiological and pathological conditions. To this end, we studied whether central and systemic NKCC1 blockade impact on central inflammatory changes differently and tested the hypothesis that NKCC1 is involved in the regulation of microglial inflammatory mediator production, neuroinflammation, and brain injury.

Interestingly, common neurological disorders including those with altered NKCC1 activity [ 6 , 7 , 22 ] display broad neuroinflammatory changes. While the impact of NKCC1 function on inflammatory cytokine production has not been widely studied, emerging evidence indicates up-regulated NKCC1 expression in response to inflammatory stimuli (LPS, IL-1β, TNF-α) [ 23 – 26 ]. Notably, while microglia are the main inflammatory cell type in the CNS, there is no information about the functional role of NKCC1 in microglia and whether microglial NKCC1 could provide a significant contribution to central NKCC1 actions under inflammatory conditions or after brain injury.

While bumetanide is routinely used to block neuronal NKCC1 in experiments performed in vitro [ 3 , 8 , 12 ], the view that the drug might have direct effects on central neurons in vivo has raised numerous questions. First, the poor pharmacokinetic properties of bumetanide, including a low penetration across the blood–brain barrier (BBB) implies that when applied systemically at clinically relevant doses, the drug may not reach pharmacologically relevant concentrations in the brain parenchyma, a prediction that has been experimentally verified [ 10 , 13 , 14 ]. Another important point in the present context is that NKCC1 is expressed within the immature and mature CNS mainly in nonneuronal cell types [ 15 ], such as oligodendrocytes and their precursors [ 16 , 17 ], ependymal cells, and astrocytes [ 18 – 21 ]. Thus, even if applied directly into brain tissue, bumetanide or any other NKCC1 blockers would be expected to inhibit this ion transporter in a wide variety of cell types.

Members of the plasmalemmal cation-chloride cotransporter (CCC) family, such as the neuron-specific K + /Cl − extruder, KCC2, and the ubiquitously expressed Na + -K + -2Cl − cotransporter, NKCC1 (coded by the Slc12a2 gene), have received a steeply increasing amount of attention in research on central nervous system (CNS) diseases, ranging from neuropsychiatric diseases to epilepsy, stroke, and dementia [ 1 – 6 ]. Notably, there is an abundance of studies, which have shown that the mRNA and protein expression levels as well as the functionality of NKCC1 are enhanced in injured and posttraumatic neurons [ 7 , 8 ]. This, in turn, has raised the obvious possibility that assuming a pathophysiological role for NKCC1, therapeutic actions might be achieved by pharmacological inhibition of this transporter. Indeed, data in a number of experimental studies based on systemic application of the selective NKCC1 blocker, the loop diuretic bumetanide, have suggested that this drug has therapeutic actions in diverse neuropathological conditions. These include neonatal seizures, temporal lobe epilepsy, autism spectrum disorders, schizophrenia, and brain edema after traumatic or hypoxic/ischemic injury [ 1 – 4 , 7 – 12 ]. In most of these studies, bumetanide has been suggested to exert its therapeutic effects by acting directly on NKCC1 expressed in central neurons.

(A, B) Microglial NKCC1-deficient mice (KO) show larger infarct volume as assessed on cresyl violet-stained brain sections and more severe neurological outcome compared to WT mice. ( C, D) Microglial NKCC1 deletion results in higher levels of IL-1α and IL-1β 24 hours after MCAo. (E) Cytokine levels in the cortex do not differ 8 hours after MCAo in KO mice compared to WT. (A) Scale: 1 mm. (B) Unpaired t test, N (WT) = 7, N (KO) = 9; *: p < 0.05, **: p < 0.01. (C) Mann–Whitney test, *: p < 0.05; N (WT) = 7, N (KO) = 8. ( D) Scale: 50 μm. ( E) Kruskall–Wallis test followed by Dunn’s multiple comparison test; N = 6/group. *: p < 0.05, **: p < 0.01. Data underlying this figure can be found in S1 Data . KO, knockout; MCAo, middle cerebral artery occlusion; ns, not significant; WT, wild type.

Because Na + -coupled Cl − importers and their upstream regulatory serine–threonine kinases (WNK-SPAK-OSR1) are involved in maintaining intracellular ionic homeostasis as well as regulation of cell volume [ 2 , 49 ], inhibiting these cotransporters is a subject of interest in ischemic stroke therapy and in other forms of acute brain injury (brain trauma, SAH) where cerebral edema is a major contributor to poor clinical outcome [ 2 , 49 ]. Thus, we investigated whether microglial NKCC1 deficiency influences the severity of brain injury after experimental stroke in microglial NKCC1 KO mice subjected to transient, 45-minute long middle cerebral artery occlusion (MCAo). Twenty-four hours after reperfusion, elimination of microglial NKCC1 was not associated with obvious alterations in astroglial GFAP levels or AQP4 levels in perivascular astrocyte endfeet ( S6 Fig ). However, we observed a significant increase in brain edema, which was accompanied by a 47% increase in lesion volume ( Fig 6A and 6B ; p = 0.0262) and poor neurological outcome in NKCC1 KO mice ( Fig 6B ; Bederson’s score: WT: 1.2 ± 0.2, KO: 2.1 ± 0.2; composite neurological score: WT: 14.4 ± 1.5, KO: 20.5 ± 1.5). Furthermore, experimental stroke induced a 4-fold up-regulation of IL-1α and 4.8-fold increase in IL-1β expressing NKCC1 KO microglia cells compared to WT animals, while the overall microglia density and the number of activated microglia were not different at 24-hour reperfusion ( Fig 6C and 6D ). We also assessed whether increases in IL-1 production take place early after brain injury before the majority of the infarct is formed. Cytometric bead array did not reveal altered cytokine production in cortical homogenates shortly (8 hours) after MCAo ( Fig 6E ). Taken together, these results indicate that microglial NKCC1 deficiency results in augmented brain inflammation, brain edema, and increased brain injury after experimental stroke.

To investigate the potential compensatory mechanisms that may arise in response to NKCC1 deletion in microglial cells, we evaluated the expression of some relevant genes that are reportedly expressed by microglia and are known to regulate ion concentrations, membrane potential or response to osmotic stress ( S5 Fig ; [ 57 , 58 ]). Slc9a1, Slc8a1, Slc12a6, Clic1, Clcn3, Kcnk13, Kcnk6, Kcnj2, Kcna3, and Sgk1 mRNAs did not show altered expression between WT and NKCC1 KO microglia ( S5A and S5C Fig ), but we found a more than 2-fold increase in Lrrc8d, the D subunit of VRAC in response to NKCC1 deletion ( S5A Fig ). Importantly, Lrrc8d (VRAC) is involved in increasing the permeability for a broad range of uncharged organic osmolytes [ 59 ]. Interestingly, intracisternal LPS treatment resulted in a marked reduction of NKCC1 and P2Y12R mRNA levels ( Fig 1E ), in line with the down-regulation of Slc9a1, Slc8a1, Lrrc8d, Clic1, Clcn3, Kcnk13, Kcnk6, Kcnj2, Sgk1 genes ( S5B Fig ) of which Slc9a1, Clic1, Kcna3, Kcnj2, Kcnk13 genes are known to play a role in maintaining ramified morphology and resting state under physiological conditions [ 35 , 39 , 58 , 60 ].

(A) Schematic representation of the experiment. Perforated patch-clamp recordings were performed on isolectin B4–labeled microglial cells in acute hippocampal slice preparations. Current responses to a train of voltage steps from −140 to 60 mV with 20 mV increments and a duration of 100 ms were measured in voltage-clamp mode (holding potential: −40 mV) both in normotonic, and after 5-minute perfusion with hypotonic ACSF (50% dilution). ( B) Example traces of recordings from WT (black: normotonic, gray: hypotonic ACSF) and NKCC1 KO (purple: normotonic, violet: hypotonic ACSF) animal. Traces show responses at −140 and +60 mV voltage steps in both conditions. ( C) Average I-V curve responses from WT in normotonic (N = 11 cells, black with SEM) and in hypotonic (N = 8 cells, gray with SEM) condition. ( D) Average I-V curves from NKCC1 KO in normotonic (N = 11 cells, purple with SEM) and in hypotonic (N = 8 cells, violet with SEM) condition. ( E) Input resistance of WT versus KO cells in normotonic condition. ( F) Input resistance of WT versus KO cells in hypotonic condition. ( G) Resting membrane potential in normotonic condition of WT versus NKCC1 KO microglial cells measured in current-clamp mode (0 pA injected current). ( E) Mann–Whitney; N (WT) = 11 cells from 8 animals, N (KO) = 11 cells from 7 animals; *: p < 0.05. ( F) Mann–Whitney; N (WT) = 8 cells from 6 animals, N (KO) = 8 cells from 5 animals; *: p < 0.05. ( G) Mann–Whitney; N (WT) = 10 cells from 8 animals, N (KO) = 11 cells from 7 animals. Data underlying this figure can be found in S1 Data . KO, knockout; ns, not significant; WT, wild type.

To study the consequences of NKCC1 deletion on microglial ion regulation and membrane currents, we made perforated patch-clamp recordings on native microglia in acute hippocampal slice preparations from WT and microglial NKCC1 KO mice. With this method, the intracellular Cl − concentration remains unaffected by the pipette solution. Recordings were done in voltage-clamp mode at a holding potential of −40 mV, while voltage steps (−140 mV to +60 mV with 20 mV increments and 100 ms duration) were delivered to construct I-V plots for individual cells ( Fig 5A ). An exposure to hypotonic ACSF solution (50% dilution, 5 minutes) was used to identify the outwardly rectifying current in response to cellular swelling, which is considered to be mainly mediated by the largely Cl − permeable volume-regulated anion channels (VRACs) [ 53 – 55 ]. Our results showed that current responses under normotonic and hypotonic conditions ( Fig 5B ) were qualitatively similar to those published previously regarding WT microglia in acute slice preparations [ 56 ]. I-V curves ( Fig 5C and 5D ) indicated a much higher resistance of the KO versus WT cells, especially at more positive voltage levels under both normotonia and hypotonia. The slope of I-V curves between the −40 mV and 0 mV voltage steps ( Fig 5E and 5F ) showed significantly higher input resistance of KO microglia both in normotonia (WT: 1.91 GΩ, q1: 1.71, q3: 2.38 versus KO: 3.00 GΩ, q1: 2.20, q3: 4.50; Mann–Whitney, p < 0.05) and hypotonia (WT: 0.78 GΩ, q1: 0.66, q3: 1.11 versus KO: 2.47 GΩ, q1: 0.94, q3: 4.25; Mann–Whitney, p < 0.05). We also measured resting membrane potential of microglia ( Fig 5G ) where no statistically significant difference was observed between WT and KO (WT: −20 mV, q1: −21, q3: −18 versus KO: −19 mV, q1: −26, q3: −17; Mann–Whitney, not significant). In order to compare the responses of WT and KO microglial cells to cell swelling, we isolated the swelling-induced current component by subtracting the currents obtained at each point of voltage in normotonic conditions from those recorded during the hypotonic response ( S4A Fig ). The resulting I-V plot of the swelling-induced currents showed a reversal potential (E swell −a rough estimate of the Cl − equilibrium potential, E Cl- ), which was more negative in the KO cells (WT: −47 mV, q1: −53, q3: −38 versus KO: −68 mV, q1: −91, q3: −46; Mann–Whitney, p < 0.05; S4B Fig ). Because of the variation of the individual data points related to reversal and resting potentials, we calculated the driving force (DF = V rest − E swell ) for each WT and KO cell ( S4B Fig ), which provides a more direct parameter of the efficacy of ion regulation than ionic reversal potentials [ 1 ]. We found significantly higher DF values in the case of NKCC1 KO versus WT microglia (WT: 28 mV, q1: 21, q3: 33 versus KO: 50 mV, q1: 30, q3: 65; Mann–Whitney, p < 0.05), supporting the view that the lack of microglial NKCC1 results in a lower intracellular Cl − concentration.

To test the impact of microglial NKCC1 deletion on the production of central inflammatory mediators, we examined LPS-induced cytokines and chemokines in microglial NKCC1 KO and WT mice 24 hours after intracortical LPS injections in brain homogenates ( Fig 4E ). Importantly, deletion of microglial NKCC1 markedly potentiated LPS-induced levels of G-CSF (3.53-fold), KC (2.12-fold), IL-1α (3.98-fold), IL-1β (3.96-fold), IL-6 (4.35-fold), and TNF-α (3-fold) in the cerebral cortex compared to controls ( Fig 4E ). In contrast, intracortical LPS treatment did not alter cytokine levels or T cell, B cell, monocyte, or granulocyte numbers in the spleen or liver ( S3A–S3C Fig ). Similarly to that seen after intracortical pharmacological inhibition of NKCC1 by bumetanide, neither the number of CD11b + CD45 int microglial cells nor blood-borne CD11b + CD45 high leukocytes, CD11b + Ly6C high monocytes, or CD11b + Ly6G high granulocytes were altered as a result of microglial NKCC1 deficiency ( Fig 4F ). We also tested whether increased baseline IL-1β and NLRP3 mRNA levels seen in NKCC1 KO microglia ( Fig 4A ) were maintained in these cells 24 hours after intracisternal LPS administration. We found that significantly increased NLRP3 mRNA levels (by 100%) were still present in NKCC1 KO microglia isolated from the brain by using magnetic separation, while a strong trend (p = 0.057) to increased IL-1β mRNA by 50% was also observed ( Fig 4G ). These findings corroborated our previous results on cortical NKCC1 blockade by bumetanide, confirming that the inhibition of microglial NKCC1 markedly potentiates central cytokine production upon inflammatory challenges, with a likely role for the NLRP3 inflammasome in the excessive production of IL-1β.

(A) Baseline NLRP3 and IL-1β mRNA expression is increased in isolated NKCC1 KO microglia compared to WT cells. ( B) Experimental outline of automated morphological analysis, cytokine array, and flow cytometry. (C, D) Automated morphological analysis shows that activated NKCC1-deficient microglia are slightly smaller than their WT counterparts. (E) LPS-induced cytokine levels are significantly higher in the cortices of microglial NKCC1 KO mice than in WT. ( F) Flow cytometric dot plots show that microglial NKCC1 deficiency does not alter the number of CD11b + , CD45 int microglia (P4 gate) or numbers of infiltrating CD11b + , CD45 high leukocytes (P5 gate), CD11b + , Ly6C high monocytes (P9 gate) and CD11b + , Ly6G high granulocytes (P7 gate) in response to intracortical LPS administration. See corresponding data on peripheral cytokine levels and immune cell populations in S3 Fig . (G) Increased NLRP3 and IL-1β mRNA levels are sustained in NKCC1 KO and WT microglia 24 hours after intracisternal LPS administration . (A) Unpaired t test; *p < 0.05, **p < 0.01; N (WT) = 6, N (KO) = 5. ( C) Scale: 25 μm. ( D) Mann–Whitney test, N (WT) = 108 cells from 6 mice, N (KO) = 92 cells from 4 mice, **p < 0.01, ***p < 0.001. ( E) Unpaired t test, *: p < 0.05; N (WT) = 7, N (KO) = 6. (F) Unpaired t test, N (WT) = 4, N (KO) = 4. (G) Unpaired t test; *p < 0.05; N (WT + LPS) = 5, N (KO + LPS) = 5. Data underlying this figure can be found in S1 Data . IL-1β, interleukin-1β; KO, knockout; LPS, lipopolysaccharide; ns, not significant; WT, wild type.

To assess the effect of NKCC1 deletion on microglial inflammatory states and responses, we first examined the expression of IL-1β and NLRP3. Assembly of the NLRP3 inflammasome is known to be a key step for processing of pro-IL-1β by caspase-1, and the release of mature IL-1β from microglia or macrophages [ 50 , 51 ]. qPCR data revealed elevated expression of both IL-1β and NLRP3 in NKCC1 KO microglia even in the absence of inflammatory stimulus ( Fig 4A ). Next, we examined the impact of NKCC1 deficiency on morphological characteristics of microglia after activation by endotoxin that primes microglial inflammatory responses and leads to morphological transformation [ 52 ]. Unbiased, automated morphological analysis showed that both WT and NKCC1 KO microglia respond to intracortical LPS administration with dramatic morphological conversion ( Fig 4B and 4C ). However, NKCC1-deficient microglia were significantly smaller than their WT counterparts, displaying a 10% smaller cell surface, 12.5% smaller cell volume, and 13.7% smaller branch volume ( Fig 4D ). According to these data, microglial NKCC1 is likely to be involved in shaping cell morphology and regulation of cell volume after inflammatory challenges.

(A) Outline of the 2-photon imaging experiment performed in Cx3CR1 +/GFP (microglia reporter) mice. Imaging was performed to assess baseline microglial process motility and response to laser-induced injury followed by Bum administration into the cisterna magna and identical measurements at a different, undisturbed cortical region. ( B) Representative images of microglial responses at selected time points, taken from S1 Video . The lesion site is marked with a circle that completely covers the area of the lesion. The ablation zone (i.e., the inner circle) was excluded from the analysis (see Materials and methods ). ( C) Mean base-state velocity distribution of processes from manual image tracking. Data show a 7% higher mean velocity of cell processes after Bum administration. ( D) Evaluation of microglial process coverage over time in the lesion-centered circular area generated by a custom image analysis pipeline. Solid lines show the proportion of the area covered by microglial GFP signal. Dashed curves show the calculated values of coverage from the best-fitting curves. Horizontal lines are the predicted maximal coverage values for the lesion site; vertical line is the time point of the lesioning. ( E) Calculated postlesion velocities of microglial process recruitment using automated image analysis followed by model fitting. Data show a significant decrease (31.7%) in mean frontline velocity of cells with Bum administration. ( B) Scale: 50 μm. ( C) Mann–Whitney test, N Control = 73, N Bum = 76 processes from 5 mice, **: p < 0.01. ( E) Mann–Whitney test, N Control = 40, N Bum = 56 fitted values from 5 and 7 mice, respectively, **: p < 0.01. Data underlying this figure can be found in S1 Data . Bum, bumetanide; GFP, green fluorescent protein.

To assess whether NKCC1 is involved in the regulation of dynamic microglial process surveillance and injury-induced microglial responses, we performed in vivo 2-photon imaging in microglia reporter Cx3CR1 +/GFP mice in combination with bumetanide treatment ( Fig 3 ). We tracked prelesion (baseline) process motility followed by focal laser-induced lesioning, which was repeated after bumetanide administration into the cisterna magna in a different (undisturbed) part of the cerebral cortex in the same animals ( Fig 3A and 3B ). The cisterna magna application protocol had been extensively tested in previous studies for effective drug penetration into the brain parenchyma and direct actions on microglia by using PSB0739, a selective P2Y12R antagonist [ 31 ]. Bumetanide caused a small but observable (7%) increase in the mean velocity of microglial processes ( Fig 3C ). In contrast, lesion-induced recruitment of microglial processes showed a marked, 31.7% reduction after bumetanide treatment (N = 7) ( Fig 3D and 3E and see S1 Video ). These experiments indicate that beyond shaping cell morphology, microglial NKCC1 also regulates dynamic microglial actions both under physiological and pathological conditions.

Microglia are known to respond to physiological and pathological challenges with fast morphological transformation [ 47 ], while the role of NKCC1 in cell volume regulation has been highlighted in previous studies concerning other cell types [ 48 , 49 ]. To study the role of NKCC1 in microglial cell shape, we performed automated morphological analysis using perfusion-fixed brain sections obtained from WT and microglial NKCC1 KO mice ( Fig 2F and 2G ). First, we aimed to test whether microglial NKCC1 deficiency has any impact on baseline cell morphology. Our data revealed that NKCC1-deficient microglia displayed a 23.8% higher cell surface, 19% higher branch volume, and 33.5% more branches compared to WT cells ( Fig 2F and 2G ), while we observed no changes in the cell body volume of KO versus WT microglia ( Fig 2G ). According to these data, microglial NKCC1 is likely to be involved in shaping baseline cell morphology.

To study the functional role of NKCC1 in microglia in vivo, we generated a novel transgenic mouse line in which microglial NKCC1 was deleted ( Fig 2B ). Slc12a2 expression in isolated microglia derived from tamoxifen-inducible microglial NKCC1 knockout (KO) mice was markedly reduced in comparison with wild-type (WT) littermates (WT: 1.46 ± 0.11, KO: 0.50 ± 0.016; Fig 2C ) as assessed by qPCR. Reduced NKCC1 protein expression was also confirmed by unbiased fluorescent density measurements after systematic random sampling of a great number of individual microglial cells from Iba1/NKCC1 stained slices. In fact, a discrete NKCC1 labeling was seen in the plasma membrane of microglial cells in WT mice, which was absent in microglial NKCC1 KO mice ( Fig 2D and 2E ).

(A) NKCC1 mRNA expression levels in newborn and adult microglia isolated from C57BL/6J mice compared to neural progenitors derived from E17 embryonic hippocampi. Note that NKCC1 mRNA levels decrease dramatically during in vitro maintenance (DIV10). ( B) We generated a novel microglia-specific conditional NKCC1 KO transgenic mouse line by crossing NKCC1 fl/fl (exon 8 of the Slc12a2 gene was flanked with lox P sites) and Cx3CR1-Cre ERT2 mice. ( C) NKCC1 mRNA levels in isolated NKCC1 KO microglia was markedly reduced in comparison to WT cells. ( D, E) NKCC1 protein expression in a large number of randomly sampled NKCC1 KO microglia cells is markedly reduced compared to WT cells. Inserts show plasma membrane localization of NKCC1. ( F, G) Automated morphological analysis and maximum intensity projections of confocal images. Inserts show cells marked with white asterisks in 3D. Arrowheads indicate altered branch structure of NKCC1 KO microglia. Automated morphological analysis shows that features of NKCC1-deficient microglia significantly differ from WT microglia. (A) One-way ANOVA, followed by Holm–Sidak’s post hoc test. N (Neuronal progenitor) = 3, N (Newborn microglia DIV0) = 7, N (Newborn microglia DIV10) = 6, N (Adult microglia DIV0) = 5, N (Adult microglia DIV10) = 6. **: p < 0.01. ( C) Unpaired t test, N = 3/group. **: p < 0.01. ( D) Mann–Whitney test, N (WT) = 142 cells from 2 mice, N (KO) = 83 cells from 1 mouse. ****: p < 0.0001. ( E) Scale: 2 μm, in inserts: 1 μm. ( F, G) Scale: 25 μm; Mann–Whitney test, N (WT) = 162 cells from 3 mice, N (KO) = 281 cells from 5 mice. **: p < 0.01, ***: p < 0.001, ****: p < 0.0001. Data underlying this figure can be found in S1 Data . DIV, days in vitro; KO, knockout; ns, not significant; TMX, tamoxifen; WT, wild type.

Available RNA-seq data as well as our results ( Fig 1E ) demonstrate that the Slc12a2 gene is expressed by microglia [ 21 , 40 ]. In line with this, our further qPCR data revealed that mRNA levels of NKCC1 in microglial cells isolated from newborn (1.052 ± 0.054) or adult (0.89 ± 0.098) C57BL/6J mice are comparable to those in neural progenitors derived from embryonic brains (1.30 ± 0.05; Fig 2A ). However, NKCC1 mRNA expression markedly decreased when cells were grown in culture for 10 days ( Fig 2A ), making it difficult to study NKCC1 function in microglia using in vitro techniques (see also [ 58 ]).

Microglia are the primary source of inflammatory cytokines in a number of neuropathologies, and IL-1β, also produced by microglia, is a key proinflammatory cytokine regulating the brain’s cytokine network [ 42 , 43 ]. To investigate the cellular source of IL-1β in this experimental model, we injected LPS intracortically into Cx3CR1 +/GFP (microglia reporter) mice. Multilabel immunostainings confirmed that Cx3CR1-GFP microglia coexpressing P2Y12R and CD45 displayed cytoplasmic IL-1α and IL-1β production ( Fig 1D ). These findings collectively suggested that microglia are likely to be major mediators of central bumetanide actions under inflammatory conditions. Supporting this, central LPS administration resulted in a significant decrease of NKCC1 (encoded by Slc12a2) mRNA levels, along with a decline of P2Y12R mRNA, a key purinergic receptor reportedly associated with homeostatic microglial actions, microglial activation, and response to injury ( Fig 1E ) [ 31 , 44 – 46 ].

We next tested whether the effects of central bumetanide administration on increasing cytokine production might be explained by altered LPS-induced recruitment of leukocytes. However, while flow cytometry revealed increased number of infiltrating CD45 high leukocytes, in particular CD11b + Ly6G high granulocytes, and CD11b + Ly6G − Ly6C high monocytes in response to central LPS injection compared to vehicle administration ( Fig 1C ), this was not altered by central bumetanide. The number of CD45 int CD11b + microglia was not affected either ( Fig 1C ), indicating that the increased LPS-induced cytokine production observed upon central bumetanide treatment is not due to altered inflammatory cell numbers in the cerebral cortex. Unbiased densitometric analysis revealed that GFAP and AQP4 immunopositivity 24 hours after intracortical LPS injection were not altered by central bumetanide treatment ( S2 Fig ), suggesting that while bumetanide may also act on astroglial NKCC1 [ 18 – 20 ], marked changes in astrocyte phenotypes or perivascular astrocyte endfeet are unlikely to explain the effect of bumetanide on central inflammatory responses in the present study.

To compare the effects of systemic versus central NKCC1 blockade by bumetanide, we next investigated the impact of central (intracortical) bumetanide administration on LPS-induced cytokine responses, by coinjecting bumetanide with LPS (50 μM bumetanide in 200 nl final volume, rate: 200 nl/10 minutes; Fig 1B ) into the cerebral cortex. As expected, intracortical LPS without bumetanide resulted in a 20- to 50-fold increase in inflammatory cytokines/chemokines in the cerebral cortex compared to vehicle. Surprisingly, on top of this, bumetanide markedly potentiated LPS-induced G-CSF, KC, IL-1β, and IL-1α levels in the brain (by 86.1%, 31.2%, 82.5%, and 72.4%, respectively), in sharp contrast with the effects of ip. bumetanide treatment ( Fig 1A ). Intracortical bumetanide had no effect on systemic cytokine levels ( S1B Fig ).

(A) Mice were subjected to either ip. or cor. LPS injections, while NKCC1 was blocked by ip. Bum administration. Central LPS injection triggers high cytokine (G-CSF, IL-1α, IL-1β) and KC responses in the brain compared to ip. LPS injection, which is blocked by ip. Bum administration. ( B) Central NKCC1 inhibition by cor. Bum administration significantly increases GCSF and IL-1β levels. See also S1 Fig for effects of systemic vs. central blockade of NKCC1 on LPS-induced cytokine responses in the periphery. ( C) Flow cytometric dot plots show that cortical administration of Bum does not affect the number of microglia (CD45 int /P5 gate), and recruitment of leukocytes (CD45 high /P4 gate), including monocytes (CD11b + , Ly6C high /P9 gate), and granulocytes (CD11b + , Ly6G high /P7 gate) upon central LPS injection. ( D) The main source of IL-1α and IL-1β in the brain are microglia cells. Confocal images of Cx3CR1 +/GFP brain slices show IL-1α-CD45-P2Y12R (above, red arrowheads) and IL-1β-CD45-P2Y12R (below, red arrowheads) labeled cells after cortical LPS injection-induced inflammation. All data are expressed as mean ± SEM. ( E) NKCC1 (encoded by Slc12a2) and P2Y12R gene expression is down-regulated in microglia isolated from adult mice 24 hours after cisterna magna LPS application. ( A) Kruskall–Wallis followed by Dunn’s multiple comparison test; *p < 0.05; N (veh.) = 5, N (veh. + ip. LPS) = 5, N (Bum + ip. LPS) = 5, N (veh. + cor. LPS) = 5, N (Bum + cor. LPS) = 9. ( B) Unpaired t test; *p < 0.05; N = 9/group; data were pooled from 2 independent studies. ( C) One-way ANOVA followed by Tukey’s multiple comparison test; *p < 0.05; N (veh.) = 5, N (cor. LPS) = 6, N (cor. Bum + LPS) = 6. ( D) Scale: 25 μm. ( E) Unpaired t test; **p < 0.01, ***p < 0.001; N (WT) = 6, N (WT + LPS) = 5. Data underlying this figure can be found in S1 Data . Bum, bumetanide; cor., cortical; ip., intraperitoneal; KC, chemokine; LPS, lipopolysaccharide; ns, not significant; veh., vehicle; WT, wild type.

To investigate whether systemic blockade of NKCC1 actions by bumetanide could alter inflammatory responses in the CNS, we injected mice either intraperitoneally or intracortically with bacterial lipopolysaccharide (LPS), while NKCC1 actions were blocked by systemic (intraperitoneal (ip.), 2 mg/kg) bumetanide administration ( Fig 1A ). As expected, LPS administration by ip. injection triggered marked inflammation in the spleen and the liver ( S1A Fig ), but caused low cytokine production in cortical brain tissues, which was not influenced by ip. bumetanide treatment ( Fig 1A ). However, bumetanide resulted in a small but significant increase in LPS-induced IL-1β production in the spleen ( S1A Fig ). In contrast, intracortical LPS administration triggered a robust inflammatory response in the brain as seen earlier [ 41 ], which was reduced by ip. bumetanide treatment. This was demonstrated by lower G-CSF, KC, IL-1β, and IL-1α levels in the brain (by 39.8%, 43%, 54.6%, and 41%, respectively), while systemic cytokine levels were not altered ( S1A Fig ).

Discussion

To our knowledge, we show for the first time that NKCC1 is functionally active in microglia in the adult mouse brain and that microglial NKCC1 is involved in shaping both baseline and reactive microglia morphology and responses. We present evidence that NKCC1 regulates microglial membrane conductance and adaptation to swelling-induced volume changes in a cell-autonomous manner, while the lack of NKCC1 results in primed microglial inflammatory states as evidenced by elevated NLRP3 and IL-1β levels. In line with this, while systemic NKCC1 blockade attenuates LPS-induced inflammation in the brain, central pharmacological inhibition or genetic deletion of microglial NKCC1 potentiates inflammatory cytokine production. Microglial NKCC1 deletion also increases brain injury, inflammation, and cerebral edema and leads to worse neurological outcome after acute brain injury induced by experimental stroke. Thus, microglial NKCC1 emerges as an important regulator of central inflammatory responses.

As pointed out in the Introduction, there is little direct information on whether systemically applied bumetanide might exert therapeutic effects by directly blocking NKCC1 in the brain and, particularly, in central neurons. To unravel the effects of pharmacological inhibition of NKCC1 in the periphery and the brain, we investigated the systemic versus central effects of bumetanide in response to bacterial endotoxin (LPS) administration, which is widely used to trigger inflammatory cytokine responses [41,52]. Intraperitoneally injected bumetanide reduced intracortical LPS-induced proinflammatory cytokine/chemokine responses in the brain, similarly to that found in the case of LPS-induced lung injury attributed to NKCC1-mediated macrophage activation [61]. These findings are also consistent with the suggested beneficial therapeutic effects of systemic NKCC1 blockade in CNS pathologies, which are associated with inflammation [1,7,8,10–12,62].

Importantly, intracortical administration of bumetanide had exactly the opposite effect on intracortical LPS-induced cytokine production to that seen after systemic bumetanide treatment, resulting in elevated cytokine levels in the cortex. These markedly different outcomes clearly demonstrate the difficulty to interpret pharmacological interventions in the absence of data on the precise cellular targets of NKCC1 actions in the brain and highlight the need for cell-specific NKCC1 targeting studies.

Because microglia are the key source of inflammatory cytokines in the brain, we tested whether microglial activity could be directly related to NKCC1 function. Indeed, we observed comparable Slc12a2 expression in microglia, isolated from either newborn or adult mouse brains, to that seen in neural progenitors. We also revealed NKCC1 protein in cortical microglial cells, which is in line with recent observations showing microglial NKCC1 expression in the superficial spinal dorsal horn in rats [63], and also supported by single-cell transcriptomic studies [21,40]. Further on, we demonstrated that NKCC1 levels are markedly down-regulated in microglia in response to LPS, alongside with decreased P2Y12R levels (a purinergic receptor regulating microglial cell–cell interactions under both physiological conditions and in response to injury [21,44,64–66]), suggesting that impaired microglial NKCC1 function alters microglial responses to injury or inflammatory stimuli.

To investigate the cell-autonomous effects of NKCC1 in microglia, we generated a new, microglial NKCC1 KO transgenic mouse line. We found that the absence of NKCC1 resulted in a more branching microglial morphology in resting state and a more amoeboid shape under inflammatory conditions. These results paralleled enhanced baseline process motility in the intact brain, in vivo, and significantly reduced process recruitment to the site of the acute lesion, when microglial NKCC1 was blocked by intracisternal bumetanide during 2-photon imaging. While the use of bumetanide was required for in vivo imaging of microglia due to the absence of microglial reporter protein expression in microglial NKCC1 KO mice, the cisterna magna application protocol had been previously shown to effectively target parenchymal microglia by using PSB0739, a selective P2Y12R antagonist [31]. As centrally administered bumetanide may also reach other NKCC1-expressing cells (e.g., neurons and astrocytes), we cannot exclude the possibility that microglial responses might have been influenced indirectly by other cell types in this experiment. Nevertheless, the net effect of central NKCC1 blockade by bumetanide leads to markedly impaired microglial process response to acute injury. Associations of NKCC1 with actin cytoskeleton dynamics [67,68] are well documented, which take place via controlling F-actin organization through Cofilin-1 and RhoGTPase activity [67]. Collectively, our results suggest that regulation of targeted process outgrowth in microglia is NKCC1 dependent, in addition to its known dependency on microglial P2Y12R, while THIK-1, a potassium channel that regulates baseline microglial surveillance and process dynamics, is not required for this process [35].

Importantly, we also found that the absence of microglial NKCC1 renders microglia to express higher levels of NLRP3 and IL-1β mRNA even without any exposure to inflammatory stimuli. It is well established that changes in microglial membrane potential and ion currents markedly determine cell volume regulation, process dynamics, and inflammatory responses of microglia. For example, depolarization is associated with reduced branching in microglia [35,69], while potassium or chloride efflux are essential for NEK7-NLRP3 interaction during the assembly of the NLRP3 inflammasome [38,50], which is required for the production of mature IL-1β [37,38,50]. The regulation of these processes appears to be highly complex. The TWIK2 two-pore domain K+ channel (K 2P ) mediates K+ efflux triggering NLRP3 activation in lung macrophages [37], while THIK-1 has been identified as the main K+ channel in microglia [35], although RNA profiling [21,40] also revealed high expression of TWIK2 in this cell type. In line with this, blocking THIK-1 function inhibits the release of IL-1β from activated microglia, consistent with potassium efflux being necessary for inflammasome activation [35].

To reveal how microglial NKCC1 may contribute to these processes, we investigated whether deletion of NKCC1 leads to altered expression of ion channels or transporters that have been linked with microglial ion- and volume regulation and with inflammatory cytokine production (see references in S4D Fig). We could not observe changes in THIK-1 or TWIK2 mRNA levels by qPCR, but the subunit D of VRAC was up-regulated in KO animals. Subunit D is involved in the transport of uncharged organic osmolytes by VRAC, and it may be important for volume reduction without a major release of inorganic ions [59] for a cell coping with potentially lowered inorganic ion levels due to NKCC1 loss. Changes in the expression of other VRAC subunits may also be important for microglial inflammatory processes. For example, LRRC8A-containing anion channels are known associate with NADPH oxidase 1 (Nox1) and regulate superoxide production and TNF-α signaling in vascular smooth muscle cells [59,70]. Thus, changes in microglial ROS production in the absence of NKCC1 will need to be investigated in future studies.

To examine the effects of NKCC1 deletion on microglial membrane properties, we conducted perforated patch recordings in hippocampal slices from WT and KO animals. Importantly, we found that NKCC1 KO cells had an input resistance almost twice as high as that measured in WT under both normotonia and hypotonia. We also characterized the swelling-induced current for WT and KO microglia, which is considered to be mainly mediated by the largely Cl− permeable VRACs [53–55]. In agreement with an expected decrease in cellular Cl− uptake upon NKCC1 deletion [1], E swell was significantly more negative in the KO versus WT microglia, which was further supported by comparing the driving forces of swelling-induced currents in individual cells, yielding higher values in KO cells. However, the driving force at V rest had a positive polarity (outward current) in all cells of both genotypes, which indicates that in addition to NKCC1 and VRAC, an active Cl− extruder is likely to be present in microglia. While the identity of the latter cannot be deduced from the present data, there is evidence that K-Cl cotransporter (KCC) isoforms, such as the volume-sensitive KCC1, are expressed in microglia [71]. It is worth noting here that in numerous cell types, NKCC1 and KCCs are simultaneously involved in chloride regulation [72]. Our data indicate that the increased input resistance in KO was also paralleled with a decrease in the swelling-induced current. We have no mechanistic explanation for these 2 effects, but they are likely to reflect adaptations of microglia to genetic NKCC1 deletion with compensatory changes not revealed by our qPCR studies. Such a coordinated adaptive response could involve down-regulation of VRACs and/or K+ channels among others. Up-regulation of VRAC subunit D in NKCC1 KO microglia could represent a compensatory response to impaired VRAC function, as suggested by the reduced swelling-related conductance in these cells.

We also found that in line with increased NLRP3 and IL-1β mRNA levels in NKCC1 KO microglia, LPS stimulation resulted in elevated NLRP3 mRNA levels even after 24 hours upon LPS treatment and potentiated the production of IL-1β and other cytokines in the cortex of microglial NKCC1 KO mice compared to control mice. Importantly, deletion of microglial NKCC1 had an identical effect on inflammatory cytokine levels to that seen after NKCC1 inhibition by bumetanide, suggesting that central bumetanide actions involve blockade of microglial NKCC1. In line with this, we could not find significant changes in 2 key astroglial markers, GFAP and AQP4, after central bumetanide application in response to LPS treatment, arguing against a major role of astroglial NKCC1 in the observed differences on the timescale of the present study. While a role for astroglial NKCC1 in mediating central bumetanide actions has been previously reported [18,19], our present work shows that central bumetanide effects involve a substantial microglial component via NKCC1.

Proinflammatory cytokine production, particularly the effects mediated by IL-1, are known to influence acute brain injury induced by brain trauma or stroke [73–75]. We found significantly increased infarct volume (by 47%) in microglial NKCC1 KO mice after experimental stroke comparable to a striking, 60% increase in infarct size caused by highly efficient, selective microglia elimination [32]. Microglia depletion resulted in dysregulated neuronal network activity, an effect that is likely to be mediated via elimination of protective microglia–neuron interactions at somatic purinergic junctions [31]. It remains to be investigated in future studies, whether microglia–neuron interactions are altered in microglial NKCC1 KO mice, which could shape neuronal injury via diverse mechanisms related to changes in neuronal excitability [29,30,32,33]. In line with this, the absence of microglial NKCC1 may have contributed to increased neuronal injury via increased IL-1 production, as suggested by the established pathophysiological role of IL-1–mediated actions in different forms of brain injury [73,76].

In our experimental stroke studies, we could not detect increased cytokine production in the absence of microglial NKCC1 early (8 hours) after MCAo, while both IL-1α and IL-1β expression were markedly up-regulated in NKCC1 KO microglia 24 hours after reperfusion. Following stroke, impaired cell volume regulation results in cytotoxic cell swelling and edema formation, which peaks beyond the first 24 hours in both patients and experimental animals [77,78]. Cerebral edema was shown to be associated with increased phosphorylation of the SPAK/OSR1 kinases playing a key role in NKCC1 activation in various neural cell types [2,48,49] and a SPAK kinase inhibitor, ZT-1a, attenuated cerebral edema after stroke [79]. In line with this, NKCC1 is involved in homeostatic cell volume regulation [48,49], and bumetanide attenuates edema formation in response to ischemic or traumatic brain injury [2]. In light of these findings, our data showing impaired volume regulation and exaggerated cytokine production in NKCC1 KO microglia have a potentially high pathophysiological and clinical relevance.

In conclusion, we show that microglial NKCC1 plays a previously unrecognized role in shaping microglial phenotype and inflammatory responses. Strikingly, bumetanide-induced enhancement of neuroinflammation was mimicked by conditional deletion of microglial NKCC1, suggesting that microglial NKCC1 is a significant target of NKCC1 blockers, given that brain tissue levels of such drugs are sufficiently high [10,13,14]. Our results implying microglial NKCC1 in inflammation, brain edema formation, and responses to injury suggest that microglial NKCC1 actions should be considered when studying the effects of NKCC1 inhibitors in different CNS functions and pathologies. This is further emphasized by the broad clinical interest in pharmacological NKCC1 blockade. The present results also call for a reevaluation of the pharmacological effects of bumetanide in brain diseases with a significant inflammatory component.

[END]

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001526

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/