Neuromodulation of Astrocytic K+ Clearance

Potassium homeostasis is fundamental for brain function. Therefore, effective removal of excessive K+ from the synaptic cleft during neuronal activity is paramount. Astrocytes play a key role in K+ clearance from the extracellular milieu using various mechanisms, including uptake via Kir channels and the Na+-K+ ATPase, and spatial buffering through the astrocytic gap-junction coupled network. Recently we showed that alterations in the concentrations of extracellular potassium ([K+]o) or impairments of the astrocytic clearance mechanism affect the resonance and oscillatory behavior of both the individual and networks of neurons. These results indicate that astrocytes have the potential to modulate neuronal network activity, however, the cellular effectors that may affect the astrocytic K+ clearance process are still unknown. In this study, we have investigated the impact of neuromodulators, which are known to mediate changes in network oscillatory behavior, on the astrocytic clearance process. Our results suggest that while some neuromodulators (5-HT; NA) might affect astrocytic spatial buffering via gap-junctions, others (DA; Histamine) primarily affect the uptake mechanism via Kir channels. These results suggest that neuromodulators can affect network oscillatory activity through parallel activation of both neurons and astrocytes, establishing a synergistic mechanism to maximize the synchronous network activity.


INTRODUCTION
Animal survival is highly dependent on their ability to adapt to the changing environment.
To do so, animals are constantly switching between behavioral states, which are correlated to different network oscillations. We recently showed that local alterations in the extracellular K+ concentration ([K+]o) can affect the oscillatory activity of neuronal networks (Buskila et al., 2019) and that specific impairments of the astrocytic clearance mechanism can affect the resonance and oscillatory behaviour of neurons both at single cell and network levels (Bellot-Saez et al., 2018), implying that astrocytes can modulate neuronal network activity. However, the cellular and molecular mechanisms that affect this clearance process by astrocytes are still unknown.
Historically, network oscillations were considered to be highly affected by neuromodulation (Lee & Dan, 2012), and previous studies indicated an essential role for neuromodulators in mediating the shift between different behavioural states. However, we still know little about the circuitry involved in this neuromodulation, specifically at the cellular level.
Astrocytic Ca2+ signalling and glutamate clearance play an essential role in the regulation of the network activity and K+ homeostasis, which ultimately affects neuronal excitability underlying network oscillations (Wang Fushun et al., 2012;Ding et al., 2016). Recently, Ma et al. (2016) (Ma et al., 2016) showed that neuromodulators can signal through astrocytes by affecting their Ca2+ oscillations to alter neuronal circuitry and consequently behavioural output.
In line with these observations, Nedergaard's group further demonstrated that bath application of neuromodulators to cortical brain slices increased [K+]o regardless of synaptic activity , suggesting that increased [K+]o could serve as a mechanism to maximize the impact of neuromodulators on synchronous activity and recruitment of neurons into networks.
In this study, we have investigated ( To this end, we assessed the impact of specific neuromodulators on the astrocytic K+ clearance mechanism by measuring their influence on the [K+]o clearance time-course in acute brain slices and further assessed whether this impact was due to a direct activation of astrocytic receptors or due to indirect influence via the neural network.

Animals and slice preparation
For this study, we used 4-8-week-old C57/BL6 mice of either sex. All animals were healthy and handled with standard conditions of temperature, humidity, twelve hours light/dark cycle, free access to food and water, and without any intended stress stimuli. All experiments were approved and performed in accordance with Western Sydney University committee for animal use and care guidelines (Animal Research Authority #A10588).

Electrophysiological recording and stimulation
The recording chamber was mounted on an Olympus BX-51 microscope equipped with IR/DIC optics and Polygon 400 patterned illuminator (Mightex). Following staining (Fluo-4 AM, SR101) and short recovery period in the BraincubatorTM, slices of somatosensory cortex were mounted in the recording chamber, for a minimum of 15 minutes, to allow them to warm up to room temperature (~22°C) and were constantly perfused at a rate of 2 ml/min with carbogenated aCSF.
[K+]o measurements were performed in layer II/III of the somatosensory cortex, by placing the K+-selective microelectrode nearby a selected astrocyte (termed "astrocyte ") stained with SR101 ( Figure 1A). K+ clearance is temperature-dependent, with Q10 of 1.7 at 26 oC and 2.6 at 37oC, mainly due to Na+/K+ ATPase activity (Ransom et al., 2000a). Due to the absence of a selective blocker for astrocytic Na+/K+ ATPase, and our interest in assessing K+ uptake into astrocytes via Kir channels, these measurements were performed at room temperature (22oC), when Na+/K+ ATPase activity is fairly low. Various KCl concentrations, corresponding to low (~5 mM), high (~15 mM) and excessive (~30 mM), were added to physiological aCSF and locally applied at a constant distance (~10 µm) from the K+-selective microelectrode through a puffing pipette (tip diameter of 1 μm) for 0.1 sec, as previously described (Bellot-Saez et al., 2018). Preparation and calibration of the K+-selective microelectrodes were performed as detailed in (Deveau et al., 2005;Haack & Rose, 2014). In short, the voltage response of the silanized K+-selective microelectrodes was calibrated before and after experiments within the experimental chamber by placing the electrode in aCSF containing different KCl concentrations (2.5 or "normal" aCSF, 4, 10, 15 or 30 mM). Once the electrode potential reached a steady state, a dose-response curve was calculated using a half-logarithmic (Log10) plot. K+-selective microelectrodes were considered good if the recorded voltage baseline was stable and the voltage response was similar before and after its experimental usage (~10 % deviation). The K+ clearance rate was calculated by dividing the concentration amplitude with the decay time (90-10%).
To assess the impact of neuromodulators on the K+ clearance rate, [K+]o measurements were performed within the same brain slices before and after 5-minute bath application of different neuromodulators, including the cholinergic agonist Carbachol (100 μM), Histamine dihydrochloride (50 μM), Noradrenaline bitartrate (40 μM), NPEC-caged-Serotonin (30 μM) and NPEC-caged-Dopamine (10 μM). To exclude the involvement of neuronal activity, similar experiments were conducted after perfusing slices with neuromodulators and tetrodoxin (TTX, 1 μM) for 5 additional minutes. Polygon400 illuminator (Mightex) was used to uncage NPECcaged-Serotonin and NPEC-caged-Dopamine compounds by applying focal photolysis with UV light (~360 nm) in a selected area (50 µm) which includes the surroundings of the K+selective microelectrode, the KCl puffing pipette and the selected astrocytic domain with its processes, for 1 second prior to local application of KCl ( Figure 3A).

Drugs
All drugs were stored as frozen stock solutions and were added to aCSF just before recordings. Neuromodulators, including NA, Histamine, 5-HT and DA were purchased from Tocris Bioscience (In vitro Technologies Pty Ltd). Noradrenaline bitartrate and Histamine dihydrochloride were dissolved in water to a stock solution at final concentration of 100 mM.
Carbachol (Sigma Aldrich) and caged neuromodulators, including NPEC-caged-Serotonin and NPEC-caged-Dopamine, were dissolved in DMSO to a stock solution at final concentrations of 1 M or 100 mM, respectively. All stock solutions were stored at -20°C and protected from light when required. Comparrisons of K+ clearance before and after application of a certain neuromodulator or TTX were conducted using two-tailed paired student t-test, as they were conducted on same slices and in the same region. For group comparrisons between different [K+]o concentrations or treatments, in which different slices from different animals were used, we conducted oneway or two-way ANOVA followed by Tukey's post hoc test, according to the experimental design. Statistical comparisons were done with Prism 7 (GraphPad Software; San Diego, CA), and unless stated, data is reported as mean ± S.E.M. Analysis of K+ transient properties were performed using a custom-made MATLAB code (MathWorks). Probability values < 0.05 were considered statistically significant.
In order to assess the overall impact of DA on the K+ clearance rate we locally uncaged NPEC-caged-Dopamine compounds (10 µM) (Castro et al., 2013)

Noradrenaline (NA)
Astrocytes express several receptors for NA, including 1, 2 and β1-adenergic receptors, which mediate multiple processes. Activation of 1 receptors triggers the PLC/IP3 signalling cascade that results in Ca2+ release from the internal stores (Duffy & MacVicar, 1995;Ding et al., 2013) leading to enhanced activity of protein kinase C (PKC) and the cAMP response element-binding (CREB)-dependent transcription (Carriba et al., 2012), and also exacerbates glutamate re-uptake into astrocytes through GLT-1/GLAST glutamate transporters (Alexander et al., 1997). In contrast, activation of 2 receptors in astrocytes primarly increase glycogenesis and reduces cAMP activity via the inhibitory G-protein (Gi/o), thus providing high ATP levels during periods of high demand, although under certain conditions it may permit glycogenolysis (Donnell et al., 2012). However, stimulation of astrocytic β1-adenergic receptors activates Gproteins (Gs) which results in [Ca2+]i increases (Nuriya et al., 2017), cAMP accumulation, PKA activation and glycogenolysis . Moreover, activation of β1-adenergic receptors enhance the Na+-K+-ATPase activity and thus facilitates K+ clearance following high neuronal activity, yet this effect is abolished at high non-physiological levels of [K+]o (Hajek et al., 1996b).
Our results show that bath application of Noradrenaline bitartrate (40 µM) led to a decrease of the K+ clearance rate following local application of excessive (30 mM, 0.80±0.06 mM/sec,

Histamine
Astrocytes express different types of histaminergic receptors, including H1, H2 and H3, which mediate multiple processes, including glutamate clearance (Fang et al., 2014) and glucose homeostasis (Medrano et al., 1992). H1 receptors are Gq/11-coupled and therefore associated with PKC and PLC signalling pathways, which lead to Ca2+ release from the endoplasmic reticulum (ER) (Arbonés et al., 1988). H2 receptors are Gs-coupled and have been found to participate in glycogen breakdown and energy supply via activation of PKA and stimulation of AC (Hill, 1990). H3 receptors are Gαi/o-coupled and less abundant in cortical astrocytes compared to astrocytes from other brain regions (e.g. striatum, hippocampus) (Mele & Jurič, 2013). These receptors have been involved in mediating the inhibition of AC, while triggering PLA2, MAP kinase and PI3K/AKT signalling pathways (Mariottini et al., 2009;Jurič et al., 2016).
Subsequently, astrocytic [Ca2+]i elevations induce gliotransmitter release of glutamate, ATP or D-serine, thereby leading to modulation of synaptic strength and transmission in both the hippocampus (Papouin et al., 2017) and the cortex (Takata et al., 2011).
To test the impact of ACh on the K+ clearance rate, we bath applied slices with Carbachol (100 µM), a non-specific ACh agonist that binds and activates both nicotinic and muscarinic ACh receptors (Hobson et al., 1983). However, the K+ clearance rate was comparable between normal aCSF and Carbachol conditions for all [K+]o tested, as shown in Figure [K+]o, which must be cleared to maintain neuronal function. In the CNS, K+ homeostasis is maintained by astrocytic K+ clearance mechanisms, including "net K+ uptake" and K+ "spatial buffering" to distal areas through GJs (Orkand et al., 1966), however the mechanisms that affect these clearance processes and overall [K+]o dynamics are largly unknown.

Neuronal activity is accompanied by a transient local increase in
In this study, we investigated the impact of different neuromodulators known to act on both neurons and astrocytes in the K+ clearance process. Previous studies have demonstrated that neuromodulators, including DA (Ito & Schuman, 2007), ACh (Kirkwood et al., 1999) and NA (O'Donnell et al., 2012) affect neuronal excitability, leading to altered network oscillations at multiple frequencies (Constantinople & Bruno, 2011). Moreover, modulation of the cholinergic (Webster & Jones, 1988;Dort et al., 2015;Ni et al., 2016) or monoaminergic (Monti, 1993;Monti & Jantos, 2008;Carter et al., 2010) signalling pathways has been reported to affect neural network oscillatory dynamics underlying behavioural shifts, as happens during different phases of sleep (i.e. REM vs NREM) or between sleep and arousal states. Another key modulator of extracellular K+ is the Na+/K+ ATPase (NKA pump), which is expressed in both neurons and astrocytes, though with different subunit isoforms (Larsen et al., 2016). However, as there is no selective blockers for the astrocytic Na+/K+ ATPase, we did not measure its direct affect, to avoid misinterpertation of the direct impact of neuromodulators on astrocytic activity.
Recently Ding and colleagues showed that application of a cocktail of neuromodulators to cortical brain slices result in an increase of [K+]o, which did not involve neuronal activity . Moreover, different behavioural states, such as arousal and sleep that are modulated by different neuromodulators, were found to be associated with alterations in [K+]o dynamics

Mechanisms that affects the K+ clearance rate
Extracellular K+ dynamics are determined by the rate of active K+ uptake into nearby astrocytes, as well as the rate of extracellular diffusion (Gardner-Medwin, 1983). Previous reports suggested that the rate of K+ clearance can also be affected by different factors, including temperature (Ransom et al., 2000b), ammonia (Rangroo Thrane et al., 2013), glutamate (Enkvist & McCarthy, 1994) and pH, however, the cellular mechanisms affecting this clearance process are still largely unknown.
Among astrocytic K+ clearance mechanisms, K+ uptake becomes activated following low local increases in [K+]o (~3-12 mM), mostly affecting small astrocytic networks located within close proximity to the synaptic release site, and becomes saturated at [K+]o above ceiling levels (>12 mM) (Heinemann & Dieter Lux, 1977;Orkand, 1986). In contrast, the K+ spatial buffering process via GJ-mediated astrocytic networks is active when there is high accumulation of [K+] (Bellot-Saez et al., 2017). In that regard, agents that affect the clearance rate of low [K+]o (~5 mM) independent of neuronal activity are likely to play a role in the modulation of astrocytic K+ uptake mechanisms, mediated via the NKA pump and Kir4.1 channels (Hajek et al., 1996a;Butt & Kalsi, 2006;, whereas compounds that affect the clearance rate of high and excessive [K+]o (15 mM and 30 mM respectively) are more prone to regulate the K+ spatial buffering process through GJs (Wallraff et al., 2006;Pannasch et al., 2011). Indeed, our results indicate that selective blockade of Kir4.1 channels affected the [K+]o clearance rate at all concentrations tested (Fig. 2), as the uptake occurs at all concentrations. However, selective inhibition of astrocytic GJ's decrease the clearance rate only at high and excessive [K+]o (Fig. 2), consistant with previous reports (Wallraff et al., 2006;Pannasch et al., 2011). In addition, a key result in this study is that the rate of K+ clearance is concentration dependent and inversly correlated to the [K+]o (Fig. 1). This is probably due the fact that at high concentrations, K+ clearance is facilitated by GJ, as previously reported (Enkvist & McCarthy, 1994;. A previous study specified that BaCl2 mainly affect the [K+]o peak amplitude , however the amplitude was measured following high frequency stimulation that lasted for 10 sec, during which neurons constantly excert K+ to the vicinity of the recording electrode. In comparrison, our stimulus was much shorter (0.1 sec), and therefore allow direct measurement of the clearance rate without the impact of further K+ increase to the extracellular fluid.

Neuromodulators impact on the [K+]o clearance rate
The involvement of different neuromodulators in the regulation of network oscillations has been reported in many studies (Webster & Jones, 1988;Monti & Jantos, 2008;Dort et al., 2015;Ni et al., 2016), however the circuitry in which they mediate their impact remains elusive. Here we demonstrate that certain neuromodulators work in parallel on both neuronal and astrocytic networks, leading to differential impact on the K+ clearance rate. However, while some neuromodulators (e.g. 5-HT, DA and NA) exert their activity directly via astrocytes, other neuromodulators, such as Histamine expressed a more complex involvement, in which they affect the clearance rate indirectly via neuronal activity at low concentration and directly at excessive concentration (Fig. 6).
Moreover, while ACh had no impact on the [K+]o clearance rate at any of the concentrations tested (Fig. 7), all other neuromodulators significantly decrease the K+ clearance rate following excessive increase of [K+]o (Fig. 3-6). In contrast, the clearance rate following a low increase of [K+]o was affected only by DA and Histamine (Fig 4B and 6B respectively), though DA affected astrocytes direcly and the histaminergic effect was mediated by neuronal synaptic activity.
Together, these results suggest that 5-HT and NA effect on the K+ clearance rate was comparable to the impact of selective blockade of astrocytic gap junctions by Cx43 mimetic peptides, suggesting they affect only the spatial buffering process. In comparrison, DA and Histamine effect on the K+ clearance rate was comparable to the impact of selective blockade of Kir4.1 channels by BaCl2, suggesting they affect the uptake process. However, we cannot exclude the possibility that DA and Histamine affect the the spatial buffering process as well, as both processes are mediating K+ clearance under high and excessive concentrations.
Monoamines, including catecholamines (i.e. NA, DA), 5-HT and Histamine are involved in a broad spectrum of physiological functions (e.g. memory, emotion, arousal) (Cirelli & Tononi, 2000;Lowry et al., 2005;Eckart et al., 2016), as well as in psychiatric and neurodegenerative disorders (e.g. Parkinson's disease, Alzheimer's disease, schizophrenia, depression) (Panula et al., 1997;Ray et al., 2008;Heneka et al., 2010). At the cellular level, synaptic release of neuromodulators impact membrane properties as well as intracellular signalling pathways in both neurons and glial cells Ma et al., 2016), and previous reports showed that different neuromodulators can fine-tune the hyperpolarization-activated current Ih (Maccaferri & McBain, 1996;Rosenkranz, 2006;Ma et al., 2007), thereby affecting membrane resonance of individual neurons, which affect the oscillatory behaviour of single neurons and their synchronization into networks (Tseng et al., 2014). However, whether this was a direct effect of the neuromodulators on neuronal activity, or indirect via astrocytic modulation was never tested.
A recent study by Ma and colleagues show that neuromodulatory signalling in Drosophila can flow through astrocytes and promote their synchrous activation (Ma et al., 2016). They further suggest that astrocyte-based neuromodulation is an ancient feature of the Metazoan nervous sytem. Moreover, Slezak and colleagues (Slezak et al., 2018) suggested that astrocytes function as multi-modal integrators, encoding visual signals in conjuction with arousal state.
Our results support this concept, in which neuromodulators impact network oscillatory activity through parallal activation of both neurons and astrocytes, establishing a synergetic mechanism to maximise their impact on synchronous network activity and recruitment of neurons into networks.