Merits and Limitations of Studying Neuronal Depolarization-Dependent Processes Using Elevated External Potassium

Elevated extracellular potassium chloride is widely used to achieve membrane depolarization of cultured neurons. This technique has illuminated mechanisms of calcium influx through L-type voltage sensitive calcium channels, activity-regulated signaling, downstream transcriptional events, and many other intracellular responses to depolarization. However, there is enormous variability in these treatments, including durations from seconds to days and concentrations from 3mM to 150 mM KCl. Differential effects of these variable protocols on neuronal activity and transcriptional programs are underexplored. Furthermore, potassium chloride treatments in vitro are criticized for being poor representatives of in vivo phenomena and are questioned for their effects on cell viability. In this review, we discuss the intracellular consequences of elevated extracellular potassium chloride treatment in vitro, the variability of such treatments in the literature, the strengths and limitations of this tool, and relevance of these studies to brain functions and dysfunctions.


Mechanisms at Play
Elevated Extracellular KCl Induces Neuronal Depolarization, But Often, Not Activity The neuronal membrane voltage is regulated by, among other factors, the concentration of potassium and sodium ions in the intracellular versus extracellular space. At rest, K þ is more concentrated inside the neuron, while Na þ concentration is higher outside the cell. Sodiumpotassium pumps maintain these gradients, using ATP to pump Na þ out of the cell and K þ into the cell, against their gradients. Neurons at rest in vivo have a membrane voltage around -65 mV, whereas in vitro, they tend to be around -60 mV due to slightly higher KCl concentrations in commercial culture media formulations (5mM) in comparison to cerebrospinal fluid (3mM). Because the resting neuronal membrane is highly permeable to K þ , the membrane potential is sensitive to changes in the extracellular potassium concentration -increasing extracellular potassium depolarizes neurons. As the extracellular concentration of potassium rises, the magnitude of the potassium gradient is reduced, and the potassium equilibrium potential becomes more positive. As external Na þ continues to leak into the soma, global depolarization occurs. As calculated by the Goldman equation, a tenfold change in extracellular K þ from 5 to 50mM will cause a 48mV membrane depolarization under physiological conditions. Experimentally, reminiscent of the Nernst relationship, elevation of external K þ to 20, 30, and 40mM concentration in one study depolarized the membrane to -37, -26, and -19mV respectively; the specific membrane potential achieved at 10 mM KCl was not reported in text, but was approximately -56mV in Figure 2B (Wheeler et al., 2008).
What happens to neuronal activity post membrane depolarization in response to elevated external potassium? When treated with 8mM KCl, spontaneous activity is retained in cortical neurons, but the pattern changes from burst firing to tonic spike firing (Golbs et al., 2011). However, application of 15mM KCl or higher, somewhat unintuitively, leads to a complete attenuation of spontaneous neuronal activity in dissociated hippocampal neurons (Grubb and Burrone, 2010). In our laboratory, we see similar activity arrest in dissociated cortical neurons, when treated with 10mM or higher concentrations of external K þ (Figure 1). Such obliteration of neuronal activity may be explained by depolarization-sensitive inactivation of voltage-gated channels. Alternatively, or in addition, mild depolarization could deplete pre-synaptic terminals of neurotransmitters, thereby preventing subsequent rounds of activity (Grubb and Burrone, 2010).
Although 10mM or higher external K þ -induced depolarization doesn't produce neuronal activity, the sustained depolarization produces a chronic increase in an important second messenger, intracellular calcium (more in the next segment, 2.2) (Collins et al., 1991;Grubb and Burrone, 2010). Also, such sustained depolarizationcoupled inactivation of neuronal activity may have notable functional repercussions in the brain (see segment 4.3.3).

Calcium Influx
Both brief and prolonged stimulation with KCl elevate intracellular calcium in a sustained fashion for the duration of treatment Kingsbury et al., 2007;Evans et al., 2013;Tyssowski et al., 2018). When the calcium chelator EGTA is included during KCl treatment, it blocks the increase in intracellular calcium, induction of BDNF mRNA, dendritic growth, and enhanced survival brought about by KCl (Zafra et al., 1990;Ghosh et al., 1994;Tao et al., 1998;Redmond et al., 2002;. In some cases, these changes are likewise prevented by reduction of the calcium concentration in the culture medium (Zafra et al., 1990). Together, these observations suggest that most of the biological effects of elevated external potassium are mediated by intracellular calcium signaling. Notably, in contrast to the localized depolarization that comes with synaptic stimulation, KCl administration induces global depolarization. Depolarization-induced calcium influx to the soma is possible through several different channels, including NMDA receptors, calcium permeable AMPA receptors, various voltage-sensitive calcium channels (VSCCs), and intracellular Ca 2þ stores in the endoplasmic reticulum (Lyons and West, 2011). NMDA receptors are found at the synapse, but VSCCs are found globally across the plasma membraneboth at the synapse and at the soma. Studies with L-type Ca 2þ channel antagonists like the dihydropyridines (DHPs) nimodipine and nifedipine show that L-type VSCCs mediate much of the effect of membrane depolarization by elevated extracellular KCl. Nimodipene is reported to attenuate KCl-induced Ca 2þ influx by 15-20%  and nifedipine by 30-50% (Redmond et al., 2002). However, N-and P/Q-type channel blockers and NMDA receptor blockers can also reduce intracellular calcium concentrations induced by membrane depolarizations by 15-20% . Together, L-type VSCCs, N-and P/Q-type channels, and NMDARs each contribute to sustained intracellular calcium elevation in the cytoplasm and nucleus during potassium mediated depolarization Lyons and West, 2011). It is notable that although Ca 2þ influx is an absolute requirement for depolarization-induced downstream signaling, reduction of Ca 2þ influx only marginally impairs signaling, suggesting the signaling strength is non-linearly coupled with intracellular calcium levels.
One of the key downstream mediators of calcium influx is the transcription factor CREB. CREBdependent transcription signaled by Ca 2þ -CaMKII activation relies on the frequency of L-type channel openings, rather than bulk intracellular Ca 2þ levels (Wheeler et al., 2008). Furthermore, N-and P/Q-type channel inhibitors (x-Conotoxin GVIA and x-Agatoxin IVA respectively), are unable to inhibit CREB activation and CREB-dependent transcription across various extracellular Ca 2þ concentrations. In contrast, the L-type channel inhibitor nimodipine inhibits CREB Ser 133 phosphorylation at concentrations of extracellular Ca 2þ up to 20 mM . For these reasons, Ca 2þ influx through the L-type channels is thought to be primarily responsible for Ca 2þ -dependent signaling and CREB-dependent transcription induced by elevated extracellular potassium. The majority of this review will focus on signaling from L-type VSCCs.

L-Type VSCCs
Although multiple channels contribute to depolarizationcoupled Ca 2þ influx, L-type (long-lasting activation) channels provide specificity to the effects of elevated extracellular KCl in comparison to other stimulation paradigms (Ghosh et al., 1994;Lyons and West, 2011;Lyons et al., 2016). Induction of BDNF expression by KCl is completely blocked by L-type channel blockers nifedipine and nimodipine (Zafra et al., 1990;Ghosh et al., 1994;Lyons et al., 2016), but is not affected by the NMDAR blocker APV (D-2-amino-5-phosphonovaleric acid) or by CNQX (6-cyano-7-nitroquinoxaline-2,3dione), an AMPAR/kainate (aka, non-NMDAR) blocker (Ghosh et al., 1994). In contrast, BDNF induced by glutamate is sensitive to APV, but not nifedipine or CNQX (Ghosh et al., 1994). Differential regulation and expression profiles of BDNF induced by signaling through NMDARs and VSCCs may explain the observation that calcium influx through NMDARs is more excitotoxic than calcium influx through L-type VSCCs (Ghosh et al., 1994). VSCCs sustain BDNF expression longer than NMDAR signaling. While calcium influx through either channel may be potentially excitotoxic, robust BDNF induction after VSCC activation may be an effective neuroprotectant, in contrast to the transient and weaker BNDF induction by NMDAR activation (Ghosh et al., 1994). More information about L-type VSCCs may be found in Lipscombe et al. (2004), Berger and Bartsch (2014), and Kobrinsky (2015).
Calmodulin Kinase (CaMKII/CaMKIV). Calmodulin (CaM) binds the L-type VSCCs at the "IQ" motif in the carboxyl terminus of the channel . When extracellular KCl induces calcium influx through these channels, Ca 2þ binds CaM (Ca 2þ /CaM), inducing a conformational switch that allows calmodulin to activate calmodulin kinase signaling. Ca 2þ /CaM activates cytosolic CaMKK, which in turn activates cytosolic CaMKI and CaMKII, or nuclear CaMKIV (Ghosh and Greenberg, 1995;Greer and Greenberg, 2008;Hagenston and Baing, 2011). CaMKII shuttles Ca 2þ /CaM to the nucleus and phosphorylates activity-dependent transcription factors including CREB, NeuroD, NF-kB, and MECP2 (Greer and Greenberg, 2008;Cohen et al., 2018). CaMKIV, activated by CaMKK and increases in nuclear [Ca 2þ ], also activates CREB by phosphorylation at Ser 133 (Redmond et al., 2002). To successfully induce CREBmediated transcription, CaMKIV must also activate the CREB co-activator CBP (CREB binding protein), a histone acetyltransferase (Hagenston and Baing, 2011). Kinase-dead CaMKIV, elimination of CaMKII, and dominant-negative CREB each block KCl-induced IEG transcription and dendritic growth (Redmond et al., 2002;Cohen et al., 2018). Both L-type VSCCs and NMDARs evoke CaMK signaling, which is critical for propagating the L-type VSCC signaling to the nucleus. However, distinctions between signaling from NMDARs and L-type VSCCs activate transcription at different regulatory elements of IEGs, indicating the mode of Ca 2þ entry to the cell results in transcriptional specificity in the nucleus (Bading et al., 1993).
Ras/MAPK/ERK. CaM bound to the L-type VSCCs also activates Ras/MAPK/ERK signaling Hagenston and Baing, 2011). Ca 2þ /CaM activates the GTP-binding protein Ras, which binds and activates Raf (mitogen activated protein kinase kinase), which phosphorylates and activates MEK. MEK phosphorylates and activates MAPK/ERK, which translocates to the nucleus to phosphorylate TCF-family transcription factors and stimulate TCF/SRF-dependent transcription (Hagenston and Baing, 2011). Ras/MAPK signaling also activates nuclear Rsk kinases, including Rsk2, which phosphorylates CREB at Ser-133 (West et al., 2001). Ca 2þ /CaM activates Ras/MAPK and CaMK pathways in concert, and both may lead to CREB-dependent transcription, but these signaling pathways may be responsive to different timing and magnitudes of neuronal activation (West et al., 2001;Hagenston and Baing, 2011). For instance, MAPK/ERK inhibition blunts and delays the first wave of IEG induction, whereas later waves of IEG transcription are less impacted, if at all (Tyssowski et al., 2018). Also, crosstalk exists between the CaMK and the MAPK pathways. Several groups, including ours, have shown that CaMKI regulates depolarization-induced MEK-ERK activity and their physiological functions (Schmitt et al., 2004(Schmitt et al., , 2005Wayman et al., 2006;Poston et al., 2020). We have recently identified that certain ortho-hydroxylated brominated ethers, which are persistent environmental toxins widely detectable in organisms including humans, bind to and inhibit CaMKI, which subsequently impairs MEK-ERKdependent neuronal functions . Such inhibition by environmental toxins may have significant neurodevelopmental implications.
Calcineurin (CaN). Calcineurin (CaN) is a phosphatase associated to L-type channels by the AKAP family of proteins. CaN is activated when it binds Ca 2þ /CaM on its catalytic subunit, and Ca 2þ on its B subunit (Groth et al., 2003). Its affinity for Ca 2þ /CaM is much higher than the CaM-dependent protein kinases, and consequently initial elevations of intracellular Ca 2þ preferentially induce CaN-mediated phosphatase activity over CaM kinase signaling (Groth et al., 2003). Inhibition of CaN by FK506 or cyclosporine A blocks CREBdependent gene expression induced by depolarization and does not inhibit depolarization-induced Ca 2þ influx or CREB Ser-133 phosphorylation (Kingsbury et al., 2007;España et al., 2010). Calcium and cAMP signaling converge on CREB-dependent gene transcription through CRTC1/TORC, a CREB transcriptional coactivator (Kova´cs et al., 2007). Active CaN phosphorylates CRTC1, causing it to translocate from the cytosol to the nucleus, where it accumulates and initiates expression of CREB-target genes, including Sik1 (España et al., 2010,Li et al., 2009. SIK1 phosphorylates CRTC1, reversing translocation and depleting it from the nucleus. This negative feedback mechanism prevents persistent CREB/CRTC1-dependent transcription and dendritic growth in the face of long lasting neuronal activity (Li et al., 2009).

Chromatin Architecture at IEGs
In addition to activating CREB and CREB co-activators like CRTC1, CBP, and p300, extracellular KCl induces changes in the chromatin architecture at IEGs Kim et al., 2010;Pruunsild et al., 2011). Neuronal depolarization by KCl induces changes in histone H1 and H3.3, concomitant with IEG transcription (Michod et al., 2012;Azad et al., 2018). KCl also induces activity-dependent enhancers, cis-element activity, and DNA breaks regulating transcription (Pruunsild et al., 2011;Malik et al., 2014;Madabhushi et al., 2015). Promoters and enhancers of rapid IEGs -a subset of IEGs which are transcribed almost immediately after activity-are marked by paused RNA polymerase II and an accessible chromatin state (Saha et al., 2011;Tyssowski et al., 2018). Delayed IEGs -a subset of IEGs that occur post-activity later in the hour-do not have similar chromatin features at their promoters and/ or enhancers. Sustained activity-dependent transcription of delayed IEGs and other secondary response genes (Tyssowski et al., 2018) may be epigenetically regulated by chromatin remodeling (Martinowich et al., 2003) and negative feedback by histone deacetylases (Kyrke-Smith and Williams, 2018). The specifics of these transcriptional regulatory mechanisms may be highly dependent on cell identity, transcriptional history, and gene or even transcript identity.

Effects of Duration and Doses of Elevated External Potassium
Different types and durations of neuronal stimulation can induce different gene expression programs (Tyssowski et al., 2018). Moreover, the mode of Ca 2þ entry into the neuron plays a key role in determining which signaling pathways and gene expression programs are activated Martinowich et al., 2003;Kingsbury et al., 2007). While extracellular KCl concentration and membrane voltage are directly proportional (Leslie et al., 2001;Somjen, 2002), signaling strength may be non-linearly correlated with Ca 2þ influx levels (Wheeler et al., 2008). Here, we will discuss variations in the types of extracellular KCl protocols published for in vitro work, including pre-treatments, variations in KCl concentration ([KCl]), and variations in time of treatment.

Silencing Activity With Pre-Treatments
Experimentation with KCl and in vitro neuronal cultures began in the 1980s with treatments at 20-85 mM and treatment times between 10 minutes and 48 hours, with no pre-treatments to silence basal activity (Greenberg et al., 1985;Morgan and Curran, 1986;Bartel et al., 1989;Collins et al., 1989). The addition of Ca 2þ channel inhibitors like APV and CNQX used in conjunction with KCl treatments distinguished the roles of NMDARs, non-NMDARs (AMPARs), and voltage-sensitive calcium channels (VSCCs) in c-fos and Bdnf induction (Zafra et al., 1990;Bading et al., 1993;Ghosh et al., 1994). Many authors have subsequently employed inhibitors in the same manner, including APV and D-AP5 (NMDARs), CNQX and NBQX (AMPARs), nimodipine and nifedipine (L-type VSCCs), KN93 (CaMK), and PD (MEK) Redmond et al., 2002;Kingsbury et al., 2007;Evans et al., 2013;Cohen et al., 2018). These inhibitors are usually added to neuronal cultures several minutes (15-30 min) prior to KCl-depolarization to allay concerns about effectivity. For the experiments cited above, these inhibitors were used as experimental variants, and were compared to KCl-treatment without inhibition.
Historically, as KCl became a common means of depolarization, inhibitors became a means of standardizing activity levels and/or conferring specificity to the channels available for Ca 2þ influx. Pre-silencing all conditions with TTX and APV excludes NMDARs from the response to depolarization. Some cultures are treated overnight in 1mM TTX to reduce endogenous activity (Xia et al., 1996;Tao et al., 1998Tao et al., , 2002Lyons et al., 2012;Michod et al., 2012;Malik et al., 2014;Lyons et al., 2016). 100 mM APV is added 30 minutes prior to potassium depolarization to block the NMDARs (Tao et al., 1998(Tao et al., , 2002. Some experiments vary concentration and time of pre-treatment, or add other inhibitors, like CNQX (Xia et al., 1996;Grubb and Burrone, 2010;Malik et al., 2014;Cohen et al., 2018;Tyssowski et al., 2018). Experimenters must decide whether pre-treatment to silence synaptic activity is appropriate for their KCl depolarization experiments. For those attempting to isolate the activity of the L-type Ca 2þ channels, TTX þ APV þ CNQX pre-treatments may be useful. However, some treatments have their own constraints and effects; for instance, prolonged neuronal silencing with TTX triggers homeostatic changes (Turrigiano and Nelson, 2004), which may add artifacts to observations. Also, washout of TTX after prolonged application induces elevated activity (Saha et al., 2011;Lyons et al., 2016).
Pre-silencing with TTX may be unnecessary for many experimental paradigms, especially considering the activity silencing effects of KCl concentrations above 10mM (see Figure 1).

Variation in Stimulation by Concentration and Time
While 50-55 mM KCl has become the standard treatment concentration in the excitation-transcription coupling field, the range historically spans from low doses of 3 mM to high doses of 120 mM. According to the Goldman equation, 50 mM is sufficient to induce a þ48 mV depolarization in cultures maintained in media with 5 mM KCl. Experimentally, 50mM KCl has been shown to depolarize the membrane to about -30mV (Furness, 1970;Somjen, 2002). This does not bring the resting membrane potential to zero, but it is sufficient to induce robust gene expression. We have found that lower concentrations of KCl (mild depolarization) are also potent inducer of neuronal IEG transcription (Rienecker et al., 2020). Treatment times also vary widely, from 10 seconds to 48 hours. Contrary to a common presumption, prolonged treatment with reasonable concentrations of elevated external potassium does not induce apoptosis; instead, it facilitates cell survival (see "Cell Viability Under KCl Treatment" section). Furthermore, different gene programs are triggered by different duration of KCl-induced (or, any) depolarization (Tyssowski et al., 2018). However, to our knowledge, there has not been a methodical investigation into whether varied concentrations and treatment times interact to produce different gene expression profiles. We might expect variation, as different concentrations of intracellular Ca 2þ and modes of Ca 2þ entry differentially induce different signaling pathways, resulting in distinct transcriptional profiles (Rajadhyaksha and Kosofsky, 2005;Tyssowski et al., 2018;Rienecker et al., 2020). In Table 1, we have compiled and sorted selected papers by KCl concentration and time of treatment.
sophisticated equipment. Because this method induces global depolarization, it is unsuitable for distinguishing differences amongst inter-compartmental signaling events in a neuron that may exist in the brain. For example, unlike approaches such as localized glutamate uncaging, this method cannot distinguish activity-induced local events in the synapse from their dendritic, somatic, or axonal counterparts. Sustained global depolarization also cannot replicate variable stimuli on physiologically relevant timescales. Extracellular potassium is unsuitable for exploring how temporal features of action potentials are coded into specific intracellular signaling and transcriptional programs (Fields et al., 1997;Tyssowski et al., 2018). Another consideration is that experimental readouts are often population averages where cell-specific responses, which might differ based on history of activity and cell type, remain obscure. However, in nearly uniform cultures this simultaneously strengthens detection of small populational effects. As global depolarization induces calcium influx through both synapse-and soma-associated channels, multiple signaling pathways may be recruited by simple KCl treatments. Channelspecific inhibitors, gene constructs, and other tools can be used to impose specificity on KCl-induced Ca 2þ influx and signaling Redmond et al., 2002;Kingsbury et al., 2007;Evans et al., 2013;Cohen et al., 2018).

Cell Viability Under KCl Treatment
Elevated extracellular potassium promotes cell survival in a range of neuronal cell types and species. Reports of enhanced survival include peripheral neurons from chicken dorsal root, sympathetic, and parasympathetic ganglia (Bennett and White, 1979;Wakade and Thoenen, 1984;Koike et al., 1989;Collins et al., 1991;Crowder and Freeman, 1999), the mollusk giant neuron from ganglia tissue (Kostenko et al., 1982), murine cerebellar neurons (Didier et al., 1989), and neonatal rat cortical and cerebellar neurons (Lasher and Zagon, 1972;Golbs et al., 2011). Optimal potassium concentrations that promote survival lie between 25-40 mM, with survival falling off above 50 mM, and dropping to 20-50% of maximal survival at 85 mM (Collins et al., 1989). These cell survival effects are dependent on elevated potassium and Ca 2þ influx through DHP sensitive Ltype channels. Both must be sustained to promote cell survival; calcium chelators, replacement of elevated potassium with normal media, or L-type channel inhibition blocks the protective effects (Collins et al., 1989;Koike et al., 1989;D'Mello et al., 1993). Likewise, the L-type channel agonist BAYK 8644 enhances the protective effect, achieving high survival rates at lower potassium concentrations. BAYK 8644 allows L-type channels to open at significantly lower levels of depolarization, increasing Ca 2þ influx (Collins et al., 1989). While there is a strong quantitative correlation between mean sustained intracellular calcium levels and the percentage of surviving neurons (Collins et al., 1991), simple flux in intracellular calcium concentration is not the mechanism of neuroprotection. Concentrations of 90-200 mM KCl are reported to be lethal to glutamate sensitive neurons (Rothman, 1985;Choi, 1987;Mattson et al., 1988;Collins et al., 1989). Moreover, levels of intracellular calcium which are protective when induced by potassiummediated depolarization, are toxic when induced by glutamate signaling, particularly in the presence of NMDARs including GluN2 (Rothman et al., 1987;Collins et al., 1989;Martel et al., 2012). This is likely because potassium-mediated depolarization relies on Ltype VSCCs, whereas glutamate signals through NMDARs (Kudo and Ogura, 1986;Murphy et al., 1987;Choi, 1988;Collins et al., 1989). Ca 2þ influx through non-NMDARs is neither sufficient nor necessary for glutamate-induced neurotoxic injury, suggesting the mode of Ca 2þ entry determines toxicity or neuroprotection (Choi, 1988;Collins et al., 1989). Likewise, potassium-mediated and neurotrophic neuroprotection mechanisms are distinct; DHP inhibition of L-type channels does not block neuronal survival mediated by neurotrophins (Collins et al., 1989). The neuroprotective effects of elevated extracellular potassium are rather mediated by calcium-dependent intracellular signaling, although the relative importance of these different pathways may vary with neuronal identity. In neonatal rat spiral ganglion neurons treated with 25mM K þ for 96 hrs, nuclear CaMKIV promotes survival by activating CREB (see Figure 2), while CaMKII functionally inactivates the proapoptotic regulator Bad (Bok et al., 2007). At this sustained depolarization, CaN signaling is inactivated (Bito et al., 1997;Groth et al., 2003). The mechanism of elevated potassium-mediated neuroprotection may also partially depend on the treatment's effect on neuronal activity. A small elevation in extracellular potassium to 8 mM reportedly shift the frequency distribution of neuronal activity toward high-frequency bursts associated with reduced caspase-3 dependent apoptosis and increased neuronal survival (Golbs et al., 2011). In cortical neurons, this neuroprotective effect of highfrequency activity is mediated by a PI3K dependent pathway (Golbs et al., 2011). The PI3K/Akt pathway is involved in both depolarization and neurotrophin promoted survival in sympathetic neurons, but may not be a universal neuroprotective mechanism in all cell types (Crowder and Freeman, 1999). At extremely high concentrations (exceeding 85mM), elevated extracellular potassium becomes antagonistic to neuronal survival due to the osmotic effects of excess sodium and chloride influx. Removal of chloride from the external medium prevents toxicity of 90mM to 140mM K þ (Rothman, 1985). In summary, while the optimal extracellular potassium concentration for cell survival may vary between differing neuronal types, KCl is generally neuroprotective at concentrations between 25-40 mM and at lower concentrations when supplemented with BAYK 8644. Mechanistically, this neuroprotection is mediated by Ca 2þ dependent intracellular signaling including calmodulin kinases (CaMKII and CaMKIV) and the PI3K/Akt pathway, but the relative importance of different pathways to neuroprotection may also depend on the cell type and strength of the depolarization.

Implications of In Vitro KCl Research for In Vivo Functions
Extracellular KCl is useful for investigating cellular mechanisms of neurons in vitro. However, concerns are frequently raised about its relevance to events in vivo.
KCl treatments are indeed different from physiological stimuli in both dose and duration of application. Many protocols apply 50-55mM KCl for hours or days at a time, periods that do not reflect the duration of sensory stimuli or seizures. Nevertheless, many mechanisms mediating responses to elevated extracellular KCl also prove necessary for similar in vivo functions. Seizures, sensory stimuli, and learning events induce IEGs likewise induced by extracellular KCl (Morgan et al., 1987;Hunt et al., 1988;Guzowski et al., 2001;Lyons and West, 2011;West and Greenberg, 2011;Mukherjee et al., 2018;Heinz and Bloodgood, 2020). Furthermore, while sustained global depolarization by elevated KCl is not a feature of healthy neuronal signaling, it is a proposed mechanism of Leão's spreading depression of neuronal activity (Leao, 1944;Grafstein, 1956;Somjen, 2001;de Curtis et al., 2018). Below we discuss the in vivo phenomena that highlight the relevance of in vitro findings from research using extracellular KCl.

IEG Transcription Is Important for Learning and Memory, and Is
Induced in Physiological Conditions. IEG induction is a prominent consequence of elevated extracellular KCl treatment (Zafra et al., 1990;West et al., 2002;Tyssowski et al., 2018). While there are key differences between KCl and physiological stimuli, induction mechanisms discovered in vitro have held true in vivo (Greer and Greenberg, 2008). Transcriptional responses to depolarizing KCl in vitro have downstream effects on homeostatic plasticity, dendritic growth, and the remodeling of synaptic structures (Leslie et al., 2001;Redmond et al., 2002;Kova´cs et al., 2007;Grubb and Burrone, 2010;. Likewise, similar programs of IEG transcription in vivo are crucial for homeostatic plasticity, neurite outgrowth, synapse development and strength, neural adaptations to drugs of abuse, and learning and memory (Ploski et al., 2008;Lyons and West, 2011).
These IEG programs are so reliable, expression of genes such as Bdnf, Arc, and Fos are often used in immunohistochemistry as proxies of neural activity induced by learning paradigms and sensory stimuli (West and Greenberg, 2011;Lyons et al., 2016).
L-Type Channels and Ca2þ Signaling Are Involved in IEG Transcription, Learning and Memory, and Homeostatic Plasticity. Both in culture and under physiological conditions, IEG transcription is regulated both by synaptic (NMDAR, AMPAR signaling) and extra-synaptic signaling, including somatically (L-type VSCCs) distributed ion channels. In vitro, extracellular KCl generates Ca 2þ influx through L-type VSCCs and stimulates signaling to IEG transcription in the nucleus. Such signaling is highly relevant in vivo where L-type VSCCs induce IEG transcription and are involved in learning and memory, as well as in pathological brain functions, autism, aging, and pain (Shinnick-Gallagher et al., 2003;Jerome et al., 2012;Berger and Bartsch, 2014;Ghosh et al., 2017;Navakkode et al., 2018;Roca-Lapirot et al., 2018).

Models of aging and neurodegenerative diseases like
Parkinson's disease suggest alterations in Ca 2þ homeostasis through L-type VSCCs increase vulnerability to cognitive decline and synaptic dysfunction (Berger and Bartsch, 2014;Navakkode et al., 2018). Genome wide association studies (GWAS) and natural channel variants in humans link L-type VSCCs to working memory performance (Berger and Bartsch, 2014). Olfactory and fear memory in rats also depends on L-type VSCCs. L-type VSCCs induce transcription of the plasticity-related genes necessary for protein-synthesis dependent long term olfactory memory (Jerome et al., 2012;Ghosh et al., 2017). In brain slices from fear-conditioned rats, nimodipine reduces EPSC amplitude and increases paired pulse facilitation. When administered in vivo by intraperitoneal-injection, nimodipine also blocks fearpotentiated startle in a dose dependent manner (Shinnick-Gallagher et al., 2003). L-type VSCCs play a selective role in long-term, but not short term, fear conditioning memory by activating Ca 2þ /calmodulin and MAPK/ERK signaling to induce CREB-mediated transcription (Schafe et al., 2000;Bauer et al., 2002). Extracellular KCl in vitro induces Ca 2þ dependent signaling cascades involving calmodulin kinases, MAPK/ ERK, and calcineurin (Ghosh and Greenberg, 1995;Rajadhyaksha and Kosofsky, 2005;Greer and Greenberg, 2008;Hagenston and Baing, 2011;Tyssowski et al., 2018). In vivo, elimination of CaMKII prevents activity-dependent expression of IEGs like Bdnf, Fos, and Arc, inhibits persistent synaptic strengthening, and impairs spatial memory . Blockade of MAPK/ERK activation in the lateral amygdala impairs fear memory consolidation and synaptic plasticity in vivo (Schafe et al., 2000). CRTC1, aka TORC1, is downstream of CaN signaling and is crucial for contextual fear and spatial memory (España et al., 2010;Nonaka et al., 2014;Uchida et al., 2017). Knockdown of CRTC1 in CA1 and CA3 reduces longterm but not short-term contextual fear memory, implying its effect on activity-induced gene transcription, which is necessary for long-term memory. In response to neural activity and learning, CRTC1 localizes to the nucleus in a CaN dependent manner and induces Fgf1b. (Uchida et al., 2017). Furthermore, gene transcription mediated by CRTC1, including fos, Bdnf, and Nr4a2, is impaired in Alzheimer's disease transgenic mice expressing human beta-amyloid precursor protein.
These transcriptional deficits coincide with long-term spatial memory deficits in the transgenic mice (España et al., 2010).
Homeostatic alterations and remodeling of cellular structures are also consequences of global depolarization by KCl (Leslie et al., 2001;Grubb and Burrone, 2010;Evans et al., 2013;. In hippocampal cultures, chronic depolarization with high potassium moves the axon initial segment (AIS) away from the soma, dependent on the activation of L-type VSCCs (Grubb and Burrone, 2010). Changes in AIS position and activity-dependent plasticity are thought to fine tune neuronal excitability according to ongoing electrical activity (Grubb and Burrone, 2010). In Aplysia, subthreshold depolarizing changes in the presynaptic holding potential of the sensory neuron B21 increase the rate at which homosynaptic facilitation occurs. L-type VSCCdependent calcium elevation is partially responsible for these membrane potential-induced alterations in facilitation (Ludwar et al., 2020).
Potassium Hypotheses of Brain Disorders. Brain disorders such as migraine aura, epilepsy, haemorrhagic and ischaemic stroke, subarachnoid hemmorahage, and traumatic brain injury are associated with a robust phenomenon called spreading depression (SD), first described by Leão in 1944 and subsequently linked to increases in extracellular K þ in vivo (Leao, 1944;Somjen, 2001;Gupta, 2006;Fabricius et al., 2008;Richter and Lehmenku¨hler, 2008;Charles and Brennan, 2009;Cui et al., 2014;de Curtis et al., 2018). SD is a transient (60-120s), self-propagating wave of neuronal and glial depolarization accompanied by a negative shift of the current potential of about 20-35 mV and followed by prolonged quiescence (Richter and Lehmenku¨hler, 2008;Zhang et al., 2011;Cui et al., 2014). This propagation occurs in all directions at 3 mm/min, a rate similar to the spread of aura in classical migraine (Richter and Lehmenku¨hler, 2008;Cui et al., 2014). SD is accompanied by changes in blood flow, vascular caliber, and energy metabolism. While SD can occur in all neural tissues (Charles and Brennan, 2009;Dong et al., 2020), it is most often studied in the cortex, where it is referred to as cortical spreading depression (CSD). The phenomenon occurs in rat, mouse, rabbit, pigeon, and human models, among others (Leao, 1944;Fabricius et al., 2008;Richter and Lehmenku¨hler, 2008;Charles and Brennan, 2009).
The mechanistic hypothesis of SD postulates increases in extracellular [K þ ] trigger massive neuronal depolarization that leads to network hyperactivity and further K þ accumulation, which eventually depresses activity (Grafstein, 1956;Somjen, 2001;de Curtis et al., 2018). When the brain tissue is unable to clear the excessive K þ in the extracellular space, prolonged accumulation of K þ and sustained depolarization compromises synaptic transmission (de Curtis et al., 2018). Mild depolarization moves the membrane potential closer to the firing threshold and increases excitability, but stronger depolarization (as extracellular [K þ ] continues to rise) inactivates some voltage sensitive ion channels, raises the threshold, and eventually-reminiscent of in vitro findings mentioned in "Elevated Extracellular KCl Induces Neuronal Depolarization, But Often, Not Activity" section-suppresses excitability (Kager et al., 2002;Somjen, 2002;Dreier et al., 2012). During SD, increases in extracellular [K þ ] above the baseline 3 mM have been reported, ranging from 7.3 mM (Enger et al., 2015) up to 35 mM (Kraio and Nicholson, 1978), and are accompanied by a sharp drop in extracellular Na þ and Ca 2þ . This suggests K þ leaving the cells is exchanged against an influx of Na þ and Ca 2þ (Brinley et al., 1960;Nicholson, 1984;Silver and Erec nska, 1990;Somjen, 2001). Membrane potentials consequently depolarize significantly, briefly approaching zero (Somjen, 2001). In K þ induced models of CSD in vivo, K þ rise precedes CSD and intracellular increases in Ca 2þ . These neuronal Ca 2þ transients lag behind the negative direct current potential shift but last twice as long. Interestingly, astrocytic Ca 2þ increases lagged further behind negative potential shifts, and were shorter than neuronal Ca 2þ transients (Enger et al., 2015). Astrocytes can prevent the occurrence of SD through buffering of extracellular K þ (Cui et al., 2014). However, glial cells also experience the depolarization induced by SD, and extracellular [K þ ] may exceed their capacity to maintain physiologically tolerable limits. As neuronal dendrites and glia uptake K þ and swell, they restrict the interstitial space and further concentrate extracellular [K þ ] (Somjen, 2001).
Both elevations in extracellular [K þ ] and glutamate contribute to SD, but synaptic signaling is not required (Brinley et al., 1960;Charles and Brennan, 2009;Enger et al., 2015). CSD becomes less frequent and harder to trigger if P/Q calcium channels are genetically modified or specifically blocked. Likewise, blocking NMDARs by MK-801 stops the spread of CSD (Richter and Lehmenku¨hler, 2008). TTX, however, does not block the propagation of CSD, even though it suppresses action potential firing. Most likely, increases in extracellular [K þ ] and neuronal voltage-activated currents are key to the ignition and evolution of SD, as opposed to interstitial glutamate (Enger et al., 2015). However, extracellular [K þ ] is reported not to increase ahead of changes in voltage, as it should if it were the agent of propagation of SD (Somjen, 2001).
The difference between in vitro KCl treatments and in vivo SD is that SD events accompany other activity. In epilepsy for instance, SD interacts with synchronous ictal epileptic events and paroxysmal depolarization shifts (Dreier et al., 2012;Kubista et al., 2019;Zakharov et al., 2019). Thus, it is ambiguous whether the in vivo cellular consequences of SD associated with epilepsy relate to increased synchronous activity, or to the sustained depolarization that follows. Furthermore, as SD propagates across neural tissue at 3 mm/min, extracellular [K þ ] rises more slowly in physiological conditions than KCl treatments applied in vitro. The rate of depolarization by extracellular [K þ ] may also impact membrane voltage and cellular responses. Nevertheless, extracellular KCl can be applied to induce SD in experimental models in vivo (Marshall, 1959;Jing et al., 1991;Herdegen et al., 1993;Herreras and Somjen, 1993;Busch et al., 1996;Fujikawa et al., 1996;Somjen, 2001;Cui et al., 2014). While comparisons between in vitro KCl treatments and physiological events like SD must be cautious, they could be reciprocally informative.

Potential Use of Extracellular KCl in Other Fields
Fields broader than that of activity-regulated transcription may benefit from using potassium-mediated depolarization in vitro. For example, in environmental toxicology, toxins like polybrominated diphenyl ethers (PBDEs) and methylmercury (MeHg) act via pathways involving neuronal excitability, intracellular calcium concentrations, and MAPK, PKA/CREB, and PI3K/Akt signaling. PBDEs are a bioaccumulative class of brominated flame retardants with neurodevelopmental toxicity. PBDE metabolites like 6-OH-BDE-47 impair MAPK signaling  and disturb intracellular [Ca 2þ ] by causing an influx of extracellular Ca 2þ and release of Ca 2þ from intracellular stores in the ER and mitochondria (Dingemans et al., 2010). Disruption of normal MAPK signaling to IEGs and of excitability could play a role in the neurodevelopmental toxicity of PBDEs. Likewise, methylmercury increases intracellular Ca 2þ , disrupts MAPK and ERK1/2 signaling, and upregulates IEGs like fos, c-jun, and Bdnf (Bulleit and Cui, 1998;Dao et al., 2017;Fujimura and Usuki, 2017;. MeHg induced MAPK and PKA/CREB pathways trigger site specific neural hyperactivity and degeneration Usuki, 2017, 2018;. These similarities with mechanisms of potassium-mediated depolarization may prove informative.

Wrap Up
Elevated extracellular potassium is a useful in vitro treatment for elucidating the signaling, transcriptional, and plasticity-related events associated with global neuronal depolarization. Because it is applied by media replacement, it is scalable by dose and time. Many in vitro depolarization-induced mechanisms, including Ca 2þ influx through L-type VSCCs, Ca 2þ dependent signaling, IEG transcription, homeostasis, and long term potentiation, are also recruited for sensory stimuli, learning, memory, disease, and seizures in vivo. Indeed, KCl treatments have strong similarities to the potassium hypothesis of SD and epilepsy. However, researchers should justify the parameters of KCl treatments cautiously; prolonged periods, rapid change in concentration, and high doses of depolarizing KCl are not comparable to in vivo stimuli. Given the wide range of potassium stimulation protocols, more investigation is required into whether different stimulation parameters have differing results. For example, different neuronal activity patterns (short and prolonged) induce different IEG expression profiles (Greer and Greenberg, 2008;Tyssowski et al., 2018). Consequently, treatment length, cell type, and activity history of the neurons may heavily impact findings. The selected stimulation paradigm also has consequences for the final data type. Electrophysiological recordings allow internal measurements of the depolarization of single cells and can stimulate single cells to depolarize at graded stages. However, the modulation of only one or a few cells at a time is incompatible with certain molecular biology techniques that require higher cell loads, like chromatin immunoprecipitation. KCl treatment, which can be applied homogenously to a population of cells in vitro, is more appropriate for these measures. While global depolarization induced by potassium recruits multiple ion channels and several signaling cascades, experimenters can still achieve some degree of specificity. Inhibitors and genetic engineering can limit the effects of global depolarization to the mechanisms of interest. Inhibitors may be used as acute treatments or prolonged pre-treatments, depending on the intended comparisons. The body of work using elevated extracellular potassium in vitro has significantly contributed to our understanding of depolarization-transcription coupling mechanisms in neurons, and these findings are robustly confirmed by in vivo work. Consolidating an understanding of the boundaries of this stimulation protocol will ensure it remains a productive research tool.

Summary Statement
Elevated extracellular potassium chloride is widely used to achieve membrane depolarization, but there is great variability between protocols. We summarize the mechanisms of potassium-mediated depolarization, their relevance to brain functions and dysfunctions, and strengths and drawbacks of this technique.

Author Contributions
KDAR (lead) and RNS wrote the manuscript with feedback from RP. RP contributed MEA data for KCL depolarizations in Figure 1.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was conducted under the auspices of (NIH) a National Institute of Environmental Health Sciences (NIEHS), NIH grant to RNS (R01ES028738). The contents of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.