Altered Kv2.1 functioning promotes increased excitability in hippocampal neurons of an Alzheimer's disease mouse model

Altered neuronal excitability is emerging as an important feature in Alzheimer's disease (AD). Kv2.1 potassium channels are important modulators of neuronal excitability and synaptic activity. We investigated Kv2.1 currents and its relation to the intrinsic synaptic activity of hippocampal neurons from 3xTg-AD (triple transgenic mouse model of Alzheimer's disease) mice, a widely employed preclinical AD model. Synaptic activity was also investigated by analyzing spontaneous [Ca2+]i spikes. Compared with wild-type (Non-Tg (non-transgenic mouse model)) cultures, 3xTg-AD neurons showed enhanced spike frequency and decreased intensity. Compared with Non-Tg cultures, 3xTg-AD hippocampal neurons revealed reduced Kv2.1-dependent Ik current densities as well as normalized conductances. 3xTg-AD cultures also exhibited an overall decrease in the number of functional Kv2.1 channels. Immunofluorescence assay revealed an increase in Kv2.1 channel oligomerization, a condition associated with blockade of channel function. In Non-Tg neurons, pharmacological blockade of Kv2.1 channels reproduced the altered pattern found in the 3xTg-AD cultures. Moreover, compared with untreated sister cultures, pharmacological inhibition of Kv2.1 in 3xTg-AD neurons did not produce any significant modification in Ik current densities. Reactive oxygen species (ROS) promote Kv2.1 oligomerization, thereby acting as negative modulator of the channel activity. Glutamate receptor activation produced higher ROS levels in hippocampal 3xTg-AD cultures compared with Non-Tg neurons. Antioxidant treatment with N-Acetyl-Cysteine was found to rescue Kv2.1-dependent currents and decreased spontaneous hyperexcitability in 3xTg-AD neurons. Analogous results regarding spontaneous synaptic activity were observed in neuronal cultures treated with the antioxidant 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox). Our study indicates that AD-related mutations may promote enhanced ROS generation, oxidative-dependent oligomerization, and loss of function of Kv2.1 channels. These processes can be part on the increased neuronal excitability of these neurons. These steps may set a deleterious vicious circle that eventually helps to promote excitotoxic damage found in the AD brain.

Alzheimer's disease (AD) is the most common cause of dementia in the elderly. Among other mechanisms, dementia of AD type can result from progressive functional and structural derangement of synapses, dendrites, brain circuits, nodes and networks. 1,2 According to recent findings, the AD-related deregulation of brain activity encompasses the simultaneous presence of hyperactive and hypoactive neurons that reorganize their activity in functional clusters. 3 This reorganization ultimately contributes to the production of network hyperexcitability and instability, hypersynchrony, epileptiform spiking as well as the activation of compensatory circuit remodeling. 4,5 At the neuronal level, the homeostatic regulation of excitability and plasticity relies on the coordinated functioning of a wide array of ionic channels. 6 In that context, delayed rectifier K + currents (I K ) have a major role in modulating cellular excitability. Kv2.1 channels are one of the most widely expressed Kv channels in the brain and considered the main determinants of I K currents in hippocampal neurons. 7,8 Kv2.1 channels control excitability but also have an important role in neuronal apoptosis. 9,10 Kv2.1 channels show very slow activation kinetics and deactivation dynamics, preventing complete channel deactivation upon sustained neuronal activity.
Oxidative stress has been shown to act as negative modulator of Kv2.1-dependent K + currents, 11,12 an intriguing finding considering the crucial role had by oxidative stress in the early stages of AD. 13,14 Considering the AD-related hyperexcitability and oxidative stress, it is possible that the Kv2.1 channel oxidation that may occur early on in the disease can promote unbalanced neuronal excitability and activity-dependent regulation as well as network instability and dysfunction. With this conceptual framework as reference, we studied the potential relationship occurring between changes in the levels of oxidative stress, alterations of Kv2.1-dependent K + currents, and variations in neuronal intrinsic activity in hippocampal neurons obtained from 3xTg-AD (triple transgenic mouse model of Alzheimer's disease) mice. The 3xTg-AD mouse, a widely investigated preclinical model of AD, has the benefit of exhibiting an agerelated development of β-Amyloid (Aβ)-and tau-dependent pathology as well as AD-related synaptic dysfunction and cognitive impairment. 15 Results 3xTg-AD hippocampal neurons show increased [Ca 2+ ] i spike frequency. Spontaneous activity is an important determinant of information processing in neuronal circuits and in the brain. 5,[16][17][18] Neurons develop spontaneous rhythmic and bursting activity at the end of the first week in culture, 19,20 a property that has been exploited to investigate basic aspects of network functioning in vitro.
[Ca 2+ ] i transients are considered an indirect index of the action potential firing status. 21,22 With real-time microfluorimetry, we evaluated spontaneous [Ca 2+ ] i transients occurring in soma of 3xTg-AD or Non-Tg (non-transgenic mouse model) hippocampal neurons. Hippocampal neurons were loaded with the high affinity Ca 2+ sensitive probe Fluo-4 AM and transients assessed in terms of number of spikes for minute (Figure 1a and b). In Non-Tg neurons, the pattern of spontaneous activity was found to be remarkably homoge-  Figure 1c). Treatment with the sodium channel blocker, tetrodotoxin (TTX), completely blocked [Ca 2+ ] i transients, thereby indicating that the observed oscillations were dependent on action potential firing. Moreover, transients were also completely abolished by the application of the ionotropic glutamate receptor blockers CNQX (6-cyano-7-nitroquinoxaline-2,3-dione; 10 μM) and MK-801 (10 μM), thereby demonstrating that increased [Ca 2+ ] i spike frequencies depend on enhanced spontaneous synaptic activity mediated by glutamate release and receptor activation (Figure 1e).
3xTg-AD mice show decreased delayed rectifier current density. In hippocampal neurons, Kv2.1 channels are the main driver for delayed rectifier K + currents (I k ). 7,23 The channels negatively modulate neuronal activity during repetitive spiking and promote neuronal integration by suppressing hyperexcitability when repetitive signals approach the soma. Kv2.1 channels undergo an oxidativedependent loss of function, a phenomenon that may promote the increased spontaneous activity that we have observed in 3xTg-AD neurons. Thus, employing patch-clamp whole-cell recordings, we investigated Kv2.1-dependent currents in our 3xTg-AD and Non-Tg hippocampal neurons.
At first, we assessed if this baseline shift of membrane potential was present in our Tg-AD cultures and found no statistical significant differences in resting membrane potentials values between 3xTg-AD and Non-Tg neurons (Supplementary Figure 1A).
Kv2.1-dependent current density values were instead found to be significantly lower in 3xTg-AD neurons ( Figure 2d). Analysis of current activation kinetics did not show differences between the two strains ( Figure 2c).
Taken together, these findings indicate that 3xTg-AD neurons express a lower number of fully functional Kv2.1 channels on their plasma membranes. No biophysical differences were found in the residual channel population between Non-Tg and 3xTg-AD neurons.
3xTg-AD hippocampal neurons show increased ROS production triggered by NMDA receptor activation. Moderate levels of ROS production have been shown to be important modulators of neuronal signaling. 24,25 ROS overproduction promotes damage by acting on multiple intracellular targets. To determine whether the reduction in macroscopic currents that we have observed in 3xTg-AD cultures is linked to increased ROS production and subsequent ROS-dependent loss of function of Kv2.1 channels, we examined cytosolic ROS levels evoked by a mild excitotoxic challenge obtained with exposure to N-methyl-D-aspartate (NMDA). To that aim, hydroethidine (HEt)-loaded neurons were challenged for 5 min with NMDA (50 μM, in presence of 10 μM of the receptor co-activator glycine; Figure 3a).  Figure 3b). These data indicate that, upon mild excitotoxic conditions, 3xTg-AD neuronal cultures may be more prone to ROS production. To evaluate baseline levels of oxidative stress occurring in 3xTg-AD and Non-Tg neurons, we then analyzed the culture basal ROS levels by assessing 3xTg-AD hippocampal neurons display increased expression and clusterization of Kv2.1 channels on somato-dendritic membranes. Oxidized Kv2.1 channels assemble in oligomers that are held together by disulfide bridges involving Cys-73. 12 The oligomerization process has been shown to lead to progressive accumulation of dysfunctional aggregates on neuronal plasma membranes. 11,12 To evaluate whether chronic oxidative stress induces Kv2.1 channel oligomerization in our AD model, hippocampal neurons were fixed and stained with a Kv2.1 antibody and clustering analysis performed. As expected, Kv2.1 channel expression was found to be strongly restricted to neuronal soma and proximal dendrites. When compared with Non-Tg cultures, 3xTg-AD neurons showed an overall increase in channel localization on neuronal membranes (Figure 4a and b). For each neuron, we measured the surface area occupied by Kv2.1 clusters within a defined somatic region, thereby obtaining a parametric 'clusterization' index. Thus, the index represents the fraction of the total plasma membrane area that is occupied by, immunohistochemically identified, Kv2.1 channels. 3xTg-AD neurons showed higher clusterization (average clusterization area in % was 28.58 ± 2.963 (S.E.M.) in 26 neurons) A previous in vivo study has shown that Aβ oligomers can modify Kv2.1 expression and promote increased channel synthesis. 26 Moreover, increased expression has also been shown to enhance channel clusterization. 27 To assess whether the increased Kv2.1 clusterization that we found in our AD model is dependent on higher levels of channel conglomeration but not synthesis, we performed a Kv2.1 western blot analysis in the two strains. WB results indicated overlapping amounts of Kv2.1 channels in the two cultures ( Figure 4c). Interestingly, 3xTg-AD hippocampal neurons exposed to an excitotoxic-like condition showed Kv.2.1 de-clustering. This finding suggests that, upon prolonged glutamatergic stimulation, 3xTg-AD neurons are, most likely, employing homeostatic mechanisms to regain basal levels of Kv2.1-mediated excitability (Supplementary Figure 1B and C).
In summary, our Ca 2+ imaging and electrophysiological experiments support the idea that 3xTg-AD neurons may develop a progressive accumulation of conglomerates of dysfunctional channels on their neuronal surfaces.
Pharmacological blockade of Kv2.1 channels promotes I k decrease and [Ca 2+ ] i spike frequency enhancement in Non-Tg neurons and has no effect in 3xTg-AD neurons. To evaluate whether the loss of function of Kv2.1 channels that we have found in the 3xTg-AD cultures is functionally related to the observed increased excitability, we performed complementary pharmacological manipulations on Non-Tg neurons in order to induce and mimic the behavior showed by 3xTg-AD neurons. To that aim, Non-Tg cultures were treated with the Kv2.1 channel blocker Guangxitoxin (GxTx) and whole-cell I k currents recorded. GxTx is a gating modifier of   As previous studies have suggested that NAC may also act independently of its antioxidant properties, 28 we tested our experimental paradigm with an additional antioxidant molecule. 3xTg-AD and Non-Tg neuronal cultures were therefore incubated for 24 h with the ROS scavenger, Trolox  Figure 8b).

Discussion
Our study supports the idea of a causative link between the development of hyperexcitability, increased spontaneous synaptic activity, and the ROS-dependent appearance of conglomerates of dysfunctional Kv2.1 channels. These results are in line with accumulating evidence indicating that AD is characterized by the development of aberrant network Analysis of spontaneous synaptic activity in our model confirms that 3xTg-AD neurons show a pattern of high [Ca 2+ ] i spike frequency that is associated with reduced amplitudes, two phenomena indicative of hyperexcitability.

KV2.1 conglomeration and activation: the oxidative link.
Suggesting a causative role for Kv2.1 channels in the hyperexcitability that we have observed in 3xTg-AD neurons, we found significant changes in channel activity of our AD model. The result can be explained by the high levels of oxidative stress that these neurons are known to face throughout development. 13,14 In AD, oxidative stress is a detrimental feature that has been suggested to occur even before the appearance of significant levels of plaque formation, neuropathological alterations or cognitive decline. 29,32 Oxidative stress is also strongly enhanced in the brain of preclinical AD models including the 3xTg-AD mouse. 33,34 In our experiments, compared with Non-Tg cultures, 3xTg-AD hippocampal neurons were found to be more prone to ROS overproduction in conditions of NMDA receptor activation (Figure 3a and b). ROS are instrumental in promoting Kv2.1 channels oligomerization, a process that favors the formation of conglomerates of dysfunctional channels that, through this step, become resistant to proteolytic cleavage, internalization or endocytosis. 12 These mechanisms lead to increased Kv2.1 surface expression without a matching increase in channel-mediated whole-cell K + currents. Recent data indicate that when channel density increases on the plasma membrane, non-clustered channels cease to conduct. 35 In agreement with this model, we found that 3xTg-AD neurons showed increased Kv2.1 oligomerization that pairs with decreased channel conductances (Figures 2 and 4).
These findings are confirmed by a set of complementary results in which we found that, in Non-Tg cultures, GxTx treatment produces an enhanced spontaneous [Ca 2+ ] i spike frequency that overlaps with the one observed in 3xTg-AD neurons (Figure 5c and e). In Non-Tg neurons, the same treatment induced a decrease in I k currents that is similar to the one observed in 3xTg-AD neurons (Figure 5b). Moreover, GxTx treatment in 3xTg-AD neurons failed to promote any significant modification in I k current densities as well as [Ca 2+ ] i spike frequency, thereby suggesting that the Kv2.1 channel population present in 3xTg-AD neurons is largely compromised and not susceptible to further pharmacological modulation (Figure 6a and c). The slight increase in [Ca 2+ ] i spike amplitudes observed in GxTx-treated 3xTg-AD neurons (Figure 6d) may be the result of compensatory activity of other [Ca 2+ ] i regulating systems triggered by the sudden drug-mediated inhibition of residual populations of Kv2.1 channels.
The relationship between high levels of oxidative stress and the appearance of Kv2.1 loss of function is further indicated by the NAC experiments. NAC pre-treatment rescued neuronal activity, decreased [Ca 2+ ] i oscillations, and increased normalized I k currents, (Figures 7 and 8). Trolox treatment was also able to decrease spontaneous [Ca 2+ ] i oscillations ( Figure 8). All these findings support a causative role for oxidative stress in setting the stage for Kv2.1-dependent neuronal hyperactivity. We, therefore, propose a sequence of pathogenic events that is centered on ROS-dependent modulation of the plasma membrane distribution of Kv2.1 channels. Lending support to this hypothesis, Kv2.1 immunostaining showed higher levels of plasma membrane appearance and clusterization of the channels (Figure 3). Our immunostaining data show that Kv2.1 are highly restricted on 3xTg-AD somatic and proximal dendritic membranes 36,37 where appear in large clusters (Figure 4), results that are in line with previous studies. 11,12 Astrocyte-neuron interactions are central in the modulation of neuronal and synaptic physiology. Kv2.1 channels are located on the soma and principal dendrites of both pyramidal and inhibitory neurons. Furthermore, the channels are strategically present at sites in the close proximity of astrocytic processes. This intercellular distribution promotes a rapid removal of the neuronal K + that is released into the extracellular space upon intense channel activation. 38 The possibility that such sophisticated regulatory mechanism may be altered in 3xTg-AD neurons represents an intriguing hypothesis that warrants future investigation.
Another important issue concerns the possibility that ROS may affect Kv2.1 activity by employing mechanisms that go beyond channel oxidation. In that respect, previous studies have shown that activity-and calcineurin-dependent Kv2.1 dephosphorylation induces hyperpolarizing shifts. Furthermore, experimental evidence have indicated 16 different Kv2.1 phosphorylation sites, only seven of which are dephosphorylated by calcineurin, thereby indicating multiple mechanisms regulating the channel biophysical properties. 39

Conclusions
Our study indicates a scenario where AD-related mutations may promote enhanced ROS generation thereby leading to oxidative-dependent oligomerization and loss of function of Kv2.1 channels, and, ultimately, hyperexcitability. Our model also suggests that interruption of this cycle may offer promising novel therapeutic approaches for AD. Animal handling and tissue preparation. All the procedures involving animals were approved by the institutional Ethics Committee (47/2011/CEISA/COM) and performed in accordance with institutional guidelines and national (D.L. n. 116, G.U., suppl. 40, 18 February 1992) and international laws and policies. Groups of 3-4 female mice were housed in colorless cages while male mice were singlehoused. Mice were kept on a 12/12 light/dark cycle and had ad libitum access to food and water. All efforts to minimize mice suffering were adopted. Murine hippocampal cultures were prepared from embryonic mice (E16-18). Briefly, hippocampal tissues were dissected in ice-cold, Ca 2+ free dissecting medium and subsequently minced with forceps. Hippocampal tissue containing medium was then transferred in a 0.25% trypsin solution for 10 min at 37°C. Hippocampal tissue was then centrifuged at 1300 r.p.m., 4°C for 5 min. After supernatant removal, pellet was dissociated with a fire-polished glass pipette. Dissociated hippocampal neurons were then re-suspended in Neurobasal medium supplemented with 1 × B27, 5% horse serum, 5% fetal bovine serum, 0.5 mM L-glutamine and 0.2% penicillin/ streptomycin. Neurons were then plated on culture plates or dishes previously treated laminin/poly-DL-lysine. A total of 5 μM of cytosine arabinofuranoside was added to the growth medium at 3 to 5 days in vitro (DIV) to arrest and inhibit excessive glial proliferation. Media changes were performed by replacing, every three days, 25% of the medium with fresh Neurobasal (a medium that does not contain FBS, HS or B27). Experiments were performed on cultures between 14 and 19 DIV. As far as the cellular composition of our cultures, it should be underlined that the cytostatic treatment was able to halt most of the glial cell replication. However, when we assayed, with anti-GFAP (to detect glial cells) and anti-MAP-2 (microtubule-associated protein 2; to detect neuronal structures) antibodies, the presence of astrocytes in our cultures, a significant amount (around 30%) of these cells was found. Thus, our cultures represent a viable mixture of neurons and supporting glial cells, a physiological setting that allows full interaction between neurons and astrocytes.   ROS measurement. Oxygen radical production was monitored using the oxidation sensitive dye HEt (Ex λ: 530 ± 15 nm, Em λ: 575-610 nm). Stock HEt (1 mg/ml) was prepared as previously described 40 in dry DMSO and stored in frozen aliquots for use within eight weeks. Cultures were loaded in the dark with 5 μM HEt in HCSS (45 min, 25°C). After loading, cultures were washed three times in HCSS and mounted on microscope stage in a static bath of HCSS containing 5 μM HEt. HEt was dissolved in all the solution throughout all the experimental session in order to maintain dye equilibration. Images were acquired by the same imaging setup previously described (see Ca 2+ Imaging and spontaneous [Ca 2+ ] i spikes analysis section). Cells were excited at 530 nm and emission was monitored at 4590 nm. To prevent the antioxidant activity of B27, at DIV 5 neuronal cultures were switched to a B27-free Neurobasal medium.
Electrophysiology. I k current recordings were obtained with the whole-cell voltage-clamp configuration. 41 Cells were mounted on an AxioExaminer microscope, patch pipettes pulled from borosilicate glass tubing (Science Product GmBh, Hofheim, Germany) and heat-polished at the tip to give a resistance of 3-6 MΩ. Electrodes were filled with a intracellular solution composed by (in mM): 10 NaCl, 117 KCl, 2 MgCl 2 , 11 HEPES, 11 ethylene glycol-bis-(β-aminoethyl ether)-N, N,N′,N′-tetraacetic acid (EGTA), and 1 CaCl 2 , at pH 7.2. Bath solution contained (in mM): 135 NaCl, 5 KCl, 1.2 MgCl 2 , 5 HEPES, 2.5 CaCl 2 , and 10 D-glucose, at pH 7.4. 0.3 μM TTX was continuously applied. All recordings were performed at room temperature (23-25°C). Holding potential was clamped at − 60 mV to reduce the contribution of A-type currents, from this holding value the following voltage protocol was used: 1 s test pulses from − 100 to +60 mV in 10 mV steps, followed by a 500 ms tail pulse to − 30 mV. Resting membrane potential was recorded using the above mentioned solutions in absence of TTX. Immediately after whole-cell access amplifier was switch from Voltage Clamp to I = 0 and traces acquired for 10 s in gap-free mode. A sampling interval of 25 μs/point was used and currents filtered at 5 kHz. Linear components of leak and capacitive currents were canceled using the P/N method. The Nernst K equilibrium potential E K was calculated as-79.4 mV. The normalized conductance was plotted against the test potential (V) and fitted to a single Boltzmann equation G = G max /(1exp[-(V-G 1/2 )/k ]). Here, G max is the maximum conductance, G1/2 is the test potential at which the I K channels have a half-maximal conductance, and k represents the activation curve slope. In the off-line data analysis, I k currents were evaluated at the steady-state amplitude selecting the last 500 ms of the 1 s test pulse. Stimulation, acquisition, and data analysis were performed with pCLAMP 10.0 and Clampfit software and Axopatch 200B amplifier (Molecular Device).