The feeding behaviour of Amyotrophic Lateral Sclerosis mouse models is modulated by the Ca2+‐activated KCa3.1 channels

Background and Purpose Amyotrophic lateral sclerosis (ALS) patients exhibit dysfunctional energy metabolism and weight loss, which is negatively correlated with survival, together with neuroinflammation. However, the possible contribution of neuroinflammation to deregulations of feeding behaviour in ALS has not been studied in detail. We here investigated if microglial KCa3.1 is linked to hypothalamic neuroinflammation and affects feeding behaviours in ALS mouse models. Experimental Approach hSOD1G93A and TDP43A315T mice were treated daily with 120 mg·kg−1 of TRAM‐34 or vehicle by intraperitoneal injection from the presymptomatic until the disease onset phase. Body weight and food intake were measured weekly. The later by weighing food provided minus that left in the cage. RT‐PCR and immunofluorescence analysis were used to characterize microglia phenotype and the main populations of melanocortin neurons in the hypothalamus of hSOD1G93A and age‐matched non‐tg mice. The cannabinoid–opioid interactions in feeding behaviour of hSOD1G93A mice were studied using an inverse agonist and an antagonist of the cannabinoid receptor CB1 (rimonabant) and μ‐opioid receptors (naloxone), respectively. Key Results We found that treatment of hSOD1G93A mice with the KCa3.1 inhibitor TRAM‐34 (i), attenuates the pro‐inflammatory phenotype of hypothalamic microglia, (ii) increases food intake and promotes weight gain, (iii) increases the number of healthy pro‐opiomelanocortin (POMC) neurons and (iv), changes the expression of cannabinoid receptors involved in energy homeostasis. Conclusion and Implications Using ALS mouse models, we describe defects in the hypothalamic melanocortin system that affect appetite control. These results reveal a new regulatory role for KCa3.1 to counteract weight loss in ALS.


| INTRODUCTION
Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disease characterized by degeneration of upper and lower motor neurons, resulting in paralysis, inability to speak and death within 3-5 years after the first symptoms (Talbott et al., 2016). The first ALS-related gene, encoding a cytosolic Cu/Zn-binding SOD (SOD1) was reported in 1993 (Dal Canto & Gurney, 1994;Ripps et al., 1995;Rosen et al., 1993) and since 2011, mutations in more than 20 genes have been identified (e.g. C9orf72, TARDBP, FUS, HNRNPA1, SQSTM1, VCP, OPTN and PFN1). The underlying causes of ALS still need to be fully established. With regard to the nutritional aspect, hypermetabolism and low premorbid body mass index (BMI) at diagnosis have been identified as risk factors for ALS (Desport et al., 1999;Marin et al., 2010;Paganoni et al., 2011). ALS patients frequently suffer from weight loss, which is associated with shorter survival (Moglia et al., 2019;Shimizu et al., 2012). Consistently, mild obesity is associated with longer survival of ALS patients (Paganoni et al., 2011) and hypercaloric diets increase survival in both patients and ALS mouse models Dupuis et al., 2004;Ludolph et al., 2020;Wills et al., 2014). Several lines of evidence suggest hypothalamus dysfunctions in ALS (Cykowski et al., 2014;Gorges et al., 2017). However, pioglitazone treatment, which acts on the melanocortin pathway in the arcuate nucleus, failed to stimulate body weight gain in ALS patients and in SOD1 mice, suggesting that a breakdown of central processes that regulate energy homeostasis may contribute to alter appetite and body weight in ALS (Vercruysse et al., 2016). Furthermore, deregulation of the feeding behaviour and activation of microglial cells have been linked to hypothalamic neuroinflammation, leading to opposite results, such as involuntary weight loss or obesity ( Avalos et al., 2018;García-Cáceres et al., 2018;Thuc et al., 2017). Animals peripherally injected with LPS display a rapid increase in the expression of inflammatory cytokines in the hypothalamus that has been implicated in the generation of the anorexic response (Layé et al., 2000;Wisse et al., 2007). Consumption of a hypercaloric diet affects the number and size of microglia in the arcuate nucleus and the median eminence, before any changes in body weight gain is observed, suggesting a potential role of these cells in metabolic disorders (Valdearcos et al., 2017). These data support the idea that inflammatory pathways may play a causative role in metabolic dysfunction. In familial ALS-associated SOD1 mutations and in the corresponding mouse models, microglia acquire an inflammatory phenotype affecting motor neuron death (McGeer & McGeer, 2002), further myeloid cells expressing mutated SOD1 promote neurotoxicity (Boillée, 2006). In another ALS mouse model based on TDP-43 mutations, the TDP43 A315T mice showed pathological aggregates of ubiquitinated proteins in specific neurons and reactive gliosis, with the loss of both upper and lower motor neurons (Wegorzewska, 2009). Recently, we have demonstrated the involvement of the Ca 2+ -activated potassium channel K Ca 3.1 in microglia activation in ALS (Cocozza et al., 2018). In the CNS, K Ca 3.1 channels are expressed by microglial cells, where they regulate cell migration, inflammatory cytokine production and phagocytic activity in physiological and pathological conditions such as glioma, ischaemia, spinal cord injury and Alzheimer's disease (Chen et al., 2011;D'Alessandro et al., 2013;Grimaldi et al., 2016;Jin et al., 2019). Previously, we found that K Ca 3.1 inhibition in hSOD1 G93A mice attenuates the pro-inflammatory phenotype of microglia in the spinal cord, reduces motor neuron death, delays the onset of muscle weakness and increases survival (Cocozza et al., 2018). Here, we focus our attention on the most important brain region involved in the central control of feeding and energy expenditure, the hypothalamus. We treated hSOD1 G93A mice with the selective K Ca 3.1 inhibitor TRAM-34 starting at the presymptomatic stage, to investigate a possible link between microglial inflammatory signalling and alteration of the feeding behaviour. We found that chronic inhibition of K Ca 3.1 activity attenuates the pro-inflammatory phenotype of microglia in the hypothalamus, inducing the recovery of melanocortin tone and an

What is already known
• We recently demonstrated the involvement of K Ca 3.1 in microglia activation in ALS.
• The K Ca 3.1 blocker, senicapoc, is safe and well tolerated in clinical trials.

What does this study add
• New insights regarding the importance of counteracting inflammation to alleviate weight loss in ALS.
• Cannabinoid-opioid interactions are crucial for the hypothalamic regulation of feeding by K Ca 3.1 channels.

What is the clinical significance
• Identification of K Ca 3.1 channels as key modulators in feeding behaviour of ALS.
increase of food consumption and weight gain. Cannabinoid CB 1 receptors and μ-opioid receptors regulate energy balance via multiple neural pathways, promoting food intake and reward (Bermudez-Silva et al., 2012;Koch et al., 2015). Here, we suggest that these positive effects on feeding behaviour are mediated by CB 1 receptor activation, which increases ß-endorphin release by pro-  D'Alessandro et al., 2013). TRAM-34 was synthesized as previously described (Wulff et al., 2000). The long-term treatment is not toxic and does not induce changes in body weight, haematology, blood chemistry or necropsy of any major organs, in either mice or rats (Toyama et al., 2008;Wulff et al., 2000). The treatment regimen was chosen to reach a CNS concentration of TRAM-34 that effectively inhibits K Ca 3.1 channels, as previously described (Chen et al., 2011). The CB 1 receptor antagonist/inverse agonist rimonabant (SR141716, Cayman Chemical Company) was dissolved in 5% Tween 80 and 5% polyethylene glycol/saline. Mice were daily treated with 3 mgÁkg À1 of rimonabant or the same amount of vehicle (saline) by intraperitoneal injections. Animals were treated until the age described in the text. The μ-opioid receptor antagonist, naloxone hydrochloride (Tocris Bioscience) was dissolved in saline and stored at room temperature until day of use. Mice were daily treated with 7.5 mgÁkg À1 of naloxone or the same amount of vehicle (saline) by intraperitoneal injections. Animals were treated until the age described in the text.
Nesting objects were included with bedding (sawdust) materials. Food (regular chow, containing 14% protein, 5% fat and 3041 kcal/kg) intake was measured weekly, weighing food provided and left in the cage. The same food was provided to all the mouse strains used in these experiments. For this experiment, intraperitoneal injections were performed at the end of the light phase, between 6:00 AM and 7:00 PM, and food was weighted after 16 h. Transgenic animals were weighed once a week from 6 until 16 weeks of age.

| Energy balance
The experiments of indirect calorimetry were performed using the

| Hypothalamic neuronal cultures
Neuronal cultures were obtained from the hypothalamus of Postnatal Day 0-1 (p0-p1) C57BL/6 mice. Mice were killed by cervical dislocation and the hypothalami were removed and tissues were digested with 30-UÁml À1 papain (Sigma-Aldrich) at 37 C. After 20 min, the reaction was stopped by removing papain and washing twice with 2 ml of prewarmed Basal Medium Eagle (BME). Tissue was then triturated with a glass pipette to obtain single-cell suspensions, which were applied to 100 μm per 40-μm cell strainers. Cells were plated at a density of 250,000 cells per well in 24-well plates (1.9 cm 2 per well culture area) with BME. The culture medium was changed completely and carefully after 4 h with neurobasal medium (Gibco, Thermo Fisher) (2-mM glutamine, 1% B27, 100-UÁml À1 penicillin and 0.1-mgÁml À1 streptomycin). Half of the culture media was removed and replaced with fresh media every 3 days. After 9 days in culture, purity of neuronal cells, analysed as anti-tubulin β3 (TUBB3) (Covance), ranges between 60% and 70%.
Isolation of CD11b + cells from hypothalamus: Non-tg, vehicle and TRAM-34-treated hSOD1 G93A mice were deeply anaesthetized with chloral hydrate (400 mgÁ kg À1 , i.p.) and decapitated. Brains were removed, hypothalamic tissues were cut into small pieces and singlecell suspension was achieved in HBSS. The tissue was further mechanically dissociated using a glass wide-tipped pipette and the suspension was applied to a 30-μm cell strainer (Miltenyi Biotec). Cells were processed immediately for MACS MicroBead separation.
The cell suspension was loaded onto a MACS Column placed in the magnetic field of a MACS Separator and the negative fraction was collected. After removing the magnetic field, CD11b + cells were eluted as a positive fraction. Gapdh: forward (F), 5 0 -TCGTCCCGTAGACAAAATGG-3 0 , reverse (R),

| Immunofluorescence
Hypothalamic slices were prepared from hSOD1 G93A or non-tg mice treated with vehicle or TRAM-34 as described. Hypothalamic sections (20 μm) were washed in PBS; blocked (3% goat serum in 0.3% Triton X-100) for 1 h, at RT; and incubated overnight at 4 C with specific antibodies diluted in PBS containing 1% goat serum and 0.1% Triton X-100. The sections were incubated with the following primary Abs:

| Skeleton analysis
Microglia morphology was obtained by confocal microscopy using IBA1 signal. Twenty-micrometre z-stacks were acquired at 0.5-μm intervals using an FV10i laser scanning microscope (Olympus, Tokyo, Japan) at 60Â objective. Cell morphology was measured using a method adapted from that described by Morrison and Filosa (2013).
Maximum intensity projections for the IBA1 channel of each image were generated, binarized and skeletonized using the Skeletonize 2D/3D plugin in ImageJ, after which the Analyze Skeleton plugin (http://imagej.net/AnalyzeSkeleton) was applied. The average branch number (process end points per cell) and length per cell were recorded for each image with a voxel size exclusion limit of 150 applied. The areas of the soma and scanning domain were measured for each cell.

| CSF collection and neuropeptide analysis
For collection of CSF (Lim et al., 2018), mice were deeply anaesthetized with chloral hydrate (400 mgÁkg À1 , i.p.) before being placed on the stereotaxic apparatus under a dissecting microscope, monitoring the temperature and the spontaneous breathing. Once the dura over the cisterna magna was exposed (triangular in shape with usually one to two large blood vessels running through the area; either side of or between the blood vessels is optimal for capillary insertion and CSF collection), we used a microscope to see the sharp-

| Measurement of β-endorphin by ELISA
The hypothalamus of hSOD1 G93A mice treated with vehicle or TRAM-34 + rimonabant was disrupted with a homogenizer and analysed for β-endorphin content using a commercially available β-endorphin ELISA kit, in accordance to manufacturer's instructions (MyBioSource).
Briefly, after two freeze-thaw cycles were performed to break the cell membranes, the homogenates were centrifuged for 5 min at 5000Â g, 2-8 C. The supernatant was removed and assayed immediately. BCA Reagent Kit was used to measure protein concentration in samples (Thermo Scientific) and β-endorphin content was calculated as pgÁml À1 .

| Data and statistical analysis
Data are shown as the mean ± SEM. Statistical significance was assessed by unpaired Student's t-test, one-way ANOVA or two-way ANOVA for parametrical data, as indicated; Holm-Sidak test was used as a post hoc test; Mann-Whitney rank test and Kruskal-Wallis for non-parametrical data, followed by Dunn's or Tukey's post hoc tests.
A P value <0.05 was considered to be statistically significant.  It has been shown that central and peripheral inflammation affects the feeding behaviour in animal models (Braun & Marks, 2010). We have previously demonstrated that the Ca 2+ -activated potassium channel, K Ca 3.1, modulates the production of inflammatory molecules produced by spinal microglia in hSOD1 G93A mice (Cocozza et al., 2018). To investigate whether the K Ca 3.1 channels are also involved in the feeding behaviour in hSOD1 G93A mice by modulating the hypothalamic microglia phenotype, hSOD1 G93A and age-matched non-tg mice were treated with TRAM-34 (daily, 120 mgÁkg À1 ) from the age of 7 weeks (before appearance of signs) until 16 weeks (when signs are evident) (Figure 1a). After this period, mice were killed and 3.2 | Inhibition of K Ca 3.1 leads to weight gain by increasing food consumption in two mouse models of ALS: hSOD1 G93A and TDP43 A315T We have previously shown that inhibiting K Ca 3.1 channels with TRAM-34 leads to increased body weight in hSOD1 G93A mice (Cocozza et al., 2018). We now confirmed these data monitoring body weight variations in age-matched non-tg and hSOD1 G93A mice upon TRAM-34 or vehicle treatment from 7 to 16 weeks of age ( Figure 2a). As expected, vehicle-treated hSOD1 G93A mice had a slower rate of weight gain in comparison with non-tg mice, starting from 10 weeks of age ( Figure 2b). However, upon TRAM-34 treatment, hSOD1 G93A mice achieved a similar weight as age-matched non-tg mice. We hypothesized that the increase in body weight for F I G U R E 1 K Ca 3.1 channels modulate hypothalamic microglia in hSOD1 G93A . (a) Experimental scheme. (b,c) RT-PCR analyses of the relative expression level of (b) pro-inflammatory (top)  leading to the observed increase in body weight (Figure 2b). No change in food intake was observed upon TRAM-34 treatment of age-matched non-tg mice. These experiments were replicated in a different ALS mouse model based on TDP43 mutations (Wegorzewska, 2009). TDP43 A315T mice were treated with TRAM-34 (daily, 120 mgÁkg À1 ) from the age of 7 weeks (before appearance of signs) until 9 weeks (when signs are evident) (Figure 2d). After a few days, TRAM-34-treated TDP-43 A315T mice gained more weight ( Figure 2e) and consumed more food (Figure 2f)

| K Ca 3.1 inhibition changes the expression of hypothalamic neuropeptides involved in energy homeostasis
The hypothalamic control of energy homeostasis is regulated by an intricate network of neuropeptide-releasing neurons (Yeo & Heisler, 2012). To investigate specific effects induced by K Ca 3.  Figure 4b shows that a lower level of α-MSH expression is observed in hSOD G93A mice and that TRAM-34 treatment significantly increased this level. Processing of POMC precursor produces several bioactive products in addition to α-MSH; among them, β-endorphin is an agonist for μ-opioid receptors. It has been shown that targeting the cannabinoid system activating CB 1 receptor, in turn, triggers hypothalamic β-endorphin release, with effects on body weight . First, we found that blockade of K Ca 3.1 channels induced an increase of expression of the μ receptor (Oprm1) in CD11b + myeloid cells isolated from the hypothalamus of both nontg and hSOD1 G93A mice ( Figure 4c). Furthermore, we measured β-endorphin levels in the CSF and the data reported in Figure 4d show that K Ca 3.1 inhibition significantly increased β-endorphin levels in the CSF of hSOD1 G93A mice (non-tg, 78.3 ± 3.18 ngÁml À1 ; non-tg TRAM-34, 65.5 ± 5.50 ngÁml À1 ; hSOD1 G93A vehicle, 71.4 ± 2.13 ngÁml À1 ; and hSOD1 G93A TRAM-34, 85.6 ± 7.35 ngÁml À1 ). Taken together, these results suggest a possible link between microglial K Ca 3.1 inhibition and β-endorphin release due to activation of hypothalamic POMC neurons.

| K Ca 3.1 channels trigger cannabinoid-opioid interactions in feeding behaviour
To determine whether the increased expression of CB 1 receptors triggered by K Ca 3.1 inhibition is functionally relevant for feeding, mice were treated with the CB 1 receptor antagonist rimonabant (Kirkham et al., 2002), as shown in Figure 5a. Briefly, mice were treated daily (between 5:00 PM and 6:00 PM) with 120 mgÁkg À1 of TRAM-34 or with the same amount of vehicle + rimonabant (3 mgÁkg À1 , i.p.) from 7 weeks of age, for 7 days ( Figure S1). This treatment reduced food intake ( Figure 5b) and body weight (Figure 5c) with a major effect on TRAM-34-treated hSOD1 G93A mice after 48 h. We also observed that rimonabant increased the mRNA expression level of Cnr1 after 2 days (1.60-fold) and 7 days (1.56-fold) of treatment ( Figure S1d), suggesting that the major effect observed with rimonabant could be due to increased receptor levels. Interestingly, rimonabant induces tolerance to food intake after 4-5 days in both rats and mice, whereas steady reductions in body weight were still observed (Colombo et al., 1998).
In our experiments, we also found the development of tolerance after a few days ( Figure S1b,c). Rimonabant administration induced a decrease of β-endorphin secretion in TRAM-34-treated hSOD1 G93A mice, further confirming the role of this channel in modulating signalling pathways involved in feeding behaviour (Figure 5c). In addition, the μ-opioid receptor antagonist naloxone (7.5 mgÁkg À1 , i.p.) ( Figure S1d) selectively diminished food intake and body weight in TRAM-34-treated hSOD1 G93A mice already after 48 h (Figure 5d,e).

| DISCUSSION
We show that blockade of K Ca 3.1 activity induces positive effects on feeding behaviour of hSOD1 G93A mice, a model of human familial ALS. We have previously demonstrated that inhibiting K Ca 3.1 channels with TRAM-34 reduced the pro-inflammatory phenotype of spinal cord microglia in hSOD1 G93A mice, delaying the appearance of motor signs and lengthening survival time (Cocozza et al., 2018).
Recently, hypothalamic inflammation, especially in the glia component, has been associated with a dysregulation of feeding behaviour, raising questions about the role of glia in the modulation of feeding circuits (Benzler et al., 2015). Previous experiments have demonstrated a link between hypothalamic inflammation and the positive energy balance associated with obesity (Valdearcos et al., 2017), whereas others have reported an association of hypothalamic inflammation with the negative energy balance and anorexia-cachexia syndrome (Braun & Marks, 2010). Furthermore, recent findings led to the hypothesis that hypothalamic inflammation impairs neuronal mechanisms of appetite control, whether it is loss of appetite or food overconsumption (Dalvi et al., 2017;Le Thuc et al., 2016). Specifically, neuroinflammation is often associated with ALS, but not yet from metabolic perspective. Ngo et al. (2019) confirmed that appetite loss is prevalent in ALS patients and might also involve factors that extend beyond the physical disability.
Here, we tested the hypothesis that the activity of K Ca 3.1 channels participates in the hypothalamic neuroinflammation processes that dysregulate feeding behaviours and promote body weight loss, both negatively correlated with survival in ALS (Desport et al., 1999).
In accordance with this hypothesis, we observed that in acutely isolated hypothalamic microglia from hSOD1 G93A mice, K Ca 3.1 inhibition increases anti-inflammatory (Bdnf, Ym1, P2yr12 and Socs3) and   (Ludolph et al., 2020;Wills et al., 2014). A few years ago, it has been shown that targeting the cannabinoid system through CB 1 receptor activation can trigger the release of hypothalamic β-endorphin, with effects on body weight .
Because the major source of β-endorphin is Pomc, which is decreased in ALS mice, targeting the cannabinoid system to stimulate food intake may not be a successful strategy. Our observation that inhibition of K Ca 3.1 activity increases the expression of the hypothalamic CB 1 receptors in both ALS mouse models led us to hypothesize its involvement in β-endorphin release by POMC neurons and thus in the increase of feeding. In support of this hypothesis, we observed that (i) K Ca 3.1 inhibition increases β-endorphin levels in the hypothalamus, likely having orexigenic effects through activation of the μ opioid receptor  and that (ii), TRAM-34 treatment increases μ receptor expression in the hypothalamus and, specifically, also in CD11b + myeloid cells. Furthermore, the administration of the inverse CB 1 agonist rimonabant (rimonabant -to TRAM-34-treated hSOD1 G93A mice acutely halted the positive stimulation of food intake and body weight gain, before inducing tolerance. In addition, the opioid receptor antagonist naloxone specifically blocks K Ca 3.1 inhibition-induced feeding, confirming that cannabinoid-opioid interactions are crucial for the hypothalamic regulation of feeding in TRAM-34 treated hSOD1 G93A .

| CONCLUSIONS
In conclusion, our data demonstrate that K Ca 3.1 blockade promotes positive effects on feeding behaviour of hSOD1 G93A mice inducing (i) an anti-inflammatory phenotype in hypothalamic microglia, (ii) weight gain and increased food consumption, (iii) restored melanocortin tone and (iv), expression of neuropeptides involved in energy homeostasis. These data strongly support the need to identify specific targets to correct the energy metabolism defects observed in ALS mouse models and in patients. In particular, it will be important to investigate the communication pathways among cannabinoids, β-endorphin and food intake, and the interaction between POMCexpressing and μ receptor-expressing neurons. We speculate that possible mediators of this communication could be microglia-derived microvesicles (MVs). It has been shown that endocannabinoids may be released by microglia-derived microvesicles, which can activate CB 1 receptors and influence synaptic communication (Gabrielli et al., 2015).
It is important to note that a drug structurally related to TRAM-34, senicapoc, has been previously found to be safe in humans in Phase I, II and III clinical trials (Ataga et al., 2008(Ataga et al., , 2011 and would be available for repurposing, and has been deposited by Pfizer for exactly this purpose in the National Center for Advancing Translational Sciences

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

RIGOUR
This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design & Analysis, Immunoblotting and Immunochemistry and Animal Experimentation and as recommended by funding agencies, publishers and other organizations engaged with supporting research.

DATA AVAILABILITY STATEMENT
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.