Carbon Monoxide-Neuroglobin Axis Targeting Metabolism Against Neuroinammation

Microglia are the immune competent cell of the central nervous system (CNS), promoting brain homeostasis and regulating inammatory response against infection and injury. Chronic or exacerbated neuroinammation is a cause of damage in several brain pathologies. Endogenous carbon monoxide (CO), produced from the degradation of heme, is described as anti-apoptotic and anti-inammatory in several contexts, including in the CNS. Neuroglobin (Ngb) is a haemoglobin-homologous protein, which upregulation triggers antioxidant defence and prevents neuronal apoptosis. Thus, we hypothesized a crosstalk between CO and Ngb, in particular, that the anti-neuroinammatory role of CO in microglia depends on Ngb. A novel CO-releasing molecule (ALF826) based on molybdenum was used for delivering CO in microglial culture. for CO’s modulation of microglial metabolism. Finally, the metabolic shift induced by CO did not depend on alteration of mitochondrial population. In conclusion, neuroglobin emerges for the rst time as a key player for CO signalling against exacerbated neuroinammation in microglia.


Introduction
Establishing a line of immune defence in the CNS is crucial to ensure homeostasis and organism survival. Microglia are the phagocytic and immunocompetent cell population in the central nervous system (CNS), responsible for promoting the in ammatory response against infection and injury [1,2]. In response to pathogen or damage-associated molecular patterns, microglia change their phenotype and morphology, secreting several in ammatory mediators, such as cytokines (e.g. IL-6, TNF-α), and reactive oxygen and nitrogen species (ROS/RNS), which in turn can trigger cell death of their targets [1][2][3][4][5][6][7].
Recently, it has been observed that the establishment of this proin ammatory phenotype in microglia, as well as in macrophage, is accompanied by a metabolic shift: from oxidative to a more glycolytic metabolism [8][9][10][11]. Along with the resolution of the insult, microglia can also promote phagocytosis for the clearance of cell debris and tissue repair [3,12]. Thus, promoting a quick and e cient in ammatory response is a vital mechanism to preserve CNS function. Still, exacerbated or chronic neuroin ammation is a hallmark of several neurological pathologies, particularly in stroke, with drastic consequences to brain parenchyma, as well as during the progression of neurodegenerative diseases [1,2,13]. Therefore, therapeutic approaches that address the microglial in ammatory response are promising therapies against several cerebral disorders.
Likewise, CO treatment can also be cytoprotective by reinforcing cellular mitochondrial oxidative metabolism in astrocytes [21], hepatocytes [23], cancer cells [24], and macrophages [35]. Despite two decades of research about CO biology, CO's downstream players and molecular mechanisms are not fully understood. Finally, although CO holds great potential for future clinical applications, direct exposure to CO gas is not tissue-targeted and requires speci c medical equipment. Thus, several CO-releasing molecules (CORM) have been developed. CORMs are usually small organo-metallic molecules able to release CO under a controlled manner, which may be advantageous in a clinical setting [36].
Neuroglobin (Ngb) is a monomeric globin, described in 2000 by Burmester and colleagues, and was rst considered as a speci c neuronal globin [37]. Although this protein is mainly expressed in neurons, Ngb expression also increases in astrocytes and microglia, in response to stress, suggesting a potential role as cytoprotective protein [38][39][40][41]. In fact, when neurons and astrocytes are exposed to hypoxic/ischemic insults, there are increased levels of Ngb expression, which activate antioxidant defences, preserve mitochondrial membrane potential and prevent cell death [41][42][43]. In response to oxygen-glucose depletion, Ngb translocates into mitochondria and improves mitochondrial membrane potential, maintaining cell viability [44,45]. Likewise, Ngb can interact with mitochondrial proteins, such as VDAC and mitochondrial Complex III subunit cytochrome c1, preserving the mitochondrial membrane potential, mitochondrial oxygen consumption rates, and ATP production [42,[46][47][48][49][50]. Moreover, Ngb overexpression correlates with a decrease of brain lesion size in animal models of cerebral ischemia [51,52]. Upon in ammatory challenge, an increase of Ngb levels is also related with an anti-in ammatory effect in astrocytes and microglia using in vitro models [38,53]. Furthermore, the presence of Ngb is associated with a signi cant decrease of proin ammatory players and a better outcome using in vivo in ammatory models: sepsis-associated encephalopathy and transient hypoxia [54,55].
Taking into account that (i) CO and Ngb are cytoprotective molecules in CNS and both target mitochondria, (ii) expressions of Ngb and HO-1 are regulated by common transcription factors, and (iii) Ngb is a haemoglobin-homologous protein with potential for CO to bind, it is hypothesized that COinduced cytoprotective and anti-in ammatory signalling is dependent on neuroglobin expression. This study demonstrated that CO released by a new molybdenum based CORM (ALF826) promoted Ngb expression in microglia, which was accompanied by prevention of neuroin ammation and improvement of mitochondrial oxidative metabolism. CO anti-in ammatory effect was independent on ROS signalling and partially dependent on the transcription factor SP1.

Chemicals and reagents
All chemicals were of analytic grade and were obtained from Sigma, unless stated otherwise. Plastic tissue cultured dishes were from Sarstedt. Foetal bovine serum, penicillin/streptomycin solution and glutamine were obtained from Thermo Fisher Scienti c. RPMI-1640 with sodium bicarbonate was obtained from Sigma.
Cell culture BV-2 microglia cell line, kindly provided to us by Motterlini's lab, was used in this study. This microglia cell line was grown in RPMI-1640 with sodium bicarbonate, supplemented with 10% (v/v) Fetal Bovine Serum and 2% (v/v) Glutamine. Cell were maintained at 37°C, 5%CO 2 . When con uence was reached, cells were detached, by scraping them in growth medium, and a 1:4 cell passage was performed.

Genetic modulation of neuroglobin expression
To knocking down Ngb, BV-2 cells were transduced with shRNA Ngb (Sigma), using lentivirus. This created a BV-2 cell line with an shRNA Ngb integrated in its genome (Ngb KD-BV-2). BV-2 cells were also transduced with a scrambled shRNA sequence as a negative control. Likewise, to create a Ngb knock in, wild type and Ngb KD BV-2 cells were transduced with human Ngb, using lentivirus. The exogenous human Ngb sequence was integrated into their genome. Ngb expression was assessed by Western blot.
Carbon monoxide preparation, storage and exposure Carbon Monoxide-releasing molecule ALF-826 was kindly supplied by Dr Romão (ITQB-UNL and Proterris, Inc). ALF-826 was prepared in 0.1M NaHCO 3 at pH 8.4, to a nal concentration of 2.5mM. CORM-A1, a commercially available CORM, is a water-soluble molecule that releases CO at a slow rate, with a half-life of 21min, at pH 7.4 and 37ºC [56]. CORM-A1 was prepared in milli-Q water to a nal concentration of 5mM. Both CORMs were ltrated with 0.2µm lter. CORM-A1 was stored at -20°C and ALF-826 was stored at -80°C, protected from the light, to avoid release of CO. For each use, an aliquot was thawed and immediately used. Cells were exposed to 50µM of ALF-826 or to 12.5 µM of CORM-A1 for the indicated time points. Inactivated ALF-826 was obtained by leaving the aliquots in contact with the air and nonprotected from light, at RT, for 48h.

Lipopolysaccharide (LPS) treatment and NO quanti cation
To evaluate the anti-in ammatory effect of CORMs and Ngb's role, wild-type and genetic manipulated BV-2 cells were exposed to LPS (500 ng/mL) for 18h and then to CORMs for 6h, for a total of 24h. To evaluate the role of ROS signalling in CO-Ngb anti-in ammatory mechanism, cells were pre-incubated with antioxidant agents (1h) and ALF-826 for 6h. BV-2 cells were then challenged with LPS. The levels of nitrite, an indicator of the in ammatory marker NO, were measured using the culture medium by Griess assay. Total protein was collected, and proteins of interest were analysed by western blot. Finally, to understand the putative SP1 modulation of CO-induced Ngb regulation, cells were incubated with LPS for 17h, followed by exposure to 2µM Mithramycin A (Sigma) for 1h, and treatment with ALF-826 for 6h. Cell medium was collected and nitrite were analysed by Griess assay. Cell slices were analysed by immunocytochemistry.

Assessment of ROS production
To evaluate ROS production, BV-2 cells were exposed to both CORMs for the period indicated and then incubated with 10µM H 2 DCFDA (DCF) for 30minutes, at 37°C, at dark conditions. Hydrogen peroxide levels were evaluated by measuring DCF uorescence with a Tecan In nite F200 Pro (λ ex =485nm e λ em =530nm). The role of ROS signalling in Ngb upregulation was assessed by exposing cells to 0.5mM of N-acetyl cysteine (NAC) for 1h and to CORMs for 6h. Total protein was collected, and proteins of interest were analysed by western blot. ROS signalling part in Ngb-mediated CO's anti-in ammatory mechanism was disclosed by subsequently challenged BV-2 cells with LPS (500 ng/mL) for 18h, for a total of 24h. Nitrite levels were analysed by Griess assay.
Images were analysed using Fiji software [57]. The Correct Total Cell Fluorescence (CTCF) was calculated by subtracting the integrated density of the background to the integrated density of our region of interest, following the equation:

CTCF = (Region Area x Region Mean Fluorescence) -(Region Area -Background Mean Fluorescence)
For each time point, 50 cells were analysed per independent experiment. The background Mean uorescence was calculated using the average of two regions.

Mitochondrial oxygen consumption rate
Mitochondria function was evaluated by assessing oxygen consumption rate with or without treatment with LSP and CO. BV-2 cells (3300 cells/well) were plated in Seahorse XF96 Cell Culture Microplates (Agilent). Before the experiment, the cells were rinsed and incubated with Seahorse XF RPMI assay medium, pH 7.4, supplemented with 2g/L Glucose and 4mM Glutamine for 45min, at 37°C, without CO 2 .
Oxygen consumption rates were measured using a Seahorse XFe96 Analyzer (Agilent) by sequential adding 2µM Oligomycin, 1 µM FCCP and 1µM Rotenone and Antimycin A. Raw data were normalized by the Sulforhodamine B method. The results were analysed with the Seahorse Wave 2.6.1 software (Agilent).

Total DNA isolation
Cells were harvest and spin down at 500g, for 10min to remove the medium. After washing with ice-cold PBS pH7.4, pellets were resuspended in 10mM Tris 1mM EDTA pH7.8. Total DNA was extracted from cellular pellets. Cell lysis and protein digestion was performed overnight, at 37°C, by adding SDS and Proteinase K to a nal concentration of 0.5% and 0.05mg/mL, respectively. Total DNA was extracted with a phenol: chloroform: isoamyl alcohol (25:24:1) solution, washed with chloroform, and precipitated with ethanol. The procedures were based on Green, M.R. and Sambrook, J. 2017 Cold Spring Harb. Prot [58].
Quanti cation of mitochondrial population The mitochondrial population was further assessed with Mitotracker Deep Red (MTDR, Invitrogen M22426). Cells were harvest and incubated with 10nM MTDR, in growth medium, for 30min at 37°C, in the dark. Samples were evaluated with the ow cytometry analyzer BD FACSCanto II, with the Red laser 633nm. Per samples, 10,000 cells were analyzed and the median of MTDR uorescence for each sample was normalized against untreated cells.

Statistical analyses
Data are presented as mean ± SEM. Reported results were statistically evaluated using GraphPad Prism (GraphPad Software Inc.) Data was analysed using ANOVA and Student t-test in the appropriated cases. P<0.05 were considered signi cant.

Results
CO is anti-in ammatory in microglia pre-exposed to LPS Exacerbated or chronic neuroin ammation is a hallmark for several brain diseases, including stroke, traumatic brain injury, and neurodegenerative diseases. Therefore, it is important to identify new therapeutic strategies targeting neuroin ammation, and CO is a great candidate for it since it has been extensively described as an anti-in ammatory molecule in macrophages, with also some evidence in microglia. Herein, a novel CORM (ALF826), presenting a molybdenum centre in its chemical structure, was tested. Molybdenum is a trace metal found in the diet and present in the human body as the centre of a few enzymes, such as xanthine oxidase. Thus, the human body has the machinery to excrete this metal easier than other metals commonly present in CORMs.
In this work, BV-2 microglia cell line was rst exposed to lipopolysaccharide (LPS), the proin ammatory stimulus, and then CO treatment was done, for better mimicking real situation, where the neuroin ammation is established when treatment occurs. After CO treatment by the addition of ALF-826, in ammatory biomarkers were assessed ( Figure 1). Endogenous nitric oxide (NO) production occurs by the activity of nitric oxide synthases (NOS). The inducible nitric oxide synthase (iNOS) is an inducible isoform of NOS, responsible for generating NO in response to various stimuli, such as in ammation. Exposure to CO signi cantly decreased the expression of iNOS in cells challenged with LPS ( Figure 1A, B). Accordingly, CO exposure decreased the release of nitrite (an indirect measure of NO) into the extracellular medium of these cells ( Figure 1C), further indicating an anti-in ammatory effect of CO in our model. In response to in ammatory stimuli, microglia secrete several in ammatory molecules, including the proin ammatory cytokine TNFα and anti-in ammatory cytokine IL-10, which were evaluated. In fact, in LPS-challenged microglia, CO exposure increased the secretion of IL-10 ( Figure 1D) and decreased the levels of TNFα ( Figure 1E). Inactivated ALF-826 (iALF-826) partially reverted the increased nitrite production induced by in ammation, indicating that CO is implicated in the anti-in ammatory effect of ALF-826 (Supplementary Figure 1). Altogether, our data indicates that CO has cytoprotective and antiin ammatory properties in microglia, even after in ammatory stimuli have been established.

CO modulates Ngb expression
Disclosing the downstream players of CO-induced protective mechanism is crucial. It allows a deeper understanding of the biological role of this endogenous gasotransmitter, which can eventually help to better develop CO-based therapies. We hypothesized that Ngb may be a downstream player of CO's cytoprotection, namely during its response against in ammation. To disclose the role of Ngb in CO's protective mechanisms, the putative modulation of Ngb expression by CO was rst assessed. Microglial cells were exposed to ALF-826 ( Figure 2) and to CORM-A1, widely used in the literature (Supplementary Figure 2). Ngb, showed a signi cant upregulation after exposure to CO. This data indicates that CO increases Ngb expression in microglial cells.
Ngb expression is needed for CO-induced cytoprotection against LPS CO upregulates Ngb after 6h of exposure, and CO treatment for 6h also limits neuroin ammation. Thus, it was assessed whether CO-induced cytoprotection against in ammation is dependent on Ngb expression by knocking down this protein. BV2 cell line was transduced with shRNA Ngb using lentivirus generating Ngb-KD BV2 cell line. Ngb-KD BV2 cells were then pre-treated with LPS for 18h and exposed to CO for 6h ( Figure 3). As a negative control, cells were transduced with scramble shRNA (Supplementary Figure   3).
Whenever the Ngb expression was knocked down, CO treatment did not downregulate iNOS expression following LPS-challenge ( Figure 3A). Accordingly, nitrite concentration was no longer decreased in knocked down BV-2 cells following LPS stimulation and CO treatment ( Figure 3B). Furthermore, when Ngb was knocked down, levels of the anti-in ammatory IL-10 did not increase after CO treatment ( Figure  3C). Nevertheless, no alteration was observed in the levels of the proin ammatory TNFα cytokine. Thus, these data indicate that Ngb is needed for CO cytoprotection. Cells transduced with scramble shRNA, and exposed to LPS and CO, behaved similarly to wild type BV2 cells, under the same conditions. This indicates that the phenotype observed in Ngb-KD cells results speci cally from Ngb downregulation (Supplementary Figure 3A, B).
To further validate the role of Ngb, rescues experiments were performed. Knocked down BV2 cell lines were then knock in with Ngb gene, followed by cellular exposure to LPS and CO, as indicated before.
Rescuing Ngb expression restored CO protective effect, indicating that Ngb is crucial in CO-induced cytoprotection against in ammation (Supplementary Figure 3C). Thus, our data indicate that Ngb is a crucial partner for CO to promote anti-in ammatory effect.
ROS signalling role in CO-induced upregulation of Ngb Several authors have described low concentrations of ROS as important signalling molecules for CO's cytoprotective pathways [20,[30][31][32]. Furthermore, in several different cell types, such as macrophages, hepatocytes, neurons, astrocytes, and macrophages, exposure to CORMs or CO gas promotes low mitochondrial amounts of ROS, which act as signalling molecules, leading to the activation of antiapoptotic and anti-in ammatory pathways of CO [19,20,[30][31][32]. Thus, it was hypothesized that COinduced upregulation of Ngb might be dependent on ROS signalling. To disclose the role of ROS on the CO-Ngb axis, we assessed the ability of ALF-826 to generate ROS by measuring intracellular hydrogen peroxide production ( Figure 4A). ALF-826 induced signi cant hydrogen peroxide production after 30min of treatment ( Figure 4A), whose kinetics is in accordance with CO-induced ROS generation in other cell models. This suggests that ALF-826 generates ROS and may be anti-in ammatory in a ROS signallingdependent manner.
To test whether CO-induced Ngb upregulation is dependent on ROS generation, BV-2 cells were pre-treated with N-acetyl cysteine (NAC), a precursor of glutathione functioning as an antioxidant molecule. BV-2 cells were then treated with CO-releasing molecule ALF-826 for 6h ( Figure 4B). No signi cant difference in Ngb levels was found between cells exposed only to CO, and cells exposed to NAC prior to CO ( Figure 4B). Moreover, in cells challenged with LPS and treated with ALF-826, exposure to NAC did not alter CO-Ngb anti-in ammatory effect ( Figure 4C). Our data indicate that CO-induced Ngb upregulation and its consequent anti-in ammatory effects are independent of ROS signalling.

CO modulation of Ngb expression is partially dependent on SP1
Ngb expression is controlled by several transcription factors under physiological and pathological conditions. The transcriptional factor SP1 is important for Ngb upregulation upon hypoxia [59].
Furthermore, SP1 may be required for HO-1 expression [60]. Thus, it was hypothesized that CO modulation of Ngb expression is under SP1 control. Using immuno uorescence images, a kinetic study was performed to evaluate the potential activation by phosphorylation of SP1 and its translocation into the nucleus upon CO treatment ( Figure 5). Exposure of BV-2 cells to ALF-826 induced a signi cant increase on phosphorylated SP1 nuclear presence after 1h of treatment, indicating the ability of ALF-826 to promote SP1 activation ( Figure 5A, B). Still, to con rm that SP1 has a functional role on CO-Ngb antiin ammatory effect, LPS-challenged cells were treated with the SP1 chemical inhibitor mithramycin, prior to CO exposure ( Figure 5C). Although there is no statistically signi cant difference between mithramycintreated cells or not, there is a tendency for a partial reversion of CO's anti-in ammatory effect whenever SP1 is pharmacologically inhibited. This suggests a potential role of SP1 in CO-Ngb modulation of neuroin ammation. Nevertheless, SP1 may not be the single transcriptional modulator involved in Ngb-CO axis. Further studies are needed to fully understand CO transcriptional regulation of Ngb.

Ngb downregulation affects CO's modulation of mitochondrial metabolism
Under normal non-stressful conditions, microglia rely mostly on oxidative phosphorylation to generate ATP. Still, when threatened with an in ammatory stimuli, microglia seem to shift into a more glycolytic cell metabolism [8,11]. This shift is important for the activation of the in ammatory response [61]. Exposure to CO has been described as bene cial by improving the cellular metabolic status in several different models, particularly by promoting a shift into a more oxidative metabolism [21,23,24,35,62]. Ngb protective effect is also strongly linked to its interaction with mitochondria [42,[46][47][48][49]. Thus, the anti-in ammatory effect of CO-Ngb axis may dependent on modulation of mitochondrial metabolism. To disclose the role of CO-Ngb in mitochondrial metabolism, oxygen consumption was measured using Seahorse technology.
BV2 and Ngb-KD BV2 cell lines were pre-challenged with LPS and then exposed to ALF-826 for 6h. Then, cell medium was exchanged to Seahorse XF RPMI assay medium, supplemented with glucose and glutamine to the same concentration found in the growth medium. Oligomycin, FCCP, and Rotenone/Antimycin were injected sequentially into our system ( Figure 6A) to measure oxygen consumption under stressful conditions. Oligomycin is an ATP-synthase inhibitor used to estimate the oxygen consumption exclusively dedicated to ATP production. FCCP is a mitochondrial uncoupler that dissipates the proton gradient formed by the electron transport chain (ETC), allowing measurement of maximum oxygen consumption. Finally, Rotenone and Antimycin A are inhibitors of ETC complex I and complex III, respectively, and together inhibit completely ETC function and the consumption of oxygen, allowing the estimation of non-mitochondrial oxygen consumption (Supplementary Figure 4).
After LPS-challenge, microglia were exposed to CO, and their basal and maximum respiration increased up to levels closer to control without in ammatory stimulus ( Figure 6B, C). The maximal respiration indicates the maximal oxygen consumption that the cell can perform under stressful conditions. One may speculate that the metabolic improvement provided by CO in a situation of in ammatory insult suggests that CO may increase the chance of survival of this population by limiting glycolytic shift ( Figure 6C). Accordingly, cells exposed to LPS+CO had higher spare respiratory capacity and higher oxygen consumption linked to ATP production than cells exposed only to LPS ( Figure 6E, D). Whenever Ngb was knocked down, CO could no longer enhance basal and maximal respiration in LPS-challenged cells ( Figure 6B, C). Although there was a slight recovery of spare respiratory capacity in Ngb-KD cells exposed to LPS and CO, this effect did not translate into an increase on the oxygen consumption devoted to ATP production ( Figure 6E, D). On the other hand, when wild-type BV2 cells were compared to Ngb-KD cells, CO alone promoted higher oxygen consumption devoted to maximal and basal respiration, and to ATP-linked production.
Overall, these results indicate that CO promotes a better energy status in LPS-challenged cells by increasing oxidative mitochondrial metabolism. Moreover, Ngb emerges as a key factor for CO's modulation of microglial metabolism.

Mitochondrial population modulation in Ngb de cient cells
The previous data revealed an improvement of oxidative metabolism by CO under in ammatory conditions and in a Ngb-dependent manner. Therefore, the potential modulation of mitochondrial population by CO-Ngb axis was also explored. In fact, a tight regulation of mitochondria quality control is crucial for cellular functioning. To further understand the crosstalk between CO-Ngb axis and metabolism, the potential modulation of mitochondria population was quanti ed (Figure 7).
Several studies have proven that CO induces mitochondrial biogenesis in different models [63][64][65]. Thus, total mitochondrial population was assessed by quanti cation of mitochondrial DNA ( Figure 7A) and by ow cytometry using the mitochondrial dye MitoTracker Deep red (MTDR) ( Figure 7B). Surprisingly, in wild-type BV2 cells, 6h of CO exposure did not increase mitochondrial population measured by both methods (Figure 7). In conclusion, although there was a clear improvement on mitochondrial metabolism and consequently on mitochondrial function (Figure 6), ALF-826 did not increase the mitochondrial population in BV-2 microglia cells. On the other hand, mitochondrial DNA was enhanced in untreated Ngb-KD cells compared to wild type BV2 cells ( Figure 7A).
Mitochondrial population and turnover depend on the balance between the elimination of dysfunctional mitochondria and biogenesis of new mitochondria. In fact, mitochondrial quality control process depends on this balance. Thus, it can be hypothesized that CO may improve mitochondrial metabolism by modulating mitochondrial quality control, without affecting total mitochondrial population. Moreover, downregulation of Ngb enhances mitochondrial DNA, which may indicate an increase on mitochondrial proliferation. Still, further studies are needed to fully understand the connection between mitochondrial dynamics and CO-Ngb axis.

Discussion
Microglia are considered the immunocompetent cell population in the CNS, in charge of the in ammatory response upon injury or infection [1,2]. Although this in ammatory response is vital for CNS protection, chronic or exacerbated neuroin ammation results in the increase of toxic compounds and, consequently, of tissue damage. CO is an endogenous gasotransmitter cytoprotective in different cerebral disease models, namely stroke [66,67] and multiple sclerosis [27,68]. Likewise, CO has been shown to promote a microglial anti-in ammatory phenotype by directly targeting in ammatory players or mitochondria [26][27][28][29]35]. The present work demonstrated the crosstalk between CO and neuroglobin (Ngb), which is a globin protein involved in neuroprotection. In fact, CO limits neuroin ammation signalling in microglia by increasing Ngb expression.
Furthermore a novel and recently developed CORM containing a molybdenum centre (ALF-826) was used to deliver CO in microglial cells. Most CORMs have a metallic centre, which may represent a challenge for clinical application due to its putative toxicity [69]. Molybdenum is a trace metal found in the human diet and is present in the metallic centre of human enzymes (e.g. xanthine oxidase), thus the organism has the needed machinery for molybdenum excretion.
Exposure of microglial cells to ALF-826 signi cantly reduced the expression of iNOS and, consequently, the secreted nitrite levels, an indirect manner for quanti cation of NO production. ALF-826 also promoted the reduction of proin ammatory TNFα secretion and stimulated the secretion of anti-in ammatory IL-10.
Overall, the data indicate that exposure to CO via ALF-826 also limits the in ammatory response, as do other CORMs and exogenous CO gas in different models, such as macrophages or microglia [26,28,35,[70][71][72][73]. Furthermore, the anti-in ammatory effect of ALF-826 in microglia occurs even when it is applied after the in ammatory insult is established.
In microglia and astrocytes, Ngb expression increases in response to damaging conditions. Ngb showed cytoprotective effects in astrocytes, namely against hypoxic/ischemic and glucose deprivation insults [41,44]. Moreover, Ngb upregulation was associated with in ammatory resolution in microglia and astrocytes [38,53]. Thus it was hypothesized that CO and Ngb may be part of the same cytoprotective pathway and that the putative crosstalk between Ngb and CO may be part of CO-induced protective mechanism. In fact, our data indicate that CO's protection against exacerbated in ammation depends on this crosstalk. CO's upregulation of Ngb correlated with an anti-in ammatory effect and an improvement of mitochondrial oxidative metabolism, which is prevented when Ngb is downregulated.
Mitochondria are main target organelles for CO. Endogenous or low amounts of CO can partially bind to cytochrome c oxidase and generate low amounts of signalling ROS. In fact, CO is anti-in ammatory in a ROS signalling dependent manner [18,31]. Thus, one can speculate that CO modulation of Ngb levels may also depend on ROS signalling. Our assays showed that ALF-826 induced ROS generation. However, Ngb upregulationis not dependent on ROS signalling, since treatment with the antioxidant NAC did not alter Ngb expression levels, following CO treatment. Thus, one can also speculate that Ngb upregulation may be a parallel pathway or may be an upstream event to CO-induced ROS generation.
SP1 is a transcriptional factor important for regulating several housekeeping genes, as well as genes related to tumour proliferation, cell differentiation, apoptosis and angiogenesis [74]. SP1 upregulation has been associated with CO-induced upregulation of HO-1 in rat brain astrocytes-1 cell line and in primary cardiomyocytes, which is necessary for CO/HO-1-induced cytoprotection [60,75,76]. Furthermore, Ngb has potential binding sites for SP1, and upregulation of SP1 transcriptional factor correlates with Ngb upregulation [59,77]. Likewise, SP1 increased expression levels seem to be needed for Ngb upregulation under hypoxia [59]. Thus, CO may upregulate Ngb via SP1 expression and control. First, our data validated the ability of ALF-826 to induce SP1 activation by increasing SP1 co-localization in the BV2 nucleus. Nevertheless, SP1 pharmacological inhibition only partially reverts CO-Ngb anti-in ammatory effect, indicating that CO-induced Ngb upregulation may be only partially dependent on SP1. Thus, it can be speculated that the transcriptional regulation of Ngb, upon treatment with CO-releasing molecule ALF-826, may be the result of a coordinated action of different transcriptional factors. Further studies are needed to deeper understand the transcriptional regulation of Ngb.
Microglia are highly dynamic and complex cells that endure molecular and morphological alterations to perform their numerous duties [78][79][80]. Depending on their stimulus and its duration, microglia present different bioenergetic demands. These needs may have resulted in the metabolic exibility that is described in microglia [81]. Thus, in response to in ammation, microglia metabolism shifts to a more glycolytic consumption of glucose [8,11,61]. CO has been described as a metabolic modulator, by promoting a shift into a more oxidative metabolism in astrocytes, hepatocytes, and prostate cancer cells, as well as in macrophages and more recently in microglia [21,23,24,35,62]. Ngb expression has also been linked to the preservation of mitochondrial membrane potential, mitochondrial function (e.g. oxygen consumption rates) and ATP production in detrimental environments [42,[46][47][48][49]. Our data indicate that, under in ammatory conditions, exposure to CO after the insult (i) increase of basal and maximal respiration to values closer to control; (ii) enhancement of oxygen consumption associated with ATP production and (iii) improvement of the spare respiratory capacity of BV2 cells when challenged with LPS. The maximal respiration corresponds to the maximal consumption of oxygen when mitochondrial are under stressful conditions. Thus, by exposing cells to FCCP, a mitochondria uncoupler, these organelles had to compensate the loss of proton gradient by promoting an increased activity of the ECT, that resulted in a higher consumption of oxygen. CO increased maximal respiration in LPS-exposed BV2 cells, possibly due to activation of the ECT, which is essential for cellular survival. In fact, enhancement of the spare respiratory capacity, a feature that is directly dependent on the maximal respiration, has been described as an indicator of higher chances for cell survival in broblasts [82]. Thus, CO promotes a better outcome in these cells by increasing oxidative metabolism. However, when Ngb was knocked down, the previous described effect of CO on oxidative metabolism is abolished, suggesting that Ngb is needed for CO's modulation of cell metabolism. Furthermore, when comparing wild-type and Ngb knock down cells, CO treatment alone promotes higher consumption of oxygen associated with maximal and basal respiration, and ATP production. Our data showed that CO increased the oxidative metabolism in BV-2 cells challenged with LPS. Also, this modulation is dependent on Ngb expression.
Mitochondria are highly dynamic organelles, and their quality control programs directly in uence their role in cell metabolism and cellular fate. Mitochondrial population depends on two main processes: mitochondrial biogenesis and mitophagy (mitochondrial selective autophagy). Thus, the potential of CO to modulate mitochondrial population was explored in the presence or absence of Ngb. Accordingly, CO has been previously described to trigger mitochondrial biogenesis in different systems, such as cardiomyocytes [63], skeletal muscle cells [64,65], hepatocytes [34], or astrocytes [83]. Nevertheless, CO treatment or Ngb expression did not alter mitochondrial density assessed by MTDR labelling. Likewise, in wild-type BV-2 cells, CO did not change mitochondrial DNA levels as previously published for cardiomyocytes or skeletal muscle cells [63][64][65]. On the other hand, Ngb-KD BV2 cells treated with CO increased their mtDNA, but it did not result in an enhancement of mitochondrial density, indirectly assessed by MTDR labelling. The maintenance of the homeostatic mitochondrial population depends on the balance between mitochondrial biogenesis and selective mitochondrial elimination by mitophagy. Thus, one can speculate that CO improvement of mitochondrial metabolism may involve the maintenance of mitochondrial quality control by accelerating clearance of dysfunctional mitochondria, simultaneously with stimulation of mitochondrial biogenesis for maintaining mitochondrial homeostasis. Still, further studies are needed to fully disclose the link between CO-Ngb axis and mitochondrial quality control programs, particularly mitophagy.
Overall, our data indicate that CO-Ngb crosstalk is essential for CO anti-in ammatory mechanism in microglia, with cellular metabolic modulation as the underlying mechanism. The recognition CO-Ngb crosstalk and the study of its relevance in other models may open new doors for a deeper knowledge of CO signalling. Ultimately this may contribute to a future e cient application of CO as a therapeutic approach.

Availability of data and material
The datasets generated during and/or analysed during the current study are available from the corresponding author on a reasonable request.
Authors' contributions DDP designed the study, conducted the experiments, helped with data analysis and evaluation, wrote the manuscript, and approved the nal version of the manuscript. JSR helped and conducted some knockdown and knock-in experiments, helped with data analysis, and approved the nal version of the manuscript. VAS design part of the study, helped and conducted some seahorse experiments, helped with data analysis, and approved the nal version of the manuscript. JGJ helped with data analysis and approved the nal version of the manuscript. CCR provided the CORM ALF-826 (which is not available commercially), helped with data analysis and approved the nal version of this manuscript. PJO helped with data analysis and approved the nal version of the manuscript. HLAV designed the study, helped with data analysis and evaluation, wrote the manuscript, and approved the nal version of this manuscript.

Con ict of interest/Competing interest
The co-author Carlos C. Romão is a scienti c consultant of Proterris (Portugal) Lda, and has an equity position in the company. The other authors declare no con ict of interest and no competing interest.   Carbon monoxide increases the expression of neuroglobin in BV-2 cells. Cells were exposed to 50µM of ALF-826 for 6h. Neuroglobin expression was assessed by western blot. A representative blot (A) and quanti cation of Ngb expression (B) are presented. n ≥ 3, *p<0.05 versus 0h. Neuroglobin upregulation, induced by CO, is not dependent of ROS signalling. BV-2 cells were pre-treated with N-acetylcysteine (NAC) for 1h and exposed to ALF-826 for 6h. Cellular hydrogen peroxide levels (A) and Ngb expression (B) were evaluated, using H2DCFDA dye or by western blot, respectively. To test the role of ROS signalling in CO-Ngb protective effect, cells were also pre-exposed with NAC and CO, and then challenged with LPS (C). Cell medium was collected, and nitrite levels were measured by Griess assay. n ≥ 3 ***p<0.001, *p<0.05 versus untreated cells; ###p<0.001 versus LPS; $p<0.05 versus CO Figure 5 Role of SP1 in CO-Ngb axis anti-in ammatory effect. Cells were exposed to CO for the indicated time. SP1 presence in the nuclei was evaluated by immuno uorescence (A) and quanti ed (B). Nitrite levels were measured in the medium of cells pre-challenged with LPS, and incubated with 2µM mithramycin for 1h, before treatment with CO (C). n ≥ 3; ***p<0.001 versus untreated cells and ###p<0.001, #p<0.05 versus LPS, ns -no signi cant.  Mitochondrial population alteration by Ngb downregulation. BV-2 cells were pre-treated with LPS and exposed to 50µM of ALF-826 for 6h. Mitochondrial DNA was quanti ed by Q-PCR, using speci c primers for mitochondrial cytochrome b and nuclear GAPDH (A