Neuroprotective effects of violacein in a model of inherited amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by progressive death of motor neurons and muscle atrophy, with defective neuron-glia interplay and emergence of aberrant glial phenotypes having a role in disease pathology. Here, we have studied if the pigment violacein with several reported protective/antiproliferative properties may control highly neurotoxic astrocytes (AbAs) obtained from spinal cord cultures of symptomatic hSOD1G93A rats, and if it could be neuroprotective in this ALS experimental model. At concentrations lower than those reported as protective, violacein selectively killed aberrant astrocytes. Treatment of hSOD1G93A rats with doses equivalent to the concentrations that killed AbAs caused a marginally significant delay in survival, partially preserved the body weight and soleus muscle mass and improved the integrity of the neuromuscular junction. Reduced motor neuron death and glial reactivity was also found and likely related to decreased inflammation and matrix metalloproteinase-2 and -9. Thus, in spite that new experimental designs aimed at extending the lifespan of hSOD1G93A rats are needed, improvements observed upon violacein treatment suggest a significant therapeutic potential that deserves further studies.

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by progressive loss of motor neurons and paralysis that becomes lethal within 1-5 years from diagnosis [1][2][3] . Rapid evolution and the younger age of affected patients compared to Alzheimer´s and Parkinson´s diseases 3 make ALS of considerable medical and social attention despite its low incidence (1-2.6:100,000) [1][2][3] . ALS is mostly a sporadic disease that results from yet unknown interactions among the environment, genes and age 2 . However, a minor percentage of cases is linked to inheritable genetic abnormalities in which mutations in the enzyme Cu/Zn superoxide dismutase-1 (SOD1) seem responsible for up to 20% of the familial types and 1% of the total cases [2][3][4] . Interestingly, when some of the SOD1 mutations identified in ALS patients were expressed in mice and rats, it elicited the most relevant pathognomonic signs of the disease including motor neuron death, glial reactivity and progressive paralysis 5 . In addition, data obtained from hSOD1G93A models showed that motor neuron death only occurs if the mutated enzyme is simultaneously expressed in motor neurons and in astrocytes or microglial cells [6][7][8] . Furthermore, astrocytes obtained from patients of sporadic and familial motor neuron diseases were neurotoxic 9,10 , even when SOD1 is not involved 10 . Therefore, neurotoxic astrocyte phenotypes may greatly contribute to motor neuron death in ALS.
Recent evidence showed that a subset of reactive/aberrant astrocytes were isolated from the spinal cord of hSOD1G93A murine models because of their proliferative potential that allowed an oligoclonal expansion without replicative senescence 11 . Isolated astrocytes were particularly neurotoxic 11,12 , either by gain of yet unknown deleterious properties 11 or by loss of homeostatic functions such as lack of expression of glutamate transporters 11,12 . This subgroup of neurotoxic astrocytes, that was called AbAs (as the acronymum of aberrant astrocytes), clearly presented a phenotype that was distinctive from homeostatic astrocytes 11  www.nature.com/scientificreports/ of undifferentiated astrocytes such as high levels of S100β and connexin 43, low expression of glial fibrillary acidic protein (GFAP) 11 , absence of gliofilaments and abundance of microtubules 14 . AbAs proliferated faster than neonatal or adult astrocytes, they could replicate more than 20 times preserving their exceptional selective toxicity to motor neurons 11,13 , and showed permanent absence of contact inhibition that allowed them to grow in invasive tridimensional aggregates enriched in extracellular matrix 14 . Previous reports also provided powerful evidence about AbAs disturbances in cellular communication, autophagy and proteostasis as well as on their exacerbated endoplasmic reticulum (ER) stress and disrupted lipid metabolism 13,14 . Moreover, likely due to their significant mitochondrial dysfunction 14,16 , AbAs experienced the Warburg effect 15 described in cancer cells since the glycolytic metabolism was their main energetic source instead of pyruvate mitochondrial oxidation 16 . Thus, AbAs and cancer cells might share similar features including some common energy production pathways as well as increased cell proliferation rate and absence of replicative senescence. Therefore, we have proposed that drugs that inhibit cancer cell proliferation, and ideally exhibit multifunctional beneficial properties, may not only control AbAs, but also may show protective effects on ALS models. The bisilindole pigment violacein (3-[1,2-dihydro-5-(5-hydroxy-1H-indol-3-yl)-2-oxo-3H-pyrrol-3-ylidene]-1,3-dihydro-2H-indol-2-one) is produced as a secondary metabolite by diverse bacterial strains, including the Janthinobacterium sp UV13 that grows in the Antarctic glaciers 17 . Violacein is a broad-spectrum bioactive compound with demonstrated anti-microbe, immunomodulatory and antioxidant properties [18][19][20][21][22] . In addition, several reports showed that violacein has anti-tumor properties on many cancer cell lines and types 18,20,21,[23][24][25][26] in a wide range of concentrations (submicromolar to micromolar) 17,26,27 , although the causes of differential cell susceptibility remain unknown.
The proposed mechanisms for violacein actions include the increase in reactive oxygen species, the activation of caspases or of mitochondrial and non-mitochondrial apoptotic pathways 26 . Other authors report that phosphorylation of the p38 mitogen-activated protein kinase or up regulation of the nuclear factor kappa-B (NFκB) pathway may have a role in the control of cancer cells [23][24][25]28 . Antioxidant properties 19 or immunomodulatory activities seen in animal models could also underlie violacein effects 18,22,29,30 . Violacein actions linked to inflammation include the maintaining of the balance between pro-and anti-inflammatory cytokines 22 and the modulation of tumor necrosis factor alpha (TNF-α) 22,31 or interleukin (IL) 6 30 levels. Interestingly, some of these actions include the inhibition of the proteolytic activity of matrix metalloproteinase (MMP) -2 and downregulation of the interactions that control cell migration and invasion in breast cancer cell lines 31 . In addition, MMP-2 and -9 can potentiate inflammatory pathways by converting inactive cytokines into their biologically active forms by either cleaving their membrane-attachments or their inactive zymogens [31][32][33] . In turn, pro-inflammatory cytokines may induce the conversion of catalytically inactive proMMP-2 and -9 into their active forms [31][32][33] , thus chronically sustaining and potentiating inflammatory cascades 32,33 .
It has been reported that MMP-2 and -9 increased expression along ALS progression [31][32][33] and MMP-9 has been proposed as a determinant of selective motor neuron vulnerability 33 . In addition, there is strong evidence about the involvement of both MMPs in ALS pathology either through direct neurotoxic effects or indirectly by eliciting cell death upon exacerbated degradation of extracellular matrix proteins 32 or by triggering glial cell activation and disturbing their homeostatic signaling [31][32][33] . Glial reactivity increases and sustains neuroinflammation 3,[6][7][8]10 . In turn, chronic neuroinflammation 8,11,31-33 may promote the emergence of reactive neurotoxic glial phenotypes that release pro-inflammatory molecules amplifying and perpetuating CNS damaging cascades 11,34 . Violacein could disrupt this deleterious feedback through different mechanisms and targets. On one hand, it can inhibit inflammatory cascades 18,22,29,30 likely abrogating some pathological responses such as the appearance of aberrant glial phenotypes. On the other hand, anti-proliferative properties and differential cytotoxicity [23][24][25]28 may allow the selective control of the emergent neurotoxic phenotypes that share common features with cancer cells as has been proposed for some CNS neurotoxic cells. Thus, the violacein protective actions could extend to the CNS.
To validate this hypothesis, we have proposed that violacein could selectively control the aberrant glial cells named as AbAs that emerged during disease progression 11,14 in the rat ALS model hSOD1G93A 5 , and also have tested if the treatment of these animals with violacein may result in protective effects.

Methods
All methods were performed in accordance with the relevant guidelines and regulations. Table 1 11 , passaged 4 times to eliminate contaminant cells and passaged 5-10 times to propagate AbAs 11,14 .
Rat glioma C6 cells 36 expressing astrocyte markers were selected as positive controls of the anti-tumor properties of the purified violacein used in this work 18,20,21,23 because of their shared lineage and previous comparative results with AbAs and astrocytes 11 .
Animal treatment and monitoring. 150 day-old Non-Tg and Tg 5 males received a weekly intraperitoneal injection (300 µl) of violacein (300 nmole/kg) 17 or vehicle (0.5% DMSO) until ~ 200 day-old. The dose was chosen because of the AbAs selective vulnerability over AA. A weekly administration was selected over two or three injections because of less animal handling and similar protective results.
Body weight was measured weekly up to 170 day-old and then every 2 days 5 . Body weight peak and first abnormal gait determined disease onset in Tg rats 5,16 .
Animal processing. At ~ 190-200 day-old, when a Tg-untreated rat reached the end stage, animals of each experimental condition were anesthetized (90:10 mg/Kg ketamine:xylazine) 16 , guillotined, lumbar spinal cord and soleus muscles dissected and weighted. A half of each sample was fixed by immersion in 4% (spinal cord) or 0.5% (soleus) PFA (24 h, 4 °C) and submitted to morphological analysis. The other half was collected in lysis buffer 44 , sonicated, spun, protein concentration determined (Bicinchoninic acid method) and frozen (− 80 °C) until analyzed by zymography or dot blot 44 . Some animals were transcardially perfused with 4% PFA 45 .

Dot blots.
Statistics. Data were expressed as mean ± SEM. Statistical tests used from GraphPad Prism 8.4 were ordinary one-way ANOVA and Tukey's post-hoc comparisons under normality, non-parametric Kruskal-Wallis test when normality failed and Long-rank Mantel Cox test for survival analysis. Statistical signification was determined at p < 0.05.

Methodology statement.
All authors confirm that all methods described in this manuscript were performed in accordance with ARRIVE guidelines.

Results
Aberrant astrocytes (AbAs) were selectively vulnerable to purified violacein. Violacein purified from Janthinobacterium sp. UV13 (Fig. 1a) selectively affected the viability of AbAs ( Fig. 1b and S1a-c) when compared with confluent cultures of Non-Tg adult astrocytes (AA, Fig. 1c and S1d) or with C6 rat glioma cells (Fig. 1c). The estimated IC 50 value for AbAs was ~ 175 nM whereas for AA and C6 cells it exceeded 1000 nM. In addition, AbAs viability in response to violacein did not differ between 150 and 300 nM for 24 h of treatment. Therefore, we have selected the range of 0-200 nM to analyze violacein effects on some functional parameters.
Log P value suggests that violacein might cross the blood brain barrier. Based on previous evidence reporting that violacein could cross LPS-injured blood brain barrier (BBB) 30 , here we predicted its ability to cross the intact BBB 38,39 . Results obtained showed that violacein has no ionization in the physiological pH range and that its log P value (2.24, calculated by RP-HPLC) is within the optimal range for BBB penetration ( Fig. 2a; Table 2) 38,39 . In addition, the molecular weight (343.3), the number of hydrogen bond donors (4) together with the sum of all nitrogens and oxygens (6) indicates that violacein´s molecular structure fulfills the thresholds of the Lipinski's rules for molecules able to cross the BBB 39 . www.nature.com/scientificreports/  Table 1. because of the AbAs selective vulnerability over AA (Fig. 1c-h), was applied (Fig. 2b) weekly due to less animal handling and similar results than divided doses. Animals that received vehicle were called as -untreated whereas those that received violacein were named as -treated. The treatment applied did not cause visible effects in Non-Tg animals including unchanged lifespan (p > 0.9999) (Fig. 2c) or movement (mark 5/5 in the disease progression scale 16 ). Regarding Tg rats, violacein effects on survival were marginally significant when comparing -untreated with -treated groups (p = 0.0616), and the survival median determined by Log-rank Mantel-Cox test passed from 198.5 to 215.0 day-old (Fig. 2d). Violacein did not influence the body weight of Non-Tg animals (Fig. 2e). Instead, it marginally delayed the onset of body weight loss 5 (p = 0.0900) in Tg-treated (~ 190 day-old, ι, p = 0.0300) versus -untreated animals (~ 180 dayold, #, p = 0.0374) (Fig. 2e). In addition, body weight differences in Tg-treated versus Tg-untreated rats reached statistical signification at 194 day-old (*, p = 0.0410) and lasted up to the end stage (Fig. 2e).
First abnormal gait (mark 4/5) was detected at ~ 174-180 day-old for Tg-untreated and Tg-treated rats, whereas paralysis of a hind limb (mark 3/5) was seen at 178 ± 5 day-old in Tg-untreated rats and was marginally statistically delayed until 185 ± 5 day-old (p = 0.0900) in Tg-treated rats. A marginally significant delay was also observed when both limbs became paralyzed (mark 2/5) in Tg-untreated (187 ± 5 day-old) versus Tg-treated rats (200 ± 7 day-old) (p = 0.0770), respectively. Dates of end stage (mark 1/5), recognized as the inability to recover right position when turned on their back, also was marginally significant delayed and occurred at 195 ± 9 and 210 ± 10 day-old for Tg-untreated versus -treated rats (p = 0.0900), respectively.
Once determined the effects of the protocol applied on the survival and disease progression of Tg animals, we decided to analyze the effects of the dose applied on some of the main ALS signs (muscle atrophy, NMJ integrity and spinal cord decreased motor neuron density and increased glial reactivity) when Tg-untreated animals reached the end stage. At that time, animals from each experimental condition were simultaneously processed and samples obtained to perform the planned approaches.
Masses of soleus muscles dissected at the same time for each experimental condition indicated lack of violacein effects in Non-Tg rats (640 ± 113 mg and 600 ± 100 mg) for -untreated versus treated animals, p = 0.3395. Instead, Tg-treated rats showed soleus muscles heavier than Tg-untreated brothers (420 ± 30 mg versus 320 ± 7 mg, p < 0.001) (Fig. 2f). Thus, soleus mass of Tg rats was preserved in ~ 20% upon violacein treatment. In accordance, besides maintaining muscle mass in general, violacein significantly improved the gross morphology of lower limb muscles in Tg-treated rats (Fig. 2g, bottom photograph ) when comparing with agematched Tg-untreated animals (Fig. 2g, upper photograph). It also caused that soleus muscles from Tg-treated rats had an appearance closer to those of Non-Tg animals (Fig. 2g).

Violacein partially protected muscle fibers and NMJ integrity in Tg rats.
To determine whether violacein helps to prevent the ALS abnormal muscular features that are recapitulated in hSOD1G93A models 5,51-57 ; H&E and trichrome stainings were made in soleus muscles from each experimental condition (Fig. 3). In comparison with Non-Tg conditions, the Tg-untreated muscle section evidenced clear pathological signs that include the presence of atrophic fibers and degranulating mastocytes (asterisks), reduced muscle area and increased collagen deposition (blue in mid and bottom images). Instead, the muscle section from Tgtreated animals exhibited better uniformity in fiber size and shape (Fig. 3a, upper images), less connective tissue (Fig. 3, mid and bottom images), and a general appearance closer to that of Non-Tg animals. Morphometric analysis showed similar distribution of cross sectional areas in Non-Tg rats, independent on violacein treatment (Fig. 3b left). However, in Tg rats, there was a shift in the frequency toward bigger cross sections when comparing -treated with -untreated animals (Fig. 3b right). Violacein also abrogated the ~ 30% increased number of muscle fibers per area in Tg-treated versus -untreated rats (p = 0.0099) (Fig. 3c), carrying the Tg-treated values similar to those of Non-Tg animals (p = 0.338). Regarding collagen, Tg-untreated and Tg-treated showed ~ 190% (p = 0.0007) and ~ 100% (p = 0.0175) increased collagen deposition than Non-Tg rats (Fig. 3d).
When analyzing violacein actions on the NMJ architecture and components (Fig. 4), there were no effects on Non-Tg animals. However, positive impact on Tg-treated animals included a significant preservation of the typical architecture (pretzel shape) [47][48][49]52 of the postsynaptic component labelled with αBungarotoxin (Btx) that had an appearance similar to that of Non-Tg rats. When comparing Tg-treated with-untreated conditions, violacein partially improved the presynaptic NMJ component as indicated by the higher SMI31 signal that labels nerve terminals (Fig. 4a). Upon violacein treatment, quantitation corroborated the preservation of Btx positive areas (p < 0.0001) and total synaptic areas (p < 0.0001) (Figs. 4b, c), as well as the enlarged SMI31 areas in Tg-treated versus Tg-untreated rats (p = 0.0084) (Fig. 4d). However, the coverage ratio was not modified in Tg-treated versus -untreated rats (p = 0.9400), and remained as ~ 50% of coverage shown in Non-Tg rats (Fig. 4e).
As MMP-2 and -9 play significant roles in neuroinflammation and were increased in ALS in close relation with progressive atrophy 32,33,48 , we have tested violacein effects on MMP-2 and -9 immunoreactivities and activities in soleus muscle samples (Fig. 5). Immunofluorescence (Fig. 5a) showed positive signals associated to sarcolemma in Non-Tg muscle sections but Tg rats also showed punctate immunoreactivity inside the muscle fibers. Violacein caused a general decrease in MMP-2 intracellular immunoreactivity in Tg -treated versus -untreated animals (p = 0.0012) (Fig. 5b). Similar findings were found in MMP-9 signals, with violacein abrogating the increase seen in Tg-untreated samples and leading their values closer to those of Non-Tg animals (p = 0.0455) (Fig. 5b).

Violacein controlled inflammation and glial reactivity in the spinal cord of Tg-treated rats. Dot
blots of Tg-untreated spinal cord homogenates showed increased levels of TNF-α (~ 124%, p = 0.0197), IL-1β (~ 175%, p < 0.0001) and IL6 (~ 103%, p = 0.0015) when compared with Non-Tg samples (Fig. 6a). Instead, no differences were found between Tg-treated and Non-Tg samples for TNF-α, IL-1β and IL6 as indicated by the respective p values (p = 0.9991; p = 0.1905 and p = 0.7290) (Fig. 6a chart). In addition, when comparing Tguntreated with -treated rats, violacein abrogated the increases in TNF-α (p = 0.0406), IL-1β (p = 0.0015) and IL6 In the Tg-treated section, TNF-α staining was very similar to that of Non-Tg samples. Cell nuclei were labelled with Hoechst 33342. (c) Z-stack images of immunofluorescences against IL-1β (green), GFAP (red) and Tomato lectin (white) from all experimental groups. Non-Tg sections showed low but specific IL-1β expression in motor neurons and blood vessels. No significant signals were seen in astrocytes or microglial cells (magenta arrows). Instead, in the Tg-untreated section, more IL-1β positive motor neurons were seen as well as extracellular positive signals together with exacerbated glial reactivity throughout the spinal cord parenchyma. Some reactive astrocytes may express low IL-1β levels as suggested by the presence of yellow/ orange spots. Exacerbated reactive microglia appeared with swollen cell bodies and negative to IL-1β. Upon violacein treatment, IL-1β signal in the Tg-treated section seemed restricted to motor neurons at lower levels than Tg-untreated samples and blood vessels remained highly positive. In this section, glial reactivity was dramatically lower than in Tg-untreated samples as evidenced by the presence of few hypertrophic astrocytes and microglial cells with signals of low/mid reactivity. Calibration: 50 µm (b) and 40 µm (c), respectively. www.nature.com/scientificreports/ (p = 0.0373) levels (Fig. 6a). Consistently, TNF-α immunoreactivity in the spinal cord sections from Tg-treated rats was low and restricted to the motor neuron cytoplasm, whereas Tg-untreated rats showed more intense signals in motor neurons and in the extracellular parenchyma (Fig. 6b). No evident changes attributed to violacein were found in the TNF-α immunoreactivity of Non-Tg spinal cord sections. Regarding IL-1β, immunofluorescences showed positive signals inside the motor neurons and in blood vessels, but not in astrocytes or microglia of Non-Tg and Tg-treated spinal cords (Fig. 6c). Instead, in Tg-untreated rats, IL-1β was increased in motor neurons and seemed detected at very low levels in reactive astrocytes but not in reactive microglia (Fig. 6c, red and white cells). IL-1β immunoreactivity in Tg-treated spinal cord exhibited Figure 7. Very low glial reactivity in the spinal cord of Tg-treated animals. (a) Representative confocal Z-stack images of S100β immunofluorescence (green) in each experimental condition. Whereas in Non-Tg images, S100β appeared surrounding the nucleus, there was an increased density of positive swollen cells with coarse processes in Tg-untreated rats. Violacein treatment modulated both the number and appearance of S100β positive cells in the Tg-treated condition. Hoechst 33342 stained cell nuclei (blue). (b) Representative Z-stack images of GFAP immunofluroescences in all experimental conditions. Non-Tg and Tg-treated images exhibited delicate positive cell processes whereas in Tg-untreated rats there was a predominance of GFAP positive swollen cells with gross and short positive processes. (c) Confocal Z-stack images of the microglial cell marker Iba1 in all experimental conditions. Images from Non-Tg and Tg-treated sections showed positive cells with thin processes whereas in the Tg-untreated image there was an increased signal mostly present in swollen cells with short and coarse processes. Blood vessels were delicately positive to Iba1 . For each marker analyzed ((a), (b) and (c)), white arrows point positive cells representative of each experimental condition. Calibration: 100 µm, 40 µm and 20 µm for (a), (b) and (c), respectively. (d) Quantitation of S100β, GFAP and Iba1 cellular density related to Hoechst 33342 positive cells indicated that Tg-untreated versus Non-Tg samples had increases in S100β (p < 0.0001), GFAP (p < 0.0001) and Iba1 (p < 0.0001), respectively. Violacein caused less but yet significant increased values in Tg-treated versus Non-Tg conditions as indicated by S100β (p < 0.0001), GFAP (p < 0.0001) and Iba1 (p < 0.0001), respectively. In all cases, quantitation was done by measuring the cells that co-localized with Hoechst 33342 positive cell nuclei. Hoechst 33342 signal was omitted to emphasize violacein modulation of GFAP and Iba1 immunoreactivity in (b) and (c), respectively. www.nature.com/scientificreports/ similar levels and localization than Non-Tg sections and was accompanied by significant lower glial reactivity as evidenced by GFAP and Tomato lectin staining (Fig. 6c). Exploration of glial reactivity upon violacein treatment in lumbar spinal cord sections (Fig. 7) evidenced that Non-Tg (independent on violacein treatment) and Tg-treated animals exhibited astrocytes and microglial cells without any sign of reactivity. Instead, prominent astrocytosis (Fig. 7a, b) and microgliosis (Fig. 7c) appeared in Tg-untreated animals as evidenced by abundant swollen cells with short coarse processes (arrows) as well as by increased density of S100β (~ 170%, p < 0.0001), GFAP (~ 150%, p < 0.0001) and Iba1 (~ 190%, p < 0.0001) positive cells when compared with Non-Tg values (Fig. 7d). Tg-treated condition not only showed astrocytes and microglial cells very similar to those from Non-Tg conditions (Fig. 7a-c) but also minor increases in cellular density of S100β (~ 70%, p < 0.0001), GFAP (~ 50%, p < 0.0001) and Iba1 (~ 70%, p < 0.0001), respectively.
The MMP-2 signal (red) inside and around the nucleus of motor neurons highly positive to SMI32 (green) was similar in Non-Tg and Tg-treated conditions. Instead, Tg-untreated sections showed extremely increased MMP-2 in motor neurons, mainly in those that lack SMI32 signal 49 , and in the rest of the spinal cord parenchyma including the highly reactive microglial cells that were recognized for the swollen bodies positive to Tomato lectin www.nature.com/scientificreports/ (white, Fig. 8c). Tg-treated sections also appeared with more SMI32 positive motor neurons than Tg-untreated condition. Quantitation of MMP-2 levels in Tg-treated versus Tg-untreated sections showed a minor percentage of positive motor neurons (from ~ 100% decreased, p = 0.0009) and lower signals (~ 250% decreased, p = 0.0001) (Fig. 8d). In addition, microglial reactivity was reduced in Tg-treated when compared with Tg-untreated rats (~ 200% decrease, p < 0.0001) (Fig. 8d). Nissl staining of motor neurons in IX Rexed lamina showed that violacein treatment preserved the morphology of motor neurons (Fig. 9a) and that seemed responsible for the increased motor neuron density in Tg-treated versus Tg-untreated rats (~ 50%, p = 0.0111), although the last one did not reach the values shown in Non-Tg animals (~ 25% less, p = 0.0024) (Fig. 9b). Remarkably, compared with Tg-untreated samples, Tg-treated showed much less small intensely labeled nuclei corresponding to glial cells surrounding motor neurons. Determination of the number of these small nuclei confirmed a decreased glial cell density when comparing Tg-treated with Tg-untreated samples (~ 70% less, p < 0.0001) (Fig. 9c). Thus, the different approaches employed allow suggesting that violacein not only controlled neuroinflammation, it partially preserved motor neurons, and also impaired glial cell reactivity, all enlarging its protective repertoire displayed in the ALS model employed.

Discussion
Natural products are attractive sources of therapeutic agents, with the majority of commercially available drugs derived from microorganisms, plants and animals because of their multiple beneficial actions [18][19][20][21][22]30 . Violacein is a quorum sensing metabolite extracted from different bacterial strains, including the Janthinobacterium sp. UV13 17 , that bears anti-inflammatory, anti-proliferative, anti-tumor and pro-apoptotic properties in experimental models of many diseases 17,18,21,[23][24][25] . Here, we have tested if violacein could be a potential therapeutic candidate to control aberrant invading glial phenotypes that have deleterious roles in neurodegenerative diseases. Next, we studied whether violacein could exhibit protective effects on the rat hSOD1G93A ALS model. Evidence obtained shows that violacein controlled AbAs, the aberrant glial cells that emerge and proliferate in the degenerating spinal cord of the hSOD1G93A (Tg) rats and do not suffer replicative senescence once in culture 11,13 . Present data reinforces the existing literature 14,16,51 about the vulnerability of these aberrant glial phenotypes 11,13 to anti-proliferative, anti-inflammatory or antioxidant compounds 18,[22][23][24][25][26][27][28][29][30] . However, violacein reduced AbAs viability at concentrations significantly lower than those employed in many cell lines including malignant glioma cells [21][22][23][24][25][26][27][28] . Remarkably, at the concentrations that affected AbAs, there were no significant effects on the viability and functional parameters of astrocytes obtained from Non-Tg adult rats. This evidencing a therapeutic concentration window that may allow to control the aberrant glial phenotypes without disturbing the cells responsible for the maintenance of CNS homeostasis 34 .
Moreover, violacein concentrations used to kill AbAs were much lower than those of the anti-proliferative drugs previously employed in vitro and in mouse and rat hSOD1G93A models 16,51 , suggesting an impressive capacity to control AbAs through many potential mechanisms. It has been reported that violacein induced apoptosis in some cancer cells through oxidative stress and imbalanced antioxidant defenses 19,23,25 . Our results showed that AbAs increased oxidative stress and reduced glutathione levels at 200 nM, a concentration close to the IC 50 (~ 175 nM) determined in viability assays. Thus, oxidative stress may be a potential mechanism underlying the violacein effects on AbAs survival. We have also found alterations in mitochondrial functionality and potential at 30 and 50 nM violacein, suggesting it aggravated the AbAs basal mitochondrial dysfunction altering even more their capacity to obtain energy efficiently 34 . It is also possible that violacein may increase the exacerbated ER stress seen in AbAs 14 or even impaired their alternative energetic sources 34 , that are already stressed by mitochondrial dysfunction. All underscoring the violacein potential capacity to kill AbAs via different underlying mechanisms that need to be precisely identified.
The evidence that indicates that violacein could cross the intact BBB 37-39 was other remarkable finding and represents a valuable advantage from the therapeutic perspective. In this regard, when assayed in animals, violacein caused marginally significant delays in the lifespan and in disease progression of Tg-treated rats. The p values resulting from the statistical analysis of both parameters suggest that modifications in the experimental paradigm employed, either by starting it earlier or by increasing the periodicity of administrations, could allow to reach a longer lifespan or slow disease progression. Although the possibility of increasing the dose of treatments needs to be analyzed in depth if desired to selectively preserve homeostatic astrocytes and likely the other neural cells, the dose employed in this work (300 nmole/Kg, ~ 0.1 mg/Kg) is much lower than those reported as toxic (7-10 mg/Kg) 25,30 . Then, doses might be carefully increased if the greater periodicity of administrations or the treatments in younger animals fail.
Weekly administrations of 300 nmole/Kg to Tg animals resulted in better preserved muscles and NMJs as well as more spinal motor neurons, suggesting that violacein has a potential capacity to slow disease progression 5 . Related to Tg-untreated animals, Tg-treated rats showed minor loss of muscle mass as reflected in soleus bigger fiber cross sections and decreased collagen areas as well as significantly preserved NMJ components. In ALS has been described an augmented production of extracellular matrix components, especially collagens, that will lead to cumulative fibrosis 52 , a pathological process that very recently has been suggested as a common trait that correlates with disease progression 53 . It also has been reported that fibrosis in skeletal muscles impairs function and regeneration and is a main cause of muscle weakness 54 . Therefore, the control of collagen in Tg-treated rats seems other violacein relevant effect.
Regarding NMJ, its dismantling is reported as an early event in ALS 52,55-57 . Recently, it also has been proposed that there is a long time window after the onset of NMJ loss in which motor neurons are not globally degenerating and preserve their capacity to re-innervate NMJs 56,57 . Apart from the protective effects on the NMJ components, upon violacein treatment, the coverage ratio was similar in Tg-treated and Tg-untreated animals, suggesting that the synaptic function was not preserved. Thus, a new experimental paradigm also needs to extend the positive effects to the NMJ in order to preserve not only its architecture but also its functionality.
The neuroprotective effects upon violacein treatment were also observed at level of the spinal cord of Tg rats, particularly by the decreased levels of the inflammatory cytokines that were predominantly expressed in motor neurons as well as by the inhibition of two glial reactivity hallmarks, the typical morphological changes and the increased cell proliferation. A higher density of motor neurons at level of IX Rexed laminae together with decreased number of MMP-2 positive motor neurons in Tg-treated rats were also observed. These results seemed in accordance with violacein neuro-immunomodulatory and anti-inflammatory properties 22,30 , including the reduction of systemic levels of the inflammatory cytokines TNF-α, IL-1β and IL6, as well as the inhibition of MMP-2 and -9 30,31 that are elevated in limb muscles and spinal cord from hSOD1G93A ALS models and patients 58 , this protecting motor neurons and modulating inflammation 32,33 .
Violacein control of the levels and signal locations of TNF-α, IL-1β and IL6 may impact not only in their roles as direct inflammatory effectors but also when acting as bridges between different pathological mechanisms. In this sense, it may interfere with the processes in which TNF-α links inflammation and excitotoxicity 59 , or IL-1β accelerates disease progression 60 or IL6 spreads inflammation into endothelial cells 61 . Therefore,violacein could modulate/inhibit multiplying damage effectors and targets in ALS.