Amyotrophic lateral sclerosis (ALS) is a fatal adult-onset neurodegenerative disorder that affects the motor neuromuscular system leading to progressive muscle weakness, paralysis and death by respiratory failure within 3–5 years after diagnosis [1, 2]. Until now, only two therapeutic agents, riluzole and edaravone, developed as potential antiglutamatergic and free radical scavenger respectively, have been approved by the Food and Drug Administration (FDA) for treatments aimed at slowing ALS disease progression. However, while riluzole marginally extends patient life by 2–3 months [3, 4], edaravone so far has only shown a slowing of the rate of decline of the ALSFRS-R score compared with placebo in a small subset of ALS patients at early disease stage [5]. This paucity of therapies is in part due to the multifactorial nature of ALS pathogenesis that involves both cell-autonomous and non-cell-autonomous processes [6]. The cause of ALS remains unknown except for the genetic forms that represent only about 15–20% of all cases and are clinically indistinguishable from the sporadic form of ALS. The affected genes are involved in various cellular functions, suggesting a large variety of possible disease mechanisms. Although more than 20 different genes have been linked to ALS [7], to date, the most reliable and informative disease model is the transgenic mouse overexpressing human superoxide dismutase 1 (hSOD1), which recapitulates the main pathological traits of ALS [8]. Evidence from these mice and from ALS patients indicates that multiple mechanisms are involved in the death of the motor neurons (MNs) and the loss of muscular innervations. They include dysfunctions intrinsic to motor neurons such as protein aggregation, impaired RNA processing and metabolism, mitochondrial dysfunction, oxidative stress, cytoskeletal alterations and impairment of axonal transport [6, 9]. Other extrinsic mechanisms involve additional cell types of the central nervous system (CNS) (such as reactive astroglia and microglia) or the peripheral immune system (like macrophages and T-cells) that may elicit neuroinflammatory processes contributing actively to motor neuron degeneration [10–12]. Motor neurons become vulnerable to these mechanisms in adult life, through the loss of compensatory pro-survival mechanisms, creating a disbalance between the signaling pathways of survival and death. Protective compensatory mechanisms may include the activation of pro-survival pathways such as the phosphatidylinositol-3 kinase (PI3K)-Akt pathway [13] and/or the induction of an anti-inflammatory environment [14]. It has been reported that neurotrophic factors, e.g., IGF1, GDNF and VEGF [15–17], are able to rescue motor neurons in SOD1 mutant mice through activation of the p-Akt pathway. On the other hand, a newly emerging hypothesis states that adaptive immunity through Th2 cells or and T regulatory cells (Treg cells) may initially activate the protective function of microglia or macrophages in the spinal cord, thereby providing a protective milieu to sustain motor neuron viability. As the disease progresses, the decline in Treg and the increase of Th1-mediated immunity lead to a shift of the microglia/macrophages toward a more proinflammmatory phenotype, which accelerates motor neuron degeneration [18]. Consistent with this hypothesis, we recently demonstrated that in C57BL/6-SOD1^G93A mice there was a prominent immune response in the motor neurons and sciatic nerves compared to a variant of the model in a different genetic background (129Sv-SOD1^G93A mice), and this was associated with a slower disease progression rate, preservation of myelin sheath morphology and a delayed neuromuscular junction (NMJ) denervation in the C57BL/6 model [19–21]. Therefore, sustaining the initial compensatory mechanisms, in addition to inhibiting the toxic mechanisms, is likely to be the best strategy to significantly slow down the disease progression in ALS.

RNS60 is a novel experimental drug containing oxygen-filled charge-stabilized nanostructures (CSNs) generated using modified Taylor-Couette-Poiseuille (TCP) flow under elevated oxygen pressure [22]. CSNs consist of an oxygen core surrounded by layers of positive and negative electrical charges [23]. The biological effects of RNS60 have been attributed to the physical properties of the CSNs, as a control solution containing a similar level of oxygen as RNS60 (50–60 ppm) and not subjected to TCP flow modification is not biologically effective [22]. The putative mechanisms of action of RNS60 encompass activation of pro-survival pathways, increase in mitochondrial biogenesis, protection of myelin forming cells, reduction of inflammation, and increase of Tregs [22, 24]. For example, RNS60 attenuates the expression of pro-inflammatory molecules in activated mouse microglia cells through the activation of the PI3-kinase-Akt pathway [22]. Additional studies have shown that RNS60 protects dopaminergic neurons and improves motor function in a mouse model of Parkinson disease by activating the pro-survival PI3K-Akt pathway and reducing microglial activation [23]. Moreover, it showed improvement of learning and memory in a transgenic model of Alzheimer disease by protecting hippocampal and cortical neurons, preventing tau phosphorylation and upregulating neuronal plasticity [25, 26]. Another positive effect of RNS60 is its ability to increase synaptic transmission by upregulating ATP synthesis in mitochondria [27, 28] and decrease fatigability of nerve stimulated muscles [29]. Finally, it is able to inhibit disease progression in multiple sclerosis rodent models by regulating the immune responses and presumably by promoting myelin maintenance and repair [24, 30–32].

Therefore, RNS60 seems to influence multiple cellular mechanisms that are also involved in ALS pathology suggesting that it may have a potential beneficial impact in this disease as well. Accordingly, in the present study we investigated the effect of RNS60 in both in vitro and in vivo models of familial ALS linked to SOD1^G93A mutation.

The current work demonstrates, for the first time, therapeutic effects of RNS60 in cellular and animal models of ALS. RNS60 is a novel experimental drug that contains charge-stabilized nanostructures consisting of an oxygen nanobubble core surrounded by an electrical double-layer at the liquid/gas interface. The biological effects of RNS60 have been attributed to the physical properties of the charge-stabilized nanostructures as ONS60, a control solution containing levels of oxygen comparable with RNS60 (50 to 60 ppm) but not subjected to TCP flow had not been effective in several earlier studies [22, 24–26]. We confirmed this observation in the model of motor neuron/glia co-cultures exposed to LPS (Additional file 1: Figure S1).

RNS60’s putative mechanism of action encompasses the activation of pro-survival pathways, increase in mitochondrial biogenesis, myelin maintenance and repair, reduction of inflammation, and increase of Tregs [22–32]. The current work demonstrates that RNS60 has a beneficial effect in cellular and animal models of ALS presumably through the same mechanisms of action (Fig. 11).

Initially, we found that RNS60 partially protects the motor neuron damage induced by LPS-activated microglia, suggesting an anti-inflammatory property of this potential therapeutic. This is consistent with previous studies showing that RNS60 suppressed the LPS-induced expression of iNOS in microglia [22]. Next, we demonstrated that RNS60 protected motor neurons (MN) in astrocytes-spinal neurons co-cultures carrying the SOD1^G93A mutation.

In addition, and most importantly, intraperitoneal treatment of C57BL/6-SOD1^G93A mice with RNS60 beginning at the onset of disease symptoms resulted in significantly improved outcome, which was characterized by protection of motor neurons, a delay in the onset of neuromuscular deficit and paralysis, and an increase of the mean survival interval from onset. Neuroprotective effects of RNS60 treatment were seen previously in other models of neuroinflammatory diseases, in particular the MPTP mouse model of PD [23]. The protective effects of RNS60 in this model, as well as in the case of microglial iNOS expression and NF-kB activation by LPS in vitro [22], were dependent on the activation of the class 1A PI3K-Akt pathway, which has emerged as a central mechanism of action of RNS60.

The effects seen in our study are likely mediated through the same pathway based on the following observations. Using the experimental paradigm of the in vitro experiments presented here, we have previously demonstrated that MN death was counteracted by Akt activation-induced motor neuron-specific overexpression of MyrAKT3 [13]. Involvement of Akt is further supported by the fact that RNS60 also preserved the number and length of neurites, which is consistent with a previous study reporting RNS60 to increase the number of neurites as well as the length, size, and maturation of dendritic spines in primary hippocampal neurons through the activation of the PI3K-Akt pathway [26]. The most direct evidence comes from the C57BL/6-SOD1^G93A mouse model, where we measured increased phosphorylation of Akt in lumbar spinal cord and sciatic nerves of mice treated with RNS60. Increased Akt phosphorylation was paralleled by phosphorylation (inactivation) of GSK3-β, further confirming the functional activation of Akt. The activation of this pathway also has previously been linked to motor neuronal protection in vivo. For example, we reported that the specific induction of Akt3 via lentiviral vector in motor neuron sustained cell viability via inhibition of GSK3β (phosphorylated at Ser9) [13]. Moreover, delivery of insulin-like growth factor 1 (IGF-1) to SOD1^G93A mice through motor neuron-targeted viral vectors [15], or by intrathecal infusion [52], partially protected motor neurons and delayed disease progression via activation Akt in the spinal cord. Similarly, intracerebroventricular injection of VEGF in SOD1^G93A rats also improved motor performance, prolonged survival, and raised p-Akt levels in the spinal cord [17, 53].

Unexpectedly, in RNS60-treated C57BL/6-SOD1^G93A mice, higher level of p-Akt was present not only in viable motor neurons but also by GFAP-positive reactive astrocytes in the ventral horn of lumbar spinal cord at the time when the motor neurons were protected, and muscle paralysis was delayed. This may appear counterintuitive because reactive astrocytosis is considered a hallmark of neuronal death. It has been demonstrated, however, that not all astrocytes are intrinsically neurotoxic for motor neurons as the total ablation of proliferating astrocytes did not slow the disease progression and motor neuron loss in SOD1 mutant mice [54]. Likewise, a focal degeneration of astrocytes has been reported in concomitance with the loss of motor neurons in SOD1^G93A mice, and their protection through metabotropic GluR5 receptor inhibition significantly delayed the onset of the disease [55]. These protective astrocytes appeared as spheroid GFAP-positive cells that existed together with typical reactive astrocytes [55]. Interestingly, many p-Akt-positive astrocytes in RNS60-treated mice appeared with similar spheroid morphology. This suggests that RNS60 may have prevented the degeneration of these focal astrocytes through the activation of p-Akt, thus promoting motor neuron survival. Since the level of the astrocyte specific glutamate transporter 1 (GLT1), a key factor in the control of excitotoxicity, is increased by the activation of the Akt pathway [56], this could be a potential mechanism for the MN protective effect of RNS60. The exact mechanism by which RNS60 activates Akt in astrocytes is presently unknown. Studies in vitro have demonstrated that IGF1 is a potent activator of Akt in astrocytes during oxidative stress and that this is a key mechanism for IGF1 to induce neuroprotection against oxidative stress injury [57]. Thus, it is possible that even in the presence of the pro-oxidant SOD1^G93A, the activated astrocytes in the spinal cord of RNS60-treated mice may acquire a neuroprotective function through the activation of Akt. Whether this may be mediated by IGF1 or other factors needs to be investigated.

In addition to the increase of astrocytes, the number of CD68-positive microglia in the spinal cord of RNS60-treated C57BL/6-SOD1^G93A mice was also increased. CD68 is expressed on phagocytic cells that can adopt a protective M2 phenotype in response to injection of anti-inflammatory IL-4 [58]. Interestingly, we detected a concomitant increase of IL-4 in the lumbar spinal cord of RNS60-treated mice. In addition, the reduction of Ym1, a marker of the M2 microglia phenotype in SOD1^G93A mice, was reverted by the treatment with RNS60 while the pro-inflammatory cytokines IL-1β and CCL2 remained unchanged. We also observed activation of Akt in CD68-positive microglial cells only in RNS60-treated mice. This is consistent with the observation that the PI3K/Akt pathway and its downstream targets have emerged as central regulators of M2 phenotype activation in macrophages/microglia [59]. For example, HDAC inhibitors have been shown to polarize microglia toward M2 by enhancing PI3K/Akt signaling in a model of severe traumatic brain injury preventing white matter damage [60]. Interestingly, Akt activation in macrophages was associated with reduced M1 markers and disease severity in a mouse model of EAE [61].

The neuroprotective role of M2 microglia is still debated but it seems that an increase of macrophagic activity in a specific phase of the disease may be associated with debris clearing and tissue remodeling [62, 63]. This process is particularly relevant at the level of the peripheral nervous system where immune cell infiltration may promote the phagocytic activity of Schwann cells and macrophages toward the degenerated axons to permit functional nerve regeneration and NMJ innervations [64–66]. We have recently demonstrated that the recruitment of immune cells in the peripheral nervous system delays muscle denervation and prolongs the lifespan in C57BL/6-SOD1^G93A mice [19]. Here, we found that RNS60 increased macrophage recruitment in tibialis anterior muscle and this correlated with reduced NMJ denervation.

Consistent with a protective role of immune cells, 129Sv-SOD1^G93A mice showing a fast disease progression are less prone to recruit macrophages and T cells in the peripheral sciatic nerve at the disease onset [20]. We therefore hypothesize that RNS60’s ineffectiveness in delaying onset and progression of muscular deficit and motor neuron degeneration in 129Sv-SOD1^G93A mice was due to the lack of response to RNS60-induced activation of the protective immune response. This is consistent with the lack of increase of CD68 immunoreactivity in the muscle tibialis and spinal cord in these mice compared to RNS60-treated C57BL/6 strain, and with a reduction of GFAP immunoreactivity in the spinal cord compared to NS-treated mice. This hypothesis is also in agreement with the observation of a lower macrophage recruitment in 129Sv mice, compared to C57BL/6 mice, after acute spinal cord injury [67]. In addition, the lower immune response in 129Sv mice is also reflected by the fact that these mice, in contrast to C57BL/6 mice, are resistant to EAE induced by myelin oligodendrocyte glycoprotein (MOG) [68]. We recently demonstrated that the comparison of 129Sv-SOD1^G93A and C57BL/6-SOD1^G93A mouse strains is useful to study disease susceptibility and to test novel therapeutic approaches aimed at addressing disease heterogeneity [69]. This may be relevant for the use of RNS60 in clinical trials of ALS patient cohorts that show heterogeneous disease severity. For example, a recent clinical trial with edaravone showed efficacy only in a well-defined group of patients [5]. A deeper biochemical and molecular analysis of these two mouse models might help to identify potential biomarkers to predict the positive or negative response to treatment in clinical trials.

In further support of potential protective immunomodulatory mechanisms, RNS60 increased Tregs in the spleen of C57BL/6-SOD1^G93A mice. Tregs are increased at early slow progressing stages of ALS and decreased during the rapidly progressing phase both in mice and in rapidly progressing ALS patients [18]. Passive transfer of endogenous Treg from ALS mice in the early disease stage into recipient ALS mice in a more progressed stage, without ex vivo activation, were shown to sustain IL-4 levels and M2 microglia, lengthen disease duration, and prolong survival [14, 18, 70]. We observed increased levels of IL-4 and a tendency to increase of the M2 microglia marker YM1 in the spinal cord of C57BL/6-SOD1^G93A mice treated with RNS60, despite the fact that the expression levels of CD4 T cells and FoxP3 were unchanged in this compartment. Thus, we hypothesize that peripheral Treg may contribute to the protective effect of RNS60 through a possible action at the neuromuscular level; however, the mechanism involved needs to be investigated.

Schwann cells are linked to the pathology in C57BL/6-SOD1^G93A mice and seem to contribute to the protective action of RNS60. We found a significant downregulation of p-Akt in the sciatic nerves of symptomatic C57BL/6-SOD1^G93A mice, which was associated with a marked decrease of both CNPase and MBP. This supports the prior finding of a key role of Akt in regulating the PNS myelination, enhancing membrane wrapping and myelin protein synthesis [71–75]. The downregulation of p-Akt, CNPase, and MBP was inhibited or reverted by treatment with RNS60. This is in line with the evidence that RNS60 retained the expression of myelin genes and blocked demyelination in CNS of EAE mice through the promotion of oligodendrocyte survival and OPC differentiation [30–32]. A more recent report showed a direct transcriptional effect of RNS60 on expression of myelin-related genes via the activation of PI3K-Akt-CREB [24].

RNS60 has been reported to modulate mitochondrial bioenergetics. In oligodendrocytes subjected to metabolic stress, RNS60 enhances the capacity of mitochondria to produce ATP [32]. In a neuronal cell line, primary dopaminergic neurons, and in the nigra of MPTP challenged mice, RNS60 was reported to upregulate mitochondrial biogenesis [49]. Alterations in the structure and function of mitochondria are a typical trait of ALS. In particular, the expression of mutant SOD1 is causally linked to alteration in their motility, dynamics, and turnover leading to energy deficit, calcium mishandling, and oxidative stress, which are all key contributors to motor neuron death [76]. In fact, it is known that mutant SOD1 accumulates in the outer membrane of mitochondria and causes the swelling of these membranes and consequent vacuole formation [43, 44, 77]. In support of an involvement of mitochondria in the effects of RNS60 observed in this study, we found a reduction of vacuoles immunostained for mutant hSOD1 in the spinal cord of RNS60-treated C57BL/6-SOD1^G93A mice compared to the NS-treated mice. That the protective effect of RNS60 could be partly mediated by action on mitochondria is also indicated by the fact that RNS60 is ineffective in 129Sv-SOD1^G93A mice. In fact, the pathological phenotype of these mice, unlike that of the C57BL/6-SOD1^G93A mice, is largely due to a marked mitochondrial dysfunction that could render them unresponsive to the protective action of RNS60.

How the mitochondrial effects of RNS60 are related to the modulation of the PI3K-Akt pathway is unclear at this time. Reports on RNS60-induced increased synaptic transmission by upregulated mitochondrial ATP synthesis in a squid giant synapse ex vivo model [27, 28] have raised the notion that delivery of the oxygen contained in the nanobubbles may directly feed into mitochondrial ATP production [27]. Of note though, RNS60 has been reported to increase mitochondrial biogenesis and ATP generation in a neuronal cell line as well as in primary dopaminergic neurons in a PI3K-dependent manner, lending support to an Akt-mediated process [49]. In other cell types, Akt has been shown to be involved in the regulation of bioenergetics. In cultured hepatocytes, for example, activated Akt translocates to the mitochondria where it phosphorylates subunits of ATP synthase and GSK3-β [78]. Akt-induced inhibition of GSK3-β in turn leads to activation of the mitochondrial enzyme pyruvate dehydrogenase (PDH), thereby boosting mitochondrial ATP production [78]. Further studies are needed to address these details in neuronal cells and immune cells and to determine whether a direct effect of delivered oxygen, an indirect effect through PI3K-Akt activation, or both are behind the increased mitochondrial ATP generation in response to RNS60.

Lastly, treatment with RNS60, an oxygen-containing drug, did not seem to cause any additional oxidative stress, and to the contrary, resulted signs of an anti-oxidant effect. Nrf2 is a basic region leucine-zipper transcription factor that binds to the antioxidant response element (ARE) and activates a battery of genes involved in the cellular antioxidant and anti-inflammatory defense, as well as mitochondrial biogenesis [79]. Reduction of Nrf2 expression levels in motor neurons and astrocytes has been demonstrated in ALS mouse models and patients [21, 79] and is thought to contribute to chronic motor neuron degeneration. In contrast, activation of Nrf2 is neuroprotective in a mouse model of ALS [80]. Thus, the prevention of Nrf2 reduction by RNS60 observed here is likely tied to an anti-oxidant response that is associated with its protective action on MNs. The reduction of NT levels seen in RNS60-treated C57BL/6-SOD1^G93A mice supports this conclusion.

In summary, the present study demonstrated that intraperitoneal treatment with RNS60 starting from onset of the disease delays the onset of neuromuscular deficit and paralysis, and increases the survival interval from onset in C57BL/6-SOD1^G93A mice by about 17%. This effect seems to be similar to or better than that obtained with other treatments targeting neuroinflammation, mitochondrial dysfunction, and oxidative stress [81], including that of edaravone, an antioxidant which has been recently approved by FDA for ALS treatment [82]. Notably RNS60, like edaravone, displayed a delay in motor decline with a trend to survival extension when the treatment started at the onset of symptoms. This is important in light of the translation to clinical practice where patients can only be treated at the diagnosis, often after the symptom onset. Moreover, RNS60 has a simple chemical composition (NaCl, oxygen, and water), which reduces the likelihood of side effects caused by metabolic by-products and may allow it to be safe when used in combination with other compounds. These significant preclinical findings, together with the excellent clinical safety profile, make RNS60 a promising candidate for ALS therapy and support further studies to unravel its therapeutic potential and molecular mechanism of action.