Memantine prodrug as a new agent for Alzheimer’s Disease

Hydrogen sulphide has recently drawn much attention due to its potent anti-inflammatory and neuroprotective roles in brain functions. The purpose of the current study was to exploit these beneficial properties of H2S to design a new agent for the treatment of Alzheimer’s disease (AD). To pursue our aims, we replaced the free amine group of memantine with an isothiocyanate functionality as a putative H2S-donor moiety. The new chemical entity, named memit, was then tested in vitro to determine whether it retains the pharmacological profile of the “native drug”, while also providing a source of H2S in the CNS. Indeed, Memit showed the ability to release H2S through a cysteine-mediated mechanism, thus generating memantine. Moreover, the new hybrid molecule exerts protective effects against neuronal inflammation and induces a drastic fall in ROS production. In addition, memit was also able to reduce the Aβ(1-42) self-induced aggregation and exerted cytoprotective effect against Aβ oligomers-induced damage in both human neurons and rat microglia cells. Finally, similarly to memantine, the new compound promotes autophagy, a complex process required for cellular homeostasis in cell survival that results to be altered in neurodegenerative diseases. In conclusion, our study revealed that memit is a prodrug of memantine. Further in vivo studies will be necessary to fully investigate the synergic or cumulative effects due to the H2S-releasing moiety and the native drug.

In addition, H 2 S exhibits anti-inflammatory and antiapoptotic activities. Indeed, administration of NaHS in a rat model of AD proved to dramatically reduce the release of proinflammatory cytokines such as tumor necrosis factor-α (TNF-α), interleukin IL-1β and IL-6 induced by amyloid β-peptides. Moreover, it inhibits the upregulation of COX2 enzyme and the activation of the nuclear factor-κB (NF-kβ) in the hippocampus 16 , thus reiterating the high potential value of H 2 S in AD therapy. Further H 2 S neuroprotective mechanisms have been suggested: such as the reduction of lipopolysaccharide (LPS)-induced NO production in microglia via inhibition of p38-MAPK pathway 17 .
Taken together all these findings corroborate the functional involvement of H 2 S in neurodegenerative diseases 18 . Therefore, the restoration of the correct levels of endogenous H 2 S is an appealing novel approach for the development of new pharmacological therapies for AD, that are still lacking of an effective treatment. In this paper we decide to investigate the possibility to exploit the H 2 S properties to design a new agent for the treatment of AD, following the multitarget directed ligand strategy (MTDL). This approach is currently used for the design of new molecules to be used in several multifactorial pathologies such as cancer 19 , cardiovascular 20 and neurodegenerative diseases, including AD 21,22 .
Even though the combination of two or more pharmacophores into a single molecule has been extensively studied, the introduction of an H 2 S-donor moiety into the molecular scaffold of a "native" neuroprotective drug has not been explored yet 9,[23][24][25] . Consistently, to pursue our aim we selected memantine as "native drug" to combine appropriately with a H 2 S-generating moiety. Memantine is an excellent example of neuroprotective drug. It is a well-known uncompetitive voltage-dependent NMDA receptor antagonist, targeting principally NMDAR pathological hyperfunction 26 , and approved in 2003 by FDA as a treatment option for moderate to severe AD.
Starting from memantine, we replaced the free amine group with an isothiocyanate functionality as a putative H 2 S-donor moiety (Fig. 1). Indeed, recent studies showed that both synthetic and naturally occurring isothiocyanates generate H 2 S with a slow releasing rate 27,28 . Increasing studies have shown that H 2 S-generating agents for clinical use should retain some requirements such as a slow releasing rate, low and long-lasting concentration and favorable pharmacokinetics 29,30 . All these features are essential to take full advantage of the beneficial effects of H 2 S, thus avoiding the onset of toxicity.
Accordingly, a slow-releasing H 2 S donor memantine-like molecule, namely Memit, was developed and its neuroprotective properties were investigated in human neuronal-like cells.

Methods
Chemical synthesis. Commercial grade anhydrous solvents were used without further drying.
Commercially available chemicals were purchased from Sigma Aldrich and used without further purification. Evaporation was performed in vacuum (rotating evaporator). Anhydrous Na 2 SO 4 was used as the drying agent. Flash chromatography was performed on Merck 60 Å high-purity grade silica gel (0.40-63 μm). Reaction was followed by TLC, performed on Merck aluminum silica gel (60 F254) sheets. Spots were viewed under a UV lamp (254 nm). 1 H and 13 C NMR spectra were obtained using a Bruker Avance 400 spectrometer and were recorder at 400, and 101 MHz, respectively. Chemical shifts are reported in parts per million (ppm) d values, coupling constants J are reported in hertz (Hz). Data for 1 H NMR spectra are reported as follows: chemical shift (δ ppm), multiplicity, integration, coupling constant (Hz). Multiplicities are reported as follows: s = singlet, m = multiplet. Synthesis of memit was reported in Fig. 1. Briefly, to a stirred solution of Memantine (300 mg, mmol 1.67) in DCM (15 ml) and NaHCO 3 6% (15 ml) cooled at 0 °C, was added dropwise thiophosgene (1.28 ml, 16.7 mmol). The resulting mixture was stirred at 2 h at r.t, followed by TLC, then the aqueous phase was separated and extracted several times with DCM, dried over anhydrous Na 2 SO 4 and concentrated. The crude product was purified through flash chromatography (AcOEt/n-hexane 9:1 as the eluent) to get Memit as a clear oil 31  Cell cultures differentiation into neurons. H9-derived Neural Stem Cells (NSCs) were cultured as reported before 32,33 . Neurobasal serum-free medium with 5 µM retinoic acid was used to differentiate cells into neurons 32,33 .

Cellular model of inflammation and viability/proliferation assays in neuronal-like cells. The
compounds Memit and Memantine were diluted in DMSO. Neuronal-like cells, differentiated from H9-derived were incubated for 16 h with an inflammatory cocktail of LPS and TNF-α (50 ng/ml), as previously reported 32 . To verify the protective effects of Memit and Memantine, neuronal-like cells were treated with the two compounds for 24 or 72 h, then the samples were washed and incubated with LPS plus TNF-α for 16 h 32 or oligomeric Aβ 1-42 www.nature.com/scientificreports www.nature.com/scientificreports/ for 24 h. In selected experiments, NaHS was used as a positive control. Following incubation, cell viability was determined using the MTS assay 32 .
Cell proliferation/viability assays on rat microglia cells. Rat microglia cells were isolated from mixed cell culture obtained from Sprague-Dawley rat cortex, by gentle physical shaking, as described 34,35 . After isolation, cells were seeded into 96 well-plated with fresh culture media and pretreated with Memit or Memantine (10 µM) for 24 h; before washing and incubation with oligomeric Aβ 1-42 for 24 h. Following incubation time, cell viability was determined using the MTS assay 32 .
Ros production. ROS activity within the cells was determined using the fluorogenic dye 2′,7′-dichlorofluorescin diacetate (H2DCFDA, Molecular Probes, Invitrogen). NSCs were seeded in black 96-multiwell plates (5 × 10 3 cells/well). The cells were treated with Memit or Memantine at the concentration of 10 µM for 24 h; after the incubation time, cells were washed and incubated with LPS and TNF-α for 16 h. One hour prior to treatment completion, 50 μM H2DCFDA was added to the same media in the dark at 37 °C. As a positive control, H 2 O 2 was added at a concentration of 100 μM. The fluorescence intensity (excitation 485 nm and emission 520 nm) was normalized based on the number of cells stained with crystal violet 36 . In selected experiments, NaHS was used as a positive control.
Glioma cell culture and viability experiments. U-87MG cell line was obtained from the National Institute for Cancer Research of Genoa (Italy) and cultured as previously described 37 . The effects of compound treatment on U87MG cell viability were evaluated using the tetrazolium dye (MTT) colorimetric assay. Briefly, Memantine (10 μM) or Memit (10 μM) was added to 96-well plates containing 5,000 U-87MG cells/well. After incubation for 24 h, cell growth was measured by MTT (3-(4,5-dimethylthiazolyl-2)-2,-diphenyltetrazoliumbromide) proliferation assay. Absorbance at 570 nm was measured using a Bio-Rad model 550 microplate reader (BioRad Molecular Bioscience Group, Hercules, USA). The data represent the means of three independent experiments.
Western blot analysis. U-87MG cells were seeded in six-well plates at a density of 1 × 10 6 /well and grown to 70-80% of confluence with standard medium (DMEM-High Glucose). Cells were treated with vehicle (0.1% DMSO) or test compounds (i.e. 10 μM Memantine or Memit, and 1 μM Rapamycin) and incubated at 37 °C for 4 and 24 h. After washing with ice-cold PBS cells were lysed and proteins (30-40 μg) were separated on Criterion TGXTM gel (4-20%) and transferred to Immuno-PVDF membrane (Bio-Rad, Milan, Italy). The membranes were probed overnight at 4 °C with primary antibodies [1:1000, LC3A/B; p62; mTOR, p-mTOR; Akt, p-Akt(Ser 473); β-actin, Cell Signaling]. The primary antibodies were detected using appropriate secondary antibody. The peroxidase was detected using a chemiluminescent substrate (ECL, Perkin Elmer). Immunoreactive bands were quantified performing a densitometric analysis with Image Lab Software (Bio-Rad, Milan, Italy) and normalized to β-actin. Thioflavin T fluorescence assay. The ThT stock solution was prepared by adding 8 mg ThT to 10 mL phosphate buffer (10 mM phosphate, 150 mM NaCl, pH 7.0) and filtered through a 0.2 μm syringe filter. This stock solution should be stored in the dark and is stable for about one week. On the day of the analysis, the stock solution was diluted into PBS to obtain the concentration of 10 µM. The Aβ 1-42 oligomers were prepared by diluting the stock solution in PBS and the dilution was shaked for 48 hours at 37 °C. The cells were seeded in black 96-multiwell plate (3000 c/w) and treated with the compounds Memit or Memantine (10 µM) for 24 h. Following incubation, the samples were washed and incubated with Aβ 1-42 oligomers for 24 h. After this time, 200 µl of ThT 10 µM were added to each well in the dark. The ThT fluorescence intensity of each sample was recorded every 5 min using a spectrophotometer by excitation at 355 nm and emission 535 nm.
Determination of H 2 s release by amperometric assay. The characterization of the potential H 2 S-generating properties of Memit has been carried out by an amperometric approach, through the Apollo-4000 free radical analyzer (WPI) detector and H 2 S-selective mini-electrodes. 6 different recordings have been carried out at RT. Following the manufacturer's instructions, a "PBS buffer 10x" was prepared (NaH 2 PO 4 •H 2 O 1.28 g, Na 2 HPO 4 •12H 2 O 5.97 g, NaCl 43.88 g in 500 ml H 2 O) and stocked at 4 °C. Immediately before the experiments, www.nature.com/scientificreports www.nature.com/scientificreports/ the "PBS buffer 10x" was diluted using distilled water (1:10) to obtain the assay buffer and the pH adjusted to 7.4. The H 2 S-selective mini-electrode was equilibrated in 10 ml of the assay buffer, until the recovery of a stable baseline. Then, 100 µL of a DMSO solution of the tested H 2 S-releasing compounds (Memit and positive drug GYY4137) was added, at the final concentration of 1 mM; the final concentration of DMSO in the assay buffer was 1%. The generation of H 2 S was observed for 20 min. Preliminary experiments demonstrated that DMSO 1% did not produce any interference on the amperometric recording. When required by the experimental protocol, L-Cysteine (final concentration 4 mM) was added 10 min before the addition of tested compounds. L-Cysteine alone did not produce any amperometric response. The correct relationship between the amperometric currents (recorded in pA) and the corresponding concentrations of H 2 S was previously determined by suitable calibration curves, which were obtained by the use of the H 2 S donor NaHS (1-3-5-10 μM) at pH 4.0. The curves relative to the progressive increase of H 2 S vs time, following the incubation of the tested compounds, were analysed following the same procedure previously described 29 . statistical analysis. The data were analysed and elaborated by Graph-Pad Prism 6.0 (GraphPad Software Inc., San Diego, CA, USA). All data are presented as the means ± SEM. Statistical analysis was performed by one-way analysis of variance (ANOVA), followed by Student-Newman-Keuls multiple comparison post hoc test or by Bonferroni's corrected t-test for post-hoc pair-wise comparisons.

Results
Design and synthesis. Memit was developed as a novel NMDAR antagonist analogue of memantine provided of H 2 S-releasing properties. This new compound was efficiently synthesized by replacing the amine functionality of memantine with an isothiocyanate group through reaction with thiophosgene in heterogeneous phase (NaHCO 3 6%: DCM 1:1). evaluation of Memit H 2 s-releasing properties via Amperometric assay. As previously described 29 , a real-time determination of the H 2 S release was achieved by using a H 2 S-selective minielectrode incubating reference H 2 S donor (GYY4137) or Memit in the presence or in absence of L-Cysteine. As shown in Fig. 2, in the absence of L-Cys the H 2 S generation from both compounds was almost negligible, whereas after pre-incubation with L-Cys a time-dependent increasing generation of H 2 S was observed.

Memit induces neuroprotection in human neuronal cell. Neuronal-like cells, obtained from
H9-derived human neural stem cells (NSCs), were incubated with an inflammatory cocktail (lLPS plusTNF-α) in order to mimic in vitro neuroinflammation 39 . The effects of Memit (10 μM) and native drug memantine (10 μM) under physiological conditions and under inflammatory stress exposure were investigated by cellular viability experiments (Fig. 3A). The compounds did not alter cellular proliferation "per se" (data not shown). As expected, the inflammatory mixture significantly affected the proliferation of neuronal-like cells. Pre-treatment for 24 or 72 h of neuronal-like cells with the selected compounds significantly counteracted the decrease in cell proliferation elicited by the inflammatory insult. These results suggest that the new compound exerts neuro-protective effects in an experimental model of inflammation.
Memit reduces Ros production. To investigate the scavenger properties of memit, ROS levels were measured in neuronal-like cells under inflammatory conditions. Figure 3B shows that challenging neuronal-like cells with LPS-TNF-α significantly enhanced ROS accumulation. Pre-treatment with compound memit or with the native drug for 24 h almost completely counteracted inflammation-mediated ROS accumulation, thus suggesting that both compounds are able to prevent ROS accumulation. (1-40) self-induced aggregation. Accumulation of extracellular deposits of β-amyloid and abundant neurofibrillary tangles in the brain are correlated with neuron loss and progressive dementia 15 . Therefore, in order to evaluate the influence of memit on amyloid fibrils, a Th-T fluorescence assay was performed using NaHS as positive control 15,16,40 . Memit and Memantine were tested at 10 µM concentration; the results are reported in Fig. 4. As expected, Aβ1-42 significantly enhanced ThT-related fluorescence (165.5 ± 26.5% versus control). All the tested compounds exerted a significant inhibition of ThT-related www.nature.com/scientificreports www.nature.com/scientificreports/ fluorescence, thus indicating inhibition of Aβ1-42 self-aggregation. In particular Memit induced an Aβ1-42 self-aggregation inhibition comparable to its progenitor Memantine.

Memit induces autophagy in U-87MG cell line. Autophagy (ATG) is a complex and finely regulated
pathway, essential for the correct cellular physiology as it is involved in the degradation of long-lived cytoplasmic proteins, protein complexes, or damaged organelles. Impairments of the autophagic process are associated with several neurodegenerative disorders 41 , it is thus possible that the induction of ATG may be exploited as a strategy to assist neuronal survival. The mammalian target of Rapamycin (m-TOR) is an evolutionarily conserved serine-threonine kinase that is known to play a major role in the modulation of ATG 42,43 . Therefore, to explore the neuroprotective potential of Memit and Memantine we decided to evaluate their ability to induce ATG in U-87MG cells, a human glioblastoma cell line characterized by a weak ATG machinery due to the upregulation of m-TOR [44][45][46] . Expression of protein indicators of autophagy, such as LC3-II 47 , p62 48 and m-TOR, was detected The cells were challenged with Memit or memantine (10 μM) for 24 h or 72 h, and then treated LPS plus TNF-α for an additional 16 h. Cell proliferation was measured by MTS assay. The data are expressed as percentages relative to untreated cells (control), which were set at 100% (mean ± SEM, N = 3). (B) The cells were treated as in A. After treatment, ROS levels were quantified as described in the experimental section. The data are expressed as percentages relative to untreated cells (control), which were set at 100% (mean ± SEM, N = 3). **p < 0.01, ***p < 0.001 vs control; # p < 0.05, ## p < 0.01, ### p < 0.001 vs cells treated with LPS/TNF-α. www.nature.com/scientificreports www.nature.com/scientificreports/ by western blotting, using Rapamycin as a positive control. As shown in Fig. 5 a significant up-regulation of LC3II expression was observed after 4 h treatment with 10 μM Memit or Memantine (Fig. 5A), and also remained almost unchanged until 24 h treatment (Fig. 5B), suggesting that both Memit and Memantine are able to stimulate the ATG flux. Consistently, after 4 h treatment parallel decreased expressions of p62, which is degraded during autophagy, and p-mTOR were observed (Fig. 5C,D). Next, we decided to address whether the observed induction of autophagy in U-87MG cells caused by treatment with 10 μM Memit or Memantine was accomplished via inhibition of the survival PI3K/Akt/mTOR pathway 49 . To this end, we analyzed the phosphorylation status of Akt in U-87MG cells treated with test compounds. As shown in Fig. 5E, 4 h treatment with Memit and Memantine (10 μM) resulted also in a significant reduction of the phosphorylation levels of Akt-Ser 473 (*P < 0.05, **P < 0.01, versus vehicle treated cells), suggesting that, similarly to Rapamycin, both compounds may induce autophagy through the inhibition of the mTOR phosphorylation by the PI3K/AKT/mTOR pathway (full length Western blots for Fig. 5F are shown in Supplementary Fig. S1). In addition, cellular viability was determined using the MTT colorimetric assay. No significant alterations of cell viability were observed in U-87MG cells treated for 24 h with (10 μM) test compounds, as compared to vehicle (0.1% DMSO) (Fig. 6). evaluation of Memit binding activity on NMDAR. Memit was tested to assess its ability to inhibit native-target activity (i.e. NMDAR). Binding to NMDAR was assessed using a radioligand assay on cortex membranes based on the displacement of [3H]MK-801 38 , in the presence or absence of Cys (Table 1). Interestingly, while Memantine showed a significant displacement of [3H]MK-801 both in presence or in absence of Cys (Ki = 954.8 ± 74.2 nM and 328.8 ± 10.9 nM respectively), our H 2 S donor compound revealed a significant activity only in the presence of an organic source of H 2 S (Fig. 7 and Table 1). Importantly, the Ki value for memit (458.4 ± 77.7 nM) was comparable to that obtained for Memantine. These data demonstrate that memit could be considered a pro-drug of Memantine, released in the presence of Cys, and that the chemical modification did not alter the compound ability to bind NMDA.
Of note, Memit itself increased significantly its affinity for NMDA receptor in the presence of Cys (Fig. 7, Table 1), suggesting that Cys is involved in the regulation of the receptor affinity/functionality. Consistent with our hypothesis, crucial cysteine residues have been suggested to be involved in the modulation of NMDA protein Figure 5. Western blot quantification of LC3II/LC3I ratio, p62, mTOR and Akt as hallmarks of the degree of ATG activation in U87MG cells. Memantine (10 mM), Memit (10 mM) and Rapamycin (1 mM) induced increased LC3II/LC3I ratio at 4h (A) and 24h (B); a significant p62 degradation (C) and decreased p-mTOR/ mTOR ratio (D) and pAkt/Akt ratio (E). Results represent mean ± SEM of three different gels. *p < 0.05, **p < 0.01, ***p < 0.001 versus vehicle treated cells (Control). Representative western blots are shown in Panel F. www.nature.com/scientificreports www.nature.com/scientificreports/ activity and signaling events via other reactions of thiol groups 50 . In particular, the presence of cysteine in the binding buffer could involve reactions that affect ligand binding allosterically, as can be also observed by the Ki values of reference drug in the presence and in absence of Cys 50 .

Memit protects neuronal-like cells and microglia from Aβ(1-42) induced injury. Several studies
have proved that H 2 S and NMDA ligands have protective effects on Aβ-induced cellular injury 15,16 . On this basis, the effect of Memit on injured neurons was assessed. As expected, oligomeric Aβ caused a significant reduction of neuronal viability (Fig. 8, panel A). When cells were pre-challenged with Memit, a marked cyto-protective effect was observed (Fig. 8, panel A). Similar results were obtained probing cells with Memantine (Fig. 8, panel A).
Microglia, together with astrocytes, forms the main active immune defense of the CNS. During the inflammation process, induction of NF-κB occurs, accompanied by a release of inflammatory mediators such as TNF-α, IL-6 and nitrite ions. Notably, also levels of CBS and H 2 S result down-regulated 51,52 . Consistently, to further determine the neuroprotection elicited by our hybrid, we decided to evaluate effects of Memit and Memantine on rat microglia cells. Cells were pretreated with the two compounds (10 µM concentration) followed by incubating Aβ 1-42 oligomers. As depicted in Fig. 8 (panel B), pre-treatment with Memit and Memantine completely restored  www.nature.com/scientificreports www.nature.com/scientificreports/ cell proliferation. These data demonstrate that the compounds can protect both neuronal cells and microglia from Aβ-induced injury.

Discussion
The goal of the present work was to develop a new and improved molecule for the treatment of neurodegeneration by modifying the structure of a well-established drug that is already approved for therapeutic use. Thus, starting from memantine, a compound currently prescribed for the treatment of moderate-to-severe Alzheimer's disease, we created a new chemical entity, named memit, by introducing an isothiocyanate moiety capable of releasing H 2 S into its scaffold.
Although the pharmacological profile of memantine has been extensively studied (moderate affinity, uncompetitive, voltage-dependent, NMDA receptor antagonist that exhibits fast on/off kinetics 53 ), the conjunction with an H 2 S-donor moiety creates a new hybrid molecule whose properties need to be fully investigated. Putatively such a molecule should exert neuroprotective and anti-inflammatory effects through both hydrogen sulphideand NMDA-mediated mechanisms. This could result in a delaying of the neurodegeneration process.
Therefore, our initial studies were directed to assess whether memit still retains the pharmacological profile of the "native drug", while also being a source for H 2 S in the CNS. To evaluate its H 2 S donor properties we used an amperometric approach 29 , which provides a real-time determination of H 2 S-release, either in the absence or in the presence of an organic thiol, such as L-cysteine. The results of the amperometric analysis on memit showed a prolonged and persisting L-cysteine mediated release, similar to that exhibited by the reference slow H 2 S-releasing agent GYY4137, a phosphinodithioate derivative.
Next, we investigated the neuroprotective activity of memit in human neuronal-like cells and in rat microglia cells, with the former representing the main active immune defense of the CNS. In neuronal cells, the neuroprotection was evaluated following an LPS-TNFα insult, known to induce oxidative stress and consequently apoptosis and cell death. The new drug demonstrated to protect cells from both oxidative stress and cytotoxicity induced by LPS-TNFα. The observed effects can be attributed to both NMDA binding and H 2 S release. Indeed, hydrogen sulfide has been proven to attenuate neural injury and cytokine release in rats 54 and memantine has been shown to mitigate colchicine-induced inflammation in the hippocampus 55 .
In addition, Memit was able to reduce the Aβ(1-42) self-induced aggregation, with effects comparable to those of its progenitor memantine. Accordingly, memit and the native drug were also able to reduce Aβ oligomers-induced damage in rat microglia cells. Moreover, the observed reduction in Aβ accumulation elicited by Memit was shown to protect neuronal cells from the peptide-induced toxicity.
Consistent with our data, an over activation of NMDA receptor has been linked to abnormal Aβ deposition in AD and memantine has been demonstrated to decrease Aβ secretion in human cells or transgenic mice 56,57 .
Interestingly, H 2 S has been also proven to inhibit protein fibrillation of Aβ peptides by forming trisulfide bridges 58 and therefore it may contribute to Memit-elicited properties. www.nature.com/scientificreports www.nature.com/scientificreports/ The observed cytoprotection from Aβ-induced toxicity was then tested in microglia cells. Recent evidences have highlighted the pivotal role of these cells in different neuropathologies, including AD 59 . Notably, pharmacological studies suggest that promoting microglia polarization plays a key role in maintaining tissue homeostasis through different mechanisms including the suppression of inflammation 59 . In our hand, both memit and the native drug were also able to reduce Aβ oligomers-induced damage in rat microglia cells, suggesting the two drugs as promising protective agents for non-neuronal cells too.
Moreover, since a large body of evidence underlies the existence of a correlation between neurodegeneration and altered autophagic flux 60 , we also decided to investigate the ability of memit and memantine to induce autophagy in U87-MG cells. Our western blot analysis revealed that both compounds were able to significantly affect the expression of common autophagy markers such as LC3II, p62, mTOR and Akt, suggesting that autophagy could be an additional mechanism to strengthen their efficacy as neuroprotective drugs. These data are of particular relevance considering the role of AKT/mTOR pathway in the modulation of autophagy and the clearance of protein oligomers in neurodegenerative diseases 49 .
Finally, we also tested the ability of memit to bind to the NMDA receptor, the main target of memantine. Interestingly, we found that memit displayed an affinity comparable to that of the native drug only when cysteine was present in the assay medium, while no binding was observed in the absence of cysteine. These observations suggest that memit behaves as a prodrug, and that the binding is due to the production of memantine. In this sense, it has to be mentioned the reported facilitating effects of hydrogen-sulfide on NMDA receptors 61,62 . Nevertheless, the functional consequences of H 2 S-mediated NMDA receptor activation in neurons and stem cells is not strictly linked to neurotoxicity. For example, these facilitating actions are age-and receptor subunit-specific 61 and can finally promote synaptic plasticity. Furthermore, H 2 S has been demonstrated to enhance NMDA receptor-mediated currents and facilitate the induction of hippocampal long-term potentiation (LTP) 63 . Finally, activation of NMDA receptors increases proliferation and differentiation of hippocampal neural progenitor cells 64 and thus does not show toxic effects on differentiated cells.
Further studies will be needed to elucidate changes in NMDA receptor functionality that can be mediated by Memit-released hydrogen sulfide.

Conclusion
Unfortunately, to date the etiopathology of neurodegeneration is still not well understood and the therapies that are currently available lack in efficacy, as they are unable to completely block neuronal damage. Consequently, there is an enormous need for new therapeutic tools that are at least capable of slowing down the progression of the pathology. The work we presented here describes the synthesis and the preliminary in vitro evaluation of an innovative multi-functional prodrug. To the best of our knowledge, memit is the first compound capable of producing H 2 S in the CNS that was designed starting from the well-known drug memantine. The new compound we have developed has the potential of restoring H 2 S levels in the CNS and, like memantine, has the ability of inducing autophagy, thus eliciting neuroprotective effects. However, further in vivo investigations will be necessary to fully appreciate the synergistic or cumulative effects due to the H 2 S-releasing moiety and the native drug.