It is widely accepted that Glu [16, 18, 19, 45] and HCY [7, 8] induced neurodegeneration is caused by the permanent plasma membrane depolarization and cytosolic Ca2+ overload of neurons, which is the condition known as excitotoxic stress [for review 45]. Under the excitotoxic stress, the maintenance of ionic gradients on the plasma membrane by NKA consumes ATP and burdens mitochondria, which, as a consequence, is followed by the loss of mitochondrial inner membrane voltage. This particular situation could be illustrated by the disruption of the mitochondrial respiratory chain with CCCP and subsequent total loss of inner membrane voltage. Furthermore, water entry via ionotropic receptor channels to balance an accumulation of ions inside of neurons and a low Na+ membrane gradient can result in neuroinflammation with fast necrosis of neurons. In addition, a long-term elevation of cytosolic Ca2+ and mitochondrial dysfunction can trigger intracellular signaling cascades of apoptosis. Overall our data concerning the effects of Glu and HCY on neurons are consistent with this concept of excitotoxicity. We, therefore, further consider the mechanisms of neuroprotection by a subnanomolar concentration of ouabain.
Necrosis at short-term excitotoxic stress
The short-term (4 h) action of both Glu and HCY resulted in a considerable increase of necrotic neurons. This effect was abolished by the combined application of agonists with 0.1 nM or 1 nM ouabain. Presumably, in a portion of neurons, the plasma membrane depolarization and Ca2+ overload evoked by Glu or HCY engage energy-consuming ionic homeostasis, which leads to a rapid depletion of ATP resources required for the maintenance of ion balance and coupled water transport. In experiments, these processes could be monitored by the loss of mitochondrial inner membrane potential. The above explanation is consistent with the earlier observation that NMDAR induced osmotic imbalance causes cell swelling [46, 47] and rapid necrosis of neurons .
In sort-term experiments, HCY, as an excitotoxic agent, was less potent than Glu. This conclusion is supported by the observations that (1) the Ca2+ overload of neurons induced by HCY was less pronounced than as for Glu and (2) HCY caused a lesser drop of mitochondrial inner membrane voltage, which was found to be about 0.2 (∆φmit) of the total drop (φmit) caused by CCCP, than the ∆φmit value of about 0.7 found for Glu. As known, these agonists differ with respect to the activation of Glu receptors. While Glu activates all Glu receptors and does not cause desensitization of NMDARs, HCY activates only NMDARs and desensitizes those containing GluN2B subunits . The latter ones are widely expressed in extrasynaptic regions of the plasma membrane, and it is thought to provide a major contribution to neurodegeneration . Since rat cortical neurons express GluN2A and GluN2B [for review 48], the HCY effects are likely determined preferentially by the activation of GluN2A-containing NMDARs [12, 16, 20]. In contrast, Glu also activates GluN2B-containing NMDARs producing more Ca2+ to be accumulated in the cytoplasm and pronounced drop of the mitochondrial inner membrane potential, which is usually associated with mitochondrial swelling and neuronal cell death .
Subnanomolar concentrations of ouabain applied with either Glu or HCY considerably reduced the intracellular Ca2+ accumulation and related drop of the mitochondrial inner membrane potential, which rescued neurons from ATP deficit and reduced necrosis of neurons (Fig. 7a). This emergency rescue of neurons is most likely determined by an acceleration of Ca2+ export from neurons by NCXs, which, as demonstrated previously, are functionally regulated by ouabain through interaction with NKAs . Besides, ouabain considerably decreased Ca2+ transients generated in some neurons on agonist application, which coincides well with the previous data that ouabain lowers the frequency of spontaneous excitatory postsynaptic currents in cortical neurons . In addition amplitudes of Ca2+ transients are controlled by the local interplay of ouabain sensitive α3NKA, NCX and pre-membrane endoplasmatic reticulum [22, 50]. It seems unlikely that this mechanism of intracellular Ca2+ regulation is specific for NMDARs because, as shown earlier, an increase of intracellular [Ca2+] elicited by kainite activation of AMPA/kainite receptors is also well prevented by 1 nM ouabain .
Apoptosis at short-term excitotoxic stress
Short excitotoxic insults induced by Glu in our experiments caused the increase of p53, AIF, Bax, which are pro-apoptotic proteins, expression, and Cas-3 activation in neurons, which was accompanied by loss of Bcl-2. Presumably, Glu elicited elevation of intracellular [Ca2+], and loss of mitochondrial inner membrane voltage lead to the opening of Bcl-2 controlled permeability transition pore of mitochondria and releases of AIF, Cytochrome-C and other pro-apoptotic factors which initiate apoptosis . In consistence with previous studies, we may suggest that 4 h Glu insults cause an elevation of p53 expression resulting in a subsequent Bax activation , which may enhance NMDA elicited Ca2+ transients and contribute to deregulation of φmit . Previously it has been demonstrated that ouabain at nanomolar concentrations causes a reduction of p53 expression by activation of Src/mitogen-activated protein kinase (MAPK) signaling pathways upon its binding to the NKA . Therefore, ouabain induced signaling may prevent the up-regulation and mitochondrial recruitment of Bax [55, 56], which opposes Ca2+-induced mitochondrial dysfunction and apoptosis (Fig. 7b).
Cas-3 is protease-activated by both extrinsic and intrinsic (mitochondrial) apoptotic pathways  and acts at late irreversible stages of apoptosis. Prevention of Cas-3 activation by ouabain during 4 h excitotoxic insults most likely exhibits that this cardiotonic steroid may induce inactivation of most up-stream pro-apoptotic signaling cascades.
Regardless of the observations that Glu effects on Ca2+ accumulation and the loss of mitochondrial inner membrane voltage are more pronounced than the effects of HCY, both Glu and HCY caused a similar increase of neuronal apoptosis after 4 h treatment. Ouabain also produced similar protection against apoptosis and necrosis in the case of 4 h incubation with both HCY and Glu. By the use of specific inhibitors, we show that PKC, PKA or CaMKII are not involved in ouabain neuroprotection at 4 h excitotoxic stress. Most likely, anti-apoptotic mechanisms such as CREB phosphorylation by PKA or PKC  do not contribute to the ouabain effects in the short excitotoxic insults.
Apoptosis at long-term excitotoxic stress
The long-term 24 h treatment of neurons with Glu or HCY significantly increased the fraction of apoptotic but not necrotic neurons. Ouabain at 1 nM effectively prevented neuronal apoptosis against 24 h treatments by both Glu and HCY. Inhibition of CaMKII had no effect on ouabain mediated anti-apoptotic action. This differs ouabain induced neuroprotection from CGRP and forskolin elicited one [40, 58], which required CaMKII activation and cAMP-dependent PKA activation.
The inhibition of PKA or PKC blocked ouabain induced neuroprotection against HCY, but not against Glu neurotoxicity. This observation can be related to the NMDAR subtype selectivity of HCY because GluN2A, but not GluN2B NMDAR subunits, mostly contribute to HCY toxicity in cortical neurons [12, 14–16] and reflect extensive recruitment of different glutamate ionotropic and metabotropic receptors and transporters in glutamate neurotoxicity.
This indirectly supports the idea that HCY activates unique pro-apoptotic mechanisms that differ from those induced by Glu. For example during long-term action HCY, but not Glu induces GluN2A dependent sustained activation of ERK2 MAPK [8, 59], internalization of the Ca2+-impermeable GluA2-subunit of AMPA receptors and increase of intracellular [Ca2+] , which both cause permanent activation of p38 MAPK , downstream phosphorylation of Caspase-9 and apoptosis [for review 61]. Conversely, GluN2B containing extrasynaptic NMDARs over-activated by Glu induce only transient  ERK MAPK activation [17–19], while HCY causes sustained p38 MAPK activity . As a result, p38 MAPK inhibition can protect neurons from short-term, but not long-term activation of NMDARs . In addition in the case of HCY, the toxicity caused by both 4 h  and 24 h  treatments can be prevented by p38 MAPK block. Taken together, these observations favor the assumption that NKA signaling could modulate p38 MAPK via some PKA or PKC dependent pathways.
The role of PKC and PKA in ouabain effects against hyperhomocysteinemia
Previously we have demonstrated that neuronal apoptosis induced by 24 h treatment with HCY could be prevented by CGRP or forskolin, which both activated anti-apoptotic cAMP-dependent pathways. This type of neuroprotection depended mostly on PKA and CaMKII activity . Ouabain induced neuroprotection against 24 h HCY treatment of cortical neurons depended on PKA and PKC, but not CaMKII, which activity is mostly linked to the postsynaptic area [for review 64]. Therefore the long-term neuroprotection caused by ouabain must involve the apoptotic mechanisms, which are not yet activated at 4 h, but already contribute in neurotoxicity at 24 h excitotoxic stress. The ouabain effects against long-term action of HCY can involve PKA and PKC dependent effects on some pro-apoptotic cascade up-regulated by HCY, but not by Glu. Probably, HCY-specific pro-apoptotic cascade [8, 59] includes Caspase-9, and p38 MAPK, which both are susceptible to inactivation by phosphorylation with PKC [65, 66] (Fig. 7c), while Caspase-9 can be inactivated by PKA or PKC to interrupt apoptosis  (Fig. 7c, d). Subnanomolar ouabain activates PKC , which provides a possible explanation for observed ouabain induced neuroprotection in hyperhomocysteinemia.