Reduced sphingosine kinase-1 and enhanced sphingosine 1-phosphate lyase expression demonstrate deregulated sphingosine 1-phosphate signaling in Alzheimer’s disease

Alzheimer’s disease (AD) is a devastating neurodegenerative disorder which is characterized by two principal features: i) intracellular accumulation of hyperphosphorylated tau protein constituting neurofibrillary tangles (NFT) and neuropil threads; and ii) extracellular accumulation of β-amyloid (Aβ) peptide, major component of diffuse, focal and stellate deposits – the focal deposit constituting the core of the senile plaques [1]. These lesions are likely implicated in neuronal death and synaptic dysfunction associated with cerebral atrophy and cognitive impairment, characteristics of AD [2]. According to the stage of the disease, they can be confined to a specific location (i.e. entorhinal cortex, hippocampus) or be widely distributed in the brain [3‐5]. Even if definite causes are not clearly identified, several molecular mechanisms have been involved in the pathogenesis of AD: mutations of APP or of presenilins, epsilon 4 allele of ApoE, excessive Aβ production and/or reduced removal, tau protein abnormalities, oxidative stress and lipid metabolism alteration [6‐9].

Sphingolipids are ubiquitous lipid components of membranes that are metabolized to form signaling molecules associated with cellular processes important for health and disease [10, 11] (Figure 1). One of the most important of these metabolites is sphingosine 1-phosphate (S1P), which regulates pleiotropic biological activities such as proliferation, survival, migration, inflammation, or angiogenesis [10, 11]. S1P is generated from sphingosine, the backbone component of all sphingolipids and a pro-apoptotic sphingolipid [12] in a reaction mainly catalyzed by the sphingosine kinase isoform 1, SphK1 [13]. In turn, SpkK1 can be activated by multiple stimuli as IGF-1 signaling [14]. The balance between the levels of S1P and its metabolic precursors ceramide and sphingosine has been regarded as a switch that could determine whether a cell proliferates or dies [15]. S1P can be secreted and signal as a ligand of five high-affinity G protein-coupled receptors (GPCR), named S1P_1-5[16]. These receptors differ in their tissue distribution and the specific effect of S1P is driven by the predominance of the S1P receptor subtypes expressed [16]. Intracellular functions of S1P also exist with recent studies linking S1P to epigenetic regulation, calcium mobilization or activation of NF-κB [11, 17, 18]. Importantly, the agonist-induced S1P production as well as its downstream effects can be disrupted by inhibition of SphK1 gene expression or enzymatic activity illustrating that SphK1 plays a crucial role in the observed cellular effects played by S1P [19]. S1P can irreversibly be degraded into hexadecenal and ethanolamine phosphate by sphingosine 1-phosphate lyase (SPL) [20]. Interestingly, recent clinical observations have suggested an inverse relationship between SPL and SphK1 activities on the level of S1P in prostate cancer specimens implying that the overall increased S1P level commonly observed in cancer does not merely reflect overexpression of SphK1 activity, but could also be a consequence of loss of SPL expression [21].

In the brain, alteration of sphingolipid metabolism is believed to be essential for neuronal function as evidenced in a number of severe disorders besides AD including Niemann Pick disease, amyotrophic lateral sclerosis, Parkinson and AIDS dementia [24]. With regard to AD, most of the post-mortem studies have examined the level of ceramide and sphingosine, the pro-apoptotic precursors of sphingosine 1-phosphate, or enzymes responsible for their generation such as acid sphingomyelinase (which degrades sphingomyelin to ceramide) or acid ceramidase (which converts ceramide to sphingosine). For instance, acid sphingomyelinase activity and ceramide content are increased in the frontotemporal area [25‐28]. A positive correlation was found between acid sphingomyelinase activity and Aβ or phosphorylated tau in this region, suggesting that increased ceramide levels are associated with AD pathology [24]. The involvement of sphingosine is unclear with either increased [25, 29] or decreased [22, 26] content in AD as compared to normal brains. On the contrary, one clinical study has reported a decrease in S1P expression in AD tissues [22]. Interestingly, this decrease of S1P level was negatively correlated with Aβ and phosphorylated tau protein levels [22]. Moreover, ceramide level has been reported to be increased in CSF from patients with moderate severity of AD [30] whereas low levels of plasmatic ceramide have been reported in the early stages of memory impairment [31, 32]. Thus, the levels of these sphingolipids might be related to disease stage and represent an interesting pool of biomarkers for AD.

In cell culture models, a wealth of studies have firmly established the deleterious effect of ceramide on glial and neuronal cells exposed to Aβ peptides [26, 33, 34]. In addition to mediate the pro-apoptotic effect of Aβ, ceramide can also promote Aβ biogenesis by activating and stabilizing BACE-1 [35]. Conversely, S1P protects neuronal cells from apoptosis [36] notably in response to Aβ peptides [33]. Moreover, SphK1/S1P signaling was found to be a major transducer of two important growth factors, IGF-I and TGF-β1, whose neuroprotective effects against Aβ are well recognized [33, 37‐39]. With regard to the S1P receptors S1P_1-5[11], their contribution to AD has not yet been investigated. However, FTY720, an agonist of S1P_1,3,4,5, developed as an immunomodulatory drug and currently prescribed for multiple sclerosis, is able to restore passive avoidance memory in a rat model of AD as efficiently as Memantine, a finding supporting the existence of a direct action of this drug on neurons through S1P receptors [40].

Herein, we report for the first time the expression of SphK1 and SPL, the two main enzymes controlling the level of S1P, in frontal and entorhinal cortices of brains from AD patients, and their interaction with Aβ deposits distribution in the cortical layers. The expression of SphK1 and SPL was also assessed by western blot on brain tissue extracts together with SphK2, the minor isoform of sphingosine kinase, and S1P₁, the most important S1P receptor and IGF-1R, whose activation promotes activation of SphK1 and production of S1P.

While cancers are associated with alterations of cellular cycle inducing anarchic proliferation, neurodegenerative diseases are on the contrary associated with a cellular deregulation leading to neuronal death. As recently shown in a cohort study, elderly people suffering from cancer have a reduced risk of AD dementia and vice versa[45]. This observation underlines the existence of a relationship between these two major mechanisms of cellular function impairment. Interestingly, SphK1 overexpression leading to increase S1P signaling has been demonstrated to have an important role in cancer initiation, progression and resistance to therapeutics [13], whereas high levels of ceramide have been reported in AD brains [24]. Thus, in cancer and neurodegenerative diseases like AD, two opposite cellular fate outcomes could result from the imbalance of ceramide/S1P biostat [15]. Recently, Brizuela and coworkers reported that SphK1 expression was upregulated whereas SPL expression was downregulated in prostatic cancer. This original result showed that abnormal S1P level in prostatic malignant cells was not only related to overproduction by SphK1 but also to an important impairment of the elimination pathway provided by SPL [21]. In our study we reported the opposite situation, and showed for the first time that in AD, SphK1 expression was downregulated whereas SPL expression was upregulated. As a consequence of this deregulation, S1P levels should be decreased in cells and drive them to neurodegenerative processes.

In 2010, He and coworkers provided crucial information about the levels of ceramide and S1P in AD brains and assessed the expression level of enzymes implicated in ceramide/S1P metabolism but not SphK1 nor SPL [22]. The authors showed that Aβ was able to interact with sphingomyelinase and could induce in fine a decrease of S1P level. On the other hand, in vitro studies showed that Aβ, under oligomeric or fibrillary form, could trigger ceramide mediated apoptosis [22, 46]. The lack of information about SphK1 and SPL in AD and their direct involvement in S1P metabolism led us to investigate their expression within AD brains and to assess their possible relationship with Aβ deposits which represent one of the principal hallmarks of this disease.

Western blot analysis showed that SphK1 expression was reduced in AD brains compared to non demented controls. This observation supports the idea that neuropathologic processes related to AD and especially Aβ accumulation may induce deleterious effects on the expression of principal actors of the sphingosine 1-phosphate metabolism. SphK2 which is largely less implicated in the overall production of S1P than SphK1 did not show any particular modification of its expression in AD brains which is consistent with literature [47]. Morphologically, SphK1 expression was dramatically decreased within neurons populating fields in which the density of Aβ deposit was the highest. These fields corresponded predominately to cortical layers II, III where neuritic plaques are preferentially found [43] and extended to layer IV. This result was significant for neurons from entorhinal cortex that are very vulnerable [48], whereas neurons from frontal cortex seemed to be more resilient to Aβ toxicity. However, the packing density of total neurons in frontal and entorhinal cortices was correlated with the packing density of neurons with high expression of SphK1. As SphK1 expression is related to survival effects, its downregulation in AD could induce an opposite outcome. We previously showed that SphK1 activity was also reduced when cultured cells were exposed to fibrillary Aβ 25-35 [33]. All these results tend to demonstrate that Aβ deposits are directly involved in the reduction of S1P production by modulating the expression and the activity of SphK1 and could eventually shift the death/survival balance in favor of neurodegenerative processes.

Inversely, SPL which is the final enzyme in the sphingolipid degradative pathway controls the only exit point for sphingolipid intermediates. In addition to regulating S1P levels, SPL is a gatekeeper of a critical node of lipid metabolic flow [20]. In our study, Western Blot examination of SPL expression showed a higher level of this enzyme in AD brains compared to controls. This observation suggests that SPL could be highly deregulated in AD and is consistent with literature that reported upregulation of SPL mRNA expression in AD brains correlated to progression of dementia [28]. Our immunohistological study on ten AD cases confirmed these data and provided complementary information. Aβ deposits packing density was not correlated with high expression of SPL within neurons from frontal cortex but was positively correlated with high expression of SPL within neurons from entorhinal cortex. Notably, SPL deficiency leads to resistance against apoptosis induced by chemotherapy or nutriment starvation [21, 49]. In AD, two single nucleotide polymorphisms were detected in the sgpl1 gene in late-onset AD, which suggests that variation in sgpl1 expression and/or function may confer susceptibility to late-onset AD [50]. Our data indicates that increase of SPL expression in AD could be one of the consequences of Aβ accumulation. Hexadecenal and phospho-ethanolamine produced by SPL from S1P degradation have been reported to induce apoptosis, among other effects [51]. As suggested by Aguilar and Saba in 2012, SPL upregulation may be involved in accumulation of hexadecenal which could induce neurological and cognitive defects in some pathologies as for example in Sjögren-Larsson syndrome. This hypothesis suggests an important involvement of SPL deregulation in the pathogenesis of AD and leads to consider this enzyme as a promising therapeutic target.

SphK1 activation is modulated by many agonists including IGF-1 which induces the translocation of SphK1 to the plasma membrane [52]. In a previous study, we showed that the deleterious effect of Aβ exposition on SphK1 activity could be reversed by adjunction of IGF-1 to the culture medium [33]. Here we show that IGF-1R expression is dramatically reduced in frontal and hippocampal regions of AD cases compared to controls. This result is consistent with literature [23] and introduces a possible candidate for mediating signaling between Aβ and SphK1. Post-mortem studies on AD brains showed that IGF-1 deficiency and resistance is related to the stage of the disease and then could be considered as causal in the pathogenesis of AD [23]. IGF-1R impairments lead to brain amyloidosis in rodents and IGF-1R confers to cells the ability to reduce exogenously applied oligomers [53, 54]. This suggests that IGF-1R disorders are involved in Aβ accumulation and subsequent synaptic loss [55]. Here, we face a vicious circle in which Aβ induces a deregulation of IGF-1 signaling that in turn leads to overproduction of Aβ.

As S1P is able to trigger intracellular signaling pathways, it is also involved in an extracellular autocrine/paracrine signaling through five S1P receptors. Now well described, these receptors are involved in a wide range of signaling pathways including proliferation, survival, migration and cell-cell interactions [16]. Here we focused on S1P₁ as it is the most represented in brain and its activation can lead to an increase of survival/prevention of apoptosis through PI3K and Akt signaling [56]. The important decrease of S1P₁ expression in AD cases reported in our study could be related to a deregulation of S1P extracellular signaling induced by Aβ accumulation. This hypothesis is consistent with recent study which showed that FTY720, an agonist of S1P receptors with high affinity for S1P₁ was able to reverse behavioral impairment in rat model of AD [40].