Crosstalk between sphingolipids and vitamin D3: potential role in the nervous system

Abstract

Sphingolipids are both structural and bioactive compounds. In particular, ceramide and sphingosine 1‐phosphate regulate cell fate, inflammation and excitability. 1‐α,25‐dihydroxyvitamin D3 (1,25(OH)₂D₃) is known to play an important physiological role in growth and differentiation in a variety of cell types, including neural cells, through genomic actions mediated by its specific receptor, and non‐genomic effects that result in the activation of specific signalling pathways. 1,25(OH)₂D₃ and sphingolipids, in particular sphingosine 1‐phosphate, share many common effectors, including calcium regulation, growth factors and inflammatory cytokines, but it is still not known whether they can act synergistically. Alterations in the signalling and concentrations of sphingolipids and 1,25(OH)₂D₃ have been found in neurodegenerative diseases and fingolimod, a structural analogue of sphingosine, has been approved for the treatment of multiple sclerosis. This review, after a brief description of the role of sphingolipids and 1,25(OH)₂D₃, will focus on the potential crosstalk between sphingolipids and 1,25(OH)₂D₃ in neural cells.

Abbreviations

1,25(OH)₂D₃

1‐α,25‐dihydroxyvitamin D3

AD

Alzheimer disease

APP

amyloid precursor protein

Aβ

amyloid β peptide

BDNF

brain‐derived neurotrophic factor

C1P

ceramide‐1‐phosphate

CDase

ceramidase

CERK

ceramide kinase

CERS

ceramide synthase

cGSN

cytoplasmatic gelsolin

cPLA₂

cytosolic phospholipase A₂

GBA1

lysosomal glucosylceramidase

GC

glucosylceramide

GCDase

glucosylceramidase

GDNF

glial‐derived neutrophic factor

HDAC

histone deacetylases

KO

knockout

LRM

lipid‐rich microdomains

MS

multiple sclerosis

NGF

nerve growth factor

NPC

Niemann–Pick disease type C

PD

Parkinson disease

pGSN

plasma gelsolin

S1P

sphingosine 1‐phosphate

SL

sphingolipid

SM

sphingomyelin

SMase

sphingomyelinase

aSMase

acid SMase

nSMase

neutral SMase

SPHK

sphingosine kinase

Tables of Links

TARGETS
GPCRs a	ERK2
S1P1 receptor	Haem oxygenase
S1P2 receptor	HDAC
S1P3 receptor	iNOS
S1P4 receptor	JNK
S1P5 receptor	3‐ketodihydrosphingosine reductase
Ligand‐gated ion channels b	Kinase suppressor of ras (KSR)
GluN3A	Lipid phosphate phosphatase
Voltage‐gated ion channels c	P38 MAPK
voltage‐gated calcium channels	PKA
Voltage‐gated potassium channels	PKC
Nuclear hormone receptors d	PLA2
Vitamin D receptor (VDR)	Sphingolipid Δ4‐desaturase
Enzymes e	Sphingomyelin synthase
Akt (PKB)	Sphingomyelin phosphodiesterase
AMPK	Sphingosine 1‐phosphate lyase
Adenylate cyclase	Sphingosine 1‐phosphate phosphatase
BACE1	SPHK1
Caspase 9	SPHK2
Cathepsin D	SPT
CBS	UDP‐glucose ceramide glucosyltransferase
Acid ceramidase	Src
Alkaline ceramidase	Other protein targets f
Neutral ceramidase	Bcl‐xL
CERK	COUP‐TF1
Ceramide synthase	Gα
CYP24A1	RAGE
CYP27B1	TNF‐α
ERK1

LIGANDS
Aβ	IL‐6
BDNF	Miglustat
CNTF	MPP+
Fingolimod	NADPH
FTY720‐phosphate	NGF
GDNF	NT‐3
Glutathione	Sphingosine (SPH)
IGF‐1	S1P
IL‐1α	Tau
IL‐1‐β	1,25‐dihydroxyvitamin D3

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Southan et al., 2016), and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 (^a,b,c,d,e,fAlexander et al., 2015a,b,c,d,e,f).

Roles of bioactive SLs in the nervous system

Sphingolipids (SLs) have long been regarded as inactive and stable structural components of the membrane, but some of them including ceramide, sphingosine, ceramide1‐phosphate (C1P) and sphingosine 1‐phosphate (S1P) are biologically active molecules. The cellular effect of SLs results from the combination of the effects of several interconvertible SLs (Figure 1), which are localized in distinct subcellular compartments and regulate distinct cellular processes and functions, including neural cell survival, apoptosis, autophagy, differentiation, migration, inflammation and neurotransmitter release (Table 1) (Colombaioni and Garcia‐Gil, 2004; Mencarelli and Martinez‐Martinez, 2013; Young et al., 2013; Shamseddine et al., 2015; Ghasemi et al., 2016).

Figure 1.

Metabolism of sphingolipids. (A). De novo synthesis of sphingolipids leads to the formation of ceramide (Cer) and sphingosine 1‐phosphate (S1P) through four reactions catalyzed by the serine palmitoyltransferase (SPT), which condenses palmitoyl‐CoA and serine into 3‐ketosphinganine (3‐KS), the 3‐ketosphinganine reductase (3‐KR), which generates sphinganine (DHSph), the (dihydro) ceramide synthase (CerS), which acylates sphinganine to dihydroceramide (DHCer), and the dihydroceramide desaturase (DES), which converts relatively inactive dihydroceramide to ceramide. The latter is converted to sphingosine (Sph) by ceramidase (CDase). Sphingosine can be converted to S1P by sphingosine kinase (SPHK) or to ceramide by CerS. The degradation of S1P is achieved by the reversible reaction catalyzed by the S1P phosphatase and the irreversible reaction catalyzed by S1P lyase, which produces hexadecenal and ethanolamine phosphate. The cell membrane constituent sphingomyelinase (SMase/SMPD) generates ceramide from sphingomyelin (SM). Phosphorylation of ceramide by ceramide kinase (CERK) generates ceramide 1‐phosphate (C1P). In the golgi, ceramide is converted to SM by SM synthase or to glucosylceramide (GC) by glucosylceramide synthase (GCS). GlcCer is then processed to more complex glycosphingolipids (not shown). Glucosylceramidase (GCDase; also called glucosylcerebrosidase) produces ceramide from glucosylCer. (B) S1P can be exported outside the cells by ABC transporters and the putative transporter Spinster 2 (spns2) and elicits autocrine or paracrine signalling by binding to and activating GPCRs (S1P_1–5 receptors; S1PRs). G‐proteins are composed of three subunits: α, β and γ and are classified as G(q), G(i/o), G(12/13) and G(s) depending on the function of their α subunits.

Table 1.

Role of SLs in proliferation, survival, differentiation, neurodegeneration, ischaemia and inflammation, Cer, ceramide

Cell/Tissue	Method	Effect	Mechanism	Reference
Proliferation
Neural progenitor cultured cells	A, exogenous S1P	↑ proliferation	–	Harada et al., 2004
Oligodendrocyte precursors	A,siS1P1R	↑ proliferation	S1P1R	Jung et al., 2007
Neuronal progenitors retina	A	↑ proliferation	S1P	Miranda et al., 2009
Human neuroblastoma cell	A, siCERK	↓ proliferation	↓ CERK expression	Bini et al., 2012
Neuronal progenitors retina	A	neuronal progenitors retina	C1P	Miranda et al., 2011
Survival
Photoreceptor	A	↑ survival	C1P	Miranda et al., 2011
SH‐5YSY, TNFα,	A,si CERK	↑ survival	↓ CERK expression	Barth et al., 2012
Photoreceptor	A	↑ survival	S1P	Miranda et al., 2009
SH‐SY5Y, MPP+	A	↑ survival	S1P	Pyszko and Strosznajder, 2014
Mature oligodendrocyte	A,si S1P5R	↑ survival	S1P5R/AKT	Jaillard et al., 2005
Drosophila mutants	B	↑ photoreceptor survival	CDase expression	Acharya et al., 2003
Retinitis pigmentosa mouse (eye)	B	↑ photoreceptor survival	SPT inhibition	Strettoi et al., 2010
Differentiation
Neuronal progenitors retina	A	↑ differentiation	S1P	Miranda et al., 2009
PC12	A	neurite retraction ↓differentiation	S1P/S1P2R	Toman et al., 2004
PC12, dorsal root ganglion neurons	A	↑ differentiation ↑neurite outgrowth	S1P/S1P1R	Toman et al., 2004
Oligodendrocyte precursor, S1P1R KO mouse	B	↓differentiation	↓S1P1R	Dukala and Soliven, 2016
Inflammation
Microglial cultured cells, brain	A, B	↓inflammation	C2‐Cer, ROS, MAPKs, PI3K/Akt, Jak/STAT	Jung et al., 2013
Murine ischaemic brain, Cultured neurons	A,B,a	↑inflammation, ↓inflammation	SPHK1inhibition, KO, SPHK2 inhibition, KO	Zheng et al., 2015
Neurodegeneration
Purkinje cell CERS1 mutants	A	↓cell number, ↓ neurite branching	↓C18Cer, ↑C16Cer, ↑Sph, ↑dhSph, ↑dhS1P, ↑S1P	Zhao et al., 2011
Purkinje cell, aSMase KO	A	↑ death	↑SM	Horinouchi et al., 1995
Neuroblastoma cell	A,siCERS2	cell growth arrest, increased autophagy	↓CERS2, ↑C16Cer, ↓ C24Cer,	Spassieva et al., 2009
Ischaemia/Hypoxia
Rat brain,glia, chronic hypoxia	B	–	↑ Cer, ↑ aSMase, ↓GCS	Ohtani et al., 2004
Hypoxia/reoxygenation, NT‐2 neuronal precursor cells	A	–	↑ C14Cer, ↑C16 Cer, ↑ SMase, ↑CERS5	Jin et al., 2008
Ischaemia, rat hippocampus astrocyte	B	↑TNFα, IL1, IL6	↑ nSMase	Gu et al., 2013

The intracellular levels of these bioactive SLs are fine‐tuned, and alterations in the SL profile in the nervous system contribute to the development of neurological and neuroinflammatory diseases such as photoreceptor degeneration, retinitis pigmentosa, Alzheimer disease (AD), Parkinson disease (PD), multiple sclerosis (MS) (Table 2) and major depression, Huntington disease and epilepsy (Acharya et al., 2003; Desplats et al., 2007; Haughey et al., 2010; Mielke and Lyketsos, 2010; Strettoi et al., 2010; Grimm et al., 2013; Halmer et al., 2014; Pyszko and Strosznajder, 2014; Vanni et al., 2014; Gulbins et al., 2015). Moreover, inherited defects of both the synthesis and catabolism of SLs cause varying degrees of dysfunction of the CNS, such as in inherited sensory and autonomic neuropathy, Niemann–Pick disease type A and B and lisosomal storage disorders (Sabourdy et al., 2015).

Table 2.

Involvement of SLs in neurodegenerative diseases

	Method	Effect	Mechanism	Reference
AD
Cortical neurons, Aβ 1–42	A	↑exosome release, ↑apoptosis	↑nSMase,	Wang et al., 2012
Human primary neurons, Aβ	A	↑ apoptosis	↑ nSMase	Jana and Pahan, 2004
Cultured hippocampal neurons, Aβ	B	↑apoptosis	↑ C18Cer, C24Cer	Cutler et al., 2004
Presenilin knock‐in mouse, primary cultured astrocytes	A	↑cell death	↑ C20Cer, C24Cer, ↑CERS1, ↑CERS4	Wang et al., 2008
Astrocytes, frontal cortex, CSF, patients	B	–	↑Cer	Satoi et al., 2005
White matter temporal cortex, white and grey matter, patients	B	–	↑ C24:1Cer, ↓sulfatides	Han et al., 2002
Medial frontal gyrus, patients	B	–	↑ C24:0 Cer	Cutler et al., 2004
Cultured neurons, Aβ, brain, patients	A, B	↑Apoptosis	↑a,nSMase, ↑acid CDase, ↑aSMase, ↑acid CDase, ↑Cer, ↓SM	He et al., 2010
Entorhinal cortex,patients, Hippocampus, temporal grey matter	B	amyloid deposit	↓SPHK1, ↓S1P1R, ↑SPL, ↓S1P, ↓SPHK1,2, ↑C16:0 Cer	Ceccom et al., 2014
Brodman areas 46,10,20 patients,	B	–	↑ PPAP2B, ↑ SPL, ↓acid CDase, ↑CERS1,2, ↓CERS6	Katsel et al., 2007
PD
Anterior cingulate cortex, patients	B	–	↓total Cer, ↓SM, ↑CERS1 expression	Abbott et al., 2014
Anterior cingulate cortex, patients	B	↑ autophagy ↑α‐synuclein	↓glucosylCDase, ↓Cer,	Murphy et al., 2014
MS
White matter, patients	B	–	↓S1P, ↑Sph, ↑C16Cer, ↑C18Cer	Qin et al., 2010
Reactive astrocytes, patients	B	–	↑ C18:0 Cer	Kim et al., 2012
NPC	–	–	–	–
NPC−/− mouse brain	B	–	↑glucosylCer, galactosylCer, glucosylSph, GM2, GM3	Marques et al., 2015
NPC−/− mouse brain with GBA2 deletion,	B	improved motor coordination	↑glucosylCer, glucosylSph, =cholesterol, =gangliosides	Marques et al., 2015
NPC1−/− mouse brain, miglustat Patients, plasma Patients + miglustat, plasma patients + miglustat, CSF	B	–	GCS inhibition, ↑ monohexylCer, ↑ monohexylCer, ↑C16:0Cer, ↓Sph, ↑S1P, ↓Cer, ↓GM1, ↓GM3, ↑monohexylCer	Fan et al., 2013
NPC1−/− mouse brain, miglustat,	B	↑synaptic plasticity	GCS inhibition	D'Arcangelo et al., 2016
Purkinje neurons from NPC1−/− cat, miglustat	A/B	↑survival	GCS inhibition	Stein et al., 2012
NPC1−/− cat, miglustat	B	↑lifespan, ↓ motor deficit	GCS inhibition	Stein et al., 2012
Lymphocytes NPC patients, miglustat	A	correction of abnormal lipid trafficking	GCS inhibition	Lachmann et al., 2004

The table illustrates some examples of the involvement of SLs in neuroinflammation, ischaemia and in neurogenerative diseases such as PD, MS, AD and NPC, indicating the experimental method used (A: in vitro; B, in vivo; si: siRNA; Cer, ceramide; GBA2, non‐lysosomal glucosylCDase). The alterations in SL concentration or in expression/activity of enzymes involved in SL metabolism are listed under Mechanism.

The modulatory role of ceramide in growth and 1‐α,25‐dihydroxyvitamin D3 [1,25(OH)₂D₃]‐induced differentiation was first reported in leukaemic HL60 cells (Okazaki et al., 1989; Bielawska et al., 1992). Twenty years ago, it was proposed that the ratio of the intracellular content of S1P and ceramide was a major determinant of cell fate (Cuvillier et al., 1996): S1P enhances growth and survival, whereas its precursors (ceramide and sphingosine) promote growth arrest and cell death (Colombaioni and Garcia‐Gil, 2004; Mencarelli and Martinez‐Martinez, 2013; Shamseddine et al., 2015; Ghasemi et al., 2016). However, there is increasing evidence showing that the ceramides containing specific acyl chain lengths (ceramide species) have different functions (Ben‐David and Futerman, 2010; Hannun and Obeid, 2011): C18:0‐ceramide is synthesized by ceramide synthase 1 (CERS1), an isoenzyme abundant in the brain, and it has been suggested to act as a protective factor, because a lack of CERS1 caused neural death in mice cerebellum and impaired motor coordination (Zhao et al., 2011; Ginkel et al., 2012). However, CERS1 elimination decreases ganglioside levels, and this might be one of the causes of neural cell death in mice (Ginkel et al., 2012). Moreover, serum deprivation‐induced apoptosis in embryonic hippocampal cells induces an increase in C16:0‐ceramide and a decrease in C24:0‐ceramide (Garcia‐Gil et al., 2015). It is worth noting that compensatory mechanisms can occur following gene knockout (KO) or when the protein expression of a substance is reduced by administration of its siRNA. For example, the treatment of neuroblastoma cells with CERS2 siRNA results in an increase in the expression of CERS5 and CERS6 with a reduction in C24‐ceramide and sphingomyelin (SM) and increase in C14‐ and C16‐ceramide levels (Spassieva et al., 2009).

Other facts, in addition to the different acyl chain composition of the ceramides, increase the complexity of studying the role of SLs in cell fate: (i) S1P acts not only intracellularly but also as ligand of specific GPCRs (Maceyka et al., 2012) (Figure 1B); (ii) other SL metabolites, such as C1P, can also mediate apoptosis or proliferation depending on the cell type (Miranda et al., 2011; Bini et al., 2012; Presa et al., 2016); and (iii) the synthesis of SL and signalling can occur in different cellular compartments (Newton et al., 2015).

Ceramide and C1P

The apoptotic role of ceramide in the nervous system has been extensively reviewed (Colombaioni and Garcia‐Gil, 2004; Mencarelli and Martinez‐Martinez, 2013; Shamseddine et al., 2015). Ceramide is also involved in the control of autophagy (Daido et al., 2004; Spassieva et al., 2009), differentiation (Riboni et al., 1995), inflammation (Gu et al., 2013) and exosome release (Trajkovic et al., 2008; Wang et al., 2012). The generation of ceramide by activation of neutral SMase2 (nSMase2) is associated with an increase in dopamine uptake (Kim et al., 2010) and this modulates excitatory postsynaptic currents by controlling the insertion and clustering of NMDA receptors (Wheeler et al. 2009). Moreover, Caenorhabditis elegans mutants lacking ceramide synthase have been found to have defects in synaptic transmission and in synaptic vesicle cycling (Chan and Sieburth, 2012). Ceramide directly regulates the activity of several enzymes including cathepsin D, phospholipase A₂, kinase suppressor of Ras, ceramide‐activated protein serine–threonine phosphatases 1 and 2A (PP1 and PP2A), PKC isoforms and ion channels, such as the potassium channel K_v1.3 (Bock et al., 2003). It is also able to form channels in mitochondria, which are involved in the release of pro‐apoptotic factors (Colombini, 2016). It directly inhibits mitochondrial complex III and increases the generation of ROS (García‐Ruiz et al., 1997). Ceramide inhibits the Akt signalling pathway, stimulates the stress‐activated kinase JNK and up‐regulates the apoptosis‐promoting variants Bcl‐xS and caspase‐9, while correspondingly down‐regulating the antiapoptotic variants Bcl‐xL and caspase‐9b (Chalfant et al., 2002).

There is accumulating evidence supporting the involvement of ceramide in the modulation of neural plasticity. For example, spatial memory and extinction learning are impaired when sphingomyelinase 2 (SMase2) is inhibited or ceramide levels are reduced (Tabatadze et al., 2010; Carrasco et al., 2012; Huston et al., 2016). Furthermore, genetic deletion of CERS1 is associated with deficits in motor learning and spatial working memory, as well as reduced anxiety (Ginkel et al., 2012).

Astrocytes are mediators of CNS responsiveness to inflammation and injury (Claycomb et al., 2013). They display increased ceramide following ischaemia/reperfusion with nSMase2‐dependent generation of the pro‐inflammatory cytokines TNF‐α, IL‐1 and IL‐6 (Gu et al., 2013). In neural stem and progenitor cells of the developing brain, ceramide influences cell polarity, motility and apoptosis (Bieberich, 2012) and induces ciliogenesis, a critical step in differentiation (He et al., 2014).

In contrast, the phosphorylated form of ceramide, C1P, induces proliferation and promotes the survival and differentiation of photoreceptors in rat retina neuronal cultures (Miranda et al., 2011), while inhibition or down‐regulation of ceramide kinase (CERK), which appears to be the only enzyme responsible for its synthesis, results in a decreased proliferation of human neuroblastoma cells (Bini et al., 2012). Moreover, C1P directly binds and activates α‐type cytosolic phospholipase A₂ (cPLA₂), so stimulating arachidonic acid release (Pettus et al., 2004). Activation of cPLA₂ by C1P induces spinal neuronal death (Liu et al., 2014), while treatment of SH‐SY5Y cells with TNFα increases CERK activity. Depleting CERK activity blocks NADPH oxidase activation and eicosanoid biosynthesis and restores neuronal viability in the presence of TNFα (Barth et al., 2012). A CERK‐null mouse has been generated. Although CERK is highly expressed in Purkinje cells, this mouse did not display histological abnormalities or impairments in motor coordination, but their emotional behaviour was slightly affected (Mitsutake et al., 2007). Outside the nervous system, C1P stimulates the migration of macrophages via a specific plasma membrane receptor coupled to Gi proteins (Presa et al., 2016), and it is released from damaged myocardial cells possibly leading to the recruitment of stem/progenitor cells to damaged organs (Kim et al., 2013). Whether C1P is also released from the injured nervous system or whether it induces migration in neural stem cells is not known.

S1P

S1P modulates cell survival (Edsall et al., 1997), proliferation (Harada et al., 2004; Miranda et al., 2009), differentiation (Toman et al., 2004; Miranda et al., 2009) and migration (Novgorodov et al., 2007; Alfonso et al., 2015), calcium homeostasis (Sato et al., 2000; Giussani et al., 2007; Hagen et al., 2011), neurite retraction (Toman et al., 2004), angiogenic vascular maturation (Liu et al., 2000; Mizugishi et al., 2005) and cytoskeleton dynamics (Postma et al., 1996; Toman et al., 2004; Jaillard et al., 2005) (for recent reviews, see Bieberich 2012; Maceyka et al., 2012, Proia and Hla, 2015, Ghasemi et al., 2016). In addition, it is able to modulate excitability (Li et al., 2015a) by increasing glutamate release (Kajimoto et al., 2007; Kanno et al., 2010) and by regulating endocytosis and exocytosis (Chan and Sieburth, 2012; Shen et al., 2014; Riganti et al., 2016).

Regarding intracellular effects of S1P, S1P induces calcium release from the ER, inhibits histone deacetylases (HDAC), acts as a cofactor necessary for the E3 ligase activity of TNF receptor‐associated factor 2, activates recombinant human PKCδ and binds the mitochondrial protein prohibitin 2, a highly conserved protein that regulates mitochondrial assembly and function (Maceyka et al., 2012) (Figure 2). In agreement with the role of S1P in proliferation, sphingosine kinase 1(SPHK1) is overexpressed, while S1P lyase is often deleted in human cancers, including glioblastoma (Steck et al., 1995; van Brocklyn et al., 2005). SPHK1 overexpression is associated with resistance to chemotherapeutic drugs and to a poor prognosis (van Brocklyn et al., 2005).

Figure 2.

Biological function of S1P/S1P receptor signalling. Phosphorylation activates SphK1 and promotes its translocation to the membrane (dashed arrow), where S1P is generated. The bioactive lipid can be released and then bind to S1P receptors (S1PRs). Activation of each receptor subtype leads to distinct G‐protein‐mediated signalling pathways. S1P can be also formed by SPHK isoform 2 inside the nucleus, and in this compartment, it can inhibit p21 transcription and histone deacetylase activity (HDAC).

S1P functions not only inside cells but also as a ligand for five specific‐protein coupled receptors (as reviewed in Spiegel and Milstien, 2003). S1P can be exported outside the cells by transporters belonging to the ATP‐binding cassette family and the putative transporter Spinster 2 and, therefore, acts as an autocrine or paracrine factor (Maceyka et al., 2012) (Figure 1B). S1P receptors are expressed in CNS cells (neurons, oligodendrocytes, astrocytes and microglia). Signalling through S1P receptors involves the activation of Gi, Go, Gq or G12/13 (Figures 1B and 2) and, therefore, signal transduction pathways involving PLC, MAPKs, PI3K/Akt, Rac and Rho/Rho kinase (Spiegel and Milstien, 2003).

The GPCRs specific for S1P (S1P_1–5 receptors) trigger different signalling pathways and are expressed and localized differently during tissue development or following stimulation. The S1P₁ receptor regulates the migration of neural stem progenitor cells both during development (Alfonso et al., 2015) and in response to injury (Kimura et al., 2007). The S1P₁ receptor is also involved in oligodendrocyte development, morphological maturation and early myelination (Jung et al., 2007; Dukala and Soliven, 2016), while activation of the S1P₅ receptor on the oligodendrocyte progenitor cells leads to process retraction and inhibits migration (Jaillard et al., 2005; Novgorodov et al., 2007).

During nerve growth factor (NGF)‐induced neuronal differentiation, there is a relocalization of S1P receptors: while the S1P₁ receptor, which induces neurite growth, is maintained in the plasma membrane, the S1P₂ receptor is internalized (Toman et al., 2004) to prevent loss of neurites. Growth factors, such as NGF, increase SPHK activity and S1P formation and vice versa S1P can induce growth factor release (Yamagata et al., 2003; Sobue et al., 2005; Murakami et al., 2007). Deletion of genes encoding S1P₁ receptors or both SPHK1 and SPHK2 in mice severely disrupts neurogenesis and angiogenesis leading to intrauterine death (Liu et al., 2000; Mizugishi et al., 2005) highlighting the role of S1P in the development of the nervous system.

1,25(OH)₂D₃ in nervous system physiology

The active form of vitamin D₃ has hydroxyl groups in positions 1 and 25. The enzymes 1‐α hydroxylase (CYP27B1), required to synthesize 1,25(OH)₂D_3, and the 24‐hydroxylase (CYP24A1), needed to degrade 25‐(OH)D₃ and 1,25(OH)₂D_3, are present in the brain (Zehnder et al., 2001; Naveilhan et al., 1993). The 1,25(OH)₂D₃ receptor (vitamin D receptor) is expressed in both neurons and glial cells (microglia, astrocytes, oligodendrocytes, Schwann cells) in different regions of the nervous system (DeLuca et al., 2013). Neural stem cells constitutively express vitamin D receptors, which can be up‐regulated by 1,25(OH)₂D₃ (Shirazi et al., 2015). Genomic 1,25(OH)₂D₃ –mediated effects require heterodimerization between the vitamin D receptor and the retinoid X receptor. This complex binds to response elements, thus regulating the transcription of genes (Christakos et al., 2016). It increases the transcription of the genes encoding growth factors, such as NGF, glial‐derived neutrophic factor (GDNF), neurotrophin 3, brain‐derived neurotrophic factor (BDNF) and ciliary neurotrophic factor, and for enzymes involved in the synthesis of neurotransmitters (tyrosine hydroxylase, tryptophan hydroxylase 2, glutamate decarboxylase), whereas it represses that of voltage‐dependent calcium channels (DeLuca et al., 2013; Patrick and Ames, 2014; Shirazi et al., 2015). The vitamin D receptor is also localized in the caveolae and induces rapid non‐genomic effects (Figure 3). Activation of PKA, Ca²⁺/calmodulin‐dependent PK, PI3K and MAPK p38 results in the phosphorylation of neurofilaments, and in the modulation of chloride, potassium and voltage‐dependent calcium channels in rat cortical neurons (Zanatta et al., 2012). In addition, other kinases including ERK1/ERK2, ERK5 and JNK1/JNK2 and PKC and other enzymes, such as PLA₂, Src and p21ras (Bi et al., 2016; Hii and Ferrante, 2016), are also targets of 1,25(OH)₂D₃.

Figure 3.

Biological function of 1,25(OH)₂D₃/vitamin D receptor (VDR) signalling. Non‐genomic rapid actions of 1,25(OH)₂D₃ are mediated by the membrane vitamin D₃ receptor (mVDR), localized at the plasma membrane. An inactive form of VDR is present in the cytosol (VDR). mVDR activation through MARRS (membrane‐associated, rapid response steroid‐binding protein) promotes the MAPK cascade, Raf kinase with the consequent activation of PKC, PI3K and PKA. 1,25(OH)₂D₃ can interact with TGF and EGF receptors to modulate cell cycle processes. Activation of the GPCR S1P₁ receptor (S1P1R) leads to a specific Raf‐MAPK–ERK cascade that may crosstalk with the classical VDR pathways. The genomic action of 1,25(OH)₂D₃ leads to the regulation of gene expression following the nuclear translocation of VDR, the formation of the complex of VDR and 9‐cis‐retinoic acid receptor (VDR/RXR), and its binding to the vitamin D₃ response elements (VDREs).

The combination of in vitro and in vivo experiments provides compelling evidence that 1,25(OH)₂D₃ has a crucial role in synaptic transmission and neuroplasticity (Smith et al., 2006; Grecksch et al., 2009; Groves et al., 2013; Eyles et al., 2013; Patrick and Ames, 2014; Latimer et al., 2014) as well as in proliferation, differentiation and neuroprotection, as summarized in Table 2. Increasing evidence derived from studies of 1,25(OH)₂D₃ deficiency and from vitamin D receptor polymorphisms suggests that 1,25(OH)₂D₃ influences susceptibility to a number of psychiatric and neurological diseases, which include AD, PD, schizophrenia, autism, depression, amyotrophic lateral sclerosis and epilepsy, and is especially strong for MS (Eyles et al., 2013; Peterson 2014; Spedding, 2014; Burton and Costello, 2015; Jiang et al., 2015; Shen and Ji, 2015a,b). The effect of 1,25(OH)₂D₃ deficiency has been studied in female rats or mice fed a 1,25(OH)2D₃‐deficient diet during pregnancy. The overall brain size of the offspring of these animals was increased and they had larger lateral ventricles. These effects were not modified by the addition of 1,25(OH)₂D₃ to the diet after birth. In adult life, these rats demonstrated subtle alterations in learning and memory (Eyles et al., 2013; Fernandes de Abreu et al., 2010; Hawes et al., 2015). Interestingly, prenatal 1,25(OH)₂D₃‐depleted rats exhibited an impairment in latent inhibition, mimicking some features found in schizophrenia (Grecksch et al., 2009). Furthermore, in humans the offspring of mothers who have had insuffient 1,25(OH)₂D₃ during pregnancy have been found to have language impairments (Whitehouse et al., 2012). The administration of 1,25(OH)₂D₃ has been found to exert a neuroprotective effect on the cognitive decline in ageing rats (Latimer et al., 2014). The hormone prevents the development of and reversibly blocks the progression of the pathological manifestations of experimental allergic encephalomyelitis, which is the animal model of MS. This protective effect is absent in vitamin D receptor KO mice (DeLuca et al., 2013; Eyles et al., 2013). The effect of 1,25(OH)₂D₃ depends on its neuroimmunomodulatory properties (Eyles et al., 2013) and also on its action on neural cells. In fact, 1,25(OH)₂D₃ increases both neural stem cell proliferation and differentiation into neurons and oligodendrocytes, the myelinating cells of the CNS (de la Fuente et al., 2015; Shirazi et al., 2015).

In neurodegenerative diseases, including PD and AD, adult neurogenesis in the hippocampal dentate gyrus and in the subventricular zone is impaired (Winner and Winkler, 2015). Therefore, factors that can promote neurogenesis are considered potential treatments for these disorders. The anti‐proliferative and pro‐differentiating effects of 1,25(OH)₂D₃ in neural cells were first described more than 10 years ago (Brown et al., 2003; Ko et al., 2004) and were mediated through the regulation of cyclin expression and NGF production in cultured hippocampal cells (Brown et al., 2003; Ko et al., 2004). Recently, the CERK signalling pathway has been shown to be involved in the cell growth arrest promoted by 1,25(OH)₂D₃ in human neuroblastoma cells (Bini et al., 2012). In fact, the pharmacological inhibition and the silencing of CERK drastically reduced cell proliferation. 1,25(OH)₂D₃, and the vitamin D receptor agonist ZK191784 induced a significant decrease in CERK expression and C1P content. The involvement of the vitamin D receptor/COUP‐TFI/histone deacetylase complex in CERK regulation has also been reported by Bini et al. (2012). There are accumulating data suggesting that 1,25(OH)₂D₃ has complex effects on the neurogenesis of neural stem cells. Cui et al. (2007) have studied the effect of foetal 1,25(OH)₂D₃ deprivation, and observed the formation of an increased number of neurospheres in cultures from the neonatal subventricular zone. Exogenous 1,25(OH)₂D₃ added to the culture medium reduced the number of neurospheres number in control samples [in agreement with the presumed anti‐proliferative effect of 1,25(OH)₂D₃] but not in cultures from the hormone‐deprived pups (Cui et al., 2007). In contrast, in vivo experiments have shown that fetal 1,25(OH)₂D₃ deficiency leads to reduced neurogenesis in the dentate gyrus of the hippocampus (Keilhoff et al., 2010). In another model of 1,25(OH)₂D₃ deficiency, the 1α‐hydroxylase KO mice, 1,25(OH)₂D₃ increases the proliferation, but decreases the survival of neurons in the dentate gyrus in neonates (Zhu et al., 2012). The different effects probably depend on the time window of exposure and/or the different sensitivity to the hormone of distinct neurogenic niches.

Crosstalk between SLs and 1,25(OH)₂D₃ actions

One or more components of the signal transduction pathway promoted by 1,25(OH)₂D₃ affect the metabolism of SLs and vice versa (Figure 4). For example, 1,25(OH)₂D₃ regulates the expression of S1P‐ phosphatase 2 (Reardon et al., 2013) and of CERK (Bini et al., 2012). Cholecalciferol, the non hydroxylated precursor of 1,25(OH)₂D₃, induces activation of SMase, an increase in ceramide and cell death in human glioblastoma cells (Magrassi et al., 1998). However, 1,25(OH)₂D₃ also increases the transcription of neurotrophic factors, such as NGF and BDNF, which require SPHK activity to execute their neuroprotective or prodifferentiating activity (Edsall et al., 1997; Culmsee et al., 2002; Saini et al., 2005; Murakami et al., 2007) and to affect excitability (Zhang et al., 2008). Similarly, many protective or differentiating actions of 1,25(OH)₂D₃ in non‐neural cells are due to stimulation of SMase and the generation of SPHK and S1P (Okazaki et al., 1989; Kleuser et al., 1998; Manggau et al., 2001; Sauer et al., 2003).

Figure 4.

Effect of 1,25(OH)₂D₃ and Aβ on the key reactions involved in the sphingolipid metabolic pathway. The effect of 1,25(OH)₂D₃ is shown in blue and dashed lines, whereas the effect of Aβ treatment or alterations associated with AD on the key enzymes involved in the reactions are shown in red and continuous line.

1,25(OH)₂D₃ is able to modulate expression of S1P receptors: the hormone reduces the chemorepulsive S1P₂ receptor levels on circulating osteoblast precursors (Kikuta et al., 2013) and decreases the expression of S1P₃ receptors in human breast cancer cells (Dolezalova et al., 2003). Interestingly, vitamin D receptor expression is correlated with the calcitriol‐mediated reduction of migration in glioblastoma multiforme (Salomón et al., 2014), but it is not known whether this effect involves the differential expression of S1P receptors.

SLs and 1,25(OH)₂D₃ have some common targets, including cathepsins. Ceramide and C1P interact directly and activate cathepsin D (Heinrich et al., 2000; Zebrakovská et al., 2011), which is involved in cell death in many cell types. For example, gemcitabine activates acid SMase (aSMase), leading to the lysosomal accumulation of ceramide, cathepsin D activation and glioma cell death (Dumitru et al., 2009). Cathepsin D is able to migrate to the nucleus. Indeed, nuclear translocation of mitochondrial cytochrome C, lysosomal cathepsins B and D and other death‐promoting proteins has been observed within the first 60 min of generalized seizures (Zhao et al., 2010). Both cathepsin D and its inhibitor cystatin A have VRE in their promoters (Wang et al., 2005). This may explain, at least in part, the pro‐survival and pro‐death effects of 1,25(OH)₂D₃ on different cells.

Histone acetylation and methylation are often present at sites of vitamin D receptors, and 1,25(OH)₂D₃‐induced binding of the vitamin D receptor at these sites is associated with an increase in the level of histone modifications as well as in changes in chromatin packaging (Carlberg and Campbell, 2013). Similarly, S1P formed inside the nuclei by SPHK2 activation can inhibit HDACs and regulate gene transcription (Figure 2). Therefore, it could be possible that both 1,25(OH)₂D₃ and S1P epigenetically modulate the same genes (Hait et al., 2009; Huang et al., 2015a).

Furthermore, SLs are important components of lipid‐rich microdomains (LRM), also named lipid rafts, fluctuating nanoscale assemblies that can be stabilized to coalesce, forming platforms that function in membrane signalling and trafficking (Gulbins and Grassmé, 2002; Lingwood and Simons, 2010). LRM have been described in the plasma membrane, mitochondria and nuclei. In the inner nuclear membrane, LRM play a role in active chromatin anchoring, transcription factor binding and DNA duplication (Cascianelli et al., 2008; Albi and Villani, 2009; 2013; Cataldi et al., 2014). The vitamin D receptor seems to be partly localized in nuclear LRM (Marini et al., 2010; Bartoccini et al., 2011). Changes in SM levels and/or a shift in SM composition (from C24:0‐SM to C16:0‐SM) have been associated with a reduction in the expression of vitamin D receptors in nuclear LRM of tumour cells (Lazzarini et al., 2015) and embryonic hippocampal cell differentiation (Bartoccini et al., 2011). Whether these alterations in nuclear LRM are involved in neurodegeneration is not known. In the nervous system, LRM play a role in many processes, including neurotrophic factor signalling, cell adhesion and migration, axon guidance and myelin formation and stabilization (Aureli et al., 2015). Notably, recent evidence also suggests that LRM alterations are involved in neurodegenerative disorders including PD, AD, amyotrophic lateral sclerosis, Huntington's disease and prion diseases (Schengrund, 2010; Aureli et al., 2015; Marin et al., 2016).

Recent data have revealed that the crosstalk between S1P and 1,25(OH)₂D₃ also occurs in the extracellular fluids. It has been reported that patients with acute or chronic inflammation have low levels of plasma gelsolin (pGSN) (Osborn et al., 2008; Lee et al., 2009). Gelsolin has two isoforms with a similar structure and function: the cytoplasmic actin‐binding protein form, important for the regulation of cell shape and motility [cytoplasmatic gelsolin (cGSN)], and pGSN, a multifunctional protein that acts as an extracellular actin scavenger system crucial for the removal of actin released from injured cells (Chauhan et al., 2008; Carro, 2010). Although the functions of pGSN and the mechanisms of its protective action have not been clarified, it is clear that low levels of pGNS are an indicator of poor prognosis or critical care complications. Notably, pGNS is able to bind to S1P in humans. This pGSN‐S1P interaction in extracellular fluids may have several important consequences by impairing either the ability of gelsolin to bind actin or that of S1P to bind to S1P receptors. In fact, the pGSN‐S1P complex affects the S1P‐S1P₁ receptor module that regulates lymphocyte distribution and the immunomodulatory balance at inflammatory sites (Bucki et al., 2010). It has been observed that patients suffering from lymphatic meningitis show low concentrations of pGSN and a high concentration of S1P in their CSF samples (Bucki et al., 2010). Notably, another recent study has demonstrated that 1,25(OH)₂D₃ treatment can affect either S1P and pGSN. In fact, the hormone alleviates inflammation in experimental allergic encephalomyelitis, a model of MS, and this therapeutic effect might be derived from the ability of the hormone to reduce S1P (which is elevated in the CSF and spinal cord of rats with experimental allergic encephalomyelitis). However, this effect might be limited by its simultaneous action in reducing pGSN and cGSN (Zhu et al., 2014).

Taken together, the accumulating evidence suggests that 1,25(OH)₂D₃ and SLs can converge and share some targets: (i) the activation of similar pathways (through activation of protein kinases); (ii) the modulation of enzyme expression/activity (i.e. cathepsin); (iii) the control of genes encoding for key enzymes of SLs metabolism and, probably, of S1P receptors by 1,25(OH)₂D₃; and (iv) the modulation by S1P‐dependent histone acetylation of vitamin D receptor‐dependent transcription. In addition, changes in the composition of SLs in LRM can also affect the localization and therefore the function of vitamin D receptors.

SLs/1,25(OH)₂D₃ crosstalk: potential role in neurogenerative diseases

AD

The actual most common form of dementia is AD, a neurodegenerative disorder of the CNS characterized by extracellular amyloid‐containing plaques, intracellular neurofibrillary tangles consisting of hyperphosphorylated tau protein and by the death of cholinergic neurons of the basal forebrain. Amyloid plaques are mainly formed by aggregated amyloid β peptide (Aβ) generated by the hydrolysis of the amyloid precursor protein (APP), first, by β‐secretase 1 and, then, by γ‐secretase. The fibrils of the senile plaques are mainly composed of the self‐assembled Aβ1–42 peptide that forms a heterogeneous mixture of oligomers and protofibrils. The small soluble Aβ1–42 oligomers are considered to be the major neurotoxic species in AD, and it has been hypothesized that cerebral accumulation of Aβ1–42 precedes and drives the deposition of the tau protein in neuronal perikarya and their processes (Selkoe and Hardy, 2016).

Alterations in the expression or in the activity of enzymes involved in the metabolism of SLs have been found in the brain of AD patients (Table 2, Figure 4). They include SMases (Katsel et al., 2007; He et al., 2010), ceramidase (CDase) (Huang et al., 2004), S1P lyase and SPHK (Ceccom et al., 2014), serine palmitoyl transferase, UDP‐glucose ceramide glucosyltransferase and CERS1,2,6 (Katsel et al., 2007; Couttas et al., 2016). Changes in SL content (ceramide, S1P, SM, gangliosides and sulfatides) have also been reported in animal models of AD (see ref. in Grimm et al., 2013) and in brain tissue and CSF of AD patients (Han et al., 2002; Cutler et al., 2004; Satoi et al., 2005; He et al., 2010; Mielke and Lyketsos, 2010; Couttas et al., 2014; Couttas et al., 2016; Fonteh et al., 2015).

The results of in vitro experiments indicate that Aβ1–42 directly binds and activates nSMase, decreasing SM content (Grimm et al., 2005). Aβ1–42 also activates aSMase through increased ROS accumulation via NADPH oxidase activation and reduced glutathion depletion (Jazvinšćak Jembrek et al., 2015). Ceramide generated by the degradation of SM due to the activation of SMase induces neuronal apoptosis (Jana and Pahan, 2004; Satoi et al., 2005; Malaplate‐Armand et al., 2006) or impairs autophagy (Yang et al., 2014). Also ceramide increases the stability, while S1P increases the activity of β‐secretase 1 (Puglielli et al., 2003; Takasugi et al., 2011). Moreover, SM decreases Aβ1–42 production by inhibiting the γ‐secretase (Grimm et al., 2005). Therefore, some SLs might be protective by lowering Aβ levels (either by decreasing its production or by increasing its clearance), while others might increase Aβ1–42 oligomerization and toxicity. Simultaneously, APP processing also leads to changes in lipid metabolism, resulting in complex regulatory feedback cycles, which appear to be dysregulated in AD (Grimm et al., 2013).

Recently, it has been demonstrated that exosome release in neural cells requires SMase activity (Wang et al., 2012). The role of exosomes in AD is controversial. One study has shown that in vitro neuronal exosomes are able to capture Aβ, and their infusion into brains of AD mice decreases Aβ and amyloid depositions (Yuyama et al., 2015). More recently, it has been suggested that ceramide‐enriched exosomes promote the aggregation of Aβ (Dinkins et al., 2016), since an AD mouse model lacking nSMase2 exhibits a decreased exosome release associated with a reduced plaque burden and improved cognition (Dinkins et al., 2016).

Increasing evidence derived from epidemiological studies indicates that 1,25(OH)₂D₃ deficiency and vitamin D receptor polymorphisms influence susceptibility to AD (Gezen‐Ak et al., 2012; Annweiler et al., 2014), whereas Aβ1–42 may disrupt the hormone‐vitamin D receptor pathway and cause defective utilization of 1,25(OH)₂D₃ by suppressing the level of vitamin D receptors, and by elevating the level of 24‐hydroxylase, and, thereby, increasing the catabolism of the hormone (Dursun et al., 2011; 2013a). In addition to its neuroprotective effects involving calcium homeostasis, a decrease in ROS and inflammation, 1,25(OH)₂D₃ is able to exert other specific effects important for AD. For example, 1,25(OH)₂D₃ may regulate the expression of many genes associated with AD and attenuate the build up of Aβ deposits either by enhancing their clearance (transport to the blood or to the CSF) or by stimulating the phagocytosis of Aβ. It is likely that the hormone alters APP processing and prevents the defects in ACh by increasing the activity of choline acetyltransferase (thus ACh synthesis) in the brain (Briones and Darwish, 2012; Annweiler et al., 2014; Durk et al., 2014; Landel et al., 2016).

Epigenetic modifications are involved in the regulation of many genes, and aberrant epigenetic changes are associated with AD. For example, hyperacetylation of histone H4 at lysine 12 in peripheral monocytes appears to be an early event in AD‐pathology (Plagg et al., 2015). Recently, HDAC inhibitors have emerged as promising compounds for restoring the cognitive deficits in a mouse model of AD (Kilgore et al., 2010). Therefore, upon treatment with 1,25(OH)₂D₃ or/and FTY7120, a possible effect on acetylation and DNA methylation of AD‐related genes (i.e. β‐secretase 1) could result in beneficial effects against Aβ‐induced toxicity.

Niemann–Pick disease type C (NPC)

Niemann–Pick disease type C (NPC) is an autosomal recessive storage disorder due to mutations of two proteins NPC1 and NPC2, which mediate intracellular cholesterol trafficking in mammals. NPC is characterized by abnormal sequestration of unesterified cholesterol within the late endolysosomes and the accumulation of sphingosine and gangliosides, the formation of meganeurites and neurofibrillary tangles, neuroinflammation and axonal dystrophy. As the disease progresses, neuronal death of Purkinje cells of the cerebellum becomes prominent. AD and NPC share some molecular pathways, including abnormal cholesterol metabolism, and the involvement of Aβ and tau (Malnar et al., 2014; Vanier, 2015). Miglustat is an inhibitor of the enzyme glucosylceramide synthase that converts ceramide into glucosyl ceramide (Figure 1), the first step in the synthesis of gangliosides. Miglustat has neuroprotective effects on NPC models (Table 2). It has been approved for use in several cases of gangliosidosis and, recently, for NPC (Patterson et al., 2015). Therefore, SLs appear to play a role in the pathogenesis of NPC. Indeed, it has been demonstrated that SM inhibits, while ceramide increases NPC2‐mediated cholesterol transport (Abdul‐Hammed et al., 2010). Moreover, Lloyd‐Evans et al. (2008) have proposed that sphingosine, by altering calcium homeostasis, could play a role in the onset of NPC.

It is possible that 1,25(OH)₂D₃ and S1P could have some protective effect on NPC. It is already known that stem cells induce survival of cerebellar NPC1−/− cells (Lee et al., 2010) by increasing S1P and, as discussed above, that 1,25(OH)₂D₃ is able to reduce Aβ load in AD models. In addition, autophagy is dysregulated in NPC, and 1,25(OH)₂D₃ exerts some neuroprotective effects through the modulation of autophagy (Li et al., 2015a).

PD

PD is a neurodegenerative disease characterized by loss of dopamine cells in the basal ganglia, and the accumulation and/or aggregation of α‐synuclein. Mutations in genes causing lysosomal storage disorders, such as those encoding GCDase A, aSMase and NPC1, may increase the risk for developing PD (for a recent review, see Migdalska‐Richards and Schapira, 2016). Moreover, reduced GCDase activity (GBA1) has been found in patients with sporadic PD (Murphy et al., 2014; Table 2). The concentration of glucosyl ceramide and that of α‐synuclein are inversely correlated. Total ceramide and SM levels are reduced in the anterior cingulate cortex of PD patients compared with controls (Abbott et al., 2014). A shift towards ceramide containing short acyl chains and an up‐regulation of the expression of the Cer1S gene (which could be a compensatory effect to the reduction in ceramide) have also been reported (Abbott et al., 2014). S1P and 1,25(OH)₂D₃ have a neuroprotective effect on cellular models of PD (Shinpo et al., 2000; Smith et al., 2006; Pyszko and Strosznajder, 2014; see Table 3). Also 1,25(OH)₂D₃ has shown neuroprotection in different animal models of PD that have been correlated with increases in GDNF, increases in tyroxine hydroxylase expressing cells and a decrease in inflammation (Wang et al., 2001; Kim et al., 2006; Smith et al., 2006). Furthermore, 1,25(OH)₂D₃ supplementation was associated with a significantly reduced risk of PD (Shen and Ji, 2015a). The mechanism by which a deficiency in GCDase increases the risk of developing PD is still unclear, but it is known that glucosyl ceramide can stabilize α‐synuclein oligomers (Mazzulli et al., 2011) and that activation of GCDase reduces the accumulation of α‐synuclein and restores lysosomal function in vitro (Mazzulli et al., 2016). Indeed, small increases in glucosyl ceramide and glucosyl sphingosine have been reported in primary cultured cortical neurons with GCDase knockdown (Mazzulli et al., 2011), in dopaminergic neurons harbouring heterozygote GCDase/GBA1 mutations (Schöndorf et al., 2014) and in the hippocampus of PD patients without a GBA1 mutation (Rocha et al., 2015) (Table 2). Another possibility is that the changes in SL metabolism derived from GCDase deficiency impair autophagy, which has been suggested to contribute to α‐synuclein accumulation in cellular and animal models of GCase deficiency (Mazzulli et al., 2011; Schöndorf et al., 2014).

Table 3.

Effects of 1,25(OH)₂D₃ on nervous system differentiation, protection and proliferation

Differentiation
Cell	Effect	Mechanism	References
Primary embryonic hippocampal	–	–	–
cells	↑ neurite outgrowth	↑NGF	Brown et al., 2003
OPC	↑ differentiation	↑MBP	de la Fuente et al., 2015
Neuronal stem cell	↑ differentiation to	–	–
–	oligodendrocytes	↑ CNTF	Shirazi et al., 2015
Schwann cells	↑ differentiation	↑ IGF‐1	Hao et al., 2015
HN9.10e	↑ neurite outgrowth	↑ NGF, Bcl‐2	Marini et al., 2010

Protection
Animal/Cell	Stimulus	Mechanism	References
Murine experimental	–	–	–
allergic encephalomyelitis	MBP	nd	Lemire and Archer, 1991
Dopaminergic cell	MPTP+, sulfoximine	↓ROS, ↑glutathion	Shinpo et al., 2000
Mesencephalic cell	6OH‐DA,	↑TH, ↑arborization	Wang et al., 2001
Hippocampal neuron	NMDA, glutamate	↓LVCC	Brewer et al., 2001
Substantia nigra	Zn	↓lipid peroxidation,	Lin et al., 2003
–	–	↑ DA	–
Cortex	ischaemia	↑HO‐1, ↓ GFAP	Oermann et al., 2004
Cortical neuron	glutamate	↑MAP2, ↑GAP‐43,	Taniura et al., 2006
–	–	↑synapsin 1, ↑VDR	–
Rat, substantia nigra	6OH‐DA	↑DA	Smith et al., 2006
rat, mice	MPTP	↓ microglia activation,	Kim et al., 2006
–	–	↓TNFα mRNA, ↓INFγ mRNA	–
cortical cells	cyanide	↓ uncoupling, ↑Ikkb	Li et al., 2008
hippocampus	glutamate, ischaemia	↓ caspase‐3	Kajta et al., 2009
mesencephalic neuron	–	↑GDNF	Orme et al., 2013
rat hippocampus	ischaemia/reperfusion	GluN3A, ERK, pCREB	Fu et al., 2013
SH‐SY5Y	rotenone	↑ autophagy	Jang et al., 2014
–	–	↑LC3, beclin, AMPK	–
Neuron–glia	endotoxin	↓ MAPK, ↓iNOS,	Huang et al., 2015b
–	–	↓IL‐6, ↓MIP‐2 mRNA	–
mouse	MPTP	↑ autophagy	Li et al., 2015b
Cortex slices	hyperhomocysteinaemia	↓ ROS, ↓ iNOS	Longoni et al., 2016
Schwann	high glucose	↑ CBS, ↑H2S, ↓ ROS	Zhang et al., 2016
–	methylglyoxal	–	–
Tg2576 and TgCRND8 mice	–	↓plaque formation,	Durk et al., 2014
–	–	↓ lower soluble Aβ levels,	–
–	–	↑ P‐glycoprotein	–
AD mouse	(AbPP)	↓decrease memory deficit	Yu et al., 2011
–	–	↓plaque formation, ↑NGF	–
–	–	↓ inflammation	–
Ageing rats	–	↓decrease memory deficit	Latimer et al., 2014
–	–	Modulation pro‐inflammatory cytokines	Briones and Darwish, 2012
–	–	↓decrease amyloid	–
–	–	↓decrease amyloid	Yu et al., 2011
Hippocampal neurons and cortical neurons	Aβ	↓cytotoxicity ↓iNOS	Dursun et al., 2011, 2013a,b
–	–	↓ LVCC A1C, ↑VDR
Mouse retina	–	↑ phagocytosis, ↓ Aβ	Lee et al., 2012
AD macrophages	–	Modulation IL‐1, IL‐1R	Mizwicki et al., 2012, 2013
–	–	↑ phagocytosis, ↓ Aβ	–
bEnd.3 cells	–	↑Aβ1–40 brain‐to‐blood efflux	–
–	–	of amyloid‐β (Aβ) peptide	Guo et al., 2016
–	–	LRP1 and RAGE regulation	–

Proliferation
Cell/animal	Action	Mechanism	References
Neuroblastoma cells	↓	CERK	Bini et al., 2012
–	–	nd	Gumireddy et al., 2003a,b
–	–	–	Celli et al., 1999a,b
–	–	–	Stio et al., 2001
Stem cells	↑	↑NT‐3, BDNF, GDNF and CNTF	Shirazi et al., 2015
Primary embryonic	↓	–	Brown et al., 2003
hippocampal cells	–	–	–
1,25(OH)2D3 ‐deprived	–	–	–
embryos E19, brain	↑	↑cyclin D	Ko et al., 2004
–	–	↓ cyclin B, ↓p21	–
Glioblastoma cells	↑, no effect	–	Diesel et al., 2005
1,25(OH)2D3‐deprived	↓	–	Cui et al., 2007
neuroprogenitors, SVZ	–	–	–
1α‐hydroxylase knockout	↑	–	Zhu et al., 2012
Mouse, dentate gyrus	–	–	–

The table illustrates the effect of 1,25(OH)2D3 in cell types and the mechanism involved. The noxius agent is listed under Stimulus. ↑, increase; ↓, decrease; 6‐OHDA, 6‐hydroxydopamine; AMPK, AMP‐dependent PK; CNTF, ciliary derived neurotrophic factor; bEnd.3, mouse brain microvascular endothelial cell line; DRG, dorsal root ganglion; HO, hemoxygenase; iNOS, inducible nitric oxide synthase; LRP1, low‐density lipoprotein receptor‐related protein 1; MBP, myelin basic protein; MPTP, 1‐methyl‐4‐phenyl‐1,2,3,6‐tetrahydropyridine; nd, not determined; OPC, oligodendrocyte precursor cell; RAGE, receptor for advanced glycation end products, SVZ, subventricular zone.

SLs, in particular S1P, and 1,25(OH)₂D₃ display their neuroprotective actions through common effectors such as calcium regulation, synaptic modulation and growth factor expression, but whether 1,25(OH)₂D₃ and SLs could act synergistically on neuroprotection and/or neurogenesis in neurodegenerative diseases, such as AD, PD and NPC, is still unknown and deserves further investigation. Preliminary results in our laboratory indicate that the crosstalk between SLs and 1,25(OH)₂D₃ leads to a specific balance between neurodegeneration/neuroprotection in neuronal cells. In particular, in human SH‐SY5Y differentiated cells, we found that 1,25(OH)₂D₃ treatment counteracts the down‐regulation of S1P1‐mediated signalling promoted by Aβ1–42 (Pierucci et al., 2017).

Potential implications for 1,25(OH)₂D₃ and FTY720 supplementation in AD

Several observations in clinical trials have demonstrated that 1,25(OH)₂D₃ supplementation may have protective effects in AD; however, in other studies, no beneficial outcome has been reported (DeLuca et al., 2013; Landel et al., 2016), and evidence for a correlation between hypovitaminosis D and reduced neuroprotection against AD or AD progression is missing. Similarly, the ability of 1,25(OH)₂D₃ supplementation to prevent other neurodegenerative diseases, such as MS, needs further investigation. Moreover, some data suggest that the combination of 1,25(OH)₂D₃ supplementation with the anti‐neurodegenerative drug nemantidine could counteract the cognitive decline better than that of the single compound (Annweiler et al., 2014).

On the contrary, the neuroprotective effect of S1P analogues on neurodegenerative diseases is well established. Fingolimod, the commercial name of FTY720, is an analogue of sphingosine, which acts as an immunosuppressant and has recently been approved for the treatment of MS. The phosphorylation of FTY720 by SPHK generates FTY720‐phosphate, a molecule structurally similar to S1P that can bind to all the S1P receptors, except the S1P₂ receptor. In lymph nodes, it acts as a highly potent functional antagonist of the S1P₁ receptor, leading to S1P₁ receptor internalization in T cells that become unable to egress from the nodes. Also FTY720 is active on different cells of the nervous system, including neurons, astrocytes, oligodendrocytes and microglia (Brunkhorst et al., 2014), and its protective function affects the process of myelination, the activation of microglia, proliferation and migration of precursor cells, neuronal differentiation and survival (Kawabori et al., 2013). In vivo, it has been shown that experimental allergic encephalomyelitis was attenuated by FTY720 supplementation, and no effect was observed on astrocytes that did not express S1P₁ receptors. However, neurons lacking S1P₁ receptors were positively affected by the compound (Choi et al., 2011). In vitro, FTY720 decreases Aβ production in cultured neuronal cells (Takasugi et al., 2013).

Regarding its therapeutic potential in AD, it has been reported that when FTY720 supplementation was given to rats injected with Aβ1–42, there was a reduction in cell death in the hippocampus and cortex as well as an increase in memory compared with control rats (Asle‐Rousta et al., 2013; Hemmati et al., 2013). The in vivo beneficial effect on the nervous system is due to many factors including an increase in BDNF production, which leads to an increase in striatum size (Deogracias et al., 2012) and facilitates neuronal repair in diseases associated with decreased levels of BDNF, such as Huntington's disease (Di Pardo et al., 2014; Miguez et al., 2015). FTY720 also reduces α‐synuclein aggregation, in part by increasing BDNF levels (Vidal‐Martínez et al., 2016).

Recently, it has been shown that FTY720 also has inhibitory effects on epigenetic modifications by reducing HDAC and regulating gene expression programmes associated with memory and learning (Hait et al., 2014). All together, these observations lead us to speculate that 1,25(OH)₂D₃ supplementation and FTY720 could act synergistically in the prevention of neurodegenerative diseases. Preliminary studies in vivo performed in our laboratory suggest the possibility of an interaction between the hormone and FTY720. In fact, damage was reduced when 1,25(OH)₂D₃ supplementation in mice injected with a submaximal dose of Aβ1–42 was combined with FTY720 treatment. Further investigations are in progress (Meacci et al., personal communication).

In conclusion, the potential for expanding the use of 1,25(OH)₂D₃ to treat neurodegenerative diseases is worth investigating. Additionally, the therapeutic potential of the structural analogues of 1,25(OH)₂D₃ (see ref. in Leyssens et al., 2014) remains unexplored. In the long term, 1,25(OH)₂D₃ and its analogues might provide valuable tools either for basic research into the elucidation of the mechanisms of neuroprotection and for subsequent designer drug development. The combined treatments with 1,25(OH)₂D₃ and agonists/antagonists of S1P_1–5 receptors and the improvement in the characterization and quantification of ceramide species may offer significant advances in terms of understanding, and the ability to predict, the protein aggregation‐induced toxicity in vivo.

Author contributions

M.G.‐G. and E.M. conceived the outline of the review and wrote the manuscript. F.P. and A.V. contributed to some parts of the manuscript.

Conflict of interest

The authors declare no conflicts of interest.

Garcia‐Gil, M. , Pierucci, F. , Vestri, A. , and Meacci, E. (2017) Crosstalk between sphingolipids and vitamin D3: potential role in the nervous system. British Journal of Pharmacology, 174: 605–627. doi: 10.1111/bph.13726.