Secretory sphingomyelinase in health and disease

Genetic impact on S-ASM activity

A number of genetic sequence variations and mutations influence S-ASM activity and the process of generating L- or S-ASM. The repeat number variation of a hexanucleotide sequence – c.108GCTGGC(3_8)/p.37LA(3)_8) – that leads to the modification of the signal peptide of the pre-pro-form of ASM (Wan and Schuchman, 1995), was associated with S-ASM activity but not L-ASM activity in a cell culture model. In addition, analysis of a human sample resulted in a significant association of S-ASM activity with the number of hexanucleotide repeats; levels were highest in subjects homozygous for six repeats, intermediate in subjects homozygous for five repeats, and lowest in subjects homozygous for four repeats. One of the most frequent SMPD1 sequence variations, rs1050239/c.1522G>A/p.G508R (Rhein et al., 2014), has been associated with S-ASM activity but not L-ASM activity. S-ASM activity was measured in the blood plasma of healthy young adults and was highest in subjects homozygous for the major G allele, intermediate in heterozygous subjects and lowest in subjects homozygous for the A allele (Reichel et al., 2014). Importantly, the influences of these two polymorphic sites on S-ASM activity are independent (Rhein et al., 2014). The nonsynonymous variation rs141641266/c.1460C>T/p.A487V, previously assumed to be a missense mutation causing Niemann-Pick disease (NPD) type B (Simonaro et al., 2002), leads to normal L-ASM and only slightly lower S-ASM activity levels (Rhein et al., 2013).

More than 100 clinically relevant missense mutations in the SMPD1 gene leading to enzymes with decreased catalytic activity, the cause of autosomal recessive NPD types A and B (Brady et al., 1966; Schuchman, 2010), have been deposited in the Human Gene Mutation Database (www.hgmd.cf.ac.uk). While genotype/phenotype correlations are typically complex, the DR608 deletion mutation appears to protect patients against the development of the more severe neuropathic NPD form type A (Schuchman and Miranda, 1997). L-ASM activity in cultured skin fibroblasts of patients correlates with the respective phenotype, with residual L-ASM activity of <5% corresponding to the more severe NPD type A (Desnick et al., 2010), for which – similar to other neurodegenerative sphingolipidoses – effective therapeutic options are missing (Eckhardt, 2010). The determination of S-ASM activity in blood plasma can be used diagnostically, particularly to discriminate between NPD patients and NPD carriers (He et al., 2003). However, clinical data on S-ASM activity in NPD patients remain limited because in most studies only cellular L-ASM activity is determined (Vanier et al., 1980; Pavlu-Pereira et al., 2005). Because lysosomal hypertrophy is a hallmark of lysosomal storage disorders, a parallel assessment of the acidic compartment volume combined with L- and S-ASM activities could provide a basis for genotype-phenotype correlations, particularly to explain the large heterogeneity of symptom severity observed in NPD patients with an intermediate phenotype (Pavlu-Pereira et al., 2005; Wasserstein et al., 2006). Interestingly, there is growing support for a genetic convergence of lysosomal storage disorders and Parkinson’s disease (Deng et al., 2014). Mutations such as p.L304P (Gan-Or et al., 2013), which may be limited to Ashkenazi Jews (Wu et al., 2014), or variants in SMPD1, e.g., p.R591C, p.P533L (Foo et al., 2013), have not only been associated with altered ASM activities but also appear to be linked to an up to 9-fold increased risk for Parkinson’s disease (Spatola and Wider, 2014). This effect is likely mediated via ceramide because plasma ceramide and monohexosylceramide levels are increased and associated with cognitive impairment in Parkinson’s disease patients lacking a risk-conveying glucocerebrosidase mutation (Mielke et al., 2013).

Transcriptional and translational effects on ASM activity

ASM is up-regulated at the transcriptional level in different cell culture models, but only effects on L-ASM activity have been investigated (Langmann et al., 1999; Murate et al., 2002). A variety of alternatively spliced ASM transcripts have been identified, of which only one, ASM-1, encodes an enzyme with complete protein domains and L-ASM activity. The isoforms ASM-2 to -7 lack protein domains that are essential for ASM activity. Some ASM isoforms have a dominant-negative effect on cellular L-ASM activity levels (Rhein et al., 2012). However, whether and how S-ASM activity is influenced by alternative splicing has not been investigated. The unexpectedly mild NPD-B phenotype and 20–25% residual L-ASM activity observed in two patients homozygous for either p.M1_W32del or p.W32X and, consequently, complete absence of full-length ASM, suggests that in the absence of a functional first initiation codon, the second initiation codon (ATG coding for p.33M) may still produce a relatively functional enzyme (Pittis et al., 2004).

Trafficking and post-translational processing relevant for S-ASM and L-ASM activity

The generation of L-ASM versus S-ASM is determined by the differential processing and trafficking of the common precursor ASM protein (Figure 1). Upon entry of the nascent ASM polypeptide (631 amino acids corresponding to 70 kDa) into the rough endoplasmic reticulum (ER) (Hasilik, 1992), cleavage of the putative N-terminal signal peptide (Wan and Schuchman, 1995), which is postulated to extend to p.A48 (Schuchman et al., 1991) yields the pre-pro-form with a molecular weight of 75 kDa after glycosylation (65 kDa protein core). The signal peptide is essential for the production of enzymatically active mature L-ASM. ASM constructs with N-terminal truncations of the signal sequence lack catalytic L-ASM activity (Ferlinz et al., 1994). Despite the predicted absence of the signal peptide in the mature enzyme, the length of the hydrophobic core of the polymorphic ASM signal peptide affects the secretion and activity of S-ASM, whereas L-ASM activity is unaffected (Rhein et al., 2014).

Figure 1:

L- and S-ASM arise from a common ASM precursor protein.

The ASM-encoding gene SMPD1 gives rise to a common mannosylated precursor protein, pre-pro-ASM, which is cleaved to yield pro-ASM. Differential glycosylation and N-terminal as well as C-terminal processing inside the Golgi then lead to the generation of two distinct ASM forms. Whereas S-ASM is released into the extracellular space via the constitutive secretory pathway and requires Zn²⁺ ions for activation, L-ASM is shuttled into the lysosomal trafficking pathway via its mannose 6-phosphate groups and encounters Zn²⁺ ions on its way to the endolysosomal compartment. The details of the exocytotic and endocytotic steps have not been fully elucidated (dotted lines). Data regarding the molecular weight of the mature enzyme with the deglycosylated protein core in brackets are based on the most recently reported data (Edelmann et al., 2011; Jenkins et al., 2011b) and protein molecular weight calculations (protcalc.sourceforge.net) after modifications cited in the text; they may vary from other reports.

L- and S-ASM undergo differential glycosylation processing inside the Golgi giving rise to the ASM pro-form of 72–75 kDa (63–64 kDa protein core). L-ASM has a high mannose N-glycan composition, which provides stabilization and protection against proteolysis within the lysosome, whereas S-ASM possesses a complex N-glycosylation pattern (Hurwitz et al., 1994b; Schissel et al., 1998b). This differential glycosylation is evident in the susceptibility of L-ASM to endoglycosidase H, which is specific for high mannose-type N-linked oligosaccharides (Yamamoto, 1994). In contrast, S-ASM is completely resistant to endoglycosidase H, and both enzymes are susceptible to peptide-N-glycosidase F (Schissel et al., 1998b). In the cis-Golgi, L-ASM acquires mannose-phosphate residues via the typical sequential action of N-acetylglucosamine-1-phosphotransferase (Reitman and Kornfeld, 1981) and N-acetylglucosamine phosphodiesterase on the mannose residues of the precursor (Hurwitz et al., 1994b; Schissel et al., 1998b), whereas S-ASM is spared of this modification. Bovine DNase I is an example of a suboptimal substrate for the phosphotransferase that is required to confer the mannose 6-phosphate moieties for the lysosomal pathway, and this feature is assumed to give rise to an intralysosomal as well as a secretory form of bovine DNaseI. A similar phenomenon may occur for ASM (Nishikawa et al., 1997; Schissel et al., 1998b). The ASM pro-form contains six potential N-glycosylation sites (N88, N177, N337, N397, N505, N522), five of which appear to be occupied in purified ASM (Ferlinz et al., 1997). The two C-terminal N-glycosylation sites (N505 and N522) are essential for proper trafficking and affect S-ASM secretion and the enzymatic activity of L-ASM (Ferlinz et al., 1997). Although presumably not glycosylated, removal of the fifth site reduces enzymatic activity to <20% due to protein misfolding (Newrzella and Stoffel, 1996; Ferlinz et al., 1997).

The S-ASM pro-form is constitutively secreted via the default Golgi secretory pathway resulting in a 75–80 kDa extracellular enzyme with a protein core of 64 kDa (Jenkins et al., 2010, 2011b). The L-ASM pro-form is targeted mostly via cation-dependent and cation-independent mannose 6-phosphate receptors and to a lesser extent via sortilin-mediated transport to the endolysosomal compartment (Ni and Morales, 2006). The importance of the mannose 6-phosphate recognition system is evident in the 8-fold increase in the ratio of secreted to intracellular ASM activity in cells deficient in mannose phosphorylation. Alanine scanning mutagenesis of the 13 lysine residues revealed that p.K95 is critical for the specific three-dimensional arrangement required for mannose phosphorylation, and p.K95A replacement results in an overall 2-fold increase in the ratio of secreted to intracellular ASM activity via a reduction of intracellular activity and an increase in Zn²⁺-dependent secretory activity (Takahashi et al., 2005). Similarly, deglycosylation treatment or expression of a dominant-negative sortilin variant results in trapping of the enzyme within the Golgi in its degradation-vulnerable form (Bartelsen et al., 1998; He et al., 1999; Ni and Morales, 2006).

Sequencing of purified enzymes has revealed that the N-terminus of S-ASM begins at the amino acid p.H62 (numbered according to the standard database reference sequence NM_000543.4 throughout the review, p.H60 in the publication), whereas that of L-ASM begins at p.G68 (Schissel et al., 1998b) after typical N-terminal processing of lysosomal proteins (Hasilik, 1992). The positions of the N-termini correspond to molecular weights of the remaining deglycosylated full C-terminal portions of 63.6 kDa for S-ASM and 63.0 kDa for L-ASM. The removal of only a few N-terminal amino acids – six in ASM – is reminiscent of the maturation of another lysosomal enzyme, the β-chain of hexosaminidase (Quon et al., 1989). By contrast, p.G85 (61.0 kDa) has been identified as the first amino acid of purified human placental ASM (Lansmann et al., 1996). While the exact molecular weight values differ slightly between publications, the presence of markedly different subunits across human tissues (Jobb and Callahan, 1987) has not been confirmed. ASM may also be delivered from the Golgi directly to the phagosome, bypassing fusion with lysosomal compartments. Sortilin participates in this process via its short cytoplasmic tail (Wähe et al., 2010), which is similar to the cytoplasmic segment of the cation-independent mannose 6-phosphate receptor (Petersen et al., 1997; Nielsen et al., 2001).

In the endolysosomal system, the 72-kDa enzyme is processed at the C-terminus to the 65-kDa L-ASM (55 kDa for the deglycosylated protein). This processing step within the acidic compartment is similar to that for other lysosomal hydrolases (Erickson and Blobel, 1983; Arunachalam et al., 2000) and is essential to generate the catalytically highly active enzyme, most likely by promoting the coordination of lysosomal Zn²⁺. By contrast, the low activity of the non-lysosomal precursor may prevent enzymatic activity in the Golgi (Jenkins et al., 2011b). The role of the C-terminal region in the proteolytic maturation of L-ASM is indicated by the altered localization and virtual absence of Zn²⁺-independent activity in the recombinant NPD mutants p.R602H, p.R602P, and p.ΔR610 (Jenkins et al., 2011b). The preliminary detection of p.Q622 as the C-terminus of L-ASM would suggest a loss of only nine amino acids, resulting in a protein core of 62 kDa (Jenkins et al., 2011b). However, another recent study indicates that TNF-mediated stimulation leads to the direct interaction of caspase-7 with the 72-kDa pro-ASM and proteolytic cleavage of the zymogen at the non-canonical cleavage site after p.D253 within TNF receptosomes, generating an active 57-kDa L-ASM (Edelmann et al., 2011). This active L-ASM consists of a protein core of 43 kDa and corresponds to the previously proposed active form after N-terminal cleavage between the second and third glycosylation sites with minimal trimming at the C-terminus (Ferlinz et al., 1994). Unlike its lysosomal counterpart, the C-terminus of S-ASM remains intact (Jenkins et al., 2011b), enabling regulation via its unbound p.C631 (Qiu et al., 2003).

Functional inhibitors of ASM (FIASMAs), e.g., antidepressant drugs such as desipramine and fluoxetine (Kornhuber et al., 2008, 2010, 2011), lead indirectly to the inactivation of L-ASM (Hurwitz et al., 1994a; Kölzer et al., 2004b) by leupeptin-sensitive proteolytic cleavage to a 52-kDa form (Jenkins et al., 2011b). The half-life of L-ASM, as determined by the addition of the protein synthesis inhibitor cycloheximide and pulse chase experiments, of approximately 5–6 h (Hurwitz et al., 1994b; Schissel et al., 1998b; Jenkins et al., 2010) is exceptionally short compared to most lysosomal enzymes, which have half-lives of days to weeks. By contrast, S-ASM is not further processed after the Golgi and is detected as a 75–80-kDa form following its secretion into the extracellular space. In contrast to L-ASM, after secretion, S-ASM is unusually stable and remains at the same activity level over at least 25 h, with profound implications for chronic conditions in which small changes in continued secretion over time compound to high total S-ASM levels (Jenkins et al., 2010).

ASM activity is influenced by further post-translational modifications. An activating effect on L-ASM has been described for the phosphorylation of serine residue p.S510 in the C-terminal domain by the protein kinase Cδ. It has been postulated that this phosphorylation results in the translocation of L-ASM to the plasma membrane (Zeidan and Hannun, 2007a; Zeidan et al., 2008b). The ASM mutant p.S510A, in which this phosphorylation site was altered by site-specific mutagenesis, retains L-ASM activity but is not secreted (Jenkins et al., 2010). This phosphorylation site seems to be essential for constitutive as well as regulated secretion and thus the activity of S-ASM. Furthermore, stimulated exocytosis of ASM is dependent on the t-SNARE protein syntaxin 4 (Perrotta et al., 2010).

L- and S-ASM are activated in vitro via interactions with Zn²⁺ ions (Schissel et al., 1998b). Zn²⁺ acquisition itself may represent a form of L-ASM regulation because the partial Zn²⁺-dependence of L-ASM isolated from lysosomal-rich fractions but not standard cell homogenates suggests an incomplete saturation of this form with Zn²⁺ (Schissel et al., 1998b), although the isolate might have contained Zn²⁺-dependent pre-lysosomal forms in addition to the fully mature lysosomal 65-kDa ASM (Jenkins et al., 2011b). Massive overexpression of ASM, which typically results in the saturation of the mannose 6-phosphate receptor shuttling mechanism and secretion of an otherwise lysosomally targeted enzyme (Ioannou et al., 1992), yielded a Zn²⁺-dependent mannose 6-phosphate receptor-binding species in the supernatant. This finding demonstrates that Zn²⁺ acquisition is not glycosylation-dependent but occurs during lysosomal trafficking (Schissel et al., 1998b).

Eight disulfide bonds are formed among 17 cysteine residues within mature ASM, which leaves the C-terminal p.C631 unbound (Lansmann et al., 2003). Oxidation, deletion or replacement of this residue activates recombinant human S-ASM by favoring the hydration of Zn²⁺. Activation is also achieved by copper-promoted dimerization via cysteine residues and by C-terminal truncation by carboxypeptidase Y (Qiu et al., 2003) and likely resembles the cysteine switch mechanism of matrix metalloproteinases (Van Wart and Birkedal-Hansen, 1990) in post-translational regulation. However, this activation is assumed to be limited to S-ASM in which the entire C-terminus is retained; this cysteine is lost in fully mature L-ASM during the final processing step (Jenkins et al., 2011b).

Regulation of S-ASM and L-ASM secretion

The ASM enzyme consists of distinct domains, a proline-rich domain, a metallophosphoesterase domain and the C-terminal domain. Moreover, strong sequence homology (Ponting, 1994) and an analogous arrangement of disulfide bonds (Lansmann et al., 2003) suggest that ASM contains its own intramolecular sphingolipid activator protein (SAP)-type domain as the fourth functional domain and is thus, unlike other lysosomal hydrolases, independent of the presence of SAPs as a cofactor. SAPs are small, enzymatically inactive lysosomal glycoproteins that facilitate sphingolipid degradation at the inner membrane of acidic compartments by promoting substrate binding and lipid extraction (Kolter and Sandhoff, 2005; Remmel et al., 2007). The addition of SAPs can partially rescue inactive ASM with mutations in conserved amino acids within its SAP domain (Kölzer et al., 2004a).

The selective activation of L-ASM by a variety of stress stimuli has been comprehensively investigated (Charruyer et al., 2005; Rotolo et al., 2005; Zeidan et al., 2008a; Milhas et al., 2010). However, stimulating agents can have differential effects on L- and S-ASM activity. The co-regulation of S- and L-ASM activity has been addressed in several cell culture systems. Cultured human platelets exhibit increased S-ASM release but decreased L-ASM activity upon stimulation with thrombin (Romiti et al., 2000). S-ASM activity is increased and L-ASM activity is decreased in MCF7 breast carcinoma cells treated with phorbol myristate acetate or ammonium chloride (Jenkins et al., 2010). S-ASM activity is increased upon stimulation with interleukin (IL) 1β or tumor necrosis factor-alpha (TNFα), with no effect on L-ASM activity (Jenkins et al., 2010). The same pattern was observed following the treatment of human umbilical vein endothelial cells with IL-1β or interferon γ; S-ASM activity was increased by 2–3.5-fold, whereas L-ASM activity was slightly decreased (Marathe et al., 1998). Oxidized low-density lipoprotein (LDL) and oxidized LDL-containing immune complexes differentially regulate L-ASM activity and S-ASM release in macrophages (Truman et al., 2012). A special role of S-ASM in chemokine elaboration was detected following the stimulation of MCF7 cells with TNFα because the chemokine CCL5 appears to be an initial target of the pathway that is selectively induced by S-ASM activity (Jenkins et al., 2011a). Cytokines also affect the ratio of L- and S-ASM activity in a genetic manner (Rhein et al., 2014). Thus, cytokine stimulation of cultured cells may have a short-term effect on S-ASM and appear to increase S-ASM activity and secretion while decreasing L-ASM activity. This effect could evolve by redirecting the ASM precursor protein away from the lysosomal pathway into the secretory pathway. The involvement in this redirection of the cis-Golgi enzyme N-acteylglucosaminyl-1-phosphotransferase, which phosphorylates the mannose residues of the precursor protein for the lysosomal trafficking pathway, has been proposed (Tabas, 1999).

Although originally thought to localize only to the lysosome, under stress conditions, L-ASM may translocate from intracellular compartments to the extracellular leaflet of the cell membrane, where it hydrolyzes SM to ceramide to form signaling platforms (Cifone et al., 1994); however, L-ASM may also contribute to extracellular ASM activity. The designation of the ASM that is translocated to the outer leaflet of the plasma membrane as a response to specific stimuli is not consistent and may, in fact, represent S- or L-ASM or a combination of both species. While some reports refer to the enzyme as ‘secretory’ ASM, other authors suggest it is L-ASM by assuming a lysosomal origin. Although the vesicles that appear to fuse and empty their contents at the extracellular surface may contain the secretory form of ASM, direct data on specific properties or Zn²⁺-dependence are lacking in most studies. For example, extracellular ASM activity increased following the repair of myoblasts from mechanical injury as a result of vesicle docking and exocytosis (Defour et al., 2014). Cationic cell-penetrating peptides as potential molecular transporters of membrane-impermeable molecules induce the translocation of ASM to the plasma membrane, resulting in an increase in ceramide formation and, in turn, enhancing its own uptake (Verdurmen et al., 2010). Similarly, treatment of Jurkat T cells with ultraviolet light type C also induces the rapid translocation of ASM to the membrane, hydrolysis of extracellular SM and raft clustering (Rotolo et al., 2005). In addition to the ASM-independent functions of CD95, transmission electron microscopy and fluorescence-activated cell sorting analysis have demonstrated that CD95 stimulation also causes ASM to translocate from intracellular, presumably vesicular, compartments to the extracellular leaflet. This process hydrolyzes extracellularly oriented SM, which, in turn, mediates CD95 clustering as an essential step toward CD95-triggered cell death (Grassmé et al., 2001). Together with CD95, TNF-related apoptosis-inducing ligand is involved in the activation of ASM via a redox mechanism that results in the generation of ceramide and the formation of ceramide-enriched membrane platforms (Dumitru and Gulbins, 2006), potentially via the release of ASM of lysosomal origin, as indicated by cell lysate activity in the absence of Zn²⁺. The application of hydrogen peroxide as the primary form of reactive oxygen species in mammals also induces very rapid calcium-dependent ASM release by lysosomal exocytosis, as detected by the exposure of lysosome-associated protein LAMP1 (Li et al., 2012). The lysosomal origin of released ASM is also supported by the concomitant appearance of LAMP1 on the sarcolemma surface and the release of the lysosomal β-hexosaminidase in a wound repair study (Corrotte et al., 2013).

In fact, ASM has recently emerged as a major regulator facilitating wound repair in eukaryotic cells, such as from pore-forming toxins (Tam et al., 2010; Corrotte et al., 2013). Plasmalemmal injury-triggered Ca²⁺ influx induces the fusion of lysosomes with the plasma membrane and exocytosis of ASM. Through ceramide generation, ASM in turn triggers lesion removal via endocytosis and intracellular degradation (Tam et al., 2013; Andrews et al., 2014). Transcriptional silencing of ASM abolishes this process, while the addition of exogenous enzyme restores it (Tam et al., 2010). Trypanosoma cruzi subverts the ASM-dependent ceramide-enriched endosomes in the plasma membrane repair pathway to invade host cells (Fernandes et al., 2011). While ceramide production by ASM in the outer leaflet promotes tighter packing and a negative curvature that leads to inward vesiculation (Trajkovic et al., 2008), ceramide generation caused by the activation of the neutral sphingomyelinase species localized to the inner leaflet of the plasma membrane promotes an outward curvature and thus allows the cell to shed membrane-containing toxin pores into the extracellular space (Draeger and Babiychuk, 2013).

Enzyme kinetics

Experiments using a secretion-incompetent p.S510A mutant enzyme suggest distinct metabolic roles of S-ASM and L-ASM in cellular ceramide formation with respect to specific substrates: while Il-1β induces a time- and dose-dependent up-regulation of S-ASM secretion with selective production of C16-ceramide at the expense of C16-SM, elevations in L-ASM activity are associated with increases in selective, very-long-chain ceramide species, e.g., C26:1-ceramide (Jenkins et al., 2010). Michaelis-Menten kinetic analysis revealed roughly similar values of the Michaelis constant K_m for SM affinity despite the use of differently labeled substrates: 77 μm for soluble ASM in the human epidermis (Bowser and Gray, 1978), 5–25 μm for ASM purified from placenta (Jones et al., 1981, 1983; Sakuragawa, 1982), 65 μm for ASM from Bacillus cereus (Fujii et al., 2004), 47 μm for ASM from human fibroblasts (Sato et al., 1988), 25 μm for recombinant ASM from the conditioned medium of infected insect cells (Bartelsen et al., 1998), 20 μm for ASM from human CSF (Mühle et al., 2013), 0.2–0.6 μm for ASM released from human fibroblasts and mouse L-cells, respectively (Weitz et al., 1983), 11 μm with a V_MAX of 21 nmol/h/mg protein for HEK 293 whole cell lysates and 2 μm and a V_MAX of 4.3 μmol/h/mg protein for ASM immunoprecipitated from Jurkat cells (Gulbins and Kolesnick, 2000). Specific activities ranged from 600 to 2500 μmol/h/mg (Sakuragawa, 1982; Yamanaka and Suzuki, 1982; Jones et al., 1983; Weitz et al., 1985; Lansmann et al., 1996).

Defects in N-glycosylation reduce enzyme stability and activity but do not affect K_M values (Newrzella and Stoffel, 1996). Moreover, pH plays an important role in substrate binding but not catalytic velocity (Callahan et al., 1983). The stimulation of cell surface receptors can lead to an increase in V_MAX′ of 2–3-fold with little or no change in K_M values (Schwandner et al., 1998). In a therapeutically aimed approach, the dipolar aprotic solvent dimethyl sulfoxide caused a slow but marked up to 2-fold increase in sphingomyelinase activity (V_max) at pH 5.0 and 7.5 over a period of days. This activity was dependent on protein synthesis and did not substantially change the K_m value (Sakuragawa et al., 1985; Sato et al., 1988).

Reaction conditions

The influence of detergents on ASM activity was examined by Spence et al. (1989) using fetal bovine serum, with an 8–15-fold increase in response to the addition of Triton X-100 to the reaction. It was also noticeable by the discrepancy in an order of magnitude between optimal Nonidet P-40 concentrations in assays for human serum- vs. CSF-derived S-ASM (Mühle et al., 2013). Detergent also influences the immunoprecipitation of soluble ASM from human urine and placenta (Driessen et al., 1985), but details regarding the different detergent optima remain limited. Storage temperature appears to be an important parameter for comparing absolute S-ASM activities because the activity of endogenous ASM from plasma, CSF and cell lysates (Mühle et al., 2013) as well as that of human recombinant ASM in cell supernatants (Qiu et al., 2003) is higher following long-term storage at -20°C compared to -80°C. This activation is attributed to the loss of the single free cysteine residue p.C631 by chemical modification during the freezing process (Qiu et al., 2003). However, the enzymatic activity remains relatively stable during the storage of CSF (Mühle et al., 2013) as well as undialyzed urine (Seidel et al., 1978) at room temperature for several days.

The dependence on the addition of Zn²⁺ ions is supported by the stimulation of S-ASM activity by Zn²⁺, which is increased in case of prior chelation of endogenous Zn²⁺ by EDTA, and its inhibition by EDTA or the more zinc-specific chelator 1,10-phenanthroline (He et al., 1999). By contrast, typical short-term chelation of L-ASM with EDTA is insufficient to strip the enzyme of its metal ion, consistent with other zinc metalloenzymes (Little and Otnäss, 1975). In contrast to neutral sphingomyelinases, Mg²⁺ and Mn²⁺ were ineffective for ASM activity (Spence et al., 1989). The role of other metal cations has not been well investigated, with the exception of a potentially different ASM species that was partially purified from seminal plasma (Vanha-Perttula, 1988). The optimal Zn²⁺ concentrations are similar to those for matrix metalloproteinases (Sorbi et al., 1993) and lie well within the range present in the extracellular space, e.g., approximately 100 μm in human whole blood or serum (Buxaderas and Farré-Rovira, 1985), although the actual concentration depends on factors such as dietary intake or pregnancy (Donangelo and Chang, 1981). Zn²⁺ ions move freely through the cerebrospinal and extracellular fluid compartments and are differentially distributed in various regions of the brain (Takeda et al., 1994), exhibiting variations in half-life (Takeda et al., 1995) and age-dependence (Sawashita et al., 1997) that may influence local S-ASM activity. Moreover, Zn²⁺ levels are elevated in both atherosclerotic and inflammatory lesions (Mendis, 1989; Milanino et al., 1993), and Zn²⁺ is released during intense neuronal activation (Assaf and Chung, 1984; Frederickson and Moncrieff, 1994). In addition, the regulation and cell-type-dependent variation of the subcellular concentration and localization of Zn²⁺ ions (Csermely et al., 1987; Brand and Kleineke, 1996) may render S-ASM partially or fully Zn²⁺-independent under specific conditions. For example, the cysteine-rich protein metallothionein acts as a temporary cellular reservoir of free zinc. Metallothionein may keep cellular concentrations very low or release zinc to intracellular zinc-dependent enzymes in a process that is dynamically controlled by its interactions with the glutathione redox couple (Jiang et al., 1998). Consequently, exogenous Zn²⁺ ions are not essential for endothelial cell secreted S-ASM activity and cause only 2-fold stimulation (Marathe et al., 1998).

Activity at neutral pH

Given the optimal reaction rate of the enzyme in an acidic range of pH 5.0–5.5 (Spence et al., 1989; Mühle et al., 2013), activity in the neutral pH of blood or CSF of 7.4 appears counterintuitive. Indeed, the ability of S-ASM to hydrolyze SM in a neutral milieu remains controversial and has led some authors to refer to S-ASM as ‘secretory sphingomyelinase’, omitting the specification ‘acidic’ despite its proven origin from the ASM-encoding gene SMPD1. However, this name would only be unambiguous in the absence of any other reported sphingomyelinase species, e.g., neutral or alkaline enzymes that could potentially be detectable as secreted or otherwise released proteins in the extracellular space, which is currently the case (Milhas et al., 2010).

Data concerning the extent of S-ASM activity at neutral pH are inconsistent. Both S-ASM activities in saliva and tear fluid peak at pH 4–5 and are markedly decreased at pH 6, with negligible residual activities at pH 7 (Takahashi et al., 2000). This behavior corresponds well to the pH profiles described for tissue L-ASM (Schneider and Kennedy, 1967; Bowser and Gray, 1978), ASM released from human fibroblasts (optimal pH 4.4) and mouse L-cells (pH 4.8) (Weitz et al., 1983) and our data for S-ASM in human CSF (Mühle et al., 2013), serum, plasma, saliva and urine, which indicate <5% of maximal Zn²⁺-dependent activity at pH 6.5 and nearly background levels at pH 7.0 (C. Mühle, unpublished data). Purified recombinant human ASM from CHO cells binds tightly to SM at pH 4.0, while binding is not detectable at pH 8.0 (He et al., 1999). By contrast, S-ASM from fetal bovine serum (Spence et al., 1989) or recombinant ASM species (Mintzer et al., 2005) exhibit clearly detectable hydrolysis at a neutral pH of 7.4 of up to 20% of that at the optimal acidic pH. This discrepancy may be attributable to the different characteristics of human endogenous vs. purified (Spence et al., 1989) bacterial or recombinant, sometimes tagged, S-ASM (Mintzer et al., 2005) species utilized in these experiments, different reaction conditions, substrates and purity or glycosylation patterns (Jenkins et al., 2009) of the enzymes. Interestingly, neutral sphingomyelinase C from Staphylococcus aureus selectively catalyzes the hydrolysis of SM in the plasma membrane (Slotte et al., 1990). Its application to hippocampal slices results in the generation of long-chain ceramides and in a positive reversible regulation of neuron excitability via a sphingosine-1-phosphate-mediated mechanism (Norman et al., 2010). Because extracellular ASM catalyzes comparable reactions at the plasma membrane, similar activity can be expected.

Although the biological role of S-ASM is not precisely clear, the enzyme has been implicated in the metabolism of lipoprotein-bound SM to ceramide, the subsequent aggregation and retention of LDL particles and the acceleration of lesion progression as key initial steps in atherogenesis (Marathe et al., 2000a; Tabas et al., 2007; Devlin et al., 2008). After observing that SM-degrading activity of several arterial wall cell types is responsible for subendothelial retention and aggregation of atherogenic lipoproteins in rabbit and human atherosclerotic lesions (Schissel et al., 1996b), Schissel et al. (1998a) identified this activity as a Zn²⁺-dependent S-ASM capable of hydrolyzing SM associated with atherogenic lipoproteins at a neutral pH. Signs of SM hydrolysis in lesional LDL (Öörni et al., 1998) as well as ASM-induced LDL aggregation (Walters and Wrenn, 2011), depending on the sphingomyelinase-to-LDL molar ratio (Guarino et al., 2006), confirmed the role of ASM and extended its promotion of aggregation to small very-low-density and intermediate-density lipoprotein particles (Öörni et al., 2005). However, another group detected sphingomyelinase activity as an intrinsic property of the multifunctional LDL constituent apolipoprotein B-100 based on the sequence homology of the apoliprotein B-100 α2 domain with bacterial sphingomyelinase or at least as very tightly associated with LDL and absent in oxidized LDL or other lipoproteins (Holopainen et al., 2000; Kinnunen and Holopainen, 2002). A number of atherogenesis-associated modifications of lipoprotein-bound SM appear to increase its susceptibility to hydrolysis: while native LDL is only hydrolyzed and aggregated at acidic pH, oxidation, treatment with phospholipase A₂, or enrichment with apolipoprotein CIII promotes prompt hydrolysis at pH 7.4 (Schissel et al., 1998a). LDL particles undergo lipolysis by ASM more readily after prior proteolysis by the extracellular hydrolases chymase and cathepsin S, and this pre-treatment was even essential for the action of the secretory phospholipases A₂ (sPLA₂-V). This sensitization of LDL by proteolysis may enhance extracellular accumulation of LDL-derived lipids during atherogenesis (Plihtari et al., 2010). Moreover, a high SM-to-phosphatidylcholine ratio appears to promote hydrolysis and aggregation by ASM at neutral pH, as observed for plasmatic lipoproteins from apolipoprotein E KO mice, which are prone to atherosclerosis (Schissel et al., 1998a). Similarly, the high SM content in the slowly cleared remnant lipoproteins of these mice enhances their susceptibility to ASM (Jeong et al., 1998). The most prominent difference was a 10-fold increase in S-ASM activity at pH 7.4 in LDL extracted from atherosclerotic lesions compared to that in plasma LDL from the same human donor (Schissel et al., 1998a).

Moreover, extracellular SM degradation by S-ASM could occur in pockets of acidity that are present in inflammatory lesions of the arterial wall (Menkin, 1934) and in artherosclerotic plaques (Naghavi et al., 2002; Sneck et al., 2012), in proximity to glycosaminoglycans (Maroudas et al., 1988), activated macrophages (Tapper and Sundler, 1992) and osteoclasts (Silver et al., 1988) or close to the cell surface, e.g., in acidic microdomains located predominantly in the distal dendrites of oligodendrocytes (Ro and Carson, 2004). Such acidic microenvironments have been linked to lipid microdomains (Steinert et al., 2008) as the site of optimal ASM action. The translocation of lysosomal V1 H⁺-ATPase to the cell membrane is a critical contribution to the formation of a local acid microenvironment to facilitate membrane ASM activation and, consequently, redox signalosome formation in coronary arterial endothelial cells (Xu et al., 2012). In addition, substrate binding but not catalytic velocity are predominantly mediated by pH (Callahan et al., 1983). S-ASM may also be active in acidic endosomes following endocytosis of the enzyme by cell surface receptors (Kornfeld, 1987).

Human plasma itself exerts an inhibitory effect on the reaction if sample volumes >2% of the total assay volume are used. This inhibition is only partially explained by the presence of low amounts of EDTA originating from anticoagulation vials. Inhibition has also been observed to a lesser degree for lithium-heparin-plasma as well as serum (C. Mühle, unpublished data). The effect is not observed for CSF but is expected to limit the SM-hydrolyzing activity of S-ASM in the blood in vivo.

Endogenous inhibition of ASM activity

The activity of S-ASM is inhibited by a number of endogenous agents, including adenosine monophosphate and high concentrations of ZnCl₂ (6 mm) (Schissel et al., 1996a; Jenkins et al., 2010), as well as in a time- and dose-dependent manner by dithiothreitol and beta-mercaptoethanol (He et al., 1999). Interestingly, S-ASM appears to be less stable than L-ASM at elevated temperatures (Schissel et al., 1996a). Inorganic phosphate, other nucleotides, dolichol phosphate and phosphoinositides are potent noncompetitive inhibitors of ASM (Callahan et al., 1983; Watanabe et al., 1983). Interestingly, while phosphatidylinositol 3,5-bisphosphate (Kölzer et al., 2003) and phosphatidylinositol 3,4,5-trisphosphate (Testai et al., 2004) both inhibit ASM in the low mm range, the enzyme is potently activated by other phosphate-containing lipids such as endocytosed bismonoacylgycerophosphate (Linke et al., 2001) or by plasma membrane-derived phosphatidylglycerol and phosphatidic acid (Oninla et al., 2014). Cholesterol, another endogenous lipid, potently inhibits ASM (Reagan et al., 2000). Removal of lipoproteins from the culture medium of fibroblasts obtained from NPD-C patients who suffer from an additional secondary ASM deficiency restored the activity of ASM (Thomas et al., 1989). The presence of sterols is likely responsible for the inhibitory effect on ASM activity, which may be relevant to inhibitor development because various members of the sterol class, such as 7-ketocholesterol, exhibit inhibitory potential in vitro (Maor et al., 1995).

Role of ASM in microorganism infection

Viral, bacterial and parasitic pathogens are assumed to induce changes in the composition and structure of membranes and microdomains therein to permit the invasion of epithelial cells. The triggered translocation of ASM from endolysosomal compartments to the host cell surface followed by SM hydrolysis to form ceramide-rich domains seems to play a key role in microbial internalization (Riethmüller et al., 2006). ASM has been associated with infection and host defense against Pseudomonas aeruginosa (Grassmé et al., 2003), Listeria monocytogenes (Utermöhlen et al., 2003) and Cryptosporidium parvum (Nelson et al., 2006). ASM mediates the entry of Neisseria gonorrhoeae into nonphagocytic epithelial (Grassmé et al., 1997) and phagocytic (Hauck et al., 2000) cells via the enrichment of CEACAM receptors for the bacterium in ceramide domains, thus functioning as a portal of microbial entry. Infection of macrophages with Salmonella typhimurium induces a dynamic change in the cellular distribution of ASM, with a reduction in L-ASM and a surge in S-ASM activity (McCollister et al., 2007). L-ASM is rapidly activated during the entry of sindbis virus in neuroblastoma cells (Jan et al., 2000). Viruses such as human immunodeficiency virus (Popik et al., 2002) and rhinoviruses (Grassmé et al., 2005) appear to employ sphingomyelin-enriched microdomains to invade epithelial cells. SM functions as a novel receptor for the entry of Helicobacter pylori vacuolating cytotoxin into epithelial cells, and treatment with sphingomyelinase to deplete plasma membrane SM significantly reduces the sensitivity of cells to the pathogen (Gupta et al., 2008). Surface-localized ASM activity and the presence of SM are required for the efficient infection of cells by Ebola virus (Miller et al., 2012). Moreover, there is evidence that pathogens utilize their own sphingolipid repertoire as virulence factors, such as the sphingomyelinase of Streptococcus aureus (Cohen and Barenholz, 1978), or continue to modulate the sphingolipid metabolism of the host after internalization to replicate, survive, and finally infect neighboring cells (reviewed in Zeidan and Hannun, 2007b). An understanding of these interaction processes is a prerequisite for future antimicrobial therapies. In addition, ASM is required for the proper fusion of late phagosomes with lysosomes to efficiently transfer lysosomal antibacterial hydrolases into phagosomes (Schramm et al., 2008).

Involvement of ASM in other cellular processes

In erythrocytes, the activity of sphingomyelinase induces a transition from a discoid to spherical shape, followed by phosphatidylserine exposure and finally a loss of cytoplasmic content (Dinkla et al., 2012). In the context of elevated S-ASM levels in sepsis (Claus et al., 2005; Kott et al., 2014), the resulting enhanced erythrocyte clearance is likely to contribute to anemia. The sensitivity of erythrocytes increases further during blood bank storage and physiological aging (Dinkla et al., 2012).

In contrast to the vast number of experimental conditions that result in increased ASM activity, only one study has reported a condition leading to a reduction of ASM activity. Mechanical injury-triggered ASM secretion is reduced by 70% in C2C12 myoblasts upon knockdown of dysferlin expression, thus mimicking the cells of patients with the progressive muscle wasting disease dysferlinopathy. The defect in cell membrane repair in the immortalized patient myoblasts was fully reversed by treating the cells with extracellular ASM (Defour et al., 2014).

While the ASM/ceramide pathway has been well established as a bona fide system in the chemotherapeutic treatment of cancers and many commonly used chemotherapeutic agents mediate cell death via ASM-induced ceramide formation [reviewed in Henry et al., (2013)], recent evidence (Don et al., 2014) suggests that reducing ASM activity with functional inhibitors such as desipramine may induce lysosomal rupture in cancer cells (Petersen et al., 2013). Although unexpected with respect to reduced SM levels in cancerous compared to normal tissue (Hendrich and Michalak, 2003; Barceló-Coblijn et al., 2011), ASM activity is down-regulated in cancer (Petersen et al., 2013) and may consequently sensitize these cells to further decreases in enzyme activity levels. Lysosomal stability and integrity requires L-ASM activity (Kirkegaard et al., 2010), while the cytoprotective properties of ASM down-regulation in cancer cells are assumed to relate to the secreted form required for vesicle and membrane turnover (Magenau et al., 2011). In the context of reduced SM levels in cancer, the potent antitumor agent 2-hydroxyoleic acid selectively kills cancer cells by stimulating SM synthases, leading to strong augmentation and restoration of the SM concentration (Barceló-Coblijn et al., 2011).

S-ASM in brain function and behavior

The development of ASM KO mice has permitted the investigation of the role of ASM in behavior. Homozygous ASM KO mice develop normally and do not exhibit impairments until they are 8 weeks of age. Mild tremor and locomotor ataxia then emerge. Tottering with zigzag movements becomes evident and is characteristic of cerebellar dysfunction. Between 12 and 16 weeks, ASM KO mice become lethargic and unresponsive to stimuli. They increasingly experience problems feeding and much reduced weight compared to WT mice. Beyond 4 months of age, these animals exhibit severe ataxia and die between 6 and 8 months of age. ASM KO animals exhibit a significant reduction in brain volume. A striking finding was the significant loss of Purkinje cells in the cerebellum and atrophy of the mesencephalon beginning at 10 weeks of age (Horinouchi et al., 1995; Otterbach and Stoffel, 1995). Both regions appear to be involved in movement control and motivated behavior. Homozygous ASM KO mice exhibited reduced ceramide levels in the hippocampus and displayed some signs of reduced anxiety and a reduction of depression-related behaviors. Young ASM KO mice did not exhibit altered synaptic structure or functioning at the level of the hippocampus (Gulbins et al., 2013). Interestingly, heterozygous ASM KO mice, which possess approximately 50% of normal ASM activity, did not exhibit any motor problems, reduced feeding, or enhanced mortality. This finding may suggest that 50% enzyme activity is sufficient for normal brain function and a normal behavioral phenotype (Horinouchi et al., 1995), consistent with the rare phenotype observed in NPD carriers (Lee et al., 2003; Schuchman, 2007). While these findings suggest an important role of ASM in brain function and locomotor behavior, it was unclear if these effects were mediated by L-ASM or S-ASM because the KO affects both subtypes. In a subsequent study, this question was addressed by fusing the Smpd1 gene with the lysosomal Lamp1 gene and thus targeting all ASM to the lysosome. The mutant mice still exhibited low levels of L-ASM activity (11.5–18.2% of WT) but a complete absence of S-ASM activity. These mice exhibited neither brain pathology nor signs of locomotor impairment. This finding suggests that severe deficits in locomotor function in ASM KO mice are mediated predominantly by a lack of L-ASM, while S-ASM may not be required (Marathe et al., 2000b).

The role of ASM in depression/anxiety has been investigated in animal studies utilizing a transgene for ASM (tgASM). These mice exhibited higher ASM activity and ceramide production in the hippocampus than WT mice (Gulbins et al., 2013). The enhanced ceramide levels in the hippocampus were paralleled by a decline in neurogenesis, neuronal maturation, and neuronal survival (Gulbins et al., 2013), which are depression-related neuronal markers (Santarelli et al., 2003; Krishnan and Nestler, 2008). S-ASM activity was increased more than 10-fold in the CSF of tgASM mice compared to the WT controls (Mühle et al., 2013). At the behavioral level, tgASM mice exhibited a depression/anxiety-like phenotype in several tests (Gulbins et al., 2013); for a review see Kornhuber et al. (2014). This ASM effect is most likely mediated in the brain because the chronic application of C16-ceramide in the dorsal hippocampus of mice induced the same depression-like behavioral phenotype observed in tgASM mice (Gulbins et al., 2013). This finding is consistent with the effects of chronic unpredictable stress, which not only double hippocampal ceramide levels and reduced neurogenesis and neuronal maturation but also induce depression-like behavior in mice (Gulbins et al., 2013). Further support for the role of ceramide is provided by the elevated plasma ceramide levels detected in patients with major depression (Gracia-Garcia et al., 2011). Whether these effects are mediated by S-ASM or L-ASM remains to be determined.

Many antidepressant drugs are functional inhibitors of ASM (Albouz et al., 1983; Kornhuber et al., 2008, 2010, 2011). These drugs can normalize the effects of chronic unpredictable stress in WT and tgASM but not ASM KO mice. This observation suggests that ASM functional inhibition is a major pathway for the pharmacological effects of antidepressants (Gulbins et al., 2013; Kornhuber et al., 2014). This effect, however, predominantly affects L-ASM. Interestingly, evidence suggests that serum zinc levels are significantly lower in patients with major depression (Levenson, 2006; Swardfager et al., 2013a). Low dietary zinc intake has been associated with a greater incidence of depression (Vashum et al., 2014), and an antidepressant effect of zinc supplementation has been observed in laboratory rodents (reviewed in Levenson, 2006, and Swardfager et al., 2013b) and in human trials (Lai et al., 2012; Swardfager et al., 2013b). While the underlying mechanism remains a subject of speculation, a possible connection to the stimulatory effects of zinc on S-ASM activity has not been explored.

S-ASM and atherosclerosis

Inflammation plays an important role in the development of atherosclerosis. In mice, inflammation induced by lipopolysaccharide (LPS) increases S-ASM activity in the serum 3 h after injection. This increase was dependent on IL-1 activity. Serum S-ASM was also increased in mice that were treated with IL-1β or TNFα. These findings raise the possibility that an increase in S-ASM activity contributes to inflammatory cytokine activity in atherosclerosis (Wong et al., 2000). The ASM activity present in atherosclerotic lesions (Schissel et al., 1996b) is dependent on S-ASM and is associated with the extracellular arterial wall. In this bound state, S-ASM retains enzymatic activity that stimulates subendothelial retention and aggregation of atherogenic lipoproteins (Marathe et al., 1999). When mice are fed an atherogenic diet containing saturated fats and cholesterol, serum S-ASM activity is up-regulated, which facilitates the conversion of SM to ceramide (Deevska et al., 2012). However, another study found increased activity of S-ASM in rats fed for 14 weeks with a diet rich in n-3 polyunsaturated fatty acids (PUFA) vs. butter, and a strong negative correlation was observed between the serum concentration of n-3-PUFA and S-ASM activity (Drachmann et al., 2007). The association of atherosclerosis with increased age is mediated, at least in part, by S-ASM. S-ASM is increased in mice at 65 weeks of age. Apolipoprotein E-/- (ApoE^-/-) mice develop atherosclerosis at 15 weeks of age. Plasma ceramide levels are generally higher in ApoE^-/- than WT mice but decrease with age in parallel with an increase in S-ASM activity. When ApoE^-/- mice develop atherosclerosis, several ceramide species increase in the aorta compared to WT mice: C18:0, C22:0 and C24:0. At 65 weeks of age, C16:0 and C24:1 species are increased compared to WT mice. These findings indicate that an age-related increase in S-ASM activity, which results in an elevation of certain ceramide species, may contribute to age-related atherosclerosis (Kobayashi et al., 2013). When ApoE^-/- mice are treated with a recombinant adeno-associated virus (AAV) that constitutively expresses high levels of human ASM in the liver and plasma, S-ASM levels are persistently elevated. However, this phenomenon leads to only a short reduction in plasma SM and no changes in serum lipoprotein levels. Plaque formation in the aortic sinus after 17 weeks did not differ in mice treated with a control AAV. These findings suggest that S-ASM has no accelerating or exacerbating role in atherosclerotic lesion formation in the ApoE^-/- model of atherosclerosis (Leger et al., 2011).

S-ASM in diabetes

Diabetes mellitus, a metabolic disorder characterized by hyperglycemia, is associated with changes in ceramide metabolism. The induction of a diabetic state with streptozotocin causes a significant increase in blood glucose in rats. Eight weeks after induction, plasma and renal ceramide levels appear to increase. While neutral sphingomyelinase levels are unchanged in the liver and kidney, plasma S-ASM activity increases. These findings suggest that the induction of diabetes increases plasma ceramide levels by enhancing S-ASM activity (Kobayashi et al., 2013).

S-ASM in sepsis

Early studies show that ASM KO mice are protected against effects of LPS (Haimovitz-Friedman et al., 1997). However, the differential role of L- and S-ASM cannot be derived from these studies. Claus et al. (2005) find increased S-ASM activity 24 h after endotoxin application in mice; this increase did not occur after pre-treatment of mice with the FIASMA NB6. In a subsequent study, the group demonstrates an increase of the S-ASM activity after induction of sepsis in WT mice, but not in ASM KO mice (Jbeily et al., 2013). ASM KO mice exhibit enhanced basal ceramide levels in the blood but a significantly reduced ceramide response when sepsis is experimentally induced by peritoneal contamination and infection. There is an increase in bacterial burden, increased phagocytic activity and an enhanced cytokine storm in the blood after sepsis induction in ASM KO mice. Thus, in murine models of sepsis, S-ASM activity is increased, which may contribute to the effective defense of septic pathogens. S-ASM triggered hydrolysis of membrane-bound SM may play an important role in the primary defense against invading microorganisms (Jbeily et al., 2013).

S-ASM in pulmonary edema

In the context of non-cardiogenic pulmonary edema, treatment of mice with platelet-activating factor (PAF) leads to a 50% increase in lung tissue concentrations of ceramides, and this increase is thought to contribute to the development of severe edema (Göggel et al., 2004). However, in contrast to the PAF-stimulated transient accumulation of ceramide in P388D1 macrophages by de novo synthesis (Balsinde et al., 1997), this increase in ceramide is attributed to the PAF-induced release of ASM from the lungs, which results in a 28% increase in strictly Zn²⁺-dependent serum ASM activity without altering the activities of neutral sphingomyelinase or ceramide synthase. The application of ceramide-specific antisera, pharmacological disruption of ceramide formation or functional inhibition of ASM all specifically prevent PAF-triggered pulmonary edema. These data corroborate the hypothesis that PAF-induced pulmonary edema is mediated by ASM and ceramide as the second major pathway in addition to the activation of EP-3-receptors by PAF (Göggel et al., 2004).

A slow, modest but significant increase in S-ASM but not L-ASM or neutral sphingomyelinase was also noted in a pulmonary emphysema mouse model after VEGF receptor blockade in which the prompt and up to 3-fold activation of ceramide synthase was the main contributor to increased lung ceramide levels associated with alveolar septal destruction (Petrache et al., 2005). The authors speculate that, due to the high abundance of S-ASM from endothelial cells, injured lung capillary endothelial cells aggravate cellular damage in a vicious cycle, resulting in the sustained lung damage observed in patients despite smoking cessation. In a neonatal pigled model of pulmonary inflammation, S-ASM activity and ceramide concentration are reduced in serum after treatment of lungs with inhaled surfactant factor combined with the FIASMA imipramine (von Bismarck et al., 2008; Preuß et al., 2012).

Wilson disease, an autosomal recessive disorder, is characterized by an accumulation of Cu²⁺, which leads to severe consequences including progressive liver cirrhosis, neurological and psychiatric symptoms and, occasionally, anemia. Patients with Wilson disease exhibit a constitutive increase in ASM activity in the blood plasma even during treatment with D-penicillamine or trientine, which decrease total but not free Cu²⁺ concentrations (Lang et al., 2007). In isolated murine hepatocytes or cell lines, Cu²⁺ activates ASM but not neutral sphingomyelinase and triggeres ceramide release within 10–20 min after application. Both effects were absent or reduced in ASM KO cells and cells preincubated with the antioxidants N-acetylcysteine and Tiron, suggesting that Cu²⁺ may directly or indirectly stimulate ASM via oxygen radicals. In whole blood and in purified mouse leukocytes, Cu²⁺ treatment results in the stimulation and secretion of ASM into the supernatant or plasma, followed by apoptosis-like changes including shrinkage of erythrocytes and phosphatidylserine exposure. In patients, increased plasma ASM activity also correlates with a constitutively increased number of erythrocytes exposing ceramide or phosphatidylserine on the cell surface, which may mediate the rapid clearance of erythrocytes and thus result in anemia. The causal connection between Wilson disease symptoms and ASM is supported by the observation of a delayed onset of disease and protection against the development of fibrosis and liver failure in a rat model of Wilson disease treated with amitriptyline as a pharmacological inhibitor of ASM (Lang et al., 2007).

Cross-sectional clinical studies

Several studies have investigated peripheral S-ASM activities in humans under various conditions. Most of these studies analyzed serum or plasma. Some studies have also examined urine, salivary fluid, tear fluid, synovial fluid (Takahashi et al., 2000) and CSF (Mühle et al., 2013). Figure 2 summarizes all available studies using plasma or serum that also included a control group. The consideration of relative changes allows comparisons between different types of disorders. However, the studies used different methods, and S-ASM activity was related to a certain volume in most studies but to the amount of protein in others. Many of these studies suffer from the limitation of a small number of investigated cases. Furthermore, the study designs are heterogeneous; some studies used a retrospective design, while other studies used a prospective design. However, Figure 2 illustrates the variation of S-ASM in different clinical situations. Reduced activity values arise because of genetic variations in the SMPD1 gene. Very low activity values (<5%) are measured in NPD-B patients (Abe et al., 1999; He et al., 2003). Heterozygous NPD-B gene carriers exhibit intermediate S-ASM activity levels (≈20%) (He et al., 2003) (the NPD patients reported in the publications by Takahashi et al. (2000, 2002) are likely identical to those reported in Abe et al. (1999) and were therefore omitted in Figure 2). Only slightly reduced activity values (40–90%) are observed for the polymorphisms rs1050239 (Reichel et al., 2014) and rs141641266 (Rhein et al., 2013) and short repeat lengths in the signal peptide (Rhein et al., 2014). Because the minor A allele of polymorphism rs1050239 is associated with allergy, slightly reduced values of peripheral S-ASM are also observed in allergy patients (Reichel et al., 2014). Based on current information, decreased S-ASM activity is genetically determined and must therefore be regarded as a trait marker.

Until now, increased S-ASM activity has not been associated with genetic changes in the SMPD1 gene but has been related to the enhanced release of the enzyme and enhanced activation via oxidative stress, inflammation and cytokines (Marathe et al., 1998; Wong et al., 2000; Jenkins et al., 2010). Increased activity of S-ASM is induced by LPS, cytokines or oxidative stress and must therefore be regarded as a state marker of these conditions. The data in Figure 2 show no clear boundaries, and thus up to 200% activation of S-ASM is considered a mild increase. Mildly elevated activities are observed in Alzheimer’s disease (Lee et al., 2014), type II diabetes mellitus (Górska et al., 2003), stable angina pectoris, acute myocardial infarction (Pan et al., 2014), chronic heart failure (Doehner et al., 2007), post-traumatic stress disorder (Hammad et al., 2012) and in response to ionizing radiation treatment in cancer patients (Sathishkumar et al., 2005). Marked increases in activity (>200%) are observed in unstable angina pectoris (Pan et al., 2014), Wilson disease (Lang et al., 2007), sepsis (Claus et al., 2005), systemic inflammatory response syndrome (Kott et al., 2014), lymphohistiocytosis (Takahashi et al., 2002; Jenkins et al., 2013), non-alcoholic fatty liver, hepatitis C (Grammatikos et al., 2014), systemic vasculitis, inflammatory renal disease (Kiprianos et al., 2012) and in patients with alcohol dependency undergoing withdrawal (Reichel et al., 2011). In sepsis, lymphohistiocytosis, systemic inflammatory response syndrome and chronic heart failure, high S-ASM activity is associated with higher mortality (Claus et al., 2005; Doehner et al., 2007; Jenkins et al., 2013; Kott et al., 2014). In accordance with this, the plasma ceramide content is an independent risk factor for mortality in patients with chronic heart failure (Yu et al., 2015). High S-ASM activity has been related to reduced skeletal muscle strength (Doehner et al., 2007). In several studies, a negative correlation between S-ASM activity and body mass index (BMI) has been observed (Doehner et al., 2007; Reichel et al., 2011; Mühle et al., 2014), while this has not been detected in other studies (Grammatikos et al., 2014). In patients with chronic heart failure, S-ASM activity increases with age (Doehner et al., 2007), in agreement with studies conducted in mice (Kobayashi et al., 2013). By contrast, Grammatikos et al. (2014) did not observe a correlation between age and S-ASM activity. A diet rich in n-3 PUFAs reduces S-ASM activity in rats (Drachmann et al., 2007) but not in humans (measured in samples from the study by Rees et al. (2006); Lars I. Hellgren, personal communication). Indirect evidence for increased S-ASM activity based on an elevated ceramide to SM ratio in septic patients has been reported (Drobnik et al., 2003). In the plasma or CSF of patients with Alzheimer’s disease, increases in this ratio have been associated with greater cognitive progression (Mielke et al., 2011). Although increased ceramide levels could originate from an inhibition of ceramide-degrading enzymes (Satoi et al., 2005), aberrantly high concentrations of total and individual ceramide species correlate with an up-regulation of ASM transcription in patients with neurodegeneration (Filippov et al., 2012).

L-ASM activity is reduced in the lysosomes of cells from patients with I-cell disease (mucolipidosis II) who are deficient in N-acetylglucosaminyl-1-phosphotransferase activity and thus unable to add mannose 6-phosphate residues to enzymes targeted to the lysosome, including L-ASM (Wenger et al., 1976; Weitz et al., 1983). Although either increased S-ASM activity in the plasma of these patients or intracellular retention would be expected, the fate and localization of the assumed excessive enzymes remain unknown.

Longitudinal clinical studies

Several clinical studies have longitudinally investigated the activity of S-ASM, yielding interesting conclusions regarding the value of S-ASM as a biomarker. In contrast to a pronounced decrease in C-reactive protein (CRP), S-ASM activity only normalizes to a minor extent within 2 weeks in patients with systemic vasculitis undergoing immunosuppressive therapy, resulting in a dissociation of S-ASM activity and CRP-levels during the treatment of these patients (Kiprianos et al., 2012). Increased S-ASM activity was measured at baseline in male alcohol-dependent patients entering an alcohol withdrawal program. S-ASM activity steadily declined from day 0 (baseline) to days 1 and 2 and nearly reached control levels on days 7–10 (Reichel et al., 2011). Similar results were reported in an independent sample of alcohol-dependent subjects, with males exhibiting higher initial S-ASM activities compared to females (Mühle et al., 2014). In a single patient with lymphohistiocytosis, S-ASM activities slowly declined over several weeks in response to active treatment with dexamethasone (Takahashi et al., 2002). In patients with unstable angina pectoris, high S-ASM activity was detected on the day of symptom onset; only slightly normalized values were observed on day 1 after onset, and slightly increased S-ASM values were observed on the day of percutaneous coronary intervention. In patients with acute myocardial infarction who underwent conventional treatment, high S-ASM activity was observed up to at least day 7 after onset (Pan et al., 2014). In a group of septic patients, S-ASM activity was increased compared to controls at baseline. S-ASM activity decreased in survivors, while S-ASM activity further increased in non-survivors (Claus et al., 2005). Similarly, in patients with severe systemic inflammation, S-ASM activity predicted mortality in patients with low procalcitonin values (Kott et al., 2014). Serial measurements of S-ASM during the days after abdominal surgery revealed increased S-ASM values that remained high until the end of the observation period at day 5 (Kott et al., 2014). Serial measurements indicated increased ceramide/SM ratios as an indirect measure of S-ASM activity in septic patients at day 11 in non-survivors compared to survivors (Drobnik et al., 2003). These studies all suggest a relatively slow decrease in S-ASM activity in response to treatment for diseases in which S-ASM activity is initially increased. This effect may be related to the slow turnover of S-ASM. In cell culture experiments, S-ASM activity remains high over 24 h despite the inhibition of protein synthesis, whereas L-ASM activity rapidly declines (Jenkins et al., 2010). The slow turnover of S-ASM indicates that small persistent changes in the release of S-ASM result in profound and long-lasting changes in peripheral S-ASM activity. If the pathological condition is removed, such as during alcohol withdrawal treatment (Reichel et al., 2011; Mühle et al., 2014), S-ASM activity slowly decreases to control levels. This underscores the notion of increased ASM activity as a state marker of disease.

S-ASM in clinical routine

Should S-ASM activity be used as a chemistry marker in the clinical routine? A moderate reduction of S-ASM activities per se is not clearly related to the pathophysiology of disease. The association of slightly reduced S-ASM activity with allergy appears to be indirectly mediated via genetic mechanisms; however, it is unlikely that reduced S-ASM activity results in an increased risk of allergy. A severe reduction of S-ASM activity is observed in patients with NPD-B. However, all diseases that are associated with increased S-ASM activity are also associated with mechanisms that lead to activation of S-ASM, namely oxidative stress, increased cytokine production and/or endothelial damage. While ASM and its reaction product ceramide have been implicated in all of these diseases, the specific role of S-ASM in relation to L-ASM remains to be clarified. Diseases with hepatic and endothelial involvement preferentially appear to present a marked elevation of peripheral S-ASM activity. This elevation may be related to the characteristics of the endothelium as a rich source of S-ASM (Marathe et al., 1998). The mechanisms leading to marked elevations of S-ASM activity must be investigated in future studies. Because of the slow turnover of S-ASM, its measured activity may represent not only the current status but also past events.

A determination of peripheral S-ASM activity appears to be useful in suspected NPD patients and carriers and in severe acute inflammatory diseases. Increased S-ASM activity is observed in many different disease states and thus can be used as a general marker that reflects inflammation, cytokine release, oxidative stress and endothelial damage as a pathological state. Although it is not applicable as a biomarker of a single disease, S-ASM levels may be helpful for monitoring disease progression and severity and therapeutic efficiency or for evaluating the prognosis as well as the differentiation between subtypes of disorders with variable involvement of inflammatory processes. In severely diseased patients, S-ASM activity is associated with mortality and low BMI.