ASAH2;  N-acylsphingosine amidohydrolase 2;  NCDase;  Neutral CDase;  Nonlysosomal ceramidase

The pH dependency of neutral CDase purified from mouse liver is quite broad; that is, although the enzyme exhibits the highest activity around pH 7.5, weak enzyme activity is observed under acidic (pH 4.0) and alkaline (pH 10.0) conditions (Tani et al. 2000a). The enzyme hydrolyzes the N-acyl linkage of various species of Cers but not glycosphingolipids or sphingomyelin. Cers containing Sph (d18:1) are hydrolyzed much faster than ones containing dihydrosphingosine (d18:0). Cers containing phytosphingosine (t18:0) are strongly resistant to hydrolysis by the enzyme. Interestingly, neutral CDase hydrolyzes a fluorescent analogue of Cer, C12-NBD-Cer, much faster than radioisotope-labeled natural Cers. Thus, C12-NBD-Cer is now recognized as a useful artificial substrate for measurement of neutral CDase activity. The enzyme activity is significantly enhanced by the addition of detergents. Sodium cholate has the strongest effect; that is, 1.0% of the detergent increases the enzyme activity approximately fourfold in comparison with the level in the absence of the detergent. The activity of neutral CDase is strongly inhibited by Cu²⁺, Zn²⁺, and Hg²⁺; however, EDTA does not affect the enzyme activity (Tani et al. 2000a).

Neutral CDase catalyzes not only hydrolysis of Cers but also Cer synthesis involving sphingoid bases and fatty acids (El Bawab et al. 2001; Tani et al. 2000a). The reverse reaction does not require coenzyme A, indicating that the reaction mode is completely different from that of Cer synthases. The enzyme exhibits the reverse activity in a narrow pH dependency in a neutral range as compared with that of the hydrolysis activity. Although 1.0% sodium cholate is most favorable for the hydrolysis activity of the enzyme, the reverse activity is almost completely abolished in the presence of the detergent. The reverse activity is inhibited by glycerophospholipids, especially phosphatidic acid and cardiolipin. The physiological significance of the reverse activity remains unclear; however, it is worth noting that neutral CDase can synthesize Cer in vivo when overexpressed in HEK293 cells (El Bawab et al. 2001).

In 2009, the crystal structure of neutral CDase from Pseudomonas aeruginosa was determined (Inoue et al. 2009). The enzyme is composed of two domains, a NH₂-terminal domain harboring an active site and an immunoglobulin-like COOH-terminal domain. A Zn²⁺-binding site at the center of the NH₂-terminal domain functions as the active center of the enzyme, and the reaction mechanism is similar to that of the Zn²⁺-dependent carboxypeptidase Y. The reverse activity of neutral CDase also proceeds in a Zn²⁺-dependent manner. The Mg²⁺/Ca²⁺ binding site is located at the interface between the NH₂-terminal and COOH-terminal domains, and it is suggested that Mg²⁺/Ca²⁺ may be involved in stabilization of the interaction between the two domains. These catalytic mechanisms are well conserved in mammalian neutral CDases; however, crystal structure analysis of human neutral CDase revealed that eukaryotic neutral CDases have a deep hydrophobic active site pocket stabilized by a eukaryotic-specific subdomain, which is not present in bacterial neutral CDases, and this pocket enables steric exclusion of sphingolipids with larger head groups and specific recognition of the small hydroxyl head group of Cers (Airola et al. 2015).

Mammalian neutral CDase has a signal/anchor sequence at the NH₂-terminus and is distributed at plasma membranes as a type II integral membrane protein, and thus the catalytic region faces the extracellular space (Tani et al. 2007) (Fig. 2). In addition, the enzyme is partly secreted into the extracellular space after cleavage of the NH₂-terminal signal/anchor sequence, which means that the enzyme functions as both membrane-bound and soluble proteins (Fig. 2). Mammalian neutral CDases have 9–10 N-glycosylation sites in their catalytic region and are highly N-glycosylated. In addition, the enzymes have a highly O-glycosylated domain (mucin box) that follows the signal/anchor sequence, and this domain plays an important role in the plasma membrane localization of the enzyme. Although the precise mechanism by which the mucin box contributes to plasma membrane localization of the enzyme is unknown, one possible explanation is that O-glycans on the mucin box prevent the cleavage of the signal/anchor sequence by unknown proteases. It is worth noting that zebra fish neutral CDase also has a mucin box and is localized at plasma membranes (Yoshimura et al. 2004); however, the enzymes from Pseudomonas aeruginosa, Aspergillus oryzae, and Drosophila melanogaster do not have mucin box and function as secretory proteins (Ito et al. 2014). This may suggest that the structure of the mucin box in neutral CDases was acquired during evolution for localization of the enzyme at plasma membranes as a type II integral membrane protein.

In mouse tissues, mRNA expression of neutral CDase is ubiquitously detected; however, the strongest expression is observed in the small intestine including the duodenum, jejunum, and ileum (Kono et al. 2006; Tani et al. 2000b). In rat kidney, neutral CDase is distributed at the apical membranes of proximal tubules, distal tubules, and collecting ducts, whereas in the liver the enzyme is mainly detected as a soluble protein and is distributed among endosome-like organelles in hepatocytes (Mitsutake et al. 2001). Although mouse kidney neutral CDase has O-glycans, the liver enzyme does not have both the signal/anchor sequence and the mucin box, suggesting that the posttranslational modification around the NH₂-terminal region of the enzyme occurs in a tissue-specific manner and affects the intracellular distribution of the neutral CDase (Ito et al. 2014).

The subcellular localization of human neutral CDase is somewhat controversial; that is, the enzyme is localized to mitochondria when the enzyme fused to GFP at the NH₂-terminus is overexpressed in HEK293 and MCF7 cells, whereas the enzyme fused to GFP at the COOH-terminus is distributed at plasma membranes of HEK293 cells (Ito et al. 2014). This discrepancy in the intracellular distribution of the enzyme may arise from the use of overexpression systems; thus, the precise intracellular distribution of human neutral CDase in vivo remains to be determined. It should be noted, however, that neutral CDase contributes to production of Sph in mouse brain mitochondria (Novgorodov et al. 2014).

In 2006, Kono et al. reported the establishment of neutral CDase gene knockout mice (Kono et al. 2006). The neutral CDase null mice have a normal life span and do not exhibit clear abnormal phenotypes. Although loss of neutral CDase does not cause significant changes of the Cer and Sph levels in most tissues, neutral CDase null mice exhibit an increase in the C16-Cer level in the jejunum and decreases in the Sph levels in the jejunum, ileum, and colon. Neutral CDase is detected in the intestinal tract along the brush border and intestinal contents, and neutral CDase null mice are defective in intestinal digestion of dietary Cer; thus it is suggested that neutral CDase is involved in the catabolism of dietary sphingolipids (Kono et al. 2006). Intestinal neutral CDase activity is increased in patients with ulcerative colitis, and neutral CDase null mice exhibit an increase in systemic inflammation in a mouse model of colitis; thus it is suggested that neutral CDase may be involved in protection against inflammatory bowel disease (Snider et al. 2012). In contrast, in colon cancer cells, inhibition of neutral CDase causes an increase in the Cer level, which induces apoptosis and autophagy (Garcia-Barros et al. 2016). Furthermore, neutral CDase null mice are protected from colon carcinogenesis induced by azoxymethane, a potent carcinogen of colon cancer, suggesting that neutral CDase is involved in the regulation of initiation and development of colon cancer. Thus, neutral CDase may become a novel target for colon cancer therapy (Garcia-Barros et al. 2016).

It has been reported that growth factors, cytokines, and glucocorticoids increase the activity and/or expression level of neutral CDase in hepatocytes and renal mesangial cells, and it is suggested that neutral CDase has cytoprotective effects against inflammatory stimuli in these cells. In contrast, in rat renal mesangial cells, an increase in nitric oxide production, which is associated with inflammatory disorders, induces degradation of neutral CDase and subsequently an increase in the Cer level (Ito et al. 2014).

Several lines of evidence have indicated that neutral CDase is involved in protection from apoptosis. For example, INS-1 cells, a rat β-cell line, are protected from cytokine-induced apoptosis by an increase in the expression level of neutral CDase (Zhu et al. 2008). Furthermore, downregulation of neutral CDase is involved in enhancement of apoptosis. In human keratinocytes, upregulation of the expression level of neutral and acid CDases caused by low-dose UVB results in protection from UVB-induced apoptosis, whereas high-dose UVB causes downregulation of these enzymes, which enhances UVB-induced apoptosis (Uchida et al. 2010). On the other hand, occasionally, neutral CDase positively regulates cell death; that is, in traumatic brain injury, activation of neutral CDase and reduced activity of Sph kinase 2 promote the production of Sph in mitochondria and, consequently, secondary brain injury (Novgorodov et al. 2014). In addition, mouse embryonic fibroblasts lacking neutral CDase are protected from nutrient- and energy-deprivation-induced necroptosis via autophagy and mitophagy (Sundaram et al. 2016). The opposite effects of neutral CDase on cell death may be attributed to the fact that the enzyme is involved in not only a decrease in the Cer level and an increase in the S1P level via Sph kinases, which suppress cell death, but also an increase in the Sph level, which enhances cell death.

Several transcription factors involved in regulation of the neutral CDase expression level have been identified. The first exon of the mouse liver and kidney neutral CDase genes is different from that in the brain, suggesting that the promoter region of the gene is different in each tissue, and transcriptional regulation of the gene occurs in a tissue-specific manner (Okino et al. 2002). Hepatocyte nuclear factor-4α (HNF-4α) and glucocorticoid receptor functionally bind to the liver- and kidney-type promoter regions, respectively. Human neutral CDase promoter does not show any significant similarity with the mouse neutral CDase promoter and has functional transcriptional response elements that bind to transcription factors, including AP-1, NF-Y, AP-2, Oct-1, and GATA. In HEK293 cells, serum stimulates the neutral CDase expression level via the c-Jun/AP-1 signaling pathway (Ito et al. 2014). Both mouse and human neutral CDase promoters do not have TATA and CAAT boxes, typical features of a housekeeping gene.

This review summarizes the current understanding of the structure and cellular and molecular functions of neutral CDase. The genetic information of neutral CDase is highly conserved in organisms from bacteria to mammals. In prokaryotes, neutral CDase is only found in specific bacteria, such as Pseudomonas aeruginosa and Mycobacterium tuberculosis (Okino et al. 1999). Although they cannot synthesize sphingolipids by themselves, the enzyme is thought to be involved in host infection by these bacteria (Ito et al. 2014). In contrast, eukaryotic organisms synthesize sphingolipids by themselves, and neutral CDases play important roles in the catabolism of Cer and regulation of sphingolipid-mediated signaling. Interestingly, the physiological importance of neutral CDase seems to differ among organisms. For example, the deletion of neutral CDase is embryonic lethal in fruit fly, Drosophila melanogaster; however, neutral CDase null mice do not exhibit notable phenotypic defects under normal physiological conditions (Ito et al. 2014; Kono et al. 2006). One possible explanation for this is that acid and/or alkaline CDases can complement the loss of neutral CDase because the neutral enzyme is a pan-CDase in fruit fly; however, mammals have acid and alkaline CDases (Ito et al. 2014). However, the subcellular localization of neutral CDase is quite different from those of acid and alkaline enzymes; that is, a neutral enzyme is plasma membrane-bound and secretory protein, whereas acid and alkaline enzymes are localized at lysosomes and the ER/Golgi, respectively. This means that transport of Cer, a substrate for CDases, between organelles is critical for the complementation of functions among CDases. Thus, understanding how Cer is transported between organelles will become an important issue for elucidation of the functional interaction of these CDases.

The balance of the cellular contents of Cer/Sph/S1P is thought to regulate diverse cellular responses, such as apoptosis, proliferation, and cell differentiation, and thus CDases play crucial roles in various signaling systems through regulation of the balance of these lipid signaling molecules. Recent studies indicated the possibility that neutral CDase will become a promising target for cancer therapy because inhibition of neutral CDase suppresses colon carcinogenesis through increased apoptosis of the cancer cells (Garcia-Barros et al. 2016). In this case, neutral CDase is involved in suppression of apoptosis; however, occasionally, neutral CDase positively regulates cell death (Novgorodov et al. 2014; Sundaram et al. 2016). Like neutral CDase, SMases and Sph kinases also play pivotal roles in regulation of the balance of the Cer/Sph/S1P contents and subsequently cell death and survival; thus, for elucidation of the functional complexity of neutral CDase, comprehensive understanding of the regulation of SMases and Sph kinases, together with neutral CDase, is required.