In the present study, SM deacylase from rat skin was purified to homogeneity with an apparent molecular mass of 43 kDa, an enrichment of >14,000-fold, and maximal pH and pI values of between 5.5 and 6.0 and around 8.0, respectively. The purified SM deacylase followed normal Michaelis–Menten kinetics with Vmax
of 14.1 nmol/mg/h and 110.5 µM, respectively. These properties of pH dependency and molecular weight (MW) are consistent with those (pH = 4.7, MW = 40 kDa) observed in our previous study of AD skin [19
]. However, the pI value of SM deacylase was different from our previous report using analytical IEF of a homogenate of the SC of AD skin, which detected pI values of SM deacylase, GlcCDase, aSMase and aCDase of 4.2, 7.4, 7.0 and 5.7, respectively [19
]. In this study, the enzymatic features of SM deacylase were definitely changed during the IEF, i.e., the pI value shifted from 5.0 in the skin homogenate to 8.0 in the purified enzyme with an enhancement of activity by ~200-fold after IEF (Table 1
). Those data suggest that acidic inhibitory proteins or N-linked carbohydrate moieties are removed from the pI 4.2 protein complex during the purification. This supports the possible removal of N-linked carbohydrate moieties from the purified SM deacylase. Another contrasting enzymatic property includes the substrate specificity for ceramide in which the unpurified SM deacylase active fractions (pI 4.2 fraction) from the SC of AD skin did not exhibit any activity (as detected using 14
C-labeled palymitoyl-SPH of aCDase [19
]). In contrast, the purified SM deacylase in this study followed normal Michaelis–Menten kinetics for aCDase activity (the hydrolysis from ceramide to SPH) (as measured by the release of SPH using LC-MS-MS) with its Km
values of 91.6 µM and 1.32 nmol/mg/h, respectively. He et al. [31
] already reported that their purified recombinant human aCDase protein has an acidic pH optimum and follows normal Michaelis–Menten kinetics with Vmax
of 27.8 µmol/mg/h and 389 µM, respectively. The very small Vmax
value of our purified SM deacylase for the enzymatic activity of aCDase indicates that our purified SM deacylase has a relatively weak aCDase activity compared with the previously reported one [31
]. We thought it likely that the significantly lower than expected Vmax for acid ceramidase activity is due to removal of the α-subunit via unknown mechanisms. On the other hand, SM deacylase can also play a role as GCer deacylase to hydrolyze GCer to GS and free fatty acid [19
] but determination of that activity is under investigation as further studies. Analysis by MALDI-TOF MS/MS using a protein spot with SM deacylase activity separated by 2D-SDS-PAGE allowed its amino acid sequence to be determined and then identified as the β-subunit of aCDase, which consists of α- and β-subunits linked by one disulfide bond (C31/C340). The identification of SM deacylase as the β-subunit of aCDase was also corroborated by the Western blotting of samples demonstrating that the purified SM deacylase was detectable with the β-subunit antibody at ~43 kDa in the reduced conditions, while recombinant human aCDase was detectable with the β-subunit antibody at ~43 kDa and at ~52kDa in the reduced and non-reduced conditions, respectively. Consistently, breaking the disulfide bond (C31/C340) of recombinant human aCDase by dithiothreitol (DTT) elicited the activity of SM deacylase with ~40 kDa upon gel chromatography. These results strongly support the conclusion that the purified SM deacylase is identical to the β-subunit of aCDase. aCDase catalyzes the hydrolysis of ceramides to yield SPH and fatty acid to regulate many cellular processes and had been purified to apparent homogeneity from urine [31
] and placenta [33
] whose full-length cDNA was determined by Koch et al. [34
]. The protein is N-glycosylated with rat aCDase with four glycosylation sites, whereas human aCDase has an additional two sites [35
]. Three specific glycosylation sites were shown to be required for autocleavage to occur [35
] whose structure reveals a configuration where two of them form extensive bridging interactions between the α- and β-subunits in the proenzyme [35
]. In the skin, aCDase exists especially in the epidermis, including the SC [5
] and plays an important role in SPH-1-phosphate-related signaling in keratinocytes [38
] as well as in the ceramide-degrading process in the SC [5
]. Autoproteolytic cleavage and activation in human and in rat aCDases have been documented in which the formation of a hydrogen bond between Asp-162 and Cys-143 elicits a conformational change, allowing Arg-159 to act as a proton acceptor, which in turn results in facilitating an intermediate thioether bond between Cys-143 and Ile-142 (Met-142 in rats), the site of aCDase cleavage into α- and β-subunits, which is an essential requirement for the activation [36
]. This was also corroborated by the fact that treatment of recombinant human aCDase with the cysteine protease inhibitor methylmethane thiosulfonate abrogated both the cleavage and the enzymatic activity [39
]. As for the natural activation mechanisms of aCDase, Gebai et al. [36
] speculated in their crystal aCDase model that the three-dimensional configuration of the substrate-binding channel in activated aCDase after autocleavage appears to be specific for ceramide, as acyl residue-containing sphingolipids with bulky head groups such as SM and GCer would result in steric hindrance and be unable to work as a substrate for aCDase [36
]. As for the role of the S-S cross-linking (C31/C340) between the α- and β-subunits in functional aCDase activity, Gebai et al. [36
] also reported that human and rat proenzyme aCDases contain six cysteines, four of which form two disulfide bonds. One disulfide bond stabilizes a turn in the β-subunit (C388/C392), whereas the other covalently latches the N-terminal end of the α-subunit linker to the β-subunit (C31/C340). Thus, the α- and β-subunits are intimately associated with burying a total of 894 Å2
, maintaining a heterodimeric state even after autocleavage. Taken together with the significantly lower than expected Vmax of the purified SM deacylase for acid ceramidase activity, this evidence indicates that the α-subunit of aCDase is essentially required for accelerating the enzymatic reaction but not for expressing its activity.
It is likely that in the epidermis of healthy skin, the proenzyme aCDase undergoes autocleavage into α- and β-subunits in the intracellular lysosomal system without breaking the S-S bond between the α- and β-subunits to acquire aCDase activity. Consistently, western blotting using antibodies to the α- and β-subunits reveals the aCDase protein with a 50 kDa molecular weight in the SC of healthy skin (data not shown). In contrast, in the epidermis of AD skin, as depicted in Figure 8
, we thought it likely that the β-subunit would be generated both by auto-cleavage of the covalent peptide bond between Ile-142 in the α-subunit and Cys-143 in the β-subunit and by breaking the S-S bond (C31/C340) between the α- and β-subunits of aCDase via unknown mechanisms, which leads to the induction of SM deacylase activity. Therefore, it is intriguing to determine whether the β-subunit of aCDase can be detected at 40 kDa in AD skin under the non-reduced conditions, but this is still under investigation. In this regard, we have already reported that the detectable activity of aCDase occurs in healthy SC with a decreased level in AD SC, which is paralleled by the decreased level of ceramide [5
], suggesting the possibility that the expression and activation of SM deacylase following both cleavages may contribute to the diminished activity of aCDase in the AD SC. Thus, it is probable that both cleavages result in deleting the hindrance in the enzymatic active pocket against acyl residue-containing sphingolipids with bulky head groups such as SM and GCer and leads to the expression of the activities of SM deacylase and possibly GCer deacylase, which occur as enzymatic deacylation reactions in the same active pocket as aCDase.
Based upon the above findings, we hypothesize two possible causative biological factors that underlie the expression of SM deacylase in AD skin as follows: (1) The formation of the S-S bond between the α- and β-subunits of aCDase could be impaired in AD skin, presumably due to a possible point mutation of the aCDase proenzyme, although no such point mutations are currently known to exist. (2) Breaking the S-S bond could occur more easily in AD skin than in healthy skin because of unknown mechanisms.
In conclusion, our finding that the pathogenic ceramide-degrading enzyme SM deacylase, discovered as a causative factor for down-regulating ceramide synthesis in the SC of AD skin, is identical to the β-subunit of aCDase provides an essential and deep insight into understanding the pathogenesis of AD. This should facilitate therapeutic approaches for possible specific inhibitors of SM deacylase that could be applied topically or orally to essentially abrogate the ceramide deficiency in AD skin, which would result in the essential cure of AD.