A facile method for controlling the reaction equilibrium of sphingolipid ceramide N-deacylase for lyso-glycosphingolipid production.

Lyso-glycosphingolipids (lyso-GSLs), the N-deacylated forms of glycosphingolipids (GSLs), are important synthetic intermediates for the preparation of GSL analogs. Although lyso-GSLs can be produced by hydrolyzing natural GSLs using sphingolipid ceramide N-deacylase (SCDase), the yield for this reaction is usually low because SCDase also catalyzes the reverse reaction, ultimately establishing an equilibrium between hydrolysis and synthesis. In the present study, we developed an efficient method for controlling the reaction equilibrium by introducing divalent metal cation and detergent in the enzymatic reaction system. In the presence of both Ca2+ and taurodeoxycholate hydrate, the generated fatty acids were precipitated by the formation of insoluble stearate salts and pushing the reaction equilibrium toward hydrolysis. The yield of GM1 hydrolysis can be achieved as high as 96%, with an improvement up to 45% compared with the nonoptimized condition. In preparative scale, 75 mg of lyso-GM1 was obtained from 100 mg of GM1 with a 90% yield, which is the highest reported yield to date. The method can also be used for the efficient hydrolysis of a variety of GSLs and sphingomyelin. Thus, this method should serve as a facile, easily scalable, and general tool for lyso-GSL production to facilitate further GSL research.

sodium cholate were purchased from Sigma-Aldrich. Tween 80 was purchased from Beijing Dingguo Changsheng Biotech, China. All of the other chemicals were of analytical or higher grade. Sep-Pak tC18 cartridges (500 mg sorbent) were purchased from Waters.

Protein expression and purifi cation
Recombinant SA_SCD was heterologously expressed in Escherichia coli BL21 (DE3) ( 35 ). E. coli cells containing pET23b-SA_ SCD were grown overnight at 37°C in Luria-Bertani medium containing 100 g/ml ampicillin. Auto-induction medium (ZYM-5052) ( 40 ) containing 100 g/ml ampicillin was then inoculated with the cultures, which were grown at 37°C. When the cultures reached 2.2-2.4 OD 600 , they were transferred to 16°C and grown for another 20 h. After induction, the cells were harvested by centrifugation at 8,000 rpm for 10 min at 4°C and resuspended in lysis buffer containing 20 mM Tris-HCl (pH 8.0), 500 mM NaCl, and 20 mM imidazole (20 ml buffer per 1 g cell pellet). The cells were lysed by sonication and centrifuged at 12,000 rpm for 30 min. The resulting supernatant was collected, and SA_SCD was purifi ed using a Ni 2+ -chelating affi nity column. The purifi ed protein was dialyzed against storage buffer (25 mM Tris-HCl [pH 7.4] and 10% glycerol) and stored at Ϫ 80°C. The protein concentration was determined with the Bradford method using BSA as a standard.

Enzyme activity assay
SCDase activity was measured using GM1 as the substrate. The reactions contained 100 nmol of GM1 and appropriate amounts of enzyme solution in 100 l of 40 mM sodium acetate buffer (pH 5.8) and 0.08% Triton X-100. After 5 min of incubation at 37°C, the enzymatic reactions were terminated by boiling for 5 min. The reactions were analyzed by HPLC as described below. One unit of SCDase activity was defi ned as the amount of enzyme that released 1 mol of lyso-GM1 per minute under the above conditions.

Optimization of enzymatic hydrolysis
To improve GSL enzymatic hydrolysis, 100 l reactions were assembled in 35 mM sodium acetate buffer (pH 5.8) as described below. The divalent metal cations were Ca 2+ , Mn 2+ , Co 2+ , and (31)(32)(33)(34), but only those from Pseudomonas sp. TK4 (PS_SCD) and Shewanella alga G8 (SA_SCD) have been well characterized (33)(34)(35). Both enzymes have been used in the preparation of lyso-GSLs ( 22,36,37 ). Previously, we found that SA_SCD showed higher catalytic effi ciency and broader fatty acid specifi city, making it a better biocatalyst than the commercial PS_SCD ( 35 ). However, enzymatic hydrolysis yields were generally low as a result of the equilibrium between hydrolysis and synthesis. Kurita et al. (38) reported that GSL hydrolysis has been improved by an aqueous-organic biphasic system in which the fatty acids released from the hydrolysis reaction were diffused into the water-immiscible organic phase, enhancing GSL hydrolysis in the aqueous phase. However, the yield of the biphasic system is decreased in preparative-scale reactions ( 22,36 ), in which case the extraction of fatty acid is not so effi cient because of higher substrate concentration and larger reaction volume.
In the present study, we developed an easier, alternative method that can also improve SCDase hydrolysis effi ciency but is more compatible with preparative-scale reactions. We found that the combination of Ca 2+ and taurodeoxycholate hydrate (TDC) enhanced GSL hydrolysis significantly. The utility of this method for the hydrolysis of various GSLs and SM, as well as in the large-scale preparation of lyso-GM1, was demonstrated. The mechanism of the method was discussed.

High concentrations of divalent metal cations signifi cantly enhance GM1 hydrolysis
SCDase catalyzes reversible reactions in which the N-acyl linkage of the ceramide moiety to various GSLs is either cleaved or synthesized. The equilibrium between the hydrolytic and synthetic reactions generates a low hydrolytic yield for lyso-GSLs, typically 40-60%. In our previous studies, we found that the presence of 5 mM divalent metal cations (e.g., Ca 2+ , Mn 2+ , Co 2+ , and Mg 2+ ) could enhance SA_SCD hydrolytic activity and inhibited the reverse activity ( 35 ). It is suggested that these divalent metal cations might play a critical role in controlling the reaction equilibrium. Here, we did a systematic study for the effects of various divalent metal cations (Ca 2+ , Mn 2+ , Co 2+ , and Mg 2+ ) on the reaction. As shown in Fig. 2A , addition of all the tested cations led to a signifi cant improvement of GM1 hydrolysis, and Ca 2+ shows the strongest effect. GM1 hydrolysis increased with increasing Ca 2+ concentration. Compared with the 40.7% yield in the absence of Ca 2+ , the Mg 2+ . Ca 2+ concentrations were 2.5-100 mM. The detergents were Triton X-100, TDC, sodium cholate, and Tween 80; TDC concentrations were 0.07-0.7% (w/v), and GM1 concentrations were 0.5-8 mM. Reactions containing approximately 74 mU of SA_SCD were incubated at 37°C for 12 h before HPLC analysis of the reaction products as described below.

Quantitative HPLC analysis of GM1 and GM3
GM1 and GM3 hydrolysis was analyzed by HPLC using an Eclipse Plus C18 column (5 m, 4.6 mm × 100 mm; Agilent Technologies) (supplementary Fig. 1; supplementary Table 1). For analysis of GM1 hydrolysis, the mobile phase consisted of solvent A (0.03% trimethylamine dissolved in water, adjusted to pH 7.5 with phosphoric acid) and solvent B (acetonitrile). A gradient was used at a 1 ml/min fl ow rate: 0-2 min with 20% A and 80% B, 2-7 min with 20-15% A and 80-85% B, 7-9 min with 15% A and 85% B, and 9-13 min with 20% A and 80% B. For analysis of GM3 hydrolysis, the mobile phase consisted of solvent A/solvent B (18:82) at a 1 ml/min fl ow rate. The UV detection wavelength was set to 205 nm for both GM1 and GM3, and hydrolysis was calculated as follows: represent the concentrations of either GM1 or GM3 before and after hydrolysis, respectively.

Analysis of fatty acids in the precipitate
The reactions were centrifuged at 12,000 rpm for 5 min, the precipitates were collected and combined with 10 l of 20 mM heptadecanoic acid in dimethoxyethane as an internal standard, and the mixtures were evaporated to dryness. The samples were sonicated and resuspended in 100 l of 2 M HCl in 2-propanol, and the fatty acids were extracted using 200 l of hexane. Quantitative analysis of fatty acids was performed using HPLC after derivatization with 2-bromoacetophenone. Derivatization was accomplished by drying 50 l of the extracted supernatant, mixing it with 50 l of 25 g/l 2-bromoacetophenone in acetone and 50 l of 25 g/l triethylamine in acetone for 30 min incubation at 70°C, and adding 50 l of 10 g/l acetic acid in acetone, followed by a fi nal 30 min incubation at 70°C. The derivatized samples were then dried and dissolved in 100 l of methanol. in Fig. 3B , hydrolysis of 1 mM GM1 increased with increasing TDC concentrations. Maximal GM1 hydrolysis ( ‫ف‬ 95%) was achieved in the presence of 0.28% TDC and was not further increased with the addition of up to 0.7% TDC.

Adequate TDC is crucial for effi cient hydrolysis at high substrate concentrations
To facilitate the production of lyso-GSLs on a preparative scale, higher substrate concentrations are preferred for the reactions. The hydrolysis of various concentrations of GM1 was investigated, and the optimal TDC concentration was revealed to be dependent on substrate concentration ( Fig. 4 ). Although 0.07% TDC was suffi cient to promote the hydrolysis of 0.5 mM GM1, the effi ciency began to decrease for concentrations of substrate greater than 1 mM and declined to 29.5% at 4 mM GM1. The addition of 0.28% TDC enhanced the hydrolysis of 4 mM GM1 to 90.8%, and hydrolysis decreased again when the substrate concentration was increased to 6 mM. Further increasing the TDC concentration to 0.56% signifi cantly improved hydrolysis at high substrate concentrations, reaching levels of 91.3% hydrolysis even at 8 mM GM1 ( ‫ف‬ 12.7 mg/ml).

Hydrolysis of various sphingolipids is also enhanced by this method
The hydrolysis of various GSLs (GM1, GM3, GluCer, and GalCer) and SM was signifi cantly improved using the optimized conditions ( Table 1 ). The hydrolysis of the gangliosides GM1 and GM3 was enhanced to as high as 96%, and the hydrolysis of GluCer, GalCer, and SM improved from 3.07, 8.22, and 23.32% to 66.34, 85.44, and 66.08%, respectively. These results demonstrate that this new method can be used as a general strategy for improving the hydrolysis of various sphingolipids.

Application to large-scale preparation of lyso-GM1
To test the feasibility of this method on a preparative scale, 100 mg of GM1 was hydrolyzed using the optimized hydrolysis conditions. After 12 h of reaction time, 75 mg of pure lyso-GM1 was isolated using Sep-Pak tC18 cartridges (90% yield). The yield was much larger than has previously reported for the aqueous-organic biphasic system (62-72% yield with 2 weeks of reaction time) ( 22,36 ). In addition of 100 mM Ca 2+ enhanced the yield to 81.6% ( Fig. 2B ).

Detergents are essential for effi cient enzymatic hydrolysis
Detergents have also been shown to play important roles in enzymatic hydrolysis of GSLs ( 38 ). Although the GM1 hydrolysis yield was 81.6% in the presence of 0.07% Triton X-100 and 100 mM Ca 2+ , it dropped sharply to 17.5% in the absence of Triton X-100 ( Fig. 3A ). Therefore, we further investigated the effects of various detergents on the lyso-GM1 production. As shown in Fig. 3A , TDC, sodium cholate, and Tween 80 also improved GM1 hydrolysis in the presence of Ca 2+ , further demonstrating the crucial role of detergents. TDC had the strongest effects of the tested detergents, improving the hydrolysis to 88.2% ( Fig. 3A ). Interestingly, TDC-based enhancement of hydrolysis required the presence of Ca 2+ : in the absence of Ca 2+ , GM1 hydrolysis was only 2.9%, 14-fold lower than the same case of Triton X-100 ( Fig. 3A ), suggesting that the enhancement of hydrolysis results from the joint effects of TDC and Ca 2+ .
To further enhance GM1 hydrolysis, the concentration of TDC in the reaction mixture was optimized. As shown  This method is applicable to many GSLs and is easily scalable for large-scale preparation of lyso-GSLs. Kurita et al. (38) developed an aqueous-organic biphasic system to increase GSL hydrolysis. The fatty acids released from GSLs were extracted from the aqueous phase with an organic solvent to push the reaction equilibrium toward hydrolysis; however, it was necessary to keep a low substrate concentration, or else the effect of fatty acid extraction would be reduced. Additionally, at least a 5-10 times greater volume of organic solvent relative to the aqueous buffer was required for effi cient enhancement of hydrolysis. Consequently, on a preparative scale, the reaction system would necessarily be very large, which would be diffi cult to handle for downstream purifi cation. These limitations make the aqueous-organic biphasic system ineffi cient for the preparation of lyso-GSLs, especially on a larger scale. As two examples, lyso-GM1 was prepared at the 10 to 20 mg scale using the biphasic system ( 22,36 ). After 2 weeks of reaction, the yield of lyso-GM1 was 62-72% ( 22,36 ), which is similar that produced by chemical catalysis ( 19 ). In contrast, 90% lyso-GM1 yield was achieved after 12 h of reaction time in this report, highlighting the increased effi ciency of the new method.
Moreover, the principle of the strategy we used is entirely different from that used by Kurita et al. ( 38 ). It has been known for a long time that fatty acids can form waterinsoluble complexes with divalent cations ( 41 ), so we speculated that the addition of divalent metal cations might reduce the level of free fatty acids in the reaction and promote hydrolysis. Indeed, all of the tested divalent metal cations (Ca 2+ , Mn 2+ , Co 2+ , and Mg 2+ ) enhanced GSL hydrolysis, and high concentrations of Ca 2+ were optimal. A considerable quantity of released fatty acids was observed in the precipitate at the bottom of the reaction tube, which is consistent with our explanation.
Interestingly, we also found that the presence of detergent was crucial for the enhancement of GSL hydrolysis. Without detergent in the reaction, GSL hydrolysis was very low even with high Ca 2+ concentrations. The importance of detergent has also been found in other GSL-hydrolyzing enzymes ( 42,43 ), although the mechanism is not known ( 42 ). However, the conformation of GSL substrate is clearly affected by detergent, which may assist the enzyme to recognize the substrate ( 42,43 ). Furthermore, the effi cient hydrolysis of high concentrations of GM1 required an adequate amount of TDC. This might be because the GSL has a better accessibility for SA_SCD when detergent and GSL were mixed at an optimal ratio in the micelle. Similar effects have been observed for other lipid-processing enzymes, such as sphingomyelinase ( 44 ), snake venom phospholipase A2 ( 45 ), and phospholipase C ( 46,47 ).
Lyso-GSLs are important molecules in sphingolipid research because they can be used as intermediates for synthesizing GSL analogs, which are useful tools for GSL biology study and drug development, among other uses. The new method described here provides an easy, efficient, and general means of producing lyso-GSLs on a preparative scale and might facilitate GSL research and the development of new drugs. fact, to the best of our knowledge, it is the highest reported yield of lyso-GM1.

Enhancement of hydrolysis might result from the precipitation of stearic acid
After the hydrolysis of GM1, an obvious precipitate was observed at the bottom of the reaction tube ( Fig. 5A ). We speculated that this precipitate could be stearic acid calcium salt ( Fig. 1 ). To test this hypothesis, we used an HPLC-based method to quantitatively analyze the stearic acid in the precipitate. Indeed, stearic acid was detected in the precipitate, and the amount observed was directly proportional to the amount of GM1 hydrolyzed in the reaction ( Fig. 5B ), explaining why the reaction equilibrium was shifted toward hydrolysis.

DISCUSSION
SCDase can hydrolyze GSLs to generate lyso-GSLs, but its application has been limited as a result of the low yield caused by the reaction equilibrium. In this study, we present an easy method for controlling the reaction equilibrium through the addition of Ca 2+ and TDC to the reaction. Reactions contained 0.5 mM GM1, 0.5 mM GM3, 0.2 mM GluCer, 0.2 mM GalCer, or 0.2 mM SM in 35 mM sodium acetate buffer (pH 5.8) with 0.07% Triton X-100 before optimization or in 35 mM sodium acetate buffer (pH 5.8) with 0.28% TDC and 100 mM CaCl 2 after optimization. Values represent the mean ± SD (n = 3).