Quantification of 3-ketodihydrosphingosine using HPLC-ESI-MS/MS to study SPT activity in yeast Saccharomyces cerevisiae[S]

Serine palmitoyltransferase (SPT) catalyzes the rate-limiting step of condensation of L-serine and palmitoyl-CoA to form 3-ketodihydrosphingosine (3KDS). Here, we report a HPLC-ESI-MS/MS method to directly quantify 3KDS generated by SPT. With this technique, we were able to detect 3KDS at a level comparable to that of dihydrosphingosine in yeast Saccharomyces cerevisiae. An in vitro SPT assay measuring the incorporation of deuterated serine into deuterated 3KDS was developed. The results show that SPT kinetics in response to palmitoyl-CoA fit into an allosteric sigmoidal model, suggesting the existence of more than one palmitoyl-CoA binding site on yeast SPT and positive cooperativity between them. Myriocin inhibition of yeast SPT activity was also investigated and we report here, for the first time, an estimated myriocin Ki for yeast SPT of approximately 10 nM. Lastly, we investigated the fate of serine α-proton during SPT reaction. We provide additional evidence to support the proposed mechanism of SPT catalytic activity in regard to proton exchange between the intermediate NH3+ base formed on the active Lys residue with surrounding water. These findings establish the current method as a powerful tool with significant resolution and quantitative power to study SPT activity.

known for its eukaryotic counterparts. Different from bacterial SPT, which is a soluble homodimer, eukaryotic SPTs exist as multi-subunit complexes, including the membraneassociated heterodimeric catalytic subunits and the regulatory subunits (13,14). This was demonstrated by the purification of the SPOTS complex [SPT subunits (Lcb1/ Lcb2), Orm1/2, Tsc3, and Sac1] in yeast and the identification of Tsc3 as positive and Orm1/2 as negative regulators of SPT activity (15)(16)(17)(18). These regulatory mechanisms are conserved in the higher eukaryotic system with the identification of small subunits of mammalian SPT, ssSPTa/ssSPTb, as positive and mammalian ORMDL 1-3 as negative regulators of SPT (18,19).
Accurate measurement of SPT activity is essential to study its mechanism of action and regulation. Multiple such assays have been developed, including using radiolabeled serine or palmitoyl-CoA as substrate to measure their incorporation into the organic phase consisting mainly of radiolabeled 3KDS, measurement of release of free CoA accompanied by SPT reaction, and HPLC to measure the fluorescent derivatives of dihydrosphingosine (DHS) (20)(21)(22)(23)(24)(25). These established methods have been widely used and have contributed greatly in studying SPT activity in various organisms and their mechanism of action.
In this study, we developed a method using HPLC coupled with positive ESI followed by MS/MS to directly measure 3KDS generated by SPT. The current method represents a simplified procedure without additional derivation steps and it eliminates the usage of radioactive reagents and the associated cost and environmental issues. The method also showed high sensitivity and emerged as a reliable and quantitative tool to study SPT activity. With this tool, we obtained new insights into yeast SPT activity, including the possible existence of multiple palmitoyl-CoA binding sites in yeast SPT and the positive cooperativity between them. We report here an estimated myriocin K i value for yeast SPT for the first time. Also the method enabled us to test the proposed catalytic mechanism of SPT and the results showed that the -hydrogen of serine was completely replaced by protons from surrounding water during SPT reaction from yeast Saccharomyces cerevisiae.

Lipid analysis
Yeast was cultured in synthetic medium with 2% glucose to logarithmic phase before adding 5% TCA and was kept on ice for 15 min. Cells were spun down at 4,000 g for 5 min and washed with water twice. C17-SPH [internal standard (IS)] (50 pmol) was added to the cell pellet before lipid extraction. Cells (2-4 × 10 8 ) were resuspended in 1 ml extraction buffer containing 65% isopropanol, 13% diethyl ether, 26% pyridine, and 0.13% ammonium hydroxide and disrupted by 300 l of glass beads with a bead-beater for 3 min. The suspension was then incubated at 65°C for 15 min before centrifugation at 4,000 g for 5 min to collect the supernatant. The pellet was extracted one more time by resuspending in 1 ml of extraction buffer. The supernatants from these two extractions were combined. Solvents were dried under nitrogen gas and resuspended in 200 l mobile phase B solution (2% formic acid and 1 mM ammonium formate in methanol). An aliquot of 10 l was subjected to lipid analysis.
Lipid analysis was performed using a Thermo Fisher LC/MS system, including an Accela autosampler, a Thermo pump, and a TSQ Quantum triple quadrupole mass spectrometer. Lipid samples were injected and eluted from Agilent Poroshell 120 EC-C18 [4.6 × 50 mm, 2.7 m particle size column with 50-98% methanol gradient generated from mobile phase A (2% formic acid and 2 mM ammonium formate in water) and mobile phase B (2% formic acid and 1 mM ammonium formate in methanol)]. The mass spectrometer was operated in multiple reaction monitoring positive ionization mode, monitoring the following transitions: m/z 286. 2 3 for DHS (+3). Data were collected and processed using Xcalibur software. The transitions are also summarized in Table 1.
Quantification of sphingoid bases is based on the target analyte/IS peak-area ratio. The target analyte concentration is calculated based on the calibration curve generated by a series of peak area ratios between synthetic target analyte and IS plotted against the analyte concentration using a linear regression model. The 3KDS and DHS were quantitated using calibration curves for each synthetic compound, respectively. No synthetic standards are available for 3KDS (+2) and 3KDS (+3) and they were quantitated using 3KDS calibration curves.

Yeast microsome membrane preparation
Yeast microsome membrane fractions were prepared by the method adapted from Pinto, Wells, and Lester (26). Briefly, yeast cells cultured to logarithmic phase were collected by centrifugation and washed twice in 50 mM potassium phosphate buffer (pH 7.0). The pellet was resuspended in lysis buffer containing 50 mM potassium phosphate (pH 7.0), 2.5 mM EDTA, 5 mM DTT, and 1× protease inhibitor cocktail and disrupted with 0.5 mm-diameter glass beads using Biospec Mini-Beadbeater-16 for six cycles of 1 min beating followed by 2 min cooling on ice. Unbroken cells and debris were removed by centrifugation at 4,000 g for 10 min. The supernatant was centrifuged at 100,000 g for 1 h at 4°C. The pellet was homogenized with the lysis buffer and centrifuged at 100,000 g for 1 h. The resulting pellet was resuspended in the lysis buffer. Protein concentration was determined by BCA assay.

Measurement of incorporation of deuterated L-serine into deuterated 3KDS and DHS in yeast culture
Yeast were cultured to logarithmic phase at 30°C in synthetic medium supplemented with 2% glucose and amino acids required for the auxotrophic marker. Cells were collected by centrifugation and resuspended in fresh medium at a density of 1 OD600 (2 × 10 7 cells/ml). Around 7.6 mM of deuterium-labeled serine were added to the culture. Approximately 2-4 × 10 8 cells were collected at each time point (0, 15, 30, 60, 90, and 120 min) and treated with 5% TCA followed by washing twice with water. The 3KDS and DHS generated from deuterated serine substrate were detected as early as 15 min.

Detection and separation of 3KDS and DHS using HPLC-ESI-MS/MS
Sensitive and specific detection of 3KDS is essential to measure SPT activity. DHS is the downstream product of 3KDS in the de novo sphingolipid synthesis pathway, formed by reduction of 3KDS through 3KDS reductase (KDSR) (27). The 3KDS and DHS have similar structures with a 2 Da difference in molecular mass caused by the replacement of the keto group with a hydroxyl group at the C3 position. Clear separation and detection of each analyte is essential for accurate quantitation of 3KDS free of DHS contamination. Synthetic 3KDS and DHS compounds were applied to Agilent Poroshell 120 HPLC column EC-18 and eluted with methanol gradient in the presence of ammonium formate over a 1 h time course. A specific signature fragmentation pattern for each compound, 3KDS with a parent m/z at 300.  Table 1. Under these conditions, 3KDS and DHS showed 0.3 min difference in retention time and they showed similar intensity (Fig. 1A, B). Transition specificity was verified, as no signal in DHS transition was observed when injecting only 3KDS, and, reciprocally, no signal in 3KDS transition was observed when only DHS was injected. C17-SPH was used as IS for quantification of 3KDS and DHS. The ion product spectrum and selected reaction monitoring (SRM) peak of C17-SPH are shown in supplemental Fig. S2. Calibration lines were established with 0.3-20 pmol pure compounds in the presence of 2.5 pmol IS (Fig. 1C,  D). The amount of compounds detected below the linear range was also plotted with the peak area ratio between analyte of interest and IS against their concentrations (supplemental Fig. S3). The observed detection limit for pure 3KDS and DHS compounds was around 10 fmol with a signal-to-noise ratio above 1,000.
After successfully separating and detecting 3KDS and DHS, we analyzed the whole cell lipid extract from the WT yeast, BY4741 ( Fig. 2A). The 3KDS was detected at a level comparable to DHS, which is surprising, as it stands in contrast to the general concept that 3KDS is a very transient intermediate sphingoid base in the cell that normally falls below the detection limit. We next performed the same analysis using a yeast mutant in the same strain background, DAmPTSC10. TSC10 is an essential gene that encodes yeast KDSR, the enzyme that reduces 3KDS to form DHS, and the DAmPTSC10 strain is a mutant with decreased abundance of TSC10 transcript due to perturbation at the mRNA level (28). The results showed that attenuating TSC10 levels in the DAmPTSC10 strain resulted in marked accumulation of 3KDS and attenuation of DHS ( Fig. 2A), thus further validating the specificity of the current 3KDS detection method. Whole cell lipids from several other commonly used S. cerevisiae WT strains were also measured, and comparable amounts of 3KDS and DHS were detected in all the strains tested, suggesting that the existence of a stable pool of 3KDS is a common phenomenon in yeast (Fig. 2B).

Detection and separation of 3KDS (+2D) and DHS (+2D) generated from L-Ser (3,3-D2) using HPLC-ESI-MS/MS
Different from prokaryotic SPT, which can be recombinantly expressed and purified, most characterizations of eukaryotic SPT activity have been performed using crude cell lysates or microsome fractions, due to the difficulties in purifying soluble eukaryotic SPT. In this study, all the in vitro SPT assays were performed using yeast microsome fractions. In order to distinguish the endogenous 3KDS associated with the microsomal fraction from those acutely generated, we applied deuterated L-serine (3,3-D2) as the SPT substrate (Fig. 3A). Figure 3B shows the result of an in vivo labeling experiment where deuterium-labeled 3KDS and DHS, 3KDS (+2D) and DHS (+2D), can be detected. They separate the same way on the column as their nondeuterated counterpart, and m/z 302.2 > 270.3 and m/z 304.3 > 286.3 are the new transitions used for 3KDS (+2D) and DHS (+2D), which generate the same fragmentation patterns. With this method, endogenous 3KDS and DHS could also be monitored at the same time using the aforementioned ESI-MS/MS collision parameters (data not shown).

SPT in vitro assay to study yeast SPT kinetics and its inhibition by myriocin
SPT activity from BY4741 microsomes was measured by conversion of L-Ser (3,3-D2) and palmitoyl-CoA into 3KDS (+2D) (Fig. 4). SPT kinetics in the presence of different concentrations of L-Ser (3,3-D2) were measured by plotting the initial velocity of the reaction against L-Ser (3,3-D2) concentration (Fig. 4A). Yeast microsomes (200 g) and 200 M of palmitoyl-CoA were used in all the assays. Under all the conditions tested, product formation was linear with time for the first 15 min and, therefore, a 5 min time point was used to stop the reaction in order to obtain the initial velocity. The data were fitted using the Michaelis-Menten equation yielding a K m value for serine of 1.37 ± 0.4 mM and V max of 387.6 ± 34.2 pmol/min/mg yeast microsome. This K m for L-Ser (3,3-D2) is slightly lower than that reported for yeast SPT activity using the radiolabeled serine (26) and it is very similar to the reported K m value for purified SPT from S. paucimobilis, where SPT kinetics were measured by monitoring the release of free CoA (10).
SPT kinetics, in the presence of different concentrations of palmitoyl-CoA, were measured by plotting the initial velocity of the reaction against various palmitoyl-CoA concentrations (Fig. 4B). Yeast microsomes (200 g) and 15.2 mM of L-Ser (3,3-D2) were used in all of these assays. The data points were fitted better using the allosteric sigmoidal equation with an R-squared value of 0.96 (solid curve) than using the Michaelis-Menten equation, which yielded an R-squared value of 0.88 (dotted curve). The Hill coefficient was determined at 2.6 ± 1.2. This result suggests the existence of multiple binding sites for palmitoyl-CoA and positive cooperativity between the sites. The structure of SPT from S. paucimobilis shows that it is a homodimer with two active substrate binding pockets (29). The catalytic subunit of yeast SPT is a heterodimer composed of Lcb1 and Lcb2 (17). Although no crystal structure is yet available for yeast SPT, it is reasonable to believe the existence of two substrate-binding pockets, just like its prokaryotic counterpart. Although the active Lys residue that is critical for its catalytic activity is missing from Lcb1, it is unlikely that this would negate binding of palmitoyl-CoA to this subunit.
Only marginal deuterium-labeled DHS signal [DHS (+2D)] was detected in this assay, and the signal did not increase with the increasing concentration of either palmitoyl-CoA or L-Ser (3,3-D2), suggesting insignificant KDSR activity under the current experimental conditions. With the current method, we were also able to detect endogenous 3KDS levels, and the results showed that they were maintained at a constant level during the course of the assay (data not shown), further supporting the conclusion that there is no significant KDSR activity in the system and that 3KDS is stable over the time of the assay. Thus, the 3KDS (+2D) generated from L-Ser (3,3-D2) represents SPT activity faithfully.
Myriocin is a potent SPT inhibitor that binds to SPT directly and competes with serine to form the external aldimine with PLP in the catalytic sites (30,31). Using the SPT in vitro assay, we measured the inhibitory effects of myriocin  on yeast SPT by calculating the K i (Fig. 5). SPT reaction rates in the presence of 0.95 mM L-Ser (3,3-D2) and various amounts of myriocin, ranging from 0 to 128 nM, were measured and plotted against myriocin concentrations. A K i value of 10.3 ± 3.2 nM was calculated using the Morrison equation. In applying this equation, the K m value for L-Ser (3,3-D2) was set to 1,370 M, which was determined in an experiment without myriocin. The concentration of yeast SPT, which is required for the Morrison equation, was estimated at 0.0125 M on the basis of 54,500 Lcb2 molecules per cell, reported by Ghaemmaghami et al. (32), and with the assumption that the microsome preparation efficiency was 50%. This estimated K i value for yeast SPT is much lower than the reported myriocin K i value (967 ± 98 nM) for the purified recombinant SPT from S. paucimobilis (31), suggesting a tighter binding of myriocin to yeast SPT.

Fate of -hydrogen of serine during SPT reaction
The mechanism of SPT has been intensively studied and supplemental Fig. S4 shows the proposed PLP-dependent catalytic mechanism of SPT with L-Ser (2,3,3-D3) as the serine substrate (7-11, 33, 34). This model is composed of the following steps: Internal aldimine formed between PLP and the active lysine residue (presumably Lys366 of Lcb2 for yeast SPT) is displaced by the external aldimine formed between PLP and incoming serine substrate, in this case, L-Ser (2,3,3-D3). Binding of palmitoyl-CoA causes deprotonation of the external aldimine with the C2-deuterium extracted and transferred to Lys forming Lys-NH 2 D + , which could become Lys-NH 3 + if exchange between deuterium with surrounding water occurs. Upon nucleophilic attack on the thioester, -keto acid is formed accompanied by CoASH release. CO 2 is then released to form KDS product quinonoid, which is then reprotonated by abstraction of a proton from Lys-NH 3 + or Lys-NH 2 D + . Lastly KDS is released, and the formation of internal aldimine between Lys and PLP is restored.
It has been proposed that the -hydrogen of serine is readily replaced by a proton from medium during the SPT reaction. This is supported by in vitro SPT assays using deuterated serine at the C2 position [such as L-Ser (2,3,3-D3) in one study (35) or L-Ser (2,3,3-D3, 15N) in another (36)] as substrate and rat liver microsomes/whole cell of yeast (Hansenula ciferri)/ocean alga Emiliania huxleyi as SPT source. Instead of +3 or +4 products, labeled 3KDS product was only detected in the +2 or +3 form through GC-MS or LC-MS for each study, showing a complete loss of deuterium at the C2 position. In contrast, other studies have argued that the -hydrogen of serine is retained during SPT reaction (37). That study measured the incorporation of tritium-labeled serine (at both the  and  position) into sphingolipids in rats and results from three out of seven experiments showed -tritium was incorporated on the C2 position of sphingosine at the same ratio as -tritium incorporation into the C1 position of sphingosine.
The currently developed assay, with an exquisite ability to distinguish the masses of the products, allowed for a direct test of the fate of serine -hydrogen during SPT reaction. According to the proposed SPT catalytic mechanism, we would expect a complete replacement of a C2-deuterium with a hydrogen if there is a complete exchange between Lys-NH 2 D + with bulk water. If no exchange between Lys-NH 2 D + with bulk water occurs, we would expect to see a mixture of 3KDS products, some in which the C2-deuterium remains and some with it being replaced by one of the two hydrogens from the NH 2 D + . In order to test this with the newly developed HPLC-ESI-MS/MS method, we used L-Ser (2,3,3-D3) instead of L-Ser (3,3-D2) to determine the fate of the deuterium at the C2 position. In addition to detecting 3KDS (+2D) and DHS (+2D) using m/z 302.2/270.3 and m/z 304.3/286.3 collision pairs, we also detected significant amounts of 3KDS (+3) and DHS (+3) using m/z 303.2/271.3 and m/z 305.3/287.3 collision pairs in the in vivo labeling experiment (Fig. 6). We then confirmed this by the in vitro SPT assay (Fig. 7). The 3KDS (+3) and 3KDS (+2D) generated at each concentration of L-Ser (2,3,3-D3) were quantified and the ratio between them was calculated ( Table 2). The 18% ratio between 3KDS (+3) and 3KDS (+2D) is consistent with the probability of 3KDS (+2D) with one C13 atom per 3KDS (+2D) with primary C12 atoms. Because the natural abundance of C13 is 1.08 molecules per 100 atoms of C12 molecules, the probability for the positive ion fragment of 3KDS (+2D) containing one C13 atom is 18% (1.08% × 17). The results suggest that the signal detected by the 303.3/271.3 collision pair represents 3KDS (+2D) with one C13 atom instead of 3KDS (+3D). In order to further exclude the possibility that the signal detected with m/z 303.3/271.3 represents 3KDS (+3D), we collected MS/MS data using both L-serine with deuterium at the C2 position is a common reagent and constantly used as a substrate to study SPT activity. It is possible to cause a kinetic isotope effect (KIE) because this C2 deuterium is involved in the SPT catalytic reaction. In order to assess whether there was any KIE when using L-serine with C2-deuterium, an SPT kinetic assay was performed using L-Ser (2,3,3-D3) as substrate. The result showed that the K m value is essentially the same by measuring SPT velocity using either the 3KDS (+2D) or the 3KDS (+2D) with one C13 product (Fig. 7). This K m value, 1.3 ± 0.5, for L-Ser (2,3,3-D3) is also very similar to the K m for L-Ser (3,3-D2). The reaction rate for 3KDS (+2D) production from L-Ser (3,3-D2) and that of 3KDS (+2D) production from L-Ser (2,3,3-D3) are also very similar (Figs. 4A, 7A). These results suggest that there is no KIE when using L-serine with C2-deuterium as substrate.
In summary, some previous studies claim a complete proton exchange during SPT reaction between the lysine base and the surrounding solvent (35,36) and some argue against it (37). Our study supports a complete proton exchange during yeast SPT reaction by showing the generation of 3KDS (+2D) as the sole SPT product when using the L-Ser (2,3,3-D3) as substrate. , and parent/product ion pairs are shown in the legends on the right of the SRM peak. Also shown are the chemical structure of each possible compound [3KDS (+3D) or 3KDS (+2D) with one C13 isotope, the C13 position is labeled randomly and it could be at any position from C2 to C18 for 3KDS and C1 to C18 for DHS] and their fragmentation scheme. The 3KDS (+3) (5 pmol) and DHS (+3) (15 pmol) are generated from an in vivo labeling study where approximately 2 × 10 8 yeast cells were cultured in synthetic medium with 2% glucose to logarithmic phase before adding 7.6 mM L-Ser (2,3,3-D3) for 120 min. Total lipids were extracted and subjected to HPLC-ESI-MS/MS.

DISCUSSION
In this study, we developed a HPLC-ESI-MS/MS method to quantitatively measure 3KDS and DHS. The current approach shows great sensitivity and specificity in detecting cellular sphingoid bases. This is demonstrated by the detection of both cellular 3KDS and DHS in the absence of fumonisin B1, a ceramide synthase inhibitor that is commonly used to boost sphingoid base levels (22).
The robust quantification power of the current method is demonstrated by the detailed in vitro analysis of yeast SPT activity, which yielded several interesting observations, namely, the allosteric sigmoidal curve for palmitoyl-CoA binding, the higher myriocin binding affinity, and the complete exchange between serine -hydrogen with the protons from surrounding bulk water during yeast SPT reaction. All of these provide new insights into the catalytic mechanism of eukaryotic SPT.
A point mutation in the bovine KDSR, FVT1, has been reported to associate with bovine spinal muscular atrophy (38). More recently, mutations in human KDSR have been reported to cause recessive progressive symmetric erythrokeratoderma (39) and keratinization disorders that associ-ated with thrombocytopenia (40). We believe that the method developed here is a useful tool to further characterize these diseases by directly measuring the 3KDS level in patients. The 3KDS is considered an intermediate sphingolipid product with no cellular function of its own. Direct detection of a cellular pool of 3KDS will help to explore its possible role as a bioactive molecule.
The existence of diverse SPT substrates other than serine and palmitoyl-CoA as well as the enzyme's multiple regulatory subunits suggest that the enzyme complex is highly regulated in eukaryotic cells (26,(41)(42)(43). The approach we developed here has the potential to be modified so that it could be used in in vivo assays to help to elucidate the detailed regulatory mechanism of SPT. Fig. 7. Kinetics of SPT activity from WT yeast microsomes measured by production of 3KDS (+2) and 3KDS (+3) using L-Ser (2,3,3-D3) as substrate. Initial SPT velocity measured by production of 3KDS (+2D) (A) or 3KDS (+3) (B) per milligram of microsome per minute were plotted against different concentrations of L-Ser (2,3,3-D3) ranging from 0 to 15.2 mM using the Michaelis-Menten equation function from Garph-Pad Prism 5. Also shown are the K m and V max .