New fluorogenic probes for neutral and alkaline ceramidases

New fluorogenic ceramidase substrates derived from the N -acyl modification of our previously reported probes (RBM14) are reported. While none of the new probes were superior to the known RBM14C12 as acid ceramidase substrates, the corresponding nervonic acid amide (RBM14C24:1) is an efficient and selective substrate for the recombinant human neutral ceramidase, both in cell lysates and in intact cells. A second generation of substrates, incorporating the natural 2-( N -acylamino)-1,3-diol-4-ene framework (compounds RBM15) is also reported. Among them, the corresponding fatty acyl amides with an unsaturated N -acyl chain can be used as substrates to determine ACER1 and ACER2 activities. In particular, compound RBM15C18:1 has emerged as the best fluorogenic probe reported so far to measure ACER1 and ACER2 activities in a 96-well plate format.


Introduction
Besides their classically accepted role as structural cell components, sphingolipids are also a well-recognized group of bioactive lipids with significant implications in cell signaling and disease (1). This is the basis of the so-called sphingolipid rheostat, in which the relative levels of sphingosine (So) and sphingosine-1-phosphate (S1P) dictate the cell fate (2). Thus, while S1P results from phosphorylation of So by specific kinases (3), the intracellular levels of So are regulated by ceramidases, a group of amidohydrolases that catalyze the hydrolysis of ceramides (Cers) into So and a fatty acids (4). According to their optimum pH, ceramidases are classified as acid, neutral or alkaline. They differ in their primary structures, since they are coded by different genes. The five human ceramidases described so far, one acidic (AC), one neutral (NC,) and three alkaline (ACERs) differ also on their tissue distribution. (5) Thus, while AC is ubiquitously expressed, NC is highly expressed the small intestine, while the three ACERs are exclusively expressed in the skin cells. Very recently, the structure of ACER3 has been disclosed (6). Apart from their optimum pH, ceramidases show different selectivity for Cers depending on the length and degree of saturation of the N-acyl side chain (4). Thus, AC prefers C12 and C14 Cers, NC prefers C16 and C18, while ACER1 prefers C20 to C24 chains (7). The remaining ACERs are less selective, since ACER2 deamidates Cers with saturated acyl chains (8), as well as dihydroceramides (dhCers) with C18 and C20 monounsaturated chains (9), while ACER3 also hydrolyses phytoceramides with unsaturated acyl chains (10).
The role of some CDases as therapeutic targets (4) raises the interest for selective CDase inhibitors and, hence, the development of efficient protocols amenable to HTS formats. In this context, in previous works (11,12), a series of first generation fluorogenic substrates (RBM14) with different N-acyl chains (from C8 to C16) were described as efficient CDase substrates with different selectivities. However, while RBM14C12 has been characterized as an efficient AC substrate (11,13), the remaining substrates proved less selective towards NC, another interesting therapeutic target (14) for which no selective inhibitors have been developed so far.
Moreover, selective substrates for the alkaline ceramidases ACER1-3 are also lacking.
Considering the above precedents and given our continued interest in the development of selective probes for the different ceramidases, we want to report on a series of new fluorogenic substrates derived from our previously reported RBM14 probes (11)(12)(13), as well as new probes derived from an homologated version thereof (RBM15) that also incorporates the 2-(N-acylamino)-1,3-diol-4-ene framework present in the natural Cers ( Figure 1).
In order to widen the scope of our previously reported RBM14 probes (11)(12)(13), longer and/or unsaturated N-acyl chain amides have been used in this study, in particular those from oleic acid (C18:1-ω9), erucic acid (C22:1-ω9), and nervonic acid (C24:1-ω9). In addition, results from a recent report seem to indicate that ACER1 -/cells accumulate -hydroxy eicosanoic ceramide (15). Since this observation can be related with a higher selectivity of ACER1 towards this type of side chain, we have also synthesized the corresponding (R) and (S)-N-(α-hydroxyeicosanoyl) amides. For the sake of comparison, the same N-acyl substitution patterns have been applied to the new RBM15 series (Figure 1).

Synthesis
The probes described in this work were obtained by acylation of RBM14 (11) or RBM15 with the suitable carboxylic acid. Probe RBM15 was obtained from deprotection of the corresponding N-Boc precursor (RBM15-N-Boc), obtained as described in (16). anhydrous DCM (5 mL) was added dropwise to a solution of RBM14 or RBM15 (190 µmol) in DCM (5 mL) and Et3N (100 µL, equivalent to 717 µmol) under argon atmosphere. For the amides of (±)-α-hydroxyeicosanoic acids, the solution of EDC and HOBt was added dropwise over a mixture of the starting amine and the carboxylic acid. In all cases, the reaction mixture was stirred at room temperature overnight and the mixture was concentrated in vacuo. The resulted crude was flash chromatographed (silica gel) using a step gradient of DCM/MeOH (from 0 to 3 % MeOH at 1 % increments). The required amides were obtained as oils or waxy solids. Full characterization and NMR spectra can be found in the Supplemental Data. harboring the mouse ACER3 gene using 5 µl/well lipofectamine and opti-MEM, following the manufacturer's instructions. To overexpress ACER1 and ACER2, tetracycline 0.05 µM were added to HeLa T-Rex TET-ON cells for 24 h before use.

Cell lysates
Cell pellets were resuspended in the appropriate volume of a 0.25 M saccharose solution with the protease inhibitors aprotinin (1 mg/ml), leupeptin (1 mg/mml) and PMSF (100 mM). The suspension was submitted to three cycles of a 5 s sonication (probe) at 10 watts/5s resting on ice. The cell lysate was centrifuged at 600 g for 5 min. The supernatant was collected, and protein concentration was determined as specified below.

Cell lysates obtained from HeLa T-Rex ACER1-TET-ON, HeLa T-Rex ACER2-TET-ON and
HEK293T overexpressing ACER3 cells were transferred to ultracentrifuge tubes and were spun at 100000 g for 1 h at 4°C. Pellets were resuspended in a 0.25 M saccharose solution and protein concentration was determined as specified below.

Determination of protein concentration
Protein concentrations were determined with BSA as a standard using a BCA protein determination kit (Thermo Scientific) according to the manufacturer's instructions.

Ceramidase activities
Activity assays were carried out in 96-well plates at a final volume of 100 µl/well. Reaction

Ceramidase activity in intact cells
To determine ceramidase activity in intact cells, 5x10 5 cells per well were seeded in a 12-well plate. HT29 cells were transfected with 1 µg/well of a plasmid harboring the ASAH2 gene using 2 µl/well lipofectamine and opti-MEM, following the manufacter's instructions. Then tetracycline 0.05 µM were added for 24 h before use. Medium was replaced by MBCD at 1, 5, and 10 mM in DMEM with NaChol 500 M for 30 min. Then RBM14-C24:1 was added at a final concentration of 40 µM. The plate was incubated for 3 h at 37°C in 5% CO2. The reaction was stopped with 125 µl/well of MeOH and then 500 µl/well of NaIO4 (2.5 mg/ml in glycine-NaOH buffer, pH 10.6) was added. After incubation at 37°C for 1 h in the dark, 500 µl/well of

Statistics
Comparison between two means has been carried out with the unpaired two-tailed t-test and statistical differences are marked with asterisks. For comparison of more than two means, data have been analyzed by one-way ANOVA test followed by Bonferroni's multiple comparison test. Statistical differences between means are pointed out with different letters atop each bar (same letter indicates no statistical difference).

Design and synthesis of the probes
Probes RBM14 were obtained by N-acylation of the free base with the required carboxylic acids, following previously reported protocols (12,13). Probes RBM15 were inspired in our "second generation" probes for S1P lyase, in which the installation of a vinyl unit between the amino diol moiety and the umbelliferone reporter led to a significant increase of affinity, probably due to the closer similarity of the vinylogated probe with the natural substrate (16). However, even under these optimized conditions, a fluorescence of around 65% of the theoretical maximum was observed (Supplemental Figure S2).

Probes RBM14 and RBM15 as AC substrates
Probes RBM14 and RBM15 ( Figure 1) were tested as AC substrates in FD-AC cell lysates overexpressing AC. These probes were compared with RBM14C12, reported as AC substrate in a previous paper (12,13). None of the new RBM14 probes were superior to RBM14C12 as AC substrates in lysates from FD-AC cells, as evidenced by the fluorescence values observed for the released umbelliferone ( Figure 3A). This was also true for probes RBM15, since RBM15C16, the best AC substrate of this series, gave values around three times lower than those from RBM14C12 ( Figure 3B). These results are in agreement with the known preference of AC for C12-C14 Cers (7) and also with the fact that long chain Cers are not AC substrates (17).

Probes RBM14 and RBM15 as NC substrates
Probes RBM14 and RBM15 were tested as substrates in human recombinant NC and HT29 cell lysates overexpressing NC and compared with RBM14C16, which was reported in a previous by guest, on April 30, 2019 www.jlr.org Downloaded from paper as the most efficient NC substrate, among the RBM14 probes (12). In all the experiments, similar results were obtained with recombinant NC (Figure 3C,D) and with cell lysates ( Figure   3E,F), although the corresponding C16 and C18:1 probes of both series (RBM14 and RBM15) were somewhat better substrates in cell lysates, in comparative terms with the remaining probes.
The kinetic parameters in human recombinant NC are shown in Table 1. Among RBM14 substrates, RBM14C24:1 proved slightly superior in terms of affinity (KM) and efficiency (Kcat).
The preference of human recombinant NC for this N-acyl side chain was surprising, given the preference of the enzyme for C16 and C18 Cers (7,18). On the other hand, the vinylogated probes RBM15 showed a slightly lower affinity in comparison with the corresponding RBM14 probes, albeit with similar or even superior Kcat/KM parameters. Interestingly, despite the vinylogated probes RBM15 were generally well tolerated by human recombinant NC, a striking drop in affinity was observed for RBM15C16, with KM>250. Finally, both (R) and (S)-α-hydroxyeicosanoyl amides were well tolerated by NC, which showed a slight selectivity, in terms of Kcat, for the (R) isomer in the RBM15 series.

Probes RBM14 and RBM15 as ACERs substrates
Three different ACERs have been reported with an optimum activity at pH around 9. Because NC exhibits some residual CDase activity at basic pH, ACER overexpressing systems were used to test the ability of our probes to act as ACER substrates. In a previous work (12), we showed that RBM14 probes of short to medium N-acyl chain lengths (C8 to C16) were inactive as ACER1 and ACER2 substrates, showing only a moderate activity as ACER3 substrates. In  (Table 1) and a high selectivity for NC ( Figure 5A). The above promising results prompted us to test this substrate in intact HT29 cells. However, no activity was observed after the initial assays (results not shown), which was attributed to the inability of the probe to cross the cell membrane. Encapsulation into liposomes, similarly as reported by us for a related probe (16), was not feasible due to the low solubility of RBM14C24:1 in the required solvent system. Gratifyingly, the use of methyl-β-cyclodextrin (MBCD), a well-know system used to increase the solubility of non-polar substances, such as fatty acids, lipids, vitamins and cholesterol in several cell culture applications (19) proved efficient to allow the use of RBM14C24:1 as NC substrate in intact cells ( Figure 5B). A concentration-dependent effect for MBCD was observed, and an optimal concentration of 5 mM could be established.

Discussion
Cer is considered the hub of the intricate set of pathways that are implicated in the biosynthesis and degradation of sphingolipids (20). Apart from the biosynthesis from simple structural moieties in the so-called de novo pathway, Cer can also be generated by the degradation of complex sphingolipids. Moreover, Cer, as well as dihydroCer (21), can also be degraded by CDases, a type of amido hydrolases that selectively convert Cers and dihydroCers into So and sphinganine (Sa), respectively, by hydrolysis of the N-fatty acyl chain present in the above sphingolipids. Interestingly, the removal of the N-acyl chain not only is important to control the relative levels of sphingolipids inside the cell but also to dramatically alter their intracellular mobility as a result of the changes in the physical properties after N-deacylation (22). Among the different CDases, overexpression of NC has been implicated in colon carcinogenesis (14) and, hence, it has emerged as a new therapeutic target. In this context, the discovery of selective inhibitors of this enzyme should be boosted by the development of efficient, sensitive analytical methods amenable for high-throughput screening (HTS) protocols. Based on our previous works addressed at the development of fluorogenic probes for CDases (11)(12)(13), we were interested in expanding the scope and selectivity of our first generation probes (RBM14) by using alternative N-acyl chains, based on the reported preferences of CDases for the different N-acyl Cers. Based on these premises, a collection of potential RBM14 substrates has been synthesized and tested on our different CDase models. Interestingly, despite NC is selective for C16 and C18 Cers (10), the N-acyl unsaturated probe RBM14C24:1 emerged as the first selective NC substrate for recombinant hNC (Table 1 and Figure 5A). Moreover, by using MBCD as adjuvant, this substrate is also suitable to measure NC activity in intact cells. The fact that the C24:1 acyl chain of nervonic acid increases the selectivity of RBM14 probes towards NC is in full agreement with the reported structural data for this enzyme, in which a 20Å deep, hydrophobic active site pocket determines the higher selectivity of NC towards long chain acyl ceramide substrates (23). Based on these considerations, the C24:1 acyl chain would be too long to fit into the 13Å cavity found in AC active site (24).
To further expand the applicability of our probes, we designed a second generation set of potential substrates (compounds RBM15, see Figure 1). As a differential feature, these modified substrates incorporate the characteristic C4-C5 vinyl unit present in the natural Cers. Since probes RBM15 can be regarded as the vinylogated analogs of probes RBM14, their design also fits into the essence of the well-established principle of vinylogy (25). In our effort to find selective CDase substrates, probes RBM15 were tested in our five CDase models and compared with the corresponding RBM14 counterparts. The most interesting results were obtained with probes RBM15 having a long unsaturated N-acyl chain (18:1 and 22:1), which behaved as useful ACER1 and ACER2 substrates ( Figure 4A,B) and were totally devoid of activity on AC ( Figure 3A,B) and ACER3 ( Figure 4C). Since RBM15C18:1 showed much weaker activity as NC substrate ( Figure 3C-F), this probe has emerged as the best fluorogenic probe reported so far to measure ACER1 and ACER2 activities in a 96-well plate format. Although less studied than other CDases, recent studies have shown the implication of ACER1 in mammalian skin homeostasis and whole-body energy homeostasis (26) and that of ACER2 in DNA damage (27) and cell differentiation processes (28).
In summary, the fluorogenic probes reported in this work complement our previously reported probe for AC (RBM14C12) (11)(12)(13) by enlarging the scope of applicability with the discovery of RBM14C24:1 as a highly selective probe for NC and probe RBM15C18:1 for ACER1 and