RP-HPLC-fluorescence analysis of aliphatic aldehydes: application to aldehyde-generating enzymes HACL1 and SGPL1.

Long-chain aldehydes are commonly produced in various processes, such as peroxisomal α-oxidation of long-chain 3-methyl-branched and 2-hydroxy fatty acids and microsomal breakdown of phosphorylated sphingoid bases. The enzymes involved in the aldehyde-generating steps of these processes are 2-hydroxyacyl-CoA lyase (HACL1) and sphingosine-1-phosphate lyase (SGPL1), respectively. In the present work, nonradioactive assays for these enzymes were developed employing the Hantzsch reaction. Tridecanal (C13-al) and heptadecanal (C17-al) were selected as model compounds and cyclohexane-1,3-dione as 1,3-diketone, and the fluorescent derivatives were analyzed by reversed phase (RP)-HPLC. Assay mixture composition, as well as pH and heating, were optimized for C13-al and C17-al. Under optimized conditions, these aldehydes could be quantified in picomolar range and different long-chain aldehyde derivatives were well resolved with a linear gradient elution by RP-HPLC. Aldehydes generated by recombinant enzymes could easily be detected via this method. Moreover, the assay allowed to document activity or deficiency in tissue homogenates and fibroblast lysates without an extraction step. In conclusion, a simple, quick, and cheap assay for the study of HACL1 and SGPL1 activities was developed, without relying on expensive mass spectrometric detectors or radioactive substrates.

. Long-chain aliphatic aldehydes are produced during various metabolic processes. A: Schematic representation of aldehyde production in mammals during various enzymatic reactions, such as microsomal oxidation of long-chain alcohols, lysosomal degradation of prenylated proteins, peroxisomal ␣ -oxidation of 3-methyl-branched fatty acids and 2-hydroxy long-chain fatty acids, attack of plasmalogens by MPO-derived hypochlorous acid, and microsomal breakdown of phosphorylated sphingoid bases. Double black arrows point to multiple steps. B: Scheme for the Hantzsch reaction.

Acidic pretreatment of tissue homogenates
Mouse livers and brains were homogenized (1 part/20 vol) in isotonic buffer [0.25 M sucrose, 5 mM MOPS (pH 7.2), 0.1% (v/v) ethanol], and 40 l homogenates were exposed to either 0.3 M HCl or NaCl for 15 min at 65°C. Subsequently, the acid-treated samples were neutralized with NaOH. The generated aldehydes were then analyzed by RP-HPLC.
In the present work, we investigated the Hantzsch reaction for carbonyl compounds in order to allow the use of nonlabeled substrates and an inexpensive derivatization reagent (see Fig. 1B ). Tridecanal (C13-al) and heptadecanal (C17-al) were employed as model compounds and cyclohexane-1,3-dione (CHD) as 1,3-diketone, followed by RP-HPLC of the fl uorescent decahydroacridine derivatives. The fi rst application of CHD to measure aldehydes, in casu formaldehyde , was described by Sawicki and Carnes in 1968 ( 22 ).

Animals and cells
Animal studies were approved by the University Ethics Committee, and all animal experiments were performed in accordance with the institutional and national guidelines for care and use of laboratory animals. Swiss Webster mice were kept under conventional conditions in the animal housing facility of the KU Leuven, with a 12 h light and dark cycle, and fed a regular chow diet. At the time of the experiment, mice were anesthetized with an overdose of Nembutal (approximately 150 g/g body weight), and tissues were dissected. Human fi broblasts ( 23 ) and murine fi broblasts, derived from C57BL6 or Sgpl1 Ϫ / Ϫ mice ( 24,25 ), were cultured as described in the cited references . dimethylsulfoxide, tetrahydrofuran, acetonitrile, and dimethylformamide, were also stimulatory but not to the same extent (ethanol, dimethylsulfoxide), or they displayed a sharp concentration optimum (data not shown). For shorter aldehydes, e.g., C13-al, the requirement for methanol was less stringent ( Fig. 2F ). From the data obtained, we evaluated the best assay conditions, which allowed for the quantifi cation of both C13-al and C17-al. The selected parameters, based on three subsequent rounds of optimizations on standards, were 1.2 M ammonium acetate, 33 mM CHD, 14% (v/v) acetic acid, 33% (v/v) methanol, and heating at 65°C for 60 min. Other factors, which could cause interference when analyzing biological samples, were considered, such as plasmalogen breakdown at low pH or high temperature (see below) or peroxidation of polyenoic acids by extreme heat; hence, 65°C instead of 90°C was chosen (see also Ref. 28 ). The selected conditions allowed for the quantifi cation of C17-al in picomolar range (low sensitivity, 10-150 pmol; medium sensitivity, 1-10 pmol) ( Subsequently, we checked to which extent aldehydes with rather similar chain length could be discriminated: a good resolution of different long-chain aldehydes was obtained, and the elution time related to the number of carbon atoms ( Fig. 4 ).
The presence of proteins can affect the recovery of aliphatic aldehydes (e.g., Schiff base with amino groups, hydrophobic interactions, trapping in pelleted proteins or in interphase during extraction), and negative effects of proteins have been observed also in CHD-based assays containing no or less methanol [59% recovery for C6-al in presence of 16.7 mg rat liver/ml assay mixture ( 28 ); 65-72%

RP-HPLC analysis of aldehydes
According to our optimized protocol, assay mixtures [100-200 l spiked with internal standard (IS, 0.5-2.0 nmol)] were mixed with 1 vol of ice-cold HPLC-grade methanol, followed by 1 vol derivatization medium [3.6 M ammonium acetate, 100 mM CHD, 42% (v/v) acetic acid] , capping and heating for 60 min at 65°C. Upon cooling, 2 vol of methanol, relative to the initial sample volume, were added to reach a methanol concentration of 60%. The latter step was omitted when optimizing the derivatization of aldehyde standards. After a 1000 g spin, the supernatant was transferred to inserts of HPLC-vials (Alltech) and 1/50 up to 1/5 of the total volume was injected on a Symmetry® C18 column (4.6 × 150 mm; 5 m; 100 Å; Waters) equilibrated with 80% (v/v) methanol. Bound derivatives were eluted with a gradient (80-100% methanol) over 12 min and detected by fl uorimetry (Waters 2475 Multi-Wavelength Fluorescence Detector; Ex 390 nm; Em 460 nm; emission mode; EUFS 3000 or lower). Wavelength settings were based on Ex/Em spectra obtained in Aminco SPF500 fl uorimeter for the decahydroacridine derivatives of formaldehyde and tetradecanal (C14-al) (data not shown).

Optimization of assay conditions
To date, the longest aldehyde that has been analyzed via the Hantzsch reaction using CHD as 1,3-diketone was decanal (C10-al) ( 28,29 ). Based on this report, different concentrations of components, as well as pH and heating parameters, were tested to optimize the detection of C13-al and C17-al ( Fig. 2A -E ). In addition, the presence of methanol as a cosolvent, not investigated in previous reports, was demonstrated to be essential for the quantitative conversion of C17-al into its fl uorescent derivative ( Fig. 2F ). Other cosolvents, such as ethanol,  biological samples, it is suggested to use less than 4 mg tissue or less than 0.8 mg protein per 200 l assay (see Fig. 5D ).

Application to biological samples
First, we verifi ed whether our optimized method was suitable for the measurement of enzyme-produced aldehydes. As expected, incubation of recombinant human HACL1 with 2-hydroxyoctadecanoyl-CoA resulted in the formation of C17-al ( Fig. 6A ); similarly, the cleavage of sphinganine-1-phosphate by recombinant human SGPL1 could be demonstrated ( Fig. 6B ).
Then, given our intention to measure HACL1 activity in cells or tissues of patients or mice with peroxisomal disorders ( 31 ), the possible interference of plasmalogens, which are present in variable amounts in different tissues and whose levels depend on functional peroxisomes, was verifi ed. Indeed, the vinyl-ether bond of plasmalogens, present at 1-10 mol/g in some tissues (brain, sebaceous glands), is sensitive to acidic conditions ( 14 ), resulting in formation of aldehydes and, consequently, in a high background when analyzing enzymatically formed aldehydes. Comparison of the fatty aldehyde profi les of liver or brain samples with or without pretreatment with acid showed that plasmalogen hydrolysis was minimal when derivatization recovery for C7-al in presence of 50 l human plasma/ml assay mixture ( 30 ); chain length-dependent decrease of recovery, down to 8% for C10-al, in presence of 0.33 ml human plasma/ml assay mixture ( 29 )]. Nonetheless, recovery of medium-chain aldehydes (C13-al) was excellent under our conditions [33% (v/v) methanol during derivatization] in presence of increasing amounts of tissue homogenate. However, recovery of C17-al decreased (15% at 2 mg tissue/assay). Attempts to prevent this by adding detergents or increasing the amount of cosolvent were not successful (data not shown). During these trials, extra methanol was added to the derivatized samples: this increased the amount of C17-al signal considerably, without ( Fig. 5A ) or with proteins ( Fig. 5B ). Hence, the formed C17-al derivative is partially lost after cooling and spinning, perhaps due to solubility problems, and this can be counteracted by adding extra methanol after derivatization, or in the opposite sense, worsened by adding water. With this modifi cation, recoveries of IS with brain or liver homogenate were improved ( Fig. 5C, D ). More importantly, the ratio of endogenous aldehydes over C17-al ( Fig. 5E ) or C13-al (data not shown) was linear with the amount of tissue analyzed, up to 16 mg tissue/assay, which allows for reliable quantifications. When applying the method for the analysis of enzyme-derived aldehydes in was performed under optimized conditions ( Fig. 7A , B ), but it was increased at 90°C (data not shown). The acidic pretreatment caused the expected generation of aldehydes, especially hexadecanal (C16-al) and octadecanal (C18-al) in brain ( Fig. 7B ). Through derivatization of synthetic plasmalogen (1-O-1'-(Z)-octadecenyl-2-oleoyl-snglycero-3-phosphocholine), the amount of hydrolysis was estimated at 6.2% ( Fig. 7C ). Hence, the peaks seen in the nontreated conditions likely represent both free fatty aldehydes ( 32 ) and plasmalogen-derived aldehydes. This might, depending on tissue/cell type investigated and on the aldehyde produced, call for correction by appropriate blank incubations (see Fig. 6D, F ). Compared with the classical derivatization with DNPH, the CHD method is less aggressive. In fact, derivatization with CHD is performed at pH ‫ف‬ 5.5, while derivatization with DNPH, being only soluble at low pH, is generally carried out under acidic conditions (e.g., 1-2 N HCl or H 2 SO 4 ). Hence, peroxidation of polyenoic acids, resulting in formation of secondary products, such as 2,4-hexadienal, is less with CHD, as was also pointed out by Yoshino et al. ( 28 ). Moreover, as shown in this work, plasmalogen hydrolysis is limited. Furthermore, DNPH-based measurements rely on a photometric detector, sensitivity being determined by its value (23,000-28,700; solvent dependent), whereas CHD relies on a fl uorimetric detector (lowest limit at highest sensitivity, approximately 5 fmol). In addition, CHD is preferred over DNPH, as the latter is classifi ed as a dangerous chemical (risk class E4), whereas CHD is considered a safe chemical (at least in the European Union ). Moreover, CHD is much cheaper than dansylhydrazine or other (even more expensive) fl uorescent hydrazides (33)(34)(35)(36)(37) that have been used to analyze (short-chain) fatty aldehydes via HPLC. Like DNPH, these compounds react optimally under acidic conditions. Subsequently, we measured HACL1 activity in mouse liver homogenates ( Fig. 6C ). A gender dependency, so far not reported, was found (106 ± 12 mU/g liver in females, n = 4, and 71 ± 19 mU/g liver in males, n = 6). The values obtained were comparable to those previously measured in rodent liver homogenates using long-chain 2-hydroxyacyl-CoA as substrate ( 5,6,23 ). Linearity of HACL1 activity in mouse liver homogenates is limited to 5 min (at 1 mg  5 and 2 nmol, respectively) were derivatized following the optimized protocol in the presence of 33% (v/v) methanol. Upon derivatization, equal volumes of water, water/methanol, or pure methanol were added to the assay to obtain the indicated methanol concentration. After centrifugation, the supernatant was subjected to RP-HPLC, and the fl uorescent derivatives were detected by fl uorimetry (Ex 390 nm, Em 460 nm). Values represent means of two determinations. B: Effect of methanol on standards in presence of proteins. Brain homogenates (2 mg tissue) were spiked with C13-al (triangles) and C17-al (open squares) before derivatization. Upon derivatization, mixtures were processed as described for (A). Integrated peak areas of C13-al and C17-al derivatives are plotted versus fi nal methanol concentration. C. Recovery of IS in presence of proteins. Increasing amounts of brain and liver homogenates were spiked with C13-al and C17-al before derivatization. Water/methanol or methanol was added to the derivatized sample in order to have a fi nal concentration of either 30% or 60% (v/v). Derivatives were separated by RP-HPLC. Recoveries of C13-al and C17-al derivatives were plotted versus amount of tissue per assay. Open triangles: C13-al, 60% methanol; fi lled triangles: C13-al, 30% methanol; open squares: C17-al, 60% methanol; fi lled squares: C17-al, 30% methanol. D. Recovery of IS in presence of proteins. Increasing amounts of brain and liver homogenates were spiked with C13-al (triangles) and C17-al (open squares) before derivatization. Methanol was added to the derivatized sample to a fi nal concentration of 60% (v/v). Derivatives were separated by RP-HPLC. Integrated peak areas of C13-al and C17-al derivatives were plotted versus amount of tissue per assay. E. Analysis of endogenous aldehydes. Increasing amounts of brain and liver homogenates were spiked with C17-al before derivatization and separation of the fl uorescent derivatives by RP-HPLC. The ratio of endogenous C16-al (squares) and C18-al (open triangles) over C17-al standard was plotted versus amount of tissue per assay. Fig. 6. Analysis of enzymes producing fatty aldehydes. A: Human recombinant HACL1 (5 g purifi ed protein) was incubated at 37°C for 5 min with 2-hydroxyoctadecanoyl-CoA as substrate (40 M). A blank reaction was also included, where no substrate was added. The reaction mixtures were fortifi ed with IS (C13-al; 1.5 nmol) before derivatization. B: Human recombinant SGPL1 (0.8 mg protein of soluble E. coli extract) was incubated at 37°C for 60 min with sphinganine-1-phosphate as substrate (40 M). A blank reaction was also included, where no substrate was added. The reaction mixtures were fortifi ed with IS (C17-al; 2 nmol) before derivatization. C: Liver homogenates of Swiss mice (1 mg tissue, corresponding to ‫ف‬ 0.2 mg protein) were incubated at 37°C for 5 min with 2-hydroxyoctadecanoyl-CoA (40 M), mixtures were spiked with IS (C13-al; 0.74 nmol), and formed aldehydes were analyzed as described in (A). D: Liver homogenates of Swiss mice (2 mg tissue, corresponding to ‫ف‬ 0.4 mg protein) were incubated at 37°C for 60 min with sphinganine-1-phosphate (40 M), and formed aldehydes were analyzed as described in (B). E: Human fi broblasts were grown either in standard medium (Ct) or in the presence of 1 mM oxythiamine (OT) for two days before HACL1 activity measurement. Fibroblast lysates (0.2-0.5 mg protein) were then incubated at 37°C for 5 min with 2-hydroxyoctadecanoyl-CoA (40 M). Blank reactions were also included, where no substrate was added. The reaction mixtures were spiked with IS (C13-al; 0.74 nmol) before derivatization. Note the presence in fi broblasts of endogenous C16-al and C18-al, eluting at 7.4 and 8.7 min, respectively. F: Lysates of mouse fi broblasts ( ‫ف‬ 0.2 mg protein), obtained from wild-type or Sgpl1 Ϫ / Ϫ C57BL6 embryos, were incubated at 37°C for 60 min with sphinganine-1-phosphate (40 M). Blank reactions were also included, where no substrate was added. The reaction mixtures were spiked with IS (C17-al; 2 nmol) before derivatization, followed by analysis of the generated C16-al. Assays without substrate serve to correct for nonSGPL1-derived C16-al. tissue/assay), while activity of the recombinant enzyme was linear at least up to 30 min. This difference can be explained by the presence of a high acyl-CoA hydrolase activity in mouse liver homogenates (approximately 1.8 U/g mouse liver under the HACL1 assay conditions), responsible for the degradation of the substrate during incubation. SGPL1 activity could easily be measured in mouse liver homogenates (4.2 ± 1.9 mU/g liver in females, n = 3; 3.4 ± 0.4 mU/g liver in males, n = 3) ( Fig. 6D ). In agreement with previous reports ( 13 ), SGPL1 activity in liver homogenates (<4 mg tissue/assay) was linear up to 1 h of incubation.
Of clinical importance, lyase activities could also be measured in human fi broblast lysates [187 ± 29 pmol/ min/mg protein for HACL1 (n = 4), and 19 pmol/min/ mg protein for SGPL1 (n = 2)]. Hence, the assay can be employed to study enzyme defi ciencies in patients. To simulate HACL1 defi ciency, so far not yet reported, HACL1 activity was measured in human fi broblasts exposed to oxythiamine (OT), which blocks the coenzyme functions of thiamine. In fact, OT-diphosphate competes with TPP for binding to apo-enzymes ( 23 ). As reported before, OT exposure resulted in inhibition of ␣ -oxidation ( 23 ) and reduced HACL1 activity ( Fig. 6E ). Likewise, SGPL1 defi ciency in man has not been diagnosed, but a clear drop in aldehyde production was seen in Sgpl1 Ϫ / Ϫ mouse fi broblasts ( Fig. 6F ).
Note that the use of sphingosine-1-phosphate as SGPL1 substrate is not recommended for this assay. The recovery of the produced 2-hexadecenal from homogenates is not complete (unpublished data), as was also noticed by Berdyshev et al. ( 19 ). Furthermore, the Hantzsch reaction with synthetic 2-hexadecenal does not proceed well, giving rise to spurious peaks, whereas derivatization with 2,4dinitrophenylhydrazine works smoothly (unpublished data). Both observations are likely due to the reactivity of the ␣ , ␤ -unsaturated carbonyl group toward sulphydryl and amino compounds. Adduct formation of ␣ , ␤ -unsaturated aldehydes with proteins and DNA has long been known ( 38 ), and more recently, it has also been demonstrated for the lyase metabolite of sphingosine-1-phosphate ( 39,40 ).
In conclusion, a simple and cheap assay for aliphatic aldehydes is described, without an extraction step or need for a mass spectrometric detector, and it is generally applicable to enzymes producing aldehydes, as demonstrated for HACL1 and SGPL1. For quantitative measurements, an appropriate IS has to be selected, depending on the chain length of the produced aldehyde and the endogenous fatty aldehydes present in tissue/cells investigated. Likewise, the assay can be adapted to other analytical techniques in which aldehydes are quantifi ed, such as plasmalogen content/composition analyses.