High-performance liquid chromatography determination of glyoxal, methylglyoxal, and diacetyl in urine using 4-methoxy-o-phenylenediamine as derivatizing reagent

Analytical Biochemistry 449 (2014) 52–58

Contents lists available at ScienceDirect

Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio

High-performance liquid chromatography determination of glyoxal, methylglyoxal, and diacetyl in urine using 4-methoxy-o-phenylenediamine as derivatizing reagent Armando Gómez Ojeda a, Katarzyna Wrobel a, Alma Rosa Corrales Escobosa a, Ma. Eugenia Garay-Sevilla b, Kazimierz Wrobel a,⇑ a b

Department of Chemistry, University of Guanajuato, 36000 Guanajuato, Mexico Department of Biomedical Sciences, University of Guanajuato, 36000 Guanajuato, Mexico

a r t i c l e

i n f o

Article history: Received 18 September 2013 Received in revised form 20 November 2013 Accepted 11 December 2013 Available online 20 December 2013 Keywords: Glyoxal Methylglyoxal Diacetyl Urine HPLC

a b s t r a c t Bioanalytical relevance of glyoxal (Go) and methylglyoxal (MGo) arises from their role as biomarkers of glycation processes and oxidative stress. The third compound of interest in this work is diacetyl (DMGo), a component of different food products and alcoholic beverages and one of the small a-ketoaldehydes previously reported in urine. The original idea for the determination of the above compounds by reversed phase high-performance liquid chromatography (HPLC) with fluorimetric detection was to use 4-methoxy-o-phenylenediamine (4MPD) as a derivatizing reagent and diethylglyoxal (DEGo) as internal standard. Acetonitrile was added to urine for matrix precipitation, and derivatization reaction was carried out in the diluted supernatant at neutral pH (40 °C, 4 h); after acidification, salt-induced phase separation enabled recovery of the obtained quinoxalines in the acetonitrile layer. The separation was achieved within 12 min using a C18 Kinetex column and gradient elution. The calibration detection limits for Go, MGo, and DMGo were 0.46, 0.39, and 0.28 lg/L, respectively. Within-day precision for real-world samples did not exceed 6%. Several urine samples from healthy volunteers, diabetic subjects, and juvenile swimmers were analyzed. The sensitivity of the procedure proposed here enabled detection of differences between analyte concentrations in urine from patients at different clinical or exposure-related conditions. Ó 2013 Elsevier Inc. All rights reserved.

Introduction Oxidative stress and glycation reactions play a detrimental role in the development of chronic degenerative diseases and in aging processes. In routine clinical control of patients, the evaluation of appropriate biomarkers provides information on endogenous processes and also on adherence to dietary recommendations. For such purposes, urine is a convenient sample because of its easy noninvasive collection and relatively simple chemical matrix as compared with other biofluids or tissues. It has been demonstrated that the analysis of spot samples or samples of morning void urine offers reliable and useful results in the follow-up of patients in human exposure-related and epidemiological studies [1–3]. The two small a-ketoaldehydes, methylglyoxal (MGo)1 and glyoxal (Go), are generated endogenously, mainly during metabolic ⇑ Corresponding author. Fax: +52 473 7326252. E-mail address: [email protected] (K. Wrobel). Abbreviations used: MGo, methylglyoxal; Go, glyoxal; AGE, advanced glycation end product; DMGo, diacetyl (or dimethylglyoxal); 4MPD, 4-methoxy-o-phenylenediamine; DEGo, diethylglyoxal (or 3,4-hexanedione); HPLC, high-performance liquid chromatography; TEA, triethylamine; SPE, solid phase extraction; FLD, fluorimetric detector; RSD, relative standard deviation; DL, detection limit; QL, quantification limit. 1

0003-2697/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ab.2013.12.014

conversion of glucose and oxidative degradation of lipids [4,5]. Their presence in fermented beverages and in food products should also be mentioned as a potential exogenous source [6,7]. Within the cell, MGo and Go form adducts or cross-links with biomolecules, thereby compromising their biological activity and inactivating antioxidant machinery [8]. Under normal physiological conditions, both molecules are efficiently scavenged by the glyoxalase system, aldose reductase, betaine aldehyde dehydrogenase, and 2-oxoaldehyde dehydrogenase [9,10]; however, the impairment of enzymatic defense as well as the increased concentrations of MGo and Go have been associated with chronic diseases and aging [8,11]. Because the two compounds act as the precursors of advanced glycation end products (AGEs) and are considered as biomarkers of lipid peroxidation [4,12], their determination in clinical samples is relevant. In particular, it has been proposed that monitoring of MGo and Go in diabetic patients would help to assess the risk of progression of diabetic complications [13–16]. Diacetyl (DMGo), the third compound of interest in the current work, has also been associated with oxidative and carbonyl stress as a potential mediator of electron transfer reactions, an intermediate of Maillard processes and precursor of AGEs [17]. This minor metabolite of acetaldehyde derived from ethanol is easily reduced

53

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

to acetoin and 2,3-butanediol [18]; therefore, it has not always been detected in biological samples [19,20]. On the other hand, diacetyl is present in many beverages and various food products as a metabolite of microbial fermentation or as a buttery flavor additive [21,22]. Its determination in urine would be of interest in studies on alcohol toxicity and addiction [17] and in evaluation of recent exposure to exogenous sources [21]. In this regard, recent findings showed that exogenous dicarbonyl compounds react with digestive enzymes, which reduces their bioavailability and might favor their elimination in urine [23]. Quantification of a-ketoaldehydes in clinical samples has often been reported; however, only few studies focused on urine analysis (Table 1) [20,24–31]. It is worth noting that urine has been suggested as the most practical sample for such analysis because spontaneous de novo formation of MGo from triose phosphates occurring in more complex biological matrices can be avoided [27,32,33]. Brief reviews and comparisons of analytical procedures, the great majority of them based on suitable precolumn derivatization followed by chromatographic or electrophoretic separation, can be found in the introductory parts of several previous articles [24,25,29,31,34] and in comprehensive reviews [9,35]. In regard to liquid chromatography, the use of 1,2-diamino-substituted aromatic compounds that yield fluorescent quinoxalines should be highlighted [36]. In particular, 1,2-diaminobenzene [12,33], 1,2-diamino-4,5-dimethoxybenzene [27,32,37], 1,2-diamino-4,5methylenedioxybenzene [38], 4,5-dichloro-1.2-diaminobenzene [20], and 2,3-diaminonaphthalene [30] has been reported so far. As already mentioned, diacetyl has rarely been detected in urine (Table 1), and some authors used this compound as internal standard [25,27,29]. It is worth noting, however, that 4-methoxy-ophenylenediamine (4MPD) was proved to be useful for fluorimetric determination of diacetyl in wine [39]. The goal of this work was to establish a new procedure for the determination of Go, MGo, and DMGo in urine at physiological levels. To this end, diethylglyoxal (DEGo) was proposed as internal standard, 4MPD was examined as a derivatizing agent, and the fluorescent quinoxalines were separated by reversed phase highperformance liquid chromatography (HPLC). Using an original sample pretreatment, the results obtained in the analysis of realworld samples demonstrated that the proposed procedure would

enable quantification of the three compounds in samples from subjects presenting diverse exposure-related or clinical conditions. Materials and methods Instrumentation An Agilent series 1200 liquid chromatographic system equipped with a quaternary pump, a well plate autosampler, a column oven, a fluorimetric detector, and a ChemStation (Agilent Technologies, Palo Alto, CA, USA) was used; the chromatographic column (Kinetex C18, 150  3 mm, 2.6 lm) and the C18 guard column were obtained from Phenomenex (Torrance, CA, USA). Chemicals and samples All chemicals were of analytical reagent grade. Deionized water (18.2 MX cm, Labconco, Kansas City, MO, USA) and HPLC-grade acetonitrile (Fisher Scientific, Pittsburgh, PA, USA) were used throughout. The standard solutions containing 1 mg/mL Go (ethanedial, Fluka), MGo (2-oxopropanal, Sigma), and DMGo (butane-2,3-dione, dimethylglyoxal, Fluka) were prepared in deionized water. The following Sigma reagents were also used: 4MPD (derivatizing reagent), hydrochloric acid, acetic acid, potassium phosphate dibasic, boric acid, sodium hydroxide, 2-mercaptoethanol, sodium chloride, and triethylamine (TEA). The first morning urine samples were provided by volunteers characterized as follows: three healthy adults, these same adults after alcohol ingestion the night before, three members of a youth swimming team, and three diabetic patients. Additional samples from healthy individuals were used for the evaluation of the method detection and quantification limits. Procedures Small dicarbonyl compounds are unstable yet ubiquitous; hence, special care was needed during preparation of standards, reagents, and samples. In particular, all aqueous solutions were purified by derivatization of potentially present dicarbonyls with

Table 1 Some examples of analytical procedures proposed for the determination of Go, MGo, or DMGo in urine. Urine

Reagent

Analytical technique

Detection limits and concentrations found (range or mean ± SD) Go

MGo

References DMGO

DL

c

DL

c

DL

c

Not specified Not specified Healthy

DCDB TRI DDB2

GC–ECD HPLC–FLD HPLC–FLD

– 32 pmol –

– 11 pmol –

1.73 ± 0.04 nmol/mga 2.10 lM nd

[20] [28] [27]

DDP

HPLC–DAD/FLD

5.30 lg/L

–

nd

[26]

Not specified Diabetic Diabetic Control diabetic

DAN DAP DDB1 TBA

SBSE–HPLC–DAD GC–FID GC–FID CE–AD

15 ng/L – 20 lg/L 1.0 lg/L

– 50 lg/L 10 lg/L –

nd nf nf nd

[30] [29] [25] [24]

Control diabetic

TRI

HPLC–FLD

0.16 lg/L

0.43 lg/L

nf

[31]

Healthy diabetic

4MPD

HPLC–FLD

0.46 lg/L

nd 1.50 lM 20–100 lM 1.4–7.2 mg/L 2.2 ± 0.7 lg/mga 24.07 lg/L 0.30–0.90 lg/mga 94.1 ± 3.2 lg/L 170–250 lg/L 190–360 lg/L 11.7–12.2 lg/L 29.4–127.2 lg/L 0.1–0.3 lg/mga 2.0–3.8 lg/mga 17.3–27.0 lg/L 53.8–249 lg/L

860 lg/L 99 pmol –

Healthy

nd 13.18 lM 50–250 lM 2.9–14.9 mg/L 4.7 ± 1.4 lg/mga 19.02 lg/L 0.43–1.50 lg/mga 268.9 ± 6.3 lg/L nd 170–400 lg/L 20.1–21.1 lg/L 64.1–71.4 lg/L 0.30–1.1 lg/mga 0.57–0.84 lg/mga 17.0–43.2 lg/L 71.2–175 lg/L

0.28 lg/L

13.2 ± 1.6 lg/L 64.6 ± 3.4 lg/L

This work

6.71 lg/L 25 ng/L 40 lg/L 10 lg/L 0.2 lg/L 0.44 lg/L 0.39 lg/L

Note: SD, standard deviation; DL, detection limit; c, concentration; nd, not determined in this work; nf, not found; AD, amperometric detection; CE, capillary electrophoresis; DAN, 2,3-diaminonaphthalene; DAP, 1,2-diaminopropane; DCDB, 4,5-dichloro,1.2-diaminobenzene; DDB, 2,3-diamino-2,3-dimethylbutane; DDB2, 1,2-diamino-4,5-dimethylenedioxybenzene; DDP, 5,6-diamino-2,4-hydroxypyrimidine sulfate; DMB, 1,2-diamino-4,5-dimethoxybenzene; SBSE, stir bar sorptive extraction; TBA, 2-thiobarbituric acid; TRI, 6-hydroxy-2,4,5-triaminopyrimidine. a Normalized to urine creatinine.

54

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

4MPD and subsequent elimination of respective quinoxalines by solid phase extraction (SPE) (Supelclean LC-18 SPE Tube, 3 mL, product no. 57012, Supelco). For 4MPD purification, a cartridge was conditioned with 3 mL of acetonitrile followed by 3 mL of acetonitrile/water (30:70), and such prepared cartridge was dried. Afterward, 1 mL of 4MPD (20 mg/mL) in acetonitrile/water (30:70) was passed at a flow rate of 0.5 mL/min, and 10 lL of 2mercaptoethanol was added to the final solution. For purification of phosphate solutions, an aliquot of 4MPD was added. The mixture was left overnight at 60 °C and then purified from quinoxalines by SPE after cartridge conditioning with 3 mL of acetonitrile and 3 mL of water. Working solutions of Go, MGo, DMGo, and DEGo (5 mg/L each) were prepared daily using freshly purified deionized water. After collection, 1.5 mL of urine sample was moved to an Eppendorf tube, transported to the laboratory under light protection, and stored at 20 °C for up to 1 month before analysis. An aliquot of urine or calibration standard solution (200 lL) was mixed with 20 lL of internal standard (5 lg/mL DEGo) and 200 lL of acetonitrile. The sample was frozen at 20 °C for 30 min and centrifuged (10,000g, 5 min). For derivatization, 150 lL of supernatant was mixed with 240 lL of water and 60 lL of phosphate buffer (500 mM, pH 7.4), and after vortex 10 lL of purified 4MPD (20 mg/mL) was added. The mixture was incubated at 40 °C for 4 h. Finally, the sample was acidified with 10 lL of HCl (3 M), diluted with 200 lL of acetonitrile, saturated with sodium chloride, vortexed, and centrifuged (10,000g, 5 min), with 75 lL of acetonitrile layer being mixed with 150 lL of mobile phases A and C (1:1) containing 5% 2-mercaptoethanol. The injection volume to an HPLC–FLD (fluorimetric detector) system was 20 lL. Separation of Go, MGo, DMGo, and DEGo quinoxalines was accomplished using three mobile phases (A [water], B [acetonitrile], and C [0.8%aceticacid and 0.6%TEA], pH 4.3) and the following gradient elution: 0 to 1 min, 70% A, 20% B, 10% C; 1 to 6 min, 50% A, 40% B, 10% C; 6 to 8 min, 20% A, 70% B, 10% C; 8 to 10 min, 10% A, 80% B, 10% C; 10 to 10.2 min, 70% A, 20% B, 10% C; 10.2 to 12 min, 70% A, 20% B, 10% C; with a total flow rate of 0.5 mL/ min. For fluorimetric detection, excitation and emission wavelengths were set at 344 and 420 nm, respectively. The calibration was carried out using a series of mixed standard solutions containing Go, MGo, and DMGo at 0, 25, 50, 75, 100, 150, and 200 lg/L each and 500 lg/L internal standard; however, it is worth noting that the linear response was observed at least up to 1000 lg/L without changing the attenuation of the FLD. For recovery evaluation, 3 and 6 lL of the mixed standard solution (5 lg/mL of each analyte) was added to urine samples, yielding 75 and 150 lg/L of each compound added to urine. Statistical analysis The results presented are the means obtained for three replicates. Standard deviations were calculated and graphs were plotted using Microsoft Excel 2007. Data were tested for statistical significance using the unpaired t test included in Microsoft Excel 2007. The significance was established at P < 0.05. Results and discussion The main difficulties in the analysis of Go, MGo, and DMGo in clinical matrices encompass their low concentrations, high reactivity, possible polymerization or formation of adducts with sample components, and de novo generation during sample handling as well as a risk of sample contamination from reagents, air, and water [9,27,35]. In this regard, sample pretreatment, derivatization, contamination control, and protection against oxidation

become the most critical parts of any analytical procedure. In particular, the conditions applied at procedural stages before and during derivatization should be as mild as possible, avoiding drastic changes of chemical conditions and prolonged heating. In this work, all solutions were purified prior to their use, as described in Materials and methods, and urine samples were stored in new plastic tubes at 20 °C under light protection. The internal standard was added directly to the sample; DEGo was applied for this purpose owing to its structural similarity to the three aketoaldehydes and because of its absence in biological samples. As described in the introductory paragraphs, several derivatives of 1,2-diaminobenzene had been used before to obtain fluorescent quinoxalines; however, depending on the chemical structure and biological sample, quite different reaction conditions were reported [12,20,27,32,33,36,37]. Specifically in the analysis of urine, the pH values applied during derivatization ranged from less than 2.0 pH 10.0, the reaction times varied between 10 min and 15 h, and the temperatures ranged from room temperature up to 75 °C [20,24–31] (Table 1). Aiming at a high yield of fluorophore formation at neutral pH, 4MPD was examined as a derivatizing agent (the reaction scheme and respective fluorescence spectrum of quinoxaline are presented in Figs. 1S and 2S of the online Supplementary material). To avoid oxidative modification of this reagent, 2-mercaptoethanol was added to the solution (Fig. 3Sa shows a photo of different urine samples after the addition of 4MPD without antioxidant protection). It is worth mentioning that sodium bisulfite and hydrosulfite were also tested as antioxidants, but their presence impeded the derivatization reaction (the chromatograms obtained for Go and MGo standard mix after the addition of 2-mercaptoethanol, sodium bisulfite, and sodium hydrosulfite are presented in Fig. 3Sb). In the first approach, the effects of temperature, time, and pH applied during derivatization were examined individually for each compound using approximately 20:1 molar excess of 4MPD reagent. Without pH adjustment, for relatively short periods of time (<1 h) and a temperature range from 25 to 80 °C, fluorescence signals of four compounds were increasing gradually; however, starting from 60 °C, poorer repeatability was observed and the signals tended to decrease, especially in the presence of urine matrix. Afterward, longer periods of incubation were examined setting temperature at 40 °C, and the results obtained are presented in Fig. 1. As can be observed, signal repeatability for all four compounds after 4 h of incubation was excellent (relative standard deviation [RSD] <1.5% for 10 independent replicates). According to Akira and coworkers [27], prolonged reaction time in diluted urine would favor liberation of a-ketoaldehydes from their possible adducts and formation of 1,2-diaminobenzene derivatives, so 4 h of incubation at 40 °C was adopted in this work. Fig. 2 shows the effect of pH (in the range of pH 2.0–10.0, Britton–Robinson buffer, 0.1 M) on the individual signals of four a-ketoaldehyde derivatives; this parameter was essential for Go and MGo, and the observed changes might be associated with polymerization processes that make both compounds less reactive upon derivatization [40,41]. In this regard, our results are consistent with computational studies reporting different polymerization pathways for MGo and Go [40]. In the cited work, hydrated structure was found to be unstable for MGo and its dimerization via aldol condensation was thermodynamically favored, which might explain the decrease of MGo fluorophore signal in alkaline conditions (Fig. 2). On the contrary; for Go, fully hydrated structures were thermodynamically favored and the polymerization pathway proposed was by acetal dimerization [40], in agreement with lower derivatization yield of Go observed in acid medium (Fig. 2). Analytical signals of DMGo and DEGo were not affected by pH changes (Fig. 2) because both molecules have two alkyl groups in their structures that obstruct polymerization. It should be noted that

55

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

Go

MGo

DMGo

150

DEGo

125

FLD, Ex=344, Em =420

FLD, Ex=342nm, Em=425nm

200 160 120 80

MGo DMGo DEGo 0

10

20

30

40

50

60

Acetonitrile (%) 0

Go

2

4

6

Time, (h)

MGo

DMGo

DEGo

250

FLD, Ex=344, Em=420

Go

50

0

Fig.1. Fluorescence signals of glyoxal, methylglyoxal, and diacetyl (500 lg/L each) and diethylglyoxal (1000 lg/L) after derivatization with 4MPD obtained for different reaction times at 40 °C. For clarity of presentation, final reaction conditions were used while varying incubation times (see Materials and methods).

200 150 100 50 0

75

25

40 0

100

2.0

4.0

6.0

8.0

10.0

pH Fig.2. Effect of pH applied during derivatization reaction on quinoxaline signals of glyoxal, methylglyoxal, and diacetyl (500 lg/L each) and diethylglyoxal (1000 lg/ L). For clarity of presentation, final reaction conditions were used while varying pH values (see Materials and methods).

neutral pH and incubation at 4 °C for 24 h has been previously reported for MGo and Go derivatization with 2,3-diaminonaphthalene in blood plasma [11,42]; the pH conditions selected in this work are very similar, but reaction with 4MPD could not be completed within a reasonable time at temperatures lower than 40 °C. These relatively mild reaction conditions (pH 7.4, 40 °C, 4 h in dark) seem to be well suited for urine in terms of avoiding possible transformations and de novo formation of analytes [32,43]. In the analysis of more complex clinical samples such as blood serum, plasma, liver, and other tissues, acid treatment has been recommended prior to derivatization for release of reversibly bound a-ketoaldehydes, inactivation of enzyme activity, and matrix precipitation [32,37,43]; however, in the case of urine, simple dilution [27,28,32], SPE [24,31], and organic solvent addition [25,29] have also been used. The original idea in this work was to employ acetonitrile for matrix precipitation and then for the recovery of derivatized analytes from the acidified mixture by salt-induced phase separation. As presented in Fig. 3, for all four compounds considered in this work, acetonitrile present during

Fig.3. Effect of acetonitrile concentration in the reaction mixture on fluorescence signals obtained for quinoxalines of glyoxal, methylglyoxal, and diacetyl (500 lg/L each) and diethylglyoxal (1000 lg/L). For clarity of presentation, final reaction conditions were used while varying acetonitrile concentrations (see Materials and methods).

derivatization reaction had a negligible effect on fluorescence signals up to its concentration of 20% (v/v). Based on the above considerations and experiments, the final protocol proposed for sample processing prior to chromatographic analyses was simple, was not rigorous, and did not require harsh chemical conditions. In addition to the detailed description given in Materials and methods, it should be stressed that the calibration solutions and urine samples were processed identically, always in the presence of internal standard. After precipitation with acetonitrile, urine was diluted 1:3 to promote derivatization reaction [27] and to lower acetonitrile concentration (Fig. 3). Once the reaction was completed, the sample was acidified with 10 lL of HCl (3 M), another portion of acetonitrile was added, the mixture was saturated with sodium chloride, and after vortex and centrifugation quinoxalines were recovered in the upper acetonitrile layer, whereas polar sample components stayed in the lower aqueous layer (Fig. 4S in the Supplementary material shows chromatograms obtained for urine with and without pretreatment). Even though such salt-induced phase separation cannot be readily controlled, possible imprecision at this stage was compensated by the method of internal standard. Reversed phase liquid chromatography separation of quinoxalines has been reported using gradient elution with mobile phases containing diluted trifluoroacetic acid or slightly acidic buffers (phosphate, formate, actetate; pH <3.5) and methanol or acetonitrile as organic modifier [9,27,32,33,37,38,44,45]. In this work, a core–shell technology column was used (Kinetex C18, 150  3 mm, 2.6 lm) aiming at the enhanced efficiency and minimization of peak broadening with respect to the performance typically achieved on classic analytical columns. The proposed earlier aqueous mobile phases were tested in combination with acetonitrile and varying pH in the range from 2.0 to 7.5; TEA was added for the peak shape improvement. The conditions finally selected were listed in Materials and methods; baseline resolution of four compounds was achieved with the retention times of 5.833 ± 0.003 min for Go, 6.987 ± 0.004 min for MGo, 7.880 ± 0.003 min for DMGo, and 11.120 ± 0.004 min for DEGo, and a total chromatographic run was completed within 12 min. It is worth noting that the separation of four compounds considered in this work had not been undertaken before; however, chromatographic runs from 10 min for one quinoxaline (MGo) [37] up to 45 min for three quinoxalines (Go, MGo, and DMGo as internal standard) [27] were reported previously. Typical chromatograms obtained for calibration solutions are presented in Fig. 4A, and

56

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

the evaluated analytical parameters are summarized in Table 2. These parameters were assessed following International Conference on Harmonization (ICH) procedures [46]. Specifically, detection limit (DL) and quantification limit (QL) were evaluated based on signal-to-noise ratio; the criteria of 6 and 10 standard deviations were adopted for DL and QL, respectively, and the signal obtained for the lowest calibration standard was used. To evaluate calibration DL and QL baseline was acquired from the chromatogram of calibration blank (Fig. 4A), whereas for method DL and QL five times diluted urine from healthy volunteers was used as a blank (Fig. 4B). As can be observed, the calibration DLs of the procedure proposed here are comparable to those recently reported in other studies focused on urine analysis (Table 1 and Refs. [24,31]).

FLD, Ex=344, Em=420

10

A

MGo

8

Go

DEGo

DMGo

6 4 2 100

200

300

400 500 Time, (s)

600

700

B FLD, Ex=344, Em=420

40

2.5

MGo

2.0

30

Go

1.5

DMGo

DEGo

1.0

20

100

200

300

400

500

600

700

800

10 0 100

FLD, Ex=344, Em=420

14

200

C

12 10

300

400 500 Time, (s)

600

700

MGo Go

DEGo

DMGo

8 6

4

4 2 300

400

500

Time, (s)

600

700

Fig.4. Typical chromatograms obtained by the proposed HPLC–FLD procedure. (A) Calibration solutions containing glyoxal, methylglyoxal, and diacetyl at three concentration levels: (——) blank; (. . ...) 75 lg/L; (___) 150 lg/L. Diethylglyoxal (500 lg/L) was used as internal standard. (B) (——) Five times diluted urine from healthy subjects (blank); (___) this same urine without additional dilution: 14.3 ± 0.9 lg/L Go, 11.0 ± 0.7 lg/L MGo, and 16.1 ± 0.9 lg/L DMGo. Diethylglyoxal (500 lg/L) was used as internal standard. (C) (___) Swimmer urine (69.2 ± 2.1 lg/L Go, 56.1 ± 1.8 lg/L MGo, and 42.7 ± 1.1 lg/L DMGo); this same sample after twopoint standard addition: (——) 75 lg/L and (-..-) 150 lg/L of Go, MGo, and DMGo added. Diethylglyoxal (500 lg/L) was used as internal standard.

Urine samples from several volunteers were analyzed, and in Fig. 4C typical chromatograms obtained after two-point standard addition are presented for urine of a youth swimmer (Fig. 5S in the Supplementary material shows urine chromatograms with and without internal standard, demonstrating that DEGo is not present in the sample and so can safely be used as internal standard). The recovery results obtained for this and also for another urine sample (healthy adult) are given in Table 3; as can be observed, the percentage values were in the ranges of 95.5% to 103.0% for Go, 91.9% to 97.8% for MGo, and 94.4% to 99.5% for DMGo, indicating acceptable accuracy. Because glucose is present in urine, it was necessary to check for its possible oxidative degradation to Go, MGo, or DMGo during sample treatment. To this end, the proposed procedure was carried out using aqueous solutions of glucose at concentration levels of 1, 50, and 100 mM (normal urine glucose up to 0.8 mM [47]). The solutions were cleaned up by SPE prior to analysis, as described in the ‘‘Procedures’’ section. None of the three analytes was detected analyzing 1 mM glucose, and only Go was found (below calibration QL) for 50 mM glucose. It is worth mentioning that glucose reagent contained traces of Go and MGo (quantitative results obtained for Go, MGo, and DMGo in glucose solutions with and without SPE treatment are presented in Table 1S of the Supplementary material). To examine the stability of a-ketoaldehydes in urine, an aliquot of fresh urine was divided into five portions; one of them was analyzed immediately, and the other four portions were kept in Eppendorf tubes at 20 °C under light protection. Triplicate analysis was performed for each portion on the next day, after 1 week, after 1 month, and after 3 months, respectively. No statistically significant differences were detected between the results obtained during first month, whereas after 3 months the concentrations of the three compounds decreased markedly (the results obtained in this experiment are presented in Table 2S). Finally, the stability of quinoxaline derivatives was examined in a similar approach but comparing analytical signals obtained immediately after sample preparation with those obtained when the acetonitrile extract was left for 6, 12, and 24 h at room temperature. The results obtained indicated that the processed urine samples should be injected to the HPLC–FLD system no later than 12 h after completing their preparation. In Table 4, quantitative results for three healthy volunteers, three diabetic patients, and three youth swimmers are presented. It is worth noting that for the healthy adults and diabetic patients, the results obtained in this work were within the ranges reported recently (Table 1). No data are available on urine a-ketoaldehydes in association with swimming training; however, the increase of oxidative stress biomarkers was recently reported in juvenile swimmers [48], and it is also known that dicarbonyl species are generated in pool water as a product of disinfection reactions [30,49], so relatively high concentrations of Go, MGo, and DMGo determined in youth athletes call for further investigation. It should be noted that the statistical t test (P < 0.05) revealed significantly higher concentrations of Go in diabetic patients and youth swimmers as compared with control healthy adults. For MGo, the concentrations found in diabetic subjects, athletes, and adults after alcohol ingestion were elevated with respect to control adults; diabetic patients also had significantly higher MGo concentration levels in urine as compared with adults after alcohol ingestion. As already mentioned in the introductory paragraphs, diacetyl is generated during ethanol metabolization, and this compound might be a key metabolite involved in ethanol toxicity and addiction mechanism [17]. In this work, diacetyl was found in urine from one healthy volunteer tested (sample 3), but after alcohol ingestion this compound was present in two volunteers (samples 1a and 3a in Table 4). On the other hand, diacetyl is classified as a precursor of AGEs [17] and, interestingly enough, was found in one diabetic urine sample at a relatively high concentration as compared with

57

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

Parameter

Glyoxal

Methylglyoxal

Diacetyl

Tret ± SD (min) Calibrationa SE for slope (%) SE for intercept (%) R2 Calibration DL (lg/L) Calibration QL (lg/L) Method DL (lg/L) Method QL (lg/L) CVb (%) CVc (%)

5.833 ± 0.003 y = 4.009c + 0.002 3.6 12.5

6.987 ± 0.004 y = 3.570c + 0.001 3.2 10.7

7.880 ± 0.003 y = 3.397c + 0.002 2.9 10.1

clusive biomedical interpretation; however, the results presented in Table 4 demonstrate that the proposed procedure can be recommended as an analytical tool in such studies. Specifically, this procedure is capable of determining three a-ketoaldehydes at physiological levels in urine and also enables the detection of differences between subjects presenting diverse exposure-related or clinical conditions.

0.9999 0.46

0.9999 0.39

0.9997 0.28

Conclusions

0.77

0.65

0.47

2.31

2.01

1.82

3.88

3.35

3.06

1.1 2.7

0.5 1.8

0.6 1.6

Table 2 Analytical figures of merit evaluated for the proposed procedure.

Note: SD, standard deviation; SE, standard error. a Linear regression fit: y, peak areas ratio (analyte/internal standard); c, analyte concentration (lg/L). b Coefficient of variation representing repeatability, evaluated based on six replicates for analyte concentration 500 lg/L. c Coefficient of variation representing intermediate precision, evaluated based on six replicates for analyte concentration 500 lg/L [46].

other samples. The three urine samples from swimmers contained DMGo at similar concentrations as in adults after alcohol ingestion but at lower concentrations than in diabetic subjects. It is clear that the experimental data obtained in this work do not enable any con-

In this work, reversed phase liquid chromatography with fluorimetric detection was applied for the determination of Go, MGo, and DMGo in urine using DEGo as internal standard. It was demonstrated that derivatization with 4MPD can be accomplished at neutral pH and 4 h of incubation at 40 °C, thereby minimizing the risk of changes in analyte concentrations before and during the course of reaction. To avoid the addition of strong acid directly to urine, acetonitrile was used for matrix precipitation, and once quinoxalines were formed acidification enabled the recovery of analyte derivatives in this same solvent by means of the salt-induced phase separation. Baseline resolution of four quinoxalines was achieved within 12 min, which together with the protocol of sample treatment makes the whole procedure relatively fast and simple. It should be stressed, however, that special care was needed to avoid sample contamination and oxidative degradation. The QLs evaluated for three a-ketoaldehydes enabled their determination at physiological levels; the procedure would be suitable for detection of concentration differences among samples/subjects in different clinical or exposure-related studies.

Table 3 Results of recovery experiments. Analyte added (lg/L)

Glyoxal

Methylglyoxal

Diacetyl

Mean ± SD (lg/L)

Recovery (%)

Mean ± SD (lg/L)

Recovery (%)

Mean ± SD (lg/L)

Recovery (%)

Urine 1 (healthy adult) 0 75 150

31.4 ± 2.1 103 ± 6 179 ± 4

– 95.5 98.4

17.3 ± 1.6 86.2 ± 2.1 164 ± 7

– 91.9 97.8

nd 70.8 ± 2.3 145 ± 6

– 94.4 96.7

Urine 7 (youth swimmer) 0 75 150

69.2 ± 2.1 146 ± 4 224 ± 6

– 102 103

56.1 ± 1.8 126 ± 3 198 ± 6

– 93.2 94.6

42.7 ± 1.1 114 ± 4 192 ± 5

– 95.1 99.5

Note: Mean values for triplicate analysis are given with respective standard deviation (SD) values. nd, not determined in this work.

Table 4 Results obtained for glyoxal, methylglyoxal, and diacetyl in different urine samples obtained in triplicate analysis. Urine sample

Healthy adults 1 2 3

Glyoxal (lg/L)

Methylglyoxal (lg/L)

Diacetyl (lg/L)

Individual

Mean

Individual

Mean

Individual

Mean

31.4 ± 2.1 17.0 ± 0.8 43.2 ± 2.1

30.5 ± 13 (a,b)

17.3 ± 1.5 21.2 ± 1.1 27.0 ± 1.2

21.8 ± 5 (c,d,e)

nd nd 13.2 ± 1.6

–

60.0 ± 40

51.1 ± 1.9 32.7 ± 2.1 52.1 ± 0.9

45.3 ± 11 (c,f)

37.3 ± 0.8 nd 42.9 ± 1.2

–

Adults after alcohol ingestion 1a 44.8 ± 2.2 2a 29.2 ± 1.8 3a 106 ± 7 Diabetic subjects 4 5 6

71.2 ± 1.8 87.2 ± 2.3 175 ± 12

111 ± 56(a)

53.8 ± 1.1 241 ± 12 249 ± 14

181 ± 111 (d,f)

nd nd 64.6 ± 3.4

–

Youth swimmers 7 8 9

69.2 ± 2.1 91.3 ± 2.7 84.6 ± 2.5

81.7 ± 11 (b)

56.1 ± 1.8 68.3 ± 1.6 49.9 ± 1.5

58.1 ± 9.4 (e)

42.7 ± 1.1 37.1 ± 0.9 34.2 ± 0.9

38.0 ± 4.3

Note: For each pair of groups, mean values were statistically compared, and the same letter indicates differences found at P < 0.05. nd, not detected in this work.

58

HPLC determination of Go, MGo, and DMGo in urine / A.G. Ojeda et al. / Anal. Biochem. 449 (2014) 52–58

Acknowledgment Financial support from the National Council of Science and Technology, Mexico (CONACYT, project 178553) is gratefully acknowledged. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ab.2013.12.014. References [1] G.S. Qian, R.K. Ross, M.C. Yu, J.M. Yuan, Y.T. Gao, B.E. Henderson, G.E. Wogan, J.D. Groopman, A follow-up study of urinary markers of aflatoxin exposure and liver cancer risk in Shanghai, People’s Republic of China, Cancer Epidemiol. Biomarkers Prev. 3 (1994) 3–10. [2] T. Cvetkovic, P. Vlahovic, D. Dordevic, L. Zvezdanovic, D. Pavlovic, G. Kocic, D. Sokolovic, The significance of urinary markers in the evaluation of diabetic nephropathy, J. Med. Biochem. 27 (2008) 376–382. [3] H.J. Laambers Heerspink, A.H. Brantsma, D. de Zeeuw, S.J.L. Bakker, P.E. de Jong, R.T. Gansevoort, Albuminuria assessed from first-morning-void urine samples versus 24-hour collections as a predictor of cardiovascular morbidity and mortality, Am. J. Epidemiol. 168 (2008) 897–905. [4] N. Shangari, P.J. O’Brien, The cytotoxic mechanism of glyoxal involves oxidative stress, Biochem. Pharmacol. 68 (2004) 1433–1442. [5] P.J. Thornalley, Endogenous a-oxoaldehydes and formation of protein and nucleotide advanced glycation endproducts in tissue damage, Novartis Found. Symp. 285 (2007) 229–246. [6] D. Tan, Y. Wang, C.Y. Lo, S. Sang, C.T. Ho, Methylglyoxal: its presence in beverages and potential scavengers, Ann. N.Y. Acad. Sci. 1126 (2008) 72–75. [7] Y. Wang, C.T. Ho, Flavour chemistry of methylglyoxal and glyoxal, Chem. Soc. Rev. 41 (2012) 4140–4149. [8] N. Rabbani, P.J. Thornalley, Glyoxalase in diabetes, obesity, and related disorders, Sem. Cell Dev. Biol. 22 (2011) 309–317. [9] I. Nemet, L. Varga-Defterdarovic, Z. Turk, Methylglyoxal in food and living organisms, Mol. Nutr. Food Res. 50 (2006) 1105–1117. [10] D.L. Vander Jagt, L.A. Hunsaker, Methylglyoxal metabolism and diabetic complications: roles of aldose reductase, glyoxalase-I, betaine aldehyde dehydrogenase, and 2-oxoaldehyde dehydrogenase, Chem. Biol. Interact. 143–144 (2003) 341–351. [11] Y. Han, E. Randell, S. Vasdev, V. Gill, V. Gadag, L.A. Newhook, M. Grant, D. Hagerty, Plasma methylglyoxal and glyoxal are elevated and related to early membrane alteration in young, complication-free patients with type 1 diabetes, Mol. Cell Biochem. 305 (2007) 123–131. [12] Y.K. Shiryaeva, V.V. Krylin, V.N. Titov, Chromatographic determination of methyl glyoxal in blood plasma as the test for glycotoxicity and accumulation of glycation end-products, Bull. Exp. Biol. Med. 153 (2012) 114–117. [13] Z. Turk, Glycotoxines, carbonyl stress, and relevance to diabetes and its complications, Physiol. Res. 59 (2010) 147–156. [14] K. Nakayama, M. Nakayama, M. Iwabuchi, H. Terawaki, T. Sato, M. Kohno, S. Ito, Plasma a-oxoaldehyde levels in diabetic and nondiabetic chronic kidney disease patients, Am. J. Nephrol. 28 (2008) 871–878. [15] M.G. Distler, A.A. Palmer, Role of glyoxalase 1 (Glo1) and methylglyoxal (MG) in behavior: recent advances and mechanistic insights, Front. Genet. 3 (2013) 250–255. [16] P.J. Beisswenger, K.S. Drummond, R.G. Nelson, S.K. Howell, B.S. Szwergold, M. Mauer, Susceptibility to diabetic nephropathy is related to dicarbonyl and oxidative stress, Diabetes 54 (2005) 3274–3281. [17] P. Kovacic, A.L. Cooksy, Role of diacetyl metabolite in alcohol toxicity and addiction via electron transfer and oxidative stress, Arch. Toxicol. 79 (2005) 123–128. [18] M. Otsuka, T. Mine, K. Ohuchi, S. Ohmori, A detoxication route for acetaldehyde: metabolism of diacetyl, acetoin, and 2,3-butanediol in liver homogenate and perfused liver of rats, J. Biochem. 119 (1996) 246–251. [19] M. Otsuka, N. Harada, T. Itabashi, S. Ohmori, Blood and urinary levels of ethanol, acetaldehyde, and C4 compounds such as diacetyl, acetoin, and 2,3butanediol in normal male students after ethanol ingestion, Alcohol 17 (1999) 119–124. [20] M. Otsuka, S. Ohmori, Simple and sensitive determination of diacetyl and acetoin in biological samples and alcohol beverages, J. Chromatogr. 577 (1992) 215–220. [21] M.W. Rigler, W.E. Longo, Emission of diacetyl (2,3 butanedione) from natural butter, microwave popcorn butter flavor powder, paste, and liquid products, Int. J. Occup. Med. Environ. Health 16 (2010) 291–302. [22] T. Wainwright, Diacetyl—A review: Part I—Analytical and biochemical considerations; Part II—Brewing experience, J. Inst. Brewing 79 (1973) 451– 470. [23] A. Amoroso, G. Maga, M. Daglia, Cytotoxicity of a-dicarbonyl compounds submitted to in vitro simulated digestion process, Food Chem. 140 (2013) 654–659.

[24] J. Zhang, H. Zhang, M. Li, D. Zhang, Q. Chu, J. Ye, A novel capillary electrophoretic method for determining methylglyoxal and glyoxal in urine and water samples, J. Chromatogr. A 1217 (2010) 5124–5129. [25] L.A. Zardari, M.Y. Khuhawar, A.J. Laghari, Capillary GC analysis of glyoxal and methylglyoxal in the serum and urine of diabetic patients after use of 2,3diamino-2,3-dimethylbutane as derivatizing reagent, Chromatographia 70 (2009) 891–897. [26] A. Espinosa-Mansilla, I. Duran-Meras, F. Canada Canada, M. Prado Marquez, High-performance liquid chromatographic determination of glyoxal and methylglyoxal in urine by prederivatization to lumazinic rings using in serial fast scan fluorimetric and diode array detectors, Anal. Biochem. 371 (2007) 82–91. [27] K. Akira, Y. Matsumoto, T. Hashimoto, Determination of urinary glyoxal and methylglyoxal by high-performance liquid chromatography, Clin. Chem. Lab. Med. 42 (2004) 147–153. [28] A. Espinosa-Mansilla, I. Duran-Meras, F. Salinas, High-performance liquid chromatographic–fluorometric determination of glyoxal, methylglyoxal, and diacetyl in urine by prederivatization to pteridinic rings, Anal. Biochem. 255 (1998) 263–273. [29] M.Y. Khuhawar, L.A. Zardari, A.J. Laghari, Capillary gas chromatographic determination of methylglyoxal from serum of diabetic patients by precolumn derivatization with 1,2-diamonopropane, J. Chromatogr. B 873 (2008) 15–19. [30] N.R. Neng, C.A.A. Cordeiro, A.P. Freire, J.M.F. Nogueira, Determination of glyoxal and methylglyoxal in environmental and biological matrices by stir bar sorptive extraction with in-situ derivatization, J. Chromatogr. A 1169 (2007) 47–52. [31] M.C. Hurtado-Sanchez, A. Espinosa-Mansilla, M.I. Rodrıguez-Caceres, E. Martın-Tornero, I. Duran-Meras, Development of a method for the determination of advanced glycation end products precursors by liquid chromatography and its application in human urine samples, J. Sep. Sci. 35 (2012) 2575–2584. [32] A.C. McLellan, S.A. Phillips, P.J. Thornalley, The assay of methylglyoxal in biological systems by derivatization with 1,2-diamino-4,5-dimethoxybenzene, Anal. Biochem. 206 (1992) 17–23. [33] F.W. Chaplen, W.E. Fahl, D.C. Cameron, Method for determination of free intracellular and extracellular methylglyoxal in animal cells grown in culture, Anal. Biochem. 238 (1996) 171–178. [34] M.A. Mirza, A.J. Kandhro, S.Q. Memon, M.Y. Khuhawar, R. Arain, Determination of glyoxal and methylglyoxal in the serum of diabetic patients by MEKC using stilbenediamine as derivatizing reagent, Electrophoresis 28 (2007) 3940–3947. [35] T. Shibamoto, Analytical methods for trace levels of reactive carbonyl compounds formed in lipid peroxidation systems, J. Pharm. Biomed. Anal. 41 (2006) 12–25. [36] S. Hara, M. Yamaguchi, Y. Takemori, T. Yoshitake, M. Nakamura, 1,2-Diamino4,5-methylenedioxybenzene as a highly sensitive fluorogenic reagent for adicarbonyl compounds, Anal. Chim. Acta 215 (1988) 267–276. [37] I. Nemet, L. Varga-Defterdarovic, Z. Turk, Preparation and quantification of methylglyoxal in human plasma using reverse-phase high-performance liquid chromatography, Clin. Biochem. 37 (2004) 875–881. [38] M. Yamaguchi, J. Ishida, Z. Xuan-Xuan, M. Nakamura, T. Yoshitake, Determination of glyoxal, methylglyoxal, diacethyl and 2,3-pentadedione in fermented foods by high-performance liquid chromatography with fluorescence detection, J. Liq. Chromatogr. 17 (1994) 203–211. [39] A.J. Maroulis, A.N. Voulgaropoulus, C.P. Hadjiantoniou-Maroulis, Fluorimetric determination of biacetyl, Talanta 32 (1985) 504–506. [40] H.E. Krizner, D.O. De Haan, J. Kua, Thermodynamics and kinetics of methylglyoxal dimer formation: a computational study, J. Phys. Chem. 113 (2009) 6994–7001. [41] S. Houdier, M. Barret, F. Domine, T. Charbouillot, L. Deguillaume, D. Voisin, Sensitive determination of glyoxal, methylglyoxal, and hydroxyacetaldehyde in environmental water samples by using dansylacetamidooxyamine derivatization and liquid chromatography/fluorescence, Anal. Chim. Acta 704 (2011) 162–173. [42] H. Odani, T. Shinzato, Y. Matsumoto, J. Usami, K. Maeda, Increase in three a, bdicarbonyl compound levels in human uremic plasma: specific in vivo determination of intermediates in advanced Maillard reaction, Biochem. Biophys. Res. Commun. 256 (1999) 89–93. [43] A. Dhar, K. Desai, J. Liu, L. Wu, Methylglyoxal, protein binding, and biological samples: are we getting the true measure?, J Chromatogr. B 877 (2009) 1093– 1100. [44] E.W. Randell, S. Vasdev, V. Gill, Measurement of methylglyoxal in rat tissues by electrospray ionization mass spectrometry and liquid chromatography, J. Pharmacol. Toxicol. Methods 51 (2005) 153–157. [45] J. Degen, M. Hellwig, T. Henle, 1,2-Dicarbonyl compounds in commonly consumed foods, J. Agric. Food Chem. 60 (2012) 7071–7079. [46] ICH Harmonized Tripartite Guideline, Validation of analytical procedures: Text and methodology (Q2/R1), International Conference on Harmonization, 2012, . [47] C.A. Burtis, E.R. Ashwood, Tietz Textbook of Clinical Chemistry, W. B. Saunders, Philadelphia, 1999. [48] S.S. Koparan, Increased oxidative stress inflammatory response in juvenile swimmers after a 16-week swimming training, Eur. J. Exp. Biol. 3 (2013) 211–217. [49] C. Zwiener, T. Glauner, F.H. Frimmel, Method optimization for the determination of carbonyl compounds in disinfected water by DNPH derivatization and LC–ESI–MS–MS, Anal. Bioanal. Chem. 372 (2002) 615–621.