Fluoroalkyl chloroformates in treating amino acids for gas chromatographic analysis

Fluoroalkyl chloroformates in treating amino acids for gas chromatographic analysis

Available online at www.sciencedirect.com Journal of Chromatography A, 1186 (2008) 391–400 Fluoroalkyl chloroformates in treating amino acids for ga...

471KB Sizes 34 Downloads 120 Views

Available online at www.sciencedirect.com

Journal of Chromatography A, 1186 (2008) 391–400

Fluoroalkyl chloroformates in treating amino acids for gas chromatographic analysis ˇ Petr Huˇsek ∗ , Petr Simek, Petr Hartvich, Helena Zahradn´ıcˇ kov´a Biology Centre, Laboratory of Analytical Biochemistry, Academy of Sciences of the Czech Republic, Braniˇsovsk´a 31, ˇ e Budˇejovice, Czech Republic CZ-370 05 Cesk´ Available online 15 December 2007

Abstract Novel fluoroalkyl chloroformates with three and four carbon atoms were investigated for the immediate conversion of amino acids into hydrophobic derivatives in water-containing media. Derivatization conditions were extensively studied and optimized sample preparation protocols elaborated. More than 30 amino acids were treated with the particular reagent in isooctane by simply vortexing the reactive organic phase with a slightly basified aqueous medium containing pyridine or 3-picoline as a catalyst. Outstanding separation of nearly all components on 5% phenylmethylsilicone phase in gas chromatographic (GC) analysis with mass spectrometric (MS) or flame ionization detection (FID) required <10 min. Quantitation characteristics involving linearity in the range of 0.1–100 nmol, regression coefficients of 0.999–0.953 (histidine), MS limit of detection (LOD) reaching 0.03 pmol at proline to nearly 20 pmol at glutamic acid, plus electron impact (EI) spectra and diagnostic SIM fragment ions of the derivatives are reported. The novel method is simple, robust and rapid, enabling to treat amino acids in aqueous environment and to analyze them in <15 min. © 2007 Elsevier B.V. All rights reserved. Keywords: Amino acids; Derivatization; Pentafluoropropyl- and heptafluorobutyl chloroformates; Gas chromatography

1. Introduction Amino acids exhibit structural diversity and high polarity for which their analysis in aqueous matrices has been a respectable testing field for any novel analytical method. Starting from the pioneering method of using ion-exchange separation and ninhydrin detection in the 1950s [1], numerous procedures were developed for alternative determination of the derivatized amino acid molecules by GC, high-performance liquid chromatography (LC) and later capillary electrophoresis (CE) also. Most of the methods were reviewed recently [2,3]. In GC, derivatization with chloroformates in aqueous media within seconds became highly popular not only for amino acid analysis [4–6]. In LC, a lot of reagents were tested and more or less successfully applied for a precolumn partial modification of amino acid molecule. For example, 6-aminoquinolyl-Nhydroxysuccinimidyl carbamate proved to be especially useful in LC–UV analysis of 27 amino acids separated in 8 min, mainly



Corresponding author. Tel.: +420 387775287; fax: +420 387775287. E-mail address: [email protected] (P. Huˇsek). URL: http://www.bclab.eu (P. Huˇsek).

0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.11.117

due to the progress in preparing small particle size sorbents and optimization of the gradient elution [7]. Native amino acids, along with other physiological compounds in complex biological matrices, are increasingly subjected to direct analysis by means of LC–MS/MS techniques [8–10]. Utilizing perfluorinated acid modifiers in mobile phases, LC separation of amino acids was substantially improved, while the ion current signals of the analytes of interest were augmented. Nevertheless, off-line precolumn derivatizations still remain attractive in LC–MS/MS for some reasons; for example, carboxylic groups of amino acids are often butylated [5,11] and amino groups alkylated, mostly with aryl reagents [12]. Electrochemical detection of native amino acids presents an alternative recently, reviewed [13] that using evaporative light scattering is rather limited to several compounds [14]. The application of CE suffers from reduced injection volumes, so that very sensitive detectors are required. Native amino acids can be determined by electrospray MS/MS [15], by conductivity detection [16], and the semi-derivatized ones using laser-induced fluorescence (LIF) detection [17,18]. A rapid chiral analysis of derivatized amino acids was recently reported using micellar electrokinetic chromatography on a plastic microchip [19]. Employment of matrix-assisted laser desorption-ionization (MALDI) MS did not require any deriva-

392

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

tization or separation; data acquisition was accomplished in several seconds [20]. The method was found reproducible for quantitative amino acid analysis but problems encountered with insufficient linearity, e.g. for plasma amino acids. This paper points to a further utilization of the chloroformate methodology toward amino acid analysis. As fluoroalkyl anhydrides of low carboxylic acids conferred molecules sufficient volatility and favorable features for GC analysis in the past [2], the fluoroalkyl chloroformates should do the same; however, with the substantial advantage of not requiring anhydrous conditions. Trifluoroethyl and pentafluorobenzyl chloroformates were already used for chiral analysis of amino acid enantiomers [21,22] or plasma amino acid profiling [23,24]. The latter method was only partially successful, mainly due to low alkylation of basic amino acids with the reagent. Italian researchers [25] put most effort into synthesis of fluoroalkyl chloroformates with bulky alkyls, such as octafluoropentyl and tridecafluorooctyl. This new family of reagents enabled a straightforward derivatization of highly polar low-molecular compounds, and succeeded in GC–MS identification of disinfection by-products in drinking water. The reagents were neither designed nor applied for amino acid analysis, apparently due to the heavy alkyls. Such achievements prompted us to examine the anticipated advantageous features of two novel reagents, the pentafluoropropyl chloroformate (PFPCF) and heptafluorobutyl chloroformate (PFPCF, HFBCF). AS they are not commercially available, their synthesis had to be ordered. Pilot studies revealed a high reactivity toward amino acids in aqueous media, even without alcohol addition, while creating highly volatile derivatives with molecular weights <850 Da. The analytes showed favorable chromatographic properties as allowing e.g. direct GC determination of such difficult amino acids as homocystine and N-methylhistidines, the latter being GC-recorded using chloroformates for the first time. Optimization of the reaction conditions and validation of the final procedures are presented in this study. 2. Experimental 2.1. Chemicals Pyridine, 3-methylpyridine (3-picoline), 2,2,4-trimethylpentane (isooctane), was delivered from Sigma–Aldrich (Prague, Czech Republic). Fluorinated alcohols, such as, 2,2,3,3,3-pentafluoro-1-propanol (PFPOH), 2,2,3,3,4,4,4heptafluoro-1-butanol (HFBOH) and benzotrifluoride were obtained from Fluorochem (Glossop, UK), all of analyticalreagent grade. HFBCF and PFPCF were synthesized in the ˇ e Budˇejovice, Czech Republic) and are Biology Centre (Cesk´ available on request.

(3MH, 1MH), were individually weighed and divided into two groups. One mix was prepared in 0.1 mol/l sodium carbonate (Ala, Gly, Val, Leu, Ile, Pro, Asp, Glu, Phe, Lys), the other with remaining protein amino acids in 0.1 mol/l HCl. Each stock solution contained amino acids at an equimolar concentration of 20 mmol/l, neutral pH was obtained upon mixing, and working solutions were prepared by mere dilution. Further, additional amino acids listed in Table 5 were admixed occasionally to document the efficacy of the procedure and the separation ability of the capillary column. Norvaline (Nval) at an amount of 10 nmol was employed as an internal standard (IS), hexadecane (C16) was additionally admixed as another standard in the case of FID analysis. 2.3. Instrumentation A GC–MS quadrupole mass spectrometer DSQ (Thermo Electron, San Jos´e, CA, USA), equipped with EI ionization, was employed for analyses of derivatized samples using a programmable temperature vaporizing (PTV) injector (Thermo Electron). A 0.5–1 ␮l-sample aliquot was injected in splitless mode (closed for 30 s) into a glass multiple baffle liner (120 mm × 2 mm i.d.). FactorFour VF-5 ms fused-silica capillary column 30 m in length (0.25 mm i.d./0.25 ␮m df ) was purchased from Varian (Lake Forest, CA, USA) and cut into two pieces; a length of 20 m was employed for GC–MS analysis, and 10 m for GC-FID. Helium carrier gas flow rate was 1.4 ml/min, injector temperature was 220 ◦ C. The ion source temperature was held at 210 ◦ C and the transfer line at 250 ◦ C. The oven was programmed from 60 ◦ C to 280 ◦ C, at a linear increase of 20 ◦ C/min. Detection employed EI mode (70 eV) in a full scan regime (58–850 Da), SIM mode was used for quantitation and characteristic fragments ions monitored for each compound. A GC-2014 chromatograph equipped with AOC-20i autoinjector, both from Shimadzu (Kyoto, Japan), was employed for GC-FID analyses. Derivatized amino acids were separated on low polarity of 5% phenylmethylsilicone phases, using either the aforementioned VF-5 column, 10 m in length, or a 30 m long thin-coated (0.25 mm i.d./0.10 ␮m) Zebron ZB-5 ms column (Phenomenex, Torrance, CA, USA). The VF-5 column served to separate the PFPCF derivatives over the range of 90–250 ◦ C at a linear raise of 20 ◦ C/min and constant velocity of 42.8 cm/s (1.26 ml/min). The ZB-5 column was run under the same conditions for PFPCF derivatives, but between 100 ◦ C and 250 ◦ C and a rise of 15 ◦ C/min for the separation of the HFBCF ones. Helium at 50 kPa or 167 kPa, was used as carrier gas; injection port and FID temperatures were kept at 220 ◦ C and 300 ◦ C, respectively. Injection of 1.5–2 ␮l was carried out in a split mode of 1:9. 2.4. Derivatization tools

2.2. Standards Standards of 19 protein amino acids except Arg (3-code abbreviations used throughout the text, see Table 5), plus three more, i.e. ornithine (Orn), 3-methyl- and 1-methylhistidine

Reactions were carried out in 6 × 50 mm glass culture tubes (Kimble/Kontes, Vineland, NJ, USA) or in 1.1-ml tapered polypropylene tubes (Continental Laboratory Products, San Diego, CA, USA) without any closures. Adjustable 50 and 100-

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

␮l Brand Transferpettor pipettes with glass capillaries were supplied by Merck (Prague, Czech Republic), and engaged for manipulation with the reactive reagents and their mixtures in the organic solvent. A vortex mixer (50–2400 rpm) was supplied by P-Lab (Prague, Czech Republic). The reactive organic medium for the phase-transfer derivatization was a mixture of isooctane with HFBCF (5:3, v/v), or, for the PFPCF reagent, isooctane/PFPOH/PFPCF, 7:1:4 (alternatively, isooctane/PFPCF, 2:1).

2.6. Warnings

2.4.1. One-step derivatization procedure It employs 3-picoline as a catalyst for the reaction medium. Amino acid standards in aqueous medium (1–10 ␮l) are covered with 200 ␮l of 0.05 mol/l aqueous sodium hydroxide containing 5% (v/v) picoline, and 50 ␮l of the reactive organic medium is added. The content is vortexed for about 5 s, during which the cloudy emulsion and milky organic medium turn clear. The reaction course is terminated by admixing 70–100 ␮l of isooctane containing 2% picoline and by additional vortexing for about 3–5 s. An aliquot of the upper phase is then aspirated for GC analysis. The narrow glass culture tubes served favorably in this respect.

3. Results and discussion

2.4.2. Two-step derivatization procedure This approach postpones the addition of the catalyst to a second stage; the broader reaction PP-tubes are here preferred. First, to 150 ␮l of 0.05 mol/l aqueous sodium carbonate, 50 ␮l of the particular reactive organic medium is admixed and the two-phase system is vigorously shaken, rather intermittently, to create an emulsion; such an effect may occur within 10–60 s. After about 1 min the second medium, i.e. 50 ␮l of 0.1 mol/l sodium carbonate with admixed pyridine (5:1, v/v) is added, followed by vortexing for about 5–7 s to clear the milky organic phase. After the addition of 70–100 ␮l of isooctane and mixing for about 3 s, an aliquot of the organic phase is taken for the analysis. 2.4.3. Re-derivatization This step was carried out to check the efficacy of the reaction. The aqueous phase, remaining after derivatization, was further treated as follows: the organic phase with the derivatives is fully removed, 100 ␮l isooctane added, briefly vortexed and aspirated. Then, 10 ␮l of 1 mol/l sodium hydroxide and 10 ␮l of picoline are admixed and the content is gently mixed. Following the addition of 50 ␮l of the particular reactive organic medium, the sample is further processed as described under the one-step approach. 2.5. External standard To determine relative molar responses (RMR) of the derivatives in the FID system, hexadecane (C16) dissolved in isooctane (2 mmol/l) was added as external standard in half molar amount of that of the amino acids (e.g. 5–10 nmol of amino acids). Such a molar ratio is recorded on the enclosed chromatograms, peak heights were chosen as the calculation base.

393

Two precautions upon working with chloroformates in general should be taken into account. Firstly, to prevent corrosion of the syringe plunger, continuous rinsing of the syringe with a weak alcohol such as propan-2-ol after each injection is highly recommended. Secondly, the manipulation of highly reactive and toxic reagents should be done exclusively in a fume-hood, gloves are recommended.

3.1. Optimization of reaction conditions 3.1.1. Pyridine vs. picoline catalysis in the one-step procedure The nature of the catalyst appeared to influence reaction yields in the one-step procedure significantly. Picoline, as a more hydrophobic base than pyridine, aided in improving reaction yields because of a smooth dispersion in both the phases, creating thus a cloudy emulsion over the whole vial content. The more the catalyst was diluted in the aqueous medium, the slower the reaction proceeded, and a better conversion rate was achieved. Finally, 5% of picoline (or, alternatively, 4% of pyridine) in 200 ␮l of the basic medium was established as optimum. As a result, with picoline replacing pyridine the yields of aliphatic amino acids and Pro were 15–20% higher, those of basic amino acids and Asp 30–40% higher, and the Glu-diester was about three times higher. In contrast to pyridine, picoline is transferred partially into the organic phase and the yield of Tyr has been slightly lowered due to this (see Tables 1 and 2). Shaking out the base from the organic phase by adding 50 ␮l of 1 mol/l aqueous HCl helps in Tyr recovery. A certain drawback is a lower purity of picoline, resulting in the appearance of some extraneous peaks at the rear of the chromatographic record. The interferences could not be distilled off, and no cleaner product was available on the market. 3.1.2. Alcohol presence Unlike reactions carried out with alkyl chloroformates in aqueous media, where the presence of the corresponding alcohol was a prerequisite of esterification [2,4–6], fluoroalkyl chloroformates facilitate esterification directly. Furthermore, the yields of aliphatic amino acids, especially that of Ala, were highest without any alcohol added. Not only in the case of HFBCF but also with PFPCF, where a 1:2 (v/v) mixture with isooctane states an alternative. Alcohol does not need to be added externally as immediate decomposition of the reagent upon contact with the organic base is the source of alcohol itself. Esterification of the carboxylic groups proceeds then via alcoholysis, as experienced earlier [26]. With PFPCF, however, a small amount of PFPOH in the organic medium promoted ester formation of Glu and increased yields of His/MHs. In contrast, a higher percentage of alcohol (>10%, v/v) led to lowered yields of Ala, other aliphatic amino acids, Lys and Orn. It follows the reaction of the dibasic amino acids on alcohol is opposite to that of His/MHs. Alcohol presence inhibits also alkylation of the aliphatic hydroxyl, so

394

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

Table 1 Average values (n = 5) of FID relative molar responses (RMR, relative to hexadecane) of HFBCF-treated amino acids and precision of derivatization at levels 1 and 20 nmol of each amino acid in equimolar mixture expressed by RSD% Amino acid

One-step procedure 1 nmol

Ala Gly Val Leu Ile Pro Thr Ser Asn Asp SerOR Glu Met Phe Gln 3MH Orn 1MH His Lys Tyr C–C Trp

Two-step procedure 20 nmol

1 nmol

20 nmol

RMR

RSD%

RMR

RSD%

RMR

RSD%

RMR

RSD%

32.2 27.0 42.8 43.7 42.1 37.6 19.0 12.3 18.9 20.4 20.3 9.3 40.2 60.7 11.2 24.1 16.4 23.6 32.9 38.5 56.6 58.7 62.3

2.2 2.9 2.4 1.4 1.4 1.6 2.3 2.1 1.5 5.2 2.8 5.2 1.2 1.9 2.8 2.9 2.6 2.0 2.5 2.1 2.4 1.8 1.2

32.6 27.1 43.0 43.6 42.3 36.7 19.1 12.2 19.0 20.2 20.2 9.5 40.3 60.2 11.2 23.7 15.0 23.4 31.4 38.0 55.8 59.4 63.3

2.4 2.6 1.9 1.7 1.5 2.7 2.9 1.8 3.0 4.8 3.2 4.8 1.2 2.2 3.2 4.1 2.9 2.3 2.8 2.2 2.1 1.5 1.4

34.4 29.1 42.1 44.3 43.5 45.9 21.8 12.9 16.2 20.0 17.3 21.8 39.2 60.7 10.8 23.1 33.9 22.5 26.3 44.6 62.8 57.1 62.5

1.6 2.5 2.6 2.0 1.8 1.3 1.8 2.5 2.2 2.9 3.5 5.8 1.5 1.1 2.9 3.1 1.8 2.3 3.0 1.3 1.9 2.5 2.1

34.1 29.3 43.5 44.2 43.3 45.3 21.7 12.7 16.2 20.1 17.7 21.5 39.4 60.6 10.4 23.4 34.9 22.9 26.4 45.2 62.0 57.6 62.5

2.0 3.0 3.2 2.1 2.5 1.9 2.7 3.6 2.4 2.5 3.5 5.7 2.1 1.6 2.3 3.1 3.1 2.0 3.3 2.4 1.6 2.6 1.8

Table 2 Average values (n = 5) of FID relative molar responses (RMR, relative to hexadecane) of PFPCF-treated amino acids and precision of derivatization at levels 1 and 20 nmol of each amino acid in equimolar mixture expressed by RSD% Amino acid

One-step procedure 1 nmol

Ala Gly Val Leu Ile Pro Thr Ser Asp Asn Glu Met Phe Gln Orn 3MH His 1MH Lys Tyr C-C Trp

Two-step procedure 20 nmol

1 nmol

20 nmol

RMR

RSD%

RMR

RSD%

RMR

RSD%

RMR

RSD%

28.8 22.3 39.6 40.1 39.7 34.0 19.5 13.5 14.8 19.3 5.6 33.5 52.3 10.2 18.4 20.2 31.5 19.1 25.7 51.4 40.5 55.6

3.1 4.2 2.6 4.5 1.5 2.4 3.1 5.3 4.7 4.2 8.8 4.9 3.2 4.3 4.4 5.8 9.7 3.7 4.3 3.9 4.7 4.2

28.5 22.1 39.5 41.7 40.2 34.8 19.6 13.7 15.1 19.5 5.1 33.9 52.8 10.2 18.3 20.8 30.4 19.2 26.7 50.9 40.3 56.1

2.6 3.4 3.0 4.1 3.1 2.6 6.1 3.5 5.1 3.0 10.1 2.1 2.3 5.5 4.1 3.5 7.5 3.4 5.2 3.0 4.6 4.5

27.2 23.1 39.8 41.6 39.6 39.3 20.0 12.5 18.6 18.9 12.2 35.4 52.6 10.1 28.8 22.3 21.4 15.9 34.0 54.8 39.5 56.8

3.5 3.0 1.4 1.5 2.3 1.0 3.9 3.6 3.4 4.4 5.2 2.0 3.0 4.2 2.5 2.2 4.6 1.8 2.3 1.7 3.8 3.0

26.8 22.5 39.5 41.9 41.0 39.2 20.1 12.4 18.8 18.7 12.4 36.3 52.1 10.3 28.2 23.2 21.6 15.8 33.9 54.5 39.3 57.4

2.4 3.4 1.9 1.8 0.8 0.8 2.5 3.4 2.6 2.7 3.8 2.0 1.9 4.1 1.7 3.1 3.1 3.1 2.5 2.4 2.3 3.0

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

395

that a shift to Ser/Thr, i.e. products with free OH-group, is augmented. Vice versa, the O-alkylated products (SerOR/ThrOR) are elevated in the absence of alcohol. The ratio of Thr to ThrOR is clearly shifted to the former with both reagents; the same is valid for the PFPCF-derivative of Ser. With HFBCF, however, the dominant product is SerOR, being thus more suitable for quantitation than Ser. It points also to a higher reactivity of HFBCF in comparison with PFPCF.

half due to the internal cyclization to 3-amino-piperidin-2-one, which is post-alkylated at the amino and imino groups. This side product is eluted before the parent Orn (the larger peak between 19 and 20 in Fig. 1, that behind 17 on Fig. 2). For the particular application, both the approaches should be tested and the one more suitable to a certain task should be applied. It is evident from the tabulated values that the one-step mode suits better to His, the two-step one to Glu, Orn and Lys.

3.1.3. One-step vs. the two-step procedure Two approaches with different reaction conditions afforded nearly identical yields. Regarding the one-step method, optimum derivatization yields were achieved with picoline in mere water. In diluted hydroxide, however, the yields of Glu and His/MHs were augmented, whereas diluted carbonate caused an opposite effect. Unlike that, aqueous carbonate proved to be the best medium in the two-step approach, in which the terminal amino groups of Lys and Orn were successively alkylated. On the other hand, a combination of pyridine with aqueous hydroxide proved to be detrimental for most amino acids, except again for His/MHs, the yield of which increased. The different yields of His in the one-step and two-step approach are evident from Tables 1 and 2. Carbonate remains the medium of choice for Scontaining amino acids, Cys and Hcy, which are always alkylated on the thiol group. In the hydroxide presence Hcy turns partially to thiolacton, being eluted behind the parent product. The same occurs with Orn, the response of which is reduced nearly by

3.1.4. Derivatization yield and reaction complexity Responses of the derivatives in FID were related to hexadecane, the response of which was set to 100, and RMR values are given in Tables 1 and 2. Recorded amounts in the Figs. 1 and 2 are about 10 pmol out of 10 nmol of each amino acid treated. Considering the calculation according to carbon number, and accepting the exclusion of the responses of the non-effective carbonyl and oxygen adjacent methylene group, the fluoroalkyls should contribute to the response to a certain extent [27]. If not, responses of HFBCF and PFBCF analytes should be equal, which is evidently not the case. Vice versa, should the fluoroalkyls respond equally like the parent alkyls, RMR would be higher. To assess a possible non-completeness of derivative formation, we have carried out the re-derivatization step as described under Experimental. Indeed, the underivatized fraction came to about 10–15% at aliphatic amino acids, Pro, and cystine, but 40–50% at Asp and Glu. Glu is preferably cyclized to pyroglutamic acid, elution of which from the column succeeded more or less (retention

Fig. 1. GC-FID chromatogram (30-m ZB-5 column) of amino acid standards (mix 1) treated with HFBCF via the one-step procedure (above), and the two-step procedure (below). Peaks: 1 = Ala, 2 = Gly, 3 = Aba, 4 = Val, 5 = Baiba, 6 = Leu, 7 = Ile, 8 = Pro, 9 = Thr, 10 = Ser, 11 = Asn, 12 = Asp, 13 = Tpro, 14 = SerOR, 15 = Glu, 16 = Met, 17 = Cys, 18 = C16, 19 = Phe 20 = Hcy, 21 = Gln, 22 = Orn, 23 = 3MH, 24 = His, 25 = 1MH, 26 = Lys, 27 = Tyr, 28 = C–C + Trp.

396

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

Fig. 2. GC-FID chromatogram (30-m ZB-5 column) of amino acid standards (mix 1) treated with PFPCF via the one-step procedure. Peaks: 1 = Ala, 2 = Gly, 3 = Aba, 4 = Val, 5 = Baiba, 6 = Leu, 7 = Ile, 8 = Pro, 9 = Thr, 10 = Ser, 11 = Asp, 12 = Asn, 13 = Tpro, 14 = Glu, 15 = Met, 16 = Phe, 17 = C16, 18 = Gln, 19 = Orn, 20 = 3MH, 21 = His, 22 = 1MH, 23 = Lys, 24 = Tyr, 25 = C–C, 26 = Trp.

close to that of Ser). The unreacted portion at other amino acids was <10%, no substantial difference between two procedures was noticed. The response of Glu was lower in the one-step approach, probably due to the fast reaction course and rapid decomposition of the reagent. The reaction of fluoroalkyl chloroformates with amino acids is a very complex phenomenon and to achieve a complete derivatization of each amino acid in the mixture is impossible. There are several reaction courses running simultaneously, i.e. the ester formation via alcoholysis of the mixed anhydride, the alkylation of thiol and phenolic hydroxyl groups, that of the terminal amino groups of Lys and Orn and of imidazolyl group of His, the side-product formation at Glu and Orn due to pre-

ferred cyclization. Furthermore, the rate of reagent degradation upon contact with the organic catalyst further complicates the reaction conditions. As a result, the composition of the reactive organic phase, that of the catalyst in the basified medium and the chosen volumes of both phases could be set only experimentally, and they resulted in two complementary derivatization protocols. To summarize, by changing the reaction conditions regarding the concentration of inorganic and organic bases, it is possible to promote yields of some targeted amino acids, but it is always at the cost of others. To get a maximum conversion of each amino acid to the desired product in the whole mixture is not possible. However, the unreacted portion is predominantly connected with

Table 3 Repeatibility of GC–MS/SIM determination of amino acids derivatized with HFBCF and PFPCF Amino acid

Ala Gly Val Leu Ile Pro Thr Ser Asn Asp SerOR Glu Met Phe Gln Orn 3MH His Lys Tyr CC Trp

HFBC reagent (amount in nanomoles)

PFPC reagent (amount in nanomoles)

One-step procedure

Two-step procedure

One-step procedure

Two-step procedure

0.5

5

50

0.5

5

50

0.5

5

50

0.5

5

50

6.0 2.3 4.6 4.2 6.7 4.3 6.6 15.6 6.4 8.6 8.9 18.2 8.2 4.5 11.3 8.4 12.9 6.5 7.0 5.8 8.3 8.3

4.7 2.1 4.2 5.7 4.5 1.2 4.3 10.7 6.7 5.6 6.4 10.8 5.7 3.7 7.5 7.3 6.4 5.0 6.4 5.0 7.5 5.6

3.9 6.5 3.9 2.5 4.4 3.6 3.7 20.5 10.9 7.4 5.1 7.4 3.1 6.4 4.0 5.9 5.9 5.0 4.0 4.7 6.8 4.9

4.4 4.0 4.7 3.5 6.0 3.7 5.8 13.6 8.6 7.6 8.4 11.3 6.7 3.3 9.3 7.9 10.1 7.9 6.0 5.0 8.0 8.4

1.8 4.4 3.1 3.5 5.5 3.2 5.7 9.2 7.1 6.8 5.2 9.1 3.5 3.4 7.0 8.1 7.9 8.1 4.9 4.5 7.4 4.2

1.9 2.0 2.3 1.3 2.4 2.1 3.9 7.0 3.7 2.2 3.0 3.5 1.7 2.8 6.6 4.5 2.9 4.5 3.6 3.5 4.4 3.2

4.1 6.8 6.9 5.3 5.7 4.0 13.1 12.4 8.4 8.3 17.0 10.1 11.2 6.2

1.7 4.0 3.2 2.5 4.6 3.4 8.0 8.1 4.6 6.5 6.1 8.7 3.5 3.0 12.0 8.8 5.7 7.9 6.7 3.4 4.2 3.9

1.9 3.7 1.0 2.6 2.2 2.0 3.7 1.8 3.3 8.8 1.2 15.5 2.7 3.3 4.4 7.1 1.8 2.5 5.2 4.6 6.4 2.7

1.8 4.3 3.4 1.1 2.4 5.1 7.3 10.6 8.5 6.4 7.0 7.6 8.2 2.5 11.3 5.8 10.6 9.5 5.7 4.5 7.7 7.3

4.4 4.1 4.6 3.4 2.6 3.8 7.5 9.4 6.7 5.9 6.2 9.9 7.5 2.5 6.2 5.6 7.4 9.1 3.3 3.8 7.0 7.1

4.0 2.9 3.4 1.5 2.1 2.0 5.2 7.2 4.6 3.8 5.6 6.9 3.3 2.7 4.4 4.2 5.3 9.9 2.8 3.8 7.2 5.2

The values represent RSD% from five samples derivatized at each denoted level.

9.0 12.5 11.2 8.7 4.6 10.9 6.2

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

non-esterified carboxyl so that the compound remains as a salt in the aqueous phase. 3.2. Method validation 3.2.1. Repeatability of derivatization The repeatability of derivative formation of each amino acid in equimolar mixture was measured at levels 1 and 20 nmol in GC-FID, and at 0.5, 5 and 50 nmol in GC–MS/SIM analysis. Results are given in Tables 1–3, amino acids were sequenced according to the elution from the OV-5 phase. Variations are given in terms of RSD and the values represent the mean from five samples derivatized at each level. Most of the variations in FID analysis are less than 5%, a larger variation with values often exceeding 10% was measured in the MS/SIM regime, particularly at the lowest levels at the one-step method. Highest RSD gave Glu derivative, most probably due to double ester formation accompanied by a partial cyclization to pyroGlu. Somewhat lower variations were observed with the HFB derivatives (Table 3). 3.2.2. Linearity of response Utility of the novel approach was tested over a broad range of masses covering three orders of magnitude. In MS-SIM the range monitored was from 0.1 to 100 nmol, in FID from 0.5 to 100 nmol of each amino acid in equimolar mixtures. Ten nanomoles of Nval was added as I.S. into the samples prior to derivatization. As complex biofluid matrices such as body fluids or food products would rarely contain 1 ␮mol of each amino acid per

397

1 ml, mixtures containing half of the protein amino acids at such a level and half at a lower level were prepared and subjected to derivatization. In GC-FID as well as in GC–MS, the linear calibration curves were obtained with respective regression coefficients (r2 ). Values better than 0.990 were achieved with all amino acids in the former case (not tabulated), while in GC–MS/SIM the r2 values were lower (Table 4), evidently due to involving of five times lower level of 0.1 nmol. The lowest value denoted His (0.953), Ser/SerOR and Gln (0.972) derivatized with PFPCF via the two-step mode, and Asp, Glu and Pro (0.972) at the one-step procedure. With the HFBCF onestep approach, Ser showed the lowest value of 0.960 but SerOR instead was the preferable product for quantitation. The response appeared as rather nonlinear at the lowest level (injected amounts <1 pmol), particularly for amino acids with nontreated or labile polar functional groups i.e. Gln, Ser, 1MH and 3MH. 3.2.3. Limit of detection (LOD) LOD was defined as the amount of compound in picomoles resulting in a peak with a signal-to-noise ratio of three. LOD values determined in MS-SIM analysis are given for the particular analyte in Table 4; amino acids are sequenced according to the retention of PFPCF-analytes on the 20-m VF-5 column (the HFBCF derivatives were correspondingly reordered). A broad range of values is evident, ranging from 0.03 pmol for PFPCFPro to 19.38/18.75 pmol for HFBCF/PFPCF-Glu, derivatized via the one-step method. The high LOD value of Glu is due to the low formation of the expected diester following the onestep procedure. A few polar amino acids like Gln with free amide, Ser with free hydroxyl, and methylhistidines (especially

Table 4 Studies on linearity of amino acid determination by GC–MS/SIM in the range of 0.1–100 nmol of each amino acid Amino acid

Ala Gly Val Leu Ile Pro Thr Ser Asp Asn SerOR Glu Met Phe Gln Orn 3MH His Lys Tyr CC Trp

HFBCF reagent

PFPCF reagent

One-step procedure

Two-step procedure

One-step procedure

Two-step procedure

r2

LOD

r2

LOD

r2

LOD

r2

LOD

0.994 0.991 0.994 0.999 0.992 0.988 0.982 0.960 0.975 0.898 0.981 0.985 0.989 0.994 0.982 0.989 0.983 0.988 0.993 0.990 0.990 0.992

0.52 0.20 0.46 0.50 0.47 0.09 5.60 8.44 1.17 0.89 1.32 19.38 12.73 0.19 12.80 2.32 9.35 7.21 1.01 0.33 4.94 0.48

0.994 0.990 0.991 0.999 0.991 0.985 0.994 0.980 0.991 0.989 0.976 0.992 0.990 0.988 0.984 0.981 0.984 0.991 0.983 0.996 0.983 0.992

0.28 0.28 0.48 1.13 1.85 0.22 4.11 5.76 0.62 1.96 1.76 2.71 5.43 0.67 6.52 1.82 6.17 3.46 1.55 1.00 7.51 0.47

0.992 0.986 0.990 0.994 0.996 0.973 0.988 0.995 0.972 0.990 0.994 0.972 0.994 0.978 0.997 0.975 0.987 0.983 0.990 0.989 0.989 0.996

0.08 0.24 0.45 0.98 1.37 0.03 3.67 7.51 1.43 1.08 1.08 18.75 10.39 0.11 10.17 1.15 3.76 1.31 0.94 0.33 2.95 0.61

0.981 0.993 0.990 0.994 0.991 0.979 0.980 0.971 0.986 0.985 0.972 0.978 0.990 0.981 0.972 0.984 0.985 0.953 0.969 0.993 0.980 0.993

0.52 0.54 1.77 1.00 1.37 0.14 2.81 7.52 2.49 2.38 3.92 3.20 3.42 0.53 17.19 1.11 3.34 2.71 1.23 0.58 3.28 0.27

LOD values expressed in picomoles per initial sample amount.

398

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

Table 5 Mass fragment ions of PFPCF- and HFBCF-treated amino acid derivatives and their relative abundances Amino acid/peptide

␣-Aminobutyric acid Alanine Asparagine Aspartic acid ␤-Aminoisobutyric Cystine Cysteine Cystathionine Glutamine Glutamic acid Glycine Homocystine Histidine Homocysteine Homoserine Hydroxylysine Hydroxyproline Isoleucine Leucine Lysine Methionine 1-Methylhistidine 3-Methylhistidine Norvaline Ornithine Orn (lactam) Phenylalanine Prolylhydroxyproline Proline Selenomethionine Serine (free OH) Serine (O-alkyl) Threonine (free OH) Threonine (O-alkyl) Thiaproline Tryptophan Tyrosine Valine

Abbreviation

ABA Ala Asn Asp Baiba C–C Cys CTH Gln Glu Gly HCC His Hcy Hser Hyl Hyp Ile Leu Lys Met 1MH 3MH NVal Orn Phe PHP Pro SeMet Ser SerOR Thr ThrOR Tpro Trp Tyr Val

HFB

PFP

Fragment ions (% abundance)

Fragment ions (% abundance)

284(100) 270(100) 295(100) 254(100) 256(100) 496(100) 285(100) 328(100) 84(100) 282(100) 256(100) 510(100) 307(100) 282(100) 100(100) 269(100) 312(100) 312(100) 312(100) 310(100) 283(100) 95(100) 95(100) 298(100) 296(100) 296(100) 91(100) 296(100) 296(100) 282(100) 283(100) 268(100) 283(100) 282(100) 287(100) 130(100) 333(100) 298(100)

84(15) 113(10) 113(29) 496(84) 112(33) 113(91) 113(94) 282(61) 282(95) 310(40) 113(16) 282(61) 362(31) 82(34) 283(42) 69(41) 294(36) 283(99) 256(93) 256(44) 61(84) 96(7) 96(19) 256(91) 256(21) 139(52) 330(79) 297(9) 297(8) 123(62) 286(56) 113(35) 100(28) 227(57) 314(100) 131(9) 289(61) 283(40)

113(9) 227(7) 95(17) 296(53) 113(16) 268(41) 300(57) 113(30) 327(66) 82(35) 212(10) 314(45) 113(28) 283(32) 56(41) 112(33) 113(20) 256(73) 270(29) 113(17) 282(50) 150(7) 350(18) 113(27) 113(12) 69(48) 131(19) 70(5) 113(5) 109(57) 86(53) 183(12) 113(20) 113(57) 86(25) 129(2) 107(13) 98(16)

234(100) 220(100) 245(100) 396(100) 206(100) 396(100) 133(100) 278(100) 84(100) 232(100) 206(100) 410(100) 257(100) 232(100) 100(100) 219(100) 262(100) 233(100) 206(100) 260(100) 61(100) 95(100) 95(100) 248(100) 246(100) 246(100) 91(100) 246(100) 246(100) 232(100) 233(100) 218(100) 233(100) 232(100) 264(100) 130(100) 283(100) 248(100)

133(14) 133(14) 133(45) 204(74) 112(22) 133(66) 235(73) 232(60) 232(73) 260(45) 133(21) 232(60) 133(41) 233(35) 233(61) 69(32) 244(27) 262(91) 262(92) 206(53) 233(93) 284(56) 285(33) 206(77) 206(35) 139(53) 280(63) 247(8) 133(8) 410(93) 86(61) 133(52) 100(21) 133(89) 237(98) 131(9) 239(55) 233(53)

84(13) 177(8) 95(17) 246(39) 133(20) 218(36) 250(44) 133(27) 277(70) 82(26) 162(8) 264(37) 312(37) 82(33) 128(45) 206(29) 69(15) 206(75) 220(31) 133(21) 75(51) 285(7) 476(27) 133(19) 133(22) 96(40) 131(13) 133(4) 160(7) 123(51) 236(50) 69(18) 83(20) 177(65) 133(28) 246(3) 472(15) 98(21)

3.2.4. EI mass spectra The EI mass spectra of all amino acid derivatives were recorded and the principal diagnostic fragment ions and their relative abundance tabulated in alphabetical order in Table 5.

noticeable loss of reactivity. The derivatized amino acids did not show changes upon standing at room temperature in the autosampler overnight. His appeared to be most sensitive over time; after a week’s storage in a refrigerator its response, if any at all, was lowered significantly. There was help in replacing the extraction medium by another organic solvent. If isooctane in the sample was blown down to a small drop of about 10–20 ␮l, and replaced by the aforementioned liquid, His was not lost even after 10 days of storage. Such a replacement might be useful, provided that longer storage was required.

3.3. Stability of reagents and derivatives

3.4. Separation system

According to the findings of Vincenti et al. [25], the chloroformates with bulky fluorinated alkyls were slowly decomposed, losing the original reactivity stepwise within months. On the contrary, the PFPCF and HFBCF reagents were found to be stable even after 2 years of storage in a refrigerator without any

From numerous silicone phases, the OV-5 phase, for example, the popular and widely used 5% phenylmethylsilicone, proved to be very effective in separating both, PFPCF and HFBCF derivatives of a set of nearly 30 amino acids (except Arg that does not elute from the column). In our search for a suitable column

1MH) inclined to losses or even disappearance at the lowest injected amount, which might be caused by the sorption of the particular analyte or breakdown of the derivative in the GC injector.

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

399

Fig. 3. TIC chromatogram (20-m VF-5 column) of GC–MS/EI analysis of amino acid standards (mix 2) treated with HFBCF via the two-step procedure. Peaks: 1 = Ala, 2 = Gly, 3 = Aba, 4 = Val, 5 = Baiba, 6 = Nval, 7 = Leu, 8 = Ile, 9 = Pro, 10 = Thr, 11 = Ser, 12 = Asn + HSer, 13 = Asp, 14 = ThrOR, 15 = TPro, 16 = SerOR, 17 = Glu, 18 = Met, 19 = Cys, 20 = SeMet, 21 = Phe, 22 = Hcy, 23 = Gln, 24 = Orn + 3MH, 25 = His, 26 = 1MH, 27 = Lys, 28 = Tyr, 29 = Hyl, 30 = C–C + PHP, 31 = Trp, 32 = PHP-OR.

Fig. 4. TIC chromatogram (20-m VF-5 column) of GC–MS/EI analysis of amino acid standards (mix 2) treated with PFPCF via the two-step procedure. Peaks: 1 = Ala, 2 = Gly, 3 = Aba, 4 = Val, 5 = Baiba, 6 = Nval, 7 = Leu, 8 = Ile, 9 = Pro, 10 = Thr, 11 = Ser, 12 = Asp, 13 = Asn, 14 = HSer, 15 = SerOR, 16 = TPro, 17 = Glu, 18 = Met, 19 = Cys, 20 = SeMet, 21 = Phe, 22 = Hcy, 23 = Gln, 24 = Orn, 25 = 3MH, 26 = His, 27 = 1MH, 28 = Lys, 29 = Tyr, 30 = Hyl, 31 = C–C, 32 = PHP, 33 = Trp.

from several distributors, the thin-coated ZB-5 capillary, 30 m in length, was found favorable for the separation of both derivative types (Figs. 1 and 2). With one exception, HFBCF-treated cystine and Trp are coeluted. With HFBCF-treated amino acids, coelution of the pairs 3MH/Orn, and 1MH/His took place on the 20-m VF-5 column, as apparent in Fig. 3. It was not the case

with PFP-analytes (Fig. 4), which were separated even on the 10-m-long column of that kind (not shown). Elution of several polar and heavier molecules in GC–MS analysis, such as PHP, Hyl, MHs, and homocystine was diminished to a certain extent (Figs. 3 and 4; homocystine was not shown but its elution in the GC-FID system was successful).

400

P. Huˇsek et al. / J. Chromatogr. A 1186 (2008) 391–400

4. Conclusions

References

The first-time synthesized fluoroalkyl chloroformates immediately attack protic functional groups of amino acids in aqueous media. Carboxylic groups are converted directly to esters, primary and secondary amino groups to carbamates and thiol and phenolic groups to carbonates at low picomole to high nanomole levels. The process is catalyzed by pyridine or 3picoline, and hydrogen chloride and carbon dioxide are released as by-products. The reaction conditions were extensively studied and the following findings experimentally proved. First, both pyridine and 3-picoline were proven as efficient catalysts allowing facile esterification of the carboxylic group in water or aqueous buffers; their optimum concentration in the medium was 3–5 vol%. Second, the molar ratio between the basic aqueous medium, i.e. the catalyst plus carbonate/hydroxide, and the “acidic” reagent (i.e. the liberated hydrogen chloride) should be close to one, with a slight excess of the base. Third, treating the aqueous phase with the reagent only was found clearly inferior as leading to low reaction yields, probably due to the difficult immiscibility of the heavy reagent drop at the vial bottom. Fourth, a marked improvement of the yields was achieved upon co-mixing the reagent (densities >1.4 g/ml) with a light hydrocarbon such as isooctane. Fifth, admixing of the corresponding fluorinated alcohol into the organic medium proved to be counterproductive with HFBCF but somewhat assisted with the PFPCF reagent. Sixth, the excessive rise of pH upon using hydroxide or higher carbonate concentrations (>100 mmol/l) led to lowered yields of aliphatic and dibasic amino acids. Likewise, higher concentrations of pyridine or picoline diminished yields of the acidic amino acids. Last, the presence of some neutral salts, such as natrium chloride or perchlorate, did not influence the yields of particular amino acids negatively. In summary, the novel derivatization reagents proved to be perfectly suitable for converting amino acids into species with excellent features in GC analysis. First application focused on chiral amino acid analysis in cyanobacterial cultures [28], the next one aimed at enantiomer analysis of a pharmaceutically important peptide [29]. Current studies deal with simultaneous screening of serum methylmalonic acid, homocysteine and related metabolites in cobalamin deficiency stages.

[1] D.H. Spackman, W.H. Stein, S. Moore, Anal. Chem. 30 (1958) 1190. [2] I. Molnar-Perl (Ed.), Quantitation of Amino Acids and Amines by Chromatography—Methods and Protocols (Journal of Chromatography Library, vol. 70), Elsevier, Amsterdam, 2005. [3] T. Fukushima, N. Usui, T. Santa, K. Imai, J. Pharm. Biomed. Anal. 30 (2003) 1655. [4] P. Huˇsek, FEBS Lett. 280 (1991) 354. ˇ [5] P. Huˇsek, P. Simek, LC–GC N. Am. 19 (2001) 986. ˇ [6] P. Huˇsek, P. Simek, Curr. Pharm. Anal. 2 (2006) 23. [7] T.E. Wheat, E.E. Grumbach, J.R. Mazzeo, LC–GC N. Am. 6 (2006) 24. [8] M. McCooeye, Z. Mester, Rapid Commun. Mass Spectrom. 20 (2006) 1801. [9] M. Zoppa, L. Gallo, F. Zacchello, G. Giordano, J. Chromatogr. B 831 (2006) 267. [10] M. Piraud, C. Vianey-Saban, C. Bourdin, C. Acquaviva-Bourdain, S. Boyer, C. Elfakir, D. Bouchu, Rapid Commun. Mass Spectrom. 19 (2005) 3287. [11] B. Casetta, D. Tagliacozzi, B. Shushan, G. Federici, Clin. Chem. Lab. Med. 38 (2000) 391. [12] Z.F. Liu, P.E. Minkler, D. Lin, L.M. Sayre, Rapid Commun. Mass Spectrom. 18 (2004) 1059. [13] P. Jandik, J. Cheng, N. Avdalovic, J. Biochem. Biophys. Methods 60 (2004) 191. [14] K. Petritis, M. de Person, C. Elfakir, M. Dreux, Chromatographia 60 (2004) 293. [15] T. Soga, Y. Kakazu, M. Robert, M. Tomita, T. Nishioka, Electrophoresis 25 (2004) 1964. [16] V. Samcov´a, P. T˚uma, Electroanalysis 18 (2006) 152. [17] P. Lochman, T. Adam, D. Friedecky, E. Hlidkova, Z. Skopkova, Electrophoresis 24 (2003) 1200. [18] J.P. Xie, J.Y. Zhang, H.X. Liu, J.Q. Liu, J.N. Tian, X.G. Chen, Z.D. Hu, Biomed. Chromatogr. 18 (2004) 600. [19] K.W. Ro, J.H. Hahn, Electrophoresis 26 (2005) 4767. [20] M.A. Alterman, N.V. Gogichayeva, B.A. Kornilayev, Anal. Biochem. 335 (2004) 184. [21] I. Abe, N. Fujimoto, T. Nakahara, Chem. Lett. (1995) 113. [22] I. Abe, N. Fujimoto, T. Nishiyama, K. Terada, T. Nakahara, J. Chromatogr. A 722 (1996) 221. [23] J.T. Simpson, D.S. Torok, S.P. Markey, J. Am. Soc. Mass Spectrom. 6 (1995) 525. [24] J.T. Simpson, D.S. Torok, J.E. Girard, S.P. Markey, Anal. Biochem. 233 (1996) 58. [25] M. Vincenti, N. Ghiglione, M.C. Valsania, P. Davit, S.D. Richardson, Helv. Chim. Acta 87 (2004) 370. [26] J. Wang, Z.H. Huang, D. Gage, J.T. Watson, J. Chromatogr. A 663 (1994) 71. [27] V. Felt, P. Huˇsek, J. Chromatogr. 197 (1980) 226. ˇ [28] H. Zahradn´ıcˇ kov´a, P. Huˇsek, P. Simek, P. Hartvich, B. Marˇsa´ lek, I. Holoubek, Anal. Bioanal. Chem. 388 (2007) 1815. ˇ [29] H. Zahradn´ıcˇ kov´a, P. Hartvich, P. Simek, P. Huˇsek, Amino Acids (in press).

Acknowledgements The Financial support of the Grant Agency of the Czech Republic in frame of the projects nos. 203/04/0192 and 303/06/1674 is highly acknowledged.