*Article* **GC-MS Studies on Derivatization of Creatinine and Creatine by BSTFA and Their Measurement in Human Urine**

**Olga Begou † , Kathrin Weber † , Bibiana Beckmann and Dimitrios Tsikas \***

> Institute of Toxicology, Core Unit Proteomics, Hannover Medical School, 30625 Hannover, Germany; mpegolga@chem.auth.gr (O.B.); kathrin.weber89@gmx.net (K.W.); beckmann.bibiana@mh-hannover.de (B.B.) **\*** Correspondence: tsikas.dimitros@mh-hannover.de

† These authors contributed equally to this work.

**Abstract:** In consideration of its relatively constant urinary excretion rate, creatinine (2-amino-1 methyl-5*H*-imidazol-4-one, MW 113.1) in urine is a useful endogenous biochemical parameter to correct the urinary excretion rate of numerous endogenous and exogenous substances. Reliable measurement of creatinine by gas chromatography (GC)-based methods requires derivatization of its amine and keto groups. Creatinine exists in equilibrium with its open form creatine (methylguanidoacetic acid, MW 131.1), which has a guanidine and a carboxylic group. Trimethylsilylation and trifluoroacetylation of creatinine and creatine are the oldest reported derivatization methods for their GC-mass spectrometry (MS) analysis in human serum using flame- or electron-ionization. We performed GC-MS studies on the derivatization of creatinine (d<sup>0</sup> -creatinine), [*methylo*-<sup>2</sup>H<sup>3</sup> ]creatinine (d<sup>3</sup> creatinine, internal standard) and creatine (d<sup>0</sup> -creatine) with *N*,*O*-*bis*(trimethylsilyl)trifluoroacetamide (BSTFA) using standard derivatization conditions (60 min, 60 ◦C), yet in the absence of any base. Reaction products were characterized both in the negative-ion chemical ionization (NICI) and in the positive-ion chemical ionization (PICI) mode. Creatinine and creatine reacted with BSTFA to form several derivatives. Their early eluting *N*,*N*,*O*-*tris*(trimethylsilyl) derivatives (8.9 min) were found to be useful for the precise and accurate measurement of the sum of creatinine and creatine in human urine (10 µL, up to 20 mM) by selected-ion monitoring (SIM) of *m*/*z* 271 (d<sup>0</sup> -creatinine/d<sup>0</sup> -creatine) and *m*/*z* 274 (d<sup>3</sup> -creatinine) in the NICI mode. In the PICI mode, SIM of *m*/*z* 256, *m*/*z* 259, *m*/*z* 272 and *m*/*z* 275 was performed. BSTFA derivatization of d<sup>0</sup> -creatine from a freshly prepared solution in distilled water resulted in formation of two lMate-eluting derivatives (14.08 min, 14.72 min), presumably creatinyl-creatinine, with the creatininyl residue existing in its enol form (14.08 min) and keto form (14.72 min). Our results suggest that BSTFA derivatization does not allow specific analysis of creatine and creatinine by GC-MS. Preliminary analyses suggest that pentafluoropropionic anhydride (PFPA) is also not useful for the measurement of creatinine in the presence of creatine. Both BSTFA and PFPA facilitate the conversion of creatine to creatinine. Specific measurement of creatinine in urine is possible by using pentafluorobenzyl bromide in aqueous acetone.

**Keywords:** BSTFA; creatine; creatinine; derivatization; quantification; silylation; TMS; validation

#### **1. Introduction**

Creatinine (2-amino-1,5-dihydro-1-methyl-4*H*-imidazol-4-one, MW 113.12; see Scheme 1) is the end-product of creatine catabolism. Creatinine is excreted in the urine with a fairly constant rate and is generally used for the correction of renal excretion rates of endogenous and exogenous substances. This correction is indispensable in clinical studies when urine specimens from spontaneous micturition must be analyzed [1]. The mean concentration of creatinine in urine samples of healthy adults is approximately 12–13 mM, with men excreting higher amounts of creatinine than women [1]. Besides the spectrophotometric method based on the famous Jaffé reaction [2] many different analytical methods are currently available for creatinine. They include spectrophotometric, enzymatic and instrumental methods

**Citation:** Begou, O.; Weber, K.; Beckmann, B.; Tsikas, D. GC-MS Studies on Derivatization of Creatinine and Creatine by BSTFA and Their Measurement in Human Urine. *Molecules* **2021**, *26*, 3206. https://doi.org/10.3390/ molecules26113206

Academic Editor: Paraskevas D. Tzanavaras

Received: 7 May 2021 Accepted: 22 May 2021 Published: 27 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

based on HPLC, GC-MS, LC-MS and LC-MS/MS [3–19]. Lawson [20] and Siekmann [21] demonstrated by electron ionization (EI) that creatinine reacts with silylation reagents to form its *N*,*N*,*O*-*tris*(trimethylsilyl) derivative. Björkhem and colleagues used trifluoroacetic anhydride for the derivatization of creatinine and its GC-MS analysis [22]. Trimethylsilylation derivatization reactions used in MS-based methods were found to be associated with interferences due to formation of several derivatives [23]. To our knowledge, the GC-MS measurement of urinary creatinine as *N*,*N*,*O*-*tris*(trimethylsilyl) derivative by negative-ion chemical ionization (NICI) or positive-ion chemical ionization (PICI) has not been reported thus far.

**Scheme 1.** Schematic of the expected derivatization reaction and products of (**A**) unlabeled creatinine (d<sup>0</sup> -creatinine), (**B**) deuterium-labelled creatinine ([*methylo*-<sup>2</sup>H<sup>3</sup> ]creatinine, d<sup>3</sup> -creatinine) and (**C**) unlabeled (d<sup>0</sup> -creatine) and with *N*,*Otris*(trimethylsilyl)trifluoroacetamide (BSTFA) to form their *N*,*N*,*O*-*tris*(trimethylsilyl) creatinine derivatives (**A**,**B**) and *N*,*N*,*N*',*O*-tetrakis(trimethylsilyl creatine derivative (**C**).

In the present study, we investigated in detail the derivatization of unlabelled creatinine (d0-creatinine), commercially available [methylo-2H3]creatinine (d3-creatinine) and unlabelled creatine (d0-creatine) by *N*,*O*-*bis*(trimethylsilyl)trifluoroacetamide (BSTFA), one of the oldest trimethylsilylation reagents for amino acids [24]. It is well known that BSTFA reacts with many functional groups, notably hydroxyl, carboxyl and amine groups. Based on this knowledge we expected that BSTFA will react with creatinine and creatine to form *N*- and *O*-derivatives (Scheme 1).

In our study, we used GC-MS in the NICI and in the PICI mode, confirmed the formation of the expected *N*,*N*,*O*-*tris*(trimethylsilyl) derivatives and identified several derivatives of creatinine and creatine that have not reported thus far. As creatinine and creatine are in a pH-dependent equilibrium and inter-convertible, our results suggest that BSTFA and GC-MS are not specific for creatinine and creatine but allow measurement of their sum. Using d3-creatinine as the internal standard we demonstrate that creatinine can be reliably quantitated in 10-µL aliquots of human urine by GC-MS as *N*,*N*,*O*-*tris*(trimethylsilyl) derivative with minimum labour.

#### **2. Materials and Methods**

#### *2.1. Chemicals and Materials*

Unlabeled creatine (d0-creatine), unlabeled creatine phosphate 4×H2O, unlabeled creatinine (d0-creatinine) and trideuterocreatinine, i.e., [methylo-2H3]creatinine (d3-creatinine; declared isotopic purity of >99 atom% <sup>2</sup>H) were obtained from Aldrich (Steinheim, Germany). Stock solutions of d0-creatine, d0-creatinine and d3-creatinine (each 20 mM) were prepared in deionized water and stored in a refrigerator at 8 ◦C. BSTFA was obtained from Macherey-Nagel (Düren, Germany). Glassware for GC–MS (i.e., 1.5 mL autosampler glass vials and 0.2 mL microvials) and a fused-silica capillary column Optima 17 (15 m × 0.25 mm I.D., 0.25 µm film thickness) were purchased from Macherey-Nagel.

#### *2.2. Derivatization Procedure for Creatinine in Human Urine Samples*

Urine samples used in method development and validation were obtained from healthy volunteers being members of the researcher group and authors of this manuscript. Urine samples (1-mL aliquots) were kept frozen at −18 ◦C until analysis. Prior to sample derivatization, urine samples were thawed and centrifuged (5800× *g*, 5 min). Urine (10 µL) and synthetic creatinine-containing samples (usually 10 µL) were evaporated to complete dryness under a stream of nitrogen. Subsequently, the samples were treated with 100 µL absolute ethanol and the solvents were evaporated to dryness by a stream nitrogen gas to remove remaining water. Then, the residues were reconstituted with pure BSTFA (100 µL), the glass vials were tightly closed and heated for 60 min at 60 ◦C in a thermostat. After cooling to room temperature, aliquots (about 90 µL) were transferred into 1.8-mL autosampler glass vials equipped with 200-µL microinserts. Aqueous solutions (usually 10 µL aliquots) of creatinine and creatine were derivatized as described above.

#### *2.3. GC–MS Conditions*

In this work, we used a GC-MS method previously used in our group for amino acid derivatives [25]. GC-MS analyses were performed on a single-quadrupole mass spectrometer model ISQ directly interfaced with a Trace 1310 series gas chromatograph equipped with an autosampler AS 1310 from ThermoFisher (Dreieich, Germany). The following oven temperature program was used with helium as the carrier gas at a constant flow rate of 1 mL/min: 0.5 min at 40 ◦C, then increased to 210 ◦C at a rate of 15 ◦C/min and to 320 ◦C at a rate of 35 ◦C/min, respectively, and held at 320 ◦C for 1 min. Interface, injector and ion-source were kept at 300 ◦C, 280 ◦C and 250 ◦C, respectively. Electron energy was set to 70 eV and electron current to 50 µA. Methane (2.4 mL/min) was used as the reagent gas for NICI and PICI. Aliquots (1 µL from derivatization mixtures) were injected in the splitless mode by means of the autosampler using a 10-µL Hamilton needle, which was cleaned automatically three times with toluene (5 µL) after each injection. Quantitative analyses were performed in the selected-ion monitoring (SIM) mode. The peak area (PA) values of d0-creatinine and d3-creatinine were calculated automatically by the GC–MS software (Xcalibur and Quan Browser). The concentration of d0-creatinine was calculated by multiplying the peak area ratio (PAR) of d0-creatinine to d3-creatinine with the concentration of d3-creatinine added to the sample. Statistical analyses and graphs were performed and prepared by GraphPad Prism 7 (San Diego, CA, USA).

#### *2.4. HPLC Analysis of Creatine, Creatinine and Creatine-Phosphate in HCl Solutions*

We used a HPLC method previously reported by our group for creatinine measurement in human urine [16]. HPLC analyses were carried out on the an HPLC system consisting of an Agilent 1100 Series binary pump G1312A, an Agilent 1100 Series Degaser G1322A, an Agilent 1100 Series oven column Colcom G1316A, an Agilent 1100 Series VWD detector (all Agilent, Waldbronn, Germany, and an MP3 autosampler (Gerstel, Mülheim, Germany), ChemStation for LC-Systems Rev.B.0402SP1 (212) and Gerstel Maestro Version 1.3.20.41.13.5 were used to control the HPLC system and evaluate the analyses. HPLC analyses were performed on a Kinetex 5 µm EVO C18 100 Å column (250 × 4.6 mm) from Phenomenex (Aschaffenburg, Germany) at a fixed column temperature of 20 ◦C. The mobile phase was 100 mM sodium acetate, pH 7.5, 10 vol% methanol and was pumped at a flow rate of 1.0 mL/min. Samples (20 µL) were injected by means of the autosampler. The effluent was monitored at 210 nm. The analysis time was 5 min. The retention time was 2.073 ± 0.018 min for creatine-phosphate, 2.252 ± 0.007 min for creatine and 2.547 ± 0.007 min for creatinine.

#### **3. Results**

#### *3.1. Generation of GC-MS Spectra and Characterization of Derivatization Products of d0-Creatine and of d3-Creatinine*

Each 100 nmol of d0-creatinine and d3-creatinine taken from their aqueous solutions were combined, the solvent was evaporated to dryness and derivatized with 100 µL BSTFA as described above. Derivatization resulted in a yellow-colored clear solution. The sample was analyzed by GC-MS in the PICI and NICI mode consecutively by injecting 1-µL aliquots of the BSTFA solutions corresponding each to 1 nmol of d0-creatinine and d3-creatinine (assuming quantitative derivatization). GC-MS spectra were generated by scanning the quadrupole in the mass-to-charge (*m*/*z*) ratio range of 50–650 and 50–1000 (1 scan per s). We observed two chromatographic peaks with the retention time of 8.6 min and 8.9 min (major peak) and their GC-MS spectra contained paired *m*/*z* values differing by 3 Da due to the three deuterium atoms in methyl group of d3-creatinine (Figure 1).

The most intense anions in the NICI mass spectrum (Figure 1A) of the GC peak eluting at 8.9 min were *m*/*z* 271 and *m*/*z* 274 (base peaks). Less intense anions were found at *m*/*z* 199 and *m*/*z* 202, and very weak ions (intensity < 1%) were *m*/*z* 326 and *m*/*z* 329, presumably due to molecular anions of the derivatives (i.e., [M]−). These data indicate the presence of the unlabeled methyl group in d0-creatinine and of the deuterium-labeled methyl group of d3-creatinine in this peak (Figure 1A). The NICI spectrum of this GC peak also contained weak anions at *m*/*z* 144 and *m*/*z* 186 that do not carry the original methyl group of creatinine (Figure 1A). The PICI mass spectrum of the GC peak eluting at 8.9 min contained intense cations at *m*/*z* 272, *m*/*z* 275, *m*/*z* 256 and *m*/*z* 259, less intense cations at *m*/*z* 300 and *m*/*z* 303, weaks ions at *m*/*z* 312 and *m*/*z* 315, and very weak ions (intensity < 1%) at *m*/*z* 330 and *m*/*z* 333, presumably due to the protonated molecules of the derivatives (i.e., [M+H]<sup>+</sup> ) (Figure 1B).

These data indicate the presence of the unlabeled methyl group in d0-creatinine and the deuterium-labeled methyl group of d3-creatinine in this peak (Figure 1). Comparison of the total ion intensity in the NICI and PICI mass spectra (1.85 × 10<sup>6</sup> versus 9.92 × 10<sup>5</sup> , Figure 1) suggests that NICI may allow for a more sensitive detection of creatinine than PICI. Proposed fragmentation mechanisms of the *N*,*N*,*O*-trimethylsilyl derivatives in the PICI are shown in Scheme 2.

The smaller GC peak eluting at 8.6 min had closely comparable NICI and PICI mass spectra to those of the *N*,*N*,*O*-*tris*(trimethylsilyl) derivative (data not shown). These observations suggest that the GC peak with the retention time of 8.6 min is an isomer of the *N*,*N*,*O*-*tris*(trimethylsilyl) derivative of creatinine.

**Figure 1.** (**A**) Negative-ion chemical ionization (NICI) and (**B**) positive-ion chemical ionization (PICI) GC-MS spectra generated from a mixture of d<sup>0</sup> -creatinine (blue) and d<sup>3</sup> -creatinine (red) after derivatization with *N*,*Otris*(trimethylsilyl)trifluoroacetamide (BSTFA) at 60 ◦C for 60 min (each 1 nmol injected). The retention time (*t*R) of the GC-MS peak was 8.9 min. Insets indicate the proposed structures of the derivatives and ions. See Scheme 2.

**Scheme 2.** Proposed fragmentation mechanisms for the *N*,*N*,*O*-trimethylsilyl derivatives of d<sup>0</sup> -creatinine (**A**, blue) and d3 -creatinine (**B**, red) of the GC-MS peak with the retention time of 8.9 min in the PICI mode. The numbers give the molecular weight of the neutral substances and the cations. See Figure 1B.

The GC peak with the retention time of 8.7 min was only detectable in the PICI mode. The PICI mass spectrum of this peak contained three pairs of cations differing by 3 Da due to the presence of d3-creatinine, i.e., *m*/*z* 314 and *m*/*z* 317 (base peaks), *m*/*z* 330 and *m*/*z* 333 ([M+H]<sup>+</sup> ), and *m*/*z* 358 and *m*/*z* 361 ([M+C2H4+H]<sup>+</sup> ) (Figure 2). Adducts such as C2H<sup>4</sup> (28 Da) are common in PICI of amines such as dimethyl amine and derive from the reactant gas methane [26]. Presumably, the adduct is on the non-ring amine group. These observations suggest the GC peak eluting at 8.7 min is a creatinine derivative with three trimethylsilyl (TMS) groups, most likely the *N<sup>2</sup>* ,*N<sup>3</sup>* ,*O<sup>4</sup>* -*tris*(trimethylsilyl) derivative. It cannot ionize in the NICI mode, presumably because of the inability to form anions by loss of an H atom or by capturing an electron due to the lack of electron-capturing atoms and functional groups in the derivative. The cations with *m*/*z* 314 and *m*/*z* 317 seem to be very stable and do not fragment. The cations *m*/*z* 55, *m*/*z* 57, *m*/*z* 73 and *m*/*z* 147 are shared by d0-creatinine and d3-creatinine and are likely to be associated with the TMS groups (see also [23]) of the derivatives (see also Figure 1B).

**Figure 2.** PICI GC-MS spectrum generated from a mixture of d<sup>0</sup> -creatinine (blue) and d<sup>3</sup> -creatinine (red) after derivatization with BSTFA at 60 ◦C for 60 min. The retention time (*t*R) of the GC-MS peak was 8.7 min. Insets indicate the proposed structures for the mass fragments. See also Figure 1B.

#### *3.2. Generation of GC-MS Spectra and Characterization of Derivatization Products of d0-Creatine*

Derivatization of d0-creatine with BSTFA (60 ◦C, 60 min) resulted in the formation of three GC-MS peaks with the retention times of the d0-creatinine. The NICI and PICI mass spectra of these derivatives were virtually identical with those of the d0-creatinine derivatives (data not shown). In order to investigate the potential formation of additional derivatives of d0-creatine we extended the upper *m*/*z* scanning range to 1000 and the acquisition time to 16 min. We observed two GC-MS eluting at 14.08 min (minor peaks) and 14.72 min (major peaks) in the NICI and PICI mode. The corresponding GC-MS spectra of these d0-creatine derivatives and the relatively high difference in their long retention times suggest that these peaks are not derivatives of d0-creatine or d0-creatinine (Figure 3). A possible explanation could be the formation of a creatinyl-derivative by the reaction of two creatine molecules and/or by the reaction of a creatine molecule and with a molecule of creatinine formed from creatine during the derivatization. The peak with shorter retention time could be due to its TMS ether functionality compared to the keto group.

**Figure 3.** NICI (**A**,**C**) and PICI (**B**,**D**) GC-MS spectra generated from the BSTFA derivatization (60 ◦C, 60 min) of a freshly prepared solution of d<sup>0</sup> -creatine in deionized water upon its evaporation to complete dryness. The retention times (*t*R) of the GC-MS peak were 14.08 min (minor peak, red) and 14.72 min (major peak, blue). Insets indicate the proposed structures for the mass fragments. The same oven column temperature program was used as in Figures 1 and 2.

#### *3.3. Standardization of [methylo-2H3]Creatinine*

The isotopic purity of stable isotope-labelled analogs is of particular importance in quantitative analyses [27]. The isotopic purity of the commercially available [*methylo*-<sup>2</sup>H3]creatinine was verified as follows.

Nine separate samples containing each 100 nmol of d0-creatinine and d3-creatinine were derivatized with BSTFA (100 µL) and 1-µL aliquots of their solutions were analyzed by SIM of *m*/*z* 256, *m*/*z* 259, *m*/*z* 271, and *m*/*z* 274 (peak with retention time 8.9 min). Analysis of the sample containing d0-creatinine generated a mean PAR of 0.02223 ± 0.00329 (RSD, 15%; *n* = 9) for *m*/*z* 259 to *m*/*z* 256, and a PAR of 0.01302 ± 0.00209 (RSD, 16%; *n* = 9) for *m*/*z* 274 to *m*/*z* 271. Analysis of the samples that contained d3-creatinine produced a mean PAR of 0.000588 ± 0.0001411 (RSD, 24%; *n* = 9) for *m*/*z* 256 to *m*/*z* 259 and a mean PAR of 0.01108 ± 0.0002861 (RSD, 2.6%; *n* = 9) for *m*/*z* 271 to *m*/*z* 274. These observations indicate the presence of only very low amounts of d0-creatinine in the commercial [*methylo*-<sup>2</sup>H3]creatinine and confirm its declared isotopic purity (>99 atom% <sup>2</sup>H).

#### *3.4. Method Linearity, Precision and Accuracy*

For quantitative analyses of creatinine, we selected the *N*,*N*,*O*-*tris*(trimethylsilyl) derivative of creatinine with the retention of 8.9 min. The structure of this creatinine derivative is most likely *N<sup>2</sup>* ,*N<sup>2</sup>* ,*O<sup>4</sup>* -*tris*(trimethylsilyl). The structure with the ring-*N<sup>3</sup>* atom of creatinine, which is not derivatized, allows both PICI and NICI. In the NICI mode, SIM

of *m*/*z* 271 for d0-creatinine and *m*/*z* 274 for d3-creatinine was performed. A representative GC-MS chromatogram is shown in Figure 4 and indicates peaks with closely comparable intensity (2.66 × 10<sup>6</sup> versus 2.56 × 10<sup>6</sup> ) due to injection of nominally 1 nmol of each analyte. In the PICI mode, SIM of *m*/*z* 256 and *m*/*z* 272 for d0-creatinine and of *m*/*z* 259 and *m*/*z* 275 for d3-creatinine was performed. The dwell-time was 108 ms for all ions and the electron multiplier voltage was set to 2025 V.

**Figure 4.** Partial GC-MS chromatograms from the analysis of an equimolar mixture of d<sup>0</sup> -creatinine (blue) and d<sup>3</sup> -creatinine (red) after derivatization with BSTFA at 60 ◦C for 60 min (each 1 nmol injected). SIM of *m*/*z* 271 for d<sup>0</sup> -creatinine and *m*/*z* 274 for d<sup>3</sup> -creatinine was performed in the NICI mode.

Stock solutions (each 20 mM) of d0-creatinine and d3-creatinine were freshly prepared in Ampuwa deionized water. Dilutions of the stock solution of d0-creatinine were prepared using Ampuwa water providing d0-creatinine concentrations of 0, 2, 4, 6, 8, 10, 14 and 20 mM. Each 10-µL aliquots of these solutions were combined with each 5-µL aliquots of the 20 mM d3-creatinine stock solution. After evaporation to dryness under a stream of nitrogen gas, reconstitution of the residue in absolute ethanol and renewed evaporation to dryness, derivatization with 100 µL BSTFA each was performed (60 min, 60 ◦C). Then. 1-µL aliquots of the samples were injected in the splitless mode and analyzed in the PICI mode by SIM of *m*/*z* 256, *m*/*z* 259, *m*/*z* 272 and *m*/*z* 275. The amounts injected were 1 nmol for d3-creatinine in each sample and varying amounts of d0-creatinine (i.e., 0.0, 0.2, 0.4, 0.6, 0.8, 1.0, 1.4, 2 nmol). These analyses were performed by three persons in triplicate for each concentration. The precision (relative standard deviation, RSD) ranged between 0.1% and 8.4%. Linear regression analysis between the PAR *m*/*z* 256 to *m*/*z* 259 (*y*) or the PAR *m*/*z* 272 to *m*/*z* 275 (*y*) and the amount of d0-creatinine (nmol) (*x*) for all data resulted in straight lines with the regression equations *y* = 0.033 + 0.0087*x* (*r* <sup>2</sup> = 0.9945) and *y* = 0.001 + 0.0098*x* (*r* <sup>2</sup> = 0.9937), respectively (Figure 5). The reciprocal values of slopes of the straight lines were 115 nmol and 102 nmol and correspond to the nominal amount of d3-creatinine of 100 nmol used in the linearity experiment. Thus, SIM of *m*/*z* 272 and *m*/*z* 275 yields a higher mean accuracy than SIM of *m*/*z* 256 and *m*/*z* 259 (87% vs. 98%) (Figure 5).

**Figure 5.** Linear relationships between the peak area ratio (PAR) values and d<sup>0</sup> -creatinine amounts obtained by SIM of *m*/*z* 256 and *m*/*z* 272 for d<sup>0</sup> -creatinine and of *m*/*z* 259 and *m*/*z* 275 for d<sup>3</sup> -creatinine in the PICI mode. The indicated d0 -creatinine amounts and each 100 nmol d<sup>3</sup> -creatinine were derivatized with BSTFA (100 µL) at 60 ◦C for 60 min and 1 µL aliquots of the reaction mixture were injected in the splitless mode. Data are shown as mean ± standard deviation (*n* = 9). These analyses were performed by three persons in triplicate for each d<sup>0</sup> -creatinine amount. For more details, see the text.

#### *3.5. Measurement of Creatinine in Human Urine in the NICI Mode*

The method was validated in human urine samples in the NICI mode by the three persons who performed the experiment described above. Three healthy volunteers (#1, #2, #3) donated urine samples by spontaneous micturition. To 10-µL urine aliquots, d0 creatinine was added to reach final added concentrations of 2, 4, 6, 10, 14 and 20 mM. d3- Creatinine was also added to these samples to reach a fixed concentration of 10 mM in each urine sample. After evaporation to dryness under a stream on nitrogen gas, reconstitution of the residues in 100 µL aliquots of absolute ethanol and renewed evaporation to dryness, derivatization each with 100 µL BSTFA was performed (60 min, 60 ◦C) and 1-µL aliquots were injected and analyzed by SIM of *m*/*z* 271 and *m*/*z* 274 in the NICI mode. Linear regression analysis between the PAR of *m*/*z* 271 to *m*/*z* 274 (*y*) and the concentration of d0-creatinine (mM) (*x*) resulted in straight lines (Figure 6). The reciprocal slope values of the straight lines were 10.9 mM for urine #1, 11.0 mM for urine #2, and 10.2 mM for urine #3. Based on the nominal concentration of d3-creatinine of 10 mM in the urine samples, the mean accuracy is calculated to be 109%, 110% and 102% in the three human urine samples in the concentration range investigated. The *y* axis intercept values indicate mean basal creatinine+creatine concentrations of 1.7, 1.8 and 1 mM, respectively.

**Figure 6.** Linear relationships between the peak area ratio (PAR) values and the varying d<sup>0</sup> -creatinine concentrations added to human urine samples donated by three healthy volunteers and regression equations. d<sup>3</sup> -Creatinine was added at the fixed concentration of 10 mM and served as the internal standard. SIM of *m*/*z* 271 and *m*/*z* 274 for d<sup>0</sup> -creatinine and d<sup>3</sup> -creatinine was performed in the NICI mode, respectively. The analyses were performed by the three persons who performed the analyses shown in Figure 5. For more details, see the text.

#### *3.6. HPLC Analysis of Creatinine in HCl Solutions of Creatine*

The aim of these analyses was to estimate the extent of formation of creatinine from creatine and creatine-phosphate in hydrohloric acid solutions of varying molarity and incubation time at room. Linear relationships between the response (*y*), i.e., peak area, mAU×min at 210 nm, and the creatinine concentration in µM (*x*) was observed: *y* = 7.4 + 4.92 *x*, (*r* <sup>2</sup> = 0.9999) (range, 0–1000 µM). This regression equation was used to measure the concentration of creatinine in creatine solutions in hydrochloric acid (Figure 7). The concentration of creatinine in freshly prepared 5000 µM creatine solutions ranged between 2 and 7 µM and increased with increasing HCl molarity and incubation time up to 56 µM at 1 M HCl and 360 min (Figure 7A) and up to 2500–3000 µM after 63 days in 250–1000 mM HCl solutions (Figure 7B). The sigmoidal creatinine-incubation time profile in a 5000 µM solution of creatine in 25 mM HCl is shown in (Figure 7C). The highest creatinine concentration was determined to be 2150 µM after 69 days. Creatine-phosphate was found to be stable in deionized water. Similar experiments with HCl-solutions of creatine-phosphate did not result in formation of considerable amounts of creatinine (data not shown). Except for creatine and creatinine we did not detect appearance of additional peaks within HPLC run time of 5 min and UV absorbance detection at 210, 232 and 250 nm.

**Figure 7.** (**A**–**C**) Creatinine formed upon incubation of 5 mM creatinine in deionized water and in the indicated HCl solutions for the indicated times at room temperature (22–25 ◦C). Analyses were performed by HPLC with UV absorbance detection at 210 nm. Note the double decadic logarithmic scale in panel (**C**).

#### **4. Discussion**

Silylation is one of the most widely used derivatization reaction in analytical chemistry, notably in GC-based methods. Silylation reagents such as BSTFA and MSTFA are not specific, but react with different functionalities of organic compounds, especially of hydroxyl and amine groups, to form *O*- and *N*-trimethylsilyl derivatives [23]. Such derivatives are volatile and thermally stable in non-aqueous systems, best properties in GC-based analytical methods.

Creatinine, 2-amino-1-methyl-5*H*-imidazol-4-one (Scheme 1), is an endogenous substance, the final metabolite of creatine catabolism. Creatinine can be formed chemically from creatine by acid-catalyzed cyclization (Scheme 1). The most significant field of interest in creatinine is Clinical Chemistry. Serum creatinine serves as an indicator of kidney function. Urinary creatinine is of particular importance in clinical, pharmacological and epidemiological studies, where biomarkers must be measured in urine collected from spontaneous micturition, i.e., when the urine volume and the time between two urine collections are unknown. This particular importance is because creatinine is excreted in the urine with a relatively constant rate primarily via glomerular filtration mainly depending on age and gender. The great interest in creatinine in various disciplines led to the development of many analytical methodologies based on different principles. As an organic amine, derivatization of creatinine improves its physicochemical properties so that its analysis becomes feasible by GC also coupled with mass spectrometry (MS) [3–23]. Thus, GC-MS was used several decades ago for the quantitative measurement of creatinine in biological samples including serum and urine using stable isotope-labelled analogs of creatinine [20–22].

Using trimethylsilylation (no conditions reported), Lawson found by GC-MS and EI that creatinine is converted to a single derivative, which was identified as the *N*,*N*,*O*-TMS derivative [20]. The EI mass spectrum of this derivative contained two ions at *m*/*z* 329, ‒

which is the the molecular radical cation [M]•<sup>+</sup> , and *m*/*z* 314 due to the loss of methyl radical ([M-CH3] + ) from one of the three TMS groups [20]. This derivative obviously corresponds to the derivatives of do-creatinine of d3-creatinine in our study with the retention time of 8.9 min. Siekmann extracted creatinine from human serum samples by cation-exchange resin Ag 50W-X2, derivatized by *N*-methyl-*N*-trimethylsilyl-trifluoroacetamide (MSTFA) in anhydrous pyridine (1:1, *v*/*v*) by heating (40 min, 60 ◦C) [21]. Siekmann reported on the formation of a single GC-MS peak (by SIM), of which the EI spectrum was very similar to that reported by Lawson [20], supporting the formation of a *N*,*N*,*O*-TMS derivative of creatinine. Neither Lawson nor Siekmann reported in their papers analogous analyses with creatine.

●

Our observations strongly suggest that derivatization of creatinine (60 min, 60 ◦C) with pure BSTFA, i.e., in the absence of any solvents such as pyridine generates at least three derivatives. The derivative eluting at 8.9 min is most likely *N*<sup>2</sup> ,*N*<sup>2</sup> ,*O*<sup>4</sup> -*tris*(trimethylsilyl) creatinine, identical with that proposed by Lawson [20] and Siekmann [21]. The second major derivative formed under the same derivatization conditions is most likely *N<sup>2</sup>* ,*N<sup>3</sup>* ,*O<sup>4</sup> tris*(trimethylsilyl)-creatinine with the retention time of 8.7 min (Scheme 3). This derivative has not been reported thus far. *N<sup>2</sup>* ,*N<sup>3</sup>* ,*O<sup>4</sup>* -*tris*(Trimethylsilyl)-creatinine elutes in front of *N2* ,*N<sup>2</sup>* ,*O<sup>4</sup>* -*tris*(trimethylsilyl)-creatinine presumably because all derivatizable N atoms of creatinine are derivatized.

Our study also strongly suggests that derivatization of creatine with pure BSTFA under same conditions (60 min, 60 ◦C) generates the same two derivatives eluting at 8.7 min and 8.9 min. The results from HPLC analyses of creatine solutions in deionized water and hydrochloric acid solutions suggest that creatine cyclizes to form creatinine, yet a very low extent. One may therefore assume that the derivatives *N<sup>2</sup>* ,*N<sup>3</sup>* ,*O<sup>4</sup>* -*tris*(trimethylsilyl) and *N2* ,*N<sup>2</sup>* ,*O<sup>4</sup>* -*tris*(trimethylsilyl) are formed during the BSTFA derivatization step.

To the best of our knowledge, the present study is the first to demonstrate the formation of two new derivatives from creatine via BSTFA derivatization (60 min, 60 ◦C). These derivatives emerge from the column 5 to 6 min later than the above-mentioned derivatives of creatinine and creatine. Our study strongly suggests that both late-eluting TMS derivatives stem from a creatine-creatinine adduct. As no creatinine was initially present

in the creatine sample, the detected creatine-creatinine is likely to have been formed by alternative mechanisms. One possible mechanism could involve formation of the *O*-TMS ester of creatine (Scheme 4), which is likely to be formed more rapidly and to a higher extent than the *N*-TMS [24]. Subsequently, free amine groups may attack the chemically activated carboxylic group to make the creatinine residues. Thus far, only one group has reported on the synthesis of creatinyl-amino acids derivatives such as creatinyl-glycine, which has been reported to be neuroprotective [28]. In the NICI and PICI mass spectra of these derivatives we obtained mass fragments being each by 4 Da (see Figure 3). An explanation for this finding could be loss of 4 H atoms in total on the three trimethylsilyl groups of the terminal guanidine group. We do not know whether this results from the derivatization or ionization irrespective of the ionization mode. Such a phenomenon has not been reported thus far. Yet, there is an indication that this may occur in *N*,*N*-di-trimethylsilyl derivatives [29–31]. Thus, in the EI mass spectra of the per-trimethylsilylated 1-phosphono-2-amino-ethane (MW 413) and *O*-phosphorylethanolamine (MW 429) the cation *m*/*z* 174 was observed, which was assigned to [CH2=N(Si(CH3)3)2] + . These spectra also contained *m*/*z* 172 with intensity ratio of 2:1. A possible structure for *m*/*z* 172 could be [CH2=N(Si(CH3)2CH2)2] + .

**Scheme 4.** Proposed chemical structures for the formation of creatinyl-creatinine derivatives from the derivatization of creatine with pure *N*,*O*-*tris*(trimethylsilyl)trifluoroacetamide (BSTFA) at 60 ◦C for 60 min. See the NICI and PICI mass spectra of these derivatives in Figure 3.

Silylation of this compound with a perdeuterated silylation reagent shifted these cations to *m*/*z* 192 and *m*/*z* 188 [30], strongly supporting the bridging of two methyl groups of the neighboring TMS groups on the amine group. BSTFA and other silylation agents can react with various functionalities [23], including acetamide groups such as that of acetaminophen (paracetamol) to generate its *O*,*O*-di-TMS derivative [31], and their derivatives undergo multiple fragmentations and rearrangements during ionization such as EI [32].

Pentafluoropropionic anhydride (PFPA) is another useful derivatization reagent in GC-MS. Like BSTFA, PFPA also reacts with amine, hydroxylic and carboxylic groups for instance of amino acids [33]. The *N*-pentafluoropropionyl derivatives are considerably more stable than the *O*-pentafluoropropionyl derivatives [34]. As BSTFA derivatization does not allow discrimination between creatinine and creatine, we tested the utility of PFPA. Under conditions previously reported for amino acids [25,32], i.e., heating the analytes in PFPA-ethyl acetate (1:4, *v*/*v*; 65 ◦C, 30 min), we observed each only one peak from creatine, d0-creatinine and d3-creatinine. The GC-NICI-MS spectra of creatine and d0-creatinine derivatives were virtually identical: *m*/*z* 221 (6 %; [M−HF−H2O]−), 239 (100 %; [M−HF]−) and *m*/*z* 259 (6 %; [M]−; C7H6F5N3O2); the GC-MS spectrum of the d3-creatinine derivative eluted a few seconds earlier: *m*/*z* 224 (6 %), 242 (100 %) and *m*/*z* 262 (6 %). These results indicate the formation presumably of *N*<sup>2</sup> -pentafluoropropionyl from both, creatine and creatinine. These observations suggest that PFPA reacts with the carboxylic group of creatine to form the mixed anhydride. Subsequently, the *N*<sup>2</sup> -imine group attacks intramolecularly the carboxylic group, with pentafluoropropionic acid leaving the molecule, analogous to the BSTFA derivative of creatine.

As far we are informed, creatinine has not be measured by GC-MS in human urine after derivatization with BSTFA. Our study indicates that creatinine can be quantified precisely and accurately by GC-MS in only 10-µL aliquots of human urine using d3-creatinine as internal standard in relevant concentration ranges. The method does not require any organic solvent or base like pyridine for derivatization and/or extraction for GC-MS analysis. Excess BSTFA serves as a solvent, in which the TMS derivatives are readily soluble, yet no other charged endogenous constituents present in urine. As BSTFA is highly reactive towards numerous substances [23], it is possible that many endogenous substances also form volatile TMS derivatives that do not accumulate in the GC column.

Currently available data suggest that specific measurement of creatinine in urine and other biological samples is possible by using 2,3,4,5,6-pentafluorobenzyl (PFB) bromide (PFB-Br) in aqueous acetone (60 min, 50 ◦C) [34]. Creatinine reacts with PFB-Br to form a single derivative, i.e., *N*<sup>2</sup> -PFB-creatinine. Interestingly, we found that the *N*<sup>2</sup> -PFBderivatives of d0-creatinine and d3-creatinine react with PFPA (65 ◦C, 30 min) to form their *N*2 -PFB,*N*<sup>3</sup> -PFP derivatives (MW=439.21, C14H7F10N3O2; MW=442.23, C14H4D3F10N3O2, respectively) with relative retention time of 1.29 with respect to the *N*<sup>2</sup> -PFB-derivatives.

#### **5. Conclusions**

BSTFA is known for many decades as a useful derivatization reagent for the GC-MS analysis of creatinine, but its utility to measure creatinine in human urine has not been reported thus far. This study investigated the derivatization of creatinine and its precursor creatine with BSTFA. Both substances react with BSTFA (60 ◦C, 60 min) to form three derivatives of virtually identical structures. Creatinine and creatine were found to react with PFPA (65 ◦C, 30 min) to form a single *N*-pentafluoropropionyl derivative. These observations indicate that BSTFA and PFPA is much more effective in the conversion of creatine to creatinine than lowering the pH by inorganic acids such as hydrochloric acid. Our findings suggest that BSTFA and PFPA are not useful for the simultaneous measurement of creatinine and creatine. The *N*,*N*,*O*-trimethylsilyl derivative of creatinine and creatine with the retention time of 8.9 min is useful for their quantitative measurement in human urine both in the NICI and PICI mode using trideuteromethyl creatinine as internal standard. Under the same derivatization conditions, creatine reacts with BSTFA and forms two creatinyl-creatine derivatives with retention times of 14.07 min and 14.72 min, suggesting intermediate formation of creatinine and its conjugation with creatine. Our results show that methyl groups of the TMS residues react to form -CH2-CH2-bridges and are supported by previous reports on alkyl amines. Such a cyclization reaction is more likely to occur during EI, PICI and NICI, rather than during the derivatization with BSTFA. The possibility that creatinine, but not creatine, reacts with PFB-Br and the *N*<sup>2</sup> -PFB-creatinine derivative reacts with PFPA to form *N*<sup>2</sup> -PFB,*N*<sup>3</sup> -PFP-creatinine offers the possibility to

measure biological creatinine and creatine simultaneously by GC-MS. This could be of particular importance in the area of Clinical Chemistry and in clinical trials.

**Author Contributions:** Conceptualization, D.T.; methodology, O.B., K.W. and B.B.; software, O.B. and D.T.; validation, O.B., K.W. and D.T.; formal analysis, D.T.; investigation, O.B., K.W. and B.B.; resources, D.T.; data curation, O.B., K.W., B.B. and D.T.; writing—original draft preparation, D.T., O.B. and K.W.; writing, O.B., K.W. and D.T.; visualization, O.B. and D.T.; supervision, D.T.; project administration, D.T.; funding acquisition, D.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study, due to the use of spot urine samples donated by three volunteers being authors of this article.

**Informed Consent Statement:** Patient consent was waived due to the use of spot urine samples donated by three volunteers being authors of this article.

**Data Availability Statement:** The study did not report any data.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Not available.

#### **References**


#### *Article* **GC-MS Discrimination of Citrulline from Ornithine and Homocitrulline from Lysine by Chemical Derivatization: Evidence of Formation of** *N***<sup>5</sup> -Carboxy-ornithine and** *N***<sup>6</sup> -Carboxy-lysine**

**Svetlana Baskal, Alexander Bollenbach and Dimitrios Tsikas \***

Core Unit Proteomics, Institute of Toxicology, Hannover Medical School, 30625 Hannover, Germany; baskal.svetlana@mh-hannover.de (S.B.); bollenbach.alex@gmail.com (A.B.) **\*** Correspondence: Tsikas.dimitros@mh-hannover.de

**Abstract:** Derivatization of amino acids by 2 M HCl/CH3OH (60 min, 80 ◦C) followed by derivatization of the intermediate methyl esters with pentafluoropropionic anhydride (PFPA) in ethyl acetate (30 min, 65 ◦C) is a useful two-step derivatization procedure (procedure A) for their quantitative measurement in biological samples by gas chromatography-mass spectrometry (GC-MS) as methyl ester pentafluoropropionic (PFP) derivatives, (Me)m-(PFP)n. This procedure allows in situ preparation of trideutero-methyl esters PFP derivatives, (d3Me)m-(PFP)n, from synthetic amino acids and 2 M HCl/CD3OD for use as internal standards. However, procedure A converts citrulline (Cit) to ornithine (Orn) and homocitrulline (hCit) to lysine (Lys) due to the instability of their carbamide groups under the acidic conditions of the esterification step. In the present study, we investigated whether reversing the order of the two-step derivatization may allow discrimination and simultaneous analysis of these amino acids. Pentafluoropropionylation (30 min, 65 ◦C) and subsequent methyl esterification (30 min, 80 ◦C), i.e., procedure B, of Cit resulted in the formation of six open and cyclic reaction products. The most abundant product is likely to be *N*<sup>5</sup> -Carboxy-Orn. The second most abundant product was confirmed to be Orn. The most abundant reaction product of hCit was confirmed to be Lys, with the minor reaction product likely being *N*<sup>6</sup> -Carboxy-Lys. Mechanisms are proposed for the formation of the reaction products of Cit and hCit via procedure B. It is assumed that at the first derivatization step, amino acids form (*N*,*O*)-PFP derivatives including mixed anhydrides. At the second derivatization step, the Cit-(PFP)<sup>4</sup> and hCit-(PFP)<sup>4</sup> are esterified on their *C* 1 -Carboxylic groups and on their activated *N*ureido groups. Procedure B also allows in situ preparation of (d3Me)m-(PFP)<sup>n</sup> from synthetic amino acids for use as internal standards. It is demonstrated that the derivatization procedure B enables discrimination between Cit and Orn, and between hCit and Lys. The utility of procedure B to measure simultaneously these amino acids in biological samples such as plasma and urine remains to be demonstrated. Further work is required to optimize the derivatization conditions of procedure B for biological amino acids.

**Keywords:** amino acids; derivatization; esterification; GC-MS; pentafluoropropionic anhydride; ureide

#### **1. Introduction**

Analysis of amino acids, dipeptides, and tripeptides such as glutathione by gas chromatography-mass spectrometry (GC-MS) requires suitable derivatization reactions to convert them into volatile and thermally stable derivatives [1–9]. Derivatization of amino acids with 2 M HCl in methanol (CH3OH) (60 min, 80 ◦C) yields their mono- and di-methyl esters. Subsequent reaction with pentafluoropropionic anhydride (PFPA) in ethyl acetate (30 min, 65 ◦C) generates the *N*- and *O*-pentafluoropropionyl (PFP) derivatives. The methyl ester (Me) PFP derivatives ((Me)m-(PFP)n) obtained by this procedure (here designated as procedure A) are useful for the quantitative measurement of biological amino acids

**Citation:** Baskal, S.; Bollenbach, A.; Tsikas, D. GC-MS Discrimination of Citrulline from Ornithine and Homocitrulline from Lysine by Chemical Derivatization: Evidence of Formation of *N*<sup>5</sup> -Carboxy-ornithine and *N*<sup>6</sup> -Carboxy-lysine. *Molecules* **2021**, *26*, 2301. https://doi.org/ 10.3390/molecules26082301

Academic Editor: Paraskevas D. Tzanavaras

Received: 24 March 2021 Accepted: 13 April 2021 Published: 15 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

by GC-MS [7]. However, the carbamoyl-amino acids citrulline (Cit) and homocitrulline (hCit) (Figure 1) are converted under these reaction conditions into the methyl esters of ornithine (Orn) and lysine (Lys), respectively [7]. Analogously, glutamine (Gln) and asparagine (Asn) are converted into glutamate (Glu) and aspartate (Asp), respectively [7]. For not yet fully understood reasons, the derivatization procedure A was found to be not useful for the GC-MS analysis of *N*<sup>G</sup> ,*N*′G-dimethylarginine (symmetric dimethylarginine, SDMA), in contrast to its structural isomer *N*<sup>G</sup> ,*N*G-dimethylarginine (asymmetric dimethylarginine, ADMA) and to their precursor arginine. This difficulty was in part overcome by using a single derivatization reaction with PFPA, which most likely generates the *tetrakis*(pentafluoropropionyl) derivative of SDMA, i.e., SDMA-(PFP)<sup>4</sup> [8]. This derivatization reaction, i.e., (*N*,*O*)-pentafluoroprionylation, enables quantitative measurement of SDMA in human urine, but requires the use of commercially available stable-isotope labelled SDMA analogue such as [*N*<sup>G</sup> ,*N*′<sup>G</sup> - <sup>2</sup>H6]dimethylarginine [8] and is less sensitive compared to the GC-MS analysis of ADMA as Me-PFP derivative. Interestingly, the tripeptides glutathione and its analogue ophthalmic acid were also found to react with PFPA under the same derivatization conditions, which enabled their GC-MS analysis [9].

**Figure 1.** Upper panel, procedure (**A**) Schematic of the reactions of citrulline and homocitrulline with 2 M HCl/MeOH forming the methyl esters of ornithine and lysine, respectively. Lower panel, procedure (**B**) Schematic of the two-step derivatization of citrulline (left) and homocitrulline (right) first with PFPA/EA to form their PFP derivatives with the proposed formulas Cit-(PFP)<sup>4</sup> and hCit-(PFP)<sup>4</sup> , respectively. Subsequently, these derivatives react with 2 M HCl/MeOH (procedure B) to form reaction products that were characterized structurally by GC-MS in the present study. Cit, citrulline; hCit, homocitrulline; MeOH, methanol; PFPA, pentafluoropropionic anhydride; PFP, pentafluoropropionyl residue; EA, ethyl acetate.

The aim of the present study was to find derivatization conditions that would allow discrimination of Cit from Orn, and of hCit from Lys. Our previous observations that SDMA can be measured in human urine by GC-MS by using PFPA/EA as the first derivatization step [8] prompt us to investigate whether the derivatization of Cit and hCit with PFPA/EA as the first step may also be useful for their GC-MS analysis and for their discrimination from Orn and Lys, respectively. Analogous to SDMA, we assumed intermediate formation of Cit-(PFP)<sup>4</sup> and hCit-(PFP)<sup>4</sup> (Figure 1). In order to investigate potential reactions of the putative intermediates, we coupled the PFPA/EA derivatization with the classical esterification with 2 M HCl/CH3OH and with 2 M HCl/CD3OD to prepare stable-isotopelabelled analogs of Cit and hCit. De facto, this resulted in a reversed order of the original two-step derivatization procedure A, which is specified as procedure B in the present work (Figure 1). In most investigations using derivatization procedure B, we used experimental conditions previously found to be optimum for the derivatization and GC-MS analysis of amino acids and the tripeptides glutathione and ophthalmic acid [7–9].

#### **2. Materials and Methods**

#### *2.1. Chemicals, Materials and Reagents*

All amino acids (chemical purity, 95 to 98%) were obtained from Sigma-Aldrich. Tetradeuterated methanol (CD3OD, 99% at <sup>2</sup>H) and pentafluoropropionic anhydride were supplied by Aldrich (Steinheim, Germany). Methanol was obtained from Chemsolute (Renningen, Germany). Hydrochloric acid (37 wt%) was purchased from Baker (Deventer, The Netherlands). Ethyl acetate was obtained from Merck (Darmstadt, Germany). Glassware for GC-MS (1.5 mL autosampler glass vials and 0.2 mL microvials) and the fused-silica capillary column Optima 17 (15 M × 0.25 mm I.D., 0.25 µM film thickness) were purchased from Macherey–Nagel (Düren, Germany). Separate stock solutions of amino acids were prepared by dissolving accurately weighed amounts of commercially available amino acids in deionized water. Stock solutions were diluted with deionized water as appropriate.

For the preparation of unlabelled methyl esters and deuterium-labelled methyl esters of amino acids, two derivatization reagents were prepared. To 80 mL ice-cold CH3OH were added 16 mL of 37 wt% HCl slowly under gentle mixing. Analogously, to 80 mL ice-cold CD3OD, 16 mL of 37 wt% HCl were added slowly under gentle mixing. The concentration of HCl in these methanolic solutions was each 2 M. In the present article, these solutions are denoted as 2 M HCl/CH3OH and 2 M HCl/CD3OD, respectively. The PFPA-ethyl acetate reagent (PFPA/EA) was prepared daily by diluting pure PFPA in ethyl acetate (EA) (1:4, *v*/*v*).

#### *2.2. Derivatization Procedures A and B for Amino Acids and Generation of GC-MS Spectra*

*Procedure A.* Solid amino acids were derivatized first with 2 M HCl/CH3OH or 2 M HCl/CD3OD and then with PFPA/EA in autosampler glass vials. Briefly, residues were reconstituted in 100 µL aliquots of a 2 M HCl/CH3OH or 2 M HCl/CD3OD solution and the glass vials were tightly sealed. Esterification was performed by heating the samples for 60 min at 80 ◦C. After cooling the samples of the esterification reaction to room temperature, solvents and reagents were evaporated to dryness under a stream of nitrogen. Aliquots (100 µL) of the PFPA/EA solution were added, and the glass vials were tightly sealed and heated for 30 min at 65 ◦C to prepare *N*-pentafluoropropionic amides of the methyl esters. Then, residues were treated first with 200 µL aliquots of 400 mM borate buffer, pH 8.5, and immediately thereafter with 200 µL aliquots of toluene, followed by immediate vortex-mixing for 60 s and centrifugation (4000× *g*, 5 min, 18 ◦C). Aliquots (150 µL) of the upper organic phase were transferred into autosampler glass vials equipped with microinserts, and the samples were sealed and subjected to GC-MS analysis.

*Procedure B.* Solid amino acids were derivatized first with PFPA/EA (30 min, 65 ◦C) and then with 2 M HCl/CH3OH or 2 M HCl/CD3OD (30 min, 80 ◦C). Briefly, aliquots (100 µL) of a freshly prepared PFPA/EA solution were added, the glass vials were tightly sealed and heated for 30 min at 65 ◦C to prepare *N*-pentafluoropropionic amides of the

methyl esters. After cooling the samples to room temperature, solvents and reagents were evaporated to dryness under a stream of nitrogen. Then, residues were reconstituted in 100 µL aliquots of a 2 M HCl/CH3OH or 2 M HCl/CD3OD solution and the glass vials were tightly sealed. Esterification was performed by heating the samples for 30 min at 80 ◦C. After cooling to room temperature, solvents and reagents were evaporated to dryness under a stream of nitrogen. Residues were treated directly with toluene (200 µL), shortly vortex-mixed, aliquots (150 µL) of the upper organic phase were transferred into autosampler glass vials equipped with microinserts, and the samples were sealed and subjected to GC-MS analysis.

#### *2.3. Generation of GC-MS Spectra*

GC-MS spectra were obtained using negative-ion chemical ionization (NICI) after separate derivatization of 5 nmol of each amino acid using both derivatization procedures as described above. The derivatives were extracted with toluene (1 mL), 1 µL aliquots containing 5 pmol of each analyte (assuming quantitative yield) were injected in the splitless mode, and mass spectra were generated in the scan mode in the mass-to-charge (*m/z*) range 50 to 650 (1 s per scan). The GC-MS software Xcalibur and Quan Browser were used. ChemDrawProfessional 15.0 was used to draw chemical structures and to convert structures into names. GraphPad Prism 7.0 (San Diego, CA, USA) was used in statistical analyses and to prepare graphs.

#### *2.4. GC-MS Conditions*

All analyses were performed on a GC-MS apparatus consisting of a single quadrupole mass spectrometer model ISQ, a Trace 1210 series gas chromatograph, and an AS1310 autosampler from ThermoFisher (Dreieich, Germany). The injector temperature was kept at 280 ◦C. Helium was used as the carrier gas at a constant flow rate of 1.0 mL/min. The oven temperature was held at 40 ◦C for 0.5 min and ramped to 210 ◦C at a rate of 15 ◦C/min and then to 320 ◦C at a rate 35 ◦C/min. Interface and ion-source temperatures were set to 300 ◦C and 250 ◦C, respectively. Electron energy was 70 eV and electron current 50 µA. Methane was used as the reagent gas for NICI at a constant flow rate of 2.4 mL/min. In quantitative analyses, the dwell time was 100 ms for each ion in the selected-ion monitoring (SIM) mode and the electron multiplier voltage was set to 1400 V.

#### **3. Results**

#### *3.1. Derivatization of Citrulline and Structural Characterization of Its Reaction Products by GC-MS*

Scanning of the Cit samples derivatized by procedure B resulted in the elution each of six GC-MS peaks using CH3OH (Supplementary Materials Figure S1A) and CD3OD (Figure S1B). In the latter case, the peaks I, II, V, and VI eluted a few seconds in front of the peaks of the Cit sample derivatized with CH3OH, indicating the presence of deuterium atoms in these peaks [7] (see Table 1). The almost identical retention times of the minor peaks III (retention time, 9.36 min) and peaks IV (retention time, 9.67 min) suggest that they are not methyl esters, but rather cyclic compounds.

The mass spectra of the peaks I (retention time, 8.36 min, 8.33 min) contained four corresponding ions that differed by 3 Da each, suggesting the presence of a single methylated carboxylic group (Figure S1(A1,B1)) (Table 1). A tentative structure of this molecule could be (*S*)-3-amino-2-oxopiperidine-1-Carboxylic acid (non-derivatized).



<sup>a</sup> Bold numbers indicate mass fragments with the highest intensity in the mass spectrum (i.e., base peaks).

Derivatization of Cit by procedure B resulted in the formation of the peaks II (retention time, 8.67 min, 8.63 min) (Supplementary Materials: Figure S1(A1,B1)) (Table 1). Peaks II had virtually the same mass spectra as the unlabelled Me-PFP (d0Me-PFP) and the labelled Me-PFP (d3Me-PFP) derivatives of Orn (Figure S1(C1,C2)), indicating conversion of Cit to Orn by both procedures as observed previously using procedure A [7].

The mass spectra in combination with the retention times of the peaks III and the peaks IV suggest that the peak III corresponds to (*S*)-3-amino-4,5-dihydropyridin-2(3*H*)-one (Figure S1(A3,B3)) and peak IV corresponds to (*S*)-3-aminopiperidin-2-one (Figure S1(A4,B4)) (Table 1).

The mass spectra of the minor peaks V (retention time, 10.57 min, 10.52 min) contained corresponding ions that did not differ (*m/z* 162) or did differ by 3 Da each (*m/z*, 301/298; *m/z*, 489/486) suggesting the presence of an intact methylated carboxylic group and presumably a fragmented methyl ester (Figure S1(A5,B5)) (Table 1). A tentative structure of this molecule could be (*S*)-2-amino-5-(Carboxyamino)pentanoic acid, which could be trivially named *N*<sup>5</sup> -Carboxy-ornithine.

The most intense GC-MS peaks of Cit derivatized by procedure B were the peaks VI, which eluted at 10.75 min (using CH3OH) and 10.71 min (using CD3OD) (Figure S1(A6,B6)). The GC-MS spectra of these peaks contained several corresponding mass fragments that differed by 3 Da (*m/z* 298/301) or 6 Da (*m/z* 296/290, *m/z* 336/330, *m/z* 355/349) suggesting the presence of two carboxylic groups in these ions (Table 1).

#### *3.2. Effects of the PFPA/EA Derivatization Time in Procedure B on the Reaction Products*

The derivatization conditions used in procedure A in the present study were found to be optimal in previous studies [7,8]. In this experiment, we investigated the effect of the derivatization time of the pentafluoropropionylation reaction of Cit in procedure B. For this, two sets of 10 µL aliquots of Cit samples in distilled water (50, 100, 150, 200, 250 µM) were derivatized first with PFPA/EA at 65 ◦C for 30 min and subsequently with 2 M HCl/CH3OH for 10, 20, 30, 40, and 60 min at 80 ◦C. A 100 µM Cit sample in distilled water served as internal standard and was derivatized in parallel under the same conditions using 2 M HCl/CD3OD. After toluene extraction, GC-MS analysis was performed by SIM of *m/z* 298 and *m/z* 301 for peak I, *m/z* 418 and *m/z* 421 for peak II (i.e., Orn), *m/z* 298 and *m/z* 301 for peak V, and *m/z* 330 and *m/z* 336 for peak VI (see Table 1).

The results of this experiment are illustrated in Figures 2–4. The peak area of the internal standards varied between 11 and 16% (*m/z* 301, Peak I), between 7 and 14% (*m/z* 421,

Peak II), between 10 and 21% (*m/z* 301, Peak V), and between 13 and 17% (*m/z* 336, Peak VI). There were significant time effects for all monitored internal standards (*P* < 0.0001, two-way ANOVA) and a Cit concentration effect for Peak II (*P* = 0.043, two-way ANOVA). With the exception of the Peak I, the peak areas of the Peaks II, V, and VI had a minimum at the esterification time of 30 min. When combining all data of the incubation times of the esterification, the peak area ratios (*y*) of all peaks depended linearly upon the Cit concentration (in µM) (*x*): *y* = −0.223 + 0.011 *x*, *r* <sup>2</sup> = 0.9943 for Peak I, *y* = 0.241 + 0.0098 *x*, *r* <sup>2</sup> = 0.9712 for Peak II, *y* = −0.469 + 0.012 *x*, *r* <sup>2</sup> = 0.9944 for Peak V, and *y* = −0.415 +0.013 *x*, *r* <sup>2</sup> = 0.9923 for Peak VI. Linear relationships were observed between the peak area ratio (PAR) values (*y*) of the individual peaks and the derivatized Cit concentration (*x*) resulted in straight lines for all derivatization times of the esterification step (Figure 4). Linear regression analysis of the mean PAR of all peaks vs. The concentrations of derivatized Cit resulted in the regression equation *y* = −0.217 + 0.0117 *x*, *r* <sup>2</sup> = 0.9973. The reciprocal of the slope value of this regression equation indicates a mean concentration of 85.3 µM for the sum of internal standards (nominal concentration, 100 µM).

Taken together, these results demonstrate the principle applicability of the procedure B for the quantitative GC-MS analysis of Cit in aqueous solutions.

**Figure 2.** Time profiles of the peak areas of the internal standards (generated from 100 µM Cit) upon derivatization of aqueous Cit (0, 50, 100, 150, 200, 250 µM; see insert) using procedure B, i.e., first with PFPA/EA for a fixed time of 30 min at 65 ◦C and subsequently with 2 M HCl/CH3OH (2 M HCl/CD3OD for the internal standards) at 80 ◦C for the indicated times. (**A**) Peak I; (**B**) Peak II; (**C**) Peak V; (**D**) Peak VI. See also Table 1 and Figure 3.

**Figure 3.** Partial GC-MS chromatograms from the analysis of an aqueous citrulline sample (250 µM) derivatized by procedure B, i.e., first with PFPA/EA (30 min, 65 ◦C) followed with 2 M HCl in CH3OH or CD3OD (60 min, 80 ◦C). Selected ion monitoring (SIM, 100 ms) of (**A1**,**A2**) *m/z* 298 and *m/z* 301 for Peak I, *m/z* 418 and *m/z* 421 for Peak II, and *m/z* 330 and *m/z* 336 for Peak VI was performed. (**B1**,**B2**) SIM (100 ms) of *m/z* 298 and *m/z* 301 for Peak V. See also Table 1 and Figure 2.

**Figure 4.** Linear regression analysis between the peak area ratio (PAR) values (*y*) for (**A**) Peak I, (**B**) Peak II, (**C**) Peak V, and (**D**) Peak VI to the corresponding internal standards (generated from 100 µM Cit) and the Cit concentration (*x*) upon derivatization of aqueous Cit (0, 50, 100, 150, 200, 250 µM) first with PFPA/EA for a fixed time of 30 min at 65 ◦C and subsequently with 2 M HCl/CH3OH (2 M HCl/CD3OD for the internal standard) at 80 ◦C for 10 (green circles), 20 (blue squares), 30 (black upper triangles), 40 (black lower triangles), 60 (red diamonds) min. Insets indicate the regression equations. Note that Peak II corresponds to the Orn derivative. See also Table 1 and Figure 3.

#### *3.3. Derivatization of Homocitrulline and Structural Characterization of Its Reaction Products by GC-MS*

Scanning of the hCit samples derivatized by procedure B resulted in the elution each of an intense GC-MS peak using CH3OH (Figure S1D) and CD3OD (Figure S1E) and two minor peaks (Table 2).

**Table 2.** GC-MS retention times (*t*R, min) and most intense ions in the mass spectra of the three reaction products of homocitrulline derivatized with procedure B (PFPA/EA then 2 M HCl/CH3OH or 2 M HCl/CD3OD). For comparison, synthetic lysine standard (Lys-Std) was also derivatized with procedure B. See also Figure S1.


<sup>a</sup> Bold numbers indicate mass fragments with the highest intensity in the mass spectrum (i.e., base peaks).

The major GC-MS peaks eluted at 9.54 min and 9.52 min, respectively. The mass spectra of these peaks are very similar to those obtained from the derivatization of hCit by procedure B, as well as to those of the d0Me-PFP and d3Me-PFP derivatives of synthetic Lys confirming previous observations of the conversion of hCit to Lys [7] (Figure 1). The minor GC-MS peaks eluting at 9.21 min and 9.18 min could correspond to (*S*)-3-amino-2-oxoazepane-1-Carboxylic acid (Figure S1(D1,E1)). The minor GC-MS peaks eluting at 11.47 min and 11.42 min could correspond to *N*<sup>6</sup> -Carboxy-lysine (Figure S1(D3,E3)), analogous to *N*<sup>5</sup> -Carboxy-ornithine obtained from Cit using procedure B. The small differences in the retention times is indicative of the presence of deuterium atoms in the earlier eluting peaks.

#### **4. Discussion**

Procedure A allows for the reliable quantitative determination of amino acids and their metabolites in biological samples by GC-MS [7,10]. During the first esterification step, however, Cit and hCit undergo almost complete conversion to the methyl esters of Orn and Lys, respectively. The same happens to Gln and Asn, which are converted to the methyl esters of Glu and Asp, respectively [7]. These observations strongly indicate that the carbamide groups of Cit, hCit, Gln, and Asn are labile under the strong esterification conditions. This circumstance prevents simultaneous measurement of Cit, Orn, hCit, Lys, Gln, Glu, Asn, and Asp [7]. We have hypothesized that reversing the order of the derivatization procedure A may present a way to prevent the abovementioned conversions. In the present study, we investigated this possibility for Cit and hCit using procedure B, i.e., first pentafluoropropionylation and subsequently esterification, using previously optimized derivatization conditions [7]. Cit and hCit reacted to form five and three reaction products, respectively. The tentative chemical structures of these reaction products are illustrated in Figure 5.

**Figure 5.** Proposed structures of the reaction products of citrulline (left panel) and homocitrulline (right panel) using procedure B (first PFPA/EA then 2 M HCl/MeOH). The red-marked structures were not found to be derivatization products of homocitrulline. The red numbers between the structures are the relative retention times of the homocitrulline products to the corresponding citrulline reaction products.

One major reaction product of Cit was identified as Orn. This observation suggests that pentafluoropropionylation prevents conversion of Cit to Orn, albeit not entirely. The major reaction product of hCit was identified as Lys. The reaction products of hCit corresponding to the Cit-derived peaks III, IV, and V were not observed (Figure 5). These observations suggest that pentafluoropropionylation prevents conversion of hCit to Lys to only a minor extent. The conversion of Cit to four reaction products in addition to Orn suggest that pentafluoropropionylation of Cit enables additional reactions during the second reaction step of procedure B. The different reaction behaviour of Cit and hCit could be due to the longer side chain of these homologue amino acids: 3 vs. 4 CH<sup>2</sup> groups. It is assumed that this structural difference plays a major role in the formation of cyclic reaction products (Figure 5). Interestingly, procedure B resulted in the formation of *N*<sup>5</sup> -Carboxy-Orn from Cit as a major reaction product and *N*<sup>6</sup> -Carboxy-Lys from hCit as a minor reaction product. Because of the commercially unavailability of synthetic standards of *N*<sup>5</sup> -Carboxy-*L*-Orn and *N*6 -Carboxy-*L*-Lys, we were not able to unequivocally demonstrate the formation of these reaction products. Nevertheless, these putative reaction products enable discrimination of Cit from Orn, and of hCit from Lys, respectively. It is interesting to note that the physiological occurrence and the biological significance of the free amino acids *N*<sup>5</sup> -Carboxy-Orn and *N*<sup>6</sup> -Carboxy-Lys (Chemical Entities of Biological Interest (ChEBI):43575) have not been reported thus far. However, a *N*<sup>6</sup> -Carboxy-Lys residue was found to be present

in the active site of class D β-lactamases and to play a significant role in the hydrolysis of β-lactam antibiotics [11,12]. Our study provides useful information for forthcoming studies on these uncommon amino acids.

Based on the results of our study, we propose potential mechanisms that may explain the reaction products of Cit and hCit during the derivatization procedure B. Being a highly reactive derivatization reagent, PFPA is likely to react with all functional groups of free amino acids and those in tripeptides [8,9]. We therefore assume that PFPA/EA reacts with all functional groups of Cit to form its *N*,*N*,*N*,*O*-(PFP)<sup>4</sup> derivative (Figure 6). An intact Cit-(PFP)<sup>4</sup> derivative was not observed in our study. An explanation could be that the remaining Cit-(PFP)<sup>4</sup> extracted into toluene decomposed during the injection in the hot injector (280 ◦C). This is more likely to happen to the *O*-PFP residue, as *N*-PFP residues of derivatized amino acids are considerably stable [7]. A more plausible explanation for our observations is that the *O*-PFP residue of the Cit derivative is a mixed anhydride of PFPA and the carboxylic group Cit. As such, the Cit-(PFP)<sup>4</sup> derivative is likely to undergo several reactions with 2 M HCl/CH3OH (Figure 6). The reaction of the Cit- (PFP)<sup>4</sup> derivative with 2 M HCl/CH3OH will always generate its *C* 1 -Carboxy-methyl ester. Analogously, the reaction of the Cit-(PFP)<sup>4</sup> derivative with 2 M HCl/CD3OD will generate the *C* 1 -Carboxy-trideutero-methyl ester. This provides a way to prepare deuterium-labelled internal standards for quantitative analyses. Especially the *N*-PFP residue on the carbamide functionality of the Cit-(PFP)<sup>4</sup> derivative opens ways for additional reactions, which leads to the formation of open reaction products including *N*<sup>5</sup> -Carboxy-Orn from Cit and *N*<sup>6</sup> - Carboxy-Lys from hCit and several cyclic reaction products that can be utilized both in analytical and organic preparative chemistry (Figure 6).

The reaction time of the esterification reaction performed at 80 ◦C has an effect on the yield of individual reaction products. In a proof-of-principle experiment, we found that procedure B is useful for the quantitative analysis of Cit in aqueous solution for several esterification times. Yet, the quantitative determination of Cit, Orn, hCit, and Lys in biological samples by GC-MS using procedure B remains to be optimized and validated. Our preliminary studies suggest that the derivatization procedure B can be extended to Gln and Asn, which are converted into Glu and Asp, respectively. The derivatization procedure B possess the potential to simultaneously quantitate a large number of biological amino acids and their metabolites by GC-MS using in situ prepared (d3Me)m-(PFP)<sup>n</sup> or commercially available stable-isotope labelled amino acids as internal standards.

**Figure 6.** Proposed reaction products of the very first, not yet identified *N*,*N*,*N*,*O*-(PFP)<sup>4</sup> derivative of Cit using procedure B. Blue and red arrows indicate the carbonyl moieties, which are attacked by methanol, and the green arrow indicates the intramolecular attack of *N*<sup>5</sup> on *C* 1 carbonyl group.

**Supplementary Materials:** The following are available online, Figure S1: Separate derivatization of citrulline and homocitrulline (5 nmol each) by procedure B, i.e., first with PFPA/EA and then 2 M HCl/CH3OH or 2 M HCl/CD3OD, and structural characterization of their reaction products by GC-MS.

**Author Contributions:** Conceptualization, D.T.; methodology, S.B.; software, S.B. and D.T.; validation, S.B., A.B. and D.T.; formal analysis, D.T.; investigation, S.B. and A.B.; resources, D.T.; data curation, S.B. and D.T. writing—original draft preparation, D.T., S.B. and A.B.; writing, S.B., A.B. and D.T.; visualization, S.B. and D.T.; supervision, D.T.; project administration, D.T.; funding acquisition, D.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study, due to the use of human urine samples originally collected in previously ethically approved study.

**Informed Consent Statement:** Subject consent was waived due to the use of human urine samples originally collected in previously ethically approved study.

**Data Availability Statement:** The study did not report any data.

**Acknowledgments:** We thank Bibiana Beckmann for administrative assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Samples of the compounds are not available from the authors.

#### **References**

