Evaluation and optimization of the derivatization reaction conditions of glyphosate and aminomethylphosphonic acid with 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate using reversed-phase liquid chromatography

Abstract


Introduction
Glyphosate (N-phosphonomethyl-glycine; GLP) is a broad-spectrum systemic herbicide widely used as a crop desiccant.It is one of the most commonly used herbicides in agriculture due to the development of glyphosate-resistant genetically modified crop varieties [1,2] and it is also widely applied for urban and residential weed control [3].
The global glyphosate market is projected to reach USD 12.54 billion by 2024, with a compound annual growth rate of 6.8% over the forecast period, from 2019 to 2024 [4].
GLP is considered to be non-persistent in the environment as it degrades by microbial organisms in water and soil to form the primary metabolite product, aminomethylphosphonic acid (AMPA).However, due to its intensive use worldwide, its impact on the environment is increasingly significant and it is regularly found in surface waters.The half-life time of GLP depends upon the environmental conditions (e.g., temperature, water depth, macrophytes, and sediment ratio) and can range between 2 and 215 days in soils and between 2 and 90 days in waters [3,5,6].Both compounds are usually analyzed together as it is uncommon to detect GLP without the presence of AMPA [3].
Another point of interest is the toxicity and carcinogenicity of GLP.Since its introduction, most regulatory assessments have established that GLP has a relatively low toxicity in mammals [7].However, in March 2015, the International Agency for Research on Cancer (IARC) listed glyphosate as "probably carcinogenic to humans" (group 2A), a category that "is used when there is limited evidence of carcinogenicity in humans and sufficient evidence of carcinogenicity in experimental animals" [8].
Therefore, GLP carcinogenicity is controversial.One review evaluating fourteen carcinogenicity studies in rats and mice reached the conclusion that glyphosate is not of concern with regards to its carcinogenic potential in humans [9].After the IARC evaluation, the European Union (EU) conducted a detailed assessment of all available information and concluded that glyphosate is unlikely to pose a carcinogenic hazard to humans and the evidence does not support classification with regard to its carcinogenic potential according to Regulation No 1272/2008 [10].Therefore, GLP is currently approved in the EU and can be used as an active substance in plant protection products until December 2022.This controversy has raised a social debate that has increased the interest in the analysis of GLP and AMPA in environmental samples.
There are different measurement methods for the determination of GLP and AMPA, which have been recently reviewed by Valle et al. [11].Due to the polar and ionic characteristics of GLP and AMPA, liquid chromatographic techniques with derivatization of the target compounds are the most suitable.Another review found that 99% of the liquid chromatographic methods used for the determination of these compounds were performed with derivatized species [12].The two most common derivatization reagents used are 9-fluorenylmethylchloroformate (FMOC) and o-phatalaldehyde (OPA), with around 75% of the applications using one of these two derivatization reagents [12].
The derivatization reaction with OPA is fast (<1 min) but the derivatives are unstable after a few minutes.Therefore, this reagent is usually applied as a post-column derivatization, after chromatographic separation of the target compounds with a strong cation-exchange column [13,14].OPA presents other significant limitations since it is sensitive to air oxidation, degrades over time, and should be prepared fresh for optimum sensitivity or stored under an inert gas to maintain its activity for one week [14,15].Moreover, this reagent only reacts with primary amines and GLP is a secondary amine and so an intermediate step must be applied before derivatization using an oxidizing solution (e.g., hypochlorite solution) to oxidize GLP to glycine [13,14].
Since the study of Moye and Boning [16], FMOC has become the most common derivatization reagent for HPLC determination of GLP and AMPA [12,17,18].This derivatization reaction is slower than with OPA and different reaction times have been proposed, ranging from 10 min [19] to overnight [20].Pinto et al. [21] performed a twolevel factorial experimental design to assess the significance of different variables during the derivatization reaction and found that the variable derivatization time did not have a significant effect, suggesting a reaction time of 10 min.However, other studies found optimum reaction times of 30 min [18,22,23] and 60 min [24].Once the pre-fixed derivatization time is achieved, the solution must be acidified to stop the reaction and to obtain stable derivatives for some days [20][21][22][25][26][27].Due to the stability of FMOC derivatives, this reagent is applied as a pre-column derivatization reaction.Another challenge with the use of FMOC is the formation of derivatization by-products, such as FMOC-OH [18,22,24,25], which are formed by hydrolysis and decarboxylation of the excess of reagent.These by-products have strong fluorescence at the same excitation and emission wavelengths of the derivatives, which may complicate the chromatographic determination.For this reason, and to decrease matrix effects, the removal of interferents, by liquid-liquid extraction [17,28] or solid phase extraction [18], are recommended to maintain a good specificity of the method.
Recently, another derivatization reagent usually applied for amino acid determination has been proposed as a rapid and simple alternative to conventional OPA and FMOC derivatization procedures [29,30].In this case, 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) was applied as the derivatizing reagent with very promising results (Figure 1).This reagent reacts with both primary and secondary amines, usually in a simple and fast way, yielding stable and fluorescent derivatives.It has the advantage that the excess of reagent is rapidly hydrolyzed to 6-aminoquinoline (AMQ), which does not need to be removed before analysis given that AMQ has a different emission spectrum to the derivatized amines, allowing for selective detection in the presence of a large excess of AMQ [31].Unfortunately, the first study evaluating AQC for GLP and AMPA derivatization did not study the derivatization reaction and the authors directly applied the commercial indications of the supplier of the reagent for the analysis of amino acids [29].Given this, we have performed the optimization of the derivatization reaction conditions with AQC using a two-factorial experimental design, combined with liquid chromatography with ultraviolet detection for the separation and detection of the analytes.

Reagents and Solutions
GLP and AMPA (analytical-grade) were purchased from Sigma-Aldrich (Germany).
Stock solutions containing either one or the two analytes at 200 mg•L -1 in 0.1 M HCl were prepared.Acidification with HCl to pH 1 has been found to be necessary in the case of environmental samples since multivalent cations form stable complexes with GLP and AMPA, which are not derivatized [11,20,[25][26][27]32,33]. All standard and working solutions were stored at 4ºC in polypropylene material given that GLP binds to active sites on glass when it is not derivatized [22,24,27].Working solutions were freshly prepared before use.
The AQC derivatization reagent (AccQ•Tag TM derivatization kit) was purchased from Waters Corporation (USA).Methanol for HPLC was from Fisher Chemical (Fisher Scientific, UK).Deionized water was from a Milli•Q Ultrapure water system (Millipore Iberica, Spain).Ammonium acetate for HPLC LiChropur was obtained from Merck (Germany).Trifluoroacetic acid (TFA, 99%) was from Fluka (Germany).
Mobile phases were a mixture of (A) an aqueous buffer solution containing 50 mM ammonium acetate, adding TFA until pH=5.0, and (B) methanol.All mobile phase solutions were filtered through 0.45 μm filters (Whatman, Germany) and degassed before use.

Instrumentation
The chromatographic experiments were performed using two HPLC systems: a SpectraSYSTEM (Thermo Scientific, USA) liquid chromatograph and an 1120 Infinity LC Compact system (Agilent Technologies).UV detection was performed at 254 nm.
After assessing different HPLC separation conditions, the following gradient was chosen for this study: 10% B for 3 min, 10-50% B in 8 min, 50-100% B in 3 min, 100% B for 4 min, 100-10% B in 3 min and re-equilibration for 4 minutes.Injection volume was 10 μl and flow rate was set at 1.0 ml•min -1 .Oven column temperature was maintained at 40ºC.Thereafter, the derivatization time and temperature were modified and assessed to find the optimum derivatization conditions for GLP and AMPA.For this reason, a two-level full factorial design (2 k , where k denotes the number of factors) was applied to evaluate the effect of these two variables.This factorial design permits the study of the effect of each variable and the presence of interactions between the variables.The reaction time (variable t) was studied in the range from 0.5 to 20 min, and the temperature (variable T) was evaluated in the range from room temperature (24ºC, as the laboratory was thermostated at this temperature) to 55ºC.The central point (axial point, 10 min, 40ºC) was also measured and considered as an experiment.All experimental points were performed in triplicate to assess the precision, and all the experiments were performed in a random order.Significance was set at 0.05.

Quality control
Method blank controls were prepared following the same derivatization procedure as for samples, using HCl 0.1 M as the blank matrix.Duplicate standards were used as control samples to confirm the stability of the results obtained.
Before proceeding with the HPLC procedure, a standard was analyzed sequentially until stable retention times were obtained for GLP and AMPA (<0.1 min variation in retention times in three consecutive injections).Once the stability of the system was obtained, a method blank was injected to confirm that no system contamination took place.After each 5-6 samples, a method blank and a quality control were analyzed to confirm the stability of the system.

Validation study
The linearity of the derivatization method was assessed by analyzing six standards in the 1-150 mg•l -1 range.Repeatability and inter-day precision were determined applying a nested design [35,36], evaluating three replicates daily at a low concentration (10 mg•l -1 ) and three at a high concentration (150 mg•l -1 ) on three consecutive days.The limit of detection (LOD) was determined by analyzing six independent replicates of blanks spiked at 5-10 mg•l -1 , taking the standard deviation (SD) as SD blank .The 3.3SD blank criterion was then applied [37].

Evaluation of the HPLC conditions
For the assessment of the liquid chromatographic conditions, derivatization was performed according to the supplier's recommendations for amino acid analysis (10 min at 50ºC).As was also described for the separation of amino acids derivatized with AQC [38], the Gemini NX column (ethyl-bridged column) gave lower retention times and reduced back pressure than a conventional C18 column due to its greater surface area and carbon loading.For both columns, it was observed that injection volumes of 20 μl at room temperature resulted in excessive sample loading and AMPA showed a split peak (Figure 2a).The increase of the column oven temperature to 40ºC helped to reduce the peak-widths but AMPA still gave peak splitting.The use of an oven temperature of 40ºC and an injection volume of 10 μl solved this problem with the ethyl-bridged column (Figure 2b).However, the AMPA peak still gave excessive fronting tailing with the conventional C18 column.It was necessary to reduce the injection volume to 5 μl to obtain adequate peak shapes with this column, which led to a large decrease in the sensitivity for GLP.Therefore, the Gemini column was chosen as being the most appropriate one for this study, with the column thermostated at 40ºC and injecting 10 μl.
Previous studies using AQC as a derivatization reagent for amino acids have shown that the retention time of the hydrolyzed excess of reagent (AMQ) is highly dependent on the pH of the mobile phase [31, [38][39][40][41][42], and variations in the retention time as large as 7 min were obtained with differences of only 0.35 units in the pH of the mobile phase [38].We have evaluated the effect of the pH on the separation of GLP, AMPA, and the excess of reagent AMQ with the ammonium acetate mobile phase used in this study.This salt resulted in a solution with neutral pH and, in these conditions, peak-shapes for GLP and AMPA, and the resolutions obtained, were adequate.However, this solution is not a buffer and its pH is highly labile, which means that changes in the pH of the injected solutions will modify the pH of the mobile phase, possibly having significant effects on chromatographic separations.Moreover, if mass spectrometry detection is required, ammonium acetate solutions electro-sprayed in positive ion mode are likely to undergo acidification as they are not buffered [39].For these reasons, the addition of modifiers to obtain a buffered mobile phase was evaluated.Ammonium acetate provides buffering around pH 4.75 (the pK a of acetic acid) and pH 9.24 (the pK a of ammonium), and both pH ranges were evaluated (Figure 3), with the addition of TFA to reach pH around 4.8 and ammonium to obtain pH values around 9.2.The results showed that retention time and peak shape for GLP were not affected at the different pH values evaluated.It was confirmed that retention time for the excess of hydrolyzed reagent increased significantly with the increase in the pH of the mobile phase (Figure 3a) [38], and the retention time for AMQ at pH=9.2 (adjusted with ammonium) was >18 min (Figure 3b), which resulted in an excessive total analysis time.Moreover, it was observed that the AMPA peak was split and deformed at this pH (Figure 3b), making these conditions inadequate for its quantification.
When the 50 mM ammonium acetate solution was buffered around pH 4.76 with the addition of TFA, good chromatographic peaks and resolutions for GLP and AMPA were obtained, and the retention time for AMQ was decreased to <14 min.Depending on the gradient applied for the separation, a co-elution between AMQ and AMPA were obtained at pH 4.5-4.7 (Figure 3a).The increase of the pH to 5.0 was enough to avoid any co-elution between these two compounds and the AMQ peak was always baseline separated after the AMPA peak (Figure 2b).For this reason, pH=5.0 (ammonium acetate 50 mM with addition of TFA until reaching the target pH) was chosen for the buffered solution A.
The evaluation of different flow rates ranging between 0.6 and 1.2 ml•min -1 showed that there were no significant differences at any of the flow rates evaluated.The number of theoretical plates and resolution values obtained increased slightly when the flow rate was decreased, but the parameters obtained were appropriate at all flows evaluated.Therefore, the flow rate was set at 1 ml•min -1 , also taking into account the total analysis time and the back pressure generated.

Optimization of the derivatization reaction
Cohen and Michaud demonstrated the ability of AQC to derivatize amino acids [31].
This reagent has the advantage that it reacts quickly with primary and secondary amines, and the excess of reagent is also rapidly hydrolyzed (<1 min) to yield AMQ, N-hydroxysuccinimide (NHS) and carbon dioxide.In a preliminary study [29], this reagent has shown promising results for the pre-column derivatization and analysis of GLP and AMPA, but the reaction conditions suggested by the supplier for the derivatization of amino acids were applied without a study into the effect of the In the present study, an experiment domain was defined to evaluate the influence of derivatization time and temperature on the yield of derivatized compounds.A two-level full factorial design was used to study not only each variable individually, but also the existence presence of interactions between the two variables.Table 1 summarizes the results obtained and the significances (p-values) are given.As can be seen, neither of the two variables evaluated were significant (p>0.15) in the range evaluated for the two target analytes.Despite a slight curvature and an interaction between time and temperature being observed for AMPA, the calculations revealed that it was not significant (p=0.162).
The stability of the derivatized compounds stored at 4ºC was evaluated daily for seven consecutive days and no significant differences in the peak areas measured were observed for either analyte (p>0.35).In conclusion, it is seen that the derivatization reaction is achieved in a few seconds at 24ºC, yielding stable compounds for at least one week.This finding shows that the reaction conditions for AQC are significantly simpler than those of clearly simplifies the reaction conditions required for AQC when compared with FMOC and OPA.

Validation
The efficiency of the conditioning procedure and the stability of the chromatographic column was evaluated from the retention time of controls over a five day period (n=3 13 each day).Residual standard deviations (RSD) <0.3% were obtained for GLP and <0.5% for AMPA.
The quality parameters obtained for the selected AQC derivatization/HPLC-UV procedure are shown in Table 2. Linearity was confirmed from the evaluation of the residual distribution, and determination coefficients (R 2 ) were always >0.999.Statistical evaluation of the results showed that intercept values were always non-significantly different from zero.Repeatability (RSD<8% at the low level and <4% at the high level) and inter-day precision (RSD<9% at the low level and <5% at the high level) were considered acceptable.No target compounds were detected in the blank controls evaluated.
The LODs of the method obtained (7-8 mg•l -1 , applying the 3.3SD blank criterion [37]) were of the same order as those obtained for the analysis of derivatized amino acids with AQC and also using HPLC-UV [38].Table 2 also gives the LODs obtained applying the signal-to-noise (S/N) ratio because this criterion is widely applied in chromatography.However, the S/N is an instrumental LOD and method limits should only be derived from instrumental limits when the analytical procedure does not contribute significantly to variability and bias of analytical results [43].The results indicate that the derivatization reaction introduces significant variability to the results that must be taken into account.The limits obtained with UV detection are too high excessive for the analysis of waters since the European quality standards for drinking water set a parametric value of 0.1 μg•l -1 for pesticides [44], a value that is many times exceeded when analyzing surface waters [20,29,45].A more sensitive detector, such as fluorescence or mass spectrometer, would be required for this application, although this has no effect on the optimized derivatization conditions found in the present study.In this respect, the study of Caretta et al. [29] has already demonstrated that LODs in the range 0.05-0.2μg•l -1 can be reached with a mass spectrometry detector.

Conclusions
It has been found that the use of the derivatization reagent AQC yields highly stable derivatives for glyphosate and AMPA with fast and simple reaction conditions (a few seconds at room temperature), which permits the reaction to be performed directly in the injection vial.The requirements for the derivatization reaction with this reagent present some advantages with respect to the most common derivatization reagents used for this determination (FMOC and OPA).Due to the high dependence of the retention time of the hydrolyzed excess of reagent on the pH of the mobile phase, it is important to use buffered mobile phases at pH 5.0 when using this derivatization reagent.

8
The method involves derivatization with the AccQ•Tag TM kit.Therefore, derivatization of GLP and AMPA was first conducted according to the manufacturer's instructions for amino acid analysis.Briefly, 10 μL of a stock mixture was mixed with 70 μL of borate buffer and 20 μL of AQC (AccQ•Tag reagent, previously dissolved in 1.0 mL of diluent)[34].The derivatization reaction was performed for 10 min at 55ºC in a dry heating block.
temperature.As was found in the preliminary study of Cohen and Michaud [31], AQC usually reacts with primary and secondary amines in just a few seconds.In the case of amino acids, a longer reaction time (10 min) is required due to the fact that tyrosine needs a temperature of 50ºC for 10 min to obtain a stable derivative.

Figure captions Figure 1 .
Figure captions

Figure 2 .
Figure 2. Chromatograms obtained with the Gemini column in the analysis of a

Figure 3 .
Figure 3. (a) Effect of the mobile phase A pH on the retention time for the derivatized

F
Sanchez J.M., Fontàs C., Assessment of a new derivatization reagent for the analysis of glyphosate and aminomethylphosphonic acid, Proceedings of the 1 st Iberian Meeting on Separation Sciences and Mass Spectrometry, 2019, pp.219 Commission of the European Communities, Council directive on the quality of water intended for human consumption, Council Directive 98/83/EC 1998.45.Botta F., Lavison G., Couturier G., Alliot F., Moreau-Guigon E., Fauchon N.,

Table 1 .
Statistical results obtained for the experimental design, p-values are given for the main effects and interactions.
(b)Analysis of seven independent fortified blanks