Evaluating the “wrong-way-round” electrospray ionization of antiretroviral drugs for improved detection sensitivity

The presence of antiretroviral drugs (ARVDs) in the aquatic environment poses a significant health risk to the ecosystem. The dilution of these compounds during wastewater treatment processes, followed by discharge into the environment, results in extremely low concentrations in the range of ng/L. Therefore, to enable detection of these low concentrations, it is important to determine the most efficient electrospray ionization (ESI) mode using the right mobile phase modifier and to establish a selective extraction procedure. In this study, we compared the ESI intensity in the positive and negative mode using both formic acid (FA) and ammonium hydroxide (NH4OH) as mobile phase modifiers. The results revealed a phenomenon known as the “wrong-way-round” (WWR) ESI in which high intensity [M + H]+ ions were detected under basic conditions using NH4OH as modifier and, similarly, high intensity [M-H]− ions were detected under acidic conditions using FA as modifier. Furthermore, mixed-mode strong cation (MCX) and mixed-mode strong anion (MAX) exchange sorbents were evaluated for extraction recoveries, which yielded extraction recoveries between 60 and 100%. Finally, the recoveries obtained using mixed-mode ion exchange sorbents compared to ion production during the ESI process provide evidence that ions produced in solution do not necessarily reflect the ions that are produced during the ESI process. Based on the results of this study, it is recommended to evaluate the optimal ionization mode under basic and acidic conditions, instead of defaulting to the use of acidic modifiers with positive ion detection. Supplementary Information The online version contains supplementary material available at 10.1007/s00216-022-04499-1.


Introduction
The accurate quantification of ARVDs present in wastewater effluent and environmental surface water has become a necessity because of the potential risks that continuous and inadvertent low-level exposure poses to non-targeted organisms [1,2] and the potential development of ARVD-resistant HIV-1 strains through unknown, low-level exposure to ARVDs in drinking water and food sources [3]. Liquid chromatography (LC) coupled with tandem mass spectrometry (MS/MS) equipped with an ESI source is the preferred method for quantifying ARVDs in aqueous samples [4]. Typically, detection of ARVDs is predominantly in the positive ion [M + H] + mode using ammonium formate, acetate, or formic acid as mobile phase modifiers [4]. The reason for predominantly using the positive ion mode is based on the premise that ARVDs contain weakly basic functional groups, i.e., amides with favorable protonation on the oxygen [5][6][7] and primary and secondary amines with preferred protonation on the nitrogen [8] in an acidic solution. In addition, it is generally assumed that the charge of the ion produced during the electrospray process depends on the mobile phase pH [9]. However, it has been shown that high intensity [M + H] + ions can be produced under strongly basic LC conditions and, similarly, that high intensity [M-H] − ions can be produced under strongly acidic LC conditions. This interesting ionization behavior was first observed by Kelly et al. [10] in the analysis of proteins and by Hiraoka and co-workers [11] in the analysis of amino acids. The term WWR ESI was later coined by Mansoori and co-workers [12] to describe this phenomenon. Since then, the WWR ionization phenomenon has been evaluated in number of studies, showing improved detection sensitivities in most cases when used for quantification of small molecules. Tso and Aga [13] reported improved signal to noise ratios for [M + H] + ions in a matrix when using basic mobile phases modified with NH 4 OH for the detection of tetracyclines and sulfonamides. Similarly, improved sensitivity by a reduction in background noise was reported in screening for doping substances in positive ionization mode under basic conditions, also with the use of NH 4 OH as modifier [14].
Considering the above, the aim of this study was to reevaluate the optimal ionization mode for selected ARVDs shown in Fig. S1, by evaluating ionization under both acidic and basic chromatographic conditions. Furthermore, to investigate the ionization of these compounds in solution and to determine their extraction recoveries from aqueous samples, a MCX and a MAX sorbents were used.

Preparation of stock solutions and working solutions
Table S1 provides a summary of the solvents used to prepare each stock solution, and the volumes spiked into a 100 mL working solution of 20% aqueous methanol to improve the poor solubility of hydrophobic ARVDs. The final working concentration of each ARVD ranged from 0.31 to 0.32 μg/mL.

Direct injections (without column)
A 20 μL aliquot of each ARVD stock solution was diluted 20 times with 50% aqueous acetonitrile containing 0.1% FA (pH 2.5) or 5 mM NH 4 OH (pH 10.7). From each of those dilutions, 10 μL was individually injected using the same dilution solvents as mobile phase at a flow rate of 200 μL/ min. An MS scan was performed with a range of 100 to 1000 amu in both the positive and negative ionization mode.

Sample extraction
Oasis® MCX (6 mL, 150 mg, 30 μm, 1 meq/g) and Oasis® MAX (6 mL, 150 mg, 30 μm, 0.25 meq/g) exchange sorbents from Waters™ via Microsep (Johannesburg, Gauteng, South Africa) were used for sample extraction. Figure S2 illustrates the generic extraction protocol (Oasis®) used to determine the extraction recovery. Briefly, following the equilibration and conditioning steps with MeOH and H 2 O, respectively, the samples were acidified with 2% aqueous FA or basified with 5% aqueous NH 4 OH, as required, before loading onto the respective MCX and MAX sorbents. Following loading, the MCX and MAX sorbents were first washed with 2% aqueous FA or 5% aqueous NH 4 OH, respectively, followed by a MeOH wash for both sorbents. The compounds were subsequently eluted form the MCX and MAX with 5% methanolic NH 4 OH or 2% methanolic FA, respectively. Samples were then evaporated and reconstituted in 20% MeOH for LC-MS analysis.

Liquid chromatography-mass spectrometry analysis
Analyses were performed on an Agilent 1260 chromatography system equipped with an autosampler, solvent manager, and column oven. Separations were performed using a Kinetex® EVO C 18  to 90% in 16 min, followed by re-equilibration for 5 min at 99% A for a total analysis time of 22 min. The flow rate was 0.4 mL/min, and the injection volume was 2 μL.
The LC system was connected to an Agilent MSD/XT single quadrupole mass spectrometer with an atmospheric jet stream ESI (AJS-ES) source. The capillary and nozzle voltages were 1.5 kV and 1 kV, respectively, for both positive and negative ionization modes. The sheath gas temperature, sheath gas flow, drying gas temperature, drying gas flow, and nebulizer pressure were 300 °C, 12 L/min, 320 °C, 5 L/min, and 34 PSIG. The in-source fragmentation voltage was optimized for each ARVD.

Results and discussion
Initially, each ARVD was injected (without a column) in a mobile phase containing either 0.1% FA or 5 mM NH 4 OH in aqueous ACN (1:1, v/v) to determine their respective positive and negative ions. The results are summarized in Table S2, which shows the [M + H] + and [M-H] − ions in the positive and negative mode, respectively, for most of the ARVDs. However, RTV and its metabolite, RTVM, produced a predominate formate adduct [M + COOH] − in the negative ion mode using 0.1% FA as modifier. Furthermore, 8,14-EFVM produced no [M + H] + peak, but instead produced low-abundance sodium [M + Na] + and ammonium [M + NH 4 ] + adducts under acidic and basic conditions, respectively. Following this, a reversed phase LC-MS method was developed for both polarities under acidic and basic conditions using a Kinetex® EVO C 18 column, due to its high pH stability. The m/z ratios used in the positive and negative ion mode are shown in Table S3. An overlay of the resulting chromatograms is presented in Fig. 1. The chromatogram produced in the negative ion mode using 5 mM NH 4 OH as modifier was omitted due to a high background noise (data not shown).
Comparing the resulting retention times revealed an overall increase in retention of the ARVDs under basic conditions. This was especially true for the early eluting 3TC that showed a threefold increase in the retention factor. However, under basic chromatographic conditions, coelution of RTV and RTVM was observed, in contrast to what was seen under acidic conditions when these compounds were resolved.
To visually inspect the differences in ionization intensity (peak area) between the four different chromatographic runs, extracted ion chromatographic data for each ARVD were exported to Excel for retention time alignment. The results are illustrated in Fig. 2, which shows an overlay of the extracted ion chromatograms obtained under acidic and basic conditions in the positive and negative ionization mode. The top left-hand corner of Fig. 2 shows the peak numbers (also shown in color) for both ionization modes under acidic and basic mobile phase conditions. Furthermore, note that the fold increase (> #x) shown in the top right-hand corner of each tile is only shown for the WWR ESI, i.e., using the same ionization mode under different LC conditions. For example, 3TC has a sixfold increase in peak area using NH 4   Evaluating the "wrong way round"electrospray ionization of antiretroviral drugs for improved… --

3
conditions during the ESI process. The most likely theory is that gas phase NH 4 + ions are produced during the electrospray process, which serve as a proton donor when NH 4 OH is used as modifier [13]. Using sulfonamides as a model compound, this theory was deduced from the observation that basic modifiers other than NH 4 OH, i.e., sodium hydroxide and tetramethylammonium hydroxide produced no [M + H] + peak, proposing that a basic modifier with an ionizable gas phase proton is necessary [13]. On the other hand, the production of [M-H] − ions under acidic conditions could be explained by the gas-phase proton affinity of the formic acid anions [21].
The frequent use of FA for the detection of ARVDs in the positive ionization mode can be attributed to various factors. Firstly, as already mentioned, ARVDs containing amides, primary and tertiary amines are considered weakly basic compounds, and therefore, it is assumed that the addition of an acid in solution will enhance the subsequent ESI process [5,8,28]. Secondly, concern over the stability of earlier silica-based LC columns that can undergo hydrolysis at high mobile phase pH values made the use of acidified mobile phases commonplace. However, more recently, many manufactures produce columns that are stable at high pH values.
To demonstrate that the ions produced in solution do not necessarily reflect the nature of the ions produced during the ESI process, MCX and MAX extraction sorbents were used. For experimental details refer to the "Sample extraction" section and Fig. S2. For maximum retention of ARVDs on a MCX or MAX sorbent, the working solution was either acidified with 2% FA or basified with 5% NH 4 OH, respectively. Following the respective washing steps, the ARVDs were desorbed from the MCX and MAX sorbent using 5% NH 4 OH and 2% FA, respectively. The results are illustrated in Fig. 3. This shows that 3TC, FTC, NVPM, NVP, RTVM, and RTV are highly protonated in acidic (2% FA) solution, yielding recoveries ranging from 60 to 95% when using MCX sorbents. On the other hand, AZT, AZTG, and EFV yielded recoveries > 96% using a MAX sorbent under basic conditions, pointing to the formation of deprotonated ARVDs under basic conditions.
The extraction of 8-EFVM and 8,14-EFVM using a MAX sorbent revealed the production of a second LC peak for both at an increased retention time, illustrated in Fig. S3a (asterisk). Consequently, determining the extraction recoveries for these analytes were not possible. To determine the cause of this anomaly, the 2 mL working solution was mixed with either 4 mL 5% aqueous NH 4 OH to mimic the loading and washing steps or 4 mL 2% methanolic FA used for the final elution step. The samples were then left standing for approximately 10 min (duration of a MAX extraction) at room temperature (23 °C) before injection. The 5% aqueous NH 4 OH dilution were injected directly, whereas the 2% methanolic FA eluate was diluted with water (1:4, v/v) to make it amenable for reversed phase LC. Furthermore, to determine if heating during the final evaporation step at 50 °C caused the formation of double peaks, the final eluate, 2% methanolic FA, was also diluted (1:4, v/v) with water and injected without the evaporation step. Figure S3b represents the chromatogram of the working solution prior to extraction. Temperature was not the cause since the same LC results were obtained before and after evaporation illustrated in Fig. S3a.   Fig. 2 An overlay of the extracted ion chromatograms for each ARVD obtained in the positive and negative ionization mode using 5 mM NH 4 OH and 0.1% FA as mobile phase modifiers Furthermore, the 2% methanolic FA dilution produced the same results as the working solution shown in Fig. S3b. However, the 5% aqueous NH 4 OH dilution produced the same additional peaks compared to the MAX extraction for both 8,14-EFVM and 8-EFVM at an increased retention time shown in Fig. S3c. This anomaly was not observed for EFV. Despite this, it is clear from Fig. S3a that both 8-EFVM and 8,14-EFVM has a strong affinity for MAX sorbents under basic conditions. Table 1 summarizes the ARVD ions that are formed in solution in the presence of an acid or base and the preferred ion exchange sorbent yielding the highest recoveries. In addition, the ions formed in highest abundance for each analyte during the ESI process are also shown. It is clear from these results that the equilibrium concentrations of ions in solution do not necessarily reflect the ions produced by electrospray ion droplets, thereby supporting the WWR ESI phenomenon.

Summary
In this study, we have shown improved detection sensitivities by using the WWR ESI for the detection of ARVDs. Using a basic NH 4 OH-modified mobile phase for detection of [M + H] + ions and using an acidic FA-modified mobile phase for detection of [M-H] − ions increased the detection sensitivity multiple fold when compared to the conventional use of acidified mobile phases for the detection of [M + H] + ions. The significant reduction in m/z signal intensity observed when FA was used for detection of ARVDs in positive mode, which has been used in most of the previous studies, implies that ARVDs with low abundances can go undetected when analyzing surface waters and wastewater effluent. In addition, by having improved sensitivities, smaller sample volumes can be used for extractions, thereby reducing the amount of consumables used. Using the WWR ionization technique can also reduce analysis time, and the use of solvents in that detection of positively and negatively charged ions can be done in a single analysis method using NH 4 OH as a modifier as was seen for the simultaneous detection of antibiotics and estrogens in a single analysis method [13].
We also developed a selective extraction procedure using MCX and MAX sorbents, which yielded extraction recoveries exceeding 60%. Although these extraction recoveries were only determined in MilliQ water, it serves as a good starting point for the extraction of ARVDs from complex samples such as wastewater effluent and aqueous environmental samples. Furthermore, the WWR ionization is further supported when comparing the specific mixed-mode ion exchange sorbent recoveries with the ions produced during the ESI process, proving that pre-ionization of compounds in solution does not necessarily enhance the signal intensity achieved during the ESI process.   Table 1 A summary of the ion exchange sorbent used for each ARVD and the preferred modifier for optimal ionization during the ESI process Evaluating the "wrong way round"electrospray ionization of antiretroviral drugs for improved… --1191

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.