Application of large volume injection for sensitive LC-MS/MS analysis of seven artificial sweeteners in surface waters

Graphical abstract Image, graphical abstract


Specifications Table
Subject Area Surface water analysis More specific subject area LVI-LC-MS/MS for artificial sweetener analysis Method name Not applicable. A new method is presented. Name and reference of original method There are no special resources. All experimental details are given in the manuscript to reproduce the method. Resource availability Surface water analysis

Method details
Artificial sweeteners as sugar substitutes have become popular in today's calorie-conscious society. The significant increase in application results in the presence of some persistent compounds in the aquatic environment, detectable in the effluent of waste water plants and even in mineral water [1][2][3][4] . The structural features of artificial sweeteners are highly diverse, and thus complicate the chromatographic separation of multiple compounds with a single technique [5] . Therefore, a method was developed using mixed-mode chromatography on a stationary phase with C18-akyl and anion exchange properties. The separation of (i) three anionic sulfamates (acesulfame, cyclamate, saccharin), (ii) two zwitterionic dipeptides (aspartame, neotame), and (iii) two polar derivates of the natural products sucralose and neohesperidin dihydrochalcone (NHDC) was achieved. Although the use of hydrophilic interaction liquid chromatography (HILIC) columns has proven to be a successful approach for the analysis of polar sweeteners [ 6 , 7 ], the mixed-mode separation showed improved peak shapes and HILIC is not compatible with the direct injection of large volumes when the injection solvent is water, as investigated in detail by Ruta et al. [8] .
The simultaneous investigation of these seven partly persistent artificial sweeteners in surface water requires a highly sensitive detection technique, commonly accompanied with a preconcentration strategy. Solid phase extractions (SPE) for analytes with a broad polarity range can be complex as well as costly and time-consuming [9] . Hence, the implementation of a large volume injection (LVI) represents a suitable alternative to extraction techniques. Furthermore, the developed mixed-mode method operated in reversed-phase mode allowing LVIs up to 500 μL. This increase in injected volume combined with a refocusing at the column head improved the limits of detection and quantification to the low and sub-ng L 1 concentration range. Common LVI issues like analyte breakthrough, peak broadening due to ineffective refocusing at the column head and high matrix stress for the stationary phase were considered during the method development. Consequently, the ultra-trace analysis of artificial sweeteners in surface water without elaborate sample preparation steps was enabled. The combination of mixed-mode chromatography and LVI represents a versatile and fast technique for the qualitative and quantitative investigation of artificial sweetener, in the environment with high sensitivity.

LC-MS/MS analysis
The analysis of artificial sweeteners was performed on a 1100 series HPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a multidraw upgrade kit (400 μL extension, Agilent) and a 100 μL sample loop. The mixed-mode separation was conducted in reversed-phase chromatography mode on a Kaseisorb LC ODS-SAX Super column (150 × 2.0 mm, 3.0 μm; TCI) at 40 °C and a flow rate of 0.25 mL min 1 . The stationary phase comprised C18-alkyl chains with an embedded anion exchange group. The injection volume was 10 μL and was subsequently increased to large volume injections of 500 μL. Mass spectrometric detection in positive/negative switching mode was performed using an EVOQ Elite TM triple quadrupole-mass spectrometer (Bruker, Bremen, Germany). The heated electrospray ionization source operated at ±40 0 0 V with cone and heated probe temperature at 350 °C with a cone gas flow of 20 units, a probe gas flow of 40 units and a nebulizer gas flow of 60 units. For the quantification of artificial sweeteners, an optimization of the respective multiple reaction monitoring (MRM) experiment was conducted; the results are shown in Table 1 . Compass Hystar 4.0 and MS workstation (both Bruker) were used for the LC-MS instrument control, data acquisition and data evaluation.
The optimization of the mixed-mode chromatography comprised (i) variation of the buffer composition, (ii) adjustment of the pH and (iii) gradient improvement for chromatographic resolution and reduced separation time. Increasing buffer concentrations from 5 to 20 mM ammonium formate at pH 3.5 resulted in a reduced retention time for the three sulfamates acesulfame, cyclamate and saccharin. The retention time of the further analytes remained unaffected. Consequently, the higher salt concentration leads to reduced interactions between negatively charged sulfamates and the anion exchange group of the stationary phase. Moreover, peak shapes were improved due to the suppression of the anion exchange mechanism (in particular for NHDC). Subsequently, the impact of three different pH (2.7, 3.5, 4.3) was investigated in a close pH range to maintain the positive/neutral charge of the dipeptides aspartame and neotame. The latter showed the overall highest retention, which is traced Table 1 Optimized parameters for precursor ion selection (Q1), product ions (Q3) and the respective Q2 collision energy for the MRM experiments regarding quantification of seven artificial sweeteners using electrospray ionization in negative mode (top) and positive mode (bottom). The mass transitions used for quantification are shown in bold. Please note that for cyclamate just one intense MS/MS fragment could be obtained and thus, only one MRM transition was used.  back to ion exchange mechanism in addition to hydrophobic interactions. With respect to decrease the analysis time, less retention was desired. However, the variation of the pH did not affect neither the anion exchange groups of the stationary phase nor the analytes to a significant extent. Only minor changes like an increased retention for dipeptides was observed for higher pH, correlating with the net charge of these molecules in the eluent (modifying from positive to neutral, slightly improving C18 interaction). Considering the chromatographic results from (i) and (ii), the gradient optimization was conducted at pH 3.5 and starting with 85/15 (v/v) 20 mM ammonium formate and ACN. The final two-step gradient started with 15% ACN from 0 to 5 min and increased in concentration to 40% and 80% ACN as shown in Fig. 2 .

Large volume injection
The large volume injection (LVI) is a valuable alternative to preconcentration techniques for improved sensitivity [10] . LVI are defined having an injection volume > 10% of the column void volume [11] . In case of analytes covering a wide polarity range, the LVI can outperform issues of analyte loss and poor recovery rates after offline-solid phase extractions. For the implementation of an LVI, the consideration of the increased sample loop volume and starting conditions with a minimum of mobile phase elution strength was required. Furthermore, the six-port valve remained in the "inject" position after sample injection to flush the loop during the chromatographic run, reducing the possibility of carry-over effects without the need of a further pump. Consequently, the response time for the gradient elution was delayed by 2 min (500 μL loop, 0.25 mL min 1 flow rate).
The adaption of the starting conditions to 5% ACN led to a refocusing and concentration of the compounds at the head of the analytical column. Overall, a reasonable increase in separation time and a reduced chromatographic resolution (analyte window of only 2.5 min) were observed after the injection of 500 μL. Nonetheless, all artificial sweeteners were successfully separated. The final gradient conditions for LVI were 5% ACN constant for 3 min followed by a first gradient step to 60% within 2.5 min and a second step to 80% ACN within 11 min (see Fig. 3 , dotted line). Hence, the application of LVI increased the analysis time by ≈5 min, while the injection volume was 50x higher. Furthermore, the organic content in the effluent was increased from 40% to 70-80% after the method modification improving nebulization and ionization efficiency of the electrospray ionization source.
According to the applied flow rate, the transfer time from the sample loop to the analytical column was two minutes. Therefore, a possible peak broadening due to diffusion and inefficient column head refocusing was investigated, and for all seven artificial sweeteners the signals from 50 to 500 μL in 50 μL steps were monitored. The peaks of acesulfame at varying injection volumes are shown in Fig. 4 . This compound eluted in the center of the retention window and showed the overall broadest signal within the equimolar concentrated standards. For the three representative injection volumes (50, 150 and 500 μL), only a slight shift in retention time and minor peak broadening was observed. Hence, the column head refocusing was successful and the chromatographic resolution was maintained. Thus, a significant improvement of detection limits was achieved. Consequently, the sample preparation was simplified to a membrane filtration (0.2 μm polytetrafluoroethylene (PTFE)). Please note, that due to the marginal sample preparation, a faster degradation of the column performance due to the increased matrix load can be assumed and has to be balanced against the time-saving sample preparation (e.g., in terms of costs). After method development and the described sensitivity and reproducibility testing, a proof of concept study was performed on > 40 samples from surface waters (data not shown here). All samples were measured three times after membrane filtration and no adverse effects could be determined (e.g., retention time shifts, deteriorated peak shapes).

Sensitivity and reproducibility
The applicability of LVI for all seven artificial sweeteners was monitored from 50 to 500 μL. The obtained correlation between peak area and injected volume showed good linearity ( Table 2 , Fig. 4. Impact of the increasing injection volumes on the chromatography. Acesulfame is shown as a representative having the broadest initial peak shape at 10 μL injection volume.

Table 2
Comprehensive overview of LVI implementation (first column), empirically determined limits of detection and quantification (second column) and obtained parameters for the reproducibility of 500 μL injections (third column).
LVI corr [-] σ [-] Rel. LVI corr ) with relative standard deviations below 7% for six out of seven artificial sweeteners. Regarding the multiple injections necessary for higher volumes, the deviations were remarkably low. Hence, LVI is scalable in this injection volume range to improve limits of detection (LODs) and limits of quantification (LOQs). The latter are shown in the middle of Table 2 for an injection volume of 500 μL. LODs and LOQs were determined empirically due to low noise signals in some MRM transitions and are in the low and sub-ng L 1 range. For the obtained values at least a S/N ratio of 3 and 10 was ensured, respectively. The two columns on the right in Table 2 depict the reproducibility of 500 μL injections ( n = 9, c = 500 pM). Overall, the relative standard deviations were below 8% and only acesulfame exhibited a higher value (20%), possibly due to the broader peak and integration inaccuracies. This assumption gets endorsed by low relative standard deviations for aspartame (5%) and neotame (4%), both showing narrow signals. The presented combination LVI and mixed-mode chromatography is a valuable tool for reliable ultra-trace analysis of artificial sweeteners in surface water samples without elaborate sample preparation steps. The incorporation of stable-isotope labeled standards could improve reproducibility and also could compensate for potential matrix effects affecting quantification.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.