Identification of Rotenone and Five Rotenoids in CFT Legumine Piscicide Formulation via High Resolution Mass Spectrometry and a New High-Throughput Extraction Procedure

Abstract

The piscicide CFT Legumine is applied to freshwater systems around the world to control invasive fish species. Rotenone, a potent inhibitor of mitochondrial cellular respiration, is the active ingredient of the piscicide; however, other rotenoids of unknown persistence and toxicity account for an equivalent amount by weight. This work identified six distinct rotenoids in CFT Legumine using liquid chromatography coupled with high resolution orbitrap mass spectrometry and optimized a rapid surface water sampling procedure for their analysis. The rotenoids were identified as rotenone and its isomer deguelin, their 12α-hydroxylated products rotenolone and tephrosin, as well as 6α,12α-dehydrorotenone and 6α,12α-dehydrodeguelin. The optimized procedure, extraction with Spin-X nylon membrane microcentrifuge filters followed by elution with acetonitrile, achieved recoveries ranging from 101 – 107 % and 97 – 145 % for all six rotenoids at high (125 nM, ~50 ppb) and low (25 nM, ~10 ppb) concentrations of CFT Legumine, respectively. Overall, this method provides a rapid sampling procedure necessary for monitoring rotenoid persistence in surface water to ensure safe and efficacious application of the pesticide.

1.0. Introduction

Rotenone (ROT, Fig. 1a) is an inhibitor of mitochondrial cellular respiration and is commonly applied to freshwater systems in North America [1–3], Europe [4–6], Africa [7], Pacific Islands [8,9], Australia [10–12], and New Zealand [13–16], among many others worldwide not documented in the peer-reviewed literature, for the control of invasive fish species. In Alaska, the majority of rotenone is applied as CFT Legumine formulation; most of this use is to control Northern Pike (Esox lucius), an apex predator illegally introduced into regional waterways throughout the western United States and Southcentral Alaska. Northern Pike threaten local ecosystems and disrupt fishing industries through heavy predation of desirable indigenous species such as salmon and trout [17–19]. While cultural controls for invasive fish infestations are commonly practiced, chemical control has remained the most effective method to remove Pike, and other invasive species such as Goldfish, Fathead Minnows, and Muskies, from impacted watersheds. Therefore, it is imperative that fishery managers have access to rapid sampling and robust analytical methods for the determination of rotenone. This helps ensure efficacious piscicide applications, the safe reintroduction of native fishes, and enables accurate exposure considerations for humans that wish to recreate in and source drinking water from the dosed waterbodies.

Fig. 1.

Structures of the rotenoids (a) rotenone (ROT), (b) deguelin (DEG), (c) rotenolone (ROH), (d) tephrosin (TPS), (e) 6α,12α-dehydrorotenone (DHR), and (f) 6α,12α-dehydrodeguelin (DHD) identified in CFT Legumine using liquid chromatography tandem high resolution orbitrap mass spectrometry

CFT Legumine is a formulated plant extract manufactured by Zoëcon, among others, known to contain active ingredient rotenone at 5% w/w plus 5% “cube resins other than rotenone” [20]. Previous studies have identified rotenoids in cube resin [21] and nusyn-Noxfish formulations [22]; however, the distribution of rotenoids present in CFT Legumine formulation sold commercially for piscicidal use have not been reported. This may vary from those presented in prior studies depending on factors such as the source plant biotype, the plant extraction method, and the storage conditions of the formulation prior to application. Given that several rotenoids are known to exhibit NADH inhibitory properties and may have synergistic effects that enhance toxicity, it is important to provide detailed information on which rotenoids other than rotenone are being applied in CFT Legumine during piscicide dosing events [23,27].

Rotenone concentrations in natural waterbodies are commonly determined from laboratory analysis of grab samples that are collected in the field, placed on ice, transported quickly to the laboratory, and analyzed by high performance liquid chromatography (HPLC) within 7 days. Once water samples reach the lab, prior to analysis by HPLC, particulates need to be removed by pre-filtration. This preparation step generally requires dilution in methanol or acetonitrile because rotenoids may adhere to filter membranes, causing poor recoveries [22]. A common approach to streamline analysis is to dilute pre-filtered water samples in the organic solvent that matches the starting mobile phase conditions of the chromatography method. This saves an extra step in the sample preparation process. After elution, detection occurs by UV/Vis diode array or mass spectrometry; with modem triple quadruople systems, limits of quantification (LOQ) can often approach < 1 ppb. In cases where rotenone concentrations are below limits of quantitation or a laboratory does not possess an LC/MS, solid phase extraction can be used to concentrate the extracts using conventional C18 extraction disks. However, the actual methanol percent required can vary by water quality, making this step inconsistent in post-filtration recovery due to sorption of rotenone to filters. Additionally, solid phase extraction techniques can require excessive personnel time and organic solvents, which is time intensive and requires high solvent loads. Recently, Sandvik et al. developed a rapid rotenone sampling method that utilizes solid-phase extraction (preconcentration) on nylon Spin-X filters, followed by elution with acetonitrile and water (50:50, v/v), and finally analysis via HPLC diode array detection [6, 24]. This method provides rapid and reproducible determination of rotenone residues in freshwater systems while reducing the cost and waste generated by the analysis. However, this method does not provide recoveries for the other rotenoids in CFT Legumine, for which at least two, deguelin and rotenolone (12 alpha hydroxy rotenone), are known NADH inhibitors with lesser-known environmental persistence [23]. Additionally, other rotenoids besides these may account for the remaining five percent of “other cube resins” in CFT Legumine.

This work aims to address a current gap in knowledge and provide management agencies with 1) new information on what primary rotenoid components CFT Legumine contains, and 2) present a new sample preparation method capable of recovering and identifying these rotenoids in CFT Legumine. Aim 1 was accomplished using High Resolution Accurate Mass Q Exactive liquid chromatography hybrid quadrupole orbitrap mass spectrometry; aim 2 was accomplished using a rapid SPE method with an unconventional sorbent material, “Spin-X” centrifuge filters for simultaneous analysis of all identified rotenoids in the piscicide.

2.0. Materials and Methods

2.1. Chemicals

Rotenone (>98%) and deguelin (>98%) were purchased from Sigma Aldrich (St. Louis, MO) and Cayman Chemical Inc. (Ann Arbor, MI), respectively. CFT Legumine (5%) was provided gratis by Zoëcon (Burleson, TX). Methanol (MeOH; Optima for LC-MS), acetonitrile (ACN; Optima for LC-MS), formic acid (Optima for LC-MS) were purchased from Fisher Scientific (Waltham, MA). Water (LC-MS) was obtained from VWR (Radnor, PA).

2.2. ESI-UHPLC-Q Exactive Analysis of CFT Legumine Rotenoids

The rotenoid composition of CFT Legumine was determined at Thermo Fisher Scientific using a Vanquish UHPLC liquid chromatography system coupled to a Q Exactive orbitrap mass spectrometer controlled by Foundation v. 3.1, Xcalibur v. 4.1, Exactive Series v. 2.9, and SII for Xcalibur v. 1.4 (Thermo Fisher Scientific, Waltham, MA). 2 μL injections were made onto a Vanquish Accucore C18+ column (2.1 x 150mm, 1.5 μm) (Thermo Fisher Scientific, Waltham, MA) at 50 °C and separated using a gradient of water (A) and acetonitrile (B) with 0.1% formic acid at 0.4 mL min⁻¹. The mobile phase composition was held at 35% B for 30 seconds before ramping to 98% B at 7 min, holding for 3 min, and equilibration at 35% B for 5 min; total run time was 15 min. Retention times for rotenolone (ROH), tephrosin (TPS), rotenone (ROT), deguelin (DEG), 6α,12α-dehydrorotenone (DHR), and 6α,12α-dehydrodeguelin (DHD) were 3.58, 3.70, 4.16, 4.32, 4.94, and 5.15 min, respectively. Compounds were ionized via positive electrospray ionization. The spray voltage, S-lens RF level, and capillary temperature were 3.5 kV, 60, and 320 °C, respectively. Auxiliary gas flow rate and heater temperature were 10 psi and 400 °C. Sheath and sweep gas flow rates were 10 and 2 psi. Full scan MSI (120 – 500 m/z) and parallel reaction monitoring (PRM) acquisition modes with resolution of 70,000 were used. Maximum integration time and isolation width were 50 ms, and 1.2 amu. Ions of 395.1000 and 393.1500 m/z were fragmented with a collision energy of 30 V, and their product ions were monitored from 50 – 420 m/z.

2.3. Syringe Tip Filtration and Spin-X Extraction

Stock solutions containing either 250 μM or 25 μM ROT and DEG in MeOH were used to prepare daily high (125 nM, ~50 ppb) and low (12.5 nM, ~5 ppb) working solutions of ROT and DEG in water (methanol ≤ 0.1%), respectively. During preliminary investigations, 500 μL aliquots of each working solution were diluted 1:1 and 2:1 with either MeOH or ACN prior to filtration with 0.2 μm polytetrafluoroethylene (PTFE; VWR, Radnor, PA) or 0.45 μm cellulose acetate (CA; VWR, Radnor, PA) syringe tip filters (n=5) followed by analysis via liquid chromatography tandem triple quadrupole mass spectrometry (ESI-LC-QQQ, Section 2.5).

For Spin-X extractions, 500 μL of high concentration working solution was loaded into separate Costar Spin-X microcentrifuge tubes (Coming; Coming, NY) containing either 0.2 μm nylon or CA filters and centrifuged at 8000 rpm (3575 g) for 3 min. The inserts were transferred to clean microcentrifuge tubes and the aqueous phase was discarded. Next, 500 μL of extraction solvent was added to each insert, the centrifugation step was repeated, and 200 μL of the filtrate was transferred into a 250 μL autosampler vial insert for ESI-LC-QQQ analysis. Gradients of both methanol and acetonitrile with water (50:50, 60:40, 70:30, 80:20, 90:10, 100:0; v/v) were used as extraction solvents for both nylon and CA filters in triplicate.

A stock solution containing 250 μM ROT in methanol, prepared from CFT Legumine, was used to prepare high (125 nM, ~50 ppb) and low (25 nM, ~10 ppb) concentration working solutions of ROT as CFT Legumine. The optimal extraction procedure, employing nylon filters and elution in 100% ACN (details in Section 3.2), was repeated in triplicate with each working solution.

2.4. Rotenoid Stability on Spin-X Filters

A high concentration working solution of CFT Legumine was prepared and centrifuged through 0.2 μm nylon Spin-X filters as described previously (section 2.3). The filter inserts were transferred to clean microcentrifuge tubes and stored in the dark at ambient temperature (21 °C) or frozen (−20 °C). Rotenoids were extracted from the filters with ACN and analyzed via ESI-LC-QQQ (section 2.5) after 3, 7, and 15 days.

2.5. ESI-LC-QQQ Analysis of CFT Legumin Rotenoids

Rotenoids (Fig. 1) were analyzed using a 1260 Infinity series liquid chromatography system connected to a 6410A triple quadrupole mass spectrometer controlled by MassHunter v. B.05.02 (Agilent, Santa Clara, CA). 10 μL injections were made onto a Luna C18(2) column (250 x 4.6 mm, 5 μm) (Phenomenex, Torrence, CA) at 30 °C. Analytes were eluted at 0.8 mL min⁻¹ using a gradient of water (A) and acetonitrile (B) containing 0.1 % formic acid. The mobile phase composition was ramped from 60 to 98% B over 12 minutes and held for 2 minutes before resetting to 60% B and holding for 6 minutes; total run time was 20 minutes. Retention times for ROH, TPS, ROT, DEG, DHR and DHD were 8.05, 8.40, 9.44, 9.97, 11.75, and 12.49 min, respectively. Compounds were ionized via electrospray ionization at 5 kV and 350 °C with nebulizer gas and curtain gas flows of 11 L min-1 and 35 psi, respectively. Multiple reaction monitoring (MRM) acquisition parameters for each analyte are provided in Table S1. Separate calibration curves were prepared using either ROT and DEG standards or CFT Legumine; calibration curves were linear (R² > 0.996) with 1/x weighting to keep accuracy within ± 10%.Limit of detection (LOD) and limit of quantitation (LOQ) were determined for ROT (0.24 nM and 0.76 nM) and DEG (0.18 nM and 0.59 nM) by multiplying the standard deviation of seven replicate injections of the analytical grade standards by 3.1427 (Student T-statistic for 98% confidence intervals) and 10, respectively. Responses for ROT and DEG were similar, however, the relative response of DEG to ROT ranged from 1.14 – 1.23 over the calibration range 2.5 – 250 nM, which accounts for the slight decrease of the LOD. Though neither LOD nor LOQ could be determined for the remaining four rotenoids, a signal to noise ratio greater than 10 was ensured prior to a sample inclusion in the semi-quantitative analysis using calibration curves prepared with CFT Legumine.

3.0. Results & Discussion

3.1. Identification of CFT Legumine Rotenoids

Six rotenoids (Fig. 1) were identified via ESI-UHPLC-Q Exactive (section 2.2). A total ion chromatogram of CFT Legumine, MS1 scans, and product ion scans for each compound are available in the supplemental information (Fig. S1–S4). ROT (Fig. 1a) and DEG (Fig. 1b) were the two primary rotenoids identified in the piscicide. MS1 full scans contained a [M+H]⁺ peak of 395.1486 m/z with a mass accuracy of −0.8678 ppm compared to the exact mass of 395.1489 m/z for the formula C₂₃H₂₃O₆. The peaks at 417.1305 and 433.1045 m/z corresponded to [M+Na]⁺ and [M+K]⁺, respectively. The most abundant product ions were 213.0909 m/z ([M-C₉H₁₀O₄+H]⁺, C₁₄H₁₃O₂, mass accuracy −0.6309 ppm) and 192.0781 m/z ([M-C₁₂H₁₁O₃+H]⁺, C₁₁H₁₂O₃, mass accuracy 0.2568 ppm), consistent with the observed triple quadrupole MRM transitions of 395/213 and 395/192 m/z identified for ESI-LC-QQQ analysis of ROT and DEG in this work and by others (Table S1) [3, 25, 26,27]. Furthermore, ESI-LC-QQQ analysis of analytical grade ROT and DEG alongside CFT Legumine confirmed these assignments by retention time.

The earliest eluting rotenoid components were identified as the 12α-hydroxylated products of ROT and DEG, rotenolone (ROH, Fig. 1c) and tephrosin (TPS, Fig. 1d), respectively. Like ROT and DEG, ROH and TPS are isomers and shared identical MS1 and product ion scans. The precursor ions of 393.1330 and 393.1331 m/z for ROH and TPS were assigned the formula C₂₃H₂₁O₆, corresponding to the [M-H₂O+H]⁺ ion with calculated mass accuracies of −0.6870 and −0.4138 ppm. The precursor ion spectra were characterized by the prominence of 365.1384 and 365.1380 m/z (mass accuracy 0.1112 and −0.8451 ppm) corresponding to the [M-H₂O-CO+H]⁺ fragment. No matches for these spectra could be found in the mzCloud database; however, these accurate mass results are consistent with known LC-QQQ MRM transitions for rotenolone at 393/365 m/z [3, 25, 26]. The current work is also in agreement with the recent study from Said et al. that identified 393/365 m/z as the primary mass transition for deguelin using a Q-Exactive Fourier-transform high-resolution mass spectrometer [27].

The final two rotenoid components of the piscicide CFT Legumine were another pair of isomers with precursor ions of 393.1330 and 393.1331 m/z corresponding to an ion with the formula C₂₃H₂₁O₆ (mass accuracy −0.4461 and −0.7736 ppm). While these compounds ionize to the same exact mass as ROH and TPS, longer retention times and unique product ions of 361.107 m/z ([M-CH₄O+H]⁺, C₂₂H₁₇O₅, mass accuracy −0.2378 ppm) and 333.112 m/z ([M-C₂H₄O₂+H]⁺, C₂₁H₁₇O₄, mass accuracy 0.4399 ppm) indicate that these compounds are likely not hydroxylated and that the ion formed is the molecular ion [M+H]⁺ rather than the dehydration [M-H₂O+H]⁺. Once again, no spectra matches were identified using the mzCloud database, though similar chromatographic selectivity factors [28] between these rotenoids (α=1.06) and the previous two pairs (α=1.05 and 1.06 for ROH vs TPS and ROT vs DEG, respectively) suggests that rotenoids 5 and 6 are differentiated by ROT and DEG-like ring structures (Fig. 1). The compounds 6α,12α-dehydrorotenone (DHR, Fig. 1e) and 6α,12α-dehydrodeguelin (DHD, Fig. 1f), previously identified in cube resins [21, 29], Tephrosia vogelii [30], and Tephrosia candida [31], are consistent with these chromatographic observations and the [M+H]⁺ formula C₂₃H₂₁O₆.

3.2. Syringe Tip Filtration vs. Spin-X Extraction

Spike recoveries were assess using syringe-driven filter units to assess whether common laboratory filter units could consistently sorb rotenone at <50 ppb. Recovery results for the dilution and filtration of aqueous ROT and DEG samples using syringe tip filters ranged from 49.50 – 86.39 % and 58.89 – 90.24 %, respectively (Table S2). In general, recovery of ROT and DEG using syringe tip filters was low regardless of the sample concentration (125 nM vs 12.5 nM), dilution (1:1 vs 2:1), solvent (MeOH vs ACN), or filter (0.2 μm PTFE vs 0.45 μm CA) combination used; with MeOH and CA performing the poorest in terms of accuracy and precision for both ROT and DEG regardless of analyte concentration. Notably, none of the tested filtration procedures resulted in ROT recovery > 90% indicating that simple filtration through syringe-driven units is not a sufficient means of recovering ROT; a stronger non-polar sorbent is required to achieve ideal method detection limits of < 2 ppb required for water quality monitoring [32]. However, solid phase extraction methods generally require more time and materials which increases the cost of analysis and impose limits on the frequency with which water sampling may be performed post-application. Such indirect costs complicate attempts to develop accurate models for rotenoid attenuation in the environment that are helpful for ensuring safe and efficacious application of the piscicide.

To address this, Sandvik et al. presented a rapid sampling procedure employing Spin-X microcentrifuge filters for the on-site determination of ROT in surface waters [6, 24]. This work expands the scope of the method to include six rotenoids that were identified in CFT Legumine by Q Exactive orbitrap analysis (section 3.3). Extraction recoveries for ROT and DEG using cellulose acetate or nylon filters eluted with 50 – 100% solutions of MeOH or ACN (Fig. 2 and Table S3) indicate that nylon filters with ACN achieved recoveries >90%. Using ACN solutions to elute sorbed analytes from the filters greatly improved method precision, and we recommend using ACN instead of MeOH wherever possible. Of the five methods that achieved acceptable recovery (> 90%) and coefficients of variation (< 8%), ACN solutions were used as the extraction solvent and four used nylon filters. Cellulose acetate filters provided lower recovery and precision than nylon filters, particularly when used in combination with MeOH based solutions; consistent with the observations made from preliminary syringe tip filter experiments.

Fig. 2.

Average extraction recoveries of (a) ROT and (b) DEG from water using (▪) cellulose acetate and (▫) nylon filters with mixtures of MeOH:H2O and ACN:H2O as the elution solvent; asterisks (*) indicate treatments with acceptable accuracy (recovery > 90%) and precision (RSD < 8%). Error bars represent one standard deviation (n=3)

The best results were obtained when using nylon filters and 100% ACN, resulting in recoveries of 94.02 ± 1.84 % and 94.53 ± 1.44 % for ROT and DEG, respectively. Follow-up extractions using working solutions of CFT Legumin show that recovery of ROT and DEG was reproducible at both high and low concentrations (Table 1). Due to the unavailability of analytical standards, accurate concentrations of the remaining rotenoids in CFT Legumine could not be established. However, percent recovery could be calculated semi-quantitatively by comparing the integrated MRM chromatograms between ACN-eluted extracts and working solutions and assuming signal linearity between 10-500 pg on-column. As seen in Table 1, the method performed well for the remaining four rotenoids with recoveries ranging from 101 – 107 % and 97 – 145 % for high and low working solutions. Therefore, this method provides excellent recoveries for all these target rotenoids.

Table 1.

Average recoveries (n=3) of the six rotenoid components identified in CFT Legumin from water at high and low fortification levels using the optimized Spin-X extraction procedure

Compound	% Recovery (125 nM) AV±SD (%RSD); n=3	% Recovery (25 nM) AV±SD (%RSD); n=3
ROT	101.24 ± 3.09 (3.05)	107.50 ± 3.67 (3.41)
DEG	97.80 ± 0.98 (1.00)	107.82 ± 0.20 (0.18)
ROH	102.56 ± 3.39 (3.30)	97.91 ± 3.73 (3.81)
TPS	103.47 ± 2.76 (2.67)	105.76 ± 4.57 (4.32)
CFT 5	101.46 ± 3.62 (3.57)	102.43 ± 1.93 (1.89)
CFT 6	107.25 ± 12.02 (11.21)	145.44 ± 12.35 (8.49)

3.3. Rotenoid Stability on Spin-X Filters

While the immediate on-site determination of piscicide residues would be ideal, it is in many cases not practical; particularly when working in remote locations where the transport and operation of sensitive equipment would be challenging or cost prohibitive. Therefore, the stability of CFT Legumine rotenoids on Spin-X filters was investigated in order to determine if field filtration could be performed to avoid shipment of aqueous grab samples. As shown in Fig. 3, the 12α-hydroxy rotenoids ROH and TPS were stable through a 15-day period when stored in the dark, though the remaining four rotenoids, including ROT, showed rapid drop in recovery within the first three days regardless of storage temperature. Recovery was greater after 7 and 15 days of storage if the samples were frozen rather than left at room temperature; however, precision was drastically reduced. Loss of precision may have been the result of insufficient thawing time after removing from the freezer as filters were only re-equilibrated to room temperature for approximately 15 minutes prior to extraction. Sandvik et al. reported ROT was stable in ACN:H₂O (50:50, v/v) at 4 °C, therefore, if detection limits are not a concern rotenoid stability may be improved by dilution of the ACN extracts [24]. Based on these results, if aqueous samples cannot be rapidly received and analyzed by a laboratory they should be filtered and extracted in the field and preserved in a conventional freezer (approximately −20 °C) until analysis.

Fig. 3.

Relative concentration of (a/b) ROT-like and (c/d) DEG-like rotenoids on Spin-X filters stored in the dark at (a/c) 21 °C and (b/d) −20 °C over 15 days. Points and error bars represent the average and standard deviation (n=3; n=2 for day 15 of plots b & d due to a sample preparation error)

4.0. Conclusion

This work identified six distinct rotenoids in the piscicide CFT Legumine using liquid chromatography coupled to high resolution orbitrap mass spectrometry. ROT and DEG were the primary rotenoids identified along with their 12α-hydroxylated, ROH and TPS, and 6α, 12α-dehydro analogues DHR and DHD (Fig. 1). Because these rotenoids are of unknown persistence with potentially synergistic toxicities [27] a rapid extraction method employing nylon Spin-X microcentrifuge filters was optimized for their simultaneous analysis via liquid chromatography tandem triple quadrupole mass spectrometry. This method provides good recovery of all six analytes ranging from 101 – 107 % and 97 – 145 % at high (125 nM, ~50 ppb) and low (25 nM, ~10 ppb) concentrations of CFT Legumine, respectively. Future work employing NMR spectroscopy on isolated components of CFT Legumine is necessary for final confirmation of the rotenoid structures identified in this work. Additional validation of this method including an investigation of the impact surface water physicochemical properties, such as dissolved organic matter content, on rotenoid recovery should be conducted.

Supplementary Material

10337_2020_3987_MOESM1_ESM

Page S2: Table S1. Triple quadrupole mass spectrometer acquisition parameters

Fig. S1. ESI-LC-Q Exactive total and extracted ion chromatograms for CFT Legumine

Pate S3: Fig. S2. ESI-LC-Q Exactive MS1 and product ion scans for ROH and TPS

Fig. S3. ESI-LC-Q Exactive MS1 and product ion scans for ROT and DEG

Page S4: Fig. S4. ESI-LC-Q Exactive MS1 and product ion scans for DHR and DHD

Page S5: Table S2. Recovery of ROT and DEG after dilution and filtration of aqueous samples using syringe tip filters.

Page S6: Table S3. Recovery of ROT and DEG from cellulose acetate and nylon Spin-X filters using different extraction solvents