Emerging Lyngbya wollei toxins: A new high resolution mass spectrometry method to elucidate a potential environmental threat

Abstract

Mass spectrometric methods for the quantitative and qualitative analyses of algal biotoxins are often complicated by co-eluting compounds that present analytically as interferences. This issue is particularly critical for organic polyamines, where co-eluting materials can suppress the formation of cations during electrospray ionization. Here we present an extraction procedure designed specifically to overcome matrix-derived ion suppression of algal toxins in samples of Lyngbya wollei, a filamentous benthic algae known to produce several saxitoxin analogues. Lyngbya wollei samples were collected from a large, persistent harmful algal bloom in Lake Wateree, SC. Six known Lyngbya wollei-specific toxins (LWT1–6) were successfully resolved and quantified against saxitoxin using hydrophilic interaction liquid chromatography coupled with triple quadrupole and quadrupole time-of-flight mass spectrometry. The parent ions [M²⁺ – H⁺]⁺ were observed for LWTs 1–6 and the [M]²⁺ ion was observed for LWT5. High resolution mass spectra and unique fragmentation ions were obtained for LWTs 1–6. A dilution factor of 50 resulted in a linear calibration of saxitoxin in the algae matrix. Ion suppression was resolved by sample dilution, which led to linear, positive correlations between peak area and mass of the extracted sample (R² > 0.96). Optimized sample extraction method and instrument parameters are presented.

1. Introduction

Analogues of the neurotoxic alkaloid saxitoxin, also known as paralytic shellfish toxins are members of a class of naturally occurring secondary metabolites produced by freshwater cyanobacteria and marine dinoflagellates (Harada et al., 1982; Oshima et al., 1987). Saxitoxin is a selective sodium channel blocker that has been documented to be extremely toxic to a wide range of species, including humans (Carmichael, 1994; Jochimsen et al., 1998; Kao, 1993; Landsberg, 2002; Negri et al., 1995). To date, there are at least 57 known analogues of saxitoxin (Wiese et al., 2010) with varying levels of toxicity (by mouse bioassay), produced by a variety of genera including Anabaena (Al-Tebrineh et al., 2010; Humpage et al., 1994; Onodera et al., 1996), Cylindospermopsis (Lagos et al., 1999), Aphanizomenon (Jackim and Gentile, 1968; Mahmood and Carmichael, 1986; Sawyer et al., 1968), Planktothrix (Pomati et al., 2003), and Lyngbya (Dell’Aversano, 2011). Lyngbya wollei (Farlow ex Gomont) Speziale & Dyck, a filamentous, benthic cyanobacteria, is a source of saxitoxin analogues known as Lyngbya wollei toxins (Carmichael et al., 1997; Cowell and Botts, 1994; Foss et al., 2012b; Onodera et al., 1997; Yin et al., 1997) (LWTs, Figure 1 and Figure S1).

Figure 1.

Molecular structures of the Lyngbya wollei toxins (LWTs) 1–6 (free base).

The qualification and quantification of saxitoxin analogues can be challenging. Effect based assays, including the mouse bioassay (Cusick and Sayler, 2013; Schantz et al., 1958; Turner et al., 2012), in-vitro cell viability assays (Gallacher and Birkbeck, 1992; Jellett et al., 1992; Kogure et al., 1988; Manger et al., 1993), enzyme-linked immune sorbent assay (Chu et al., 1992; Humpage et al., 2010), and receptor binding assay (Davio and Fontelo, 1984; Doucette et al., 1997; Usup et al., 2004; Van Dolan et al., 2012) can provide qualitative and quantitative toxicity information (Cusick and Sayler, 2013). However, these techniques report concentrations as summed saxitoxin equivalents and do not report the relative concentrations of different structural analogues. The latter is key for developing a detailed understanding of the relevant toxin biosynthetic pathways and understanding the synergistic toxic effects possible from complex toxin sources like Lyngbya wollei.

Mass spectrometric techniques are alternatives that typically have a wider linear dynamic range than most immunoassays and offer a higher level of structural information than effect-based assays (D’Agostino et al., 2019; Dell’Aversano, 2011; Dell’Aversano et al., 2005; Dell’Aversano et al., 2019; Foss et al., 2012b; Lajeunesse et al., 2012). However, the use of electrospray ionization mass spectrometry for the identification and quantification of polyamines like the saxitoxin analogues can be complicated by the presence of co-extracted materials that affect ionization efficiency in the source (Dell’Aversano, 2011; Foss et al., 2012b). The most commonly observed effect is suppression, resulting in non-linear calibration curves for the saxitoxin analogues (Onodera et al., 1997; Annesley, 2003). Here we report the development of a new mass spectrometry method for accurate and precise measurement of Lyngbya wollei toxins at ng/L detection limits that overcomes ion suppression.

2. Materials and Methods

2.1. Materials

All chemicals were used as received. All aqueous solutions used 18 MΩ cm⁻¹ (Barnstead E-pure) water. All glassware was acid washed in 2 M HCl/0.1 M oxalic acid prior to oxidation in a muffle furnace to ensure it was trace metal and organic free. Acetonitrile (HPLC grade) was obtained from VWR BDH chemicals. Ammonium formate (98+%) was purchased from Alfa Aesar. Formic acid (certified ACS), glacial acetic acid (certified ACS PLUS), and hydrochloric acid (certified ACS Plus) were purchased from Fisher. Saxitoxin dihydrochloride in dilute hydrochloride standard solutions were obtained from NIST and Abraxis Inc.

2.2. Sampling

Lyngbya wollei grab samples were collected on November 6, 2018 from a surface floating mat on Lake Wateree, SC. Samples were collected in sterile 500-mL collection bottles and stored at 0°C during transport (less than three hours). Samples were processed immediately when received. Samples were rinsed lightly under deionized water to remove entangled detritus, drained of excess water, frozen in liquid nitrogen, and lyophilized. After lyophilization, algae samples were homogenized using a tissue grinder and stored at −20 °C until extraction.

2.3. Extraction method

Freeze dried algae samples were returned to room temperature and their masses obtained with an analytical balance. Sample mass varied from 25–300 mg of dry weight algae. All samples were mixed with 10 mL 0.1 M acetic acid, sonicated for 15 minutes and centrifuged for 5 minutes. The resulting supernatant was removed, passed through a 0.45-micron nylon filter and analyzed by liquid chromatography-mass spectrometry. We also investigated the role of a strong acid as the extraction solvent as a potential variable that could affect the measured concentration of LWTs in field samples. Extractions and subsequent dilutions were repeated using 0.1 M HCl and 0.01 M HCl. The mass to volume ratio was kept constant at 10 mg of dry weight algae for every 1 mL of extraction solvent in the extraction procedure to produce extracts for subsequent sample dilution experiments.

2.4. Chromatography and Mass Spectrometry

Toxins were analyzed using a Waters (Milford, MA, USA) Acquity ultra performance liquid chromatograph (UPLC) coupled with a Xevo triple quadrupole (TQ) mass spectrometer equipped with an electrospray ionization (ESI) source in positive ion mode. High resolution analyte confirmation was performed on an Agilent (Santa Clara, CA, USA) 1290 Infinity II ultrahigh performance liquid chromatograph (UHPLC) system coupled to an Agilent 6545 quadrupole (Q)-time-of-flight (TOF) tandem mass spectrometer with electrospray ionization (ESI) in positive ion mode. High resolution Q-TOF data was processed using Agilent B.08.00 software. Separations were performed on a BEH Amide (2.1×150 mm) 1.7 μm particle size column (Waters). The mobile phases were aqueous 5.6 mM formate buffered pH 3.5 (A) and 95:5 acetonitrile:water 5.6 mM formate buffer pH 3.5 (B). The gradient LC method used for both instruments (TQ and Q-TOF) was as follows: 80% B held for one minute, ramped to 60% B over the next 3 minutes and held for 2 minutes, at 7 minutes ramped back to the original conditions over one minute (80% B) and held for re-equilibration for 8 minutes (16 minutes total).

Saxitoxin was optimized manually by direct infusion into the source, using an optimized cone voltage of 0.5 kV and cone energy of 80 V. The source temperature was 150 °C. The desolvation temperature, extractor voltage, desolvation gas flow, cone gas flow, and collision gas flow are shown in Table S1. Limit of detection for saxitoxin on the UPLC-TQ was 0.1 ppb (Table S2). Due to the lack of commercial standards, optimization of source conditions on the UPLC-TQ mass spectrometer was done by optimizing chromatographic peak areas. A series of source methods was developed varying capillary and cone voltages to determine the optimal ionization parameters for LWT1, 4, 5, and 6 (Figure S2). A capillary voltage of 0.5 kV was optimal for all LWTs observed, consistent with the voltage for saxitoxin. LWT1 experienced in-source fragmentation at cone potentials significantly above 30 V, therefore this value was maintained at 30 V at the retention time window corresponding to LWT1 elution. The other LWTs did not experience this issue, and a cone energy of 80 V was optimal for LWT4, LWT5, and LWT6. The limit of detection was calculated for saxitoxin

Instrument parameters previously optimized for identifying small molecules by UHPLC-QTOF were utilized for LWT confirmation (Huang et al., 2018). The mass spectrometer was operated at a fragmentation voltage of 110 V, capillary voltage of 4000 V, gas temperature of 300 °C, drying gas flow of 12 L min⁻¹, and nebulizer pressure of 35 psi, with a m/z scan range of 50 to 750. During initial analyte screening, the collision energy was ramped from 0, 20, to 40 eV every scan to obtain both MS and MS/MS spectra for each peak. Once LWTs of interest were identified in algae extracts, targeted analysis was performed with a collision energy of 30 eV to obtain high resolution MS/MS spectra.

3. Results

3.1. Toxin Identification

Previous published studies on the analysis of LWTs were the result of mass spectrometers with unit-mass resolution, along with NMR (nuclear magnetic resonance) spectroscopy (Dell’Aversano et al., 2005; Foss et al., 2012a; Foss et al., 2012b; Onodera et al., 1996). In our study, saxitoxin analogues were initially detected in Lyngbya wollei extracts by full scan analysis at unit mass resolution on the UPLC-TQ mass spectrometer (Figure 2). LWTs lose a proton during ionization, forming an [M²⁺ – H⁺]⁺ parent ion, which is consistent with previously published work on doubly charged, low molecular weight poly amines by ESI-MS (Castro et al., 2001; Pizzutti et al., 2016; Wang et al., 2008). High resolution analysis by UHPLC-QTOF-MS was utilized for confirmation of the identity of saxitoxin and the suspected LWTs (Figure S3–S7). High resolution data provided exact masses of precursor and product ions, correlating to a specific molecular formula for each peak, which allowed additional confidence in the chemical structures for these toxins (Table 1).

Figure 2.

LC-MS chromatogram for LWT1 (RT = 4.79 min), LWT4 (RT = 6.64 min), LWT5 (RT = 6.11 min), and LWT6 (RT =5.79 min).

Table 1.

High resolution fragmentation data for A: LWT1, B:LWT4, C: LWT5, and D: LWT6 (ESI+).

A: LWT1
m/z	Formula	Mass loss	Formula Loss	Theoretical Mass	ppm Mass Error
379.1040	C11H19N6O7S			379.1030	2.53
299.1464	C11H19N6O4	79.9576	−SO3	299.1462	0.67
281.1350	C11H17N6O3	97.9690	−SO3, −H2O	281.1351	0.43
240.0981	C10H14N3O4	157.9895	−SO3, −H2O, −C2H4O2	240.0979	0.92
221.1145	C9H13N6O	175.0161	−SO3, −H2O, −C2H4O2, −NH3	221.1145	0.00
204.0880	C9H10N5O	139.0059	−SO3, −CH5N3	204.0880	0.20
197.1034	C8H13N4O2	182.0006	−SO3, −CH4N2, −C2H3O2	197.1033	0.51
180.0771	C8H10N3O2	199.0269	−SO3, −CH5N3, −C2H4O2	180.0768	1.94
162.0663	C8H8N3O	217.0377	−SO3, −CH5N3, −C2H4O2, −H2O	162.0662	0.68
110.0712	C5H8N3	269.0328	−SO3, −C2H4O2, −C4H8N3O2	110.0713	0.64
102.0661	C3H8N3O	277.0379	−C2H3O, −C6H8N3O5S	102.0662	0.88
72.0556	C2H6N3	307.0484	−C9H13N3O7S	72.0556	0.28
60.0557	CH6N3	319.0483	−C10H13N3O7S	60.0556	1.33
B: LWT4
m/z	Formula	Mass loss	Formula Loss	Theoretical Mass	ppm Mass Error
241.1405	C9H17N6O2			241.1408	1.04
223.1295	C9H15N6O	18.0110	−H2O	223.1302	3.09
205.1192	C9H13N6	36.0213	−H2O, −H2O	205.1196	2.05
177.0886	C7H9N6	64.0519	−H2O, −H2O, −C2H4	177.0883	1.58
164.0821	C8H10N3O	77.0584	−H2O, −CH5N3	164.0818	1.58
152.0819	C7H10N3O	89.0586	−CH4N3, −CH3O	152.0818	0.39
136.0867	C7H10N3	105.0538	−CH4N3, −H2O, −CH2O	136.0869	1.62
122.0711	C6H8N3	119.0694	−CH4N3, −CH3O, −CH2O	122.0713	1.39
110.0712	C5H8N3	131.0693	−C4H8N3O, −OH	110.0713	0.64
94.0650	C6H8N	147.0755	−C3H9N5O2	94.0651	1.28
80.0492	C5H6N	161.0913	−C4H11N5O2	80.0495	3.44
72.0556	C2H6N3	169.0849	−C4H11N5O2	72.0556	0.28
69.0447	C3H5N2	172.0958	−CH3O, −C5H9N4O	69.0447	0.29
60.0555	CH6N3	181.0850	−C8H11N3O2	60.0556	2.00
C: LWT5
m/z	Formula	Mass loss	Formula Loss	Theoretical Mass	ppm Mass Error
299.1460	C11H19N6O4			299.1462	0.77
281.1353	C11H17N6O3	18.0107	−H2O	281.1357	1.28
257.1240	C10H17N4O4	42.0220	−CH2N2	257.1244	1.67
239.1157	C10H15N4O3	60.0303	−H2O, −CH2N2	239.1139	7.65
204.0880	C9H10N5O	95.0580	−H2O, − H2O, −NH2, −C2H3O	204.0880	0.05
197.1030	C8H13N4O2	102.0430	−C2H4O2, −CH2N2	197.1033	1.52
179.0927	C8H11N4O	120.0533	−C2H4O2, −CH2N2, −H2O	179.09274	0.22
150.0755	C11H20N6O4	—	—	150.0768	8.00
138.0673	C6H8N3O	161.0787	−C2H4O2, −CH2N2, −H2O, − C2H3N	138.0662	8.04
96.0442	C5H6NO	203.1018	−CH2N2, − CH4N3, −C2H4O2, −C2H3O	96.0444	1.98
83.0604	C4H7N2	216.0856	−C2H3O2, −C5H9N4O2	83.0604	0.36
72.0552	C2H6N3	227.0908	−C9H13N3O4	72.0556	5.83
60.0557	CH6N3	239.0903	−C10H13N3O4	60.0556	1.33
D: LWT6
m/z	Formula	Mass loss	Formula Loss	Theoretical Mass	ppm Mass Error
283.1513	C11H19N6O3			283.1513	0.04
241.1301	C10H17N4O3	42.0212	−CH2N2	241.12952	2.41
224.1032	C10H14N3O3	59.0481	−CH5N3	224.1030	1.03
205.1194	C9H13N6	78.0319	−H2O, −C2H4O2	205.1196	1.07
190.0958	C8H10N6	93.0555	−H2O, −C2H4O2, −CH3	190.0962	1.84
181.1082	C8H13N4O	102.0431	−C2H4O2, −CH2N2	181.1084	1.05
177.0883	C7H9N6	106.0630	−C2H4O2, −H2O, −C2H4	177.0883	0.11
164.0825	C8H10N3O	119.0688	−C2H4O2, −CH5N3	164.0818	4.02
146.0713	C8H8N3	137.0800	−C2H4O2, −CH5N3, −H2O	146.0713	0.21
136.08679	C7H10N3	147.0645	−C2H4O2, −CH3N3, −CH2O	136.0869	0.96
122.0713	C6H8N3	161.0800	−C2H4O2, −CH3N3, −C2H4O	122.0713	0.25
110.0713	C5H8N3	173.0800	−C2H4O2, −C4H8N3O	110.0713	0.27
102.0655	C5H8N3	181.0858	−C2H3O, −C4H8N3O3	102.0662	6.76
94.0651	C6H8N	189.0862	−C5H9N5O2, −H2O	94.0651	0.21
80.0495	C5H6N	203.1018	−C6H11N5O2, −H2O	80.0495	0.31
72.0554	C2H6N3	211.0959	−C9H13N3O3	72.0556	3.05
60.0556	CH6N3	223.0957	−C10H13N3O3	60.0556	0.33

The structural similarities between the LWTs and saxitoxin means that similar product ion profiles were seen across multiple analytes of interest. For example, fragments with m/z 72.0556 and m/z 60.0556, corresponding to elemental formulae of C₂H₆N₃ (< 6 ppm mass error) and CH₆N₃ (< 2 ppm mass error), respectively, were observed for all LWTs (Table 1). Fragmentation of LWT1 resulted in ten fragment ions. The sulfur-containing functional group was lost from each fragment ion observed. Thirteen fragment ions were observed for LWT4 (Table 1). Similarly to LWT1, the most abundant fragment was from loss of the OH functional group, as well loss of nitrogenous fragments (such as CH₅N₃ and CH₄N₃). Six identical product ions were observed between LWT4 and LWT6.

Fragment ions obtained for LWT5 and saxitoxin were nearly identical (Table 1 and Table S3). Eleven fragment ions were observed for LWT5, of those eleven, eight were identical ions to saxitoxin. The three unique product ions were the result of losses from locations other than the carbamate ester (for saxitoxin) or the acetyl ester (for LWT5). Resulting mass differences in these product ions differed by 1 Da. For example, the loss of water from LWT5 resulted in the product ion m/z 281.1357, whereas the loss of water from saxitoxin resulted in m/z 282.1309. The agreement between fragments for saxitoxin and LWTs increases the confidence in the LWT identifications. Importantly, this agreement suggests parallels between the ionization chemistry of these two families of analytes, supporting the use of saxitoxin as a quantification standard for the LWTs, in the absence of commercially available standards for these toxins.

LWT2 and LWT3 are structural isomers (hereafter referred to as LWT2/3) with the same molecular weight. They were not detected using the UPLC-TQ mass spectrometer, but at least one isomer was detected with the QTOF mass spectrometer during the high resolution analysis of the algae extract (Figure S8, Table S4). LWT2/3 are structural isomers, and likely were not separable with the chromatographic approaches used in this study; thus, it is possible that both isomers were present, but coeluted. Similar to LWT1, LWT2/3 has a sulfur-containing functional group which was lost in four out of the six product ions observed. LWT2/3 were 2 orders of magnitude lower in peak intensity, relative to LWT1, 4, 5, and 6 (which had peak intensities on the order of 10⁵) on the QTOF mass spectrometer, and LWT2/3 were undetectable at these concentrations on the UPLC-TQ mass spectrometer.

3.2. Matrix Effects

The initial extraction procedure used 0.1 M acetic acid, sonication for 15 min, and filtration before analysis by UPLC-MS. However, the method response (peak area) from this extraction procedure showed a non-linear relationship vs the mass of the algae extracted (R² < 0.6838) for LWT4, 5, and 6. Furthermore, peak area vs mass extracted showed a negative slope for LWT5 (Figure 3). This behavior is consistent with the presence of co-extracted materials in the sample acting to suppress the ionization of the LWTs in the mass spectrometer (Annesley, 2003).

Figure 3.

Normalized MS peak area vs. mass of lyophilized L. wollei. The peak area of each toxin was normalized against the MS response from the lowest mass of algae.

A commercially available standard, saxitoxin, was added to the sample as a surrogate LWT to probe for the nature of matrix effects. Saxitoxin calibration curves were prepared from 0.1 M acetic acid and algae extracts for comparative purposes (Figure 4). The slope for saxitoxin in acetic acid was 7820 ppb⁻¹ with an R² of 0.9923, whereas in the algae matrix, the slope was 926 ppb⁻¹ saxitoxin with an R² of 0.2455. The concentration dependence was determined by diluting saxitoxin-spiked algae extracts in 0.1 M acetic acid (Figure 5) over a range of dilution factors (corresponding to a dilution factor of zero to 1000). Comparison of peak areas vs log dilution factor for the LWTs (Figure 5) showed a parabolic function for LWT 4, 5, and 6.

Figure 4.

Calibration curve for saxitoxin, a model LWT, in 0.1 M acetic acid (red) and in algal extract (blue).

Figure 5.

Normalized MS peak area of LWTs and saxitoxin in the algal matrix as a function of (log10) dilution factor. Peak area is normalized against the area at a dilution factor of one. Inset: Calibration curve of saxitoxin, a model LWT, in an extract of algae diluted by a factor of 50.

Examination of the dilution factor effect revealed that sample dilution of 50 or more reduced the interfering species to a level essentially below “detection” based on their ability to disproportionately affect the standard. The full calibration curve for saxitoxin was repeated at a dilution factor of 1:50 for extract:acetic acid. The corresponding curve (Figure 5 inset) was linear with a slope of 7975 ppb⁻¹ and an R² of 0.9959. In order to test the linearity of our toxins at a DF of 50, the mass of dry algae was varied over the range of 25 – 300 mg, extracted, and the resulting extracts were diluted by a factor of 50. The peak area vs mass was linear (R² > 0.9522) for all toxins observed (Figure 6).

Figure 6.

Normalized MS peak area vs. mass of lyophilized Lyngbya wollei. The peak area of each toxin was normalized against the MS response from the lowest mass of algae (25 mg). Response as a function of dry mass is linear for all 4 toxins (r₂ = 0.952 to 0.994).

3.3. Extraction in Hydrochloric acid

As shown in the literature, the choice of extraction solvent is important, and the presence of strong acids can alter the apparent toxin profile. We investigated the role of such strong acids in relation to the matrix effects experienced by the LWTs. LWT5 and LWT6 were affected strongly by the addition of HCL as the extraction solvent. This change caused an increase in the required dilution by a factor of 2 to ensure dilution of the matrix beyond visible suppression effects (Figure S9). Therefore, our results indicate the addition of HCl in any capacity during LWT analysis is inadvisable.

4. Discussion

The high-resolution fragmentation analysis presented here provides an unprecedented range of fragment ion options that can be used to conclusively indicate the presence and retention time of LWTs 1 through 6 in a sample, even in the absence of commercial standards. These same transitions can be used for MRM-based analyses in quantification. The presence of ion suppression factors can vary within environmental matrices. Smaller, polar molecules are more susceptible to ion suppression, as well as amine analytes. Biological samples are more likely to contain nonvolatile and less volatile solutes, leading to a change in spray droplet solution properties, which can be a major source of ion suppression when using electrospray. Given that saxitoxin analogues are small, polar molecules, usually extracted from natural samples, matrix effects will most likely be present in any algae extract, therefore, the current results suggest analysis of saxitoxin and its analogues extracted from algae should be subjected to a dilution factor-based assay. A strategy to obtain linear saxitoxin calibration within the algae matrix as well as linear toxin vs. mass correlations is demonstrated. The current work suggests a safe dilution factor of at least 10 for the Lake Wateree based samples, however for our work we chose 50 to ensure adequate dilution of the matrix during lake turnover events.

Accurate risk assessments for Lyngbya wollei are extremely difficult due to reference standards for the mixture of toxins produced by this algae being commercially unavailable. Effect-based assays for the analysis of these toxins remain largely non-specific, fail to provide a molecular toxin profile, and often require secondary verification by mass spectrometry. The combination of a lack of standards for quantification and qualification make risk assessment and remediation a gamble each time this species is encountered, as historically the relative concentrations of LWTs are variable and unpredictable. Utilizing a dilution factor-based assay, as presented here, provides selective detection with minimal sampling processing to avoid interferences from ion-suppressing matrix effects.

Highlights.

• High resolution mass spectra and unique fragmentation ions obtained for the algal toxins observed

• Dilution of algal extracts resulted in linear biomass relationships

• Saxitoxin exhibited similar ion suppression to the extracted algal toxins

• Optimized sample extraction method developed to decrease ion suppression

Abbreviations:

LWT

Lyngbya wollei toxin

UPLC-TQ

Ultra performance liquid chromatograph – triple quadrupole

UHPLC-QTOF

ultrahigh performance liquid chromatograph – quadrupole time-of-flight

ESI

electrospray ionization

MS/MS

tandem mass spectrometry