Anal Bioanal Chem (2015) 407:1463–1473 DOI 10.1007/s00216-014-8367-6
RESEARCH PAPER
Liquid chromatography–high-resolution mass spectrometry for palytoxins in mussels Patrizia Ciminiello & Carmela Dell’Aversano & Emma Dello Iacovo & Martino Forino & Luciana Tartaglione
Received: 16 September 2014 / Revised: 14 November 2014 / Accepted: 21 November 2014 / Published online: 9 December 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Palytoxins from Ostreopsis cf. ovata (a putative palytoxin and ovatoxins) are emerging toxins in the Mediterranean basin and are not yet regulated, although there is evidence that they can accumulate in seafood and thus enter the human food chain. This poses serious concerns for human health, because palytoxin itself is among the most potent marine toxins known. In 2009, the European Food Safety Authority (EFSA) announced the need for optimization of efficient analytical methods for detecting palytoxins and for preparing standards. Herein, we propose a procedure including a one-step extraction, solid-phase-extraction (SPE) cleanup, and liquid chromatography-high resolution mass spectrometry (LC–HRMS) detection of individual palytoxins in mussels. The method enabled efficient chromatographic separation of individual compounds, including structural isomers, with good sensitivity, reproducibility, and linearity in a large dynamic range (14–1000 ng mL−1 in matrix). As a result, the putative palytoxin from Ostreopsis cf. ovata was identified as an isomer of palytoxin itself and re-named isobaric palytoxin. The whole procedure (sample preparation and LC–HRMS analysis) proved able to detect palytoxins in both spiked and natural mussel samples at levels as low as 70 μg kg−1 in crude mussel extracts and 15 μg kg−1 after SPE clean-up. Although full validation of the method is currently prevented by the unavailability of palytoxin(s) certified standards and reference material, this study constitutes a first step towards achieving this.
Electronic supplementary material The online version of this article (doi:10.1007/s00216-014-8367-6) contains supplementary material, which is available to authorized users. P. Ciminiello : C. Dell’Aversano (*) : E. Dello Iacovo : M. Forino : L. Tartaglione Department of Pharmacy, University of Napoli Federico II, Via D. Montesano 49, 80131 Napoli, Italy e-mail: [email protected]
Keywords Palytoxin . Putative palytoxin . Isobaric palytoxin . Ovatoxins . LC–HRMS . Mussels
Introduction Since the late 1990s, the benthic dinoflagellate Ostreopsis cf. ovata has repeatedly bloomed in the Mediterranean Sea [1]. On the basis of liquid chromatography–mass spectrometry (LC–MS) evidence, several Mediterranean strains of this alga were identified as producers of palytoxins (ovatoxin-a to f, and a putative palytoxin), but with variable toxin profiles and contents from strain to strain [2–4]. Most Mediterranean O. cf. ovata strains analyzed produce ovatoxin-a (OVTX-a) as the major component of their toxin profile (≥50 %), followed by OVTX-b, OVTX-d/e, OVTX-c, and a putative palytoxin (pPLTX), listed in decreasing order of concentration [5–7]. As well as the above most common toxin profile, some strains were found not to produce OVTX-b and c, whereas others produced OVTX-f [4, 8, 9]. Ovatoxins differ from palytoxin in only a few structural details, and toxicological data on these compounds are still lacking. OVTX-a has recently been isolated as a pure compound and its structure fully elucidated (Fig. 1) [9–11], but only high-resolution-mass-spectrometry (HRMS) data are available (Electronic Supplementary Material (ESM) Table S1) for OVTX-b to f, which have not yet been isolated [4, 6, 7]. The presence of palytoxin-producing algae in the Mediterranean Sea has raised serious concerns for public-healthprotection authorities for at least three reasons: 1. Palytoxin itself [12] is one of the most potent marine toxins known, presenting very high acute toxicity in mice after intra-venous, intra-peritoneal, and intra-tracheal administration [13];
1464
P. Ciminiello et al. O
OH
OH
O
OH
O
OH
HO
H2N
OH
OH O
OH
OH Me OH
HO
OH OH
OH A moiety O HO
( )n
N H
O
Me
OH
OH
HO
Me
OH OH
OH
O
HN OH
Me
OH OH
OH
O OH
R1
OH
R4
B moiety Me
OH O
OH
O O
Me
OH R2 O
HO
Me
OH
OH OH
OH R3
OH
OH
OH
OH
n
R1
R2
R3
R4
Palytoxin
1
OH
H
OH
OH
Ovatoxin-a
1
H
OH
H
H
Fig. 1 Structure of palytoxin and ovatoxin-a. Stereochemistry of palytoxin is reported. Stereochemistry of ovatoxin-a is inverted at positions C-9, C-26, and C-57
2. Concurrently with Ostreopsis spp. blooms, serious health problems have been recorded in humans after inhalation of aerosols [14, 15] and/or dermal contact [16]; and 3. Palytoxins can accumulate in marine edible animals, thus entering the human food chain [17]. Palytoxins have been believed to be the cause of a few fatal human seafood poisonings in the Tropics [18, 19]. Reported signs and symptoms included bitter and/or metallic taste, myalgia, vomiting, diarrhea, bradycardia, tingling of the extremities, and delirium, among others. However, in such case reports the presence of palytoxins in leftovers was not conclusively proved, because of the lack of a reliable and sensitive analytical method for their detection in seafood. The subject has become even more controversial because the oral acute toxicity of palytoxin by gavage has been recently estimated to be relatively low [20, 21], and the presence of palytoxin in several blue humphead parrotfish implicated in symptomatic “palytoxin-like poisoning” cases in Japan has been excluded [22]. Currently, there is no regulation on palytoxins either in the European Union or in other regions of the world. However, because of the potential health risks posed by palytoxins to seafood consumers, monitoring programs for these toxins in seafood have been enacted in many Mediterranean countries. Within such programs, LC–MS studies of seafood from the French Mediterranean coast [23] and Southern Italy (Ciminiello et al., personal communication) have identified mussels and sea-urchins as the most commonly contaminated edible animals. Sporadic occurrence of palytoxins has been detected in
octopus and viscera of small fish [17]. Average levels of contamination (calculated as the sum of OVTXs and pPLTX) of most of the French samples (72 %) fell in the 80–400 μg kg−1 range, whereas over 50 % of the Italian samples were in the 80–600 μg kg−1 range. In Italy, in most cases ovatoxin-a was the only toxin detected (83 % of samples), and the remaining 17 % of samples contained it in association with ovatoxin-d/e. Alerted by the repeated occurrence of palytoxinproducing species in the Mediterranean, the European Food Safety Authority (EFSA), in its statement on the PLTX group of toxins [24], suggested a safe maximum tolerance level for palytoxins of 30 μg kg−1 in shellfish, creating the need for preparation of reference material and for optimization of analytical methods for selective and sensitive detection of palytoxins in shellfish. Liquid chromatography combined with either fluorescence (LC-FLD) or tandem-mass-spectrometry (LC-MS/MS) detection has been used by several groups to detect palytoxins in algal samples [25, 26], and only an LC-MS/MS screening method including periodate micro-scale oxidation has proved successful for detecting palytoxins in shellfish at levels below the EFSA limit (LOD=10 μg kg−1) [27]. This method, however, necessitates a derivatization step and cannot distinguish between the individual palytoxin congeners that might have different toxicity [20]. In this work, we report on the optimization of an analytical procedure for selective determination of individual palytoxins in mussels, which includes extraction, clean-up, and LC–HRMS. Method linearity, dynamic range, matrix effect, limit of detection (LOD) and quantitation (LOQ), and overall procedure recovery and reproducibility were evaluated using the only commercially available palytoxin standard. The proposed approach was deliberately based only on full HRMS experiments, without exploiting the MSn potential of our instrument (LTQ Orbitrap XL) at its fullest. This choice was made with the purpose of providing a procedure widely accessible to users who rely only on less expensive HRMS systems (e.g. TOF, Orbitrap exactive, etc.). With the objective of using the analytical method to analyze Mediterranean mussels, we had first to define the chromatographic conditions for an efficient separation of ovatoxins, which are all closely related in structure, ionization, fragmentation, and chromatographic behavior [26]. Although a full validation of the whole procedure is currently prevented by the lack of both palytoxin and ovatoxin certified standards, the main achievements of this study are efficient chromatographic separation of all known ovatoxins and reliable quantitation of palytoxins in mussel extracts, either as-such or after a clean-up step, depending on the level of contamination of mussel samples.
LC-HRMS for palytoxins in mussels
Materials and methods Chemicals and materials Mobile phases were prepared using Water Chromasolv Plus for HPLC and Acetonitrile E-Chromasolv for HPLC for UV ≥99.9 % (Sigma Aldrich, Italy). All the organic solvents used in extraction and clean-up procedures and the glacial acetic acid (Laboratory grade) were by Carlo Erba (Italy). Ammonium hydroxide (28 % NH3 in water, 99.99 %) and additives (Ca, Mg, Sr, Na, and K acetates) were from Sigma–Aldrich (Italy). Palytoxin standard (from Palythoa tuberculosa) was purchased from Wako Chemicals GmbH (Germany). This standard is not certified, and samples of the same lot were used to prepare calibration curves and in spiking experiments. A previously characterized crude extract of O. cf. ovata [9] containing a combination of palytoxins was used as reference. Blank mussels were used for spiking experiments before and after extraction or clean-up. Wild mussels collected along the Campania coasts (Italy) in summer 2013 concurrently with an O. cf. ovata bloom were used to test the developed procedure. Analytical method Liquid chromatography–high-resolution mass spectrometry (LC–HRMS) All analyses were performed on a hybrid linear ion trap LTQ Orbitrap XL Fourier-transform mass spectrometer (FTMS) with an ESI ION MAX source (Thermo-Fisher, USA) combined with an Agilent 1100 LC binary system (USA). Three columns were tested—Gemini C18, 3 μm, 2 ×150 mm; Kinetex C18, 2.6 μm, 2.10×100 mm (Phenomenex, USA); and Poroshell 120 EC-C18, 2.7 μm, 2.1×100 mm (Agilent, USA)—under a variety of isocratic and gradient conditions. Mobile phase was A=H2O, 30 mmol L−1 acetic acid, and B= 95 % MeCN–H2O, 30 mmol L−1 acetic acid. Flow was 0.2 mL min−1 and injection volume was 5 μL. For the Poroshell column only, we investigated the effect of pH by analyzing PLTX standard (50 ng mL−1) under gradient elution at different-pH mobile phase, namely pH 6.6 (mobile phase A=H2O, B=95 % MeCN–H2O), pH in the range 2.8–5.0 (mobile phase A=H2O, B=95 % MeCN–H2O, both eluents containing acetic acid in the range 0.02–170 mmol L−1), and pH 8.4 (mobile phase A=H2O, B=95 % MeCN–H2O, both eluents containing ammonium hydroxide 19.8 μmol L−1). In addition, we evaluated the effect of additives by separately adding Mg, Sr, Na, and K acetates in the 1–200 nmol L−1 range and Ca acetate in the 1–2000 nmol L−1 range to the mobile phase (A=H2O, B=95 % MeCN–H2O, both eluents containing 30 mmol L−1 acetic acid). The optimized elution conditions for each column and retention times (Rt) of each toxin are summarized in Table 1. The degree of separation
1465
between closely eluting palytoxins was calculated as resolution (R) using the following equation, where RtA 10. Matrix interference was calculated by comparing MF and MM
1466
P. Ciminiello et al.
Table 1 Retention times (Rt, min) of PLTX, isobaric PLTX (previously referred to as putative PLTX) [31] and OVTX-a to e, measured on the three tested columns under different elution conditions. Total run time and equilibration (Eq.) time are also reported Gemini C18
Kinetex C18
Poroshell 120 EC-C18
Fast gradienta Rt (min)
Slow gradientb Rt (min)
Gradientc Rt (min)
Gradientd Rt (min)
Isocratice Rt (min)
PLTX Isobaric PLTX OVTX-a OVTX-b OVTX-c OVTX-d
6.29 6.22 6.33 6.31 6.22 6.28
11.31 10.78 11.45 11.28 10.90 11.07
10.14 8.40 10.95 10.29 8.90 9.58
8.55 6.64 9.62 8.83 7.10 7.78
9.38 6.94 10.92 9.79 7.51 8.36
OVTX-e Run time Eq. time
6.28 15 10
11.15 31 10
9.77 21 9
8.30 21 14
9.04 15 –
Details of columns and mobile phase are given in the Experimental section a
20–100 % B in 10 min; 100 % B for 5 min
b
20–50 % B in 20 min; 50–80 % B in 10 min; 80–100 % B in 1 min; 100 % B for 4 min
c
25–30 % B in 15 min; 30–100 % B in 1 min; 100 % B for 5 min
d
28–29 % B in 5 min; 29–30 % B in 10 min; 30–100 % B in 1 min; 100 % B for 5 min
e
28 % B
response (in terms of peak area) for each concentration level. Ion suppression (IS %) was assessed as:
ðpeak area of MM standardÞ 100− 100 ðpeak area of M F standardÞ
Sample preparation Extraction Mussels were homogenized in a Waring blender and subsampling was done immediately after blending while still well-mixed. Each aliquot (1 g) was extracted with 3 mL MeOH–H2O 8:2 (v/v), sonicated for 3 min while cooling in an ice bath, and then centrifuged at 6500 rpm for 10 min. Each supernatant was decanted and saved for analysis (musseltissue concentration = 0.3 g mL−1). Aliquots of blank mussel extracts were spiked with PLTX after extraction and used as MM standards for quantitation. Accuracy and reproducibility of the extraction was assessed over three replicates, spiking mussel tissue (1 g) with PLTX standard at four concentrations (60, 150, 500, and 1000 μg kg−1). The spiked tissue was extracted as described above and the extracts were analyzed in triplicate by LC–HRMS versus PLTX MM standards. Accuracy was assessed in terms of recovery, calculated as ((mean measured concentration)/(spiked concentration))×100. Reproducibility was expressed in terms of relative standard deviation (RSD).
Solid-phase-extraction (SPE) clean-up Several SPE cartridges were tested, namely Strata-X 500 mg/6 mL, Strata-XL 500 mg/6 mL, Strata-X 200 mg/6 mL (Phenomenex, USA), OASIS HLB LP 6 cc (500 mg) (Waters, USA), and PolyLC INC (LabService Analytica, Italy), under different combinations of the following loading, washing, and eluting conditions: 1. Load: MeOH–H2O 2:8, 1:9, 5:95 2. Wash: MeOH–H2O 1:1, 4:6, 3:7, 1:9, H2O 100 % 3. Elute: MeOH 100 %, MeOH–H2O 9:1, 8:2, 8:2 with 0.2 % acetic acid, MeOH with 1 % acetic acid, MeOH– H2O 8:2 with 0.1 % trifluoroacetic acid, isoPrOH–H2O 8:2, isoPrOH–H2O–acetic acid 40:59:1, 70:29:1. Accuracy of the SPE procedure was assessed on post-cleanup parallel blank and spiked (30, 60, and 150 μg kg−1) mussel extracts; in such experiments aliquots of the SPE elutes of blank mussel extracts were spiked with PLTX after the cleanup and used as PLTX MM standards for quantitation. All SPE elutes of the spiked mussel extracts (including load, wash, and elutes) were analyzed by LC–HRMS versus MM standards, both as-such and after concentration under N2 flow. Accuracy and reproducibility of the whole procedure (extraction, clean-up, and analytical method) were assessed by contaminating blank mussel tissue with PLTX standard at three concentrations (60, 30, and 15 μg kg−1) and repeating the whole procedure over five replicates, in parallel with a
LC-HRMS for palytoxins in mussels
blank. LC–HRMS analyses were performed using a Poroshell column under gradient elution. Matrix-matched standards of PLTX were used for quantitation. LOD and LOQ (μg kg−1) were measured and/or extrapolated from the linear regression curve as reported above.
Results and discussion The LC–MS/MS method for determination of palytoxin in algal samples that we previously developed on a triplequadrupole (TQ) MS was used as the starting point of this work [5]. It used a reversed-phase column (Gemini C18) eluted with aqueous acetonitrile and 30 mmol L−1 acetic acid under gradient conditions. When used to analyze palytoxins in mussels, this method had an LOQ of 228 μg kg−1 on TQ MS [28]; this value was inadequate to detect palytoxins at the safety level suggested by EFSA. Another disadvantage was that, under the LC conditions used, all the palytoxins almost co-eluted. This chromatographic behavior, considering the ever-increasing number of O. cf. ovata toxins discovered and their complex ionization profile (high number of multiply charged ions, mixed cationized species, and high 13C isotope contribution) [26], prevented the accurate quantitation of some palytoxin analogues by TQ MS. Hence, the objective of our work was to develop a procedure (extraction, clean-up, and LC–HRMS method) able to quantify palytoxins in mussels in quite a large dynamic range, covering the EFSA level (30 μg kg−1) and the levels of palytoxins in seafood measured by the monitoring programs (80–600 μg kg−1). Chromatographic separation of potentially interfering palytoxins was a major challenge to be addressed. To this end, the efficiency of three reversed-phase columns was tested. A previously characterized Mediterranean O. cf. ovata extract [9] and blank mussel extracts spiked with it were used as references. Some refinements to the extraction procedure of palytoxin from mussels [28] and optimization of the clean-up procedure were also performed. The whole procedure (extraction, clean-up, and analytical method) was tested on mussels spiked with PLTX before extraction at 0.5, 1, and 2 times the EFSA-suggested level, and on real samples of mussels collected in summer 2013 in the Gulf of Napoli (Italy). Analytical method Chromatographic separation An O. cf. ovata algal extract containing several palytoxins (OVTX-a 52 %, OVTX-b 29 %, OVTX-c 5 %, OVTX-d 7 %, OVTX-e 6 %, and pPLTX 1 %) at levels of 10 μg mL−1 (total toxin content) was injected on a Gemini, a Kinetex, and a Poroshell reversed-phase column. Figure 2 shows the obtained
1467
LC–HRMS results in the form of extracted ion chromatograms (XICs) of the most intense [M + H + Ca]3+ ions for each compound, and Table 1 reports retention times of individual toxins for each chromatographic system. All the O. cf. ovata toxins co-eluted on the Gemini column under fast gradient conditions (Fig. 2a). This LC behavior did not affect unambiguous identification of each toxin on the basis of its full HRMS spectrum [26]. However, with regards to quantification, the structural isomers ovatoxin-d and -e were quantifiable only as a combined total, and most ovatoxins of different elemental formula were not accurately quantifiable because they formed different adduct ions having similar exact masses (ESM Table S1). For example, the calcium adduct of ovatoxin-a (m/z 895.8225) differs by only 5.6 mDa from the magnesium adduct of ovatoxin-d/-e (m/z 895.8281). Considering that the resolving power (RP) of the Orbitrap mass analyzer diminishes as the square root of m/z [29], even at the highest resolution of 100,000 (effective RP at m/z 896=66,809) the FWHM of the m/z 896 peak is 13 mDa. This value impedes complete separation of the overlapping ion peaks of OVTX-a (m/z 895.8225) and OVTX-d/-e (m/z 895.8281), thus impeding accurate quantification of OVTX-a on the basis of XIC of its [M + H + Ca]3+ ion. Similar problems occur for most palytoxins (ESM Table S1), and could be overcome by selecting the [M + 3H − nH2O]3+ ions (n=1–3); these are, however, barely detectable on our LC– HRMS system. As a consequence, if the Gemini column is used under fastgradient conditions, HRMS/MS experiments are needed to accurately quantify individual compounds and appropriate selection of fragment ions must be performed [4, 7, 10, 26]. A slow gradient on the Gemini column (Fig. 2b) enabled us to chromatographically separate most of the potentially interfering toxins, although all of them still eluted in a quite narrow range (approximately 1.2 min) [26]. Under such conditions, a fivefold decrease in the response (measured on palytoxin standard) was observed compared with the fast-gradient elution. In addition, a heavy chromatographic overlapping occurred when slightly more concentrated algal extracts (>12 ng mL−1) were analyzed (data not shown). In this case a dilution step was required. Compared with the Gemini, the Kinetex column provided a better chromatographic separation (Fig. 2c, Table 1). All the toxins eluted in a 3.4 min range with sharper peaks, although OVTX-d and e were still barely separated; in an attempt to separate them, several gradients were tested (data not shown), but none of them succeeded and they resulted in total run times exceeding 40 min. The Poroshell column provided the best chromatographic resolution (Fig. 2d, e) among the tested columns. All the toxins—including OVTX-d and -e—were well separated and eluted in a 4 min range under gradient elution (Fig. 2d), with an average resolution of 1.2 between closely eluting
1468
P. Ciminiello et al.
a)
OVTX-a Rt = 11.45
b)
OVTXs and pPLTX Rt = 6.22-6.33
OVTX-a Rt = 10.95
c)
OVTX-b Rt = 11.28 OVTX-e Rt = 11.15 OVTX-d Rt = 11.07
OVTX-d Rt = 9.58
OVTX-c Rt = 10.90
4
5
6 Time, min
7
8
8
9
9
pPLTX Rt = 8.40 10
12
13
14
7
8
9
OVTX-a Rt = 9.62
OVTX-e Rt = 8.30 OVTX-c Rt = 7.10
OVTX-e Rt = 9.04
OVTX-d Rt = 8.36
9
13
OVTX-b Rt = 9.79
pPLTX Rt = 6.94
8 Time, min
12
OVTX-a Rt = 10.92
OVTX-b Rt = 8.83
OVTX-d Rt = 7.78
7
11
Kinetex slow gradient
e)
pPLTX Rt = 6.64
6
10 Time, min
Gemini slow gradient
d)
5
11 Time, min
Gemini fast gradient
OVTX-b Rt = 10.29
OVTX-c Rt = 8.90
pPLTX Rt = 10.78
3
OVTX-e Rt = 9.77
10
11
6
OVTX-c Rt = 7.51
7
8
9 Time, min
10
11
12
Poroshell isocratic
Poroshell gradient 3+
Fig. 2 Extracted ion chromatograms (XICs) of [M + H + Ca] ions of pPLTX (herein renamed isobaric palytoxin) and OVTX-a to -e under different LC conditions: Gemini, fast (a) and slow (b) gradients; Kinetex, gradient (c); Poroshell, gradient (d) and isocratic (e)
compounds. Under these conditions equilibration time had an important effect on the reproducibility of retention times (Rt), with a minimum of 14 min being required. The reproducibility of Rt of each toxin measured overnight was suitable (RSD 1.1–1.5; n=6). The Poroshell column was also able to separate palytoxins under isocratic conditions over a 15 min run (Fig. 2e), with an average resolution of 1.7 between closely eluting compounds. Reproducibility of Rt (RSD 1.0–1.3; n=6) was comparable to that for gradient conditions. Once an efficient chromatographic separation was obtained, the effect of pH on analyte response (in terms of peak area) was investigated in the pH 2.8–8.4 range, analyzing PLTX MF standard (50 ng mL−1) under gradient elution. Under acidic conditions (acetic acid added to mobile phase in the range 17.5–170 mmol L−1, pH 2.8–5.0), response (in terms of peak area) changed significantly whereas Rt was only slightly affected (Rt 8.55±0.36 min). The highest response was obtained at pH 3.1 (30 mmol L−1 acetic acid), whereas response decreased by approximately five times at lower (170 mmol L − 1 acetic acid, pH 2.8) and higher (17.5 mmol L−1 acetic acid, pH 3.3) pH. At pH 6.3 (no acid or base added to mobile phase), response in terms of peak area was higher than that at pH 3.1 but the PLTX peak (Rt 7.02 min) was very broad, spreading over 6 min. This might result in a heavy chromatographic overlapping of different
ovatoxins when a real sample (algal or mussel extract) is analyzed. At pH 8.4 (ammonium hydroxide 19.8 μmol L−1 added to mobile phase) response was very weak and the PLTX peak (Rt 13.94 min) had significant fronting; in addition, high background noise was present in the associated full MS spectrum, with only the [M + H + K]2+ ion of PLTX clearly detectable. Figure S1 in the ESM reports LC–HRMS analysis of the PLTX MF standard (50 ng mL−1) at pH 3.1, 6.3, and 8.4. As a result, a pH of 3.1 (corresponding to 30 mmol L−1 acetic acid added to the mobile phase) was regarded as optimum. The use of additives was considered, to make response less susceptible to pH variations and to improve MS response through reduction of the ion population in full HRMS spectra. Because the use of ammonium acetate had already proved unable to fulfill such objectives [5], Ca, Mg, Sr, Na, and K acetates were separately added to the mobile phase in the nmol L−1 concentration range. Regardless of the additive used, the [M + H + Ca]3+ ion of palytoxin dominated the full HRMS spectra, in agreement with previous findings that palytoxin has the highest affinity for Ca [10, 30]. No significant improvement of the response was observed, whereas background noise increased significantly until it completely suppressed the PLTX signal. During the above analyses, different chromatographic behavior in terms of retention times was observed between the
LC-HRMS for palytoxins in mussels
1469
PLTX standard and the pPLTX in the O. cf. ovata extract (Table 1). This was a clear indication that the two molecules were indeed isomers. To verify this hypothesis, given the good chromatographic separation of toxins achieved with the Kinetex (gradient) and the Poroshell (gradient and isocratic) columns, the O. cf. ovata extract was spiked with palytoxin standard at 500 ng mL−1 and analyzed again. The two molecules eluted at different retention times in all experimental settings (Fig. 3a), although they produced superimposable full HRMS spectra (Figs. 3b, c), thus confirming their isobaric relationship. Structural insights into putative palytoxin have been recently gained from LC–HRMSn evidence, and the compound has been re-named “isobaric palytoxin” [31]. Linearity, limits of detection (LOD) and quantitation (LOQ), and matrix interference Palytoxin MF and MM standards at different concentrations were analyzed using full HRMS on the three columns to assess linearity (R2 over three replicates), LOD and LOQ (ng mL−1) extrapolated from the linear regression curve, experimentally measured LOQ, and matrix interference (IS %). First, the effect of the resolving power (RP) of the mass analyzer (RP=15,000; 30,000; 60,000; 100,000) on PLTX response was investigated, using the Kinetex column. Table S2 in the ESM shows the LOQ in MF and MM standards extrapolated from the calibration curves and matrix interference measured at different RP settings. A similar
b)
matrix effect (IS % in the range 50–68 %) occurred at 60,000 and 100,000 RP, whereas at 15,000 and 30,000 RP the PLTX signal was totally suppressed by the matrix at 25 ng mL−1 and linearity was poor (
RESEARCH PAPER
Liquid chromatography–high-resolution mass spectrometry for palytoxins in mussels Patrizia Ciminiello & Carmela Dell’Aversano & Emma Dello Iacovo & Martino Forino & Luciana Tartaglione
Received: 16 September 2014 / Revised: 14 November 2014 / Accepted: 21 November 2014 / Published online: 9 December 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Palytoxins from Ostreopsis cf. ovata (a putative palytoxin and ovatoxins) are emerging toxins in the Mediterranean basin and are not yet regulated, although there is evidence that they can accumulate in seafood and thus enter the human food chain. This poses serious concerns for human health, because palytoxin itself is among the most potent marine toxins known. In 2009, the European Food Safety Authority (EFSA) announced the need for optimization of efficient analytical methods for detecting palytoxins and for preparing standards. Herein, we propose a procedure including a one-step extraction, solid-phase-extraction (SPE) cleanup, and liquid chromatography-high resolution mass spectrometry (LC–HRMS) detection of individual palytoxins in mussels. The method enabled efficient chromatographic separation of individual compounds, including structural isomers, with good sensitivity, reproducibility, and linearity in a large dynamic range (14–1000 ng mL−1 in matrix). As a result, the putative palytoxin from Ostreopsis cf. ovata was identified as an isomer of palytoxin itself and re-named isobaric palytoxin. The whole procedure (sample preparation and LC–HRMS analysis) proved able to detect palytoxins in both spiked and natural mussel samples at levels as low as 70 μg kg−1 in crude mussel extracts and 15 μg kg−1 after SPE clean-up. Although full validation of the method is currently prevented by the unavailability of palytoxin(s) certified standards and reference material, this study constitutes a first step towards achieving this.
Electronic supplementary material The online version of this article (doi:10.1007/s00216-014-8367-6) contains supplementary material, which is available to authorized users. P. Ciminiello : C. Dell’Aversano (*) : E. Dello Iacovo : M. Forino : L. Tartaglione Department of Pharmacy, University of Napoli Federico II, Via D. Montesano 49, 80131 Napoli, Italy e-mail: [email protected]
Keywords Palytoxin . Putative palytoxin . Isobaric palytoxin . Ovatoxins . LC–HRMS . Mussels
Introduction Since the late 1990s, the benthic dinoflagellate Ostreopsis cf. ovata has repeatedly bloomed in the Mediterranean Sea [1]. On the basis of liquid chromatography–mass spectrometry (LC–MS) evidence, several Mediterranean strains of this alga were identified as producers of palytoxins (ovatoxin-a to f, and a putative palytoxin), but with variable toxin profiles and contents from strain to strain [2–4]. Most Mediterranean O. cf. ovata strains analyzed produce ovatoxin-a (OVTX-a) as the major component of their toxin profile (≥50 %), followed by OVTX-b, OVTX-d/e, OVTX-c, and a putative palytoxin (pPLTX), listed in decreasing order of concentration [5–7]. As well as the above most common toxin profile, some strains were found not to produce OVTX-b and c, whereas others produced OVTX-f [4, 8, 9]. Ovatoxins differ from palytoxin in only a few structural details, and toxicological data on these compounds are still lacking. OVTX-a has recently been isolated as a pure compound and its structure fully elucidated (Fig. 1) [9–11], but only high-resolution-mass-spectrometry (HRMS) data are available (Electronic Supplementary Material (ESM) Table S1) for OVTX-b to f, which have not yet been isolated [4, 6, 7]. The presence of palytoxin-producing algae in the Mediterranean Sea has raised serious concerns for public-healthprotection authorities for at least three reasons: 1. Palytoxin itself [12] is one of the most potent marine toxins known, presenting very high acute toxicity in mice after intra-venous, intra-peritoneal, and intra-tracheal administration [13];
1464
P. Ciminiello et al. O
OH
OH
O
OH
O
OH
HO
H2N
OH
OH O
OH
OH Me OH
HO
OH OH
OH A moiety O HO
( )n
N H
O
Me
OH
OH
HO
Me
OH OH
OH
O
HN OH
Me
OH OH
OH
O OH
R1
OH
R4
B moiety Me
OH O
OH
O O
Me
OH R2 O
HO
Me
OH
OH OH
OH R3
OH
OH
OH
OH
n
R1
R2
R3
R4
Palytoxin
1
OH
H
OH
OH
Ovatoxin-a
1
H
OH
H
H
Fig. 1 Structure of palytoxin and ovatoxin-a. Stereochemistry of palytoxin is reported. Stereochemistry of ovatoxin-a is inverted at positions C-9, C-26, and C-57
2. Concurrently with Ostreopsis spp. blooms, serious health problems have been recorded in humans after inhalation of aerosols [14, 15] and/or dermal contact [16]; and 3. Palytoxins can accumulate in marine edible animals, thus entering the human food chain [17]. Palytoxins have been believed to be the cause of a few fatal human seafood poisonings in the Tropics [18, 19]. Reported signs and symptoms included bitter and/or metallic taste, myalgia, vomiting, diarrhea, bradycardia, tingling of the extremities, and delirium, among others. However, in such case reports the presence of palytoxins in leftovers was not conclusively proved, because of the lack of a reliable and sensitive analytical method for their detection in seafood. The subject has become even more controversial because the oral acute toxicity of palytoxin by gavage has been recently estimated to be relatively low [20, 21], and the presence of palytoxin in several blue humphead parrotfish implicated in symptomatic “palytoxin-like poisoning” cases in Japan has been excluded [22]. Currently, there is no regulation on palytoxins either in the European Union or in other regions of the world. However, because of the potential health risks posed by palytoxins to seafood consumers, monitoring programs for these toxins in seafood have been enacted in many Mediterranean countries. Within such programs, LC–MS studies of seafood from the French Mediterranean coast [23] and Southern Italy (Ciminiello et al., personal communication) have identified mussels and sea-urchins as the most commonly contaminated edible animals. Sporadic occurrence of palytoxins has been detected in
octopus and viscera of small fish [17]. Average levels of contamination (calculated as the sum of OVTXs and pPLTX) of most of the French samples (72 %) fell in the 80–400 μg kg−1 range, whereas over 50 % of the Italian samples were in the 80–600 μg kg−1 range. In Italy, in most cases ovatoxin-a was the only toxin detected (83 % of samples), and the remaining 17 % of samples contained it in association with ovatoxin-d/e. Alerted by the repeated occurrence of palytoxinproducing species in the Mediterranean, the European Food Safety Authority (EFSA), in its statement on the PLTX group of toxins [24], suggested a safe maximum tolerance level for palytoxins of 30 μg kg−1 in shellfish, creating the need for preparation of reference material and for optimization of analytical methods for selective and sensitive detection of palytoxins in shellfish. Liquid chromatography combined with either fluorescence (LC-FLD) or tandem-mass-spectrometry (LC-MS/MS) detection has been used by several groups to detect palytoxins in algal samples [25, 26], and only an LC-MS/MS screening method including periodate micro-scale oxidation has proved successful for detecting palytoxins in shellfish at levels below the EFSA limit (LOD=10 μg kg−1) [27]. This method, however, necessitates a derivatization step and cannot distinguish between the individual palytoxin congeners that might have different toxicity [20]. In this work, we report on the optimization of an analytical procedure for selective determination of individual palytoxins in mussels, which includes extraction, clean-up, and LC–HRMS. Method linearity, dynamic range, matrix effect, limit of detection (LOD) and quantitation (LOQ), and overall procedure recovery and reproducibility were evaluated using the only commercially available palytoxin standard. The proposed approach was deliberately based only on full HRMS experiments, without exploiting the MSn potential of our instrument (LTQ Orbitrap XL) at its fullest. This choice was made with the purpose of providing a procedure widely accessible to users who rely only on less expensive HRMS systems (e.g. TOF, Orbitrap exactive, etc.). With the objective of using the analytical method to analyze Mediterranean mussels, we had first to define the chromatographic conditions for an efficient separation of ovatoxins, which are all closely related in structure, ionization, fragmentation, and chromatographic behavior [26]. Although a full validation of the whole procedure is currently prevented by the lack of both palytoxin and ovatoxin certified standards, the main achievements of this study are efficient chromatographic separation of all known ovatoxins and reliable quantitation of palytoxins in mussel extracts, either as-such or after a clean-up step, depending on the level of contamination of mussel samples.
LC-HRMS for palytoxins in mussels
Materials and methods Chemicals and materials Mobile phases were prepared using Water Chromasolv Plus for HPLC and Acetonitrile E-Chromasolv for HPLC for UV ≥99.9 % (Sigma Aldrich, Italy). All the organic solvents used in extraction and clean-up procedures and the glacial acetic acid (Laboratory grade) were by Carlo Erba (Italy). Ammonium hydroxide (28 % NH3 in water, 99.99 %) and additives (Ca, Mg, Sr, Na, and K acetates) were from Sigma–Aldrich (Italy). Palytoxin standard (from Palythoa tuberculosa) was purchased from Wako Chemicals GmbH (Germany). This standard is not certified, and samples of the same lot were used to prepare calibration curves and in spiking experiments. A previously characterized crude extract of O. cf. ovata [9] containing a combination of palytoxins was used as reference. Blank mussels were used for spiking experiments before and after extraction or clean-up. Wild mussels collected along the Campania coasts (Italy) in summer 2013 concurrently with an O. cf. ovata bloom were used to test the developed procedure. Analytical method Liquid chromatography–high-resolution mass spectrometry (LC–HRMS) All analyses were performed on a hybrid linear ion trap LTQ Orbitrap XL Fourier-transform mass spectrometer (FTMS) with an ESI ION MAX source (Thermo-Fisher, USA) combined with an Agilent 1100 LC binary system (USA). Three columns were tested—Gemini C18, 3 μm, 2 ×150 mm; Kinetex C18, 2.6 μm, 2.10×100 mm (Phenomenex, USA); and Poroshell 120 EC-C18, 2.7 μm, 2.1×100 mm (Agilent, USA)—under a variety of isocratic and gradient conditions. Mobile phase was A=H2O, 30 mmol L−1 acetic acid, and B= 95 % MeCN–H2O, 30 mmol L−1 acetic acid. Flow was 0.2 mL min−1 and injection volume was 5 μL. For the Poroshell column only, we investigated the effect of pH by analyzing PLTX standard (50 ng mL−1) under gradient elution at different-pH mobile phase, namely pH 6.6 (mobile phase A=H2O, B=95 % MeCN–H2O), pH in the range 2.8–5.0 (mobile phase A=H2O, B=95 % MeCN–H2O, both eluents containing acetic acid in the range 0.02–170 mmol L−1), and pH 8.4 (mobile phase A=H2O, B=95 % MeCN–H2O, both eluents containing ammonium hydroxide 19.8 μmol L−1). In addition, we evaluated the effect of additives by separately adding Mg, Sr, Na, and K acetates in the 1–200 nmol L−1 range and Ca acetate in the 1–2000 nmol L−1 range to the mobile phase (A=H2O, B=95 % MeCN–H2O, both eluents containing 30 mmol L−1 acetic acid). The optimized elution conditions for each column and retention times (Rt) of each toxin are summarized in Table 1. The degree of separation
1465
between closely eluting palytoxins was calculated as resolution (R) using the following equation, where RtA 10. Matrix interference was calculated by comparing MF and MM
1466
P. Ciminiello et al.
Table 1 Retention times (Rt, min) of PLTX, isobaric PLTX (previously referred to as putative PLTX) [31] and OVTX-a to e, measured on the three tested columns under different elution conditions. Total run time and equilibration (Eq.) time are also reported Gemini C18
Kinetex C18
Poroshell 120 EC-C18
Fast gradienta Rt (min)
Slow gradientb Rt (min)
Gradientc Rt (min)
Gradientd Rt (min)
Isocratice Rt (min)
PLTX Isobaric PLTX OVTX-a OVTX-b OVTX-c OVTX-d
6.29 6.22 6.33 6.31 6.22 6.28
11.31 10.78 11.45 11.28 10.90 11.07
10.14 8.40 10.95 10.29 8.90 9.58
8.55 6.64 9.62 8.83 7.10 7.78
9.38 6.94 10.92 9.79 7.51 8.36
OVTX-e Run time Eq. time
6.28 15 10
11.15 31 10
9.77 21 9
8.30 21 14
9.04 15 –
Details of columns and mobile phase are given in the Experimental section a
20–100 % B in 10 min; 100 % B for 5 min
b
20–50 % B in 20 min; 50–80 % B in 10 min; 80–100 % B in 1 min; 100 % B for 4 min
c
25–30 % B in 15 min; 30–100 % B in 1 min; 100 % B for 5 min
d
28–29 % B in 5 min; 29–30 % B in 10 min; 30–100 % B in 1 min; 100 % B for 5 min
e
28 % B
response (in terms of peak area) for each concentration level. Ion suppression (IS %) was assessed as:
ðpeak area of MM standardÞ 100− 100 ðpeak area of M F standardÞ
Sample preparation Extraction Mussels were homogenized in a Waring blender and subsampling was done immediately after blending while still well-mixed. Each aliquot (1 g) was extracted with 3 mL MeOH–H2O 8:2 (v/v), sonicated for 3 min while cooling in an ice bath, and then centrifuged at 6500 rpm for 10 min. Each supernatant was decanted and saved for analysis (musseltissue concentration = 0.3 g mL−1). Aliquots of blank mussel extracts were spiked with PLTX after extraction and used as MM standards for quantitation. Accuracy and reproducibility of the extraction was assessed over three replicates, spiking mussel tissue (1 g) with PLTX standard at four concentrations (60, 150, 500, and 1000 μg kg−1). The spiked tissue was extracted as described above and the extracts were analyzed in triplicate by LC–HRMS versus PLTX MM standards. Accuracy was assessed in terms of recovery, calculated as ((mean measured concentration)/(spiked concentration))×100. Reproducibility was expressed in terms of relative standard deviation (RSD).
Solid-phase-extraction (SPE) clean-up Several SPE cartridges were tested, namely Strata-X 500 mg/6 mL, Strata-XL 500 mg/6 mL, Strata-X 200 mg/6 mL (Phenomenex, USA), OASIS HLB LP 6 cc (500 mg) (Waters, USA), and PolyLC INC (LabService Analytica, Italy), under different combinations of the following loading, washing, and eluting conditions: 1. Load: MeOH–H2O 2:8, 1:9, 5:95 2. Wash: MeOH–H2O 1:1, 4:6, 3:7, 1:9, H2O 100 % 3. Elute: MeOH 100 %, MeOH–H2O 9:1, 8:2, 8:2 with 0.2 % acetic acid, MeOH with 1 % acetic acid, MeOH– H2O 8:2 with 0.1 % trifluoroacetic acid, isoPrOH–H2O 8:2, isoPrOH–H2O–acetic acid 40:59:1, 70:29:1. Accuracy of the SPE procedure was assessed on post-cleanup parallel blank and spiked (30, 60, and 150 μg kg−1) mussel extracts; in such experiments aliquots of the SPE elutes of blank mussel extracts were spiked with PLTX after the cleanup and used as PLTX MM standards for quantitation. All SPE elutes of the spiked mussel extracts (including load, wash, and elutes) were analyzed by LC–HRMS versus MM standards, both as-such and after concentration under N2 flow. Accuracy and reproducibility of the whole procedure (extraction, clean-up, and analytical method) were assessed by contaminating blank mussel tissue with PLTX standard at three concentrations (60, 30, and 15 μg kg−1) and repeating the whole procedure over five replicates, in parallel with a
LC-HRMS for palytoxins in mussels
blank. LC–HRMS analyses were performed using a Poroshell column under gradient elution. Matrix-matched standards of PLTX were used for quantitation. LOD and LOQ (μg kg−1) were measured and/or extrapolated from the linear regression curve as reported above.
Results and discussion The LC–MS/MS method for determination of palytoxin in algal samples that we previously developed on a triplequadrupole (TQ) MS was used as the starting point of this work [5]. It used a reversed-phase column (Gemini C18) eluted with aqueous acetonitrile and 30 mmol L−1 acetic acid under gradient conditions. When used to analyze palytoxins in mussels, this method had an LOQ of 228 μg kg−1 on TQ MS [28]; this value was inadequate to detect palytoxins at the safety level suggested by EFSA. Another disadvantage was that, under the LC conditions used, all the palytoxins almost co-eluted. This chromatographic behavior, considering the ever-increasing number of O. cf. ovata toxins discovered and their complex ionization profile (high number of multiply charged ions, mixed cationized species, and high 13C isotope contribution) [26], prevented the accurate quantitation of some palytoxin analogues by TQ MS. Hence, the objective of our work was to develop a procedure (extraction, clean-up, and LC–HRMS method) able to quantify palytoxins in mussels in quite a large dynamic range, covering the EFSA level (30 μg kg−1) and the levels of palytoxins in seafood measured by the monitoring programs (80–600 μg kg−1). Chromatographic separation of potentially interfering palytoxins was a major challenge to be addressed. To this end, the efficiency of three reversed-phase columns was tested. A previously characterized Mediterranean O. cf. ovata extract [9] and blank mussel extracts spiked with it were used as references. Some refinements to the extraction procedure of palytoxin from mussels [28] and optimization of the clean-up procedure were also performed. The whole procedure (extraction, clean-up, and analytical method) was tested on mussels spiked with PLTX before extraction at 0.5, 1, and 2 times the EFSA-suggested level, and on real samples of mussels collected in summer 2013 in the Gulf of Napoli (Italy). Analytical method Chromatographic separation An O. cf. ovata algal extract containing several palytoxins (OVTX-a 52 %, OVTX-b 29 %, OVTX-c 5 %, OVTX-d 7 %, OVTX-e 6 %, and pPLTX 1 %) at levels of 10 μg mL−1 (total toxin content) was injected on a Gemini, a Kinetex, and a Poroshell reversed-phase column. Figure 2 shows the obtained
1467
LC–HRMS results in the form of extracted ion chromatograms (XICs) of the most intense [M + H + Ca]3+ ions for each compound, and Table 1 reports retention times of individual toxins for each chromatographic system. All the O. cf. ovata toxins co-eluted on the Gemini column under fast gradient conditions (Fig. 2a). This LC behavior did not affect unambiguous identification of each toxin on the basis of its full HRMS spectrum [26]. However, with regards to quantification, the structural isomers ovatoxin-d and -e were quantifiable only as a combined total, and most ovatoxins of different elemental formula were not accurately quantifiable because they formed different adduct ions having similar exact masses (ESM Table S1). For example, the calcium adduct of ovatoxin-a (m/z 895.8225) differs by only 5.6 mDa from the magnesium adduct of ovatoxin-d/-e (m/z 895.8281). Considering that the resolving power (RP) of the Orbitrap mass analyzer diminishes as the square root of m/z [29], even at the highest resolution of 100,000 (effective RP at m/z 896=66,809) the FWHM of the m/z 896 peak is 13 mDa. This value impedes complete separation of the overlapping ion peaks of OVTX-a (m/z 895.8225) and OVTX-d/-e (m/z 895.8281), thus impeding accurate quantification of OVTX-a on the basis of XIC of its [M + H + Ca]3+ ion. Similar problems occur for most palytoxins (ESM Table S1), and could be overcome by selecting the [M + 3H − nH2O]3+ ions (n=1–3); these are, however, barely detectable on our LC– HRMS system. As a consequence, if the Gemini column is used under fastgradient conditions, HRMS/MS experiments are needed to accurately quantify individual compounds and appropriate selection of fragment ions must be performed [4, 7, 10, 26]. A slow gradient on the Gemini column (Fig. 2b) enabled us to chromatographically separate most of the potentially interfering toxins, although all of them still eluted in a quite narrow range (approximately 1.2 min) [26]. Under such conditions, a fivefold decrease in the response (measured on palytoxin standard) was observed compared with the fast-gradient elution. In addition, a heavy chromatographic overlapping occurred when slightly more concentrated algal extracts (>12 ng mL−1) were analyzed (data not shown). In this case a dilution step was required. Compared with the Gemini, the Kinetex column provided a better chromatographic separation (Fig. 2c, Table 1). All the toxins eluted in a 3.4 min range with sharper peaks, although OVTX-d and e were still barely separated; in an attempt to separate them, several gradients were tested (data not shown), but none of them succeeded and they resulted in total run times exceeding 40 min. The Poroshell column provided the best chromatographic resolution (Fig. 2d, e) among the tested columns. All the toxins—including OVTX-d and -e—were well separated and eluted in a 4 min range under gradient elution (Fig. 2d), with an average resolution of 1.2 between closely eluting
1468
P. Ciminiello et al.
a)
OVTX-a Rt = 11.45
b)
OVTXs and pPLTX Rt = 6.22-6.33
OVTX-a Rt = 10.95
c)
OVTX-b Rt = 11.28 OVTX-e Rt = 11.15 OVTX-d Rt = 11.07
OVTX-d Rt = 9.58
OVTX-c Rt = 10.90
4
5
6 Time, min
7
8
8
9
9
pPLTX Rt = 8.40 10
12
13
14
7
8
9
OVTX-a Rt = 9.62
OVTX-e Rt = 8.30 OVTX-c Rt = 7.10
OVTX-e Rt = 9.04
OVTX-d Rt = 8.36
9
13
OVTX-b Rt = 9.79
pPLTX Rt = 6.94
8 Time, min
12
OVTX-a Rt = 10.92
OVTX-b Rt = 8.83
OVTX-d Rt = 7.78
7
11
Kinetex slow gradient
e)
pPLTX Rt = 6.64
6
10 Time, min
Gemini slow gradient
d)
5
11 Time, min
Gemini fast gradient
OVTX-b Rt = 10.29
OVTX-c Rt = 8.90
pPLTX Rt = 10.78
3
OVTX-e Rt = 9.77
10
11
6
OVTX-c Rt = 7.51
7
8
9 Time, min
10
11
12
Poroshell isocratic
Poroshell gradient 3+
Fig. 2 Extracted ion chromatograms (XICs) of [M + H + Ca] ions of pPLTX (herein renamed isobaric palytoxin) and OVTX-a to -e under different LC conditions: Gemini, fast (a) and slow (b) gradients; Kinetex, gradient (c); Poroshell, gradient (d) and isocratic (e)
compounds. Under these conditions equilibration time had an important effect on the reproducibility of retention times (Rt), with a minimum of 14 min being required. The reproducibility of Rt of each toxin measured overnight was suitable (RSD 1.1–1.5; n=6). The Poroshell column was also able to separate palytoxins under isocratic conditions over a 15 min run (Fig. 2e), with an average resolution of 1.7 between closely eluting compounds. Reproducibility of Rt (RSD 1.0–1.3; n=6) was comparable to that for gradient conditions. Once an efficient chromatographic separation was obtained, the effect of pH on analyte response (in terms of peak area) was investigated in the pH 2.8–8.4 range, analyzing PLTX MF standard (50 ng mL−1) under gradient elution. Under acidic conditions (acetic acid added to mobile phase in the range 17.5–170 mmol L−1, pH 2.8–5.0), response (in terms of peak area) changed significantly whereas Rt was only slightly affected (Rt 8.55±0.36 min). The highest response was obtained at pH 3.1 (30 mmol L−1 acetic acid), whereas response decreased by approximately five times at lower (170 mmol L − 1 acetic acid, pH 2.8) and higher (17.5 mmol L−1 acetic acid, pH 3.3) pH. At pH 6.3 (no acid or base added to mobile phase), response in terms of peak area was higher than that at pH 3.1 but the PLTX peak (Rt 7.02 min) was very broad, spreading over 6 min. This might result in a heavy chromatographic overlapping of different
ovatoxins when a real sample (algal or mussel extract) is analyzed. At pH 8.4 (ammonium hydroxide 19.8 μmol L−1 added to mobile phase) response was very weak and the PLTX peak (Rt 13.94 min) had significant fronting; in addition, high background noise was present in the associated full MS spectrum, with only the [M + H + K]2+ ion of PLTX clearly detectable. Figure S1 in the ESM reports LC–HRMS analysis of the PLTX MF standard (50 ng mL−1) at pH 3.1, 6.3, and 8.4. As a result, a pH of 3.1 (corresponding to 30 mmol L−1 acetic acid added to the mobile phase) was regarded as optimum. The use of additives was considered, to make response less susceptible to pH variations and to improve MS response through reduction of the ion population in full HRMS spectra. Because the use of ammonium acetate had already proved unable to fulfill such objectives [5], Ca, Mg, Sr, Na, and K acetates were separately added to the mobile phase in the nmol L−1 concentration range. Regardless of the additive used, the [M + H + Ca]3+ ion of palytoxin dominated the full HRMS spectra, in agreement with previous findings that palytoxin has the highest affinity for Ca [10, 30]. No significant improvement of the response was observed, whereas background noise increased significantly until it completely suppressed the PLTX signal. During the above analyses, different chromatographic behavior in terms of retention times was observed between the
LC-HRMS for palytoxins in mussels
1469
PLTX standard and the pPLTX in the O. cf. ovata extract (Table 1). This was a clear indication that the two molecules were indeed isomers. To verify this hypothesis, given the good chromatographic separation of toxins achieved with the Kinetex (gradient) and the Poroshell (gradient and isocratic) columns, the O. cf. ovata extract was spiked with palytoxin standard at 500 ng mL−1 and analyzed again. The two molecules eluted at different retention times in all experimental settings (Fig. 3a), although they produced superimposable full HRMS spectra (Figs. 3b, c), thus confirming their isobaric relationship. Structural insights into putative palytoxin have been recently gained from LC–HRMSn evidence, and the compound has been re-named “isobaric palytoxin” [31]. Linearity, limits of detection (LOD) and quantitation (LOQ), and matrix interference Palytoxin MF and MM standards at different concentrations were analyzed using full HRMS on the three columns to assess linearity (R2 over three replicates), LOD and LOQ (ng mL−1) extrapolated from the linear regression curve, experimentally measured LOQ, and matrix interference (IS %). First, the effect of the resolving power (RP) of the mass analyzer (RP=15,000; 30,000; 60,000; 100,000) on PLTX response was investigated, using the Kinetex column. Table S2 in the ESM shows the LOQ in MF and MM standards extrapolated from the calibration curves and matrix interference measured at different RP settings. A similar
b)
matrix effect (IS % in the range 50–68 %) occurred at 60,000 and 100,000 RP, whereas at 15,000 and 30,000 RP the PLTX signal was totally suppressed by the matrix at 25 ng mL−1 and linearity was poor (
