Degree Project C in Chemistry, 15c Comparison of matrix effects for clean-up methods for determination of PSP toxins in bivalve mollusks with HILIC UPLC-MS/MS

(1)

Comparison of matrix effects for clean-up methods for determination of PSP toxins in bivalve mollusks with HILIC UPLC-MS/MS

Sofia Blom 18 juni 2015

Supervisor: Heidi Pekar Subject specialist: Per Sjöberg

(2)

Sammandrag

Produktionen av bivalva blötdjur är en viktig industri i Europa. Odlade skaldjur är ett bra val av livsmedel, både ur miljö- och näringssynpunkt. Dock finns det risk vid intag.

Algblomning i haven kan leda till att marina biotoxiner, däribland PSP toxiner, acku- muleras i skaldjuren. Om människor konsumerar musslor eller ostron kontaminerade med dessa neurotoxiner kan det leda till förlamning och även dödsfall.

Den officiella kontrollmetoden inom EU för PSP toxiner i musslor är ett mustest med tveksam detektionsgräns. Det är därför av intresse att utveckla en kemisk analysmetod som är känslig och reproducerbar för dessa toxiner. Rapporten fokuserar på att analy- sera matriseffekter hos tre olika typer av bivalva blötdjur, blåmussla, hjärtmussla och ostron. Matriseffekter är ett fenomen som förekommer vid analys med elektrospray masspektrometri. Matriseffekter orsakar att signalen från toxinerna förstärks eller för- svagas av den matris som analyten befinner sig i.

Två olika uppreningsmetoder för det ändamålet har utförts och jämförts i den här rapporten. Masspektrometri kopplat till vätskekromatografi har använts som analysmetod för alla prover för båda uppreningsmetoder. Den första uppreningsmetoden gav vari- erande resultat och den andra uppreningsmetoden gav nästan enbart försvagning av resultaten då den inte kunde utföras fullständigt. Variation mellan arterna uppstod ock- så. Följaktligen kan inte samma standard för matriserna användas för de olika arterna vid framtida analyser.

(3)

Acknowledgements

This thesis is the result of my degree project C in chemistry of my bachelor degree at Uppsala University. It was performed at the National Food Agency in Uppsala, from 23 of March to 25 of May 2015. The orientation is analytical chemistry, and also the corresponding Institution at Biomedical Centre (BMC), Uppsala University.

I got the opportunity from my mentor Heidi Pekar (PhD. Senior Chemist) and her colleagues Annette Johansson (PhD. Senior Chemist) and Siv Brostedt (Laboratory Engineer) to perform my degree project C at the chemistry department, who I would like to thank for their great guidance, patience and support through the whole project.

I got the chance to participate in a project funded by the Swedish Scientific Research Council. I would like to thank Per Sjöberg as a subject specialist for reading my report and inspect my topic. I would like to thank the staff at the chemistry department at the National Food Agency for their interest in- and for discussing my task. I would like to thank the institution of analytical chemistry at BMC for lending of chemicals. I would also like to express gratitude to Heidi Pekar for her commitment in my project and her supervision, feedback and advices.

To my acknowledgements, I would also like to put my thesis in the real context and project that it is a part of. The project was announced by EFSA, and delegated from the Swedish Scientific Research Council to my mentor Heidi Pekar. The title of that project is: Accelerating European legislation for control of PSP toxins in shellfish: replacing more than 20 000 in vivo rodent tests with chemical methods using LC-MS/MS, LC-q TOF and rapid tests", Project number: K2015-79X-22646-01-3.

(4)

Abbreviations and list of words

PhD Doctor of philosophy

EFSA European Food Safety Authority

WHO World Health Organisation

PSP toxins Paralytic Shellfish Poisoning toxins

WHO World Health Organisation

Bioaccumulate Process where environmental toxins are stored in organisms

TEF Toxic Equivalence Factors

Analyte Compound in a sample which is supposed to be analysed

Matrix All content in sample except analyte

Extraction Process where compounds of sample is isolated from a mixture

SPE Solid Phase Extraction

Cartridge Column for sample clean-up

Eluate Sample content after SPE eluation

HLB Hydrophilic Lipolytic Balanced

Vacuum mainfold Box with ventilators used for SPE

UPLC Ultra Performance Liquid Chromatography

MS Mass spectrometry/Mass spectrometer

HILIC column used for Hydrophilic Interaction Liquid Chromatography

ESI Electrospray

Gradient Elution Process where mobil phases for LC are gradually changed in elution concentration

Chromatogram Signal vs. time from a chromatographic separation

Peak area Area of detected signals in a chromatogram

Retention time Time from start of analysis until detection for a compound

(5)

Table of content

1 Introduction 6

1.1 PSP toxins . . . 6

1.2 Analytical methods . . . 7

1.3 HILIC UPLC-MS/MS method. . . 7

1.4 Matrix effects . . . 8

2 Experimental 8 2.1 Chemicals and consumables . . . 8

2.2 Instrumentation . . . 9

2.3 Sample Preparation . . . 9

2.4 Sample clean-up . . . 10

2.5 Extraction and SPE clean-up for method 1 . . . 11

2.5.1 Sample extraction . . . 11

2.5.2 Optimised graphitised carbon SPE clean-up . . . 11

2.6 Extraction and SPE clean-up for method 2 . . . 12

2.6.1 Sample extraction . . . 12

2.6.2 Oasis HLB SPE . . . 13

2.6.3 Activated carbon SPE . . . 13

2.7 Preparation of spiked samples . . . 14

2.8 Analysis using HILIC UPLC-MS/MS . . . 14

2.8.1 Sample analysis . . . 16

2.9 Calculation of results . . . 16

3 Results 18 3.1 Interpretation of figures . . . 18

3.2 Method 1: analysis with LC-MS/MS . . . 19

3.3 Method 2: analysis with LC-MS/MS . . . 22

4 Discussion 24 4.1 Sample extraction . . . 24

4.2 Clean-up steps for method 1 and 2 . . . 25

4.3 LC-MS/MS method . . . 26

4.4 Postcolumn Infusion trial . . . 26

5 Conclusion 27

6 References 27

7 Appendix 29

(6)

1 Introduction

1.1 PSP toxins

The production of shellfish in Europe amounts up to 800 000 tons yearly^[1]. That includes mainly bivalve mollusks, such as mussels, oysters and cockles. The World He- alth Organization (WHO) predicts a growing demand and cultivation of shellfish. A growing industry also implies increased control of marine biotoxins in shellfish. This thesis will focus on Paralytic Shellfish Poisoning toxins (PSP toxins). Such neurotoxins consists of several variants, Figure 1 show those which have been focused on in this report and Table S1 in appendix include a detailed list of the names. They are produced by unicellular algae, e.g. dinoflagellates. Toxin producing algae occur in all kinds of water environments. The areas of interest in Sweden are coastal waters in Skagerack and Kattegatt, where the Swedish cultivation of bivalves is located. Algae blooms have been reported from early history but is accelerated by excessive fertilization of the water environment caused by discharge from waste water and agriculture^[1]. Most likely, the problem with incidents of human poisoning will also increase due to more frequent harmful algae blooms as an effect of overfertilization^[2].

Figure 1. Picture of the toxin compounds. A) and B) show the main part of the mo- lecule. Specific sidechains for the different variants are shown in the tables.

PSP toxins bioaccumulate in mollusks, after they have filtered toxic water with algae.

Symptoms of PSP toxins poison in human starts with numbness around the mouth and the most severe cases leads to paralysis and death^[2]. It can therefore be a serious ha- zard to public health. The control of shellfish produced all over the world is therefore mandatory and very important^[3]. The maximum residue limit (MRL) for PSP toxins in

(7)

bivalve mollusks is 0.8 mg STX · 2HCl eq/kg^[3]. Since PSPs are varied in their toxicity, the concentrations of toxin in foods etc. are recounted with TEFs (Toxic Equivalence Factors) based on the most toxic compound, Saxitoxin (STX)^[3].

1.2 Analytical methods

A mouse bioassay is currently the reference method for analysis of PSP toxins in shellfish. Due to both ethical and technical issues, this method is about to get replaced.

Methods that are more sensitive, reproducible and have less interferences are of certain interest^[4,5]. Chemical analysis using LC-MS/MS can be applied for the purpose^[4,5]. However, it is well known that analysis of PSP toxins in shellfish is afflicted by matrix effects causing loss of sensitivity^[4,5].

The aim of this project follows. The matrix effects from different types of bivalve mollusks using two different clean-up methods on 9 different PSP toxins will be investigated. HILIC LC-MS/MS will be used for detection and separation. Both clean-up methods include solid phase extraction (SPE). The first method performed and presented (mentioned as method 1) according to Boundy et al.^[4]involves optimised graphitised carbon SPE clean up with ASPEC and the second method (mentioned as method 2) according to Sayfritz et al.^[5]utilizes two SPE steps. The first with an Oasis hydrophilic lipophilic balanced (HLB) column and the second one with activated carbon column.This project is delmited to the extent that a full validation will not be imple- mented. Recovery test, linearity and batch-to-batch reproducibility will not be tested.

Furthermore, no unknown PSP toxin will be identified.

1.3 HILIC UPLC-MS/MS method

Ultra Performande Liquid Chromatography with a hydrophilic interaction column (HI- LIC UPLC) coupled to a tandem mass spectrometer (MS/MS) with an electrospray ionization (ESI) technique with triple quadrupole is the chosen instrumentation for chemical analysis of the analytes (substances of interest to be investigated) in biologi- cal matrices, such as bivalve mollusks. The matrix of a sample is all remaining content except for the analyte. The analytes in this project are 9 individual PSP toxins, see fig 1. The HILIC column that will be used includes a hydrophilic stationary phase. It has, consequently, other properties than the more common reversed-phase column. The HI- LIC column is well suited for separation and determination of polar compounds, hence it gives great retention of the PSP toxins^[4,5]. For HILIC columns, the eluent strength is increased when the fraction of water is increased in the mobile phase during gradient elution^[6]. Mobil phase of both water and organic character will be used, which will give efficient ionisation. Those are both requirements for the use of ESI. Quantitative detection with mass spectrometer offers many advantages. It is selective and sensitive, compared to e.g. flourescence detection^[5,7]. The matrix effects of the mollusks on the toxins are the main part to be investigated with this project and presented in this report.

The matrix effects will be compared using analysis with HILIC LC-MS/MS with the

(8)

two different clean-up methods. Matrix effects will also be compared between individual PSP toxins.

1.4 Matrix effects

Matrix effects occurs when compounds from the matrix is coeluting with molecules of the analyte. This will change the efficiency of the ionization for the ESI. Experiments have shown that matrix effects are the result of competition between nonvolatile matrix compounds and analyte ions^[7]. This occurs more specifically in the interface of ESI, on the droplet surface where parts of the sample injected goes from liquid to gas phase.

The concrete reason of why this happens is unclear. Depending on the ionization and ion evaporation, it can both lead to an increased and decreased formation of analyte ions in the interface^[7]. Consequently, the detected amount of a certain analyte will show to be too high or too low compared with the real amount. When there is an increase, it is called ion enhancement and when there is a decrease, it is called ion suppression. Thus, it has shown to be of major concern for the detection of a certain analyte which matrix it it is analysed in^[7].

2 Experimental

2.1 Chemicals and consumables

For both method the following chemicals were used. 100 % acetic acid (anhydrous for analysis, EMSURE), 98 % Ammonium acetate (for analysis), 99 % Na3C6H5O7 · 2 H2O (tri-sodium-citrate-2-hydrate) (pro analysis), 98 - 100 % Formic acid (pro analysis) and 99.9 % Methanol (LiChrosolv hypergrade for LC-MS) were all purchased from Merck. 99 % Na2WO4· 2 H₂O (sodium salt of tungstic acid) and also Optimi- sed graphitised polymer carbon Supelco ENVI-Carb 250 mg/3 ml cartridges (used for solid phase extraction) were bought from Sigma-Aldrich. Milli-q water was produced by a Milli-q purification system from Millipore (Billerica, MA, USA). Acetonitrile was from Fisher Scientific (Loughborough, United Kingdom). 28 - 30 % Ammonium hydroxide (ammonia basis) was bougth from BioChemika FLUKA. Oasis hydrophilic lipophilic balanced (HLB) SPE (pore size 80 Å, particle size 30 µm) 60 mg cartridge was purchased from Waters. Spin-X centrifuge tube 0.2 µm cellulose acetate filters was from Corning Inc. Supelclean ENVI - Carb SPE Tubes 3 ml, 250 mg cartridges was purchased from Supelco. 99.8 % Dichloromethane Pestican LAB-SCAN was from Analytical Sciences. Reference toxin standards for the different toxin variants to be investigated were dcSTX, STX, C1/C2, dcGTX2/dcGTX3, GTX2/GTX3 and GTX5. A detailed list of the toxins are presented in Table S1 in appendix. They were purchased from the National Research Council of Canada (NRC, Halifax, Nova Scotia, Canada).

BD Falcon tubes High-Clarity Polypropylene Conical Tubes of 15 resp. 50 ml from Francal Lakes, USA were used.

(9)

2.2 Instrumentation

A Retsch GM 200 Grindomix from Retsch were used for mixing mollusk sample to a batter. Ultra-Turrax Silent Crusher M (Heidolph) and 1500 Shaqer SPEX shaker: SPEX Sample Prep were used for homogenisation during sample extraction. Heraeus Multi- fuge 3SR+ Thermo Scientific centrifuge was used for centrifugation of Falcon tubes.

GX-274 ASPEC from GILSON and Vacuum mainfold for sample clean-up. Heraeus Sepatech Biofuge 13 was used for centrifugation and filtration for volumes below 1 ml. The LC-MS/MS system used for analysis was a Waters Xevo TQ-S UPLC-MS/MS system.

2.3 Sample Preparation

Approx. 1 kg of raw material from mussel, cockle and oyster were shucked and the contents were separetly placed in a mixer. The species of the raw material are shown in Figure 2. Homogenisation was performed for two minutes until the contents were turned into a smooth batter without lumps. The batter was split into samples aliquots of 4.9-5.1 g and put to 50 ml Falcon tubes. The samples were stored in a freezer until analysis. In total 9 samples were used for each of the two methods. The samples came from three different species according to Figure 2 and Table 1. Aliquots of the same 9 samples were used for both methods. A sample list of origin and treatment and the names in latin, english and swedish of the species are presented in Table 1.

Figure 2. Mussel (Mytilus Edilus)^[8], Cockle (Cerastoderma edule)^[9]and Oyster (Cras- sostrea gigas)^[10].

(10)

Table 1. Sample overview of species and origin. Three samples per species from different origins were used.

2.4 Sample clean-up

The clean-up part of the project were performed after two different methods. The first one was according to Boundy et al.^[4] and is called method 1. The second one was according to Sayfritz et al.^[5]and is called method 2. A procedure overview of method 1 and 2 is presented in Figure 3.

Figure 3. Overview of method 1 and 2.

(11)

2.5 Extraction and SPE clean-up for method 1

2.5.1 Sample extraction

An extraction solution of a total volume of 500 ml was prepared with 5 ml of acetic acid/milli-q water (1:100 v/v) with automatic pipette and graduated cylinder. Frozen sample aliquotes of 5 g were taken from the freezer and thawed in a water bath at room temperature.

3 ml of the extraction solution was added to the samples, which were then shaken for 5 minutes in a Spex shaker. The samples were then boiled for five minutes. The samples were cooled in a water bath afterwards and then shaken again in the Spex shaker for an additional five minutes. After that, the samples were centrifuged for 10 minutes on 4500 rpm (2422 g) at 20^◦C.

The extracted supernatant was pipetted to a smaller 15 ml Falcon tube. 3 ml of the extraction solution were then added to the remaining pellet and the procedure was repe- ated again, except for the boiling of the samples. The supernatant were then combined and diluted to a total volume of 10 ml.

2.5.2 Optimised graphitised carbon SPE clean-up

The clean-up step in the method was performed with solid phase extraction by ASPEC, where the Envicarb optimised graphitised carbon cartridges were used. The APSEC instrument that was used is presented in fig 3. The cartridges were conditioned first with 3 ml of acetonitrile/milli-q water/acetic acid (20:80:1 v/v/v). Secondly, they were conditioned with 3 ml of milli-q water/ 28-30 % ammonium hydroxide 1000:1 v/v). 400 µl of each sample extract was applied to the SPE catridges which then were washed with 700 µl milli-q water. Elution of the SPE cartridges was performed using 2 ml acetonitrile/milli-q water/ 100 % acetic acid (20:80:1) v/v/v). 100 µl of each sample eluate were put to vials and diluted with 300 µl acetonitrile and stored at 4^◦C, until further analysis with LC-MS/MS. All solutions were prepared with automatic pipette and graduated cylinder.

(12)

Figure 4. The instrument that was used for optimised graphitised carbon SPE clean-up.

2.6 Extraction and SPE clean-up for method 2

2.6.1 Sample extraction

A solution of a total volume of 500 ml was prepared with acetonitrile/milli-q water (80:20, v/v) with 0.1 % formic acid using a measuring cylinder and an automatic pipette. The homogenised samples were brought to room temperature before starting the experiment. There were three samples of each species (cockle, mussel and oyster). 10 ml of the extraction solution were added to each sample and homogenised for 5 min at 10.000 rpm with Silent Crusher M (Heidolph).

(13)

The Falcon tubes with sample aliqoutes were cooled with a water bath during homogenization, to avoid evaporation of the acetonitrile. Some heat was excessed during the process. The samples were then centrifuged for 15 min at 3500 rpm and the supernatants were stored in freezer for - 20^◦C over night.

The supernatants had formed two layers, one acetonitrile phase and one water phase containg the toxins. The upper, acetonitrile, phase was discarded from all samples and was not analysed. The remaining lower phases of the samples were evaporated with a flow of nitrogen gas in a heating block at 40^◦C. When the volume was reduced to just below 3 ml, the samples were then volume adjusted to 3 ml with milli-q water. The samples were stored in freezer at -20^◦C until clean-up with SPE was performed.

2.6.2 Oasis HLB SPE

The samples were taken from freezer for thawing. A solution of 5 % sodium salt of tungstic acid was made according to the following description. 500 mg of the salt NaWO4was weighted on automatic scales and diluted with 10 ml milliq-water measu- red with automatic pipette. 500 µl of each sample was treated with 50 µl of 5 % sodium salt of tungstic acid to precipitate the proteins. A vacuum mainfold with a water pump was used to perform the first step of solid phase extraction with the Oasis HLB SPE cartridges. They were conditioned with 1 ml MeOH followed by 1 ml milli-q water and then dried. 450 µl of the samples were loaded on the cartridges, and filtered through the SPE column. Centrifugation and filtration with Spin-X filters were then performed for 2 min at 10.000 rpm. Sample eluates were stored at -20^◦C for freezing over night before the second SPE step.

2.6.3 Activated carbon SPE

Sample eluates from the first clean-up step with Oasis HLB columns were put at room temperature from freezer for thawing. The cartridges for SPE with activated carbon were conditioned with the following solutions when they had been placed on the vacuum manifold. 5 ml dichloromethane/methanol (80:20 v/v) followed by 4 ml methanol and at last, 5 ml 0,5 % sodium salt of tungstic acid. 400 µl of each thawed sample eluate from Oasis HLB SPE were loaded with automatic pipette on the cartridges. The cartridges were then washed with 2 ml of 0,01 M ammonium acetate solution. The toxins were eluted with 2 ml of 0.5M citrate buffer in 10 % acetonitrile. The sample eluates were concentrated in a heating block at 40^◦C with a flow of nitrogen gas until approximately 300 µl were left. The samples were stored in freezer at -20^◦C until analysis.

The consistency of the sample eluates was very thick. Therefore, the same procedure with Oasis HLB and Activated carbon SPE was performed for a sample with milliq- water. Similar consistency on the eluate was received. The sample was tested to be solved in water, acetonitrile and methanol without result. Furthermore, the sample eluates that had gone through both steps of SPE were not used in further analysis. Sample

(14)

eluates that only had undergone the first clean up step, with Oasis HLB, were used for analysis.

2.7 Preparation of spiked samples

All sample eluates from method 1 and 2 and also sample extracts from method 1 were spiked by adding 20 µl standard solution to 180 µl sample extract or eluate. The sample extracts are samples that not have gone trough SPE clean-up with ASPEC. Two different concentrations of standard solution were prepared, presented in Table 2. The one with low concentration was added to the sample extracts and the one with high concentration was added to the sample eluates. The solutions contained a mix of the toxins dcSTX, STX, dcGTX2/dcGTX3, C1/C2, GTX2/GTX3 and GTX5 in acetonitrile, at high and low concentration according to Table 2. The concentrations are adjusted after the limits that are approved in foods, according to the National Food Agency. Samples with only solvent (milli-q water:acetonitrile, 1:3 v/v) were prepared and spiked with standard solution. All standard solutions were prepared using automatic pipettes.

Table 2. The concentration of each individual toxin in the standard mixes. The mix with Low concentration will be used for sample extract. The mix with High concentration will be used for sample eluats, after clean-up SPE.

2.8 Analysis using HILIC UPLC-MS/MS

All samples, both extract, eluates, and reference standards in solvents obtained from both methods were analyzed with the same HILIC UPLC-MS/MS method with additional blank samples containing only solvent. The HILIC UPLC-MS/MS method according to Boundy was used. Everything was performed in the same instrument, shown in fig 4. The following mobile phases were used for analysis and gradient elution, A1:

formic acid/ 28-30 % ammonium acetate/milli-q water (0.075:0.03:700 v/v/v) and B1:

formic acid/milli-q water/acetonitrile (0.01:300:700 v/v/v). In addition, a solution of formic acid/milli-q water (1:200 v/v), and a solution of methanol/milli-q water (1:1 v/v) was used for washing the HILIC column and the LCMS/ MS system after analysis. All solutions were prepared with measuring cylinder and an automatic pipette.

(15)

Figure 5. The LC-MS/MS instrument that was used and the interface.

The liquid chromatography was performed on an ACQUITY UPLC system (Waters, Manchester, United Kingdom). An ACQUITY UPLC BEH Amide column, 2.1 x 150 mm was used with a VanGuard ACQUITY UPLC BEH Amide pre-column, 2.1 x 5 mm, (Waters, Manchester, United Kingdom). The particle size for both was 1.7 µm.

The injection volume was 10 µl and the column temperature was 60^◦C. The gradient elution is shown in Table 3.

(16)

Table 3. Schematic figure for the concentration gradient used for the liquid chromatography.

Analysis was performed in dynamic Multiple Reaction Monitoring mode using a triple quadrupole mass spectrometer (MS/MS), Xevo TQ-S fromWaters (Manchester, United Kingdom). For ionization, the mass spectrometer was set to both positive and negative electrospray (ESI+, ESI-) depending on which compound that was detected. STX and dcSTX were analysed in ESI+ and C1/C2, dcGTX2/3, GTX2/3 and GTX5 were analysed in ESI-. The capillary voltage was 3.0 kV. The source offset was 50 V and the source temperature 150^◦C. The desolvation and cone gas had flows of 650 and 150 L/Hr and N2-gas generated from pressurized air was used for that. Desolvation gas temperature was 600^◦C. N2-gas was also used as nebulizing gas, at a pressure of 7.0 bars. Argon (Alphagas^{T M}, Malmö, Sweden) was used as collision gas at a flow of 0.15 ml/min. In Table S1 in appendix, the specific mass spectrometric parameters for each compound are presented.

2.8.1 Sample analysis

All spiked sample eluates that had been obtained from method 1 were analyzed with HILIC LC-MS/MS with additional samples containing standard mixes of the toxins in solvent. Also the sample extracts from method 1 (before SPE clean-up) were analysed.

All spiked sample eluates from method 2 with samples containing standard mixes of the toxins in solvent were also analysed. To clarify, it was sample eluates that only hade gone through the first clean-up step since eluates after the second step were not possible to analyse.

2.9 Calculation of results

The %-results were calculated from the peak areas from the chromatograms, as follows in equation 1. Calculations of matrix effect in the various species were done with the received %-results for each of the 9 toxins.

(17)

P eak area of analyte standard

Conc. of standard in solvent = P eak area in sample

Standard conc. in sample× M atrix ef f ect (1) The peak areas from standard are divided with the concentration in the standard, which is assumed to be 100 % and the same for all samples as well. The peak areas from the samples are also divided with the concentration in the standard. The added volume is the same for both standards and samples. Both quotients in equation 1 becomes the response factors. The first is response factor for standard and the second is for sample.

Since the denominator is the same for both response factors, the equation can then be rewritten as follows in equation 2.

P eak area of solvent

P eak area in sample = M atrix ef f ect (2) The values for matrix effects are calculated with equation 3 in accordance with Stahnke et al^[11]. Three sample values of each species (cockle, mussel and oyster) were representative for further calculations, total 9 sample values, to calculate the response factor for sample. To calculate the response factor for standard, 6 solvent values was used.

The right character (enhancement or suppression) of the effects was also received from equation 3.

( Rfsample

Rf_standard× 100) − 100 = M atrix ef f ect [%] (3)

The standard deviation for the matrix effect for each species and the solvent was calculated, using equation 4.

σ =r P(x − m)²

n (4)

The standard deviation within each species and solvent was calculated with equation 4. Where σ is the value of the standard deviation,P is the sum of each set to the right.

x is a single value of result within each species, m is the mean of all results of the species and n is the number of results, n=3. In this case, three different results from each species.

Diagrams for each toxin were drawn with the values for the means of matrix effects of the 3 species. The standard deviations within each species were also included in the diagrams. They are presented in paragraph 3.2.

(18)

3 Results

3.1 Interpretation of figures

The chromatograms of each toxin are presented in Figure 6. The response is showed on the y-axis. From the detected signals in solvent and mollusk sample, diagrams vi- sualising the matrix effects could be created. The diagrams in chapter 3 should be interpreted as follows. At no matrix effect the bar in the diagram is at zero. When the mass spectrometric signal for the studied compound is higher in the samples compared to the solvent it is called enhancement. Consequently, bars in > 0 in the diagrams represents species generating enhancement as matrix effect. Supression is visualized in the figures when the bars in the figures are < 0. Supression occurs when the mass spectrometric signal for the compound studied is lower in the sample compared to the solvent. The black lines on each bar shows the standard deviation for each species and toxin (n=3). Raw data for the diagrams are shown in appendix, table S2-S8 for the diagrams presenting the method according to Boundy and tables S9-S17 for the diagrams presenting the method according to Sayfritz. The latin names are used in the following diagrams. The english and swedish names for each species is presented i Table 1 in paragraph 2.3.

(19)

Figure 6. The chromatograms for the nine different toxins. Each peak represents a toxin and the time is presented in minutes.

3.2 Method 1: analysis with LC-MS/MS

The crude sample extracts (before clean-up with SPE) were analysed with HILIC UPLC-MS/MS. The resulting chromatograms were not possible to interpret. The matrix effects were so strong that the signals from all toxins were completely suppressed and

(20)

they were not detected. These results were as expected, since the crude sample extracts had not passed through the clean-up step of the method^[9].

The matrix effects of the toxins in the eluats from the mollusk samples after SPE carbon clean-up were visualised in diagrams. A summary of all diagrams, according to each toxin is presented in the following Figure 7. All diagrams are presented in Figure S1-S7 in appendix. Raw data for the diagrams are presented in table S2-S8 in appendix. The visualised matrix effects on dcSTX, STX, dcGTX2, dcGTX3, C1, GTX3 and GTX5 in the different species after elution using the active carbon based SPE cartridges are shown. Of the 9 toxins, 7 were representative for further investigation and possible to calculate the matrix effects for. C2 and GTX2 were excluded because of insufficient results.

(21)

(22)

In general, mussel have less matrix effect than the other two species for the toxins. For all toxins, with STX as an exception, the matrices cause either enhancement or suppression, regardless of species studied. STX and dc STX carries two positive charges.

dcGTX2, dc GTX3, GTX3 and GTX5 carry one positive charge even before ionization in the mass spectrometer. However C1 is a neutral compound. All toxins was affected by enhancement in all matrix except dcSTX that showed suppression in oyster and cockle, STX in mussel and C1 that suffered suppression in all species, see Figure 7.

Clearly, the charge of the toxins can not alone explain the findings.

The character of the matrix effect could also depend on the ionization mode and the method utilized Electrospray in positive and negative mode (ESI+ or ESI-). dcGTX2, dcGTX3, GTX3, GTX5 and C1 are all ionized in ESI- and all of them except C1 show enhancement. Interestingly, they also have similar degree of matrix effect. STX and dcSTX ionize in ESI+, they both have similar suppressing matrix effects for oyster and cockle. All these compounds can be ionized in both ESI+ and ESI- and it would have been interesting study the matrix effects in both ionization modes for all compounds.

However, it was beyond the scope of this report.

The mobile phase of the HILIC column can be considered concerning the retention order of the toxins. Concentration gradient were used for the chromatographic separation. It started with a big fraction of solution B (formic acid/milli-q water/acetonitrile (0.01:300:700 v/v/v) with acidic character and followed by a 50:50 fraction of B and A (formic acid/ 28-30 % ammonium acetate/milli-q water (0.075:0.03:700 v/v/v). Furt- her, the gradient ends with a big fraction of B. Of the two solutions, A has basic character and B acidic. Most of the toxins have a retention time around 6 - 7 minutes.

According to the concentration gradient in Table 3 on pg. 16 that occurs when the fraction of A is increasing to a 50:50 fraction between A and B. And also when the pH is increasing. C1 and C2 are the only two toxins that are eluated before, at 3 minutes when the pH still is lower. C1 and C2 are the only two toxins that have natural charge during separation, before ionization in ESI. The rest has one or two positive charges.

For some toxins the matrix effect varies significantely between the species. In general, cockle caused the most matrix effect.

3.3 Method 2: analysis with LC-MS/MS

Figure S8 - S16 in appendix show the matrix effects on the toxins for the different species visualised in diagrams. A summary of the diagrams are presented here in Figure 8. Raw data for all diagrams are presented in appendix, table S9-S17.

(23)

(24)

Sayfritz et al. is a two-step SPE method where a HLB cartridge is utilized before a carbon cartridge. Unfortunately the SPE step with carbon cartridges eluated a transparent oily substance, approximately 300 µl from the clean-up procedure. The transparent oily substance could not be dissolved in water, acetonitrile or methanol. The eluted substance is shown in Figure 22. Hence, it was not possible to analyse the sample on reversed phase HILIC-LC-MS/MS.

Figure 9. One of the sample eluats after clean-up with activated carbon column.

Consequently, the analysis were performed on samples that have been cleaned up with only the HLB cartridges.

4 Discussion

4.1 Sample extraction

The sample preparation includes homogenization of the raw material and splitting of the sample into aliquots before extraction. For the method according to Boundy et al., the extraction was performed with acetic acid in Milli-Q water and for the method according to Sayfritz et al., with a mixture of acetonitrile and formic acid in Milli-Q.

The toxins are polar and hydrophilic compounds, which is partly similar with acetic or formic acid in water. They are therefore very soluble in these solvents.

Since organic solvents are used for extraction in the Sayfritz et al. method more matrix is co-extracted such as fats and proteins compared to Boundy et al. Therefore, the

(25)

method according to Sayfritz et al. has a freeze-out step where the upper layer mostly contain acetonitrile and the lower contain water. The majority of the PSP toxins remain in the lower water layer while the upper acetonitrile phase contains more hydrophobic compounds such as fats and proteins. The upper layer was discarded after the freeze out step. However, the two layers created, were quickly mixed again when the sample was taken out of the freezer. Therefore, the freeze out step required a high degree of meticulousness and is not suitable for large sample series.

4.2 Clean-up steps for method 1 and 2

The clean-up step for method 1 includes that the sample extract are eluted through a column with graphitised carbon. Most of the matrix, such as fat and protein, are adsor- bed on the column, and the toxins are eluated.

For the clean-up steps for method 2, the sample extracts are first eluted through an Oasis hydrophilic lipophilic balanced (HLB) column and then a column with activated carbon. In general, suppression matrix effects were observed for method 2. It is expected, since the SPE HLB is less powerful than the SPE carbon cartridge in removing matrix from the sample. The samples were not sufficient cleaned and diluted for the toxins to be unaffected by the matrix. Unfortunately, the sample eluates received from the step involving the carbon SPE column could not be analysed, see Figure 9. Because they are not possible to dissolve. However, it is apparent that clean-up using the HLB cartridge is not enough since the samples showed severe suppressive matrix effect. Using only the HLB catridge would have affected the LOD (lowe limit of detection) of the method negatively. This is visualised in paragraph 3.3, where the sample matrices cause severe suppression. That effect would probably have been less prominent if the complete clean-up was possible. The second clean-up step with activated carbon was tested with milliQ-water as “sample”, resulting in the same oily and insoluble substance that was observed in the eluat.

C1, C2, STX, dcSTX, GTX2 and GTX3 showed similar degree of suppression for all species (60-100%). The only cases of enhancement are in mussel, for GTX5 and dcG- TX3. GTX5 and GTX3 did suffer suppression, much like the other toxins in oyster and cockle. Results are similar to those from Zhuo et al.^[9].

In general, mussel cause the least matrix effect of the species and cockle the most.

To sum up, the clean-up for method 1 achieved cleaner samples and more reliable results than the clean-up for method 2. But they would have been more comparable if both methods would have been completed. The SPE HLB is less powerful than the SPE carbon cartridge in removing matrix from the sample. That makes it a crucial part of the clean-up.

(26)

4.3 LC-MS/MS method

Matrix effects were visualised for the toxins in the matrices of the species of bivalve mollusks when comparing the signals for the mollusk samples and the solvent. The increase (enhancement) or decrease (suppression) of the detected toxin signals, the matrix effects, of the mollusk samples were easily monitored with the LC-MS/MS method. The LC-MS/MS method became slightly unstable when analyzing the dirty eluates from the HLB cartridge in the Sayfritz method resulting in larger standard deviations. It also led to that the retention time shifted for some of the toxins. However, they were still easily identified. Therefore it was concluded that the HILIC column se- ems sensitive to dirty sample extracts.

The analysed signals from toxins are assumed to only proceed from the added amount of toxin from the standard mix. Natural toxin concentrations in the shellfish matrices are possible. All collected samples are though bought from grocery stores and are consequently supposed to have been controlled to not contain any toxins.

This project concerns the matrix effects for clean-up methods for determination of PSP toxins in bivalve mollusks with HILIC UPLC-MS/MS. It is a part of a project that is announced by EFSA. It is furthermore delegated from the Swedish Scientific Research Council. The aim of the project is to accelerate European legislation for control of PSP toxins in shellfish. The current reference method that is used is an in vivo rodent test.

It is painful for more than 20 000 mice every year. The method is also not as sensitive as the UPLC-MS/MS method. Therefore it is of interest to replace the mouse bioassay and use chemical analysis, with MS detection instead.

4.4 Postcolumn Infusion trial

During the experiment, postcolumn infusion was tested. It was performed the following way. Instead of spiking samples and blank solvent samples, the standard toxin mix was infused (injected) directly into the interface of the mass spectrometer. It is ionized in the same way as the samples. The difference is that the unspiked samples have been separated in the LC-system and the compounds of the standard have not. Standard- and sample ions are detected in the same way. Infusion is used to compensate for matrix effects, and quantification of matrix effects is then also possible.

This was performed with a couple of standard mixes with varied concentrations. It was shown when interpretting the chromatograms, that no representative results were received. It was visualised that infusion solution only had reached the detector in the beginning. Not fluently during the injection of the sample into the interface. This could depend on the pressure from which the solution is transported in the system. It could be too low for the infusion solution to inject during the whole time of analysis.

Similar effect of infusion, as for spiking samples with standard mixes, could be expected. When toxin ions from the standard mix are detected at the same occasion as the sample ions, the matrix effect of the sample is visualised of a weaker signal of the

(27)

toxin. The toxin concentration and amount that is injected is known and constant. And the matrix, as for this experiment, is complex and contains unknown compounds. And will consequently affect the toxin signal^[11].

5 Conclusion

To sum up, the resulting matrix effects showed from the method according to Boundy et al. showed larger variation between toxins compared to the method according to Sayfritz. The sample eluats became cleaner using the first method. The results received from the method according to Sayfritz showed almost only suppression which points at that a second step of clean-up is useful. Matrix effects diminishing the signals of the toxin, which is expected^[9]. The Sayfritz et al. method requires more work and takes longer time to perform compared to Boundy et al. Also in the Sayfritz et al. method the freeze out step requires careful sample handling, limiting the number of samples that can be handled at the same time.

For the Boundy method the matrix effect differ between enhancement and suppression for the species. Therefore, the calibration standards used for quantification have to be diluted in matrix for the specific species that is being analyzed to ensure accurate results. The matrix effects observed for the toxins in mussels, cockle and oysters are too diverse and large for applying a general matrix standard or standard in solvent.

6 References

[1] Portabales Longa A. (2012). Bivale Aquaculture in Europe. http://www.oysterworldcongress 2012.com/telechargements/telechargements-2/?lang=en. Arachon Bay, France: Presen- tation 2012.

[2] Lidman, U. (2008). Toxikologi - Läran om gifter. Pozkal, Polen: Studentlitteratur.

[3] Marine biotoxins in shellfish - SUmmary on regulated marine biotoxins. Scientific Opinion of the Panel on Contaminants in the Food Chain.(2009) The EFSA Journal 1306.

[4] Boundy, M.J. et al. (2015). Development of a sensitive and selective liquid chromatography–mass spectrometry method for high throughput analysis of paralytic shellfish toxins using graphitised carbon solid phase extraction(Journal of Chromatography A).

Nelson: Cawthron Institue.

[5] Sayfritz, S.J. et al. (2008). Determination of paralytic shellfish poisoning toxins in Norwegian shellfish by liquid chromatography with fluorescence and tandem mass spectrometry detection(Toxicon). Oslo: Norwegian School of Veterinary Science.

[6] Harris, D.C. (2010). Quantitative Chemical Analysis. Eighth Edition. The United States: W.H Freeman and Company

[7] Taylor, P. J. (2005). Matrix effects: The Achilles heel of quantitative high-performance liquid chromatography–electrospray–tandem mass spectrometry(Elsevier). Brisbane:

Princess Alexandra Hospital.

(28)

[8] All-fish-seafood-recipes.com. (2015). Mussels. http://www.all-fish-seafood-recipes.com /index.cfm/fish/mussels/ [2015-05-16]

[9] Wikimedia Commons. (2014). Cerastoderma edule. https://commons.wikimedia.org/wiki/Main_Page [2015-06-18]

[10] Wikimedia Commons. (2014). Crassostrea gigas. https://commons.wikimedia.org/wiki/Main_Page [2015-06-18]

[11] Stahnke H. et al. (2009). Compensation of Matrix Effects by Postcolumn Infusion of a Monitor Substance in Multiresidue Analysis with LC-MS/MS(Anal. Chem). Ber- lin: Federal Institute for Risk Assessment (BfR), Residue Analysis.

[12] Zhuo L. et al. (2013). Determination of paralytic shellfish poisoning toxins by HILIC-MS/MS coupled with dispersive solid phase extraction(Food Chemistry, Else- vier). Fuzhou: Fuzhou University.

(29)

7 Appendix

(30)

(31)

(32)

Figure S1. Matrix effects for the different sample types on dcSTX.

Figure S2. Matrix effects for the different sample types on STX.

(33)

Figure S3. Matrix effects for the different sample types on dcGTX2.

(34)

Figure S5. Matrix effects for the different sample types on C1.

Figure S6. Matrix effects for the different sample types on GTX3.

(35)

Figure S8. Matrix effects for the different sample types on dcSTX.

(36)

Figure S9. Matrix effects for the different sample types on STX.

(37)

(38)

(39)