Quantifying Inorganic Arsenic and Other Water-Soluble Arsenic

Apr 29, 2017 - Because the toxicity of arsenic depends on its chemical form, risk assessments of arsenic exposure must consider the type of arsenic ...
0 downloads 0 Views 731KB Size
Subscriber access provided by University of Colorado Boulder

Article

Quantifying inorganic arsenic and other watersoluble arsenic species in human milk by HPLC/ICPMS Michael Stiboller, Georg Raber, Elin Lovise Folven Gjengedal, Merete Eggesbo, and Kevin A. Francesconi Anal. Chem., Just Accepted Manuscript • Publication Date (Web): 29 Apr 2017 Downloaded from http://pubs.acs.org on May 3, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Graphical abstract 213x187mm (150 x 150 DPI)

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 21

Quantifying inorganic arsenic and other water-soluble arsenic species in human milk by HPLC/ICPMS Michael Stibollera, Georg Rabera*, Elin Lovise Folven Gjengedalc, Merete Eggesbøb, Kevin A. Francesconia a

Institute of Chemistry, NAWI Graz, University of Graz, 8010 Graz, Austria Norwegian Institute of Public Health, 0403 Oslo, Norway c Norwegian University of Life Sciences, Faculty of Environmental Sciences and Natural Resource Management, 1432 Ås, Norway b

*[email protected] Abstract Because the toxicity of arsenic depends on its chemical form, risk assessments of arsenic exposure must consider the type of arsenic compound, and hence they require sensitive and robust methods for their determination. Furthermore, the assessment should include studies on the most vulnerable people within a population, such as newborns and infants, and thus there is a need to quantify arsenic species in human milk. Herein we report a method for the determination of arsenic species at low concentrations in human milk by HPLC/ICPMS. Comparison of single - and triple quadrupole mass analysers showed comparable performance, although the triple quadrupole instrument more efficiently overcame the problem of ArCl+ interference, from the natural chloride present in milk, without the need for gradient elution HPLC conditions. The method incorporates a protein precipitation step with trifluoroacetic acid followed by addition of dichloromethane or dibromomethane to remove the lipids. The aqueous phase was subjected to anionexchange and cation-exchange/mixed mode chromatography with aqueous ammonium bicarbonate and pyridine buffer solutions as mobile phases, respectively. For method validation, a human milk sample was spiked with defined amounts of dimethylarsinate, arsenobetaine and arsenate. The method showed good recoveries (99 – 103%) with detection limits (in milk) in the range of 10 ng As kg-1. The method was further tested by analyzing two Norwegian human milk samples where arsenobetaine, dimethylarsinate, and a currently unknown As species were found, but iAs was not detected.

1 ACS Paragon Plus Environment

Page 3 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

INTRODUCTION Arsenic is a major environmental health problem for people in many parts of the world, mainly as a consequence of inorganic arsenic (iAs) in their drinking water. Inorganic arsenic is a known carcinogen1 that has also been associated with several additional health problems including heart disease2 and diabetes3. Although food generally has only low levels of iAs, recent studies have indicated that health problems related to iAs exposure can be evident at concentrations much lower than previously thought,4 and consequently, the contribution of food to humans’ iAs exposure is now under scrutiny.5

Investigations into iAs in food need analytical methods able to selectively measure iAs among the organic arsenic species found naturally in many foods. These organoarsenic

species

range

from

simple

methylated

species

such

as

methylarsonate (MA), dimethylarsinate (DMA) and arsenobetaine (AB) to more complex arsenicals such as arsenosugars and arsenolipids.6 In terrestrial foods, iAs is usually the major arsenic species and reliable quantitative methods based on HPLC and detection with inductively coupled plasma mass spectrometry (HPLC/ICPMS) have been reported.7 HPLC/ICPMS can also be used to determine iAs in foods of marine origin, although here the measurement is rendered more difficult owing to the large amounts of organoarsenic species usually present in the samples.6

Information about arsenic species are now being included in risk assessments of arsenic in food.9,8 A risk assessment of a contaminant in food must take into account sensitive groups within a population such as people with specific diets and children. Children’s exposure to elemental toxicants such as mercury is closely scrutinized because of the disproportionately large detrimental effects, relative to adults, seen in growing children.10 Exposure of children to arsenic is now being similarly investigated, for example in rice formulations for babies and young children.11 Very few studies, however, have investigated arsenic species in mothers’ milk, which is the major food for infants. One study,12 performed among populations in 2 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 21

Bangladesh with extremely high iAs intake through contaminated drinking water, reported the presence of iAs in milk.

A Swedish study,13 on the other hand,

indicated that the total arsenic content of human milk was low, and currently available analytical methods could not reliably determine the As species at those low concentrations.

To enable risk assessment of iAs exposure through milk, improved methods to determine the arsenic species in milk are needed. In this paper, we report a validated quantitative HPLC/ICPMS method including a novel sample preparation procedure for the determination of trace levels of iAs and water-soluble organoarsenic species in human milk.

3 ACS Paragon Plus Environment

Page 5 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

MATERIALS & METHODS Samples and Quality Control. In the absence of a milk sample certified for arsenic content, we used the following reference materials as part of our method validation for the determination of the total As content in human milk samples: (i) Serum Control lyophilized, Level 1 from RECIPE (Munich, Germany) with a mean total As concentration of 9.87 µg As L-1 (control range 7.90 – 11.80 µg As L-1); we obtained a total As content of 9.25 ± 0.24 µg As L-1 (n=10); (ii) an in-house reference sample of human milk donated from a healthy mother (and colleague) from Graz; we obtained a total As content of 0.17 ± 0.03 µg As kg-1 (n=6); and (iii) a Skimmed Bovine Milk Powder reference material, ERM®-BD150, obtained from LGC Standards GmbH (Wesel, Germany) without total As certification. We determined a total As content of 3.50 ± 0.26 µg As kg-1 (n=10) for the skimmed milk powder. In addition, we applied the method to two human milk samples collected in Norway as part of the Norwegian HUMIS-NoMIC Study.14,15 Chemicals, Reagents, and Standards. Water (18.2 MΩ cm) provided from a Milli-Q Academic water purification system from Millipore GmbH (Vienna, Austria) was used throughout this work. Nitric acid ROTIPURAN® 68%, p.a., further sub-boiled in a MLS duoPUR sub-boiling unit (MLS GmbH, Leutkirch, Germany), pyridine ROTIPURAN®

≥99.5%, p.a., formic acid

ROTIPURAN® ≥99.5%, p.a., aqueous ammonia solution 25% ROTIPURAN®, ≥25%, p.a.,

trifluoroacetic

acid

≥99.9%,

hydrogen

peroxide

30%

(w/w)

and

dichloromethane ≥99.9%, were purchased from Carl Roth GmbH & Co. KG (Karlsruhe, Germany). Dibromomethane 99% was purchased from Sigma Aldrich (Vienna, Austria), and ammonium bicarbonate ≥99.0%, purum p.a. was obtained from Fluka Chemie (Buchs Switzerland). For the determination of total arsenic in human milk samples, a single element standard containing 1000 ± 3 mg As L-1 in 2% nitric acid from CPI International (Santa Rosa, CA, USA) was used. Single element standard with a concentration of 1000 mg L-1 of germanium from Carl Roth GmbH & Co. KG was used as internal standard. For speciation analysis, standard solutions were prepared for: arsenobetaine (AB), previously synthesized in-house, with a purity >99% (by NMR and HPLC/mass spectrometry); dimethylarsinate (DMA) 4 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 21

prepared from sodium dimethylarsinate purchased from Fluka Chemie (Buchs, Switzerland); arsenate (As(V)) prepared from Na2HAsO4. 7H2O from Merck (Darmstadt, Germany); methylarsonate (MA) prepared in-house from As2O3 and CH3I (Meyer reaction); and dimethylarsinoyl acetate (DMAA)16 and dimethylarsinoyl propionate (DMAP)17 were prepared in-house. Arsenic Speciation Analysis. For method development and quality control, a human milk sample (see above, inhouse reference sample), with a low background arsenic level and no detectable AB and iAs, was spiked with 20 ng As each of AB, DMA and As(V) by adding 0.200 g of a solution containing 100 µg As kg-1 of AB, DMA and As(V) to 19.8 g of human milk giving a spiked concentration of 1 µg As kg-1 for each arsenic compound; this spiked sample served to mimic a human milk sample influenced by seafood and terrestrial food. Further, a reconstituted milk sample was prepared from ERM®-BD150 Skimmed Bovine Milk Powder (2.65 g of milk powder with a water content of 8%, were diluted with water to 20.0 g), and the milk solution (10.0 g) was spiked with 1.5 ng of As(V) to give a spiked iAs content of 0.15 µg kg-1, as a sample containing a low concentration of iAs. All samples were stored at -80°C. Determination of LOD and LOQ. The LOD and LOQ were determined for AB (cation-exchange/mixed mode), and for DMA, MA and As(V) (anion-exchange, gradient elution) by HPLC/ICPMS based on DIN 32645 (calibration method). Therefore, standards of AB, DMA, MA and As(V) were prepared in aqueous 0.1% (v/v) trifluoroacetic acid for matrix matching, in triplicate in the concentration range 0-0.200 µg As L-1 with an 8-point calibration in increments of 0.025; the calibration was based on peak areas. Instrumentation. Microwave-assisted acid digestions were performed with an UltraCLAVE III microwave system (MLS, Leutkirch, Germany). Total As measurements were performed with an Agilent 7900 ICPMS (Agilent Technologies, Waldbronn Germany) equipped with a Scott type spray chamber and a Micro Mist concentric glass nebulizer (Glass Expansion, West Melbourne, Australia). For HPLC/ICPMS 5 ACS Paragon Plus Environment

Page 7 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

measurements of As species, an Agilent Series 1100 HPLC system, equipped with solvent degasser, binary pump and thermostated autosampler and column compartment, was connected to the Agilent ICPMS 7900 with 0.125 mm PEEK (polyetheretherketone) tubing (Upchurch Scientific, Oak Harbour, USA). HPLC/ICPQQQ-MS measurements were performed with an Agilent Series 1200 HPLC system, equipped with a solvent degasser, quaternary pump, thermostated autosampler and column compartment, coupled to an Agilent 8800 triple quadrupole ICPMS (ICPQQQ-MS, Agilent Technologies, Waldbronn Germany). A high-speed refrigerated micro-centrifuge (SCILOGEX, Rocky Hill, CT, USA) and a Hettich Rotina 420 R centrifuge (Andreas Hettich, Tuttlingen, Germany) were used for the fractionation of arsenic species in human milk samples. Solvents were removed from the human milk fractions by using a Christ RVC 2-33 CDplus vacuum lyophilisator (Martin Christ, Osterode am Harz, Germany). Determination of Total Arsenic Content. Arsenic content was determined in the milk samples and in the various fractions of the sample preparation scheme by ICPMS following a microwave-assisted acid mineralization step. Thus, portions, weighed with a 0.1 mg precision, of human milk or quality control milk samples (each ca 1 g) or human milk fractions (varying amounts), serum (ca 0.2 g), or milk powder reference material (ca 0.1 g) were placed in 12 mL quartz tubes of the UltraCLAVE microwave digestion system. Then nitric acid (2 mL) and 1 mL of an internal standard solution containing 100 µg L-1 of Ge in 1% HNO3 (v/v) were added. The tubes were covered with Teflon® caps, placed in a Teflon® rack, and transferred to the microwave system; an argon pressure of 4 x 106 Pa was applied and the acid mixture was heated to 250 °C for 30 min. After mineralization and cooling to room temperature, the samples were diluted with water to 10 mL in the quartz tubes, sealed with Parafilm® M (Bemis Company Inc., Wisconsin, USA) and mixed. The digested samples were directly analysed for total As concentration by ICPMS. Arsenic measurements were made by using ICPMS with conventional nebulization and operated with helium (4.0 mL min-1) as collision cell gas for removing polyatomic interferences from argon chloride (40Ar35Cl on 75As), and introducing 1% 6 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 21

CO2 in argon as optional gas to the plasma to enhance the arsenic signal. One channel was measured per isotope. Monitored masses were m/z 75 (75As, 3 s/point), m/z 77 & 82 (to gauge possible interference of

40

Ar35Cl on

75

As while monitoring

possible Se contribution to m/z 77 (0.5 s/point), and m/z 72 and 74 (internal standard

72

Ge and

74

Ge (each 0.5 sec/point). Standards for calibration were

prepared in 15 mL polypropylene tubes (Greiner Bio-One International GmbH, Kremsmünster Austria) containing 20% HNO3 (v/v) for matrix matching and the internal Ge standard at a final concentration of 10 µg L-1; quantification was made from a blank plus nine-point calibration covering the range 0.01 to 1 µg As L-1. Fractionation of Arsenic Species in Human Milk. A portion of ca 1 g (weighed to 0.1 mg) of human milk was placed in a 2 mL polypropylene micro-centrifuge tube (Carl Roth, Karlsruhe, Germany); then 10 µL of an aqueous 10% (v/v) trifluoroacetic acid solution were added, and the mixture vortexed and let stand for 15 min to precipitate the proteins. For the removal of lipids, 500 µL of dichloromethane or dibromomethane were added and the mixture was vortexed and allowed to stand for 15 min. Afterwards, samples were placed in the high-speed refrigerated micro-centrifuge and centrifuged at 21380 g at 4°C for 15 min. Three phases formed: a clear aqueous phase at the top, a white solid intermediate phase, and an organic phase at the bottom. Portions of the three layers were analysed for their total arsenic content using acid digestion and ICPMS as described above. The aqueous upper layer was used directly or after oxidation with hydrogen peroxide (10% (v/v) in final solution) for measurements by HPLC/ICPMS.

HPLC/ICPMS Measurements. The water-soluble arsenic compounds in the milk were separated by anion- and cation-exchange HPLC - for details see legends to Figs 2 & 4. In brief, anion-exchange HPLC was performed with a Thermo Scientific Dionex IonPac™ AS14A column and ammonium bicarbonate buffer under gradient elution conditions, and cationexchange HPLC was performed with a cation-exchange/mixed mode TCI Dual ODS7 ACS Paragon Plus Environment

Page 9 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

CX10 column with a pyridinium formate buffer under isocratic conditions. The HPLC system was coupled either to the single quadrupole ICPMS or the triple quadrupole ICP-QQQ-MS, serving as the arsenic-selective detector. The single quadrupole ICPMS was operated in no-gas mode whereas the ICP-QQQ-MS was operated in MS/MS mode with O2 as reaction cell gas at a flow rate of 30% (~ 0.3 mL min-1). Both instruments were operated in time-resolved analysis mode using sample and skimmer cones made of nickel and tuned for highest sensitivity and robust plasma conditions. 1% CO2 in argon was introduced as optional gas to the plasma to enhance the arsenic signal. One channel was measured per isotope: monitored masses were m/z 75 (0.3 s/point, for 75As), m/z 53 (0.05 s/point, 40Ar13C to monitor carbon) and m/z 77 and 82 (each 0.1 s, to ascertain possible argon chloride interferences on 75As) with the single quadrupole ICPMS. Mass transitions m/z 75  m/7 91 (0.3 s/point), m/z 53  m/z 53 (0.02 sec/point) and m/z 77  m/z 77 (0.3 sec/point) were monitored with the ICP-QQQ-MS. Quantification was performed by external calibration against standard arsenic species (AB, DMA, MA and As(V) based on peak areas in the calibration range of 0.02 to 5 µg As L-1 depending on the species. Standards were prepared in aqueous 0.1% (v/v) trifluoroacetic acid for matrix matching. In the absence of a suitable reference material certified for arsenic species, for quality control we used: our in-house human milk reference material spiked with 1 µg As kg-1 each of AB, DMA and As(V); and the reconstituted milk sample

spiked

with

0.15

µg

As

kg-1

As(V)

(see

above).

8 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 21

RESULTS & DISCUSSION We set about to develop a method for determining arsenic species in human milk, but with a clear focus on iAs because it is a known toxic substance currently of great public interest in regard to children exposed to arsenic in food.11 The quantification of arsenic species in human milk is a formidable task - human milk is characterized by low concentrations of arsenic,13 and a complex matrix comprising high levels of lipids, carbohydrates and proteins.18 Preliminary attempts to do speciation analysis on human milk using the techniques applied to arsenic species in human urine proved unsuccessful. Human urine from healthy individuals is a comparatively simple matrix and contains arsenic levels typically 20-50 fold higher than those in human milk.12 Thus, our proposed study to investigate the arsenic species in mothers’ milk called for the development of a new analytical method able to handle protein- and lipid-rich samples, and with detection limits better than the methods in current use for other biological and environmental samples. We tackled the problem in two stages: first we developed a sample preparation procedure that effectively removed the proteins and lipids from the milk, and then we developed a HPLC system that enabled us to detect iAs, our initial target As species, and other biologically relevant water-soluble As species, at low concentrations. Fractionation of human milk – Removal of proteins and lipids. In our first attempt to handle the complex human milk matrix, we tried to precipitate proteins with acetonitrile, which has been used effectively for protein precipitation in serum samples for As and Se species.20,19 However, this approach proved unsuccessful for this complex matrix because considerable amounts of proteins and lipids were still present in the human milk samples after the acetonitrile treatment, which made it difficult to access the aqueous fraction after centrifugation. The acidification with formic acid has been used to precipitate proteins in human milk samples prior to the analysis of As species in the water fraction by HPLC/ICPMS.12 When applied to the lipid- and protein-rich human milk matrix, however, this method showed practical difficulties similar to those observed with the acetonitrile treatment. 9 ACS Paragon Plus Environment

Page 11 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

To overcome the difficulties observed with the currently used methods, we employed a “one-pot” methodology incorporating two consecutive steps of protein precipitation with acid followed by liquid-liquid extraction with dibromomethane to remove lipids. We tested both 0.1% trifluoroacetic acid (TFA) and 2% formic acid for protein precipitation, with TFA proving to be the more effective at precipitating the milk proteins. Treatment with TFA yielded a clear aqueous water fraction whereas a whitish-grey phase was present after treatment with formic acid. Higher TFA concentrations (up to 1% v/v) showed no improvement in protein precipitation, but influenced negatively the HPLC performance resulting in retention time shifts and misshaped peaks. These observations were also reported by Raber et al7, when subjecting water extracts containing higher TFA concentrations to anion-exchange chromatography. To separate the lipids from our milk samples and to provide our aqueous fraction as an easily accessible top layer, we tested solvent-solvent extraction with the halogenated solvents dichloromethane and dibromomethane. After centrifugation, the aqueous fraction is found on the top of the organic and insoluble intermediate fractions making it easily and directly accessible for HPLC/ICPMS measurements. Although dichloromethane treatment produced a clean aqueous fraction, we observed 40Ar35Cl interference on mass m/z 75 resulting from traces of DCM in the water fraction when it was analysed by HPLC coupled to a single quadruple ICPMS (see below). This chloride interference was totally removed in HPLC/ICP-QQQ-MS measurements performed in MS/MS mode with O2 as reaction cell gas. To overcome the chloride interference, we also investigated the use of dibromomethane as extraction solvent. It was equally effective as DCM in removing lipids from human milk, maintaining the practical advantage of having the aqueous layer at the top, and had the additional advantage of avoiding chloride interference when a single quadrupole ICPMS is used. We then applied our fractionation method to our in-house (Graz) human milk sample and to the same sample spiked with iAs, DMA, and AB (each at 1 µg As kg-1). The human milk sample contained 0.17 ± 0.03 µg As kg-1 (n=6), of which 99 ± 9% (n=3) was found in the water layer after fractionation. The spiked milk sample 10 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 21

contained 3.29 ± 0.10 µg As kg-1 (n=6), of which 96 ± 2% (n=3) was found in the water fraction, whereas 2 ± 1% (n=3) was found in both the organic layer and the solid intermediate phase. The reconstituted milk with a total As content of 0.41 ± 0.02 µg As kg-1 (n=3) prepared from the certified reference material ERM®-BD150 Skimmed Bovine Milk Powder was spiked with 0.15 µg As kg-1 As(V). Application of our developed method returned a recovery of 103 ± 2% (n=3) of the spiked amount of As in the water fraction. When the fractionation method was applied to the two Norwegian milk samples (total As content in these samples was 1.55 µg As kg-1 and 4.46 µg As kg-1), a lower proportion of the arsenic (ca 60%) was found in the water fraction. HPLC/ICPMS For the determination of iAs and other water-soluble As species in the human milk samples we used HPLC/ICPMS under anion- and cation-exchange/mixed mode conditions. The proposed HPLC method was specifically developed to be used in combination with a single quadrupole ICPMS, but particularly in regard to chloride interference and the detection of MA, we additionally explored the capability of a triple quadrupole ICPMS for determining arsenic species in human milk. Separation conditions commonly used for arsenic species in urine or food samples (e.g. polymer-based strong anion-exchange column with phosphate or malonic acid buffers) were not successful when applied to the milk samples because the matrix was too complex even after removal of the proteins and lipids. We overcame these problems by using an anion-exchange column with a carbonate/bicarbonate buffer system as mobile phase. This chromatographic system was recently reported for the determination of As species in matrix-rich extracts of turkey meat.21 First, we investigated isocratic elution for the determination of our target analyte iAs in our spiked human milk test sample (Supporting information Figure S1). The signal at RT 2.2 min corresponds to 40Ar35Cl interference by the naturally occurring chloride in milk, which could complicate the quantitation of MA at the trace levels. The recovery of a spiked amount of As(V) in human milk was 103 ± 2% (mean ± SD; n=3). Additionally, we determined iAs in the reconstituted milk sample, because no iAs was detected in the human milk test sample. We determined a concentration for 11 ACS Paragon Plus Environment

Page 13 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

[iAs] of 0.21 ± 0.01 µg As kg-1 (n=3) in the prepared milk. Spiking the milk with As(V) returned a recovery of 101 ± 3% (n=3). The triple quadrupole ICPMS operated in MS/MS mode with O2 as reaction cell gas whereby the mass transition of m/z 75 (As+) to m/z 91 (AsO+) was monitored, removed the undesired ArCl+ interference effectively (Supporting information Figure S2). The recoveries of spiked amounts of As(V) in our human milk test sample and reconstituted milk sample were 101 ± 2% (n=3) and 102 ± 1% (n=3) respectively and comparable to that found for the single quadrupole MS but having the additional advantage of being able to measure MA under isocratic conditions free of chloride interference. The HPLC method based on isocratic elution, developed for iAs measurements, showed low detection limits and good recoveries when applied to the test milk samples, but our investigations of two human milk samples from Norway showed no detectable levels of the iAs (< 20 ng As kg-1). Rather, we observed an arsenic signal near the void volume, and two poorly resolved peaks around the retention time of DMA (Supporting information Figure S3). We then employed gradient elution aimed at improving the separation of the early eluting arsenic compounds and lowering the detection limit for iAs. Additionally, with gradient elution MA is well-resolved from the peak arising from chloride, which enabled the detection of low concentrations of MA in the presence of the naturally occurring chloride (Figure 1).

12 ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 21

Figure 1. Gradient elution anion-exchange HPLC/ICPMS chromatograms of (a) a mixed standard solution containing DMA, MA and As(V) (each at 1 µg As kg-1) and (b) our human milk test sample spiked with 1 µg As kg-1 of AB, DMA and As(V). Thermo Scientific Dionex IonPac™ AS14A (150 x 3 mm; 5 µm) column; mobile phase: gradient elution (20 - 150 mM NH4HCO3 , pH 9.20 (pH adjusted with aqueous NH3), 0-2 min: 20 mM, 2-4 min: 20-150 mM, 4 – 6 min: 150 mM; 6.1 – 9 min: 20 mM; flow rate, 0.7 mL min-1; column temperature, 40°C and injection volume, 20 µL.

Importantly, the LOD for iAs (5 ng As kg-1) is 3-fold lower than that obtained with isocratic elution conditions. Under the gradient elution conditions, we see an unknown As species distinct from the DMA signal in a Norwegian human milk sample (Figure 2). There is also an arsenic signal in the void volume, suggested the presence of neutral or cationic organic arsenic compounds in human milk. The arsenic profile of the human milk sample did not change after oxidation with hydrogen peroxide, indicating that it was not arsenite.22 Spiking of a human milk sample with DMA (Figure 2) supported the presence of an unknown As compound. Neither MA (