Comparison of a Chemical and Enzymatic Extraction of Arsenic from

Jun 9, 2005 - An Automated Hydride Generation Interface to ICPMS for Measuring Total Arsenic in Environmental Samples. Mrinal K. Sengupta and Purnendu...
0 downloads 6 Views 123KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 5241-5246

Comparison of a Chemical and Enzymatic Extraction of Arsenic from Rice and an Assessment of the Arsenic Absorption from Contaminated Water by Cooked Rice A M A N D A H . A C K E R M A N , †,‡ P A T R I C I A A . C R E E D , † A M Y N . P A R K S , †,§ M I C H A E L W . F R I C K E , †,‡ CAROL A. SCHWEGEL,† J O H N T . C R E E D , * ,† DOUGLAS T. HEITKEMPER,| AND NOHORA P. VELA| Microbiological and Chemical Exposure Assessment Research Division, NERL, ORD, United States Environmental Protection Agency, Cincinnati, Ohio 45268, and Forensic Chemistry Center, United States Food and Drug Administration, Cincinnati, Ohio 45249

Rice is a target food for arsenic speciation based analyses because of its relatively high arsenic concentration and per capita consumption rates. Improved speciation data for rice can be helpful in estimating inorganic arsenic exposures in the U.S. and in endemic populations. The inorganic arsenic exposure for cooked rice should include both the arsenic in raw rice plus the arsenic absorbed from the water used to prepare it. The amount of arsenic absorbed from water by rice during preparation was assessed using five different types of rice cooked in both contaminated drinking water and arsenic-free reagent water. The rice samples were extracted using trifluoroacetic acid (TFA) and speciated using IC-ICP-MS. The TFA procedure was able to extract 84-104% of the arsenic (As) from the five different cooked rice samples. Chromatographic recoveries ranged from 99% to 116%. The dimethylarsinic acid (DMA) and inorganic arsenic concentration ranged from 22 to 270 ng of As/g of rice and from 31 to 108 ng of As/g of rice, respectively, for samples cooked in reagent water. The overall recoveries, which relate the sum of the chromatographic species back to the total digested concentration, ranged from 89% to 117%. The absorption of arsenic by rice from the total volume of water [1:1 to 4:1 (water:rice)] used in cooking was between 89% and 105% for two different contaminated drinking water samples. A comparison of the TFA extraction to an enzymatic extraction was made using the five rice samples and NIST 1568a rice flour. The two extraction procedures produced good agreement for inorganic arsenic, DMA, and the overall recovery. Through the use of IC-ESI-MS/ MS with a parent ion of m/z 153 and fragment ions of m/z 138, 123, and 105, the structure dimethylthioarsinic acid * Corresponding author phone: (513)569-7617; (513)569-7757; e-mail: [email protected]. † U.S. EPA. ‡ Oak Ridge Postdoctoral Research Fellow. § Oak Ridge Research Fellow. | U.S. FDA. 10.1021/es048150n CCC: $30.25 Published on Web 06/09/2005

 2005 American Chemical Society

fax:

was tentatively identified in two of the rice samples using the enzymatic extraction.

Introduction Drinking water and food are the two major sources of arsenic exposure for people not exposed occupationally. The “toxicity” of an exposure is dependent on the chemical form(s) of arsenic. This has caused an increase in speciation based analyses (1) especially in dietary samples containing a mixture of arsenicals. Rice is a target food for these speciation studies on the basis of its consumption rates and inorganic arsenic concentrations (1-3). Traditionally, total arsenic determinations have been used to assess the exposure to arsenic from rice (4-8). The total arsenic concentration in rice (on an uncooked basis) and the chemical form of the arsenic vary considerably depending on the rice (3). Therefore, a total arsenic determination may be a poor indicator of the inorganic arsenic concentration in rice. In cooked rice, the total arsenic comes from the uncooked rice and any arsenic absorbed from water during food preparation (4). The percentage of water absorbed by the rice is dependent on the type of rice and the way the rice is prepared. With typical preparation procedures used in the U.S., rice absorbs approximately 100% of its mass in water during cooking. Bae et al. used a large excess of water and concluded that the rice contained more total arsenic than would be expected from the mass of water absorbed through cooking (4). Dietary matrixes contain a more diverse set of arsenicals than drinking water, and the arsenicals are present within a solid matrix. The solid matrix dictates the need to solubilize the individual species prior to quantitation. A number of extraction techniques for rice have been reported in the literature (1, 3). A chemical extraction using trifluoroacetic acid (TFA) has been shown to quantitatively remove arsenic from uncooked rice prior to speciation (3). The ideal extraction procedure would liberate only those arsenicals which are solubilized by the human digestive process and, in doing so, reflect the “bioaccessible” fraction of arsenic within a sample. The chemical aggressiveness of different types of extractions varies considerably, and the correlation between the solubilized fraction and the bioaccessible fraction is difficult to assess. One approach to assessing the correlation between the amount of As solubilized using TFA extraction and the bioaccessible fraction of arsenic in rice is to compare an enzymatic extraction to the chemical-based TFA extraction on a total arsenic basis, as well as a species-specific basis. A good correlation would provide some credence for using the much more cost effective TFA-based extraction for routine analysis. An enzymatic extraction system which employs conditions similar to those found in the human digestive tract was developed by Glahn et al. (9, 10). This technique was created for estimating bioavailability of iron in various foods. The procedure involves a two-step extraction, intended to mimic the physical and biochemical processes in the human digestive system (stomach and small intestine). This approach, used extensively for a variety of food samples, including rice, provides a reasonable estimation of human digestion (9, 10). The goal of this study is to improve the exposure and risk assessments associated with arsenic in rice, and this is accomplished in two ways. First, species-specific information is generated for five types of rice, and the absorption of arsenic from waters used in the preparation of rice is estimated. This VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5241

improves the exposure assessment by increasing pathway information and provides arsenic absorption rates that can be used to estimate exposure to arsenic from cooked rice. Second, the extraction technique used is compared, on a species-specific basis, to an enzymatic extraction which is more physiologically based. This should improve the risk estimate by providing species-specific information and a more realistic “estimate” of the bioaccessibility of the arsenic from rice.

Experimental Section Rice Cooking. The samples used in this study represent a variety of commercially available rice samples in the United States. These samples included short-grain white rice, longgrain white rice, short-grain brown rice, instant rice, and instant infant rice cereal. A 10 g portion of each sample was prepared according to package instructions. For this study, it is important to understand that the water:rice ratio used in food preparation varies from 1:1 to 4:1 (v:v) depending on the sample. Each type of rice was prepared with 18 MΩ distilled, deionized (DDI) water (0 µg/kg arsenic added) (Millipore, Bedford, MA) and contaminated water samples, containing 21.9 µg/kg As(V). In addition to these preparations, long-grain white rice was also prepared using a contaminated water containing 35.3 µg/kg As(V). After food preparation was complete, each sample was dried at 105 °C to a constant weight. These dried rice samples were ground to a fine powder in a coffee mill and stored in a desiccator prior to extraction. The concentrations reported are based on the oven-dried weight of each rice sample. Extraction of Rice. TFA Extraction. The TFA extraction was carried out using a previously developed method (3). For extraction, 0.5 g of rice powder was placed in a 50 mL centrifuge tube. A 2 mL portion of 2 M TFA (Fisher, Fair Lawn, NJ) was added to the sample. This mixture was vortexed to ensure thorough mixing and then placed in a 100 °C oven for 6 h. After 6 h, each extract was diluted to a mass of 25 g using DDI water. Prior to analysis, the extract was centrifuged and filtered through a 0.2 µm filter (Fisher). Enzymatic Extraction. To mimic cooked rice entering the digestive tract, the oven-dried rice samples were rehydrated in DDI by treating a 0.5 g sample of rice with 1.6 g of DDI in a sealed centrifuge tube and heating to 80 °C for 60 min. Extraction was then carried out using a method similar to that developed by Glahn et al. (9, 10). The hydrated rice paste was suspended in 10 mL of a saline solution, containing 120 mM NaCl (Sigma, St. Louis, MO) and 5 mM KCl (Sigma). The pH of this solution was adjusted to 2 using 5 M HCl (Fisher). A pepsin solution was prepared by dissolving 0.2 g of pepsin (Sigma) in 8 mL of 0.1 M HCl, and 0.5 mL of the resulting solution was added to each rice sample. The mixture of rice, saline, and pepsin was incubated for 60 min at 37 °C. The pH was then raised from 2 to 6 using a solution of 1 M NaHCO3 (Sigma). A pancreatin solution was prepared by dissolving 0.05 g of pancreatin (Sigma, St. Louis, MO) in 27 mL of 0.1 M NaHCO3, and 2.5 mL of this pancreatin solution was added to each extract. The pH was then adjusted to 7 using 1 M NaOH (Fisher). The rice solution was incubated for 120 min at 37 °C and then brought to a final mass of 15 g with saline solution. The extracts were filtered prior to analysis, as described for the TFA extracts. Standards. Inorganic arsenic standards, As(III) and As(V), were purchased from Spex Certiprep (Metuchen, NJ). Dimethylarsinic acid (DMA) and disodium methyl arsenate (MMA) were obtained from Chem Service (98% purity, West Chester, PA). The arsenic concentrations in all standards were verified against NIST 1640 (Gaithersburg, MD) on a total metal basis. The standard reference material 1568a rice flour was purchased from NIST (Gaithersburg, MD). Certified total 5242

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

arsenic in standard reference material (SRM) 1568a is 0.29 ( 0.03 mg of As/kg. Synthesis of Dimethylthioarsinic Acid. Hydrogen sulfide gas was bubbled through a 3 mL stock solution containing 10.2 mg/kg DMA for 30 min. The gas was generated by the action of hydrochloric acid (6 mL of concentrated HCl added to 4 mL of DDI) and iron(II) sulfide (5 g), and this reaction was carried out in a well-ventilated fume hood (11). These solutions were made fresh daily to avoid stability issues. Digestion of Rice. Samples were microwave digested prior to total arsenic determination. To accomplish this, 0.5 g of rice powder and 5 mL of HNO3 were added to the digestion vessels. The vessels were covered, and the samples were predigested overnight. A closed-vessel microwave digestion system was used to digest the samples (3). A four-stage digestion program was used, with a maximum temperature of 165 °C and a maximum pressure of 170 psi. Total digestion time was 20 min. This aggressive procedure resulted in complete dissolution of the rice matrix and provided access to the total arsenic concentration in rice. Determination of Total Arsenic. Total Arsenic in Digest. After digestion was complete, the digests were transferred to 30 mL high-density polyethylene (HDPE) bottles and diluted to 25 g with DDI. Following this dilution, a 2 g aliquot of each digest was diluted to 5 g with internal standard containing Ge and Y, giving a final concentration of 10 µg/kg Ge and Y. Total arsenic was determined for each rice digest using standard addition (single point, 5 µg/kg As(V) spike) to correct for the influence of the matrix. Two digestions of each rice sample were performed. Total Arsenic in TFA Extract. Total arsenic was also determined in the TFA extracts. Five extractions were performed on each cooked rice sample, and total arsenic analysis was performed on each filtered extract. A 5 g portion of each TFA extract was diluted to 10 g with 18 MΩ water, and the internal standard was added as described above. Total arsenic of each extract was determined using standard addition with a 5 µg/kg As(V) spike. Total arsenic was not determined on the enzymatic extracts because the 125 mM chloride matrix produced a large 40Ar35Cl isobaric interference on mass 75 (75As, monoisotopic). This large interference in combination with the low arsenic concentrations in the rice extracts degraded the precision and accuracy of this determination. Instrumental Details. The IC system used to speciate the arsenic was an Agilent 1100 system (Palo Alto, CA), and it was interfaced to a Hewlett-Packard 4500 ICP-MS instrument (Palo Alto, CA). Masses 75 and 77 were monitored using a 1 s dwell time. Total arsenic analysis for the TFA extracts and the digests was performed using a PQ3 ICP-MS instrument (Thermo Jarrell Ash, Franklin, MA). The ESI-MS/MS instrument was a ThermoFinnigan (San Jose, CA) LCQ Deca ion trap mass spectrometer equipped with an atmospheric pressure ionization source. The LCQ was operated in negative-ion electrospray mode with an ESI spray voltage of -3 kV, sheath gas of 100 psi, auxiliary gas of 40 (unitless), and heated capillary temperature of 275 °C. The mass range scanned during LC/MS analysis was 100300 Da. LC/MS/MS of the standards provided further structural confirmation by collisionally activated dissociation (CAD) of the [M - H]- at m/z 153 with a normalized collision energy of 35% and scanning the product ions from 50 to 300 Da. A centrifuge (Eppendorf 5810R, Brinkman Instrument, Westburg, NY) was used to reduce the difficulty of filtering the extracts. The incubator used for the enzyme extraction was a MAXI 14 (Hybrid Limited, Middlesex, U.K.). The oven used in the TFA extraction was a Thelco model 17 (Winchester, VA).

TABLE 1. Absorption of Arsenic Present in Contaminated Water by Rice during Food Preparation Using a Mass Balance Approach

sample instant white (DDI) instant white (water 1) short-grain brown (DDI) Short-grain brown (water 1) Short-grain white (DDI) short-grain white (water 1) instant infant rice cereal (DDI) instant infant rice cereal (water 1) long-grain white (DDI)j long-grain white (water 1) long-grain white (DDI)k long-grain white (water 2)k across matrix

total extraction digesta efficiencyb,c (µg/kg) (%) 305 345 119 178 99 162 241 371 236 310 245 342 av RSD

92 ( 6.5 89 ( 7.0 97 ( 3.4 104 ( 3.5 101 ( 4.8 88 ( 6.9 84 ( 3.1 93 ( 8.3 98 ( 1.7 88 ( 5.6 90 ( 2.7 96 ( 2.1

overall speciation evaluationb

Aschromb,d

Astotal DMA (µg/kg)

As(III) + As(V)e (µg/kg)

∑chromf (µg/kg)

270 ( 12.8 270 ( 7.7 22 ( 4.4 23 ( 1.8 32 ( 7.4 35 ( 3.4 121 ( 13.0 125 ( 5.9 146 ( 2.9 154 ( 8.6 154 ( 4.8 160 ( 4.5

31 ( 3.2 57 ( 4.7 108 ( 4.8 166 ( 7.3 84 ( 3.1 128 ( 8.5 93 ( 12.6 222 ( 13.7 83 ( 6.0 143 ( 7.0 83 ( 3.9 181 ( 2.9

301 ( 13.4 327 ( 9.4 130 ( 7.3 188 ( 5.7 116 ( 12.1 163 ( 10.2 214 ( 12.6 347 ( 19.5 229 ( 8.0 296 ( 11.6 237 ( 7.7 340 ( 6.1

93% 6.8%

chromatographic recoveryg (%) 108 ( 8.6 107 ( 9.4 113 ( 8.1 102 ( 5.0 116 ( 9.0 114 ( 11.3 106 ( 10.5 100 ( 5.3 99 ( 8.3 108 ( 3.6 108 ( 5.3 104 ( 2.7 107% 5.4%

overall recoveryh (%) 99 ( 4.4 95 ( 2.7 109 ( 6.1 106 ( 3.2 117 ( 12.2 100 ( 6.3 89 ( 10.4 94 ( 8.3 97 ( 3.7 96 ( 3.7 97 ( 3.2 100 ( 1.8

As absorption by rice from water percent As absorbed by ricei 89 98 98 105 98 98

100% 8.3%

a Concentration based on the average of two replicates for oven-dried, cooked rice. b Value based on the average of five replicates, reported as x¯ ( 2σ, for oven-dried, cooked rice. c Extraction efficiency equals [Asextraction total/Astotal digest] × 100. d MMA was not detected in any of the rice samples. e Inorganic As was calculated on the basis of the sum of the areas of the peaks for As(III) and As(V) and is reported as the sum of As(III) and As(V). The arsenic values for the TFA extracts were blank subtracted using the average value calculated for the cookware blanks. f ∑chrom is reported as the sum of DMA and inorganic As. g Chromatographic recovery equals [∑chrom/Asextraction total] × 100. h Overall recovery equals [∑chrom/Astotal digest] × 100. i Absorption of As by rice equals [(Ascontaminated rice - Asuncontamiated rice)/Aswater] × 100. j Values other than total digest based on four replicates reported as x¯ ( 2σ. k Not included in determination of across matrix averages.

Chromatographic Separations. The PRP-X100 anion exchange column (Hamilton, Reno, NV) was used in both the TFA chromatography (see separation A in Table SI-1 in the Supporting Information) and enzyme chromatography (see separation B in Table SI-1 in the Supporting Information). Separation A has been described previously (3) and was used with TFA extracts. The high chloride concentration in the enzyme extractions made it necessary to resolve chloride from MMA to avoid an isobaric interference produced by 40Ar35Cl in the plasma. Separation B provided a separation of chloride and MMA but resulted in an increase in analysis time from 15 min for the TFA extracts to 60 min for the enzymatic extracts. ICP-MS was used for all quantitations. A postcolumn injection to estimate instrument drift was used and has been described elsewhere (12). A third separation (C in Table SI-1 in the Supporting Information) was used as a secondary chromatographic separation to verify the retention time match of the standard and the unknown on two separate columns. The detection of the unknown using IC-ESI-MS was accomplished using separation B with a postcolumn addition of 0.1 mL/min methanol to improve stability. The introduction of methanol coincided with a 0.1 mL/min decrease in the 25 mM (NH4)2CO3.

Results and Discussion Blank Control for Inorganic Arsenic in TFA Extracts. The TFA extraction allows a mass balance to be calculated between the total arsenic determined after microwave/HNO3 digestion and the speciated arsenic after a TFA extraction. A good mass balance implies that arsenic in rice is being quantitatively extracted. The use of this mass balance approach, in combination with the relatively low arsenic concentration in rice, required good species-specific blank control. The means and standard deviations for all cookware and labware blanks (including 0 ng/kg for those blanks in which arsenic was below the detection limit) were 26 ng/kg ( 32 and 21 ng/kg ( 29 (xj ( 2σ), respectively. (See the Supporting Information for more details on estimating labware and cookware blank considerations.) Blank controls can be placed in perspective by realizing that the highest determined blank (92 ng/kg) represents about 6% of the inorganic arsenic in a typical rice sample. This percentage

can get as high as 15% for rice samples that contain very low inorganic arsenic concentrations. TFA Extraction of Cooked Rice. One of the objectives of this study was to evaluate the percentage of arsenic from drinking water that is absorbed by rice during cooking. The native concentration of arsenic in rice was determined by preparing the rice in DDI water, and the uptake of arsenic by rice during cooking was determined by preparing the same rice in contaminated drinking water. Table 1 provides a summary of the arsenic absorption/speciation data for five different types of rice samples. The total arsenic concentration for each of the microwave-digested samples is shown in the second column of Table 1. These concentrations were determined on a sample which was completely digested and, for this reason, provides an estimate of all the available arsenic within the rice sample. The native As concentrations ranged from 99 to 305 µg/kg. The absorption of arsenic from the cooking water by rice is evident by comparing the total digest concentrations for samples prepared in DDI and those prepared in contaminated drinking water (water 1). A separate portion of the cooked rice was then extracted using the TFA procedure, and total As concentrations were determined in the extracts. The extraction efficiency relative to the total digest concentrations is reported in column 3 of Table 1. The extraction efficiencies ranged from 84% to 104% with typical 2σ control limits of 1.7-8.3%. The repeatability of the TFA extraction efficiency can be assessed by inspection of the relative standard deviation (RSD; based on 1σ) from the replicates of each of the five matrixes. For replicate extractions on the same matrix, the RSD ranged from 1% to 4.5% with an average RSD of 2.8% (data not shown) including rice cooked both in DDI and in water 1. These data indicate that replicate variation of extraction efficiency associated with this method is instrument limited because the RSDs obtained are comparable to the stability specifications of the ICP-MS instrument. An across matrix average (93%) and RSD (6.8%) were calculated for the average extraction efficiency to estimate the matrix-dependent performance of the method. This indicates that the precision of the extraction efficiency is influenced predominately by the matrix. The arsenic in the TFA extract was also speciated using IC-ICP-MS, and these results are shown in columns 4 and 5 of Table 1. DMA and inorganic arsenic were detected in VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5243

TABLE 2. Enzymatic Extractions of Arsenic from Cooked Rice chromatographic dataa

sample

total digest (µg/kg)

As(III) + As(V)b,c (µg/kg)

DMA (µg/kg)

unknownd

∑chrome (µg/kg)

overall recoveryf (%)

instant white short-grain brown short-grain white instant infant rice cereal long-grain white

305 119 99 241 236

29 ( 2.1 90 ( 0.9 75 ( 1.7 81 ( 4.8 75 ( 6.0

226 ( 6.3 20 ( 2.9 31 ( 0.9 110 ( 2.6 105 ( 4.4

40 ( 5.0 0 0 0 46 ( 7.0

295 ( 11.2 109 ( 2.1 106 ( 2.4 191 ( 2.8 226 ( 13.8

97 ( 3.7 92 ( 1.7 108 ( 2.5 80 ( 1.1 95 ( 5.9

a Concentrations based on the average of three replicates for enzymatic extraction except for long-grain white, which is based on five replicates. Concentrations are reported as x¯ ( 2σ and are based on the oven-dried weight of cooked rice. b MDL calculated on the basis of the response of As(V). c Inorganic arsenic was reported as the sum of As(III) and As(V). The arsenic values for the enzyme extracts were blank subtracted using the average blank. d Quantitation based on the response of the MMA standard. e ∑chrom is reported as the sum of DMA, the unidentified species, and inorganic arsenic. f Overall recovery equals [∑chrom/Astotal digest] × 100.

each of the five rice samples. The short-grain brown rice and short-grain white rice samples contained proportionately higher levels of inorganic arsenic. Previous work on arsenic speciation in rice has shown wide variations in the type and amount of arsenic present in rice from different geographical areas (1, 3-8). It is clear that the ratio of inorganic arsenic to DMA is not a constant across rice matrixes (1, 3). To verify that all the arsenic species extracted by the TFA procedure elute from the chromatographic column, the sum of the chromatographic concentrations (column 6) is compared to the total arsenic concentrations determined in the extracts (data not shown). This is shown as chromatographic recovery in column 7. A 100% chromatographic recovery would indicate that all of the arsenic injected onto the column eluted off the column. In general, the chromatographic recoveries indicate that all the arsenic solubilized by the TFA was chromatographable. The matrix-dependent performance of the method was estimated by calculating an across matrix average (107%) and RSD (5.4%) for the average chromatographic recovery. This indicates that the chromatographic component of the analysis performs well independent of the matrix. The final mass balance check is the overall recovery, found in column 8 of Table 1. The overall recovery compares the sum of the chromatographic concentrations (column 6) with the total digest concentrations (column 2). This is the percentage of the total arsenic in the sample that was accounted for in the speciation analysis. The overall recoveries, in Table 1, ranged from 89% to 117% with an across matrix average of 100%. For replicate speciation determinations on the same matrix, the overall recovery RSDs (based on 1σ) ranged from 1.4% to 6.1% with an average of 3.0% (data not shown). These values were calculated using the rice cooked in DDI and in water 1. The matrix-dependent variation of the average overall recovery was estimated by calculating an across matrix RSD (8.3%) for rice prepared in DDI and water 1. These data provide the end user with method performance data which indicate the precision of the overall speciation approach in the context of a good mass balance between speciated and total arsenic. A similar analysis was performed on standard reference material 1568a rice flour with a certified total arsenic concentration of 0.29 ( 0.03 mg/kg. Using the certified As concentration in place of the total digest value, the extraction efficiency was 95% ( 7.0 (xj ( 2σ), the chromatographic recovery was 96% ( 5.4 (xj ( 2σ), and the overall recovery was 92% ( 5.2 (xj ( 2σ, n ) 3). The DMA, inorganic arsenic, and MMA concentrations were 168 µg of As/kg of rice ( 8.7, 87 µg of As/kg of rice ( 8.7, and 12 µg of As/kg of rice ( 0.2 (xj ( 2σ, n ) 3), respectively. The data in Table 1 in combination with the SRM results indicate that very little arsenic remains unextracted or unchromatographed in rice 5244

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

following the TFA extraction/speciation procedure. The use of procedures having a good mass balance improves the risk assessment for arsenic by providing the risk assessor with species-specific information on nearly all the arsenic present in a dietary ingestion of rice. The final column in Table 1 contains the percentage of the arsenic present in the contaminated drinking water that was absorbed during food preparation. The percentage of arsenic in the water used for cooking that was recovered from rice ranged from 89% to 105%. This indicates that the arsenic in the water was coabsorbed with the water taken up by the rice during the cooking process. The absorption appeared to be independent of the water:rice ratio (volume ratios of 1:1 to 4:1) used in the preparation of the rice and the type of rice studied. All data summarized in Table 1 thus far have involved food preparation with either DDI or a drinking water contaminated with 21.9 µg/kg arsenic. Rows 10-13 in Table 1 compare the data for long-grain white rice cooked in contaminated drinking water samples containing 21.9 (row 11) and 35.3 (row 13) µg/kg inorganic arsenic. The values obtained for extraction efficiency, chromatographic recovery, and overall recovery were not significantly different from each other or those of the rest of the samples analyzed. The rice absorbed 98% of the arsenic for both of the contaminated waters. The reproducibility of these values across water type provides preliminary data that indicate that the method performance was not adversely affected by the water in which the rice was cooked. Dietary arsenic exposure assessments often utilize total arsenic data and a default assumption to estimate inorganic arsenic exposures (5, 7, 8). The data in Table 1 concur with recent findings (3) that the inorganic arsenic concentration in rice is not a fixed percentage of the total arsenic. The speciated data in Table 1 can be used to estimate a mean of inorganic arsenic concentrations. This mean can be used to replace a default assumption and thereby improve the accuracy of the exposure assessment. Furthermore, the arsenic absorption from water in combination with ingestion rates (2) can be used to improve the estimate of inorganic arsenic exposure from ingestion of cooked rice. Enzymatic Extraction of Rice. Table 1 indicates that TFA quantitatively extracted the arsenic from the rice matrixes and provides species-specific information on nearly all arsenic in the rice. An additional issue to be addressed is whether the TFA extraction followed by speciation is similar to a physiologically based extraction which attempts to mimic the human digestive system. Prior to utilizing the enzymatic procedure on the five rice samples and the SRM, the relative magnitude of the enzymatic blank was evaluated (see the Supporting Information for more details on estimating labware and cookware blank considerations).

FIGURE 1. IC-ICP-MS and IC-ESI-MS/MS identification of DMTA: (A) ICP-MS chromatogram for the unknown in rice and rice fortified with synthetic DMTA standard; the chromatography used is separation B described in Table SI-1 (in the Supporting Information) with a postcolumn addition of 0.1 mL/min methanol; (B) IC-ESI-MS mass chromatogram and the CAD spectra of m/z 153 for the synthetic DMTA standard. Table 2 contains the arsenic speciation results for the five cooked (DDI water) samples using the enzymatic digestion. The total arsenic determinations on the enzyme extracts were not determined because of the 125 mM chloride (isobaric interference 40Ar35Cl; see the Experimental Section). The sample name and total digest concentrations (column 2) are carried over from Table 1. A comparison of the chromatographic data in columns 3-5 indicates that the inorganic arsenic and DMA concentrations compare well across the two extraction procedures. The inorganic arsenic concentrations for the enzymatic extraction were 83-94% relative to those of the TFA extraction, while the DMA concentrations for the enzymatic extraction were 72-91% relative to those of the TFA extraction. The poorest agreement for DMA was observed for instant white rice (84%) and long-grain white rice (72%). This poor agreement in these two rice samples was offset by the presence of an unknown species in the enzymatic extracts (see column 5). When the concentrations of this unknown species (separation B, retention time 37 min) in the enzymatic extracts were added to the DMA, the sum was within 4% of the DMA concentration reported for the TFA extracts. Finally, a comparison of the overall recovery for the two extraction techniques indicates a good mass balance between the chromatographable species and the total arsenic concentration determined after a HNO3 digestion for the five rice samples. The SRM rice flour was also analyzed using the enzymatic digestion. The DMA, inorganic arsenic, and MMA concentrations were 148 µg of As/kg of rice ( 7.3 (xj ( 2σ), 101 µg

of As/kg of rice ( 6.9 (xj ( 2σ), and 11 µg of As/kg of rice ( 0.5 (xj ( 2σ, n ) 3), respectively. The overall percent recovery was 89% ( 4.9 (xj ( 2σ, n ) 3). Two experiments were conducted regarding the unknown arsenical in two of the enzyme extracts reported in Table 2. First, the unknown arsenical was converted to DMA by treating the extract with H2O2. This conversion was studied via IC-ICP-MS. Second, a chromatographic fraction collection of the unknown peak (from the PRP-X100) was treated with 2 M TFA, and it chromatographed as DMA. Recent reports have indicated the occurrence of sulfur analogues for arsenosugars in natural products (13, 14) and for DMA in urine (11). On the basis of our experiments, a sulfur analogue of DMA was thought to be the unknown. Dimethylthioarsinic acid (DMTA) was synthesized (11) with slight modification (see the Experimental Section) and its retention time compared to the unknown in rice. Figure 1A shows an ICP-MS chromatogram for the unknown in rice and an ICP-MS chromatogram of that rice sample fortified with the synthetic DMTA standard. The spike recovery was 98% on the basis of the drift standard area counts. The retention time match indicates that DMTA may be the unknown from the enzymatic extraction of the rice. Figure 1B contains the IC-ESI-MS mass chromatogram of m/z 153 (negative ion mode) and the CAD spectra of m/z 153 for the standard. The CAD spectra indicate product ions at m/z 138, 123, and 105. The structural assignments are made in the top right-hand side of Figure 1B. VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5245

Two additional types of confirmatory tests were conducted. First, the ratio of m/z 153 to m/z 155 (ratio 4.2) is consistent with the molecule containing a sulfur (sulfur ratio 4.4). Second, ICP-MS and IC-ESI-MS/MS data were collected using a second chromatographic separation (separation C in Table SI-1, ION-120, 20 mM (NH4)2CO3, pH 9.0). The standard and the unknown coeluted at 10.2 min (using ICP-MS), and the ESI-MS/MS spectrum for the standard was identical to Figure 1B (data not shown). The elemental (Figure 1A), molecular (Figure 1B), and supporting information provide the basis for a tentative identification of the unknown in rice as DMTA. Finally, a rice sample was spiked with DMA and subjected to the enzymatic extraction process in an attempt to determine if the unknown was produced by the extraction. The DMA spike did not produce an additional unknown, which would be consistent with the natural occurrence of the unknown in the rice rather than the production of the unknown by the enzymatic extraction process.

(3) Heitkemper, D. T.; Vela, N. P.; Stewart, K. R.; Westphal, C. S. Determination of total and speciated arsenic in rice by ion chromatography and inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 2001, 16, 299-306.

Acknowledgments

(9) Glahn, R. P.; Lee, O. A.; Yeung, A.; Goldman, M. I.; Miller, D. D. Caco-2 Cell Ferritin Formation Predicts Nonradiolabeled Food Iron Availability in an In Vitro Digestion/Caco-2 Cell Culture Model. J. Nutr. 1998, 128, 1555-1561.

This work was sponsored by the U.S. EPA and U.S. FDA under Interagency Agreement DW-75-93936101. This research was supported in part by appointment to the Postgraduate Research Participation Program administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. DOE and the U.S. EPA. The U.S. Environmental Protection Agency through its Office of Research and Development and the U.S. Food and Drug Administration funded and managed the research described in this paper. It has been reviewed in accordance with EPA peer and administrative review policies and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. No official support or endorsement by FDA of this paper is intended or should be inferred.

Supporting Information Available Additional text and table. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Schoof, R. A.; Yost, L. J.; Eickhoff, J.; Crecelius, E. A.; Cragin, D. W.; Meacher, M. M.; Menzel, D. B. A Market Basket Survey of Inorganic Arsenic in Food. Food Chem. Toxicol. 1999, 37, 839846. (2) International Rice Research Institute. World Rice Statistics, 2002. http://www.riceweb.org/Ricefact2000.pdf (accessed January 2004).

5246

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

(4) Bae, M.; Watanabe, C.; Sekiyama, M.; Sudo, N.; Bokul, M. H.; Ohtsuka, R. Arsenic in cooked rice in Bangladesh. Lancet 2002, 360, 1839-1840. (5) Duxbury, J. M.; Mayer, A. B.; Lauren, J. G.; Hassan, N. Food Chain Aspects of Arsenic Contamination in Bangladesh: Effects on Quality and Productivity of Rice. J. Environ. Sci. Health, A 2003, A38, 61-69. (6) Abedin, Md. J.; Cresser, M. S.; Meharg, A. A.; Feldmann, J.; CotterHowells, J. Arsenic Accumulation and Metabolism in Rice (Oryza sativa L.). Environ. Sci. Technol. 2002, 36, 962-968. (7) Roychowdhury, T.; Uchino, T.; Tokunaga, H.; Ando, M. Survey of arsenic in food composites from an arsenic-affected area of West Bengal, India. Food Chem. Toxicol. 2002, 40, 1611-1621. (8) Meharg, A. A.; Rahman, Md. M. Arsenic Contamination of Bangladesh Paddy Field Soils: Implications for Rice Contribution to Arsenic Consumption. Environ. Sci. Technol. 2003, 37, 229-234.

(10) Glahn, R. P.; Cheng, Z.; Welch, R. M.; Gregorio, G. B. Comparison of Iron Bioavailability from 15 Rice Genotypes: Studies Using an in Vitro Digestion/Caco-2 Cell Culture Model. J. Agric. Food Chem. 2002, 50, 3586-3591. (11) Hansen, H. R.; Raab, A.; Jaspars, M.; Milne, B. F.; Feldmann, J. Sulfur-Containing Arsenical Mistaken for Dimethylarsinous Acid [DMA(III)] and Identified as a Natural Metabolite in Urine: Major Implications for Studies on Arsenic Metabolism and Toxicity. Chem. Res. Toxicol. 2004, 17, 1086-1091. (12) Creed, J. T.; Magnuson, M. L.; Pfaff, J. D.; Brockhoff, C. Determination of bromate in drinking waters by ion chromatography with inductively coupled plasma mass spectrometric detection. J. Chromatogr. A 1996, 753, 261-267. (13) Fricke, M. W.; Creed, P. A.; Parks, A. N.; Shoemaker, J. A.; Schwegel, C. A.; Creed, J. T. Extraction and detection of a new arsine containing arsenosugar in molluscs by IC-ICP-MS and IC-ESI-MS/MS. J. Anal. At. Spectrom. 2004, 19, 1454-1459. (14) Schmeisser, E.; Raml, R.; Francesconi, K. A.; Kuehnelt, D.; Lindberg, A.; So¨ro¨s, C.; Goessler, W. Thio arsenosugars identified as natural constituents of mussels by liquid chromatographymass spectrometry. Chem. Commun. 2004, 16, 1824-1825.

Received for review November 24, 2004. Revised manuscript received April 14, 2005. Accepted May 2, 2005. ES048150N