Effects of Wet-Blending on Detection of Melamine in Spray-Dried

Office of Regulatory Science, Center for Food Safety and Applied Nutrition, United States Food and Drug Administration, 5001 Campus Drive, College Par...
1 downloads 0 Views 1MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Effects of Wet-Blending on Detection of Melamine in Spray-Dried Lactose Betsy Jean Yakes, Marti M. Bergana, Peter F. Scholl, Magdi M. Mossoba, Sanjeewa R Karunathilaka, Luke K Ackerman, Jason D. Holton, Boyan Gao, and Jeffrey C Moore J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.7b00834 • Publication Date (Web): 24 May 2017 Downloaded from http://pubs.acs.org on June 13, 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.

Journal of Agricultural and Food 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 30

Journal of Agricultural and Food Chemistry

1

Effects of Wet-Blending on Detection of Melamine in Spray-Dried

2

Lactose

3 4

Betsy Jean Yakes‡, Marti M. Bergana§, †, Peter F. Scholl‡, Magdi M. Mossoba‡, Sanjeewa R.

5

Karunathilaka‡, Luke K. Ackerman‡, Jason D. Holton§, Boyan Gao+ and Jeffrey C. Moore Î, *

6 7



8

Regulatory Science, 5001 Campus Drive, College Park, MD 20740 USA

9

§

U.S. Food and Drug Administration, Center for Food Safety and Applied Nutrition, Office of

Abbott Nutrition, Division of Abbott Laboratories, Research and Development, 3300 Stelzer

10

Road, Columbus, OH 43219 USA

11

+

12

USA

13

Î

U.S. Pharmacopeial Convention, 12601 Twinbrook Parkway, Rockville, MD 20852 USA

14



Current Address: Consultant, U.S. Pharmacopeial Convention

Department of Nutrition and Food Science, University of Maryland, College Park, MD 20742

15 16

Corresponding Author

17

*Phone: +1 301 816 8288. Fax: +1 301 816 8157. E-mail: [email protected]

18

1 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

19

ABSTRACT

20

During the development of rapid screening methods to detect economic adulteration, spray-dried

21

milk powders prepared by dissolving melamine in liquid milk exhibited an unexpected loss of

22

characteristic melamine features in the NIR and Raman spectra. To further characterize this

23

“wet-blending” phenomenon, spray-dried melamine and lactose samples were produced as a

24

simplified model and investigated by near-infrared spectroscopy, Raman spectroscopy, 1H NMR,

25

and DART-FTMS. In contrast to dry-blended samples, characteristic melamine bands in NIR

26

and Raman spectra disappeared or shifted in wet-blended lactose-melamine samples. Subtle

27

shifts in melamine 1H NMR spectra between wet- and dry-blended samples indicated differences

28

in melamine H-bonding status. Qualitative DART-FTMS analysis of powders detected a greater

29

relative abundance of lactose-melamine condensation product ions in the wet-blended samples,

30

which supported a hypothesis that wet-blending facilitates early Maillard reactions in spray-dried

31

samples. Collectively, these data indicated the formation of weak, H-bonded complexes and

32

labile, early Maillard reaction products between lactose and melamine contributes to spectral

33

differences observed between wet- and dry-blended milk powder samples. These results have

34

implications for future evaluations of adulterated powders and emphasize the important role of

35

sample preparation methods on adulterant detection.

36 37

Keywords. Adulteration, DART-FTMS, H-bonding, lactose, Maillard reaction, melamine, milk

38

powder, NMR, NIR, Raman spectroscopy, wet-blending

39

2 ACS Paragon Plus Environment

Page 2 of 30

Page 3 of 30

Journal of Agricultural and Food Chemistry

40

INTRODUCTION

41

The U.S. Pharmacopeia (USP) has been leading a collaborative research project to develop

42

reference standards and analytical methods for the targeted and non-targeted detection of

43

economically-motivated adulteration in foods. As a first model, skim and nonfat dry milk

44

powders (SMP and NFDM, respectively; collectively called MPs) were recently evaluated using

45

both targeted and non-targeted techniques.1 Non-targeted methodologies are of interest due to

46

their potential utility for high-throughput, in-situ screening and ability to flag a food matrix as

47

adulterated with new or unknown compounds that might be missed by targeted methods. Non-

48

targeted procedures that offer these advantages include near infrared (NIR), mid-infrared (MIR),

49

and Raman spectroscopies, in conjunction with chemometrics.2 However, one challenge of

50

applying these vibrational spectroscopies is that highly variable and complex food matrices can

51

degrade method performance.

52 53

To evaluate the effect of matrix complexity on the efficiency and accuracy of detection while

54

also investigating the potential to detect MP adulteration at different points in the supply chain,

55

two classes of melamine-spiked MP samples were prepared and studied in our previous work.1

56

The first class had melamine mixed with MP through physical mixing (i.e., dry-blending (DB-

57

MP)), while the second was prepared by spray-drying melamine dissolved in liquid milk (i.e.,

58

wet-blending (WB-MP)). Interestingly, the WB-MP samples did not exhibit the NIR first

59

overtone, primary amine stretching vibration (6812 cm-1) expected for melamine. Instead, this

60

band was replaced by new bands that were broader and frequency-shifted, and these observations

61

were attributed to wet-blending (WB). As such, targeting the prominent 6812 cm-1 band, as

62

commonly performed,3-5 could no longer be considered appropriate for the detection of 3 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

63

melamine from all adulteration routes. The implications from the discovery of this WB effect are

64

clear: other nitrogen contaminants may also present similar spectral challenges and would

65

thereby be less likely to be detected by targeted spectroscopic methods.

66 67

Based on the potential effects of elevated temperature and pressure that are used in spray-drying,

68

a plausible explanation for these spectral shifts was the development of a different crystalline or

69

amorphous form, and this phenomenon has been reported in the literature for other compounds

70

(e.g., lactose,6 simvastatin7). Results of XRD, PLM, Raman, and NIR analyses in previous MP

71

studies were consistent with such a change only for lactose (i.e., loss in crystallinity features

72

upon spray-drying).1 A second explanation considered potential chemical reaction(s) between

73

melamine and MP constituents. Quantitative LC-MS/MS and qualitative 1H NMR data did not

74

indicate melamine degraded in WB samples. These data collectively provided evidence that the

75

WB effect was not predominantly driven by an irreversible chemical modification of melamine.

76 77

In this paper, additional interactions, specifically those that involve the primary amine in

78

melamine and a compound naturally found in MP (lactose), are explored. One interaction that

79

could have arisen from this system was hydrogen-bonding (i.e., H-bonding), where melamine

80

has been shown in literature to H-bond to other molecules and create new structures. These

81

structures can range from small to supramolecular, often depending on the mixing method,

82

pressure, and temperature, with the chemistry well-described in a number of articles.8-10 Indeed,

83

shifts in Raman and IR bands have previously been attributed to melamine H-bonding, notably in

84

work by Li et al. that showed both band broadening and shifts for the OH group of ascorbic acid

85

and NH groups of melamine in FTIR spectra.11 Mircescu et al. have also performed detailed

4 ACS Paragon Plus Environment

Page 4 of 30

Page 5 of 30

Journal of Agricultural and Food Chemistry

86

calculations and experiments using Raman spectroscopy that demonstrated how melamine

87

spectral bands can be shifted by amine protonation.12

88 89

Maillard reactions between melamine and lactose were also considered herein. Controlling such

90

reactions between lactose and amines is important in the food and pharmaceutical industries

91

because these determine the organoleptic properties and nutritional quality of foods, as well as

92

the stability of drug commodities. As such, these reactions have been well-studied, including

93

work on milk protein lactosylation as a function of pasteurization and MP storage conditions.13, 14

94

Although lactose serves well as an inactive ingredient in many pharmaceuticals, studies have

95

documented unwanted interactions of lactose excipients with drugs that contain amines.15, 16

96

Maillard reaction products have been characterized in protein-free model systems containing

97

melamine and lactose, using reverse-phase LC-MS/MS.17 Related studies have also evaluated the

98

kinetics of reactions between melamine and a variety of aldehydes.18

99 100

Previous efforts to understand the cause(s) of the WB effect in MP were limited due to the

101

complexity of the MP matrix.1 As lactose is one of the most prevalent components in MP, from a

102

chemical concentration and chemical functional group standpoint,19 a simple melamine-lactose

103

model system was developed for this study. This simple model system was used to prepare the

104

samples listed in Table 1, and, in order to maintain consistency with our previous MP study,1

105

this system was processed using similar spray-drying techniques and at equivalent melamine

106

concentrations. NIR, Raman, NMR and DART-FTMS instrumentation were then employed to

107

qualitatively analyze this simpler system and characterize spectral changes upon wet-blending.

5 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

108

This lactose-melamine model gave insight into potential interactions (i.e., H-bonding and/or

109

lactose-melamine adducts) that may cause the band disappearance and shifting phenomena.

110 111

MATERIALS AND METHODS

112

Samples. Sample information including ID, melamine and lactose concentrations, and

113

preparation type are indicated in Table 1. A lactose and melamine solution (0.23 g/mL α-Lactose

114

monohydrate, 2.82 mg/mL melamine (both reagent grade from Sigma-Aldrich, St. Louis, MO))

115

was prepared by dissolving lactose in water at 52 °C, adding melamine, and holding for 10 min

116

before spray-drying. This sample was identified as wet-blended “WB-Lac”. A lactose-only

117

control was similarly prepared; this mixture was identified as “SDLactose”. To explore whether

118

changes in melamine crystallinity could be contributing to spectral differences, a 4.35 % (w/v)

119

aqueous solution of melamine was gently warmed, to aid dissolution, and spray-dried; this

120

sample was identified as “SDMelamine”. All spray-dried samples were processed using a Buchi

121

B-290 bench-top spray-dryer (New Castle, DE) using an inlet temperature of 220 °C, outlet

122

temperature of 123 °C, feed rate of 6 mL/min, and airflow at 357 L/hr. For the dry-blended

123

melamine sample, 0.123 g of unprocessed melamine was blended into 10 g of dry SDLactose;

124

this sample was identified as dry-blended “DB-Lac”. All samples were stored in a desiccator

125

until use.

126 127

NIR Measurements. Spectra were acquired using an MPA Bruker Optics (Billerica, MA) FT-

128

NIR spectrometer equipped with an integrating sphere, diffuse reflection accessory including a

129

sample rotating cup, and PbS detector. Spectra were collected at room temperature using 16 cm-1

130

resolution in the 10,000 to 4000 cm-1 range at 32 scans. Replicate spectra (N=3) were measured 6 ACS Paragon Plus Environment

Page 6 of 30

Page 7 of 30

Journal of Agricultural and Food Chemistry

131

for each test portion in a randomized order. Instrument performance was internally verified

132

according to vendor-specific tests with a USP NIR Suitability Reference Standard (Rockville,

133

MD). Included in the OPUS software is the Optics Validation Program (OVP), which executes a

134

series of performance tests using the standards in an automated filter wheel. This program was

135

used to verify that the spectrometer was operating within specifications prior to sample

136

acquisition.

137 138

Data were preprocessed and analyzed using PLS_Toolbox software within a Matlab

139

computational environment (PLS_Toolbox_7.82, Eigenvector Research Inc., Wenatchee, WA).

140

To eliminate scaling effects, standard normal variate (SNV) normalization was applied to the

141

derivative spectra, while fourth derivative transformation (Savitzky-Golay algorithm, window

142

size = 15 points, fourth order polynomial fit) was employed to eliminate baseline artifacts.

143 144

Raman Spectroscopy. Raman spectra were acquired on a DXRxi Raman Imaging System

145

(Thermo Electron North America, Madison, WI) controlled with OMNICxi Raman Imaging

146

Software. Automated alignment and calibration procedures were performed prior to sample

147

measurement. Instrument parameters were: 780 nm laser at 24 mW, full-range grating (5 cm-1

148

nominal resolution (FWHM), 50-3300 cm-1), 10× microscope objective, and 50-µm pinhole

149

aperture. Approximately 0.05 g of each powdered sample was placed onto individual, plain glass

150

microscope slides (Fisher Scientific, Pittsburgh, PA) for analysis. For melamine samples, spectra

151

were obtained with a 0.2 s exposure and 60 scans. For lactose and mixture samples, spectra were

152

obtained with a 3 s exposure and 10 scans. For chemical imaging, a 75 × 81 µm2 area was

153

defined, and spectra were obtained via an image pixel size of 3 µm and 2 s exposure with 2 scans

7 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

154

per spectrum. Raman spectral intensities were recorded as counts per second (cps). Data analysis

155

was performed with Atlµs in OMNIC for Dispersive Raman (v. 9.2) and GraphPad Prism (v.

156

5.02, La Jolla, CA).

157 158

1

159

tetramethylsilane ((TMS), Cambridge Isotope Laboratories, Inc., Andover, MA) and vortex

160

mixed for ~1 min. Immediately following a 10 min centrifugation at 1417g (Fisher Scientific

161

Micro-Centrifuge 59A), the supernatants were transferred to NMR tubes, and the spectra were

162

acquired. To maintain similar times between the introduction of solvent and data collection,

163

samples were prepared and analyzed sequentially. For reagent melamine, an arbitrary amount of

164

melamine was added to DMSO-d6 to allow for evaluation. Proton (1H) NMR spectra were

165

collected with a 500 MHz VNMRS spectrometer equipped with a 3 mm PFG-ID probe (Varian,

166

Palo Alto, CA), utilizing a pulse width of 90° and a 10 s pre-pulse delay. A sweep width of

167

6510.4 Hz and an acquisition time of 2.517 s were employed. Each spectrum (32 scan average)

168

was collected at 25 ± 0.1 °C and internally referenced to TMS at 0.0 parts per million (ppm).

169

Spectra were processed with MestReNova software (v. 10.0.2, Mestrelab Research, Santiago de

170

Compostela, Spain).

H NMR Analysis. Samples (6.0 ± 0.1 mg) were dispersed in 600 µL of DMSO-d6 with

171 172

DART-MS. Powders were fastened to DipIt™ sample introduction tips (melting point

173

capillaries, LEAP Technologies, Carrboro, NC) using a thin film of silicone high vacuum grease

174

(HVG, Dow Corning Corp., Greensboro, NC). HVG was dissolved (17 mg/mL) in hexane

175

(pesticide residue grade, Fluka, Milwaukee, WI) and aged 48 h prior to use to allow SiO2 solids

176

to precipitate. Sample capillaries were then dipped into the supernatant, and the solvent was

8 ACS Paragon Plus Environment

Page 8 of 30

Page 9 of 30

Journal of Agricultural and Food Chemistry

177

evaporated (~15 min) to produce a thin film coating. Three HVG coated sampling capillaries

178

were dipped into vials containing each dry, powdered sample, and the capillaries were gently

179

tapped against each vial to remove loose particles. Each capillary was subjected to two

180

concentrated blasts (