Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Distribution of Animal Drugs among Curd, Whey, and Milk Protein Fractions in Spiked Skim Milk and Whey Nancy W. Shappell, Weilin L. Shelver, Sara J Lupton, Wendy Fanaselle, Jane M. Van Doren, and Heldur Hakk J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.6b04258 • Publication Date (Web): 04 Jan 2017 Downloaded from http://pubs.acs.org on January 8, 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 41
Journal of Agricultural and Food Chemistry
Distribution of Animal Drugs among Curd, Whey, and Milk Protein Fractions in Spiked Skim Milk and Whey
Nancy W. Shappell,†* Weilin L. Shelver,† Sara J. Lupton,† Wendy Fanaselle,‡ Jane M. Van Doren,‡ Heldur Hakk† †USDA-ARS, Biosciences Research Laboratory, 1605 Albrecht Blvd, Fargo, ND 58102-2765 ‡US-FDA, Center for Food Safety and Applied Nutrition, 5001 Campus Drive, College Park,
MD, 20740
AUTHOR INFORMATION *Corresponding Author: Email:
[email protected] Phone: 701-239-1233 Fax: 701-239-1430
1 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 2 of 41
1
ABSTRACT: It is important to understand the partitioning of drugs in processed milk and milk
2
products, when drugs are present in raw milk, in order to estimate the potential consumer
3
exposure. Radioisotopically labelled erythromycin, ivermectin, ketoprofen, oxytetracycline,
4
penicillin G, sulfadimethoxine, and thiabendazole were used to evaluate the distribution of
5
animal drugs among rennet curd, whey, and protein fractions from skim cow milk. Our previous
6
work reported the distribution of these same drugs between skim and fat fractions of milk. Drug
7
distribution between curd and whey was significantly correlated (R2 = 0.70) to the drug’s
8
lipophilicity (log P), with improved correlation using log D (R2 = 0.95). Distribution of drugs
9
was concentration-independent over the range tested (20 – 2,000 nM). With the exception of
10
thiabendazole and ivermectin, more drugs were associated with whey protein than casein on a
11
nmol/mg protein basis (oxytetracycline experiment not performed). These results provide
12
insights into the distribution of animal drug residues, if present in cow milk, among milk
13
fractions, with possible extrapolation to milk products.
14 15 16 17 18 19
KEYWORDS: drug residues, curd, whey, skim milk, antibiotic, anthelmintic, NSAID, partition, distribution, protein association
2 ACS Paragon Plus Environment
Page 3 of 41
Journal of Agricultural and Food Chemistry
20 21 22 23
INTRODUCTION
Over the last decade, ≥ 99.96% of the more than 3 million annual U.S. bulk milk tanker
24
samples were found to be free of violative drug residues.1 Historically, testing bulk milk
25
deliveries of tanker milk to processing plants in the U.S. has been primarily focused on drug
26
residues of beta-lactam antibiotics, which are commonly used in dairy cows when antibiotic
27
treatment is required. However, other kinds of animal drugs are also administered to dairy cows
28
when appropriate. Reports published in the National Milk Drug Residue Database by the U.S.
29
Food and Drug Administration (FDA) confirm the presence of residues from drugs other than
30
beta-lactam antibiotics in some samples from bulk tank or bulk milk pick-up tankers in the
31
United States.2 The U.S. FDA published a risk-assessment entitled “Multicriteria-Based
32
Ranking Model for Risk Management of Animal Drug Residues in Milk and Milk Products”,
33
which used a science-based analytical approach to collate and incorporate relevant available data
34
and information; and provided a decision-support tool to assist with reevaluating which animal
35
drug residues should be included in milk testing programs.3 In preparing the risk assessment,
36
data gaps, including the lack of drug residue distribution data in milk products, were identified. This
37
information, describing partitioning of animal drugs in processed milk and its by-products, is
38
necessary to ascertain the potential for human exposure. The complexity of assessing human
39
exposures to drug residues from the products of one “liquid of biological origin” – cow milk – is
40
evident when considering the multitude of products, including ice cream, yogurt, sour cream,
41
various cheeses, whey protein supplements, and more than 35 others, derived from milk or
42
whey.4 Understanding the factors driving the distribution and/or concentrating of animal drug
3 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
43
residues among milk fractions will allow for better assessment of risk for potential human
44
exposure.
Page 4 of 41
45
Literature describing distribution of drug residues during the processing of cow milk is
46
limited, increasing somewhat when including studies of other species such as dairy goat, ewe,
47
and bison.5-19 Most of the available studies have focused on antibiotics. Cerkvenik et al.8
48
reported drug concentrations in milk products, such as yogurt or whey cheese, were both
49
variable, and drug-dependent. Cayle et al.7 indicated that penicillin G (PENG) distribution was
50
predominately in the aqueous phase with little distribution into fat-rich products such as cheeses.
51
In two other penicillin related studies (other penicillins11, various antibiotics15), the physical and
52
chemical properties of the specific compound affected drug distribution. The distributions of a
53
number of parasiticides into curd and cheese, and the differential distributions in the milk
54
products were dependent on the properties of the particular drug.5-7,16-17 Within a given drug
55
class, the increase of residue concentrations in cheese relative to milk varied widely, from 1.5
56
(albendazole10) to 13-fold (triclabendazole14). In both cases however, only metabolites were
57
detected, not parent drug. In general, concern over residues in milk and milk products increases
58
for drugs with long elimination half-lives, such as rafoxanide17, which persisted in milk for 47
59
days, or ivermectin in buffalo or goat milk, with half-lives of approximately 2.5 days.5-6
60
In our first paper we described “Phase 1” partitioning of animal drugs between the fat and
61
skim fractions of pasteurized bovine whole milk, and created an empirical model that predicted
62
drug distribution based on lipophilicity (logP).20 The model fit was improved by including the
63
ionization state of the drug (logD). Seven drugs (Figure 1) that spanned four antibiotic classes
64
[erythromycin (ERY), penicillin G, oxytetracycline (OTET), and sulfadimethoxine (SDMX)]
65
two anthelmintics [ivermectin (IVR) and thiabendazole (THIA)]; and one analgesic [ketoprofen 4 ACS Paragon Plus Environment
Page 5 of 41
Journal of Agricultural and Food Chemistry
66
(KETO)] were studied. These results were the first reported measurements of this kind for many
67
of these drugs. 20 In the present report, we measure the distributions of the same seven drugs
68
between rennet skim milk curd (henceforth noted as “curd”) and whey fractions, casein proteins,
69
and whey proteins using drug fortification and fractionation studies. The goal of this study was to
70
better understand the factors that determine the distribution of animal drugs, if present in cow
71
milk, among milk fractions, with possible extrapolation to milk products.
72
MATERIALS AND METHODS
73
Chemical, Supplies, and Equipment. Details pertaining to drugs, both unlabeled and
74
radiolabeled, were reported previously. 20 Raw milk was obtained from the bulk milk tank at the
75
North Dakota State University (NDSU, Fargo, ND) Dairy Unit farm (stored for ≤ 48 h after
76
milking). Reference standards used to validate compositional analyses of various milk fractions
77
(skim and whey) were obtained from Eurofins DQCI, (Mounds View, MN). Liquid vegetable
78
rennet was obtained from New England Cheesemaking Supply Company (South Deerfield, MA).
79
Ecolite® liquid scintillation cocktail was purchased from MP Biomedicals, LLC, Solon, OH.
80
Carbo-sorb®, Permafluor®, and Monophase® were obtained from Perkin Elmer, Waltham, MA.
81
Silica gel thin layer chromatography (TLC) plates (Uniplate™, 5 x 20 cm, 250 µm) and
82
octadecyl (C-18) modified silica gel plates were purchased from Analtech, Inc. (Newark, DE).
83
Amicon® Ultra-15 centrifugal filter (UF) devices were purchased from Millipore (Billerica,
84
MA). Dremel tool and garlic press were standard retail products. A Precision® water bath
85
(Thermo Scientific, Milford, MA), MagniWhirl® water bath (BlueM Electric Co., Blue Island,
86
IL), and an Allegra X-14R centrifuge (Beckman-Coulter, Brea, CA) were used. A Tri-Carb 1900
87
TR liquid scintillation counter (LSC, PerkinElmer, Houston, TX) and Model 307 Sample
5 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 6 of 41
88
Oxidizer (Packard, Meridan, CT) were used for scintillation counting and sample oxidation,
89
respectively. A Bioscan AR-2000 Imaging Scanner (Washington, DC) was used to detect
90
radioactivity on TLC plates.
91
Safety. Radiolabeled chemicals were handled in compliance with Nuclear Regulatory
92
Commission (NRC) regulations for 14C and 3H.
93
Determination of chemical purity and confirmation of chemical integrity. Thin layer
94
chromatography analyses, and for some drugs, liquid chromatography tandem mass spectrometry
95
(LC-MS/MS) were performed before and after the experiments. Initial analyses were used to
96
evaluate dose purity, while post-incubation analyses were used to evaluate the extent of putative
97
decomposition of the drug during processing. All TLC analyses were performed under
98
conditions described in Hakk et al. 20
99
Milk processing and radiochemical analysis.
100
Drug Partitioning from Skim Milk into Curd and Whey Fractions (Phase 2). In previous work
101
(Phase 1, Hakk et al.20) experiments were performed incubating the same set of radiolabeled
102
drugs in whole milk followed by quantitation of drugs in the skim milk and fat fractions. Here, in
103
Phase 2, drugs were incubated in skim milk prepared from whole pasteurized milk by
104
centrifugation, as previously described. 20 Starting from fifteen tubes of whole milk (50 mL
105
each), the skimmed fractions were combined into three pooled, replicate skim milk samples
106
ensuring both randomness and homogeneity across the study. Partitioning of products was
107
performed as diagramed in Figure 2 (Scheme 1). Skim milk (47 mL/tube) from pools 1, 2, and 3
108
were fortified with 94 uL of 0, 10,000, 100,000, or 1,000,000 nM radiolabeled drug stock
109
solutions to final concentrations of 0, 20, 200, and 2000 nM (three replicates per concentration).
110
Due to the low specific activity of [14C] thiabendazole, 75 nM was utilized instead of 20 nM. 6 ACS Paragon Plus Environment
Page 7 of 41
Journal of Agricultural and Food Chemistry
111
One additional tube without spiked drug was prepared from each pool of skim milk to serve as
112
matrix blanks (40 mL) for Kjeldahl analyses. Drug doses were prepared in water, acetonitrile, or
113
methanol, depending on the drug solubility, with the final organic solvent concentration of the
114
milk fraction never exceeding 0.2% (v/v). In order to determine possible non-matrix related
115
analyte decomposition during equilibration and processing, duplicate water samples (20 mL
116
each) were fortified with 200 nM drug (40 µL of 100 µM stock solution). After fortification, all
117
tubes were vortexed for 5-10 sec, manually inverted and shaken to ensure complete mixing;
118
aliquots were then removed for LSC (100 µL x 3) and TLC analyses. Tubes were then shaken at
119
80 rpm for 30 min in a 38°C water bath (previously established as adequate for drug
120
equilibrium). 20 A second sampling of the 2000 nM dose of skim milk (300 µL) was taken post-
121
equilibration for TLC analysis.
122
Curding was initiated by adding 172 µL of diluted rennet (~ 5% of stock vegetable rennet
123
in nanopure water) to skim milk (~ 47 mL), vortexing, followed by a second incubation at 38° C
124
(stationary water bath, 1 hour). Rennet was also added to water as a control, following the same
125
procedure. Rennet curd was separated from whey by centrifugation (3,000 x g, 15 min, 20°C).
126
Whey and water were sampled (3 x 200 µL) and assayed for radioactivity by LSC, and drug
127
integrity assessed by TLC and/or LC-MS/MS. Curd was homogenized by passing twice through
128
a hand-held garlic press, weighed into combustion cups (5 x 0.1 g aliquots per sample) and
129
assayed for radioactivity by combustion and LSC.
130
Determination of drug associated with whey proteins (Phase 3). Centrifugal ultrafiltration
131
(10kDa molecular weight cut-off) was used to separate protein-associated drug from “non-whey
132
protein-associated” (henceforth noted as “free”) drug. Fifteen tubes of pasteurized whole milk
133
(50 mL each) and two tubes of nanopure water (20 mL each) were brought to 38°C in a 7 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 8 of 41
134
stationary water bath, and whey was obtained as described above. Three separate pools of whey
135
were created and distributed into tubes as described for skim milk preparations resulting in 15
136
tubes of 20 mL each ( triplicate replicates, one replicate per pool for each dose level, and six
137
blanks). Half the matrix blanks (0 nM) were used for determination of background radioactivity,
138
total solids, and percent lipid and half were analyzed for protein using Kjeldahl as described in
139
Hakk et al. 20 Whey was fortified using 40 µL of radiolabeled drug at 0, 10,000, 100,000, and
140
1,000,000 nM to yield final concentrations of 0, 20, 200, and 2000 nM. Again, two water tubes
141
were fortified to assess drug stability (200 nM). A 1 mL aliquot was removed from vortexed
142
samples for LSC (100 µL x 3) and TLC analyses. Drug equilibration conditions for whey were
143
the same as those used for skim milk (described above) and drug integrity was assessed post-
144
equilibration using a 300 µL aliquot of each 2000 nM whey replicate. Post-equilibration, 15 mL
145
of whey was transferred into an UF device, weighed, and centrifuged (4,000 x g, 17 min, 20°C)
146
producing retentate (~5 mL) and permeate (~10 mL) fractions. Final fraction weights were taken
147
prior to removal from UF device and density corrected to determine fraction volumes. The
148
retentate was removed from the UF device, with ~ 1mL of retentate used to dislodge residual
149
particulates on the filter, and added back to the retentate fraction prior to LSC. The filter was
150
then washed with nanopure water (1 mL x 2) and the two washes combined prior to LSC. The
151
washed retentate filters (two per UF device) were removed from the plastic housing using a
152
Dremel tool and razor blade, air-dried overnight, weighed, placed in combustion cups, and
153
oxidized in a sample oxidizer. The filter washes, permeate, and retentate were analyzed for
154
radioactivity by LSC (200 µL x 3). TLC or LC-MS/MS analyses were performed to qualitatively
155
assess drug integrity.
8 ACS Paragon Plus Environment
Page 9 of 41
Journal of Agricultural and Food Chemistry
156
Quantitation of drug-associated with casein and whey proteins. In Phase 3, drugs were
157
incubated in whey, in the absence of the major milk proteins, caseins, which comprise ~ 84% of
158
bovine milk proteins. 21 Therefore, additional experiments were performed to evaluate if drug
159
affinity for whey proteins changed in the presence of caseins (Figure 3, Scheme 2). Twenty nM
160
of drug (with exception of 75 nM for THIA) was added to 20 mL of skim milk and incubated for
161
30 min. Post-incubation 15 mL of skim milk was processed through an UF device as described
162
above for whey, enabling calculation of the percentage of drug associated with the total protein
163
(casein + whey proteins) content of the UF skim milk retentate. In a second set of incubations,
164
the drug was incubated with skim milk (20 nM in 47 mL), but was then processed with rennet to
165
produce curd and whey. This whey (15 mL) was then subjected to the UF process. Drug
166
remaining in the retentate in this case represented the proportion of drug that associated with
167
whey proteins, while in the presence of casein.
168
Differences in experimental conditions between Scheme 2, which is presented in Figure 3
169
and Phase 3 above (see Scheme 1, Figure 2) were as follows: In Scheme 2, skim milk was spiked
170
at only one drug concentration (20 nM). One matrix blank was included for each process instead
171
of 3. The skim milk matrix required extending the ultrafiltration time to 2h in an attempt to
172
obtain the v/v ratio of retentate/permeate previously obtained (~1:2) with Phase 3 whey filtration.
173
In addition, more vigorous resuspension of retentate was required after the UF of skim milk,
174
using a larger wash volume and three filter washes were performed. A fraction of the curd (5 x
175
0.1 g) was combusted and counted to assess dose recovery. Filters were also combusted and
176
counted as described above to determine non-specific binding.
177
Compositional analyses. Compositional analyses were performed on the 0 nM blanks (no drug
178
added, matrix blanks) of pasteurized skim milk, whey, curd, retentate, and permeate to yield 9 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 10 of 41
179
percent lipid, total solids, and protein concentrations. For ultrafiltration experiments, aliquots
180
from three matrix blank tubes were used for total solids and lipid composition, and aliquots from
181
another three blanks were used for protein determinations. Method validation was confirmed by
182
concomitant analysis of Eurofin standards. Procedures for total solids, fat, and protein
183
determination were previously described.20 Briefly, volumes for total solids determinations were
184
as follows: 1 mL for skim milk, whey, permeate and retentate, and 0.1 g for curd; for lipid
185
content: 5 mL for each of skim milk, whey and permeate, and 1 mL retentate (diluted with 4 mL
186
water), and 0.25 g curd (diluted with 5 mL water); for protein determination-total nitrogen: 5 mL
187
for skim milk and permeate, 1 mL for retentate, and ~0.5 g for curd; for non-protein nitrogen: 10
188
mL for skim milk, 2 mL for retentate; and for non-casein nitrogen: 10 mL for skim milk.
189
Calculations for protein-associated drug (Figure 3). In the case of whey protein-associated
190
drug, whey permeate after UF contained non-protein associated drug (‘free’ drug), while
191
retentate contained both the protein associated drug and free drug. Permeate radioactivity and
192
volume were used to calculate the dpms/mL so that the contribution from free drug could be
193
subtracted from the radioactivity of the retentate. The radioactivity on the filter (obtained post-
194
combustion) was treated as non-specific binding while the filter wash radioactivity was treated as
195
part of the radioactivity of the retentate. The percent of whey protein-associated drug was
196
calculated by dividing whey protein-associated radioactivity (‘bound’) by [dosed radioactivity
197
minus non-specific binding radioactivity] and multiplying the result by 100. Similarly, the
198
percent casein-associated drug was estimated by subtracting [percent drug associated with whey
199
protein] from [percent drug associated with total N protein equivalents]. The percentage of whey
200
protein-associated drug in the presence of casein was calculated based on the original spiked
201
skim milk radioactivity so it was comparable to the total protein-associated drug. 10 ACS Paragon Plus Environment
Page 11 of 41
Journal of Agricultural and Food Chemistry
202
Statistical analyses
203
Distribution data from all three incubation concentrations for curd/whey and retentate/permeate
204
ratios (Phase 2 and 3, respectively) were analyzed for differences by ANOVA and linear
205
regression. Retentate percent drug distribution was calculated relative to the total dpms (retentate
206
+ permeate) and compared to retentate percent volume. A preferential association with whey
207
protein was indicated by a higher percent of drug in the retentate than the percent of retentate
208
volume, as analyzed by a paired two sample mean t-test.
209
RESULTS AND DISCUSSION
210
Compositional Analyses. Curding of skim milk yielded an average of 41 ± 0.4 mL of whey and
211
5.1 ± 0.19 g of curd (COV ≤ 4%, data not shown). Skim milk and whey standards were used for
212
QA/QC of composition analyses (Table S1). Laboratory values ranged from 97 to 102% of the
213
reported standards for percent total solids, nitrogen, and true protein. However, laboratory values
214
for percent lipid were consistently higher for the standards for both skim milk (0.16% versus
215
0.09%), and whey samples (0.38% versus 0.28%, respectively). Balance inaccuracy was found to
216
be the source of the overestimation when lipid weights were extremely low (< 0.3% of the milk
217
fraction).
218
Total solids, total N protein equivalent, true protein, and casein protein in skim milk
219
prepared for incubation with drugs varied by ≤ 3%, and were ≥ to 96% of the values reported for
220
skim milk used by Hakk et al.20 and ≥ 93% of the mean reported for standards (Table S1). Due
221
to the low percentage of lipid in skim milk (≤ 0.3%) and consequent small lipid mass, the higher
222
variability in lipid content was expected. Skim milk, curd, and whey composition were as
223
follows: lipids (0.3, 2.0, and 0.2%); total solids (9, 31, and 7%); and total N (3, 24, and 1%, 11 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 12 of 41
224
respectively). True protein was calculated from total N protein equivalents in skim milk, whey
225
(whole or skim) and retentate, yielding 2.9, 0.7, and 2.3%, respectively. Casein protein in skim
226
milk was 2.3%.
227
Total N protein equivalents and true protein were determined on whey and its retentate
228
fractions. Total N protein equivalents (0.19 ± 0.01%) in permeate fractions were equivalent to
229
the non-protein N, confirming the integrity of the ultrafilters (Table S1). The increase in true
230
protein in the retentate was proportional to the concentrating factor of ~ 3 (15 mL original
231
volume to ~ 5 mL retentate volume). The starting whey was 0.91% total N and 0.68% true
232
protein, while the respective values for retentate were 2.59% and 2.32%.
233
Incubation period, drug stability, recovery, and dose response. Equilibrium conditions for all
234
drugs in whole milk were achieved within 30 min when incubated at 37° C.20 Therefore drugs
235
were allowed to equilibrate in skim milk for 30 min prior to rennet addition. Equilibration time
236
of drugs in whey was evaluated using SDMX and KETO in triplicate incubations. Drug
237
distribution was unchanged over 4 h (COV of mean % drug in retentate or permeate across 30
238
min, 1, 2 and 4 h was ≤ 4%) and drug equilibrium was established in whey to be ≤ 30 min.
239
Because equilibrium times of all drugs in whole milk were identical, the same was assumed for
240
whey.
241
Valid quantitation of drug distribution using radiolabeled drugs necessitates knowing that
242
the drug stability was maintained throughout the experimental period. The integrity of 6 drugs
243
was confirmed by radiochemical analysis of TLC plates at the following points: pre- and post-
244
incubation in skim and whey, curd, and whey retentate fractions. The confirmation of OTET
245
stability required LC-MS/MS analysis (data not shown) in addition to TLC. With the exception 12 ACS Paragon Plus Environment
Page 13 of 41
Journal of Agricultural and Food Chemistry
246
of KETO and PENG, drugs remained unchanged during milk partitioning. In 6 of 10 TLC
247
analyses of KETO in curd, a shoulder appeared on the KETO peak, representing ~ 10-12% of the
248
total activity (data not shown). In whey and retentate, radioactivity was evident in only one peak
249
with the same Rf as cold KETO. The presence of lipids within the curd fraction might have
250
interfered with the chromatographic behavior of KETO. The other drug exhibiting more than
251
one peak was PENG, which is easily degraded.11 For PENG, broad peaks, peaks with shoulders,
252
or split peaks were present in curd and whey. Reproducibility of data was evidenced by excellent
253
dose recoveries with consistently small S.D.s (curd and whey experiments 98 ± 3.0%; and UF
254
experiments 100 ± 2.2% across all doses and drugs, data not shown).
255
The distributions observed across all drugs were generally independent of dose (SI Table
256
2). While the ratios for both Phases 2 and 3 differed (P = 0.04) by dose for PENG, there was no
257
dose dependence (P > 0.05). For SDMX, distribution was dependent on dose as assessed by both
258
ANOVA (P = 0.002) and linear regression (P< 0.0002) in Phase 3, but the dose-dependent
259
decrease in distribution ratio was 0.5%, essentially biologically irrelevant.
260
Percent drug distribution from skim milk into whey and curd fractions (Phase 2). Prior to
261
consideration of curd distribution data, one must remember that these data are specific to the
262
methodology employed, and not interchangeable with processes that use lactic acid prior to renin
263
addition in the curding process, as well as other variables such as pH. In addition, while cheese
264
might be equated with curd protein, the fat content of cheese is highly variable, and therefore, the
265
estimated drug distribution into cheeses would have to take into account the fat content and the
266
lipophilicity of the specific drug.
13 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
267
Page 14 of 41
The range of drug distributing into the curd (including small amounts of whey) ranged
268
from 14% (OTET) to 75% (IVR) (Figure 4). Reproducibility of data was evident with COVs for
269
curd and whey distribution averaging 2% across doses (the highest COV of each 4% and 6%,
270
respectively). The percent of drug retained in curd appeared to be more strongly correlated with
271
log D than MW. For example, the drugs with the largest relative distribution into curd, IVR
272
(75%, log D 6.61, MW 875) and THIA (48%, log D 2.93, MW 201) had the highest log D values
273
among the drugs examined. For drugs with intermediate log D values, like SDMX (1.23) and
274
ERY (1.24), curd retained ~ 30% of the drug; while for drugs with a negative log D (OTET,
275
PENG, and KETO) ~ 13% of drug was entrained in curd.
276
Few reported data are directly comparable to our findings, as drugs were typically
277
administered to animals, and incorporation into milk and milk products (predominately cheese)
278
evaluated. One exception was a report by Cayle et al.7, evaluating whole milk fortified with
279
PENG, in which the same percentage of PENG (12%) was found in cheese (42% moisture) as
280
was observed with fortified skim milk in this study (Table S4, 12 ± 0.3%). One abstract reported
281
that chlortetracycline (structurally similar to OTET) was initially at the highest concentration in
282
the whey fraction post-mammary infusion of the drug (as detected by microbiological
283
assessment), yet in later post-infusion milkings, concentrations were higher in the casein
284
fraction. 22 Initial concentrations post-infusion may not have reflected equilibrium with milk
285
proteins in the gland cistern, and later distributions may have been the result of drug absorption
286
by the gland and de novo incorporation into milk. While we found the majority of OTET in the
287
whey fraction, the concentration was actually higher for OTET in the curd or casein fraction.
14 ACS Paragon Plus Environment
Page 15 of 41
288
Journal of Agricultural and Food Chemistry
When sheep were dosed subcutaneously with IVR, an average of 57% of the drug from
289
whole milk was retained in cheese (extrapolated moisture content of 48%) compared to 75%
290
measured in the present study after spiking skim milk.8 Cerkvenik et al8 reported a mean of 9%
291
for IVR in whey, versus this study’s 21 ± 0.7% in spiked skim milk (Table S4). A partial
292
explanation for the differences in % IVR found in whey might be differences in recovery (this
293
study a = 95 ± 2.1% across all doses, 65%).8 Another factor may be differences in the amount
294
of fat. Others reported 66% of IVR from whole sheep’s milk partitioned into cheese, with 38%
295
associated with whey.12
296
Thiabendazole (THIA) results from this study are similar to findings with albendazole, a
297
drug with a similar structure. Fletouris et al.10 orally dosed cows with albendazole, and reported
298
that 30% of the albendazole initially present in whole milk (as metabolites) was associated with
299
curd and 70% was associated with whey. In the current study using skim milk, 47% of THIA
300
associated with curd and 54% with whey (Table S4). Using Phase 120 and Phase 2 data, the %
301
THIA associated with curd and whey derived from whole milk would be ~ 37 and 40%,
302
respectively, closer to the findings of Fletouris et al10. Again, direct comparison is problematic
303
when evaluating drug distribution when starting with whole versus skim milk, especially for
304
highly lipophilic drugs such as IVR.
305
The curd produced in our experiments contained small amounts of whey. To adjust for
306
this, the moisture content of curd was assumed to have retained the same concentration of whey
307
protein as that of whey, and the value for drug concentration in whey was multiplied by the curd
308
liquid weight and subtracted from the drug concentration of curd. This value is referred to as
309
“0% moisture curd”. The difference in percentage of drug residue in “experimental” versus “0%
15 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 16 of 41
310
moisture” curd is presented in Table S3. Accounting for whey in the curd reduced the
311
percentage of curd-associated drug by 30 to 60% for all drugs but THIA and IVR, which
312
changed by less than 10%.
313
Drug concentration in curd relative to milk (Phase 2). Figure 4 shows that drug concentration
314
increases in the curd relative to the initial skim milk concentration in every case, with the
315
smallest increase found with PENG (110% of skim milk) and the largest at more than 500%
316
(IVR). Since the mass of curd produced from the starting skim milk is ~10% of the whey
317
produced, an increase in drug concentration could be expected, even without preferential affinity
318
for curd proteins. Consistent with our results, Adetunji et al.23found essentially no change in
319
PENG concentration of curd and cheese relative to whole milk concentrations at 3 of 5 milk
320
processors. Data from the same study using streptomycin and tetracycline found inconsistencies
321
across processors, with curd and cheese concentrations sometimes lower than the original raw
322
milk. This is in contrast with our findings for OTET, where a 40% increase in concentration was
323
found in the curd. It is unclear why the lack of consistency was present across processors, but it
324
may reflect differential methodologies, along with potential for more degradation of the
325
antibiotics during milk processing.
326
Literature reports of anthelmintic in curd and cheese were somewhat similar to those
327
reported here. IVR in cheese from sheep and/or buffalo whole milk had concentrations from
328
290% 12to 400% 13,6 those of the original milk. Another anthelmintic of similar structure,
329
eprinomectin (MW 914), was found in curd at 375% of the concentration of the whole sheep’s
330
milk.13 The concentration in curd relative to skim milk was intermediate for THIA (410% of
16 ACS Paragon Plus Environment
Page 17 of 41
Journal of Agricultural and Food Chemistry
331
skim, this study), in contrast to the 1300% in cheese produced from whole sheep’s milk for
332
another structurally similar drug, triclabendazole (based on metabolites).14
333
Empirical modelling of drug distribution between curd and whey fractions. The log
334
concentrations in curd to whey versus the log P or D value (pH 6.8) are presented in Figure 5,
335
Panels A and B; numerical values are reported in SI Table 3. While both log P and D were
336
correlated with the distribution ratio of drugs in curd and whey, the impact of ionizable groups
337
accounted for by the log D value resulted in a stronger linear correlation (R2 = 0.95 versus 0.70
338
for log P). Specifically, the fit improved for the four drugs with ionizable groups (PENG, KETO,
339
THIA and IVR; see SI for pKa’s) resulting in charged molecules at the pH range of milk
340
fractions examined (pH 6.6 - 6.8). Similar results were obtained for a model of drug distribution
341
between fat and skim milk.20 Figure 5 (Panels C and D) also show drug distribution ratios for
342
0% moisture curd and whey; while changes in slope and intercept for both log P and log D
343
graphs occurred, the R2 values remain essentially unchanged. Reporting drug distribution on a
344
0% moisture curd basis would allow for modeling of concentrations in curd with variable
345
moisture content.
346
When considering the distribution of incurred animal drug residues in dairy products, it is
347
essential to describe the distribution among all of the components in the original whole milk.
348
Figure 6 combines the results of the present study with those of Phase 120 to provide drug
349
partitioning relative to whole milk. For example, while IVR concentration in curd was ~33 times
350
higher than in whey (and ~7 times the original concentration from an incubation of skim milk), ~
351
87% of IVR would be removed with lipid during skimming, resulting in only ~13% of original
352
concentration for distribution between the skim milk rennet curd (~10%) and whey (~3%).
17 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 18 of 41
353
Drug-association with whey protein as determined by centrifugal ultrafiltration – retentate
354
versus permeate (Phase 3). Whey produced from whole milk was fortified with radiolabeled
355
drug, and ultrafiltered. The distribution of drug between retentate and permeate fractions are
356
shown in Figure 7. Assuming no drug-protein association, drug distribution in whey would
357
simply reflect the volume distribution of retentate and permeate (Figure 7 solid lines). The
358
distributions for all drugs were significantly different (P
435
80% of the milk proteins, and are preferentially coagulated in the curding process. By contrast,
436
in a pure casein solution in PBS of 2.66 mg/ mL, binding was reported as only 8%,11somewhat
437
smaller than our findings of 16% (Table 1). The difference may be due to experimental
438
conditions, where pure casein was dialyzed at 4° C for several days in the study by Grunwald
439
and Petz11, with the potential for degradation of PENG during dialysis. In addition, casein may
440
have become saturated since their concentration was 1/10 that of normal cow’s milk (~28 mg/ 21 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 22 of 41
441
mL). In contrast, in vivo administration of PENG to cattle resulted in 28% and 33% of PENG in
442
serum to be protein-associated (based on dialysis or ultrafiltration separation, and assay of
443
radioactivity or antibiotic activity, respectively). Interestingly, these results for bovine blood
444
proteins are similar to the present findings of 16% for casein, but lower than those reported by
445
Drug Bank25 a data base of over 8,000 drugs providing chemical, pharmacological and
446
pharmaceutical information (45 to 68%, protein unspecified). These data demonstrated that
447
PENG binding can differ widely depending on the sources of proteins.
448
No data has been published on OTET association with milk proteins, but Ziv and
449
Rasmussen17 did report the distribution of the related compound, tetracycline, in milk from goats
450
receiving the drug in vivo. Their finding that 52% of tetracycline was associated with milk
451
proteins in skim milk is much higher than reported here for curd (7%) or whey associated protein
452
(8%). The structure, MW, log P, and log D (-2.50 and -2.93 calculated at pH 6.8) values for
453
OTET and tetracycline are very similar, providing no rationale for the differences observed. But,
454
differences for protein association among OTET, tetracycline, and chlortetracycline were
455
observed by Ziv and Sulman24, when the drugs were administered in vivo to either cows or ewes.
456
The percent associated with serum was lowest for OTET (~20%), followed by tetracycline (~
457
35%), and then chlortetracycline (~48%). The values reported by the Drug Bank25 ranged from
458
20% to 67% for tetracycline.
459
While there are no literature reports for the distribution of ERY in milk, 13% of
460
spiramycin (a structurally similar antibiotic) was associated with skim milk proteins after in vivo
461
administration to goats.19 This finding is similar to the 12 and 15% found for curd protein and
462
whey protein, respectively, in the present study. Literature reports for binding to bovine serum
22 ACS Paragon Plus Environment
Page 23 of 41
Journal of Agricultural and Food Chemistry
463
protein were lower: 18% by dialysis, and 20% by UF23, although the Drug Bank25 reported very
464
high association values (75-95%) possibly due to different sources of protein or experimental
465
methods. Drug Bank25 values for the rest of the drugs were consistently higher than either whey or
466 467
casein protein, even the calculated casein association. Two antibiotics, sulfamethoxazole and
468
sulfapyridine, chemically similar to SDMX, were reported by the Drug Bank25 to bind 50% and
469
70% of unspecified protein, respectively, while the present study reported 37% of SDMX to be
470
casein-associated. The sum of curd and whey protein associated THIA was 38% in our study. To
471
the best of our knowledge, association of THIA with milk proteins has not been reported
472
previously.
473
In conclusion, this paper provides an empirical model for predicting animal drug
474
distribution between rennet skim milk curd and whey. Results from this study also characterize
475
drug residue protein-associations. However, caveats must be considered that include the in vivo
476
distribution of drugs may differ from in vitro laboratory equilibrations, specifics of industry
477
processing which may result in different distributions, and metabolites or degradation products,
478
when present, which may distribute differently. If a drug is present in cow milk, the data
479
reported here and in our first paper20 presents a quantitative model for use as a food safety tool in
480
estimating drug distribution and concentration among milk fractions for various types of animal
481
drugs.
482
ABBREVIATIONS
483
ERY, erythromycin; IVR, ivermectin; KETO, ketoprofen; OTET, oxytetracycline; PENG,
484
penicillin G; SDMX, sulfadimethoxine; and THIA, thiabendazole.
23 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 24 of 41
485
ACKNOWLEDGMENTS
486
The authors wish to acknowledge the helpful discussions provided by Dennis Gaalswyk, Yinqing
487
Ma, David Oryang, and Chi Yuen Yeung from the FDA. We wish to thank the technical
488
assistance provided by Dee Ellig, Patrick Harland, Lindsey Fransen, Amy McGarvey, Jason
489
Neumann, Colleen Pfaff, and Michael Woodworth from the ARS Biosciences Research
490
Laboratory. Appreciation also goes to Todd Molden and Thomas Brown for milk collection
491
from the NDSU dairy barn.
492
SUPPORTING INFORMATION
493
Compositional analysis of skim milk, milk fractions, and reference standards; statistical analysis of dose
494
responses; log D and P values used for regression analyses of drugs; and data summaries for individual
495
drugs for Phase 2 and 3 (Tables S5-S18) are included in the supporting information.
496
FUNDING
497
This study was collaboratively funded by an interagency agreement with the FDA and USDA
498
ARS (Interagency Agreement no. 224-14-2006).
499
NOTES
500
The use of trade, firm, or corporation names in this publication is for the information and
501
convenience of the reader. Such use does not constitute an official endorsement or approval by
502
the United States Department of Agriculture (USDA), the Agricultural Research Service, or the
503
Food and Drug Administration of any product or service to the exclusion of others that may be
504
suitable. USDA and FDA are each equal opportunity employers.
505
The authors declare no competing financial interest.
506 507
24 ACS Paragon Plus Environment
Page 25 of 41
Journal of Agricultural and Food Chemistry
508
REFERENCES
509
(1) National Milk Drug Residue Database, FY 2005 - FY2015, www.kandc-sbcc.com/nmdrd/.
510
(accessed November 21, 2016).
511
(2) Food and Drug Administration. Milk drug residue sampling survey.
512
http://www.fda.gov/downloads/AnimalVeterinary/GuidanceComplianceEnforcement/Complianc
513
eEnforcement/UCM435759.pdf, 2015 (accessed August 11, 2016).
514 515 516 517
(3) Food and Drug Administration. Multicriteria-based ranking model for risk management of animal drug residues in milk and milk products. http://www.fda.gov/Food/FoodScienceResearch/RiskSafetyAssessment/ucm443549.htm, 2015 (accessed September 4, 2015).AAAA
518
(4) Smith, K. Dried dairy ingredients. Wisconsin Center for Dairy Research. 2008.
519 520 521
(5) Anastasio, A.; Esposito, M.; Amorena, M.; Catellani, P.; Serpe, S.; Cortesi, C. Residue study of Ivermectin in plasma, milk, and mozzarella cheese following subcutaneous administration to buffalo (Bubalus bubalis). J. Agric. Food Chem. 2002, 50, 5241-5245.
522 523 524
(6) Anastasio, A.; Veneziano, V.; Capurro, E.; Rinaldi, L.; Cortesi, M.; Rubino, R.; Danaher, M.; Cringoli. G. Fate of Eprinomectin in goat milk and cheeses with different ripening times following pout-on administration. J. Food Prot. 2005, 68, 1097-1101.
525 526
(7) Cayle, T.; Guth, J. H.; Hynes, J. T.; Kolen, E. P.; Stern, M. L. Penicillin distribution during cheese manufacture and membrane treatment of whey. J. Food Prot. 1986, 49,796-798.
527 528 529
(8) Cerkvenik, V.; Perko, B.; Rogelj, I.; Doganoc, D. Z.; Skubic, V.; Beek, W. M. J.; Keukens, H. J. Fate of ivermectin residues in ewes’ milk and derived products. J. Dairy Res. 2004, 71, 3945.
530 531 532
(9) De Liguoro, M.; Longo, F.; Brambilla, G.; Cinquina, A.; Bocca, A.; Lucisano, A. Distribution of the anthelmintic drug albendazole and its major metabolites in ovine milk and milk products after a single oral dose. J. Dairy Res. 1996, 63, 533-542.
533 534 535
(10) Fletouris, D. J; Botsoglou, N. A.; Psomas, I. E.; Mantis, A. I. Albendazole-related drug residues in milk and their fate during cheesemaking, ripening, and storage. J. Food Prot. 1998, 61, 1484-1488.
536 537
(11) Grunwald, L.; Petz, M. Food processing effects on residues: penicillins in milk and yoghurt. Anal. Chim. Acta. 2003, 483, 73-79.
25 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 26 of 41
538 539 540
(12) Imperiale, F.A.; Busetti, M. R.; Suarez, V. H.; Lanusse, C. E. Milk excretion of ivermectin and moxidectin in dairy sheep: assessment of drug residues during cheese elaboration and ripening period. J. Agric. Food Chem. 2004, 52, 6205-6211.
541 542 543
(13) Imperiale, F. A.; Pis, A.; Sallovitz, J.; Lifschitz, A.; Busetti, M.; Suárez, V.; Lanusse, C. Pattern of eprinomectin milk excretion in dairy sheep unaffected by lactation stage: Comparative residual profiles in dairy products. J. Food Prot. 2006, 69, 2424-2429.
544 545 546
(14) Imperiale, F.; Ortoz, P.; Cabrera, M; Farias, C.; Sallovitz, J. M.; Lezzi, S.; Perez, J.; Alvarez, L.; Lanusse, C. Residual concentrations of the flukicidal compound triclabendazole in dairy cows’ milk and cheese. Food Addit. Contam. 2011, 28, 438-445.
547 548 549
(15) Iezzi, S.; Lifschitz, A.; Sallovitz, J.; Nejamkin, P.; Lloberas, M.; Manazza, J.; Lanusse, C.; Imperiale, F. Closantel plasma and milk disposition in dairy goats: assessment of drug residues in cheese and ricotta. J. Vet. Pharmacol. Therap. 2014, 37, 589-594.
550 551 552 553
(16) Power, C.; Sayers, R.; O’Brien, B.; Clancy, C.; A. Furey, A.; Jordan, K.; Danaher, M. Investigation of the persistence of closantel residues in bovine milk following lactating-cow and dry-cow treatments and its migration into dairy products. J. Agric. Food Chem. 2013a, 61, 87038710.
554 555 556
(17) Power, C.; Danaher, M.; Sayers, R.; O’Brien, B.; Whelan, M.; Furey, A.; Jordan, K. Investigation of the persistence of rofoxanide residues in bovine milk and fate during processing. Food Addit. Contam. 2013b, 30, 1087-1095.
557 558 559
(18) Whelan, M.; Chirollo, C.; Furey, A.; Cortesi, M. L.; Anastasio, A.; Danaher, M. Investigation of the persistence of levamisole and oxyclozanide in milk and fate in cheese. J. Agric. Food Chem. 2010, 58, 12204-12209.
560 561
(19) Ziv, G.; Rasmussen, F. Distribution of labeled antibiotics in different components of milk following intramammary and intramuscular administration. J. Dairy Sci. 1975, 58, 938-946.
562 563 564 565
(20) Hakk, H.; Shappell, N. W.; Lupton, S. J.; Shelver, W. L.; Fanaselle, W.; Oryang, D.; Yeung, C. Y.; Hoelzer, K.; Ma, Y.; Gaalswyk, D.; Pouillot, R.; Van Doren, J. M. Distribution of Animal Drugs between Skim Milk and Milk Fat Fractions in Spiked Whole Milk: Understanding the Potential Impact on Commercial Milk Products. J. Agric. Food Chem. 2016, 64, 326-335.
566 567
(21) Anderson, R.R.; Collier, R.; Guidry, A.; Heald, C.; Jenness, R.; Larson, B.; Tucker, H.A. Lactation. Chapter 5, p 178, 190-191. Iowa State University Press, Ames, IA. 1985.
568 569
(22) Rusoff, L.; Lee, G.C.; Stone, E. J. Aureomycin (Chlortetracycline) distribution in milk. J. Dairy Sci. 1957, 40:1390.
570 571
(23) Adetunji, V. O. Effects of processing on antibiotic residues (streptomycin, penicillin-G and tetracycline) in soft cheese and yoghurt processing lines. Pakistan J. Nutri. 2011, 10, 792-795.
26 ACS Paragon Plus Environment
Page 27 of 41
Journal of Agricultural and Food Chemistry
572 573
(24) Ziv, G.; Sulman, F.G. Binding of antibiotics to bovine serum. Antimicrob. Agents Chemother. 1972, 2, 206-213.
574
(25) Drug Bank Database. http://www.drugbank.ca/ (accessed March 18, 2015).
27 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
575 576 577 578 579 580 581 582 583 584
Figure Legend
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603
Figure 3. Scheme (2) of milk partitioning for determination of percent drug associated with whey or casein protein. Results indicated in boxes, OTET not assayed, as protein associated fraction too small to accurately quantify.
Page 28 of 41
Figure 1. Structures of animal drugs used in the milk partitioning studies. a) Radioactively labeled with a single, general tritium atom, i.e. location of label is unknown. b) Asterisk indicates site of radiolabel. Asterisk within an aromatic ring indicates the ring was uniformly labeled. Figure 2. Scheme (1) of milk partitioning processes that yielded curd and whey from skim milk (Phase 2) and retentate and permeate from whey (Phase 3).
Figure 4. Drug distribution and relative concentration ratios from skim milk into whey and curd fractions. Bars represent percent mean of all concentrations (n=3 concentrations; n=3 replicates per concentration) ± standard deviation of all three dose mean percentages based on disintegrations per minute (dpm) of whey and curd (at 70% moisture) fractions compared to fortified skim milk dpm. Numerical values on graph represent the mean ratio (n=3) of the drug concentration in the fraction (curd or whey) to the initial drug concentration in skim milk ± SD. Sum of stacked plots represents total, unadjusted drug recovery values. Figure 5. Regression analyses of observed log [Drug]curd or 0% moisture curd/[Drug]whey (log C/W or 0%mC/W) vs. log P and log D (pH 6.8) using natural y intercept. Graph A is the regression analysis of log C/W vs. log P. Graph B is the regression of log C/W vs. log D (pH 6.8). Graph C is the regression analysis of log 0%mC/W vs. log P. Graph D is the regression of log 0%mC/W vs. log D (pH 6.8). Drug Bank pKas accessed on 2-11-2015 (www.drugbank.ca). Log P accessed from Chemspider on 1-28-2015 (www.chemspider.com). Calculations were performed as: log Dacid = log P + log[1/(1+10pH-pKa)] or log Dbase = log P + log[1/(1+10pKa-pH)].
604 605 606 607 608 609
Figure 6. Normalized percentages of animal drugs calculated to be in the milk end products (a) milk fat, (b) curd, (c) permeate, and (d) retentate based on data generated from the current studies. Percentage values in the curd and retentate bars represent pure curd percent and drug-towhey protein associations normalized to whey percentages. SDMX bar has additional information on which milk end products comprise whole milk, skim milk, high-fat curd, low-fat curd, and whey, as a guide to where drug may partition during commercial milk processing.
610 611 612 613 614 615
Figure 7. Drug distribution and relative concentration ratios of retentate and permeate fractions produced from fortified whey originating from whole milk. Bars represent percent mean of all concentrations (n=3 concentrations; n=3 replicates per concentration) ± standard deviation of the three dose means based on disintegrations per minute (dpm) of retentate and permeate fractions normalized to the total dpm between retentate and permeate. Numerical values on graph represent the mean ratios (n=3) of the drug concentration in the fraction (retentate or permeate) 28 ACS Paragon Plus Environment
Page 29 of 41
Journal of Agricultural and Food Chemistry
616 617 618
to the initial drug concentration in whey ± SD. Horizontal lines represent retentate and permeate percentage volumes from the filtration process. All drug distribution percentages were statistically different than volume percentages, p < 0.05.
619 620 621 622 623 624 625 626 627 628 629
Figure 8. Milk partitioning scheme for all seven animal drugs, including the radiochemically assayed partitioning values, for all fractions, i.e. Phases 1 to 3. Whey and curd for Phase 2 were produced from fortified skim milk. In Phase 3, whey was produced from whole milk, then fortified and retentate and permeate fractions prepared by ultrafiltration. The wheys produced from skim milk and whole milk were similar as seen in the compositional data from Table S1. Milk fat and skim partitioning values (Phase 1) are from Hakk et al. 2016. The moisture content of curd was assumed to reflect drug associated with whey, and was subtracted out to obtain the “0 % Moisture Curd” value. The % of drug that was whey protein-associated was calculated by difference (total in retentate minus concentration in permeate times the concentrating factor of retentate).
29 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 30 of 41
630
Table 1. Percentages of the extent that seven animal drugs in the current study associated with various milk protein fractions as described in Figure 3. Comparisons are also made to available literature values, footnotes with citation and differences in methodology. This Study, Casein Associated Drug
This Study, Whey Protein Associated
PENG
16
OTETd
24
25
Literature, Milk Proteins
Literature , Bovine Seruma (D-Dialysis; UFUltrafiltrate)
Drug Bank Unspecified Protein
7
5 (whole milk)b,11 9 (skim milk)c,19 8 (pure casein)b,11
28 (D), 33(UF)
45-68
7
8
52 (skim milk)e,19
20-67g
ERY
31
11
13 (skim milk)f,19
19(D), 22 (UF) 18(D), 20 (UF)
SDMX
37
8
KETO
38
17%
99
THIA
48
7%
70h
IVR
77
21%
93
Drug
75-95 50, 70g
a
in vivo administration of drug to cows; bovine serum, UF 1:10 retentate:permeate, dialysis 1 part of serum into 2 parts of buffer b bench-top fortification of milk or casein with drug c in vivo administration to goats d In this study, protein association too low for quantitation of OTET, values presented are from Phase 2 (curd) and 3 (whey protein) determined associations. e in vivo administration to goats, similar drug: tetracycline f in vivo administration to goats, similar drug: spiramycin g similar drugs: Sulfamethoxazole (70%) sulfapyridine (50%) h similar drug: albendazole
30 ACS Paragon Plus Environment
Page 31 of 41
Journal of Agricultural and Food Chemistry
Table 2. Calculated protein associated drug (nmoles per gram of casein or whey protein). Whey Whey Casein Casein Casein Protein Protein nmol/g Assoc./ Assoc./ [Incubation] nmol/g nmol/g (Phase 2 0% Drug Whey Whey nM (Phase 3 (Scheme 2 Moisture Assoc. (w/o Assoc. (w/ w/o w/ Curd)a casein) casein) casein) casein)b 20 0.07 0.24 nac 0.3 OTET 200 0.77 2.52 na 0.3 2,000 7.97 23.3 na 0.3 20 200 2,000
0.04 0.44 4.10
0.18 1.98 19.7
0.18
0.2 0.2 0.2
0.2
PENG
20 200 2,000
0.14 1.41 14.5
0.44 4.12 43.7
0.28
0.3 0.3 0.3
0.5
ERY
0.20 1.87 18.6
0.44 4.48 37.7
0.26
0.5 0.4 0.5
0.8
SDMX
20 200 2,000
0.19 1.86 18.2
0.74 6.91 84.5
0.46
0.3 0.3 0.2
0.4
KETO
20 200 2,000 75 200 2,000
1.50 3.82 39.6
1.02 2.90 31.1
0.61
1.5 1.3 1.3
2.5
THIA
20 1.08 3.25 0.55 0.3 2.0 200 8.32 31.2 0.3 2,000 88.4 308 0.3 a These data have whey associated drug subtracted, and were calculated for “0% moisture curd” consisting of casein protein (see text for details). b These data were obtained when drug was incubated with skim milk, curded, then the whey fraction subjected to centrifugal ultrafiltration to determine drug associated with whey proteins. These values would be more reflective of incurred drug distribution. Scheme 2 was not done with OTET, as counts were too low to reliably quantitate on a protein basis. c na is not analyzed. IVR
631
31 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 32 of 41
632
Figure 1. Structures of animal drugs used in the milk partitioning studies. a) Radioactively labeled with a single, general tritium atom, i.e. location of label is unknown. b) Asterisk indicates site of radiolabel. Asterisk within an aromatic ring indicates the ring was uniformly labeled.
* * 3
a
H(G)-Oxytetracycline (OTET) Tetracycline Antibiotic
14
3
14
b
C-Sulfadimethoxine (SDMX) Sulfonamide Antibiotic
*
* 14
b
C-Penicillin G (PENG) β-lactam Antibiotic
b
C-Thiabendazole (THIA) Fungicide Anthelmintic
a
H(G)-Ketoprofen (KETO) NSAID Analgesic
**
14
b
C-Erythromycin A (ERY) Macrolide Antibiotic
3
b
H-Ivermectin B1a (IVR) Avermectin Anthelmintic
32 ACS Paragon Plus Environment
Page 33 of 41
Journal of Agricultural and Food Chemistry
Figure 2. Scheme (1) of Milk partitioning processes that yielded curd and whey from skim milk (Phase 2) and retentate and permeate from whey (Phase 3).
33 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 34 of 41
Figure 3. Scheme (2) of milk partitioning for determination of percent drug associated with whey or casein protein. Results indicated in boxes, OTET not assayed, as protein associated fraction too small to accurately quantify.
34 ACS Paragon Plus Environment
Page 35 of 41
Journal of Agricultural and Food Chemistry
Figure 4. Drug distribution and relative concentration ratios from skim milk into whey and curd fractions. Bars represent percent mean of all concentrations (n=3 concentrations; n=3 replicates per concentration) ± standard deviation of all three dose mean percentages based on disintegrations per minute (dpm) of whey and curd (at 70% moisture) fractions compared to fortified skim milk dpm. Numerical values on graph represent the mean ratio (n=3) of the drug concentration in the fraction (curd or whey) to the initial drug concentration in skim milk ± SD. Sum of stacked plots represents total, unadjusted drug recovery values.
35 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 36 of 41
Figure 5. Regression analyses of observed log [Drug]curd or 0% moisture curd/[Drug]whey (log C/W or 0%mC /W) vs. log P and log D (pH 6.8) using natural y intercept. Graph A is the regression analysis of log C/W vs. log P. Graph B is the regression of log C/W vs. log D (pH 6.8). Graph C is the regression analysis of log 0%mC C/W vs. log P. Graph D is the regression of log 0%mC C/W vs. log D (pH 6.8). Drug Bank24 pKa’s accessed on 2-11-2015 (www.drugbank.ca). Log P accessed from Chemspider on 1-28-2015 (www.chemspider.com). Calculations were performed as: log Dacid = log P + log[1/(1+10pH-pKa)] or log Dbase = log P + log[1/(1+10pKa-pH)].
36 ACS Paragon Plus Environment
Page 37 of 41
Journal of Agricultural and Food Chemistry
Figure 6. Normalized percentages of animal drugs calculated to be in the milk end products (a) milk fat, (b) curd, (c) permeate, and (d) retentate based on data generated from the current studies. Percentage values in the curd and retentate bars represent pure curd percent and drug-to-whey protein associations normalized to whey percentages. SDMX bar has additional information on which milk end products comprise whole milk, skim milk, high-fat curd, low-fat curd, and whey, as a guide to where drug may partition during commercial milk processing.
37 ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 38 of 41
Figure 7. Drug distribution and relative concentration ratios of retentate and permeate fractions produced from fortified whey originating from whole milk. Bars represent percent mean of all concentrations (n=3 concentrations; n=3 replicates per concentration) ± standard deviation of the three dose means based on disintegrations per minute (dpm) of retentate and permeate fractions normalized to the total dpm between retentate and permeate. Numerical values on graph represent the mean ratios (n=3) of the drug concentration in the fraction (retentate or permeate) to the initial drug concentration in whey ± SD. Horizontal lines represent retentate and permeate percentage volumes from the filtration process. All drug distribution percentages were statistically different than volume percentages, p < 0.05.
38 ACS Paragon Plus Environment
Page 39 of 41
Journal of Agricultural and Food Chemistry
Figure 8. Milk partitioning scheme for all seven animal drugs, including the radiochemically assayed partitioning values, for all fractions, i.e. Phases 1 to 3. Whey and curd for Phase 2 were produced from fortified skim milk. In Phase 3, whey was produced from whole milk, then fortified and retentate and permeate fractions prepared by ultrafiltration. The wheys produced from skim milk and whole milk were similar as seen in the compositional data from Table S1. Milk fat and skim partitioning values (Phase 1) are from Hakk et al.18 The moisture content of curd was assumed to reflect drug associated with whey, and was subtracted out to obtain the “0 % Moisture Curd” value. The % of drug that was whey protein-associated was calculated by difference (total in retentate minus concentration in permeate times the concentrating factor of retentate). Phase 1
Whole Milk (100%) Fortified With OTET, PENG, ERY, SDMX, KETO, THIA, or IVR
Milk Fat (4.4% w/v) OTET: 1% PENG: 1% ERY: 2% SDMX: 5% KETO: 5% THIA: 22% IVR: 81% Skim (95% v/v) OTET: 100% PENG: 98% ERY: 97% SDMX: 92% KETO: 99% THIA: 78% IVR: 13%
0% Moisture Curd OTET: 7% PENG: 5% ERY: 16% SDMX: 22% KETO: 16% THIA: 42% IVR: 73%
Phase 2 Curd (11% w/v) OTET: 15% PENG: 12% ERY: 22% SDMX: 28% KETO: 23% THIA: 47% IVR: 75% Whey (85% v/v) OTET: 86% PENG: 85% ERY: 74% SDMX: 70% KETO: 72% THIA: 54% IVR: 21%
Phase 3 Retentate OTET: 32% PENG: 32% ERY: 39% SDMX: 39% KETO: 40% THIA: 32% IVR: 100% Permeate OTET: 68% PENG: 64% ERY: 54% SDMX: 59% KETO: 57% THIA: 62% IVR: 0.5%
ACS Paragon Plus Environment
Whey Protein Associated OTET: 8% PENG: 7% ERY: 17% SDMX: 14% KETO: 23% THIA: 12% IVR: 102%
39
Journal of Agricultural and Food Chemistry
Page 40 of 41
40 ACS Paragon Plus Environment
Page 41 of 41
Journal of Agricultural and Food Chemistry
41 ACS Paragon Plus Environment