Industrial chemical bisphenol A and raw milk: a toxicokinetic study in lactating dairy sheep after repeated dietary and subcutaneous administration

• Background: Dietary intake is the predominant route for human exposure to bisphenol A and one of the food items important for humans is milk, and BPA-polluted animal feed and environments may thus affect human exposure. The aim of our study was to evaluate the BPA exposure and disposition in sheep milk after repeated dietary and subcutaneous administration of a relatively low dose (100 µg/kg of body weight per day) of BPA to a sheep. • Results: With our toxicokinetic model, we showed that most likely only free BPA passes into the mammary gland and is subsequently conjugated there. The percentage of the dose eliminated with milk was less than 0.1%, regardless of the route of BPA administration. • Conclusions: It is proven that the BPA is eliminated through the milk of lactating sheep. However, the amounts excreted in the milk that were detected in this study are minimal.


Background
Since the start of the commercial production of bisphenol A (BPA) in the 1950s until the present, the global production and consumption of this substance, regardless of the suspected negative health effects, has continued to rise [1]. With both the wide use of BPA and its leaching from many products and materials [2], it is known to be one of the ubiquitous environmental contaminants [3]. The main route of BPA exposure is thought to be oral ingestion (up to 83% of the total estimated exposure), and in 2013 canned products accounted for about 50% of the dietary exposure to BPA. Thus, cans and packaging are believed to be the main source of contamination in foods [4]. However, the products from farm animals, being directly exposed to human pollution, could still be, in some cases, an additional risk factor for human exposure.
One such product, of which production is inseparably linked to the environment and depends largely on human activities, is milk. The application of contaminated material on the soil, such as sewage sludge or industrial waste, and atmospheric deposition from nearby industrial activities, have resulted in a broad range of environmental contaminants that enter the milk chain [5]. It is also true that chemicals can enter milk even during the collection and preparation processes of dairy products [6]. For instance, BPA may be introduced during milking from plastic parts of the milking machines , or also transferred from bulk milk to plastic storage tanks [7].
Finally, BPA can also migrate as an additive from packaging material into the consumable milk. The actual levels of BPA found in commercial milk samples are presented in the review of Mercogliano and Santonicola, and are in the range between not detected (ND) to 521 ppb [5].
To the best of our knowledge, only a few in vivo studies are published regarding BPA transfer to milk, with all of them using rodent models, and all report limited excretion of BPA into milk [8][9][10][11]. Doerge et al. evaluated the lactational transfer of BPA after repeated oral dosing in rats, and found concentrations of 0.83 +/-0.26 nM of free BPA and 7.6 +/-2.8 nM of total BPA 1 hour after the administration of 100 µg/kg of b. w.. They calculated that doses delivered to pups lactationally were 300-fold lower than the dose administered to the dams [11].
The aim of our study was to estimate the transfer of BPA from feed or via subcutaneous administration to milk. To do so, one Slovenian autochthonous dairy sheep, an Istrian Pramenka, and her lamb were used in the study. Time courses of the free, conjugated and total BPA concentrations were followed in the ewe's blood plasma after repeated dietary and subcutaneous administration, as well as BPA transfer in milk. We also aimed to assess lactational transfer of BPA to the suckling lamb by estimating BPA exposure in its blood plasma.

Validation of the analytical methodology used
The validation parameters of the BPA blood plasma and milk analysis are presented in Table 1. The method was linear for BPA standards and matrices, as proved by the determination coefficients (r 2 ) of ≥0.999 and ≥0.991, respectively. Mean recoveries for free and total BPA in the blood plasma were 82.3 and 49.5%, respectively and in the milk 62.9 and 54.3%, respectively. The total BPA refers to the sum of free and conjugated BPA.
The coefficients of variation (CVs) of the concentrations detected and recovery in the fortified samples were from 1.5-24.4% under within-laboratory reproducibility conditions. Limit of detection (LOD) values were 0.05-0.1 µg/L and 0.2-0.4 µg/L for the free and total BPA determination, respectively, and differed according to a more comprehensive chromatographic background in the total BPA extracts. Toxicokinetic analysis BPA levels were checked before conducting both the first and second part of the experiment to provide a baseline for the analysis. The ewe entered the first part of the experiment with 0.05 and <0.4 µg/L of the free and total BPA in the blood plasma, respectively, and with <0.1 and 0.31 µg/L of the free and total BPA in milk, respectively.
Just before the start of the subcutaneous administration, the ewe's blood plasma contained 0.15 and 0.72 µg/L of the free and total BPA, respectively, while its milk contained <0.1 and 0.35 µg/L of the free and total BPA, respectively.

Comparison of the plasma concentration-time profiles
The maximum plasma concentration of free BPA obtained after subcutaneous administration was higher than after dietary administration. In addition, free BPA exposure was prolonged after subcutaneous administration compared to the dietary route of intake.
With the dietary route, the maximum plasma concentration (c m ax ) of free BPA was 2.15 µg/L and was obtained very quickly, at 0.33 h. For the subcutaneous route, c m ax of free BPA was 6.41 µg/L and was obtained after 2 h.
The c m ax values of BPA-conjugate were similar for both routes of exposure and were 49. Administration by the subcutaneous route led to a higher overall internal exposure to free BPA and lower internal exposure to conjugated/total BPA compared to the dietary route.
Clearance and relative bioavailability of free BPA, BPA-conjugate and total BPA obtained with noncompartmental toxicokinetic (TK) analysis are presented in Table 2.  Table 3. Blood plasma BPA-conjugate and total BPA concentration time courses were described with a one-compartment model, while a two-compartment model was more suitable for free BPA.   Regarding the suckling lamb, which drank milk after his mother was administered with BPA by the dietary or subcutaneous routes, there were only traces of BPA in the samples of its plasma.

Discussion
The purpose of this study was to investigate the toxicokinetics of BPA and to evaluate its elimination into the sheep milk after two different routes (po and sc) of repeated low dose BPA administration.
A comparison of the plasma concentration-time profile for the basic TK parameters of the two administration routes was made using the noncompartmental approach. Regarding the comparison of both routes of BPA administration, our results are similar to those in Guignard et al. [12], where the TK parameters for the same routes of administrations but with higher dose regimens were compared. The formulations for the dietary and as well for subcutaneous route of administration were similar in both studies. In our study, the c m ax of free BPA for dietary administration was obtained quickly (0.33 h). In their study, mean c m ax was attained 0-12 h for three ewes and 0.20 h for two others. For the subcutaneous route, c m ax in our study was obtained after 2 h, in their study it was obtained after 2 h for three ewes and after 1 h for one ewe. In our study, the free BPA c m ax for the subcutaneous route was three-fold higher than for the dietary route and in their work the free BPA c m ax for the subcutaneous route was 4.6 ± 1.5-fold higher than for the dietary route. Our study demonstrates a higher cumulative (AUC) internal exposure to free BPA after subcutaneous administration compared to the dietary route, which is in line with the findings of Guignard et al. [12]. In their study the relative bioavailability of BPA for the dietary as compared to subcutaneous route was 3.3 ± 0.3%. In our work, the relative bioavailability of BPA for the dietary as compared to subcutaneous route was 4.5%. Both this earlier work and the current study were also in agreement with regard to the BPA-GLUC concentration time course. Unlike free BPA, the BPA-GLUC concentration time courses are very similar for the two routes of exposure.
The comparison of our study to Guignard et al. [12] is important to ensure our data coincided well with theirs, as a limitation of our work was the use of only one animal and a very low dose of BPA, which resulted in even lower measured concentrations in plasma.
The similarity of the results from both studies thus indicates the credibility of our data. This is important, as these data were the base for our TK model, which we used to evaluate the elimination of BPA into the sheep milk. Sampling of the milk was possible only at a couple of sampling points, and thus it was not possible to make time-concentration profiles for it. However, our TK model enabled us to estimate the percentage of the dose eliminated with milk, which was less than 0.1% for free BPA, conjugated-BPA and total BPA, regardless of the route of administration. This result is comparable with the results of Snyder et al., where they found only a small fraction of the 14 C labelled BPA (0.63+/-0.13 µg/equiv/mL) 8 hours after dosing [8].
Regarding free, conjugated and total BPA, it is already indirectly proven in rats that free BPA is transferred into the mammary gland to a greater extent than bisphenol A glucuronide (BPA-GLUC) [11,13]. Given our TK models, the same was true in our study for the ewe. In the first model we were assuming passive transfer (first-order) of free BPA into milk, and in the second we were assuming that conjugated BPA would also be transported.
Based on the lower value of the Akaike information criterion (AIC), with the first model it is more likely that only free BPA is transferred into the mammary gland. Nevertheless, it was reported that the major molecular species in the milk of rats after oral administration of 14 C -BPA was BPA-GLUC [8]. The concentrations measured in milk six hours after BPA administration (dietary and subcutaneous) in our study show the same result. Six hours after dietary administration the concentration of free BPA was 0.05 µg/L and the concentration of BPA-GLUC was 0.78 µg/L. Similarly, the concentrations of free BPA and BPA-GLUC after subcutaneous administration were 0.87 and 1.89 µg/L, respectively.
Regarding the BPA-GLUC in the milk, we hypothesise that free BPA is passively transferred into the mammary gland, and subsequently conjugated in its glucuronidated form by the tissue, although with glucuronidation activities that are much lower (by more than 100,000-fold) compared with those seen in the liver [14]. There are currently no (to the best of our knowledge) known studies that have evaluated the presence of UDPglucuronosyltransferases in the ewe mammary gland, although it seems reasonable to assume that the mammary glands of all mammals are equipped with similar detoxifying mechanisms.
The above mentioned concentrations of free BPA measured in this study are well within the range with the concentrations found in raw milk measured in a recent Italian monitoring study, where the concentrations of only free BPA ranged from 0.081 -2.492 µg/L [15]. However, the concentrations measured in commercial milk samples were generally higher (from 14.0 to 521.0 µg/L) [5], meaning that the BPA load in consumption milk is greater at the end of the production line.

Conclusion
Considering the widespread consumption of milk and dairy products, the origins of milk contamination with BPA should be well investigated. To the best of our knowledge, this is the first study in which BPA elimination in milk was evaluated in sheep. Our study carried out in an animal model relevant to dairy cattle shows that BPA and BPA-GLUC are detected in the milk sample obtained from the ewe administered by two different routes of administration. We estimated that the percentage of the eliminated BPA in the milk is less than 0.1% of the administered dose for both dietary and for subcutaneous routes. were kept under natural temperature and photoperiodic conditions, with free access to water, hay and salt. In addition, the sheep was fed twice a day with 400 g plant based pellets (SchafKorn Lac, Unser Lagerhaus Warenhandels Ges., Austria). Eventual contamination of the experimental environment was checked by preliminary testing of drinking water and pellets by HPLC analysis, which revealed the slight presence of BPA of 0.02 µg/L and 5 µg/kg in these two matrices, respectively. The sheep and its lamb were, at both periods of the study, penned individually the day before the first administration until three days after the last administration. The lamb was kept with its mother, except on sampling days, when they were separated for a few hours before sampling time to collect enough milk for analysis. The animals were clinically healthy, as indicated by medical (temperature, breathing and rumination frequency, pulse rate), haematological, biochemical and faecal examinations. Fourteen days after the second experimental period, the sheep and its lamb were released in their original herd.

Chemicals
A sheep was chosen for this study due to its physiological similarities with cows, but easier manipulation. However, cows are mainly used in milk production in Europe (accounting for 96.9% of the total milk produced) [16].

Experimental design
The experiment was divided into two periods, the first being the dietary administration period and the second being the subcutaneous administration period. The same ewe was used for both exposure routes, thus a 13 days wash-out period was permitted to ensure that BPA was removed from the body of the ewe before the start of the second period.
Regarding the administration of BPA, in the first period the ewe received BPA in its diet (100 µg/day/kg of body weight) for five consecutive days (dietary route of administration).
The ewe ingested all pellets within 2-9 minutes. During the second period, the same ewe was injected in the shoulder area with 100 µg/kg of b. w. of BPA subcutaneously per day for five consecutive days (subcutaneous route of administration).
On the first day of the dietary period of the experiment, the ewe`s blood samples were taken at time 0 (before the first administration) and 0.083, 0. 16 (v/v) of acetonitrile, gradient to 12 min, 35-50% (v/v) of acetonitrile, held to 20 min, gradient to 20.5 min, 50-35% (v/v) of acetonitrile, held to 21 min. The excitation and emission wavelengths of the fluorescence spectrophotometry analysis were set at 230 and 315 nm, respectively [18]. The results were evaluated in accordance with the external standard method using a standard calibration curve as a function of chromatographic peak areas and standard concentrations. Each sample series consisted of a matrix sample, obtained before the first periodic BPA administration (a baseline sample), five to seven animal study samples in duplicate and two baseline matrix samples fortified with BPA to control the recovery rate. The measured sample concentrations were corrected for the possible baseline matrix response and for the mean recovery of the respective series and then used as final results.
Validation of the analytical methodology used was performed to demonstrate its fitness for the stated purpose. Linearity was determined by the least-squares method to calculate regression and correlation parameters for six to seven standard concentration points per calibration curve (range 1.0-100 ng/mL), and for both matrices as a correlation between measured and added concentrations (ranges 0.25-10 μg/L and 1.0-50 μg/L for free and total BPA in blood plasma, respectively, 0.5-15 μg/L for both free and total BPA in milk).
Mean recovery was evaluated by analysis of four to six fortified blank materials at two concentration levels at separate time points (blood plasma: free BPA 2 and 10 μg/L, total BPA 25 and 50 μg/L; milk: free BPA 2 and 5 μg/L, total BPA 5 and 10 μg/L). The withinlaboratory reproducibility of the method was evaluated as the CV of the determined and recovery values. The LOD value was estimated as the BPA concentration in the retention time window where the analyte was to be expected, which corresponded to 3 × noise and was corrected for the blank matrix response.

Toxicokinetic analysis
Each entity (free, conjugated, and total) plasma concentration time course until the second BPA administration was first analysed using a noncompartmental approach to obtain the estimates of the area under the concentration-time curve extrapolated to infinity (AUC), maximum concentration in plasma and time when it occurs (c m ax and t m ax , respectively). AUC was calculated using the linear trapezoidal method and extrapolated to infinity by addition of the term C last /λ z , where C last is the last quantified concentration measurement and λ z is the terminal slope of the concentration profile in the semi-log plot calculated by linear regression. t m ax and c m ax were reported as observed. AUCvalues were used to estimate clearance (CL) as CL = Dose/AUC sc and relative bioavailability after dietary administration (F r ) as F r = AUC po /AUC sc . The indexes po and sc refer to the route of administration (dietary and subcutaneous, respectively) and Dose is the single BPA dose (100 µg/kg of b. w.). Note that CL can be estimated only after intravenous administration. Our estimate of CL is therefore apparent clearance, i.e. assuming complete bioavailability after subcutaneous administration.
Subsequently, all TK data after both routes of administration were simultaneously fitted to a one-and two-compartment model with first-order absorption and elimination. The estimated parameters were clearance (CL), volume of the central and peripheral compartment (V c and V p , respectively), distribution clearance (Q), absorption rate constants after subcutaneous and dietary administration (k a sc and k a po , respectively) and relative bioavailability (F r ). Parameter fitting was performed using ADAPT II software [19] with the maximum likelihood method and a proportional variance model, where V i is the variance of the i-th data point and Y i is the value predicted by the model.
The AIC value was used to select the model.
Permeation of free, conjugated and total BPA into milk was modelled as a first order process dA m /dt = k m × C p (t), where dA m /dt is the transfer rate in µg/h, C p (t) is the BPA plasma concentration at time t, and k m is the transfer rate constant. k m was estimated by simultaneous fitting of the amounts excreted into milk up to six hours after the first subcutaneous and dietary administration, with TK parameters for the plasma data fixed to previously estimated values. The amounts excreted in milk up to 6 h were approximated by multiplication of the concentration in milk at 6 h by 0.25 L, i.e. assuming an average milk yield of 1 L/day. We tested the hypothesis that only free BPA is transferred into milk and subsequently conjugated in the mammary gland, i.e. fixing the TK parameters to the values estimated for the free BPA versus the hypothesis that conjugated BPA is also transferred, i.e. fixing the TK parameters to the values obtained for the conjugated and total BPA.

Consent for publication
Not applicable.

Availability of data and materials
The datasets analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.     Study design with two experimental periods, BPA administration and blood sampling schedule for the ewe.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.