Phytoestrogens and Their Metabolites in Bulk-Tank Milk: Effects of Farm Management and Season

Phytoestrogens have structures similar to endogenous steroids and may induce or inhibit the response of hormone receptors. The objectives of the present study were to compare the effects of long-term vs. short-term grassland management in organic and conventional dairy production systems, compare organic and conventional production systems and assess seasonal variation on phytoestrogen concentrations in bulk-tank milk. The concentrations of phytoestrogens were analyzed in bulk-tank milk sampled three times in two subsequent years from 28 dairy farms: Fourteen organic (ORG) dairy farms with either short-term or long-term grassland management were paired with 14 conventional (CON) farms with respect to grassland management. Grassland management varied in terms of time since establishment. Short-term grassland management (SG) was defined as establishment or reseeding every fourth year or more often, and long-term grassland management (LG) was defined as less frequent establishment or reseeding. The proportion of red clover (Trifolium pretense L.) in the herbage was positively correlated with milk concentrations of the mammalian isoflavone equol. Therefore, organically produced bulk-tank milk contained more equol than conventionally produced milk, and milk from ORG-SG farms had more equol than milk from ORG-LG farms. Milk produced during the indoor-feeding periods had more equol than milk produced during the outdoor feeding period, because pastures contained less red clover than fields intended for silage production. Organically produced milk had also higher concentrations of the mammalian lignan enterolactone, but in contrast to equol, concentrations increased in the outdoor-feeding periods compared to the indoor-feeding periods. There were no indications of fertility problems on ORG-SG farms who had the highest red clover proportions in the herbage. This study shows that production system, grassland management, and season affect milk concentrations of phytoestrogens. However, compared to soy products, milk concentrations of phytoestrogens are low and future studies are required to investigate if the intake of phytoestrogens from dairy products has physiological effects in humans.


Introduction
Phytoestrogens are phenolic, plant derived compounds and divided into several groups, including isoflavones, lignans and coumestans. In plants, phtyoestrogens have various functions such as defense against pathogens [1]. Phytoestrogens have structures similar to endogenous steroids and may induce or inhibit the response on hormone receptors in animals or humans. In sheep, high intake of phytoestrogens, especially formononetin has been found to impair fertility [2,3], whereas in cattle, effects are not consistent [3][4][5][6]. In humans, dietary phytoestrogens may protect against cancers or prevent osteoporosis and some may function as antioxidants. The most important source of phytoestrogens in the human diet are soy products [7], however, the intake of animal derived phytoestrogens may be significant [8].
The isoflavones, formononetin and daidzein, are to a large extent converted by rumen microorganisms to the mammalian isoflavone equol [9]. Equol has a much higher estrogen activity than daidzein [10][11][12]. Biochanin A and genistein can be converted to substances with no estrogenic activity [7,13]. The lignans secoisolariciresinol and matairesinol can be converted by intestinal microorganisms to enterodiol and enterolactone [14]. However, a recent study has shown that also ruminal Prevotella spp. can metabolize secoisolariciresinol to enterolactone [15]. Also other lignans such as pinoresinol, syringaresinol and lariciresinol are precursors of enterodiol and enterolactone [16].
Grassland management and production system affect sward botanical composition [17]. Therefore, it is likely that botanical composition affects the uptake of phytoestrogens and concentrations in milk. Red clover (Trifolium pratense L.) has high concentrations of formononetin [18], and several studies have shown high concentrations of equol in milk from cows fed silage containing red clover [19][20][21], or from cows grazing pastures containing red clover [22,23]. Elevated levels of equol have also been reported from organically produced consumer milk compared to conventional milk [24,25].
Less is known about the precursors of the mammalian lignans enterodiol and enterolactone. In the study by Höjer et al. [21], feeding silage produced from short-term grassland resulted in milk that contained less enterolactone than silage from long-term grassland, whereas in the study by Adler et al. [22] grazing a short-term grassland gave higher enterolactone concentration in milk than grazing a long-term pasture.
Although the above-cited studies indicate specific effects of botanical composition on phytoestrogen concentrations in milk, to our knowledge no attempts have been made so far to investigate the effect of botanical composition at the level of farming systems.
The objectives of the present study were to compare the effects of long-term vs. short-term grassland management in organic and conventional production systems, compare organic and conventional production systems and assess seasonal variation on phytoestrogen concentrations in bulk-tank milk.

Material and Methods Experimental Design
Twenty-eight dairy farms in central-Norway participated in the study in 2007 and 2008 and have been previously described by Adler et al. [17]. In brief, 7 organic (ORG) farms with shortterm grassland management (SG), referred to as ORG-SG-farms, were paired with 7 conventional (CON) farms with SG, referred to as CON-SG-farms, and 7 ORG-farms with long-term grassland management (LG), referred to as ORG-LG-farms, were paired with 7 CON-farms with LG, referred to as CON-LG-farms. Grassland management was defined as SG when the grassland fields of a farm were renewed every fourth year or more frequently and as LG when the fields were renewed less frequently. Grassland fields were renewed by soil tillage and seeding. Organic and conventional farms were paired on location and calving pattern, based on information from local extension services. The organic farms were certified by the Norwegian certification body Debio (Bjørkelangen, Norway) according to the EU standards for organic farming [26]. In brief, the standards for organic farming require a minimum forage intake in total dry matter intake (DMI; 50% in the first 3 months of lactation increasing to 60% thereafter) and all feeds have to be grown organically, i.e. without use of synthetic pesticides and synthetic N-fertilizers. Fertilization with animal manure is limited to 170 kg N/ha and yr. All farms participated in the Norwegian Dairy Herd Recording System and delivered milk to the same dairy company (TINE Norwegian Dairies SA, Oslo). On all farms, forages were fed ad libitum and allocated concentrate amounts were based on individual milk yields.

On Farm Analysis, Sampling and Data Collection
Data on farm characteristics were collected in farmer interviews and milk production data were collected from the Norwegian Dairy Herd Recording System. On average, the dairy farms had a farmland area of 28 ha (SD 9.6), a herd size of 19.6 dairy cows (7.9 SD), a forage area proportion of 0.86 (SD 0.150) for SG and 1.00 (SD 0.017) for LG, and a grassland age of 3 (SD 0.9) yr for SG and 11 (SD 3.8) yr for LG [17].
Herbage botanical composition before first cut silage in 2007 was estimated on 4 selected fields on each farm by the dry-weight-rank method [27], modified by Jones and Hargreaves [28]. The selected fields represented overall grassland use including fields that were cut, cut and grazed in combination, or only grazed.
In February, June and October in 2007, and February, June, August and October 2008, milk was sampled from stirred bulk-tanks and the samples were transported chilled (4°C) to the dairy.

Animals and Diets
On most farms cows were in tie-stalls and on 6 farms, mainly SG-farms, in loose-house barns. The cows were mainly fed grass silage based diets during the indoor-feeding periods (October to mid-May), and on all farms cows grazed in the outdoor-feeding periods, although many herds also had access to silage. Only herds with Norwegian Red dairy cows were included in the study, and calving time was rather evenly distributed over the year on all farms. Dairy feed rations included forage, concentrates, mineral mixtures and vitamin mixtures on all farms. Silage fermented in bulk silos or in round bales was the main forage in the indoor-feeding periods. On most farms the herbage was wilted before ensiling. In brief, the silages fed on ORG farms had had lower concentrations of crude protein (ORG: 138 g/kg of dry matter (DM), CON: 169 g/kg of DM), but higher concentrations of non-fibrous carbohydrates (ORG: 219 g/ kg of DM; CON: 175 g/kg of DM). Neutral detergent fiber concentration (mean: 570 g/kg of DM), in vitro digestibility (mean: 813 g/kg of DM) and net energy of lactation (NEL) (mean: 5.7 MJ/kg of DM) did not differ [17]. Pasture samples and concentrate samples were not analyzed for chemical composition. The cows were kept indoors at night on 8 farms (ORG-SG: 1, ORG-LG: 3, CON-SG: 1, CON-LG: 3) during the outdoor-feeding periods. Homegrown grains of barley and oats were fed in addition to commercial concentrates on most ORG-SG farms and supplemented with fishmeal on 2 ORG-SG farms.

Analysis Methods
Sample Preparation and Chemical Analysis of Milk. Milk samples intended for analysis of gross composition, urea, and FFA were preserved with 2-bromo-2-nitropropane-1,3-diol (Bronopol, D&F Inc., Dublin, CA) and analyzed by a Fourier transformed infrared spectroscopy milk analyzer (MilkoScan 6000 FTIR, Foss, Hillerød, Denmark). Milk samples intended for analysis of phytoestrogens were frozen at -20°C immediately after arriving at the dairy.
Phytoestrogen concentrations in herbage, concentrate and milk were analyzed according to the methods described by Steinshamn et al. [19]. Briefly, herbage samples were extracted with ethanol and acetate buffer (pH 5.0), then incubated with Cellulase Onozuka R-10 from Trichoderma viride (Merck, Darmstadt, Germany) at room temperature overnight, followed by centrifugation. Milk samples were defatted and deproteinized by mixing with acetate buffer (pH 5.2), heptane and acetone. The acetone/water phase was separated and evaporated to dryness and residues re-dissolved in water. Conjugates of phytoestrogens were cleaved by incubation with β-glucuronidase and sulfates from Helix pomatia type H2 (Sigma/Aldrich, St. Louis, MO) at 40°C for 4 h followed by centrifugation. Secoisolariciresinol, matairesinol, enterodiol, daidzein, enterolactone, equol, genistein, coumestrol, formononetin, prunetin and biochanin A were analyzed using the liquid chromatography (LC)-mass spectrometry (MS)/MS technique (Micromass, Manchester, UK) with standard addition [19].

Calculations and Statistical Analysis
Data of the participating herds were collected from the Norwegian Dairy Herd Recording System. Milk gross composition is the weighted average of all individual milk samples analyzed in one year on each farm. Energy corrected milk (ECM) yields were calculated as kilograms of milk per cow and year × (0.01 + 0.0122 × g of fat/kg of milk + 0.0077 × g of protein/kg of milk + 0.0053 × g of lactose/kg of milk).
Daily concentrate intake for individual cows was forwarded by the farmers to the Norwegian Dairy Herd Recording System and mean concentrate intake per year and cow was calculated. Forage intake of NEL per cow and year was estimated by the Norwegian Dairy Herd Recording System as net energy requirement for maintenance, production, activity and pregnancy and subtracting NEL supplied from concentrates. The cow fertility in the participating herds was evaluated by three indicators, age at first calving, calving interval and animals culled due to fertility problems per cow and year. All data were based on the Norwegian Dairy Herd Recording System.
If the content of phytoestrogens was below the limit of detection, half the detection limit was used to be able to run the statistical analysis. The limits of detection in milk were 0.391 μg/kg for secoisolariciresinol, 0.021 μg/kg for matairesinol, 0.018 μg/kg for enterodiol, 0.072 μg/kg for daidzein, 0.067 μg/kg for enterolactone, 0.040 μg/kg for equol, 0.092 μg/kg for genistein, 0.053 μg/kg for coumestrol, 0.012 μg/kg for formononetin, 0.014 μg/kg for prunetin and 0.023 μg/kg for biochanin A.
Milk concentrations of phytoestrogens were analyzed using the MIXED model procedure by SAS [29]. The statistical model (1) was used: where Y were the individual dependent variables for individual phytoestrogen concentrations in milk (n = 1 to 168) and μ was the average of all observations, G was the fixed effect of grassland management (i = 1, 2; where 1 = SG and 2 = LG), P(G) was the fixed effect of production system within G (j = 1, 2; where 1 = ORG and 2 = CON), M was the fixed effect of month (k = 1 through 6; where 1 = February 2007, 2 = June 2007, 3 = October 2007, 4 = February 2008, 5 = June 2008, 6 = October 2008), (GM) and (MP(G)) were interactions of the fixed effects, f was the random effect of farm within G and P (l = 1 through 28), and t was the random effect of farm pair within G (m = 1 through 14) and e ijklmn were the random residual errors, assumed to be independent and N(0, σe2). Observations for month within farm were treated as repeated observations. Contrasts were calculated for the effects of ORG-SG vs. ORG-LG, CON-SG vs. CON-LG, ORG vs. CON and indoor (k = 1, 3, 4, 6 vs. outdoor-feeding periods (k = 2, 5). Differences between means were tested with the Tukey-Kramer test.
In order to find correlations between proportions of botanical families in the herbage, concentrate DMI and milk concentrations of phytoestrogens a principal component analysis was performed by the PRINCOMP procedure in SAS [29]. Average milk concentrations of phytoestrogens and concentrate DMI were calculated for indoor-feeding and outdoor-feeding periods.
The botanical composition of fields that were cut or both cut and grazed were associated with the indoor-feeding periods, whereas the botanical composition of fields that were grazed or both cut and grazed were associated with the outdoor-feeding periods.

Botanical Composition of Diets
Fields on CON-farms that were cut or cut and grazed in combination had higher proportions of grasses in the herbage (mean 905 g of DM/kg of DM) than fields that were grazed only (mean 670 g of DM/kg of DM) ( Table 1). Compared with CON-farms, ORG-farms had lower proportions of grasses (mean 574 g vs. 826 g of DM/kg of DM) and differences between fields that were cut, cut and grazed in combination, or only grazed were small. The prevailing grass species were timothy (Phleum pratense L.) and meadow fescue (Festuca pratensis Huds.) in all management systems. Smooth meadowgrass (Poa pratensis L.) was most common on fields on ORG-SG and CON-SG farms that were grazed only. Red clover was mainly found on fields on ORG-SG and ORG-LG farms that were cut or cut and grazed in combination. In fields that were grazed only, white clover (Trifolium repens L.) was the prevailing legume species. The prevailing species belonging to other plant families were northern dock (Rumex longifolius DC.) and dandelion (Taraxacum spp.). Other botanical families than grasses and legumes were found in higher proportions in fields that were grazed only than other fields. Fields on ORG-LG farms had the highest proportions of these plant families.

Effects of Grassland Management, Production System and Season on Feed Intake, Milk Yield and Milk Gross Composition
Concentrate intake per cow and year and per 100 kg of ECM produced were on average 32% (P <0.001) and 24% (P = 0.001), respectively, lower on ORG-farms than on CON-farms ( Table 2). Forage intake per cow and year was on average 10% higher (P = 0.02) on SG-farms Table 1. Botanical composition estimated before first cut in 2007 on fields, which were cut, cut and grazed in combination or only grazed on dairy farms with organic production system (ORG) and short-term (ORG-SG) or long-term grassland management (ORG-LG) and dairy farms with conventional production system (CON) and short-term (CON-SG) or long-term grassland management (CON-LG).  Table 2. Feed intake, milk production, milk gross composition and fertility indicators on dairy farms with organic production system (ORG) and short-term (ORG-SG) or long-term grassland management (ORG-LG) and dairy farms with conventional production system (CON) and short-term (CON-SG) or long-term grassland management (CON-LG). 2 P(G) = Effect of production system within grassland management. 3 Contrast of ORG-SG vs. ORG-LG. 4 Contrast of CON-SG vs. CON-LG. 5 Contrast of ORG vs. CON. 6 Net energy lactation. 7  than on LG-farms, whereas forage intake per 100 kg of ECM produced was 22% higher (P = 0.004) on ORG-farms than on CON-farms. Milk yield per cow and year was 12% lower (P = 0.009) on ORG-farms than on CONfarms. The cows on SG-farms yielded on average 15% more (P = 0.002) milk with 4% more (P <0.001) protein than the cows on LG-farms. Lower milk concentrations of fat and higher concentrations of lactose were found in the indoor-feeding periods than the outdoor-feeding periods [17].
Grassland management did not affect milk concentrations of total or individual lignans. Compared with milk from CON-farms, milk from ORG-farms had higher concentrations of secoisolariciresinol (P = 0.008), enterodiol (P = 0.006), enterolactone (P <0.001) and total lignans (P <0.001). Milk concentrations of secoisolariciresinol (P <0.001) and matairesinol (P <0.001) were higher during the indoor-feeding periods, whereas the concentrations of enterodiol (P = 0.01), enterolactone (P <0.001) and total lignans (P <0.001) were higher during the outdoor-feeding periods. However, in August 2008 the concentrations of enterolactone were lower than in June and similar to the indoor feeding periods (Fig 1). Enterolactone was the lignan found in highest concentrations in milk samples from all farming systems (9 to 335 mg/kg milk).
Coumestrol was not affected by grassland management, production system or season.

Indicators of Fertility
Compared to CON-farms, ORG-farms had slightly longer (P = 0.03) calving intervals, but fewer (P = 0.04) cows were culled due to fertility problems. Age at first calving was similar for all farming systems. None of the fertility indicators were affected by grassland management.

Botanical Composition of Diets
Differences in seed mixtures, harvesting managements, N-fertilization level and use of herbicides may explain the differences in botanical composition between farming systems. Legumes are important forage species in organically farmed grassland systems due to their N-fixation capability. In Norway, red clover is commonly included in organically grasslands intended for harvesting by cutting or a combination of cutting and grazing, whereas white clover is often included in pastures. In conventionally managed grasslands N-fertilizers boost plant growth in early spring favoring the growth of grasses in preference to legumes. Other plant species may have been favored by absence of herbicides on organically managed grasslands, long grassland duration on LG-farms or grazing animals in all systems. Mean values in bulk-tank milk from dairy farms with organic production system and short-term (ORG-SG) or long-term grassland management (ORG-LG) and dairy farms with conventional production system and short-term (CON-SG) or long-term grassland management (CON-LG) in indoor (n = 7, error bars indicate standard error of the mean). doi:10.1371/journal.pone.0127187.g001

Effects of Grassland Management, Production System and Season on Phytoestrogen Concentrations in Milk
Isoflavones. High concentrations of equol in milk from ORG-SG farms is in accordance with other studies where cows were fed red clover containing silage [19][20][21] or grazing red clover containing pastures [22,23]. In the studies by Andersen et al. [23] and Steinshamn et al. [19], the concentrations of equol were similar to the present study, whereas in the studies by Höjer et al. [21] (experiment 1: 1,494 μg/kg, experiment 2: 716 μg/kg) and Adler et al. [22] (1,199 μg/kg), the concentrations were considerably higher than in the present study. High concentrations of equol in milk are related to high intake of the precursors formononetin and daidzein [18]. Formononetin is found in high concentrations in red clover [30] and thus the proportion of red clover in the forage is an important factor. Furthermore, the concentrations of formononetin in red clover depend on plant phenological stage, leaf-stem ratio and season [31][32]. In the studies of Andersen et al. [20,23], Höjer et al. [21], Adler et al. [22] and Steinshamn et al. [19], the formononetin concentrations in red clover grass silage or pasture ranged from 2,790 to 11,420 mg/kg DM and concentrations in white clover grass silage or pasture Phytoestrogens in Bulk-Tank Milk ranged from 142 to 201 mg/kg DM. Unfortunately, it was not possible to measure red clover intake and concentration of phytoestrogens in the herbage in the present study.
On the farming system level, similar differences as in the present study have been found between organic and conventional farms. Hoikkala et al. [25] found 411 μg/L of equol in organic skimmed milk whereas conventionally produced milk contained 62 μg/L of equol. Milk samples were collected between January and March and represent milk produced during the indoor-feeding periods. Also Antignac [24] found higher concentrations of equol in organic milk In the present study, less red clover was found in pastures compared to fields that were cut or cut and grazed in combination, and differences were greater on ORG-farms than on CONfarms. Thus, it is likely that the red clover proportion in the herbage DMI was higher during the indoor-feeding periods than the outdoor-feeding periods. Herbage botanical composition was estimated before first cut in 2007. In organic farming, red clover often makes up a higher proportion in the herbage yield of the second cut than of the first cut [33]. Therefore, red clover intake was most likely higher on ORG-farms than what could have been expected based on the botanical composition before the first cut. It has been shown that including concentrate in the diet decreases equol concentrations in milk slightly [19]. Thus, the lower concentrate intake in cows on ORG-farms compared to CON-farms may partly explain a part of the differences in milk equol concentrations. The observed differences in milk yield may also have affected the recovery of isoflavones in milk.
Phytoestrogen concentrations in feed were not analyzed in the present study, but based on reported concentrations in other studies the contribution of forages and concentrates can be estimated. Soybeans (Glycene max (L.) Merr.) have high concentrations of daidzein (105 to 560 mg/kg DM), but only small amounts of formononetin [7]. Typically, concentrations of formononetin in silages including red clover are 2,500 to 3,000 mg/kg DM, in silages with white clover about 150 mg/kg DM and in grass based silages <50 mg/kg DM, whereas concentrations of daidzein are much lower [19][20][21]34]. Inclusion of soybeans in concentrate mixtures was on average 4% and 12% in organic and conventional production, respectively. In addition, the concentrate level was higher on organically than on conventionally managed farms. Thus, the intake of the equol precursors from concentrate was likely much higher on CON-farms than ORG-farms. Thus, in organically produced milk, equol derives primarily from the forage, whereas in conventionally produced milk a large share derives from soybeans.
Lignans. The concentrations of lignans in milk were not associated with grassland management, but with production system and season. Unlike the situation for isoflavones, the reasons for the observed effects on lignans are not clear. Compared with milk from CON-farms, milk from ORG-farms had more secoisolariciresinol, enterodiol and enterolactone. A possible reason is that the diets fed on ORG-farms contained higher levels of mammalian lignan precursors such as secoisolariciresinol, mataresinol, lariciresinol and pinoresinol [35]. High concentrations of these precursors are found in flaxseed and sesame seed, but also brassica vegetables contain significant concentrations. However, brassicas were only found in small proportions (Shepherd's-purse, Capsella bursa-pastoris (L.) Medik.). Milk from the indoor-feeding periods contained more secoisolariciresinol and mataresinol, but less enterodiol and enterolactone than milk from the outdoor-feeding periods. The higher concentrations of enterodiol and enterolactone in milk from the outdoor-feeding periods indicate that plant species rich in mammalian lignan precursors may be found in higher proportions in pastures than fields cut or cut and grazed in combination. These may be unsown grasses or non-legume dicotyledons such as northern dock and dandelion. However, the principal component analysis of herbage botanical composition and milk concentrations of phytoestrogens did not show any correlations between enterolactone concentrations in milk and plant species. The low concentrations in August 2008 indicate that the intake of enterolactone precursors may depend on the phonological stage of forage plants. The high concentrations in June may be related to the lignification process during the elongation stage in grasses. Furthermore, the stability of mammalian lignan precursors in silage may have affected the intake of lignans.
Steinshamn [19] found positive effect of concentrate on enterolactone concentrations in milk. This effect may have decreased the differences between ORG-farms and CON-farms caused by differences in forage intake. In contrast, Antignac et al. [24] found no effect of productions system on enterolactone.
Botanical Composition and Phytoestrogens in Milk. The proportion of red clover in the herbage is positively correlated with milk concentrations of the mammalian isoflavone equol. Therefore, organically produced bulk-tank milk contained more equol than conventionally produced milk, and milk from ORG-SG farms had more equol than milk from ORG-LG farms. Milk produced in the indoor-feeding periods had more equol than milk produced in the outdoor feeding period, because pastures contained less red clover than fields intended for silage production. Organically produced milk had also higher concentrations of the mammalian lignin enterolactone, but in contrast to equol, concentrations increased in the outdoor-feeding periods compared to the indoor-feeding periods.
Phytoestrogens and Fertility in Dairy Cows. ORG-SG farms who had the highest proportions of red clover in the forage and the highest milk concentrations of phytoestrogens did not differ from ORG-LG farms in terms of the fertility indicators. The red clover proportion before first cut in 2007 was 17.4% for ORG-SG and even if a higher proportion can be assumed for the second cut, the average intake of red clover may be only moderate explaining no observation of impaired fertility on these farms relative to the others. The slightly longer calving intervals on ORG-farms, may be due to lower plane of nutrition as indicated in a previous field study on organic dairy farms [36]. Higher culling frequency on conventional farms may be due to higher production aims. In conclusion, neither intake of red clover or phytoestrogen concentrations in milk can be related to fertility issues in the present study.
This study shows that production system, grassland management and season affect milk concentrations of phytoestrogens. However, compared to soy products, milk concentrations of phytoestrogens are low and future studies are required to investigate if the intake of phytoestrogens from dairy products have physiologic effects in humans.