The whole foliage of B. papyrifera was mowed with 20 cm of stubble left from saplings of about 1.8 m height. The whole crop was ensiled without additive after chopping to a length of 2.5±1.5 cm, The B. papyrifera silage was fermented for 45 days before being used for the feeding experiment. Seventy-two Holstein cows with initial day in milk (DIM) = 128±65 d, body weight = 572±29 kg, milk yield = 32.16±0.95 kg and milk somatic cell counts (SCC) = 158×10³±35×10³ cells/mL were divided into four groups randomly according to day in milk, parity, milk yield, and milk SCC, and allocated in 6 pens with 3 individuals in each group and given ad libitum access to feed and water. The cattle in the control group were fed on the original total mixed ratio (TMR) in the dairy farm, and the other 3 treatment groups were fed on the new TMR with 5% (Group low [Group L]), 10% (Group middle [Group M]), or 15% (Group high [Group H]) of B. papyrifera silage preparation, separately. The four different TMRs were adjusted as isonitrogenous and isoenergetic diets, with the composition shown in Table 2. The cattle were fed approximately 110% of expected consumption at 8:00 and 17:00 separately, and the residual was collected before the next feeding and then weighed and recorded. The delivery amount was adjusted according to the previous day’s consumption. Cows were milked at 6:00, 13:00, and 19:30 in the milking parlor. The feeding adaption period was 7 days, and the experiment period was 42 days. The body condition score was measured by 2 evaluators and scores were averaged.

The composited dietary samples, including feeding ingredients, TMRs and residual, were collected on fresh matter basis every 2 weeks during the experimental period. The samples were dried in an oven at 65°C for 48 h, then kept at room temperature for 24 h, and ground to pass through 40-mesh sieve. Dry matter (DM), crude fat (ether extract), ash, neutral detergent fiber (NDF), acid detergent fiber (ADF) were determined according to the methods reported by van Soest et al [16] and Goering and Van Soest [17], respectively.

The pH value of B. papyrifera silage was 5.04 and the nutrient composition of B. papyrifera silage was 29.06% of DM, 15.01% of crude protein (CP), 58.78% of NDF, 32.15% of ADF, 6.17% of ash at DM basis (Table 1), and 4 different TMRs are shown in Table 2. Milk yield was recorded every day, and milk samples were collected weekly, 40 mL of the milk was added with Potassium Dichromate for preservation and kept in −4°C for milk component analysis. Protein, fat and lactose concentrations in milk were determined with a Foss MilkoScanFT 6000 Analyzer (Foss Electric A/S, Hillerod, Denmark), and SCC were determined with a Foss Matic FM5000 instrument (Foss Electric, Denmark).

Another 50 mL of milk was frozen at −20°C without preservative for FA profile analysis. Total lipids in milk were extracted according to the method reported by Bligh and Dyer [18], then FAs were methylated into fatty acid methyl esters (FAME) according to the method described by Carrapiso et al [19]. The FA profile was determined by gas chromatography (GC, Shimadzu GC-2010, Kyoto, Japan) equipped with flame ionization detector and split injection, 100 m length, 0.25 mm i.d., 0.2 μm thickness capillary GC column (SP-2560; Supelco Inc., Bellefonte, PA, USA).The initial oven temperature was 100°C, which was held for 13 min, then increased to 180°C at a rate of 10°C and held for 6 min, then increased to 200°C at 1°C/min and held for 20 min, and then increased to 230 at 4°C/min and held for 15.5 min. High purity nitrogen was used as carrier gas at a flow rate of 1 mL/min. The injector was set at 270°C, and the detector was set at 280°C. FAMEs were identified by compared with the retention times of the standards. The standards were FAME MixC4-C24 Unsatures (Sigma 18919, Germany).

Six blood samples (10 mL) were collected into lithium heparin Vacutainer tubes (Becton, Dickinsonand Company, Franklin Lakes, NJ, USA) using jugular venipuncture before feeding on d 4 during each week. Samples were centrifuged at 3,500 rpm for 20 minutes; the serum was aspirated and kept at −20°C for analysis of metabolites, immune and antioxidation capacity. Metabolite, immune parameters and anti-oxidation biomarkers were determined by a biochemical auto-analyzer (Hitachi automatic biochemical analyzer 7600, Tokyo, Japan) using commercially available kits: glucose (GLU), total protein (TP), albumin (ALB), globulin (GLB), non-esterified fatty acid (NEFA), blood urea nitrogen (BUN), immunoglobulin A (IgA), IgG, IgM, CAT, SOD, GSH-Px, TAC, and malondialdehyde (MDA) according to the manufacturer’s instructions. The activity of tumor necrosis factor α were measured by a micro plate reader (BioTek ELX800, BioTek, Winooski, VT, USA) with commercially available kits.

The data were compared using one-way analysis of variance with Duncan’s multiple comparisons. Significance was set at p<0.05 using SAS version 9.2 (SAS Institute Inc., Gary, NC, USA), and the data were expressed as mean and accompanied by standard error of the mean.

As shown in Table 3, cows fed on diets with B. papyrifera silage had lower dry matter intake (DMI) than those fed on diet without B. papyrifera silage (p<0.05), and cows had lower DMI as the B. papyrifera silage increased in the diet (p<0.05). Cows had similar feed efficiency in control group and treatment groups (p>0.05). There was no difference between groups in milk yield, milk fat, protein and lactose (p>0.05), while cows in Group M and Group H had significantly lower milk SCC than that in control group and Group L (p<0.05).

Parameters of serum are presented in Table 4. No significant differences were found in TP, ALB, GLB, NEFA, BUN, GLU, GSH-Px, MDA (p>0.05) between groups. Compared with the control, 10% and 15% of B. papyriferasilage significantly increased the concentration of IgA and IgG (p<0.05), and 15% of B. papyrifera silage significantly increased the concentration of IgM (p<0.05), and 15% of B. papyrifera silage significantly increased CAT, SOD, and TAC in the serum (p<0.05). But serum CAT, SOD, and TAC activity of cows was not affected by the diet with 10% or 15% B. papyrifera silage.

Compared with the control, 15% of B. papyrifera silage diets significantly decreased C4:0 and C17:0 content in milk fat (p< 0.05), increased C18:1n9c and polyunstaturated fatty acid (PUFA) concentration in the milk (p<0.05) as shown in Table 5.

DMI of cattle in the experiment decreased significantly (p< 0.05) as the content of B. papyrifera silage increased in the diet. The palatability of B. papyrifera silage might not be as good as the hay silage or flavonoids exceeded a certain level which affected the DMI of the cattle in the Group L, Group M, and Group H. However, with a lower DMI of cattle fed with B. papyrifera silage, the feed efficiency and milk yield were not affected (p>0.05). The SCC in milk is an indicator reflecting mammary health and milk quality. The physiological level of SCC in bulk cow milk is between 50×10³ and 100×10³ cells per mL, whereas 200×10³ cells per mL is taken as the threshold to define a mammary quarter between healthy and infected [20,21]. In the present study, Group M and Group H had lower milk SCC than that of the control Group and Group L, which might due to the anti-inflammatory effect of the active ingredients in B. papyrifera [22–24]. It is reported that high content of milk protein was found in high SCC milk [25] and fat content was significantly increased from 3.55% to 3.72% by an increment of SCC from 100 to 200 cells/mL [26], but a decrease of milk protein content when SCC in milk was high [5]. Lactose content in high-SCC milk was significantly lower [27] which could be a result of reduced synthesis because of the inflammation processes. In our study, no significant differences in milkfat, protein and lactose was found in difference levels of milk SCC which is consistent with the result reported by Mazal et al [28]. There might be no connection between the SCC and protein content in milk when the udders of cows are not infected.

Blood metabolites represent an integrated index of nutrient supply in relation to the utilization of nutrients. In the present study, no significant difference was found in the serum TP, ALB, GLB, NEFA, BUN, and GLU between groups (p>0.05). The concentration of TP, ALB, and ALB were in the normal range, which indicates that the animals were in healthy status with no adverse effect on their performance [29]. The concentration of plasma GLU, NEFA, and BUN are directly related to timing of feed intake. NEFA is an indicator of energy balance and increased NEFA indicates lipid mobilization to meet energy demands, and intake time of protein, especially rumen undegraded protein. No significant differences were found in the concentrations of NEFA, BUN, and GLU in cows between groups, which probably demonstrates no potential metabolic changes from being fed on the amount of B. papyrifera silage. The immune system is a complex system regulating group of cells with integrated function essential to health. It is reported that plant flavonoids modulate immune and inflammatory cell functions [30].

In the present study, cows in Group M and Group H had significantly higher IgA and IgG than those in control group and cows in Group H had higher IgM than those in control Group, which might be due to the flavonoids in the B. papyrifera silage. Oxidative stress, an imbalance between free radical generation and antioxidant system, is a main contributing factor inducing various diseases in animals [31]. Antioxidant nutrients could be used to reduce the effects of substances that induce reactive oxygen metabolites and to control oxidative stress [32]. Some specific photochemicals isolated from the fruit of B. papyrifera have potent antioxidant effects [1]. Flavonoids are a major class of plant secondary metabolites with multiple functions, such as pigmentation, antimicrobial activity and antioxidant activity [12]. It is reported that they are naturally occurring in bark [33], fruit [15] and leaves [34] of B. papyrifera. SOD is the first line of antioxidant defense that catalysis the conversion of superoxide radical to hydrogen peroxide [35], while CAT is another antioxidant enzyme that can catabolize peroxide to water [36]. 8-Hydroxy-2′-deoxyguanosine (8-OHdG) is believed to be one of the most abundant DNA lesions resulting from oxidative stress and is a biomarker of the oxidative DNA damage and repair. The diet with 15% of B. papyrifera silage significantly increased concentration of CAT, SOD, and TAC and decreased the serum concentration of 8-OHdG. Phenolic compounds and flavonoids appear to be the main components responsible for the antioxidant activity of B. papyrifera silage, and the antioxidant capacity was improved as the supplement level of B. papyrifera silage increased. The efficiency of antioxidative systems against reactive oxygen species is monitored by the estimation of TAC. It is reported that mean TAC levels in shrubs and trees were higher than those in grass, concentrate and timothy hay, and intake of high TAC plants might be beneficial for the TAC status of cattle [10,11]. The MDA is a final product of lipid peroxidation. Cows in Group H had significantly higher TAC than that in control Group, but there was no significant difference between the groups, which is in accordance with results that an increase in TAC is not parallel with an increase in MDA in cows [37]. Supplementation of B. papyrifera silage did not affect the serum concentrations of MDA in dairy cows, which might due to the length of the experiment or the dosage of B. papyrifera silage not being in an appropriate range. Some non-enzymatic antioxidants, like GSH-Px, α-tocopherol, β-carotene and uric acid represent the primary antioxidant capacity of serum [35] and could be used as a tool to evaluate the general nutritional status of animals [38]. In the present study, nutrition didn’t appear to have significant effect on the activity of GSH-Px in blood which is in accordance with the studies by Sgorlon et al [39], and it might be sensitive at tissue or cellular level.

Because of the biohydrogenation occurring in the rumen, milk and other dairy products are a major contributor to dietary saturated fatty acid, which has led to a negative consumer perception and a public health concern related to excessive intake of saturated fats [40]. There have been numerous researches carried out during the last three decades to improve the FA profiles of the products of ruminants [41–44], and there was not enough evidence of direct relationship between PUFA and health, however, further studies indicated PUFA content of the foods did affect some indicators of health status [45–47]. Compared with the control group, milk PUFA of cows are significantly higher (p<0.05), and saturated fats (C4:0, C17:0) are significantly lower (p<0.05) in group M and group H, which might be caused by the reduced biohydrogenation of unsaturated fatty acids in the rumen. It is reported that the active ingredients of B. papyrifera have antibacterial or antimicrobial effect [33], and this might have affected the rumen microflora, which reduced the hydrogenation of FA in the rumen. Dietary PUFA act as precursors for inflammation-mediating eicosanoids and resolvins, and represent direct ligands of transcription factors associated to immune response [48–50]. With more PUFA, the milk would be more beneficial for human’s health.