All data were analyzed using SAS
All data were analyzed using SAS software (version 9.3, SAS Institute Inc., Cary, NC). Because of very low FFAR1 protein detection (at the detection limit or no signal) in 3 H-BHB cows, quantification was not possible within a linear range and therefore these animals were omitted from the statistical analysis (H-BHB: n = 5; L-BHB n = 5). The MIXED procedure of SAS for repeated measures was used, considering group, time, and their interactions as fixed effects. As covariance structures the spatial power or ante-dependence were used. The Tukey-Kramer test was applied for multiple comparisons of means. In addition, Spearman correlation statistics with Fisher's z transformations for the calculation of the 95% confidence limits were calculated for FFAR1 and FFAR2 versus blood metabolites and the liver fat content (LFC) without grouping by BHB. Statistical significance was accepted at P < 0.05 and trends toward significances were considered at P < 0.10. The analysis of FFAR1 and FFAR2 in liver revealed specific protein bands of ∼31 and 50 kDa, respectively. The use of antigen-specific blocking peptides confirmed the specificity of both K-7174 (Supplemental Figure S1; https://doi.org/10.3168/jds.2016-11021). As shown in Figure 1A, H-BHB animals had lower hepatic FFAR1 abundance (P = 0.011) than L-BHB cows over the entire peripartal period. Changes over time were observed as a trend (P = 0.096). In contrast, abundance of FFAR2 protein tended to be higher in H-BHB cows than in L-BHB cows (Figure 1B; P = 0.055). In addition, the abundance of FFAR2 increased over time (Figure 1B; P = 0.025). No interactions between group and time were found for either protein. The calculation of the correlations revealed a positive correlation of FFAR1 with plasma glucose concentrations antepartum (r = 0.68, P = 0.004) as well as postpartum (r = 0.49, P = 0.009) but negative correlations with BHB postpartum (r = −0.46, P = 0.016) and LFC (r = −0.40, P = 0.040). In contrast, FFAR2 was correlated negatively with glucose postpartum (r = −3.5, P = 0.041) but positively correlated with BHB (r = 0.52, P = 0.001; Supplemental Table S2; https://doi.org/10.3168/jds.2016-11021). For further information on blood metabolites, insulin, LFC as well as energy balance, please check Supplemental Table S3 (https://doi.org/10.3168/jds.2016-11021). Immunohistochemistry using anti-FFAR1 antibodies revealed immunoreactivity for FFAR1 in association with the cell membrane of the hepatocytes (Figure 2). However, the anti-FFAR2 antibody was not suitable for immunohistochemistry under the tested conditions (data not shown). Fatty liver is an important metabolic disorder after partition in dairy cows. This metabolic disease is strongly associated with the formation of BHB as an indicator for elevated LFC as a result of increasing plasma fatty acid concentrations, as reviewed by Bobe et al. (2004). In addition, increasing concentrations of BHB are associated with decreasing glucose synthesis (Grummer, 1993), and individual differences in gluconeogenic capacity could be involved, as discussed by McCarthy et al. (2015a). By using BHB concentrations to group animals in the current study, metabolic pathways were covered, which are associated with the ligand spectrum of FFAR1 and FFAR2 (Brown et al., 2003; Kim et al., 2013; Yonezawa et al., 2013). The grouping of our cows by maximum BHB concentrations was more successful compared with fatty acid concentrations because of less individual fluctuations of BHB concentrations postpartum (data not shown). Very large variations on the day of maximum BHB but also fatty acid concentrations postpartum were found by McCarthy et al. (2015a); however, as observed in that study, the postpartum BHB area under the curve concentrations were less variable than the fatty acid concentrations. By our grouping, only plasma glucose concentrations and plasma cholesterol concentrations (trend) were lower, in contrast to fatty acids and LFC, which were numerically higher in H-BHB cows on d 18 and 30 postpartum.