In situ starch and crude protein degradation in the rumen and in vitro gas production kinetics of wheat genotypes^†

Sample processing; chemical and physical analyses

Twenty winter wheat grain genotypes were characterized according to Bundessortenamt (2011) including all qualities from elite wheat with CP concentrations exceeding 135 g/kg DM to unspecified qualities, recommended for animal feeding, with CP contents lower than 124 g/kg DM (Table 1). All genotypes were grown in one field at the experimental station of the University of Hohenheim in 2011. A detailed description of cultivation, harvesting and storage of the wheat grains is given in a companion paper by Rodehutscord et al. (2016). This communication also provides a detailed description of all laboratory analyses and for every genotype the values of physical characteristics (thousand seed weight, TW and falling numbers), proximate constituents (DM, ash, CP, ether extract and crude fibre) and fibre fractions (amylase pre-treated neutral detergent fibre, aNDFom and acid detergent fibre, ADFom). In addition, the concentrations of amino acids (AA), specific minerals (Ca, P, Mg, K, Na, Fe, Mn, Zn and Cu) as well as gross energy, phytase activity and inositol phosphates are given (Rodehutscord et al., 2016). In the present study, CP content and fibre fractions are given in Table 1. Characterization according to Bundessortenamt (2011), kernel density (KD) and enzymatic ST content are not included in Rodehutscord et al. (2016), but are relevant in the present study, and thus also given in Table 1. Kernel density was measured as described by Correa et al. (2002) using the pycnometer method, and ST concentrations in samples and bag residues were analysed enzymatically as described by N. Seifried, H. Steingaß, W. Schipprack and M. Rodehutscord (submitted).

Animals and diet

Three lactating Jersey cows fitted with rumen cannulae were used for in situ incubation studies, and two of these served as donor animals for the in vitro experiment. The study animals had an average body weight of 485 kg during the experimental period and a mean milk production of 28 l/day. The cows were housed in groups with ad libitum access to feed and water. The cows received a total mixed ration composed of 430 g/kg mixed concentrate, 200 g/kg maize silage, 180 g/kg grass hay, 130 g/kg grass silage, 20 g/kg barley straw, 20 g/kg rapeseed meal and 20 g/kg mineral mixture. The concentrate was composed of 250 g/kg maize, 250 g/kg rapeseed cake, 200 g/kg barley, 200 g/kg field beans and 100 g/kg peas. Mean DM intake was 15 kg/day.

In situ incubation study

Ruminal in situ degradation of DM, CP and ST was determined according to Madsen and Hvelplund (1994) with minor modifications. Every genotype was ground through a 2-mm sieve in a cutting mill (type SM1; Retsch GmbH, Haan, Germany). Bags for in situ incubations (Ankom Co, Macedon, NY, USA; pore size: 50 ± 3 μm; internal dimensions: 10 × 20 cm) were pre-weighed, and approximately 7 g of DM of each genotype was weighed into separate bags. Bags were closed tightly and fixed to an incubation cylinder with a weight of 900 g. To obtain enough sample material for chemical analysis, three replicates per genotype and time point were incubated in each cow for short incubation periods, and five replicates per genotype and time were used for incubation periods equal to or exceeding 8 h. Prior to incubation, bags were soaked for one minute in warm tap water (≈39 °C), and the cylinder with bags was introduced into the rumen and fixed to the cannula with a 50-cm long nylon rope. After 1, 2, 4, 8, 16, 24 and 48 h, bags were removed from the rumen and immediately dispersed in ice-cold water and rinsed with cold tap water to remove rumen content. Three bags per genotype were used for the determination of the initial 0 h time point and rinsed together with the incubated bags in cold tap water for 15 min without centrifugation using a commercial washing machine (type WM14A160; Siemens GmbH, Munich, Germany). Subsequently, bags were dried at 60 °C in an air-forced oven for 24 h and then reweighed to determine DM disappearance. For CP and ST determination, in situ residues of each genotype were merged from each animal and time point. The residues of all animals and genotypes of 24- and 48-h incubation periods were merged within time point to obtain sufficient sample material for chemical analysis.

In vitro gas production study

The method of Menke and Steingass (1988) was used for the determination of in vitro GP, with the slight modification of GP recorded after 2, 4, 6, 8, 12, 24, 48 and 72 h of incubation. Approximately 200 mg of DM of the ground wheat grains (ultracentrifugal mill; type ZM1; Retsch GmbH; sieve size 1 mm) was weighed in 100-ml graduated syringes, closed airtight with vaseline-greased plungers and warmed to 39 °C in an air-forced oven. The buffer solution was prepared and maintained in a water bath at 39 °C under continuous flushing with CO₂. Rumen fluid was collected from two cows prior to the morning feeding into pre-warmed thermos flasks. The rumen fluid from both cows was mixed, filtered through two layers of cheesecloth and added to the reduced buffer solution (1:2 v/v) under constant stirring. Thirty millilitres of buffered rumen fluid was dispensed into each syringe, which was then immediately placed into a rotating disc and oven incubated at a constant temperature of 39 °C. Three syringes with only buffered rumen fluid, termed as blanks, and three syringes with concentrate standard of known GP were included in each run, and GP was measured according to incubation times of the samples. Two repetitions in three runs were carried out for each time and genotype. After removal of outliers, a minimum of four replicates per time and genotype were used for calculation of GP kinetic parameters.

Calculations and statistical analyses

The in situ degradation kinetics of DM, CP and ST were fitted for each cow and wheat genotype, and then ED of the respective nutrients was calculated according to the equations proposed by Ørskov and McDonald (1979):

where Deg (%) is the degradation after t hours, a (%) is the rapidly disappearing fraction, b (%) is the potential degradable fraction with the rate constant of degradation c (%/h), and t is the time (h).

The ED (%) was calculated at feed passage rates from the rumen (k) of 5 (ED5) and 8%/h (ED8) as:

The GP at each incubation time was corrected for blanks and concentrate standard and fitted to an exponential equation for each genotype and syringe:

where GP (ml/200 mg DM) is the GP after t hours, b_gas is the potential GP (ml/200 mg DM), c_gas (%/h) is the rate constant of GP, and t is the time (h).

Model parameters for in situ and in vitro data were estimated by an iterative least-squares procedure using the software graphpad prism (version 5.00 for Windows, GraphPad Software, San Diego, CA, USA).

Correlation and linear regression analyses were only performed for ED8, because differences between ED5 and ED8 were negligible, and as all necessary information is provided in this study and that of Rodehutscord et al. (2016), correlations and regressions can be recalculated easily. Correlations of EDST8 and EDCP8 were tested with ED8 of DM (EDDM8), physical characteristics (thousand seed weight, TW, falling number, KD), chemical characteristics (ash, ether extract, CP, enzymatic ST content, aNDFom, ADFom), AA contents and GP time points and parameters using PROC CORR in sas/stat^® software (version 9.3 for Windows, SAS Institute, Cary, NC, USA). Stepwise multiple linear regressions to predict EDST8 and EDCP8 were performed using PROC REG from sas/stat^® version 9.3. For model 1, only physical characteristics, chemical characteristics, and AA were used for the selection process. For model 2, GP measurements and DM degradation at time point zero (DM_0 h) were added, because these factors do not rely on ruminal incubations. To create model 3, DM_0 h was replaced with EDDM8 and only information about proximate constituents, fibre fractions and GP was used. For creating model 4, EDDM8 in combination with proximate constituents and AA content of the samples was tested. The significance level for variable entry and exit in the regression models was set to α = 0.10. Best-fit regression was chosen based on the highest adjusted coefficient of determination (adjR²) and the lowest root mean square error (RMSE).

In situ degradation

Degradation parameters (a, b, c) of DM, CP and ST varied considerably among wheat grain genotypes (Table 2). The mean values for the fraction a were 31% for DM, 17% for CP and 36% for ST with a range from 21 to 40% for DM, 11 to 22% for CP and 25 to 49% for ST. The fraction b averaged 62% for DM and ST and 81% for CP, with variance between 53 to 71% for DM, 51 to 74% for ST and 75 to 89% for CP. The mean degradation rate (c) was lower (p < 0.001) for CP (21%/h) than for DM (40%/h) and ST (65%/h), with variance from 18 to 27 (CP), 29 to 54 (DM), and 38 to 99%/h (ST). The ED5 ranged between 82 and 88% for DM, 80 and 85% for CP and 91 to 96% for ST, with average values of 85, 82 and 94% for DM, CP and ST respectively. For a passage rate of 8%/h, variation among genotypes was slightly higher, with average values of 82, 76 and 91% for DM, CP and ST respectively.

In vitro gas production

The GP values at each recorded incubation time varied only slightly (Table 3). After 2- and 4-h incubation, minimum and maximum values differed between 2.7 and 3.9 ml/200 mg DM. The greatest difference between genotypes was found after 6-h incubation with 8.3 ml/200 mg DM. After longer incubation times, GP among genotypes showed a maximum difference of 4 to 5 ml/200 mg DM. Consequently, GP parameters varied only slightly with an average b-value of 82 ml/200 mg DM and variation from 78 to 84 ml/200 mg DM. The GP rates showed a mean value of 11.4%/h, a minimum value of 10.5%/h and a maximum rate of 12.3%/h.

Correlations

Significant correlations of EDST8 and EDCP8 between one another and among EDDM8, physical and chemical properties, GP and AA are shown in Table 4. Correlations of EDDM8 with EDST8 and EDCP8 were positive, and a positive correlation was also demonstrated between EDCP8 and EDST8. The latter two were inversely related to the TW of the grains and EDST8 was also inversely related with the KD. The thousand seed weight and falling number showed no relationship to any of these variables (values not shown). In most cases, correlations with chemical characteristics were not significant except for a low negative correlation of EDCP8 with the CP concentrations of the samples. Positive correlations of EDCP8 were found with aspartic acid, alanine, lysine and arginine, and negative correlations were determined for serine, glutamic acid, proline and phenylalanine. With EDST8 only proline showed a low negative relationship. Total GP after 2 and 6 h was negatively correlated with EDST8. The GP after 12 h showed a positive relationship with EDCP5, and the correlations with the GP parameters were not significant for either nutrient.

Regression analyses

Regression equations to predict EDST8 and EDCP8 are shown in Table 5. The predictions of the ED from physical and chemical measurements and AA content (model 1.1 and 2.1) were significant and showed intermediate and low values of 0.70 and 0.60 for the adjR², but low RMSE of 1.1% and 1.3% for EDST8 and EDCP8 respectively. The model for EDCP8 was very simple, with only ether extract and glutamic acid as independent variables, whereas for the prediction of EDST8, more variables (KD, ash, ether extract, CP, cysteine) were incorporated. In models 1.2 and 2.2, the DM_0 h was used in combination with different laboratory variables and the GP after 4 h (model 2.2) to improve the prediction of the ED of both nutrients. A reasonable estimation of EDCP8 (model 2.4) was found when the ST and glutamic acid contents were used in combination with the EDDM8 (adjR² = 0.75; RMSE = 1.1%). The best model for the prediction of EDCP8 (model 2.3) combined the variables aNDFom, GP after 12 h and EDDM8 (adjR² = 0.80; RMSE = 0.94%). The combination of the GP after 12 h with ADFom and EDDM8 provided a prediction equation for EDST8 with a high adjR² of 0.97 and a very low RMSE (0.32%); EDST8 prediction was further improved by adding glutamic acid, glycine and lysine to EDDM8 and ADFom (adjR² = 0.99; RMSE = 0.25%).

In situ degradation

Differences between estimated degradation parameters of genotypes were in accordance with the results of Swan et al. (2006), who found substantial differences in the fraction a and fraction c of DM and ST between two wheat lines differing in their endosperm characteristics. The authors thus concluded that selection of wheat grains with a slower degradation can be used to shift ST digestion from the rumen to the small intestine with positive effects on energy efficiency and rumen function due to reduced microbial acid production. In the present study, high variation in ST degradation parameters induced only slight variation in the estimation of ED5 (91–96%) and ED8 (87–94%) due to fast degradation rates of all genotypes. The fast degradation of wheat in the rumen can also explain the small differences observed between the ED of CP and ST when assuming different passage rates. Differences between ED5 and ED8 for ST and CP were minimal at approximately two and seven percentage points respectively. The results of Garnsworthy and Wiseman (2000) also revealed high variance in CP and ST degradation parameters in situ for eight lines of wheat grains, in contrast to our reported variation for ED8 at only 72% to 80% (CP) and 83% to 90% (ST). Garnsworthy and Wiseman (2000) thus concluded that rumen bypass ST of wheat was influenced more by total ST content than by extent of rumen degradation. While the matter has been tested and reviewed extensively in the literature, results are contradictory and it remains uncertain whether the small intestine of dairy cows has a limited ability to digest ST (Waldo, 1973; Owens et al., 1986; Nocek and Tamminga, 1991; Reynolds, 2006). Matthé et al. (2001) and Moharrery et al. (2014) suggested that intestinal ST digestibility is not negatively affected when the daily quantity of rumen bypass ST does not exceed 1.3 to 2 kg/day and that these amounts would include the majority of the practical feeding situations of ruminants. Assuming that the amount of bypass ST is limited to 1.5 kg/day, the genotypes (16 and 20) with the lowest ruminal ST degradation (87%), but only intermediate (63%) and low (61%) total ST contents, would contribute up to 32% of the tolerable bypass ST when wheat is included with 6 kg DM in the daily ration. In contrast, the genotype (1) with the highest total ST content (66%) would contribute only 18% to the total bypass ST because of its high ruminal degradation (94%). It may therefore be erroneous to consider total ST content in combination with a mean value of ruminal degradation for different wheat grains. In contrast, the UDP values of the examined wheat genotypes cover only between 4 (genotype 18) and 7% (genotype 9) of the requirements of utilizable CP at the duodenum of a high-producing dairy cow with a daily milk yield of 40 kg/day when wheat is used at an amount of 6 kg DM/day. In the case of calculating the supply of utilizable CP at the duodenum, it may therefore be sufficient to consider mean values for CP degradation in the rumen.

Comparison of in situ and in vitro measurements

The comparison of the GP and in situ degradation values showed that the variation of in situ degradation parameters was not reflected in the in vitro GP parameters. The SD of the fraction b (1.23%) and fraction c (0.57%/h) of the GP was low compared to the SD of the in situ DM degradation parameters (b = 4.85%; c = 6.31%/h). The GP rate arises from the rate of which all organic matter fractions of the wheat grain are fermented and should therefore be more linked to the ruminal DM degradation than to the degradation of individual nutrients like ST or CP of the grain. In the present paper, the GP rates of the grains were much lower (p < 0.001) than the in situ DM degradation rates and the lack of agreement of both methods is indicated by low or missing correlations of in situ parameters of DM with GP measurements (values not shown). Thus, it cannot be expected that there are strong relationships between the degradation of the individual nutrients (EDST8 and EDCP8) and the GP measurements (Table 4). In contrast, N. Seifried, H. Steingaß, W. Schipprack and M. Rodehutscord (submitted) showed that the in situ degradation rates of DM and ST of 20 genotypes of maize grains were in the same range as their GP rate in vitro and even the ranking of the genotypes was similar with both methods. Furthermore, the authors found a close correlation (r = 0.83; p < 0.001) between GP rate and ED5 of ST and these results indicate that for maize grains it is possible to use GP kinetic studies for a fast screening of in situ degradation characteristics in the rumen.

The missing relationship between fermentation rates of both methods in the present study might be due to the problem of ST particles losses through the bag pores during incubation process, when evaluating soft cereal grains. Secondary ST particle losses in situ occur in significant quantities when evaluating soft cereal grains such as wheat and barley, whereas the evaluation of maize grains is not affected by this error (Huhtanen and Sveinbjörnsson, 2006; Seifried et al., 2015). Secondary ST particle losses cannot occur in the GP technique, which is a major difference between the in situ and GP methods (Cone et al., 2002). Although the ST degradation rates were potentially overestimated in this study, the mean value (91%) and variance (87–94%) of EDST8 were in accordance with in vivo values (n = 13; mean = 90%; minimum = 83%; maximum = 95%) of ruminal ST degradation from wheat grains in the literature (Axe et al., 1987; Zinn, 1994; Lebzien et al., 1996; Philippeau et al., 1999; Matthé et al., 2003; Ueda et al., 2003; Hindle et al., 2005; Larsen et al., 2009; Šimko et al., 2011). Therefore, we conclude that overestimation of the ST degradation rates in situ only marginally influences the calculation of the EDST from wheat grains.

Correlations

Interesting correlations were detected among EDCP8 and EDST8, and chemical and physical properties of the 20 genotypes of wheat tested in the present study. Correlation between EDCP8 with several amino acids indicates that protein composition of the grain influences in situ EDCP, at least in part. The albumin and globulin protein fractions, defined as water-/salt-soluble proteins by classical Osborne fractionation, are likely to display high solubility in the rumen. These proteins contain more aspartic acid, lysine and arginine than gluten proteins (Khan and Shewry, 2009), and these AA are positively associated with EDCP8. In contrast, gluten proteins, insoluble in water and NaCl solution as defined by classical Osborne fractionation, are more pronounced in the endosperm of the grain, have higher proportions of glutamic acid and proline (Khan and Shewry, 2009) and both show negative relationships with EDCP8. McAllister et al. (1993) assumed that the protein matrix of maize grains has a substantial influence on ruminal ST degradation, and many studies have confirmed that maize types with a higher proportion of hard endosperm have a significantly lower in situ DM or ST degradation in the rumen than soft endosperm varieties (Philippeau and Michalet-Doreau, 1997; Correa et al., 2002; Lopes et al., 2009). For wheat, McAllister and Sultana (2011) found that increased kernel hardness decreased in situ DM degradation and the correlation between the rate of DM degradation and kernel hardness (greater numbers indicated softer kernels) was r² = 0.53. This is within the range as the correlation we report between TW and KD with EDCP8 and EDST8, which suggests that grains with a tighter structural arrangement show lower ED in the rumen. In contrast, Swan et al. (2006) found that in situ DM and ST disappearance of hard kernels of wheat was at least slightly faster and higher compared to soft wheat kernels. The positive correlation between EDCP8 and EDST8 in the present study also indicates that the degradation of both nutrients is related to the nature of the ST-protein matrix in the endosperm. Garnsworthy and Wiseman (2000) also found a significant correlation between the extent of ST and CP degradation (r = 0.837; p < 0.001) in the rumen and an even stronger relationship between the degradation rates of both nutrients (r = 0.953; p < 0.001). In the present study, no correlation between ST and CP degradation rates was detected (r = 0.276; p = 0.234) and only a negligible relationship was found between the EDST8 and proline content (and no other AA) of the wheat varieties. In contrast, N. Seifried, H. Steingaß, W. Schipprack and M. Rodehutscord (submitted) found high correlations between specific AA associated with the endosperm characteristics of maize varieties and the extent of ST degradation in the rumen.

Regressions

Variability in rumen bypass ST as calculated above for different wheat grain genotypes revealed that the amount of total ST content and ED in the rumen must be taken into account and that prediction equations are needed to characterize the EDST in the rumen. In the case of CP, the contribution of different UDP values of wheat varieties to the total supply of utilizable CP at the duodenum appears low. Nevertheless, estimation of EDCP can help to define the feeding value of wheat varieties for ruminants. All models for the estimation of EDCP8 and EDST8 in the rumen show low RMSE, due to low variance. The models 1.1, 2.1, 1.2 and 2.2 can be applied without the need of ruminal incubation. Therefore, models 1.1 and 2.1 are appropriate for fast screens of samples when information about nutrient content is available. For the application of model 1.2, simply rinsing of samples in bags in a commercial washing machine is needed; however, for model 2.2, in vitro GP after 4 h must be available. All information can be predicted by NIRS (Garnsworthy et al., 2000; Fontaine et al., 2002; Owens et al., 2009). Additionally, investigation into new calibrations seems promising (Herrero et al., 1997) and could accelerate prediction of wheat variety ED from the models presented herein. The linear relationship between EDST8 and EDDM8 can be described best by the function EDST8 = 20.1 + 0.87 × EDDM8 (adjR² = 0.85; RMSE = 0.79), and combination with other parameters improves the estimation of EDST8 (model 1.3 and 1.4) showing very high adjR² and low RMSE. These models are precise enough to predict EDST for the calculation of the proportion of bypass-ST delivered by individual wheat grain samples. The relationship between EDCP8 and EDDM8, represented as EDCP8 = 13.5 + 0.76 × EDDM8 (adjR² = 0.57; RMSE = 1.42), was not as good as the relationship between EDDM8 and EDST8. Estimation of EDCP8 could be further improved by incorporating additional information about GP, proximate constituents and fibre characteristics and AA which correlated well with EDCP8.