Predicting ruminal degradability of lucerne and grass forage protein from in vitro solubility with non-specific bacterial protease or pancreatin

Samples o f lucerne (25) f rom the primary growth and 2 regrowths and samples o f grass (9) f rom the pr imary growth were harvested in successive stages o f matur i ty wi th in one vegetation season and were used to test the applicability o f protein solubilization dur ing incubation w i th a non-specific protease f rom Streptomyces griseus (Sigma type X I V ) or porcine pancreatin to predict in sacco ruminal degradability o f crude protein (CP) in dried forage. The effective degradability ( E D ) o f protein in the forage, calculated at k = 0.06 h ' 1 , ranged from 63 to 88%. The conditions for protease X I V activity given by Knshnamoor thy et al. (1983) and Aufrere and Carthailler (1988), at constant enzyme concentration in the incubation medium and short incubation period, were not suitable for predicting variabil i ty in in sacco protein degradability o f lucerne due to morphological changes or g rowth type ( R 2 = 0.183, P = 0.03, R S D = 5.94). The results were better when a constant rat io o f enzyme to protein in a sample was maintained (4 U o f protease per 100 mg o f protein) and durat ion o f incubation was extended to 24 h ( R 2 = 0.713, P < 0 . 0 0 1 , RSD = 3.52). However, the best fit between enzymatic solubility and effective degradability o f lucerne protein was obtained using pancreatin (ca. 5 U o f trypsin per 0.5 g o f dry forage): R 2 = 0.830, P < 0 . 0 0 1 , RSD = 2.71. Val idat ion o f regression equations w i th samples o f grass forage indicated that solubility w i th pancreatin was superior to the action o f protease from S. griseus i n predicting ruminal degradability o f forage determined in situ i n cows ( R 2 and R S D = 0.96 and 1.64% vs. 0.59 and 5.15%, respectively). The regression equations between E D ( Y , % ) and enzymatic solubili ty o f protein ( X , % ) for the combined sets o f lucerne and grass samples (n = 34) were for pancreatin: Y =1.18 X -10.97, R 2 = 0.882, P < 0 . 0 0 1 , RSD = 2.84; for protease: Y = 1 . 0 0 X + 2.93, R 2 = 0.544, P < 0 . 0 0 1 , R S D = 5.97. K E Y W O R D S : lucerne, grass, protein, rumen degradability, bacterial protease, pancreatin, enzymatic solubility 342 ANTONIEWICZ A . M . , KOSMALA I .


INTRODUCTION
Changes in composition occurring in lucerne plants during maturation (increase of fibre and decrease of crude protein content, Andrieu et al" 1989) also induce changes in the extent of herbage protein solubility (Tamminga, 1982) and degradability (Balde et al., 1993).Thus, forages harvested at a wide range of morphological stages (from early vegetative to late generative) present a good model for testing the suitability of laboratory methods to predict ruminal degradability of protein by using linear regression.
Protein degradation has been estimated from measurements of digesta flow in the small intestine and from protein disappearance during incubation of feed in polyester bags suspended in the rumen (Van Straalen and Tamminga, 1990).Both methods require cannulated animals and are not suitable for routine screening of feedstuffs.Therefore, various solubility tests and enzymatic procedures have been developed (Broderick, 1982;Miller, 1982;Lindberg, 1985).Pichard and Van Soest (1977) proposed using a commercial protease from Streptomyces griseus, which has a broad specificity for cleavage of peptide bonds, to estimate ruminal protein degradation.Poos et al. (1980) reported that in vitro digestion of different feed proteins using a neutral fungal protease was more highly correlated with in vivo ruminal degradation than was in vitro digestion using several other commercial proteases, including that from S. griseus.The method based on the use of the latter protease was further developed by Krishnamoorthy et al. (1983), Poos-Floyd et al. (1985), Aufrere and Cartailler (1988) and Aufrere et al. (1991).However, the results of predicting protein degradation were not always satisfactory.Aufrere et al. (1991) suggested a 1 h incubation time for routine analysis of concentrate ingredients, and this duration of incubation also gave the best correlation between enzymatic and in vivo degradation in the work of Poos-Floyd et al. (1985).However, our preliminary results (Antoniewicz and Kosmala, unpublished) indicated that during 2 h incubation, the protease from S. griseus (1 mg of enzyme per 500 mg of sample) was not able to differentiate grass forage of 18 and 9% crude protein (early vegetative and late bloom stage, respectively) according to their effective ruminal degradability (81 and 69% in sacco, 1A and 73% in vitro, respectively).
The present study conducted with lucerne and grass forage was undertaken to assess the predictive ability of in vitro incubations with the protease from S. griseus as compared to the results obtained by the in sacco procedure or in vitro incubation with pancreatin.

Feed samples
Lucerne herbage (a monoculture of Medicago sativa) of primary growth and regrowths was cut at successive stages of maturity throughout the whole vegetation season to produce a set of 25 samples covering a fairly wide range of composition and quality.Grass herbage (Dactylis glomerata 0.72; Poa pratensis 0.20; Festuca pratensis and others 0.08) of primary growth was cut from early vegetative to dry stem stage.Fresh herbage samples were dried at 35°C and ground to pass a 1 mm sieve using a Wiley mill.
Dry forage was analyzed in sacco for protein degradability (Antoniewicz et al., 1995).The assay was conducted on cannulated cows, essentially as described by Orskov et al. (1980).Calculations of effective degradability (ED) were done using models based on McDonald (1981) assuming a small particle outflow rate k of 0.06 h" 1 , with no correction for microbial contamination.Lucerne forage was used to make calibrations (regression equations) for predicting ED from enzymatic solubility, and grass samples were used to validate the obtained equations.

Assay with bacterial protease
Protease from Streptomyces griseus, type XIV (5.1 units/mg, Sigma Chemical Co., St. Louis, Missouri, USA) was used.One unit represents the quantity of enzyme that will hydrolyse casein to produce colour with the Folin-Ciocalteu reagent equivalent to 1 mmol tyrosine min" 1 at pH 7.5 and 37°C.
Forage samples (0.5 g) were incubated in duplicate with enzyme in 40 ml of borate-phosphate buffer pH 8.0 (NaH 2 P0 4 -2H 2 0 8.6 g and Na 2 B 4 O 7 10H 2 O 13,17 g f 1 ).The amount of enzyme was either constant, i.e. 4 units (procedure I) or equivalent to protein content in the incubated sample, 4 U per 100 mg of protein (procedure II and III).Samples .wereincubated for 2 h (procedure I and II) or 24 h (procedure III) by shaking in a water bath at 39°C.Next, they were filtered through polyester fibre (40 [im square pore size) and the solid residues were thoroughly washed with distilled water.Feed residues on the filters were deep frozen and in this form quantitatively transferred to Kjeldahl flasks for total N determination.

Assay with pancreatin
The enzyme used was porcine pancreatin (Polfa, Warszawa, Poland, 59 U of trypsin g" 1 ), a solution of 2 g l" 1 0.1 M phosphate buffer pH 7.4 (80 ml 0.2 M NaH 2 PO 4 +420 ml 0.2 M Na 2 HP0 4 made up to 1 1 with distilled water).Samples (0.5 g) were incubated with 40 ml of enzyme solution for 24 h at 39°C and further treated the same way as in the procedure with the protease from S. griseus.

Chemical and statistical analysis
Nitrogen was determined by the Kjeldahl method using Kjeltec Auto 1030 (Tecator, Hoganas, Sweden).The results of in vitro assays were compared to the ED of protein using linear regression and the analysis of variance.Nonaccountable residual mean square variance was expressed as residual standard deviation (RSD).

RESULTS
Protein solubility of lucerne forage by the S. griseus protease depended strongly on the conditions of incubation.The relationship against ED values was poorest at a constant enzyme concentration and 2 h incubation time (procedure I, Table 1, Figure 1).It improved when the enzyme concentration was equivalent to protein content in forage: slightly at 2 h incubation (procedure II) and significantly when incubation was prolonged to 24 h (procedure III).As was expected, the closest relationship between ED values and enzymatic solubility was obtained using pancreatin (Table 1, Figure 2).Solubility with pancreatin allowed to account for 83% of variability in ED values due to maturity and growth type of lucerne forage, while the accountable variance at the best procedure (III) for protease XIV was 71%.There was a very close correlation between the protein solubility results obtained using pancreatin (Y, %) and the protease from S. griseus according to procedure III (X, %): Y = 0.83X + 14.27; n = 25, R 2 = 0.92, P<0.001, RSD =1.60.
The results of predicting protein ED values of primary growth of grass using the equations obtained for lucerne (Table 1) and protease (according to procedure III) or pancreatin are shown in Figure 3.The regression statistics are shown in Table 1.The correlation coefficients and RSD were much better for pancreatin.Accountable variance in the prediction of ED values reached 96% but only 59% when the equations for pancreatin and protease, respectively, were applied.procedure III, compare Table 1).

DISCUSSION
The protease from S. griseus shows a higher rate of protein solubilization than is observed during in sacco incubation in the rumen, especially at a substrate-saturating enzyme concentration (Krishnamoorthy et al., 1983).This could explain the higher protein solubility than degradability from more mature forages of lower protein content when procedure I was used.The low correlation observed in this situation indicated that the enzyme to substrate (feed protein) ratio is of crucial importance.When the ratio is high, the reaction follows first-order kinetics and the rate depends strongly on substrate concentration.On the other hand, when substrate concentration is higher and the ratio of enzyme to substrate lower, the reaction follows zero-order kinetics, the rate of proteolysis is almost constant and not affected by substrate concentration.It is not quite clear which reaction order best describes proteolysis in the rumen.However, there is much to support the suggestion that the reaction deviates from first-order due to limiting enzyme activity (Van Soest et al., 1982).The experiments by Broderick (1978) support first-order kinetics, while those of Nugent and Mangan (1981) show that they could be zero-order.Our results indicate that excess enzyme should be avoided, and the results of in vitro solubilization correlate better with in situ measurements at a constant enzyme to substrate ratio.As protein solubilization under these conditions is less rapid, it was reasonable to increase the reaction time to 24 h.
The better correlation with in situ degradability found when lucerne protein was solubilized with pancreatin than with bacterial protease apparently seems unjustified.However, Craig and Broderick (1980) reported disproportionate release of lysine and, especially, arginine during in vitro degradation by mixed rumen microorganisms.Also artificial trypsin substrates inhibited in vitro degradation of casein by rumen organisms (Craig, unpublished, after Broderick, 1982).These data suggest that initial cleavage by trypsin-like proteases of rumen microbes may limit ruminal protein degradation.Production by microorganisms of trypsin-like proteases may be a good explanation of the satisfactory prediction of ED using porcine pancreatin.Solubility with this enzyme accounted for 83% of variance in the protein degradability results.Poorer results obtained with the protease from S. griseus can be a consequence of a very broad specificity of this enzyme that may obscure differences among feed protein in their susceptibility to microbial proteolysis in the rumen.Another reason may be a difference in the pH of the solution (7.4 for pancreatin and 8.0 for protease).It seems important to measure in vitro protein degradability under conditions similar to those in the rumen, because the pH affects solubility of different types of protein in feeds (Krishnamoorthy et al., 1983).However, it is worth noting that the values of standard errors of estimate obtained in our study with S. griseus protease (Table 1) were much lower than those obtained by Assoumani et al. (1992) (8.1-16.5%).
With the use of S. griseus protease (according to procedure III), and also pancreatin, forages can be evaluated on a relative basis and protein solubilization results are in a reasonable agreement with those obtained by the nylon bag technique for ED estimation.The equation obtained for lucerne and applied to grass fits well in the case of pancreatin and relatively worse as far as protease is concerned.
It may be concluded that measuring solubility by protease from S. griseus at a constant enzyme to protein ratio or by porcine pancreatin provides a fairly accurate approach to predicting effective degradability of lucerne protein in the rumen.The equation obtained with pancreatin is more universal and may also be used satisfactorily for grass forage.This indicates that the extent of proteolysis of dry forage by this enzyme is not species-dependent.wegetacji (R 2 = 0,183, P=0,03, RSD = 5,94).Lepsze wyniki uzyskano przy zastosowaniu stałego stosunku enzymu do białka w próbce i wydłużeniu czasu inkubacji do 24 godz (R 2 =0,713, P < 0,001, RSD = 3,52).

Figure 1 .Figure 2 .
Figure 1.The relationship between ruminai effective degradability of protein of dried lucerne forage and protein solubility under action of protease from Streptomyces griseus: I 4 units, 2 h II 4 units per 100 mg of protein, 2 h III 4 units per 100 mg of protein, 24 h 90 T

Figure 3 .
Figure 3. Prediction of the effective degradability of protein of dried grass forage using equations obtained for lucerne forage (Pancr.-pancreatin, Prot.-protease from S. griseus acc. to