general enquiries on this form should be made...

General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: [email protected]

SID 5 Research Project Final Report

SID 5 (Rev. 3/06) of 29

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code LS3653

2. Project title

The potential of non-toxic tannins to improve the utilisation of nitrogen compounds in grass silage for ruminants

3. Contractororganisation(s)

Animal Science Research GroupChemistry and Biochemistry LaboratorySchool of Agriculture, Policy and Development, University of ReadingP.O. Box 236, 1 Earley GateREADINGRG6 6AT

54. Total Defra project costs £ 246,104.00(agreed fixed price)

5. Project: start date................ 01 February 2004

end date................. 30 September 2006

SID 5 (Rev. 3/06) of 29

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.

IntroductionPlants produce a great diversity of different tannin compounds, which are finding renewed interest because of their positive effects on animal nutrition and health. Tannins can generate rumen-escape protein which leads to better utilisation of dietary protein by ruminants, they can also increase fertility, prevent bloat, cure diarrhoea and produce anthelmintic effects. Better utilisation of dietary protein not only leads to higher N-retention but also to less environmental N pollution. Moreover, tannins tend to cause a shift in the form of excreted N resulting in less urinary N and more faecal N, which is an environmentally safer form of N. However, there are also reports that tannins produce negative effects ranging from a reduction in nutrient digestibility, to inhibition of rumen fermentation, growth reduction, signs of toxicity and even animal deaths.

Clearly, not all tannins have the same effects. The challenge is to identify the parameters that can distinguish between the beneficial and the harmful tannin types. The traditional classification into condensed and hydrolysable tannins has not proved useful for predicting their effects. Whilst by definition all tannins bind proteins, different tannins vary by several orders of magnitude in their affinities for proteins. Tannins also vary by several orders of magnitude in their solubilities in water and octanol (K ow-values). Kow-values can be used to predict the physiological fate of compounds. This project therefore investigated a) if Kow-values could be used to distinguish between beneficial and harmful tannins and b) to test the hypothesis that Kow-values combined with measurements of tannin-protein binding strengths can be used to predict the effects of tannins on proteolysis during ensiling and in the rumen.

Characterisation and protein binding properties of commercially available tanninsColumn chromatographic analysis established that commercially available tannins had a surprisingly wide range of tannin contents (1 to 78%; average: 23.2%). Furthermore, the composition of these products is also highly variable (e.g. oak products: 1 to 10%, myrabolan products: 15 to 49%, mimosa products: 26 to 39% tannins. Isothermal titration calorimetry (ITC) measurements revealed that two different types of hydrolysable tannins, i.e. gallotannins and ellagitannins, exhibited distinctly different protein binding characteristics. Bovine serum albumin (BSA) and gelatin were used as models for the main forage and seed storage proteins, Rubisco and prolamines, respectively. Gallotannins (tara and sumac tannins and pentagalloyl glucose) had similar equilibrium binding constants for the interaction with BSA and gelatin (in the range of 104 to 105 M-1 for the stronger binding sites). In contrast the ellagitannins, chestnut and myrabolan tannins, exhibited three to four orders of magnitude greater equilibrium binding constants in the interaction with gelatin (~2 × 106 M-1) than with BSA (~ 8 × 102 M-1). Overall, the data showed that relative binding

SID 5 (Rev. 3/06) of 29

constants for the interactions with BSA and gelatin are dependent on the structural flexibility of the tannin molecule. The more constrained ellagitannins, e.g. chestnut tannins, bound more strongly to the flexible gelatin than the globular BSA. The flexible tara tannins bound much more strongly to BSA than the chestnut tannins.

Two of the condensed tannins (CT) studied, mimosa and grapeseed tannins, showed similar binding isotherms to gelatin and these were comparable to those for HT – except that the CT bound much more strongly than the HT. In contrast, the two CT differed considerably in their BSA binding behaviour. More research will be needed in order to interpret the binding of CT to BSA as little is known about the molecular flexibilities of CT.

Contrary to suggestions in the literature, no relationship could be found between CT and HT tannin solubilities (Kow-values) and their protein binding strengths. Instead, it would appear that the overall flexibility of tannin molecules is the most important factor influencing the binding strength in tannin-BSA complexes. Further work will be needed to assess if BSA is a suitable model in these ITC experiments for predicting the susceptibility of tannin-Rubisco complexes to proteolysis.

Effects of tannins on proteolysis during ensiling of grass All tannin products reduced ammonia contents in grass silage relative to the control. Whilst there was considerable variation, reductions averaged 12% at the lowest rate of addition (21 g/kg DM). It was concluded that the extent of proteolytic response depended significantly on the types and quantities of added tannins.

More research will be needed to fully understand the impact of tannin structure and tannin-protein binding strengths on proteolysis in silo. An attempt has been made to explain the effects of the various tannins on the proteolysis of Rubisco based on their BSA binding strengths. Grapeseed tannins bound more strongly to BSA than the gallotannins or ellagitannins and reduced proteolysis most effectively. Tara tannins bound BSA more strongly than chestnut tannins and also reduced proteolysis more. However, this was only observed at rates of additions ranging from 65 to 195 g/kg DM. At the lowest rate of addition (21 g/kg DM), tara tannins surprisingly affected proteolysis less than the chestnut tannins. It is possible that this unexpected tara tannin effect at low concentrations is related to the enhanced proteolysis that has been reported previously for tannic acid added to alfalfa silage – both tara tannins and tannic acid belong to the gallotannin group. Further studies would be needed to investigate this phenomenon. One explanation could be that the flexible gallotannins alter the structure of the globular dietary protein making it more accessible to microbial digestion at low tannin to protein ratios. Indeed, ITC experiments provided some evidence for such an explanation at low gallotannin concentrations.

Effects of tannins on ruminal proteolysisThree tannins that reduced proteolysis effectively in silo and showed contrasting binding behaviour to BSA were subjected to in vitro studies, i.e. chestnut, mimosa and grapeseed tannins. The tannin-treated silages produced on average 48% less ammonia than the control grass silage during in vitro incubation with rumen fluid. However, simulating the use of these tannins at the point of feeding produced highly variable results. The chestnut and grapeseed tannins had no effect on in vitro proteolysis, but the mimosa tannins significantly reduced proteolysis compared to the control. These in vitro findings were in contrast to the in vivo results: no significant differences were found in the in vivo trial between adding tannins at ensiling or at the point of feeding grass silage.

It is possible, however, that the low in vivo intakes of mimosa silages were due to an inhibition of ruminal proteolysis - as indicated by the low in vitro proteolysis of the mimosa treatment that simulated addition at point of feeding. Further research would be needed to investigate if mimosa tannins – in contrast to chestnut tannins - had a negative effect on proteolytic rumen bacteria. Limited evidence exists suggesting that chestnut tannins are not toxic to ruminants and non-ruminants and do not adversely affect intestinal protein digestibility.

Effects of tannins on in vivo digestibilities and N-excretion in sheepThe N-excretion study revealed some interesting results. The chestnut and mimosa tannins appeared to generate rumen-escape protein, as urinary N losses were significantly lower and faecal N losses were significantly higher compared to the control grass silage. Mimosa and chestnut tannins reduced urinary N losses by 13% and 14%, and increased faecal N-losses by 41 and 28%, respectively. This shift in N excretion from urine to faeces is of interest for mitigating N pollution from ruminant farming systems. To our knowledge, this is the first report detailing the potential reductions in urinary N that can be achieved with these tannins. Lower urinary N content leads to lower volatile N losses to the environment. Faecal N is an environmentally safer form of N as it can contribute to the build-up of soil organic matter. However, it is well known that some tannins can produce too much rumen escape protein and high faecal N contents (possibly due to endogenous N losses) so that animal production is negatively affected. Overall, this study

SID 5 (Rev. 3/06) of 29

cast doubt on the usefulness of the mimosa tannins. Results from this Project suggest that chestnut tannins have potential for reducing N pollution from ruminant systems without compromising the utilisation of dietary protein.

Neither chestnut nor mimosa tannins produced a significant increase in N-retention. It is expected that replacing mature wether sheep with young growing ruminants with high protein requirements would lead to higher N retention and would probably produce a different response. It is also likely that greater nutritional and environmental benefits may be derived from adding tannins to alfalfa silages because of their higher protein contents than grass silages.

Summary of the main findings:1. Judging by the absence of negative effects on in vitro ruminal digestion and on DM intakes, chestnut

tannins appeared to be non-toxic to rumen microflora and to sheep.2. Chestnut tannins showed promise for reducing environmental nitrogen pollution from ruminants by

shifting the forms of excreted N from urinary to faecal N. 3. Low levels of chestnut tannins (21 g/kg DM) significantly reduced proteolysis of grass during ensiling

(ca 11%) and during in vitro ruminal digestion (60%).

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

1.0 Background

Sustainable ruminant livestock systems must promote an efficient use of resources for food production and minimise their environmental impact. Intensive dairy cow production, based on grass silage, is a major contributor to nitrogen (N) loss to the environment, estimated as 18.2 t N/t milk for grass silage-based diets (Delaby et al 1995). It is widely accepted that this low efficiency (milk N/dietary N) value reflects the low efficiency of capture in the rumen of the extensively degraded N fraction in grass silage (Givens and Rulquin 2004). However, it is also known that some natural plant products, so-called tannins, can significantly improve the utilisation of proteins by ruminants.

Tannins are a diverse group of natural plant products. Common features of plant tannins are the fact that they are polyphenols and have a capacity for binding proteins. Tannins can improve the utilisation of dietary N by ruminant livestock, by rendering the plant proteins less or more slowly susceptible to degradation in the rumen. This results in lower urinary N losses to the environment and probably more efficient conversion of dietary N compounds into microbial protein (Waghorn et al 1987). For example, tannins in sainfoin have given a 50%+ improvement in net absorption of intestinal amino acids when compared with an isonitrogenous, but tannin-free, lucerne diet (Thomson et al 1971).

However, the study of tannins is complicated because “the word tannin cannot be precisely defined in a chemical sense and for this reason it has been widely misapplied” (Haslam 1977). Tannins comprise many different molecules some have toxic effects whilst others have beneficial effects on ruminant nutrition (Mueller-Harvey 1999, 2006). Investigations into the relationships between tannin structures and their biological activities are expected to lead to a better use of plant proteins. In addition, scientific tools are needed in order to predict their effects on proteolysis.

The beneficial effects of tannins on N utilisation are keenly exploited in New Zealand through promoting the use of the tanniniferous Lotus sp. in grass-based production systems. However, the potential of tannins in legumes

SID 5 (Rev. 3/06) of 29

has not yet been fully exploited in Europe largely because the tannin containing fodder legumes, sainfoin and Lotus sp., have poorer agronomic characteristics than the tannin-free lucerne (Doyle et al 1984).

The potential of an alternative approach has hardly been investigated, i.e. adding tannins to grass at ensiling or directly to grass silage-based diets. Given recent advances in the knowledge of tannin chemistry, methods of tannin analysis and for probing tannin binding affinities for proteins (Osborne & McNeill 2001; Mueller-Harvey 2001, 2003; Frazier et al 2003) it is timely to explore the effects of tannins on proteolysis of grass silage.

This project aimed to identify groups of non-toxic plant tannins with optimum affinity for binding N-compounds in grass silage. The efficacy of these tannins with respect to the control of proteolysis in silage was investigated in vitro and in vivo. The objective of this study was to examine the potential of these tannins, when added to grass at ensiling or to grass silage-based diets, to improve the efficiency of silage N utilisation and to reduce total N excretion by ruminants.

2.0 Scientific objectives

The overall objective of this Project is to examine the potential of non-toxic plant tannins to improve the efficiency of utilisation by ruminants of N compounds in grass silage leading to a reduction in N losses to the environment. This will be met through the following sub-objectives:

Sub-objective 1: Screen a wide range of plant tannins considered to be non-toxic for their affinity to bind N compounds in grass silage.

Sub-objective 2: Test effects of added tannins during ensilage on (a) proteolysis using mini silos and (b) on in vitro susceptibility of resulting silage N to degradation and deamination.

Sub-objective 3: Test/simulate effects of adding tannins at point of feeding on in vitro susceptibility of silage N to degradation and deamination.

Sub-objective 4: Validate the findings of sub-objectives 2 & 3 using N balance studies in ruminant animals.

The identification of suitable plant tannins with the potential to improve the efficiency of silage N utilisation, and hence reduce total N excretion by ruminants, would contribute to Defra’s overall objectives of increasing the efficiency of nutrient utilisation and reducing N loss to the environment.

3.0 Completion of scientific objectives

Sub-objective 1: To screen a wide range of plant tannins considered to be non-toxic for their affinity to bind N compounds in grass silage.

Several commercially available tannins were screened for their protein-binding properties by isothermal titration calorimetry (ITC), for their molecular weights by mass spectrometry and gel permeation chromatography and for their octanol-water solubilities (Kow-values). The Kow-values of the commercial tannins (n=13) were compared to tannins from fodder legumes and fodder trees (n=13), which are known for producing positive or negative effects on protein utilisation in ruminants.

Sub-objective 2: To test effects of added tannins during ensilage on (a) proteolysis using mini silos and (b) on in vitro susceptibility of resulting silage N to degradation and deamination.

a) Eleven commercial tannin products were evaluated for their effects on proteolysis in dose response experiments. Significant effects were found for types and quantities of added tannins, and there was a significant interaction between tannin types and quantities for dry matter, ammonia-N and buffer soluble N. However, addition of a tannin-containing by-product (grape marc) did not appear to reduce proteolysis in the ensiled grass.

b) Grass ensiled with chestnut, mimosa or grapeseed tannins produced much less ammonia than the control silage during in vitro incubation with rumen fluid.

Sub-objective 3: To test/simulate effects of adding tannins at point of feeding on in vitro susceptibility of silage N to degradation and deamination.

SID 5 (Rev. 3/06) of 29

Simulating the use of chestnut, mimosa or grapeseed tannins at the point of feeding grass silage produced highly variable results. This finding was in contrast to results obtained under Sub-objective 2b. Chestnut and grapeseed tannins appeared to have no effect on in vitro degradation, whereas mimosa tannins reduced in vitro degradation.

Sub-objective 4: Validate the findings of Sub-objectives 2 & 3 (above) using N excretion studies in ruminant animals.

The effects of chestnut and mimosa tannins were evaluated in vivo. Tannins significantly reduced the apparent digestibility of DM, OM and NDF. Mimosa tannins had a significantly greater effect than the chestnut tannins in terms of intake and digestibilities, suggesting that mimosa tannins had an inhibitory effect on microbial degradation in the rumen.

The results showed that both sets of tannins generated some rumen escape protein, as urinary N losses were significantly lower whilst faecal N losses were significantly higher compared to control silage. However, there were no significant differences between adding the tannins at the time of ensiling or at the point of feeding, which was in contrast to the in vitro results from Sub-objectives 2b and 3.

Although there was no effect on total N loss, tannins caused a shift in N excretion from urine to faeces. This finding may offer an opportunity for reducing environmental pollution from ruminant farming systems, as faecal N is an environmentally safer form of N than urinary N.

4.0 Scientific approach - Sub-objective 1: To screen a wide range of plant tannins considered to be non-toxic for their affinity to bind nitrogen (N) compounds in grass silage.

4.1 Introduction Colorimetric assays are widely used for tannin analysis but suffer from serious limitations (Lowry et al 1996) and have not proved useful for measuring the nutritional effects of tannins. Moreover, the current lack of understanding of the interactions between tannins and proteins means that the activity of different types of tannins in relation to controlling proteolysis can also not be predicted. Recent research suggested that other approaches could be used to identify non-toxic tannins and to study their interactions with proteins.

Octanol-water partition coefficients (Kow-values) can be used to predict the physiological uptake of compounds (Mueller-Harvey 2006). This technique has been used to test the hypothesis that low and high K ow-values can distinguish between non-toxic and toxic tannins, respectively. Isothermal titration calorimetry (ITC) measures the binding interactions between compounds and was first applied to different tannins and proteins by researchers at Reading University (Frazier et al 2003). These interactions are characterised by their binding strengths and modes of binding, i.e. ratios of tannin/protein in complexes. ITC has been applied to test the hypothesis that the binding strengths in tannin-protein complexes and their composition would affect proteolysis.

4.2 SamplesThis project studied a set of 13 commercially available Tannin Products (TPs) and included a chestnut tannin extract which is already an approved feed additive in Switzerland (Śliwiński et al 2002). The project also included 7 tanniniferous fodder legumes and fodder trees with known nutritional effects [sainfoin (Nova), Lotus corniculatus (Viking), L. pedunculatus (Grassland Maku and Grassland Sunrise), Acacia nilotica, Dichrostachys cinerea, Piliostigma thonningii and several Leucaena sp, ] and an agro-industrial by-product from the wine industry comprising mainly of grape skins and seeds, so-called ‘grape marc’ (Reichensteiner variety).

4.3 Isolation of Tannins4.3.1 MethodTannins were isolated from commercially available TPs as follows. The TP samples (2 g) were dissolved in methanol/water (10 ml; 1:1, v/v) under a stream of nitrogen for 10 minutes. The solution was centrifuged (3200 rpm), filtered through glass wool and then applied to a Sephadex LH-20 column (Amersham Biosciences, Chalfont St. Giles, UK; 10 g Sephadex LH20 was pre-swollen in 10 ml methanol/water; column dimensions: 10 cm length x 1.5 cm diameter). Non-tannin compounds were eluted with 300 ml methanol/water (1:1, v/v) and a tannin fraction was eluted with acetone/water (150 ml; 7:3, v/v). Acetone was evaporated in vacuo (35oC) and then the tannins were freeze-dried (~24 hours) and stored frozen (-20oC).

Isolation of tannins from the tanniniferous plants and grape marc was carried out using a two-stage process. Firstly, the ground leaf/plant material (20 g) was extracted with acetone/water (200 ml; 7:3, v/v) for 40 minutes at ambient temperature. The extract was filtered through Whatman 41 filter paper (125 mm diameter) under vacuum. An equal volume of dichloromethane was added to remove lipids and pigments. Traces of acetone and dichloromethane were evaporated in vacuo (35oC) and the remaining aqueous phase was freeze-dried (~24 hours). The freeze-dried extract was dissolved in methanol/water (10 ml; 1:1, v/v) and applied to the Sephadex LH-20 column. A tannin fraction was eluted as described for the commercial TPs.

SID 5 (Rev. 3/06) of 29

4.3.2 Results and Discussion of Tannin ContentsThe tannin yields that were obtained from each of the samples are given in Table 1. Yields from some of the crude commercial TPs were surprisingly low and varied from 1.34 to 78.0 g/100 g. The fact that some of the commercial tannin products contain very low amounts of tannins has recently been verified in discussion with Dr S. Battaglia (Silvateam, Italy).

The tannin contents of the sainfoin and Lotus sp. samples were also low (mean 1.14 g/100g airdried sample), similar between samples and in accord with the literature (Hedqvist et al 2000), although some reports give higher tannin contents for these fodder legumes (Ulyatt et al 1976). The grape marc sample contained 3.05 g tannins/100g airdried sample.

4.4 Isothermal Titration Calorimetry (ITC)4.4.1 IntroductionAll interactions between molecules involve the generation or absorption of heat. ITC experiments can measure the small changes in heat (µ Joules) that result from molecular interactions between for example tannins and proteins. The collaborative research group at Reading pioneered the use of ITC for studying binding interactions between different tannins and proteins. This approach aims to discover why some, but not all, tannins generate useful rumen escape protein.

4.4.2 MethodThe ITC technique was initially applied to the crude commercial TPs (Frazier et al 2003). Given the low tannin contents in some of these products (Table 1), the present study investigated the binding interactions of well-defined tannin fractions - after isolation from Sephadex LH-20 - with two model proteins.

A total of 13 tannin fractions from the commercial TPs plus one model tannin compound, PGG (Table 1) were studied using Equation 1:

Equation 1: Corrected ITC data (at protein concentration Y1, 2, 3…) = X – [(1) – (2) + (3)]

where (1) is the titration of tannins into buffer, (2) titration of buffer into protein and (3) titration of buffer into buffer. The titrations (1) to (3) are control titrations and the data are used to correct the experimental data (X, titration of tannins into protein at concentration Y1, 2, 3…). Tannin solutions (5 g/l; prepared in 0.05 M citrate buffer (pH 6); degassed in a vacuum chamber with constant stirring) were titrated into two standard protein solutions at up to four different concentrations: bovine serum albumin (BSA; 0.003, 0.015, 0.040 and 0.075 mM) and gelatin (0.001, 0.002, 0.007 and 0.01 mM). The protein solution was placed in the 1.001 cm 3 sample cell of the calorimeter and the buffered tannin solution was loaded into the injection syringe (250 μl). Tannins were titrated into the sample cell as a sequence of 24 injections of 10 μl aliquots. The time delay (to allow equilibration) between successive injections was 3 minutes. The contents of the sample cell were stirred throughout the experiment at 200 rpm to ensure thorough mixing and all experiments were completed at 25oC. ITC measurements were carried out on a Setaram CSC Nano-ITC III instrument (Model 5300). This instrument employs a differential power compensation design, which measures the change in power required in order to maintain the system at a constant temperature. Raw data were obtained as a plot of heat (μJ) against injection number and featured a series of peaks for each injection. These raw data peaks were transformed using the instrument software to obtain a plot of observed enthalpy change per mole of injectant (ΔHobs, kJ mol-1) against tannin:protein molar ratio, the latter using mean molecular weights of the tannin fractions determined by high performance liquid chromatography-gel permeation chromatography (HPLC-GPC; see Section 4.7). Data were then used to calculate tannin-protein binding strengths and stoichiometries of the interaction.

Table 1. Tannin contents, octanol-water partitioning coefficients (Kow-values) and molecular weights (Mp; Dalton).Tannin Source Lab. No. Tannin type1) Tannin

content2)Kow-value Molecular

weight (Mp)1. Commercial Tannin ProductsVelani oak K16-41 HT: ellagitannins 7.39 0.119Grapemax K16-44 CT gallate esters 33.3 0.165 1,206Mimosa K99-1 CT: profisetinidins 34.3 0.182 1,206Mimosa K16-22 CT: profisetinidins 26.4 0.204Oak K16-3 HT: ellagitannins 1.34 0.304Chestnut K99-2 HT: ellagitannins 14.7 0.280Chestnut K16-19 HT: ellagitannins 8.84 0.322Quebracho K16-11 CT: profisetinidins 33.6 0.552 1,476Wood bark or gall nuts K16-43 HT 78.0 1.29 2,140Myrabolan K16-34 HT: ellagitannins 14.5 1.41 1,249Sumac K16-7 HT: gallotannins

(glucose core)9.28 1.96 2,069

SID 5 (Rev. 3/06) of 29

Tara K16-1 HT: gallotannins (quinic acid core)

25.9 2.99 593

Teamax-AR25 K16-46 epigallocatechin gallate (17.5%)

14.2 3.14 822

2. Tanniniferous plantsSainfoin (Onobrychis viciifolia, var. Nova)

K18-2 CT: mostly prodelphinidins

0.75 0.069

Lotus corniculatus (Viking)

K17-11 CT: mostly procyanidins

0.93 0.057

Lotus pendunculatus (Grassland Maku)


1.52 0.062

Lotus pendunculatus (Grassland Sunrise)


1.37 0.062

Leucaena leucocephala (shoot)

K14-1 CT 3) 0.551

Leucaena leucocephala (leaf)

K14-2 CT 3) 0.878

Leucaena pallida (shoot)

K14-3 CT 3) 0.518

Leucaena pallida (leaf) K14-4 CT 3) 0.402Leucaena Kx2 (shoot) K14-5 CT 3) 0.308Leucaena Kx2 (leaf) K14-6 CT 3) 0.688Piliostigma thonningii (fruits)

I37-2 CT 0.061

Dichrostachys cinerea (fruits)

I37-1 CT 0.111

Acacia nilotica (fruits) I37-4 Flavanol gallates 2.00

3. Agricultural by-productGrape marc K99-3 CT gallate esters 3.05 0.150

4. Tannin reference compoundPentagalloyl glucose PGG 4) 129 940

1), See Appendix 1 for tannin structures 2), as g/100 g airdried sample; HT, hydrolysable tannins; CT, condensed

tannins. 3), sample donated by Dr D.M. McNeill, University of Queensland, Australia 4), sample donated by Prof. A.E. Hagerman, University of Miami, Ohio, USA.

For many of the samples it proved necessary to carry out repeated analysis in order to optimise the conditions for the measurement of the tannin-protein binding interactions. In addition, the isolated tannins varied in the extent of their solubility in citrate buffer. In such instances it was not possible to follow the standard procedure of titrating tannins into protein solutions. In order to resolve this problem, the standard ITC approach was modified by titrating protein solutions into tannin solutions.

4.4.3 Results and Discussion of ITC DataThe results for the control titrations showed that it was possible to correct the ITC data for the titration of tannins into buffer since the sum of the titrations of buffer into buffer and buffer into protein tended to be close to zero. The ITC experiments yielded considerable new information on tannin-protein interactions and will be the subject of two scientific publications covering condensed and hydrolysable tannins (CT and HT). This report focuses in detail on the results observed for contrasting HTs, i.e. the tara, sumac, chestnut and myrabolan tannins. A brief summary of the CT results can be found in the comparison of Kow-values and ITC data (see Section 4.5.3).

4.4.3 a) Tannin binding to gelatinITC revealed that the binding between tannins and proteins resulted in a release of energy. Modelling of the observed energy change - also described as an enthalpy change (ΔHobs) - versus tannin:protein molar ratios showed an exothermic interaction with gelatin. The protein binding sites become saturated at molar ratios in excess of 50 tannin molecules per protein. These high molar ratios reveal multiple binding sites of the tannins on gelatin. Because of these high binding stoichiometries, it was necessary to repeat titrations with several different gelatin concentrations in order to obtain complete binding isotherms.

A common feature of each tannin-gelatin binding isotherm was a long “tail” to the sigmoidal curve at high molar ratios (e.g. >200:1 for tara gallotannins; Fig 1) before zero enthalpies were recorded, which indicate complete binding saturation. These “tails” were taken to indicate that a secondary and very weak interaction process was occurring at higher molar ratios. It was therefore necessary to employ a binding model consisting of two sets of multiple binding sites, each with a different binding strength.

SID 5 (Rev. 3/06) of 29

Interestingly, the isotherm for sumac gallotannins, and to some extent that for the chestnut ellagitannins, revealed an increase in exothermicity with subsequent tannin injections to the protein solution during the initial stages of interaction (i.e., at molar ratios <5:1 for sumac and <25:1 for chestnut tannins). This observation may be evidence of some cooperative behaviour during the initial binding events, i.e. tannin binding was enhanced by a few pre-bound tannin molecules.

4.4.3 b) Tannin binding to BSAThe results of the tannin-BSA binding isotherms (see Fig 2) showed that the interaction of the gallotannins (tara and sumac tannins) behaved similarly with respect to their binding to gelatin, albeit with lower binding stoichiometries (n1 for BSA < n1 for gelatin). However, the ellagitannins (chestnut and myrabolan tannins) showed a marked difference, because the data exhibited evidence of protein concentration dependence: at a given molar ratio the ΔHobs values are more exothermic at higher BSA concentration. After fitting the data to binding models, the results revealed that the ellagitannins interacted more strongly with gelatin than with BSA (see Table 2).

It is widely accepted that proteins which have a compact globular tertiary structure (such as BSA) have a poor affinity for tannins, whereas proline-rich proteins (such as gelatin) which have an extended random coil conformation have a high affinity for tannins. However, the equilibrium binding constants determined in this Project for the protein binding of gallotannins and ellagitannins suggest an alternative hypothesis: gallotannins bind with equal strength to gelatin and BSA, whereas ellagitannins bind strongly to gelatin and weakly to BSA. On the other hand, there are also clear differences in binding stoichiometry: i) gelatin generally binds more tannin molecules per mole of protein (n1 + n2 = 56.1 to 110.6) than does BSA (n1 + n2 = 17.9 to 33.6) and ii) strong binding sites (n1) on gelatin outnumber those on BSA by a ratio of at least 2:1 in all cases.

This study also found that the ITC binding isotherms observed for myrabolan ellagitannins were markedly different to those reported in earlier work (Frazier et al 2003). In the earlier study, ‘myrabolan tannins’ consisted of the crude commercially available TP as supplied to the leather industry, which contains also non-tannin impurities. The present study revealed that the commercial product only contained 14.5 g tannins/100g air-dried TP. The commercial myrabolan TP exihibited concentration dependence for interactions with gelatin and BSA whereas the isolated tannins only showed concentration dependence for the BSA interaction. Interestingly, the binding parameters were similar in both studies for the BSA interaction. Binding of the crude TP to gelatin resulted in significantly lower binding enthalpies. This suggests that some of the impurities in the commercial TPs bound more strongly with the tannins than gelatin. This is likely to have important implications for the extrapolation of data from pure systems to the complexities of in vivo systems.

Figure 1. ITC binding isotherms for (a) tara gallotannin, (b) sumac gallotannin, (c) chestnut ellagitannin and (d) myrabolan ellagitannin interactions with gelatin. Symbols denote different protein concentrations of 0.01 mM (circles), 0.007 mM (squares) and 0.002 mM (crosses).

Figure 2. ITC binding isotherms for (a) tara gallotannin, (b) sumac gallotannin, (c) chestnut ellagitannin and (d) myrabolan ellagitannin interactions with BSA. Symbols signify different BSA concentrations of 0.075 mM (circles), 0.015 mM (crosses) and 0.003 mM (triangles).

SID 5 (Rev. 3/06) of 29

4.5 Octanol-Water Binding Coefficients (Kow-Values)4.5.1 IntroductionOctanol-water partition coefficients (Kow-values) are used to predict the pharmacological, toxicological or environmental fate of chemicals (see Equation 2). High Kow-values indicate that an organic molecule is likely to be absorbed into body tissues, i.e. more likely to produce pharmacological effects. If absorbed in large doses such compounds may prove to be toxic. Recently, Kow measurements have been investigated as a novel approach for predicting the likely effect of tannins in animals (Mueller-Harvey 2004). Since Kow measurements are easier and cheaper to perform than ITC experiments one objective of this Project is to compare the data from these two methods and to determine the relationship between the two approaches.

Concentration in octanol[Equation 2]: Kow = ----------------------------------

Concentration in water

4.5.2 MethodAll tannins were analysed in triplicate and the procedure was controlled by the use of catechin as an internal standard. Isolated tannins (10.0 mg) were dissolved in 10.0 ml acetone/water (7:3, v/v). Acetone was removed under a stream of nitrogen at 30oC, n-octanol (3 ml) was added to the water phase (3 ml) and the solution vortex mixed. The two phases were thoroughly mixed with a vigorous nitrogen stream for 45 minutes at 30 oC. Any loss of water was replaced by making up to 3 ml with distilled water. The phases were separated by centrifuging (3,200 rpm) for 5 min. Aliquots (1.5 ml) were removed from each phase and acetone (1 ml) was added to each phase in order to ensure that tannins remained in solution. All solutions were kept at 6oC prior to analysis.

Table 2. Estimated thermodynamic binding parameters for the interaction of hydrolysable tannins with gelatin and BSA.

tara gallotannins sumac gallotannins chestnut ellagitannins

myrabolan ellagitannins PGG

MW 593 2069 780 1249 940gelatin:

χ2 6.7 2.9 6.7 2.4 3.9n1 53.0 35.4 46.0 36.4 30.8

K1 / M-1 8.0 ×103 6.9 × 105 1.5 × 106 2.2 × 106 2.8 × 105

ΔH1 / kJ mol-1 -21.0 -37.6 -22.2 -27.9 -47n2 57.6 20.7 32.3 38.8 60

K2 / M-1 1.5 × 102 4.2 × 102 1.1 × 104 8.3 × 103 7.5 × 102

ΔH2 / kJ mol-1 -20.9 -49.4 -27 -31.2 -43.8BSA:

Χ2 2.0 4.3 9.6 4.2 1.3n1 2.5 9.4 17.7 22.0 16.5

K1 / M-1 1.0 × 104 1.7 × 105 9 × 102 7.0 × 102 2.2 × 105

ΔH1 / kJ mol-1 -33.0 -30.3 -39.8 -58.1 -37.5n2 15.4 24.2 67.1

K2 / M-1 6.8 × 102 2.2 × 103 6 × 102

ΔH2 / kJ mol-1 -24.8 -29.7 -48.6

SID 5 (Rev. 3/06) of 29

MW, molecular weight; χ2 is the chi-square describing the fit of the data to the sigmoidal curve; n1 and n2, molar ratios of interacting species; ΔH1 and ΔH2, enthalpies; K1 and K2, equilibrium binding constants.

The octanol and water phases were subjected to HPLC using a Gilson HPLC system (Anachem, Luton, UK) with a Dionex UVD340S diodearray detector and Dionex Chromeleon vs 6.10 software. Aliquots (20 l) from the original acetone/water extract, and the water and octanol phases were injected onto an HPLC column (Phenomenex Gemini C18, 5 μ, 110Å, 150 mm x 4.6 mm). Water/acetic acid (99:1, v/v; solvent A) and methanol (solvent B) were used for gradient elution at 1.5 ml/min. The linear gradient profile was as follows: 5% B (0-5 min); 5-50% B (5-40 min); 50-100% B (40-45 min); 100 to 5% B (45-50 min). Absorption was recorded between 200 and 595 nm. Total peak areas of the tannins were determined at 280 nm and used for calculating the octanol/water partition coefficients (Kow-value).

4.5.3 Results and Discussion 4.5.3 a) Kow-valuesTable 1 shows that the Kow-values of the isolated tannin fractions varied from 0.119 (Velani oak, K16-41) to 3.14 (Teamax-AR25, K16-46). This suggests that several of these tannins may be non-toxic (i.e. K ow-values <1; Mueller-Harvey 2004) and are likely to bind specifically via strong hydrogen bonds (i.e. generate rumen escape protein). Table 1 also reports the Kow-values of the mimosa and chestnut tannin samples (K99-1 and K99-2 respectively), which were used in the in vivo N-balance study (see Section 7.0). The Kow-values recorded for these samples are in close agreement to a separate set of mimosa and chestnut tannin samples (K16-22 and K16-19 respectively).

The Kow-values for all sainfoin and Lotus tannins were very low (mean 0.063) and similar, suggesting that the Kow-values could not differentiate between the nutritionally valuable sainfoin and L. corniculatus tannins compared to the less valuable L. pedunculatus tannins. A similar conclusion can be drawn from the Leucaena results: L. leucocephala tannins (Kow = 0.55 to 0.88) produce useful rumen escape protein in contrast to the L. pallida tannins (Kow = 0.40 to 0.52), which have a negative effect on N absorption.

4.5.3 b) Comparison of Kow-values and ITC dataExamination of the results in Table 2 showed that no relationship could be found between the Kow-values of the HT and their ITC binding isotherms to proteins. A similar conclusion could be drawn from the condensed tannins (CT): mimosa (K16-22) and grapemax (K16-44) tannins had similar Kow-values (0.2 and 0.17, respectively; Table 1). Whilst these tannins showed similar binding to gelatin [i.e. sharp sigmoidal binding isotherms that were similar to those seen for HT tannins with evidence of multiple binding sites but with stronger binding enthalpies than those shown for HT]. However, the two CT differed considerably in their BSA binding behaviour, with each tannin giving vastly different binding isotherms. Therefore no obvious trend in terms of linking binding behaviour to K ow-values could be determined.

4.6 Determination of Tannin Molecular Weights4.6.1 IntroductionInitially, MALDI-TOF mass spectra were obtained for the tara, sumac, oak, chestnut, mimosa, quebracho and Lotus tannins. Fig 3 (see Appendix 2) reveals that the measurement of tannin molecular weights (MW) was highly dependent on the experimental conditions, i.e. the ease with which molecules were desorbed from the platen surface. Therefore, subsequent experiments concentrated on obtaining MW measurements by HPLC-GPC instead.

4.6.2 Method4.6.2 a) MALDI-TOF mass spectrometryThe tannins were dissolved at 18 mg/ml (or as a saturated solution) in acetone: water (4:1 v/v). Five parts of a 0.4 mmol/ml solution of 3-trans-indole acrylic acid was mixed with 1 part of the sample solution. Where applicable, the samples were doped with sodium (NaCl, 18 mg/ml in 4:1 acetone/water) or potassium (KCl, saturated solution in 4:1 acetone/water) ions to increase ion formation. The sample platen was loaded with 0.4 l of the final sample mixture and the tannin samples were analysed using a SAI LT3 LaserTof mass spectrometer in either linear or reflectron modes.

4.6.2 b) HPLC GPC analysisMolecular weights of tannins were determined by gel permeation chromatography (GPC) using a GPC50 instrument with a differential refractive index detector (Polymer Laboratories, Church Stretton, UK). The tannin samples were dissolved in THF (0.2 g tannins in 100 ml THF) at 5 oC overnight. Samples (100 μl) were injected into the GPC system and separated on two serially connected PLgel 3μm MIXED-E columns (300 x 7.5 mm; Polymer Laboratories) and eluted with THF at 1 ml/min at ambient temperature. Column calibration was performed with polystyrene standards (PSTY EasiVial, Polymer Laboratories).

4.6.3 Results

SID 5 (Rev. 3/06) of 29

The results are presented in Table 1 and were used to calculate the thermodynamic binding parameters between tannins and proteins (Table 2).

4.6.4 Conclusions (Sub-objective 1)Tannin contents of the commercial TPs covered a surprisingly wide range (1 to 78%) and their MW varied from 593 to 2140 Daltons. Information on tannin contents is relevant to the design of feeding trials and for interpreting data from the literature.

ITC was used to study the interactions of different tannins with two model proteins. BSA was used as a model for the globular leaf protein, Rubisco, and gelatin as a model for the flexible seed storage proteins. The results showed that the flexible gallotannins interacted equally well with the globular and flexible proteins. However, it could be shown that the more rigid ellagitannins interacted much more strongly with the flexible gelatin than with the globular BSA.

The Kow-values of the commercial TPs (0.12 to 3.14) covered a similar range as the K ow-values of tannins isolated from fodder plants and fodder trees (0.06 to 2.00). Comparisons between tannin Kow-values from feeds that produce beneficial rumen escape protein with those that do not suggested that Kow-values could not be used to predict which tannins produce rumen escape proteins. This study also revealed that there was no relationship between the Kow-values and ITC data and concluded that Kow-values could not be used to predict the binding strengths in tannin-protein complexes.

5.0 Scientific approach - Sub-objective 2a: To test effects of added tannins during ensiling: on proteolysis using laboratory-scale silos.

5.1 IntroductionTannins are natural products which bind to proteins. Such tannin-bound proteins can improve the utilisation of dietary N by ruminant livestock, by rendering plant proteins less or more slowly susceptible to degradation in the rumen. This results in lower urinary N losses to the environment and probably more efficient conversion of dietary N to microbial protein (Waghorn et al 1987). Thomas et al (1971), for example, reported that the tannins in sainfoin gave a ~50% improvement in the net absorption of intestinal amino acids when compared with an isonitrogenous Lucerne diet. Therefore, the aim of Sub-objective 2 was to study the potential of various added, crude commercial non-toxic TPs products to reduce proteolysis during ensilage. A reduction in silage ammonia-N (NH3-N) content was used as an indicator of reduced proteolysis during ensilage.

A total of three ensilage studies were completed using laboratory-scale silos during the course of the Project. Ensilage study 1 and 2 utilised fresh grass harvested in October 2004 (3rd cut; dry matter (DM), 192 g/kg; crude protein (CP), 230 g/kg DM) and May 2005 (1st cut; (DM, 266 g/kg; CP 124 g/kg DM) respectively to determine the effects of 11 commercial TPs (included in Table 1) on proteolysis during ensilage. In the third study, a by-product from the wine industry, so-called ‘grape marc’, was investigated as a potential source of tannins to reduce proteolysis during ensilage. The results of ensilage study 1 were reported in the first Project Annual Report (LS3653-SID4-Year1) and therefore this report focuses on the results obtained for ensilage studies 2 and 3.

5.2 Ensilage Study 25.2.1 Materials and MethodsGrass silage treatments were prepared from 1st cut (May 2005; DM 245.2 g/kg; 21.8 g N/kg DM) perennial ryegrass. The herbage was mown with a mower conditioner, wilted (~24 h) and then picked up with a precision chop forage harvester and stored frozen (-20oC) prior to use. A randomised complete block design was employed with 46 treatments: negative control (no added tannins), positive control (formic acid-treated; equivalent to 3 l/tonne fresh grass) and tannin-treated silages involving 11 commercial TPs added in increasing amounts (target rates: 20, 60, 120 and 180 g DM/kg grass DM) in three blocks. The tannin powders were sieved onto the grass and then incorporated using 11.5 ml distilled water/230 g fresh grass. The negative control received only 11.5 ml distilled water and the positive control received 10.8 ml distilled water plus 0.7 ml formic acid (85%). All treatments were ensiled in laboratory-scale silos (~230 g grass/silo) and kept at ~20oC for 90 d. When the silos were opened any waste material was removed and the remaining material was thoroughly mixed and sub-sampled for analysis.

5.2.1 a) Chemical analysisSub-samples of the fresh silage were dried at 100oC for 18 h to determine oven DM content (MAFF, 1986). Silage pH was determined on a water extract following incubation of 20 g fresh silage in 50 ml distilled water for 1 hour. Total N (TN) content was determined on fresh samples according to the Kjeldahl method (MAFF, 1986). The NH3-N content of the fresh silages was determined on a water extract following incubation of 10 g fresh material in distilled water overnight at approximately 3oC. The NH3-N content of the water extract was measured using a modified Bertholet reaction an auto-analyser, which is based on the reaction of NH3 with salicylate and

SID 5 (Rev. 3/06) of 29

dichloroisocyanurate (Burkard Scientific 1996). Nitroprusside was added as a catalyst and the product measured at 650 nm. The concentration of NH3-N was calculated by regression with respect to the response of ammonium chloride standards containing 2 to 20 mg N/l. Potassium chloride (0.1 M) was used as the sample matrix in the preparation of the standard solutions and silage extracts for analysis.

Further sub-samples were dried at 100oC for 8 h and then milled (1 mm screen). The dried samples were analysed for total ash content (MAFF 1986) by ashing in a muffle furnace at 550oC. Acid detergent fibre (ADF) content was determined in triplicate using the ANKOM technique and based on the method of Van Soest et al (1991).

5.2.1 b) Nitrogen fractionationsThe combined ADF residue (i.e. across three replicates) was then analysed for TN content using the Dumas technique in order to determine acid detergent soluble nitrogen (ADIN) content, a measure of unavailable N. Buffer soluble N (BSN) content was determined following incubation of the dried samples (1.2 g) in borate-phosphate buffer (pre-heated to 39ºC), pH 6.7-6.8 (modified from Licitra et al 1996; containing 12.20 g/l monosodium phosphate and 8.91 g/l sodium tetraborate). Incubation was carried out in 50 ml centrifuge tubes for 1 h at 39ºC in an incubation chamber, equipped with an orbital shaker with a shaking frequency of 50 rpm. The tubes were also shaken manually following the addition of the borate-phosphate buffer and then at 15 minute intervals throughout the incubation period. Following the lapsed time the tubes were centrifuged at 3000 × g for 10 minutes at 4ºC. The N content of the supernatant (20 ml) was determined by the Kjeldahl technique. Dried silages were used to determine trichloroacetic acid (TCA) soluble N content. Initially, the same procedure was employed as described previously for the determination of BSN content, except following centrifugation the supernatant (30 ml) was transferred to a clean centrifuge tube and 3 ml TCA solution (1000 g/l) added. The tube contents were vortex mixed, the tubes were kept for 1 h in iced water and then centrifuged at 15000 × g (15 min at 4ºC). The supernatant (20 ml) was analysed for N content according to the Kjeldahl technique. Buffer soluble protein-N (BS-PN), comprising of protein and long chain peptides, was calculated as the difference between BSN and TCA soluble N. Where appropriate, compositional values determined for the individual samples were expressed on an oven DM basis.

Three tannins were selected for further study based on their effectiveness to reduce silage NH3-N content. The resulting silages following ensilage of chestnut, mimosa and grapemax-AR25 tannins (see Table 1) at 120 g DM/kg grass DM, together with fresh grass and grass silage (negative control; no added tannins), were subjected to further analysis to fractionate the N-compounds into short chain peptides (SCP) and free amino acids (FAA). The SCP and FAA contents were determined on the TCA soluble N fraction, prepared according to the method described above. Concentrated hydrochloric acid (6 ml; ~12 M) was added to 6 ml TCA soluble N solution in glass bottles, thoroughly mixed with a vigorous nitrogen stream for 10 minutes and then the bottles were sealed using Teflon lined caps. The samples were hydrolysed by heating in a fan-assisted oven for 16 hours. Once at ambient temperature, the hydrolysed solution was transferred to 50 ml centrifuge tubes and then centrifuged at 3000 × g (10 minutes at 4ºC). An aliquot of the hydrolysed supernatant was transferred to glass round-bottom flasks (100 ml capacity) and then evaporated in vacuo (40oC). After re-suspension of the remaining residue in 0.2M lithium citrate buffer (6 ml; pH 2.2; Amersham Biosciences, Chalfont St Giles, UK), the solution was transferred to glass tubes, flushed with N and the tubes sealed and then transferred to a fridge (~3 oC) while awaiting analysis. An aliquot (6 ml) of unhydrolysed TCA soluble N was evaporated and then re-suspended in 6 ml lithium buffer as described above. Amino acid concentration of the hydrolysed and unhydrolysed TCA soluble N solutions was determined following reaction with ninhydrin at 100oC for 15 minutes and measurement of absorbance at 570 nm. Leucine was used as a standard for measuring amino acid concentration and prepared at concentrations of 0, 0.05, 0.10, 0.25, 0.50, 0.70 and 0.10 mg/ml in lithium buffer. The results were expressed as amino-N/kg DM.

5.2.1 c) Statistical analysisThe data from the randomised complete block design with 46 treatments were analysed using analysis of variance (ANOVA) with effects for treatment and block. Contrast analysis for comparing any two treatment means was also determined. Linear and quadratic effects of rate of tannin addition were also determined.

5.2.2 Results and DiscussionThe preliminary findings from this study (ensilage study 2) have been accepted for publication by the British Society of Animal Science (BSAS) at their Annual Conference in 2007 and it is envisaged that the results will be used in the preparation of at least one scientific paper for publication in a peer reviewed journal.

A small difference between the actual DM content of the grass pre-ensilage and the DM value used in the initial calculations (245.2 and 266 g/kg respectively) resulted in the TPs being added at rates of 21.7, 65.1, 130.2 and 195.3 g DM/kg grass DM (equivalent to approximately 21.2, 61.1, 115.2 and 163.4 g/kg DM respectively). The addition of tannins at a rate of 61.1 g/kg DM is similar to the mean value of 74.8 g/kg DM for the tannin-containing treatment diets employed in the N-balance study (see Section 7.0).

The analysis of the data showed that the ANOVA accounted for a large proportion of the variance across all treatments (denoted by high R2 values) and therefore there was little unexplained variation (i.e. highly controlled)

SID 5 (Rev. 3/06) of 29

within the model. As a result, highly significant differences were recorded for small differences between treatment means. Therefore, it is also necessary to moderate the interpretation of the statistical significance of the results by their practical impact.

The results of the ANOVA of the data showed that there was a significant effect of both tannin types and rate of addition at ensilage. However, differences in silage composition were largely due to tannin rate owing to the very high F-value obtained for the rate effects (Tables 4 to 6). This report focuses on presenting the results for NH3-N, BSN, TCA soluble N (or buffer soluble non-protein nitrogen, BS-NPN), buffer soluble-protein-N (BS-PN) and ADIN [the results are given for the N contents of the fractions as determined, however, the calculation of the overall BSN and BS-NPN contents would also include, for example NH3-N). Due to the large number of treatments (n=46) assessed in this study the data are presented graphically in order to aid interpretation of the results. In each case the average effect (mean across all 11 TPs) together with the effects recorded for tara (K16-1) and grapemax (K16-44) tannins will be presented in order to demonstrate the variation between tannins. This type of presentation gives a useful summary, but may not always demonstrate the extremes of variation since the effect of the individual tannins was not consistent across all the different rates of addition or silage characteristics measured.

The mean chemical composition of the grass silage treatments (mean across control and tannin-treated silages) is summarised in Table 3, together with their standard deviation (SD) and range for the each of the parameters studied. For example, the range in DM content (209 to 266 g/kg) reflects the increasing proportion of tannins within the overall treatment DM (range 21.2 to 163.4 g tannin/kg treatment DM). The mean pH value was low (4.15) and was similar across all silage treatments (range 4.07 to 4.03). The BSN fraction (mean 7.83 g/kg DM) was comprised predominantly of soluble NPN (mean 7.11 g/kg DM) and very low levels of BS-PN (mean 0.72 g/kg DM).

Table 3. Mean chemical composition (across all control and tannin treatments) of grass silage treatments (as g/kg DM unless stated otherwise).

Parameter Mean Standard deviation Minimum value Maximum valueDry matter (g/kg) 237 14.9 209 266pH 4.15 0.05 4.07 4.30Acid detergent fibre 310 19.7 266 345Ash 98.6 7.63 77.7 119Total nitrogen (TN) 19.9 1.71 16.6 25.8Ammonia-N 1.59 0.27 1.02 2.16Ammonia-N/TN (g/100g) 7.95 1.10 5.28 11.3Buffer soluble N 7.83 1.37 5.47 11.5Buffer soluble N/TN (g/100 g) 39.2 5.54 28.7 56.6Buffer soluble NPN 7.11 1.20 4.58 9.76Buffer soluble protein-N 0.72 0.72 -2.11 2.70Acid detergent insoluble N 1.01 0.31 0.56 2.93

5.2.2.a) Effect of tannins on ammoniaThe effect of tannins and rate of tannin addition on NH3-N content of the tannin-treated silages, expressed as g/kg DM and g/100 g TN, is given in Figs 4 a & b, respectively. There was a significant (P<0.001) effect of tannins and rate of tannin addition on silage NH3-N content (as g/kg DM and as a proportion of TN). However the response was greatest for the rate of added tannins, resulting in a significantly linear (P<0.001) decrease in NH 3-N with increasing amounts of tannins (average NH3-N decreased from 9.00 to 7.18 g/100 g TN with <2 to 20% tannins). The significant linear rate x tannin interaction showed that the response or slope varied across the various tannins studied. The response of the tara and grapemax tannins in relation to the average of all tannins (n=11) is illustrated in Figs 4a & b. Also, the NH3-N content decreases relative to the negative control in response to the addition of low amounts of all TPs (21.2 g/kg DM).

Figure 4. Effect of tannin and rate of addition on ammonia-N content of tannin-treated silages (a) as g/kg DM and (b) g/100 g total N.

SID 5 (Rev. 3/06) of 29

1.00

1.25

1.50

1.75

2.00

2.25

0 50 100 150 200

Added tannin (g DM/kg grass DM)

Am

mon

ia-N

(g/k

g D

M)

Tara (K16-1)

Grapemax (K16-44)Average tannin

Negative controlPositive control

(a)

6.00

6.50

7.00

7.50

8.00

8.50

9.00

9.50

10.00

0 50 100 150 200


Am

mon

ia-N

(g/1

00 g

tota

l N)

Tara (K16-1)Grapemax (K16-44)

Average tanninNegative controlPositive control

(b)

Table 4. Analysis of variance.Effect on NH3-N (g/kg DM)

df F P-value Effect on NH3-N(g/100 g total N)

df F P-value

Tannin 10 13.2 <0.001 Tannin 10 4.18 <0.001Linear (L) rate 1 642 <0.001 Linear (L) rate 1 149 <0.001L rate*tannin 10 3.70 <0.001 L rate*tannin 10 3.60 <0.001s.e.d. = 0.075 s.e.d. = 0.51

df, degrees of freedom; s.e.d., standard error of difference of any two treatment means.

5.2.2.b) Effect of tannins on buffer soluble NThe results showed that BSN was a major component of the overall N-fraction and that there was a significant (P<0.001) effect of tannins on BSN content (expressed as g/kg DM and g/100 g TN). Fig 5 shows the average effect of all tannins (n=11) compared to tara and grapemax tannins. The value of 48.9 g BSN/100 g TN recorded for the tara tannin (added at 21.2 g/kg DM) is likely to reflect the elevated NH 3-N content given above (see Fig 5b). There was a significant (P<0.001) linear decrease in BSN content with increasing amounts of tannins in the silage treatments. The data in Fig 5b suggest that the response is greatest up to ~61.1 g added tannins/kg DM and diminishes at higher tannin additions. The significant linear rate x tannin interaction showed however that the response or slope varied across the various tannins studied, highlighting differences in the potential of the various tannins to reduce proteolysis during ensilage. Lower BSN contents may result from an inhibition of microbial proteolysis or from the binding of tannins to peptides. This finding may be of practical importance since it may reduce the intake of soluble-N compounds in grass silage which are known to be used inefficiently by ruminants. Further work would be required to determine the rumen and post-rumen availability of the tannin-bound peptides and proteins.

Figure 5. Effect of tannin and rate of addition on buffer soluble-N (BSN) content of tannin-treated silages (a) as g/kg DM and (b) as g/100 g total N.

5.50

6.50

7.50

8.50

9.50

10.50

0 50 100 150 200

Adde d tannin (g DM/k g grass DM )

Buf

fer s

olub

le N

(g/k

g D

M)

Tara (K16-1)


Negative control

Positive control

(a)

30.0

35.0

40.0

45.0

50.0

0 50 100 150 200


Buf

fer

solu

ble

N (g

/100

g to

tal N

)

Tara (K16-1)


Negative control

Positive control

(b)

Table 5. Analysis of variance.Effect on BSN(g/ kg DM)

df F P-value Effect on BSN(g/100 g total N)

df F P-value

Tannin 10 28.3 <0.001 Tannin 10 7.84 <0.001Linear (L) rate 1 141 <0.001 Linear (L) rate 1 19.1 <0.001L rate*tannin 10 6.11 <0.001 L rate*tannin 10 3.38 <0.001s.e.d. = 0.50 s.e.d. = 3.21

df, degrees of freedom; s.e.d., standard error of difference of any two treatment means.

SID 5 (Rev. 3/06) of 29

5.2.2.c) Effect of tannins on buffer soluble NPN and PNThe results for the BS-NPN and BS-PN content of the tannin-treated silages is illustrated in Figs 6a & b and showed that the BSN fraction of the silages was comprised mainly of BS-NPN, with low levels of BS-PN. The response in silage BS-NPN content to the tannins and rate of addition of tannins was similar to that observed for BSN. There was a significant (P<0.001) effect of tannins on silage BS-NPN content as illustrated by the response to the tara and grapemax tannins in relation to the average of all tannin treatments (n=11; Fig 6a). Also, there was a significantly linear (P<0.001) decrease in BS-NPN with increasing amounts of tannins; the average decrease in tannin treatments ranged from 8.19 to 6.36 g/kg DM compared to the negative control of 9.17 g/kg DM. However, the significant (P<0.001) linear rate x tannin interaction showed that the response or slope varied across the various tannins studied. Using the overall linear response for the various tannins it may be possible to select those tannins that have the greatest effect for further study.

The results presented in Fig 6b showed that the BS-PN content varied over a relatively narrow range and that there was no significant effect of rate of addition of tannins on BS-PN. However there was a significant (P<0.05) difference in the response to the different tannins. The variation observed for the different tannins is likely to reflect the error involved in calculating this fraction from the difference between BSN and BS-NPN contents.

Figure 6. Effect of tannin and rate of addition on (a) buffer soluble-NPN (BS-NPN) and (b) buffer soluble protein-N (BS-PN) content of tannin-treated silages (as g/kg DM).

5.00

6.00

7.00

8.00

9.00

10.00

0 50 100 150 200


Buf

fer

solu

ble-

NP

N (g

/kg

DM

)

Tara (K16-1)

Grapemax (K16-44)

Average tannin

Negative control

Positive control

(a)

-0.5

0.0

0.5

1.0

1.5

2.0

0 50 100 150 200


Buf

fer

solu

ble

prot

ein-

N (g

/kg

DM

)

Tara (K16-1)Grapemax (K16-44)Average tanninNegative controlPositive control

(b)

Table 6. Analysis of variance.Effect on BS-NPN(g/kg DM)

df F P-value Effect on BS-PN(g/kg DM)

df F P-value

Tannin 10 23.2 <0.001 Tannin 10 2.00 <0.05Linear (L) rate 1 233 <0.001 Linear (L) rate 1 0.59 NSL rate*tannin 10 3.70 <0.001 L rate*tannin 10 1.46 NSs.e.d. = 0.41 s.e.d. = 0.58

df, degrees of freedom; s.e.d., standard error of difference of any two treatment means; NS, non-significant (P>0.05).

Fractionation of the BS-NPN component showed that it comprised almost entirely of short-chain peptides and free amino acids and that the values decreased in line with the BS-NPN contents in response to the added tannins. However, the proportion of free amino acids and peptides was similar in the negative control and tannin-treated silages (43.8 and 56.2%, respectively).

5.2.2.d) Effect of tannins on ADINThe results showed that the mean level of ADIN in the control treatments was quite low (0.98 g/kg DM). The effect of tannins and the rate of addition on ADIN contents of the silages is illustrated in Fig 7. There was no significant (P>0.05) effect of rate of added tannin on ADIN content. However there was a significant effect of tannin (P<0.001) and also a significant linear rate x tannin interaction on silage ADIN content, resulting in different slopes for the response of the different tannins studied. Most tannins had no effect on ADIN content as shown in

Fig 7 by the averaged result of all tannins (n=11). However, the ADIN content of the grapemax-treated silages was variable and it would appear that the grapemax tannins bound most strongly to grass silage proteins as these tannins resulted in higher ADIN values and lower NH3-N (Fig 4).

Figure 7. Effect of tannin and rate of addition on acid detergent insoluble-N (ADIN) content of tannin-treated silages (as g/kg DM).

SID 5 (Rev. 3/06) of 29

0.00

0.50

1.00

1.50

2.00

2.50

0 50 100 150 200


Aci

d de

terg

ent i

nsol

uble

-N (g

/kg

DM

)

Tara (K16-1)

Grapemax (K16-44)

Average tannin

Negative control

Positive control

Table 7. Analysis of variance.Effect on ADIN (g/kg DM) df F P-valueTannin 10 10.2 <0.001Linear (L) rate 1 1.03 NSL rate*tannin 10 2.27 <0.05s.e.d. = 0.18df, degrees of freedom; s.e.d., standard error of difference of any two treatment means; NS, non-significant.

5.3 Ensilage Study 35.3.1 Materials and MethodsGrass silage treatments were prepared from 1st cut (May 2005) perennial ryegrass (DM, 218.8 g/kg; N, 21.8 g/kg DM). The herbage was mown with a mower conditioner, wilted (~24 h) and then picked up with a precision chop forage harvester and stored frozen (-20oC) prior to use. Fresh grape marc, the by-product remaining following the pressing of grapes (Reichensteiner variety) for wine manufacture, comprising mainly of grape skins and seeds, was stored frozen (-20oC) immediately following collection. A sample of the grape marc was freeze-dried and milled (1 mm screen; residual DM, 909.5 g/kg; N, 10.9 g/kg DM) prior to ensilage. A randomised complete block design was employed with seven silage treatments; control (no grape marc) and increasing amounts of grape marc (target rates: 60, 120, 180, 240, 300 and 360 g DM/kg grass DM) in three blocks. The grape marc was applied to the grass immediately prior to ensilage in laboratory-scale silos (~230 g grass/silo) and kept in a temperature controlled environment (~20oC) for ~90 d. Sub-samples of the fresh silages were analysed for DM (100oC for 18 h), ash, TN (Kjeldahl method) and NH3-N contents and pH according to the methods outlined in Section 5.2). The data were analysed using ANOVA with effects for treatment and block. Analysis of contrasts (e.g. control with mean of grape marc-treatments and comparison of treatment means) and the linear treatment effect were also completed.

5.3.2 Results and DiscussionA difference between the actual DM content of the grass pre-ensilage and the estimated DM value used in the initial calculations (218.8 and 266 g/kg respectively) resulted in the grape marc being added at rates of ~73, 146, 219, 292, 365 and 438 g DM/kg grass DM (equivalent to ~68, 127, 180, 226, 267 and 304 g/kg DM). The composition of the grape marc-treated silages is presented in Table 8. Addition of freeze-dried grape marc gave a significant linear (P<0.001) increase in DM (P<0.001) and decrease in N (P<0.01) content of the silage treatments. The NH3-N content (range, 3.49 to 2.35 g/kg DM) and the proportion of TN as NH3-N (range 14.0 to 10.8 g/100 g) were significantly (P<0.001) lower in the grape marc-treated silages.

SID 5 (Rev. 3/06) of 29

Table 8. Chemical composition of grape marc-treated silages (as g/kg DM unless stated otherwise).Level of added grape marc (g DM/kg grass DM) Contrast1 Contrast2 Linear

Parameter 0 73 146 219 292 365 438 s.e.d. s.e.d. effectDM (g/kg) 196.0 214.8 216.2 222.5 233.8 239.2 246.6 3.84 2.94 ***pH 3.98 4.03 3.84 3.88 3.93 3.85 3.84 0.053 0.040 **TN 24.9 23.5 23.0 23.3 22.4 22.7 21.7 0.523 0.400 ***NH3-N 3.49 3.08 2.92 2.87 2.59 2.61 2.35 0.097 0.074 ***NH3-N/TN3 14.0 13.1 12.7 12.3 11.6 11.5 10.8 0.431 0.329 ***NH3-N/N4 0.142 0.139 0.135 0.139 0.133 0.139 0.131 0.005 0.003 NSAsh 115.4 114.3 109.7 109.0 112.3 106.9 105.5 2.21 1.69 ***Contrast1, comparison of treatment means; contrast2, comparison of control with mean of grape marc treatments; s.e.d., standard error of difference; ***, P<0.001; **, P<0.01; NS, non-significant; TN, total nitrogen; NH 3-N, ammonia-nitrogen; 3, proportion of TN as NH3-N (g/100 g); 4, weight of silage NH3-N/weight grass N ensiled (g/g).

Ensilage of grass with grape marc resulted in a significant reduction in NH3-N content of the silages, suggesting a reduction in proteolysis (amino acid deamination) of the grass N fraction during ensilage. However, this result is likely to reflect the increasing proportion of grape marc within the overall treatment DM (range 0 to 304 g/kg DM) since there was no significant effect (P=0.283) of treatment on NH3-N when the results were expressed as the weight of silage NH3-N/weight grass N ensiled (g/g) in each silo (e.g. 0.142 and 0.136 g NH 3-N /g grass N ensiled for control and mean of grape marc treatments respectively). Further work is required to determine the effect of grape marc and other tannin-containing by-products on silage N fractions and in vivo utilisation of forage N.

The effects of the tannins on the reduction of silage ammonia (using 1st cut grass) and the Kow-values recorded were used to select two tannins for use in the in vivo N-balance study, i.e. mimosa and chestnut tannins (see Section 7.0).

5.4 Conclusions (Sub-objective 2a)Effect of tannins on ammonia: All TPs, even at low concentrations (21.2 g/kg DM), reduced NH3-N concentrations in the grass silage by a mean of 12% relative to the negative control, i.e. from 2.13 to 1.88 g NH 3-N/kg DM. Grapemax and tara tannins represent some of the extremes exhibited in the response. There was also a large response for the rate of tannin addition. Moreover, there was also a significant linear rate x tannin interaction, which showed that the response varied across the various tannins.

Effect of tannins on buffer soluble nitrogen (BSN): Tannins significantly reduced BSN content, which represented the major component of the total N fraction from 46.1% of TN in the negative control to 42.2% of TN for the ‘average tannin’ at 21.1 g/kg DM. The variation in the tannin response is illustrated by the differences observed for Grapemax and tara tannins. The response was greatest up to ~61.1 g added tannins/kg DM and diminished at higher tannin additions. There was a significant linear rate x tannin interaction, which showed that the response differed across the various tannins. This suggested that there are differences in the potential of the various tannins to reduce proteolysis during ensilage.

Effect of tannins on buffer soluble NPN (BS-NPN): The BSN fraction of the silages consisted mainly of BS-NPN and only low levels of buffer soluble protein N (BS-PN). The response in silage BS-NPN to the tannins and rate of tannin addition was, therefore, similar to that observed for BSN. Grapemax and tara tannins are shown for illustrative purposes. Averaged across all treatments, tannins decreased BS-NPN from 8.19 to 6.36 g/kg DM compared to the negative control of 9.17 g/kg DM. The significant rate x tannin interaction showed that the different tannins produced different responses. Therefore, it may be possible to select those tannins that have the greatest effect for further study.

Effect of tannins on buffer soluble PN: BS-PN contents varied over a relatively narrow range and there was no significant effect of rate of tannins on BS-PN. Although there was a significant difference in the response to the different tannins, the observed variation is likely to result from the errors involved in calculating this fraction from the difference between BSN and BS-NPN.

Effect of tannins on ADIN: There was a significant effect of tannin and also a significant linear rate x tannin interaction on silage ADIN content, resulting in different slopes for the response of the different tannins. However, ADIN contents were quite low and remained similar in all treatments (0.98 g/kg DM) – except for the grapemax tannins. It would appear that the grapemax tannins bound most strongly to grass silage proteins as these tannins tended to produce higher ADIN and less ammonia.

Lower NH3-N and BSN contents may either result from an inhibition of microbial proteolysis or from the binding of tannins to peptides. This finding may be of practical importance since it may reduce the intake of soluble-N compounds in grass silage which are known to be used inefficiently by ruminants. Further work will be required to determine the rumen and post-rumen availability of the tannin-bound peptides and proteins.

SID 5 (Rev. 3/06) of 29

Grass ensiled with grape marc, instead of the commercial TPs, had significantly less NH3-N. However, this reduction was probably a reflection of the increasing proportion of grape marc in the overall DM, as there was no significant effect when the results were expressed in terms of silage NH3-N/grass N. Further work will be needed to ascertain if tannins in agro-industrial by-products can interact with the proteins of co-ensiled products.

6.0 Scientific approach - Sub-objective 2b: To test effects of added tannins during ensiling on in vitro susceptibility of resulting silage N to degradation and deamination. Sub-objective 3: To test/simulate effects of adding tannins at point of feeding on in vitro susceptibility of resulting silage N to degradation and deamination.

6.1 Introduction: The potential of tannins to improve N-utilisation in ruminants is mediated through their ability to render plant proteins less or more slowly susceptible to degradation in the rumen. Therefore, the aim of Sub-objective 2b and 3 was to undertake an initial study to determine the effect of adding tannins at ensilage or at the point of feeding on the susceptibility of the silage-N fractions to in vitro degradation in bovine rumen fluid.

6.2 Materials and MethodsThree tannins (chestnut, mimosa and grapemax tannins; see Table 1) were selected for further study based on their effectiveness to reduce silage NH3-N content. The tannins were ensiled at 120 g DM/kg grass DM in laboratory-scale silos as described previously (see Section 5.0). In addition, a negative control (no added tannins) grass silage treatment was prepared. The resulting silages were oven-dried (100oC for 8 hours) and milled (1 mm screen). Each of the treatments (1 g dried material) was weighed into 50 ml polypropylene tubes in triplicate. In order to simulate the addition of tannins at the point of feeding, the control silage was weighed into separate tubes and the equivalent weight of tannins present in the tannin-treated silages was added immediately prior to incubation with 40 ml strained rumen fluid (obtained from 3 animals). Following the addition of rumen fluid the tube contents were mixed and the tubes incubated at 39oC for 1 hour, with manual stirring every 15 minutes. The fermentation was then stopped by adding 5.5 ml 2M sulphuric acid, centrifuged at 10,000 rpm for 5 minutes. The supernatant (20 ml) was analysed for NH3-N content by steam distillation following the addition of 10 ml hot, saturated sodium borate solution (10 ml). The entire in vitro incubation process was completed on two separate occasions.

6.3 Results and DiscussionThe mean results for NH3-N release (mg/ml) following incubation of the grass silage treatments in rumen fluid for 1 hour is summarised in Fig 8. Ammonia-N concentration was used as an indicator of microbial silage protein degradation and deamination, and assuming that microbial NH3-N uptake was low within the incubation period (1 hour). The results show that the mimosa tannins generated the lowest NH3-N contents. The results were similar whether the mimosa tannins were added at the point of ensiling or immediately prior to incubation and suggests a strongly inhibitory effect against proteolysis both in silo and in vitro. The effect of the mimosa tannins is also interesting in the light of the results from the in vivo N-balance study (see Section 7.0) where significant differences were observed between the chestnut and mimosa tannins.

In each case the tannin-treated silages had much less NH3-N than the control grass silage. However the effect of the tannins simulating their use at the point of feeding was much more variable and contrasted with the in vivo N-balance study, where no significant differences were recorded for the effect of adding tannins at ensilage or at the point of feeding. Clearly, rumen micro-organisms would have had little time to adapt to the tannins in this in vitro method. However, it would appear that the chestnut and grapemax tannins – in contrast to the mimosa tannins - did not inhibit in vitro proteolysis. Further work will be required to develop in vitro models to study the effects of tannins on rumen degradation of silage N-compounds and their post-rumen availability.

Figure 8. In vitro degradation of silage nitrogen compounds in tannin-treated silages or simulating their effect at the point of feeding.

SID 5 (Rev. 3/06) of 29

0.000

0.020

0.040

0.060

0.080

0.100

0.120

-ve control Chestnut Mimosa Grapemax

Am

mom

ia-N

(mg/

ml)

ControlEnsilagePoint of feeding

6.4 Conclusions (Sub-objective 2b and 3)The three tannin-treated silages produced much less NH3-N than the control grass silage during in vitro incubation with rumen fluid. However, simulating the use of these tannins at the point of feeding produced highly variable results: the chestnut and grapemax tannins – in contrast to the mimosa tannins – appeared to have no effect on in vitro proteolysis. This suggested that chestnut and grapeseed tannins did not inhibit in vitro proteolysis by binding to silage proteins or by inhibiting proteolytic bacteria in vitro. This is of interest because chestnut tannins have been reported to be effective against pathogenic intestinal bacteria and have also been used to prevent diet-induced diarrhoea in young cattle and pigs (Dr S. Battaglia, pers. commun.; Krisper et al 1992). These findings suggest that the chestnut and grapemax tannins may confer some benefits on both animal nutrition and health.

However, these in vitro results differ from the findings of the in vivo N-balance study, where no significant differences were recorded for the effect of adding tannins (chestnut and mimosa) at ensilage or at the point of feeding. Clearly, rumen micro-organisms would have had little time to adapt to the tannins in this particular in vitro method. Further work would be required to develop in vitro models for studying the effects of tannins on rumen degradation of silage N-compounds and their post-rumen availability. Nevertheless, it is possible that the marked mimosa tannin effect on in vitro proteolysis could explain their negative effects in the in vivo N-balance study (see Section 7.0).

7.0 Scientific approach - Sub-objective 4: To validate the findings of Sub-objectives 2 and 3 by N excretion studies in ruminant animals.

7.1 IntroductionThis Project investigated the potential of non-toxic plant tannins to improve grass silage N utilisation in vivo. This included studying the effect of tannins, either when added to grass at ensilage or when incorporated in grass silage-based diets at the point of feeding. Based on the results obtained in Sub-objective 1 and 2, mimosa and chestnut tannins were selected to determine their potential to improve N utilisation in vivo.

7.2 Materials and MethodsCommercial TPs of mimosa (Valretan MB; see K99-1 in Table 1) and chestnut (Valretan TCK; see K99-2 in Table 1) tannins (75 kg of each) were purchased from SCRD, Le Havre, France in September 2005. In addition to studing the effects of tannins as silage additives, the tannins were also characterised according to the procedures given in Sub-objective 1 to determine their Kow value (see Table 1) and ITC protein binding affinities. Fresh grass (1st cut), from a perennial and tetraploid perennial ryegrass sward (sown 9 March 2004), was mown with a mower conditioner, left to wilt (~ 24 hours) and then picked up with a precision chop forage harvester. The required amount of fresh grass (~3000 kg) was placed in heavy gauge plastic bags (in 20 kg quantities) and then stored frozen at -20oC until required for ensilage.

The ensilage process was completed over a total of five days (12, 14, 18, 21 and 26 October 2005) and sufficient fresh grass was thoroughly defrosted immediately prior to ensilage. Three grass silage treatments were prepared; control grass silage (no added tannins), grass silage A (grass treated with mimosa tannins) and grass silage B (grass treated with chestnut tannins). The tannins were added to the fresh grass at a target rate of 150 g tannin DM/kg grass DM (based on an estimate of the grass DM of 245.2 g/kg). The silage treatments were prepared in 220 kg amounts using a mixer wagon. For the tannin-treated silages the grass was thoroughly mixed and then the tannin slowly added to the grass while still mixing. Ensilage was carried out in drum silos, with a capacity of ~100 kg fresh material, lined with a heavy gauge plastic bag. Grass was added to the silos in 20 kg amounts and then consolidated before adding further 20 kg amounts to the silos. After completion of the filling process, the silos were carefully sealed, and a 20 kg weight used to aid consolidation and then ensiled for ~90 days.

SID 5 (Rev. 3/06) of 29

In order determine the effect of the tannins at ensilage or point of feeding on N utilisation in vivo, five experimental diets (designated T1-T5) were prepared from the three grass silage treatments as follows:

Experiment diet T1 – Control grass silage Experiment diet T2 – Control grass silage + mimosa tannins added at the point of feeding Experiment diet T3 – Control grass silage + chestnut tannins added at the point of feeding Experiment diet T4 – Grass silage A (i.e. grass ensiled with mimosa tannins) Experiment diet T5 – Grass silage B (i.e. grass ensiled with chestnut tannins)

The treatment diets were then combined with a second control silage (C2)1 (following ensilage in a conventional clamp silo) so that the silage DM from the treatment diets (T1-T5) represented on average 49% of the forage DM within the overall treatment diet. The daily amounts offered were calculated to provide the sheeps’ metabolisable energy (ME) allowance for maintenance using the relationship given in Equation 3 below:

[Equation 3]: ME allowance (MJ ME/day) = (0.1 x liveweight) + 1.8

The study was completed using a 8 x 5 incomplete Latin Square design with eight mature wether sheep (mean initial liveweight 71.4 ±4.39 kg) and 5 periods (each lasting a total of 21 days). The daily amounts of each experiment diet (T1-T5) were weighed immediately prior to the start of each period and then stored frozen until required. All experiment diets were thoroughly defrosted prior to feeding and given in equal portions at ~09.00 and 16.30 h daily. For treatment diets T2 and T3, the tannin (61 g fresh weight/day) was mixed with the grass silage immediately prior to feeding. In addition, all diets were supplemented with 7 g sheep mineral/vitamin supplement per animal per day and all animals had free access to clean drinking water throughout. In vivo N balance was estimated by making complete collections within each experiment period of faeces and urine over a 5 day collection period following a 15 day adaptation to the treatment diets. Adaptation to the diet was carried out in individual raised pens and faeces and urine (preserved by the addition of 150 ml 2 M sulphuric acid) collection was performed in metabolism crates. During the collection period the amount of feed offered and refused each day was recorded. The N-balance study started on 4 February and was completed on 20 May 2006.

Following completion of the in vivo study, representative samples of the treatment diets, faecal and urine samples were analysed according to the methods described previously (see Section 5.0). Sub-samples of the silage treatments offered in each period were analysed for oven DM, ash, TN (Kjeldahl method) and NDF content. In addition, a composite sample of each fresh silage treatment across the five periods was analysed by near infrared spectroscopy (NIRS), based on an approach described Baker et al (1994) including prediction of fermentation characteristics and metabolisable energy (ME) content. All feed refusals were analysed for oven DM and ash content and the values used to calculate DM and organic matter (OM) intake. Sub-samples of faeces were analysed for oven DM content. Representative samples were also freeze-dried and milled (<1 mm), and then analysed for ash, TN (Kjeldahl method) and NDF content. Urine was analysed for TN (Kjeldahl method).

7.2.1 Statistical analysisThe data were analysed using ANOVA with effects for animal, period and treatment. Treatment effects were partitioned into effects of tannin treatment (mimosa and chestnut tannins), method of tannin inclusion (ensilage and feeding) and the tannin treatment x method of inclusion interaction. Contrast analysis for comparing the control vs mean of the tannin treatments was also completed. In addition, analysis of carry-over effects was estimated for all parameters.---------------

Note 1 : During period 1 of the N-balance study very high feed refusals were encountered for all treatment diets containing tannins but especially with the mimosa tannins (T2 and T4). This problem was resolved by reformulating the tannin-treatment diets so that the overall tannin content of the treatments was reduced from 150 to 75 g tannin DM/kg grass silage DM. For this reason it was necessary to complete a 5 th period (21 days) in addition to the four periods originally planned in order to complete the N-balance study.

7.3 Results and DiscussionThe mean chemical composition of the grass silages (untreated controls and tannin-treated silages), and the mimosa and chestnut tannins used in the preparation of the treatment diets (T1-T5) throughout the study is summarised in Table 9. The higher DM content of silage A and B reflects the addition of the tannins to the grass prior to ensilage. Lactic acid was the predominant fermentation acid in all silages although a reduction in total fermentation acids was reported for the tannin-treated silages. This may suggest some degree of inhibition of bacterial fermentation during ensilage. The tannins were low in N and ash (mean 3.8 and 38.4 g/kg DM respectively).

Table 9. Chemical composition of grass silages1 and tannins used in the preparation of the treatment diets (as g/kg DM unless stated otherwise).

Parameter Control silage

Controlsilage (2)

Silage A Silage B Mimosa tannins

Chestnut tannins

Dry matter (g/kg) 217 265 239 244 917 938

SID 5 (Rev. 3/06) of 29

Nitrogen 19.0 21.8 20.2 23.2 3.1 4.5Ammonia-N (g/100g N) 12.3 10.1 11.8 13.2 - -Neutral detergent fibre 524 487 515 484 - -Acid detergent fibre 363 348 384 373 - -Ash 84 83 84 91 45.8 30.9pH 3.7 3.9 3.8 3.9 - -Total fermentation acids 177 126 134 123 - -

Lactic acid 153 108 111 99.2 - -Acetic acid 12.0 15 12.0 15.0 - -Butyric acid 5.6 <1.0 5.2 3.9 - -

DOMD (g/kg) 690 700 670 680 - -ME (MJ/kg DM) 11.1 11.3 10.7 10.9 - -

1, Chemical composition predicted by NIRS based on the procedure of Baker et al. (1994); Silage A, ensiled with mimosa tannins; Silage B, ensiled with chestnut tannins; DOMD, digestible organic matter in the DM; ME, metabolisable energy.

A difference in the actual DM content of the grass pre-ensilage and the DM value used in the initial calculations (216 and 225 g/kg respectively), and small variations in the DM content of the silage treatments across the five periods resulted in the tannins being added at a mean rate of 80.8 g DM/kg grass DM (or representing 74.8 g/kg treatment DM). The mean DM and N content of the treatment diets was 220, 233, 233, 235 and 229 g/kg (average standard error of the mean (SEM), 0.51 g/kg) and 21.7, 21.2, 21.4, 21.7 and 22.5 g/kg DM (average SEM, 0.12 g/kg DM) for T1-T5 respectively. The treatment diets, T2-T5, contained 74.1, 74.6, 76.1 and 74.3 g tannins/kg DM (average SEM, 0.29 g/kg DM), respectively.

The analysis of the in vivo data showed that the ANOVA accounted for a large proportion of the variance across all treatments (denoted by high R2 values) and therefore there was little unexplained variation within the model. As a result, highly significant differences were recorded for small differences between treatment means. Therefore, it is also necessary to moderate the interpretation of the statistical significance of the results by their practical impact. Analysis showed that there were no significant carry-over effects between experimental periods. The results of the in vivo DM intake (DMI) and apparent digestibility values are summarised in Table 10, and showed that there was no significant effect (P>0.05) of method of including the tannins in the treatment diets (i.e. either at the point of feeding or ensilage). Overall there was no significant difference (P>0.05) in DMI between the control and the mean of the tannin treatment diets (704 and 682 g/day respectively). However, there was a significant (P<0.001) effect of tannins on the in vivo apparent digestibility values determined for DM, OM and NDF. In all cases, the addition of tannins resulted in significantly (P<0.001) lower digestibility values compared with the untreated control (T1). It is also of note that significantly lower values were recorded for the mimosa than for the chestnut tannins. The lower digestibility values for the grass silage fractions (e.g. mean NDF digestibility coefficient, 0.62 and 0.69 for mimosa and chestnut respectively) suggest that the mimosa tannins result in a greater inhibition of microbial degradation in the rumen, the main site of digestion in ruminants.

Table 10. Dry matter intake and apparent digestibility values of control and tannin-treated grass silage treatment diets.

Treatments Statistical significanceParameter Added at

feedingAdded at ensiling

Ave. SEM

Tannin (T)

Method (M)

T x M Contrast1

Control(T1)

Mim. (T2)

Ches.(T3)

Mim. (T4)

Ches.(T5)

DMI (g/day) 704 660 729 591 746 19.1 *** NS * NS

DOMD (g/kg)

641 568 627 576 601 10.2 *** NS NS ***

Apparent digestibility coefficientsDM 0.69 0.61 0.68 0.61 0.65 0.010 *** NS NS ***OM 0.71 0.63 0.69 0.63 0.66 0.010 *** NS NS ***NDF 0.71 0.61 0.69 0.62 0.68 0.012 *** NS NS ***Ave. SEM, average standard error of the treatment means; Tannin (T), effect of mimosa (T2, T4) vs. chestnut (T3, T5) tannins; Method ; effect of adding tannins at point of feeding (T2, T3) vs at ensilage (T4, T5); T x M, tannin x method interaction; Contrast1, comparison of control (T1) with mean of tannin treatments; DMI, dry matter intake; DOMD, digestible organic matter in the dry matter (DM); OM, organic matter; NDF, neutral detergent fibre; ***, P<0.001; *, P<0.05; NS, non-significant (P>0.05).

SID 5 (Rev. 3/06) of 29

The results of the in vivo N intake, losses and N retention are summarised in Table 11. Differences (P<0.001) in total N intake between the mimosa and chestnut tannins reflect the significantly lower DMI recorded for the mimosa-treated silages (Table 10). Overall there was no significant difference (P>0.05) in total N loss (g/day) between the control and the mean of the tannin treatment diets (15.2 and 14.2 g/day respectively), however, these figures do not account for differences in N intake. But a significant difference was detected between the mimosa (13.1 g N/day) and chestnut tannins (15.3 g N/day). The results indicate that tannins generated some rumen-escape protein as urinary N losses, as a proportion of total N intake, were significantly (P<0.001) lower (56.4 and 65.1 g/100 g) whilst faecal N losses were significantly higher (40.2 and 29.8 g/100 g) in the tannin-treated silages compared to the untreated control. Also, faecal N losses as a proportion of total N intake were significantly (P<0.01) higher for the mimosa than for the chestnut tannins (mean 42.1 and 38.2 g/100 g respectively). Therefore, tannins resulted in a shift in N excretion from urine to faeces. A lower urea content in urine is likely to lead to lower volatile N losses to the environment. As a result of the shift in the pattern of N excretion there was no significant (P>0.05) effect of treatment on N retention (i.e. difference between N intake and total N losses in urine and faeces) in mature wether sheep. However, it is possible that young growing animals (e.g. lambs) with a high protein requirement would produce a different response. Table 11. Nitrogen intake, losses and retention for control and tannin-treated grass silage treatment diets (as

g/day unless stated otherwise).Treatment Statistical significance

Parameter Added at feeding Added at ensiling

Ave. SEM

Tannin (T)

Method (M)

T x M Contrast1

Control(T1)

Mim. (T2)

Ches. (T3)

Mim. (T4)

Ches.(T5)

Nitrogen intake

16.0 14.0 15.6 12.8 16.8 0.41 *** NS ** *

Nitrogen lossTotal 15.2 13.6 14.9 12.6 15.7 0.47 *** NS NS NSIn urine 10.5 7.7 9.2 7.2 9.0 0.29 *** NS NS ***In faeces 4.7 5.9 5.7 5.4 6.7 0.25 * NS ** ***

Nitrogen loss as a proportion of N intake (g/100 g)Urine 65.1 55.5 59.4 57.6 53.2 1.86 NS NS * ***Faeces 29.8 42.3 36.5 41.9 39.9 1.31 ** * NS ***Nitrogen retention

5.10 2.19 4.10 0.57 6.88 2.28 NS NS NS NS

Ave. SEM, average standard error of the treatment means; Tannin (T), effect of mimosa (T2, T4) vs. chestnut (T3, T5) tannins; Method; effect of adding tannins at point of feeding (T2, T3) vs at ensilage (T4, T5); T x M, tannin x method interaction; Contrast1, comparison of control (T1) with mean of tannin treatments; ***, P<0.001; **, P<0.01; *, P<0.05; NS, non-significant (P>0.05). 7.4 Conclusions (Sub-objective 4)Tannins significantly reduced the in vivo apparent digestibility of DM, OM and NDF. It is of note that the mimosa tannins yielded significantly lower values than the chestnut tannins. This suggests that the mimosa tannins produced a greater inhibition of microbial degradation in the rumen, the main site of digestion in ruminants. Differences in total N intake between the mimosa and chestnut tannins reflect the significantly lower DMI for the mimosa-treated silages.

The results indicate that both tannins generated some rumen-escape protein, as urinary N losses were significantly lower whilst faecal N losses were significantly higher compared to the control. Interestingly, mimosa tannins produced significantly higher faecal N losses - as a proportion of total N intake - than the chestnut tannins. It is well known that some tannins can produce ‘too much’ rumen escape protein, e.g. Lotus pedunculatus or Acacia karroo (Mueller-Harvey 2006) and it is important to identify those types of tannins that generate optimum levels of rumen escape protein.

Tannins caused a shift in N excretion from urine to faeces. This finding is relevant in terms of pollution from ruminants, as lower urea content in urine is likely to lead to lower volatile N losses to the environment.

Although there was no significant effect of tannins on N retention (i.e. difference between N intake and total N losses in urine and faeces) in mature wether sheep, it is likely that young growing ruminant animals with a high protein requirement would retain more N and may therefore produce a different response.

8.0 Overall Discussion

SID 5 (Rev. 3/06) of 29

Plants produce a great diversity of different tannin compounds. These natural products are finding increasing interest because of their positive effects on animal nutrition and health. Tannins can generate rumen-escape protein which leads to better utilisation of dietary protein by ruminants, they can also increase fertility, prevent bloat, cure diarrhoea and produce anthelmintic effects. Better utilisation of dietary protein not only leads to higher N-retention but also to less environmental N pollution. Moreover, tannins tend to cause a shift in the form of excreted N resulting in less urinary N and more faecal N, which is an environmentally safer form of N. However, there are also reports that tannins produce negative effects. These range from a reduction in organic matter or protein digestibility, to inhibition of rumen fermentation, growth reduction, signs of toxicity and even animal deaths.

Clearly, not all tannins have the same effects. Some tannins are particularly effective at generating rumen-escape protein or as anthelmintics. Other tannins consistently produce harmful effects. The challenge is to identify the parameters that can distinguish between the beneficial and the harmful tannin types. The traditional classification into condensed and hydrolysable tannins (CT and HT) has not proved useful for predicting their effects. Similarly, the suggestion that levels below 5% are beneficial and safe is not valid for all types of tannins. Whilst by definition all tannins bind proteins, different tannins vary by several orders of magnitude in their affinities for proteins. Tannins also vary by several orders of magnitude in their solubilities in water and octanol (Kow-values). This project set out to test the hypothesis that Kow-values combined with measurements of tannin-protein binding strengths can be used to predict the effects of tannins on proteolysis during ensiling and in the rumen.

Protein binding properties of commercially available tanninsCommercial tannin products had a surprisingly wide range of tannin contents (1 to 78%; average: 23.2%). This means that some of these products are in effect only ground plant samples, whereas others are relatively concentrated tannin extracts. Tang et al (1992) using a similar procedure for isolating tannins as the present study reported that commercial extracts contained 10 to 90% tannins. A comparison between the two studies also revealed that the composition of commercially available products is highly variable; e.g. oak products had 1 and 10%, myrabolan 15 and 49%, mimosa 26 and 39% in our study and that of Tang et al (1992), respectively.

Studies using isothermal titration calorimetry (ITC) found that the binding strengths in tannin-protein complexes are affected not only by the structures of the tannins or the proteins but also by co-occurring impurities. The example of myrabolan tannins revealed that some of the impurities in the commercial product bound more strongly with the tannins than protein. This finding is likely to have implications for the extrapolation of data from pure systems to the complexities of in vivo systems.

ITC measurements also revealed that two different types of HT, e.g gallotannins and ellagitannins, exhibited distinctly different protein binding characteristics. Bovine serum albumin (BSA) and gelatin were used as models for the main forage and seed storage proteins, Rubisco and prolamines, respectively. Gallotannins (tara and sumac tannins and pentagalloyl glucose) had similar equilibrium binding constants for the interaction with BSA and gelatin (in the range of 104 to 105 M-1 for the stronger binding sites). In contrast, the ellagitannins, chestnut and myrabolan tannins, exhibited three to four orders of magnitude greater equilibrium binding constants in the interaction with gelatin (~2 × 106 M-1) than with BSA (~ 8 × 102 M-1). Overall, the data showed that relative binding constants for the interactions with BSA and gelatin are dependent on the structural flexibility of the tannin molecules; the more constrained ellagitannin structures bound preferentially to gelatin rather than BSA. These experiments, therefore, demonstrated that the flexible tara tannins bound much more strongly to BSA than the chestnut tannins.

Two of the CT studied, mimosa and grapeseed tannins, showed similar binding isotherms to gelatin and these were comparable to those for HT – except that the CT bound much more strongly than the HT. In contrast, the two CT differed considerably in their BSA binding behaviour, giving vastly different binding isotherms. More research will be needed in order to interpret their binding to BSA; for example, nothing is known about their molecular flexibilities which might influence the BSA binding of the mimosa or grapeseed tannins.

Contrary to suggestions in the literature (Haslam 1996; Mueller-Harvey 2004), no relationship could be found between CT and HT tannin solubilities (Kow-values) and their protein binding strengths as determined by ITC. Instead, it would appear that the overall flexibility of tannin molecules is an important factor that determines the binding strength in tannin-BSA complexes.

Effects of tannins on proteolysis during ensiling of grass In agreement with the literature (Salawu et al 1999; Santos et al 2000; Adesogan & Salawu 2002; Lavrencic and Levart, 2005; Tabacco et al, 2006), all tannins even at low concentrations (21 g/kg DM) reduced ammonia contents in grass silage relative to the control. Whilst there was considerable variation, reductions averaged 12% at the lowest and 38% at the highest tannin addition. For each of the tannins, increasing the amount of tannin had a highly significant effect on proteolysis. It was concluded that the extent of proteolytic response depended significantly on the types of added tannins. Most tannin treatments resulted in very low ADIN contents and

SID 5 (Rev. 3/06) of 29

averaged only 1 g/kg DM. Silages containing grapeseed tannins had most ADIN compared to the control and all other tannins. This is consistent with the strong binding observed between grapeseed tannins and BSA.

More research will be needed to fully understand the impact of tannin structure and tannin-protein binding strengths on proteolysis in silo. An attempt has been made to explain the effects of the various tannins on the proteolysis of Rubisco based on their BSA binding strengths. Grapeseed tannins bound more strongly to BSA than the gallotannins or ellagitannins and reduced proteolysis most effectively. Tara tannins bound BSA more strongly than chestnut tannins and also reduced proteolysis more. However, this was only observed at rates of additions ranging from 65 to 195 g/kg DM. At the lowest rate of addition (21 g/kg DM), tara tannins surprisingly affected proteolysis less than the chestnut tannins. It is possible that this unexpected tara tannin effect at low concentrations is related to the enhanced proteolysis that has been reported previously for tannic acid added to alfalfa silage (Santos et al 2000) – both tara tannins and tannic acid belong to the gallotannin group. Further studies would be needed to investigate this phenomenon. One explanation could be that the flexible gallotannins alter the structure of the globular dietary protein making it more accessible to microbial digestion at low tannin to protein ratios. Indeed, ITC experiments provided some evidence for such an explanation at low gallotannin concentrations.

Effects of tannins on ruminal proteolysisThree tannins that reduced proteolysis effectively in silo and showed contrasting binding behaviour to BSA were subjected to in vitro studies, i.e. chestnut, mimosa and grapeseed tannins. The tannin-treated silages produced on average 48% less ammonia than the control grass silage during in vitro incubation with rumen fluid. However, simulating the use of these tannins at the point of feeding produced highly variable results. The chestnut and grapeseed tannins had no effect on in vitro proteolysis, but the mimosa tannins significantly reduced proteolysis compared to the control. These in vitro findings were in contrast to the in vivo results. No significant differences were found in the in vivo trial between adding tannins at ensiling or at the point of feeding grass silage.

It is possible, however, that the low in vivo intakes of mimosa silages were due to an inhibition of ruminal proteolysis - as indicated by the low in vitro proteolysis of the mimosa treatment that simulated addition at point of feeding. Further research would be needed to investigate if mimosa tannins – in contrast to chestnut tannins - had a negative effect on proteolytic rumen bacteria. Limited evidence exists suggesting that chestnut tannins are not toxic to ruminants and non-ruminants, do not adversely affect intestinal protein digestibility and have a beneficial effect on gut microflora (Krisper et al 1992; Sliwinski et al 2002; Gonzalez et al 2002; Dr S. Battaglia, Silvateam, pers. commun.). Obviously, proteolysis in silo and in the rumen can be reduced either by lower protein degradability or by inhibition of proteolytic bacteria. The generation of rumen escape protein may be a delicate balance between these two processes and may need evaluation for each tannin type.

Effects of tannins on in vivo digestibilities and N-excretion in sheepThe N-excretion study revealed some interesting results. Both chestnut and mimisa tannins appeared to generate rumen-escape protein, as urinary N losses were significantly lower and faecal N losses were significantly higher compared to the control grass silage. Mimosa and chestnut tannins reduced urinary N losses by 13% and 14%, and increased faecal N-losses by 41 and 28%, respectively. This shift in N excretion from urine to faeces is of interest for mitigating N pollution from ruminant farming systems. To our knowledge, this is the first report detailing the potential reductions in urinary N that can be achieved with these tannins. Lower urinary N content leads to lower volatile N losses to the environment. Faecal N is an environmentally safer form of N as it can contribute to the build-up of soil organic matter.

However, it is well known that some tannins can produce too much rumen escape protein and quite high faecal N contents (possible resulting from endogenous N losses) so that animal production is negatively affected, e.g. Lotus pedunculatus or Acacia karroo (Mueller-Harvey 2006). Overall, our results cast doubt on the usefulness of the mimosa tannins despite their relatively low Kow-value. The significantly higher faecal N losses - as a proportion of total N intake - and the relatively low intakes of the mimosa treated silages indicate that mimosa tannins are inferior to the chestnut tannins as dietary supplements. Mimosa and quebracho tannins are structurally related (Appendix 1) and a detailed study by Komolong et al (2001) demonstrated that quebracho tannins did not produce rumen escape protein. Villalba and Provenza (2002) also showed that lambs increased their intakes if quebracho tannins were neutralised with PEG. It would appear from the present study that mimosa tannins – in contrast to the chestnut tannins - had a much more negative effect on microbial degradation in the rumen and also on the in vivo apparent digestibility of DM, OM and NDF. The results also agree with findings by Zimmer and Cordesse (1996), who observed reduced DM and OM apparent digestibilities in sheep and goats with diets containing relatively high contents of chestnut tannins (81 g/kg DM), but no effects on live weight and body condition scores.

Neither chestnut nor mimosa tannins produced a significant increase in N-retention. It is expected that replacing mature wether sheep with young growing ruminants with high protein requirements would lead to higher N retention and would probably produce a different response. Indeed, Decruyenaere et al (1996) found that

SID 5 (Rev. 3/06) of 29

chestnut tannins caused a 5% increase in the non-ammonia nitrogen flow to the duodenum in growing bulls, however there was no effect on overall apparent protein digestibility.

9.0 Main Implications of the Findings1. Judging by the absence of negative effects on in vitro ruminal digestion and on DM intakes, chestnut tannins

appeared to be non-toxic to rumen microflora and to sheep.2. Chestnut tannins showed promise for reducing environmental N pollution from ruminants by shifting the forms

of excreted N from urinary to faecal N. 3. Low levels of chestnut tannins (21 g/kg DM) significantly reduced proteolysis of grass during ensiling (ca

11%) and during ruminal digestion (60%).

10.0 Possible Future Work1. This study was primarily concerned with identifying non-toxic plant tannins that are most effective in reducing

dietary N degradation during ensilage, in vitro and in limited in vivo studies using wether sheep. However, many factors other than the rate and extent of dietary N degradation in the rumen influence the efficiency of N utilisation. Further studies will be required to assess if the chestnut tannins that caused lower proteolysis in silo and in the rumen can lead to less N-excretion from young growing ruminants by creating higher N-retention. Studies will also be required in highly productive ruminants, e.g. dairy cows, which are major contributors to environmental pollution, to confirm the effectiveness of tannins under different conditions, e.g. using silage of differing fermentation quality or fermentable energy supply to the rumen.

2. This study involved the addition of non-toxic plant tannins to grass at ensilage and at the point of feeding using fresh grass. Whilst the results from chestnut tannins appeared promising, it is likely that greater nutritional and environmental benefits may be derived from adding tannins to alfalfa or clover silages because of their higher protein contents.

3. Further work should also investigate the potential of tannin-containing forages such as sainfoin for silages and of agricultural by-products as silage additives. Recent research in Japan has demonstrated that locally available green tea wastes can be ensiled successfully with forages, whole-crop cereals and agro-industrial by-products (Kondo et al 2004a, 2004b, 2006).

4. Our in vitro studies yielded promising results with grapeseed tannins. However, an alternative, cheaper source of these tannins would be required to enable in vivo studies.

5. The present study used model proteins, i.e. bovine serum albumin (BSA) as a globular protein for Rubisco and gelatin as a proline-rich protein for salivary proteins and seed storage proteins. However, further work will be required to study the interaction of tannins with Rubisco, which is the major protein in green plants, in comparison with BSA.

6. Further development of some of the methods would underpin the above research needs. Firstly, the in vitro method that simulated addition of tannins at point of feeding needs further improvement to ensure that it matches with in vivo results. Secondly, methods to determine molecular weights (MW) of tannins using HPLC-GPC with laser scattering detection, in addition to RI detection, would yield absolute MW, rather than relative MW which are based on comparison with a structurally unrelated polymer standard.

11.0 Any action resulting from the research (e.g. IP, Knowledge transfer)Knowledge transfer:16 May 2006: Presentation to farmers at the Dairy Forum, Department of Agriculture, Reading University: ‘Defra

project: Non-toxic tannins as silage additives?’

Meetings with industrial companies:23 March 2006: Dr S. Battaglia, Director of Research (Silvateam, Italy) visited to discuss potential applications of

tannins. Silvateam is the largest producer of commercially available tannins worldwide (chestnut, mimosa and tara tannins). Discussions focussed on using the by-products from Silvateam’s production, which contain tannin residues plus considerable quantities of sugar, for nutritional and veterinary applications. He has since provided us with several free tannin samples for research purposes and indicated interest in future collaborations, e.g. LINK projects or EU projects.

30 October 2006: Dr C. Adams (Kemira, Belgium) met with Dr. A. Kelly (Kingston University) and Dr. I. Mueller-Harvey to discuss potential applications for tannins as nutrition based health products.

8 November 2006: Drs S. Battaglia and E. Hernandez (Silvateam, Italy) visited to explore a joint grant application on tannins for animal health.

Following these meetings, Drs Adams and Battaglia have exchanged e-mails on tannins for use in animal feeds.

Student research projects:Dr Mueller-Harvey is now involved in two pilot studies with colleagues in Food Biosciences, Reading University on testing the effects of several tannins from Silvateam and from agro-industrial by-products on pathogenic and beneficial intestinal bacteria (in collaboration with Prof. M. Woodward, VLA). Two final year students are studying their antibiotic and probiotic activities and their mechanisms of action.

SID 5 (Rev. 3/06) of 29

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

1. References cited in the final Project reportPlease see Appendix 3 for a complete list of the references cited in the report and Appendix 4 for a list of abbreviations.

2. Publications in refereed journals arising from the Project1. Deaville, E.R., Green, R.J., Mueller-Harvey, I., Willoughby, I., Frazier, R.A. Hydrolysable tannin

structures influence relative binding strengths to globular or random coil proteins. Biochemistry (submitted).

2. Deaville, E.R., Givens, D.I., Mueller-Harvey, I. Non-toxic plant tannins as silage additives: effect on proteolysis during ensilage. BSAS conference, Southport 2007 (submitted).

3. Deaville, E.R., Givens, D.I., Mueller-Harvey, I. ‘Grape marc’ as a source of tannins to reduce proteolysis during ensilage. BSAS conference, Southport 2007 (submitted).

3. Publications in refereed journals planned from the Project1. Deaville, E.R., Green, R.J., Mueller-Harvey, I., Willoughby, I., Frazier, R.A. Condensed tannin binding

to globular or random coil proteins. Biochemistry.2. Deaville, E.R., Givens, D.I., Mueller-Harvey, I. Tannins affect nitrogen fractions in silage. Anim. Feed

Sci. Technol.3. Deaville, E.R., Mueller-Harvey, I. Effects of added chestnut and mimosa tannins in grass silage on

nitrogen excretions from sheep. Livestock Prod.

Presentations at conferences and meetingsThe financial contribution by Defra to our research on tannins was acknowledge at three conferences:1. Mueller-Harvey, I. 2005. Plenary presentation: Tannin-containing plants for animal nutrition and health.

XVII International Botanical Congress. 17-23 July 2005. Vienna, Austria. p. 225.2. Mueller-Harvey, I. 2005. Tannins in animal health & nutrition: opportunities for temperate and tropical

regions. In: International Conference - Integrating Livestock-Crop Systems to Meet the Challenges of Globalization. Animal Husbandry Association of Thailand and British Society for Animal Science, Khon Kaen, Thailand, 14-18 November 2005.

3. Mueller-Harvey, I. 2006. 'Sustainable farming practices for optimising the synthesis of natural and veterinary products'. Ethnoveterinary Medicine Conference: Harvesting Knowledge, Pharming Opportunities. British Society of Animal Science, Writtle College, Chelmsford, Essex, U.K. 14-15 Sep 2006. Proceedings, p. 18.

SID 5 (Rev. 3/06) of 29

SID 5 (Rev. 3/06) of 29

general enquiries on this form should be made...

Documents