Effect of carbohydrate source on microbial nitrogen recycling in growing

Abstract


Introduction
Monogastric herbivorous are hindgut fermenters, i.e. the fermentation compartment is located after the enzymatic digestion area and thus microbial biomass synthesised in the large intestine is mostly wasted with the faeces (Hörnicke, 1981).In order to recycle this nutritious material, lagomorphs, such as rabbits have developed a physiological mechanism, termed caecotrophy, involving ingestion of soft faeces (caecotrophes), originating from a selective retention process actively performed in the main fermentation compartment, the caecum, and proximal colon (Leng and Hornicke, 1976;Fioramonti and Ruckebusch, 1976;Hornicke, 1981).Caecal fermentation may be manipulated through the diet to improve the microbial contribution to rabbit protein nutrition.Potential nutritional factors include dietary components able to resist digestion in the small intestine and reach the caecum, such as structural carbohydrates (fibre) or resistant starch (Gidenne, 1997;Belenguer et al., 2002).
Quantification of the contribution of caecotrophy to the host amino acid (AA) metabolism has been limited by the accuracy of the methodology (Belenguer et al., 2008).Conventional procedure involves preventing caecotrophy using a neck collar to collect the soft faeces.Unfortunately, collar fitting in growing rabbits may alter dry matter intake (Carabaño et al., 2000a) and caecotrophes production (Fioramonti and Ruckebusch, 1976;Gidenne and Lebas, 1987;Gidenne and Lebas, 2006).Caecotrophy reached 26% of feed intake when estimated without a neck collar, but only 12 to 15% with the collar fitting procedure (Gidenne and Lebas, 2006).Another direct procedure, a serial slaughtering protocol (Gidenne, 1987), or indirect methodologies based on microbial markers utilization (internal -purine bases (PB), Balcells et al., 1998;or external -microbial 15 N-lysine, Belenguer et al., 2005) allow estimates of microbial-N recycling in vivo without altering physiological behaviour, although the former technique produces only mean values and not individual estimations.
This study aimed to examine the effect of the dietary inclusion of two types of carbohydrates (starch or fibre), characterized by its high (wheat and sugar beet pulp (SBP)) or low (maize and alfalfa hay (AH)) digestive utilization, on caecal fermentation and microbial-N recycling, using either direct or indirect procedures.

Material and methods
Protocols and animal handling were approved by the "Comité Ético del Servicio de Biomedicina y Biomateriales" of the University of Zaragoza.

Animals and diets
Thirty-two New Zealand White newly weaned male rabbits (35-45 days), with a mean initial weight (W) of 0.91±0.017kg were used.Animals were randomised among four experimental diets (8 animals per diet), housed individually in cages and submitted to a 12:12 h light:dark schedule, starting the light period at 8:00.
The four experimental diets were formulated to contain a similar energy (11.9-13.1 MJ digestible energy/kg dry matter, DM, estimated as described by Villamide et al., 2009), protein (170-185 g crude protein/kg DM), and lysine (10.2 mg/kg, estimated) content, using two sources of structural carbohydrates (fibre), alfalfa hay (AH) and sugar beet pulp (SBP).These fibrous ingredients were combined either with maize or wheat grain, as sources of non-structural carbohydrates (starch), at a constant fibre/grain sources ratio (0.80/0.20), i.e. containing about four times more fibre (SBP plus AH) than grain (wheat or maize).Diets were supplemented with 15 N-labelled NH4Cl (0.7 %; 15 NH4Cl, 10 + atom % 15 N ISOTEC, Inc USA) during the experimental period but not during the adaptation period.Ingredients and composition of the four experimental diets are presented in Table 1.Diets were offered at 8:00 in the morning at a constant level (approximately 100 g/d; 90% ad libitum or 2.25 times maintenance requirements, defined as 410 kJ metabolizable energy/kg W 0.75 /d; Xicatto and Trocino, 2010) for the whole experimental period and animals had free access to drinking water.

Experimental procedures
Each experimental period lasted for 35 days, 28 days for dietary adaptation and seven days for experimental measurements.Rabbits were penned individually during the adaptation period and in metabolism cages for the experimental trial.The following schedule was adopted on each experimental design: adaptation to the metabolic cage (days 1-2), digestibility measurements and urine collection (days 3-6), and collar fitting to prevent caecotrophy and allow collection of the soft faeces for the last 24 h (day 7), according to the method of Carabaño et al. (2010).A wooden round neck collar (50 mm i.d. and 270 mm e.d., weighing approximately 67 g) was fitted at 8:00 and maintained for 24 h.Animals were weighed at the beginning of the experiment and then once a week until the end of the trial when they were slaughtered between 9:00 and 12:00.

Sample collection
During the period of experimental measurements daily urine was collected individually under 1 M H 2 SO 4 (50 ml/L, final pH<3), weighed, diluted (to 1 L), sampled (100 mL) and stored at -20ºC.Faeces were also collected daily, together with caecotrophes on day 7, then weighed and immediately frozen at -20ºC.
Animals were slaughtered by cervical dislocation, dissected and the caecum excised and weighed.Caecal content pH was measured with a glass electrode pH-meter.
Immediately afterwards, two samples of caecal contents were taken (1 g each), acidified with either 0.2 M HCl or 0.5 M H 3 PO 4 , and both stored at -20ºC for ammonia and volatile fatty acid (VFA) determination, respectively.The remaining caecal contents were weighed (20 to 50 g), diluted in a methylcellulose solution (9 g NaCl/L, 1 g methylcellulose/L) and chilled at 4ºC for 24 h to dislodge and isolate adherent bacteria as previously described (Belenguer et al., 2005).The resultant microbial pellet was frozen at -20ºC and then freeze-dried for subsequent analysis.Samples of liver were also taken and stored at -20ºC for 15 N-AA enrichment determination.

Chemical analyses
In feeds and faeces, either hard or soft, DM was determined by drying at 60ºC to constant weight.Organic matter (OM) was estimated by ashing samples at 550 ºC for 8 h.Nitrogen (N) was measured by the Kjeldhal method.Neutral and acid detergent fibre (NDF and ADF) and lignin were determined according to Van Soest et al. (1991) after an amylase pre-treatment.Caecal volatile fatty acids (VFA) concentration was analysed by gas chromatography, following the procedure described by Jouany (1982) and ammonia concentration by the method proposed by Chaney and Marbach (1962).
Urinary purine derivatives (PD: allantoin, uric acid, hypoxanthine and xanthine) were analysed by reverse-phase HPLC (Balcells et al., 1992).Purine bases (PB: adenine and guanine) in feeds, bacterial extracts and soft faeces were determined by the same HPLC technique, with the modifications proposed by Martín-Orúe et al. (1995).Amino acid 15 N-enrichments in caecotrophes and liver were measured by GC-C-IRMS as described previously (Belenguer et al., 2005).

Calculation and statistical analyses
Duodenal flow of PB was estimated from the urinary excretion of PD following the predictive model proposed by Balcells et al. (1998) as modified by Belenguer et al. (2008).The microbial lysine intake and the contribution of microbes to tissue lysine were estimated as described previously (Belenguer et al., 2005(Belenguer et al., , 2008)).Briefly, the microbial lysine intake was estimated assuming a true ileal digestibility of 0.82 (Carabaño et al., 2000b) and 0.88 (Storm et al., 1983) for dietary and microbial lysine, respectively.The microbial contribution to tissue lysine was calculated as the ratio of 15 N-enrichments in liver and microbial lysine.Data were analysed by ANOVA as a complete randomised design, following a 2 x 2 factorial structure, with the source of fibre (F; AH vs SBP), the source of starch (St; wheat vs maize) and their interaction as main effects.Analyses of 15 N-AA enrichment were done as a split-plot design, comparing main plot factors (fibre, F and starch, St) against the animal error term (Є1), whereas 15 N enrichment between substrates (Sub; liver and caecotrophes) nested within animals (A) and their interactions were compared against the residual error term (Є2).
The model was as follows: When methodologies to estimate microbial N recycling were compared, data were analyzed using a similar split-plot design with method (M) used instead of substrate.
Data were analyzed using the using the MIXED procedure of the statistical software SAS (2004).The level of statistical significance was set at P<0.05.

Results
No significant effect of the interaction between fibre and starch (F x St) was observed in most studied parameters, and therefore only the effects of the main factors (fibre and starch) are presented in the tables unless otherwise stated.

Digestive parameters
Average growth rate was 26.5±0.69g/d and final live weight was 1.90±0.024kg, and no differences among experimental groups were recorded (Table 2).The lower dry matter intake in animals fed SBP diets (-10%) than in those receiving AH diets (P<0.01) was counterbalanced by a higher DM and OM total tract apparent digestibility (P<0.001).In consequence, digestible organic matter intakes were similar, although NDF and N were also better digested in SBP (0.53 and 0.78, respectively) than in AH diets (0.43 and 0.74; P<0.001).
Digestibility of DM, OM and N were higher for wheat than maize diets (0.70, 0.71 and 0.77 vs 0.68, 0.68 and 0.75, respectively; P<0.05 for DM and OM, and P<0.01 for N), while there were no significant differences between starch sources for digestible OM and N intake and NDF digestibility.

Characterization of caecal fermentation
Animals fed SBP had more fresh caecal contents (99.1 vs 80.6 g/d; P<0.05) and higher empty caecum weight than those receiving AH diets (37.1 vs 32.9; P<0.01), but there were no differences between starch sources (Table 2).
Rabbits with a larger caecum also showed higher total VFA concentrations (89.6 vs 67.5 mmol/L in rabbits fed SBP and AH diets, respectively; P<0.01) and a lower pH (5.69 vs 6.18; P<0.01).Molar proportions of the main VFA did not differ between diets (acetic, 0.79, butyric, 0.14, and propionic 0.06).Ammonia concentration averaged 3.90±0.431mg/dL, and was independent of dietary treatment.
Caecotrophes showed a two-fold higher protein concentration (335 vs 144 g/kg DM) but only half the NDF concentration (297 vs 567 g/kg DM) of hard faeces.

Urinary excretion and duodenal flow of purine compounds
Urinary PD was composed almost entirely of allantoin and uric acid (92 and 8%) with trace amounts of xanthine and hypoxanthine.Allantoin and uric acid excretions averaged 839.5±40.31and 68.6±4.22 µmol/kg W 0.75 , respectively, with both affected by diet.Total PD excretion was greater in rabbits fed AH than in those receiving SBP as the main source of fibre (1035.4vs 780.7 µmol/kg W 0.75 , respectively; P<0.01), but no differences were detected between starch sources (Table 3).In addition, rabbits fed AH diets showed a greater duodenal flow of PB than those fed SBP diets (1.55 vs 1.17 mmol/kg W 0.75 ; P<0.01) due to both the greater ingestion of dietary PB (0.83 vs 0.62 mmol PB/d for AH and SBP-fed rabbits, respectively) and a higher estimated duodenal flow of microbial PB (1.07 vs 0.81 mmol/kg W 0.75 for animals fed AH and SBP diets, respectively; P<0.05).

Microbial 15 N-lysine incorporation
Ingestion of the isotope (0.7 g/d 15 NH4Cl) resulted in enriched AA in caecotrophes and liver (Table 4).For essential AA the isotope enrichments were always greater in caecotrophes than in liver.With non-essential AA, however, there was no clear pattern, with either lower (tyrosine, proline, aspartate, glutamate), similar (glycine), or higher (alanine, serine) enrichments in liver.
The effect on 15 N-AA enrichment was always higher (P<0.05) for both caecotrophes and liver in animals fed SBP compared to those receiving AH as the main source of fibre.Effects of starch type were less pronounced, but for most amino acids, including lysine, enrichments (atom % excess) were greater in maize-fed rabbits when SBP was the main fibre source, whereas no significant differences were observed with AH, as indicated by the significant fibre x starch interaction.For instance, this occurred for lysine either in caecotrophes (0.56 vs 0.50) or liver (0.22 vs 0.18; Table 4).
The contribution of microbial AA to tissues was estimated using lysine and threonine, AA that do not undergo amination or transamination, as described by Belenguer et al. (2005), although in the present study enrichments were measured in caecotrophes rather than caecal bacteria.Microbial contributions to liver lysine and threonine both averaged 0.37 (0.37±0.008 and 0.37±0.014,for lysine and threonine, respectively).Microbial lysine contribution to liver was not affected by the experimental treatment (Figure 1).

Microbial N recycling
Dry matter and N excretion through the caecotrophy process and microbial N recycling, estimated by the three approaches, are presented in Table 5.Average soft faeces excretion, determined by caecotrophe collection, was 12.7±0.71g DM/d, with a high interindividual variation (from 5.8 to 22.7 g/d), and was unaltered by the experimental treatment.
Microbial N recycling values differed among estimation protocols, with the lower values obtained by direct collection (collar method; 0.50±0.032g/d) compared to those derived from PD excretion (1.22±0.080g/d) or 15 N-lysine incorporation (1.39±0.064g/d; P<0.001).The effect of either the source of fibre or starch was not significant for any of the estimation methods.Residual variations (coefficient of variation, CV) were 37, 35 and 18%, for PD excretion, direct collection and microbial 15 N-lysine incorporation, respectively.

Digestive and fermentation parameters
Animals adapted well to the experimental treatments, although growth rate was lower than in similar trials (26.6±0.69 vs 30-38 g/d; de Blas and Villamide, 1990;Gidenne and Bellier, 2000) where rabbits were fed ad libitum, while in our case feed supply was restricted (90% of ad libitum intake), in order to minimize individual feed intake variability.
SBP contains a greater proportion of hemicellulose plus water insoluble pectins (Gidenne, 2003) than AH.The AH has a greater degree of lignification, which may stimulate the transit of digesta through the digestive tract (Gidenne et al., 2001).These differences may explain variations in intake and digestibility and the high fermentation rates observed in SBP-fed rabbits (Gidenne et al., 2010), evidenced by a lower pH (-8%) and a higher VFA concentration (+33%; Fraga et al., 1991;Gidenne, 1997).
Likewise, maize and wheat grains differ in the physical structure of the starch granule.In maize the protein matrix that surrounds starch is thicker, which restricts the enzyme accessibility and reduces digestibility compared to wheat (Rooney and Pflugfelder, 1986;Blas and Gidenne, 2010).

Indirect approaches to estimate microbial contribution
PD excretion values were within the range described previously with similar animals and diets (Belenguer et al., 2008), although lower than values derived from older and heavier animals (Abecia et al., 2005).As previously observed (Belenguer et al., 2002), urinary PD was greater with AH than in SBP-fed animals, due to a higher intake of dietary PB.A similar effect was observed in total and microbial duodenal PB flow.
Ingestion of a 15 N-labelled NH4Cl results in production of labelled microbial AA-N (Belenguer et al., 2005) with a differential 15 N enrichment depending on metabolic inflows from labelled or unlabelled sources (Atasoglu et al., 2004).In this regard, amino acid enrichment in caecotrophes was greater in SBP-fed animals than in those animals receiving AH diets.We might speculate that availability of dietary [non-labelled] N for caecal microbes would be higher in AH-fed animals due to either more dietary resistant protein (Gidenne and Ruckebusch, 1989) and/or greater protein ingestion.Alternatively, SBP diets could have provided more fermentable fibre (such as hemicellulose) reaching the caecum, which would elicit an increased level of microbial biosynthesis and consequently recycling of endogenous labelled N sources.
Although starch is digested mainly in the small intestine (Merino and Carabaño, 1992;Carabaño et al., 1997), the resistant starch structure in maize granule (Blas and Gidenne, 2010) may enhance the starch flow at the ileum (Gidenne et al., 2005) and the microbial activity within the caecum.Thus caecotrophes seemed to show a higher level of 15 N incorporation in maize than in wheat-fed rabbits, although this effect was significant only with SBP as the main source of fibre (+12%), suggesting that the starch effect could be enhanced by presence of fermentable fibre.
In tissues, the AA 15 N-enrichments rely on the contribution of microbial AA, derived mostly from caecotrophy (Belenguer et al., 2005), plus amination/transamination processes.Average tissue enrichments were lower in the essential AA, such as lysine, that do not undergo transamination within tissues (Bender, 1985) and was then used to estimate microbial contribution (mostly through caecotrophy) to AA absorption.
Differences in absolute AA enrichments in caecotrophes and tissues between fibre sources (SBP vs AH, Table 4) were not reflected in the microbial contribution to liver lysine, which seems to suggest that the ingested amount of caecotrophes relative to total intake might be similar among treatments.If this is the case, variations between fibre sources in 15 N-AA enrichments in caecotrophes may be due to differential 15 N/ 14 N availability from either labelled (ammonia, body proteins) or unlabelled (dietary) sources in the caecum.
Regarding the source of starch, no significant effect was observed on microbial AA contribution to tissue lysine (Figure 1), despite the greater microbial 15 N incorporation in caecotrophes (and also in liver) in maize than in wheat-fed rabbits when SBP was the main source of dietary fibre.
The use of 15 N-lysine enrichments in caecotrophes instead of caecal bacteria may overestimate microbial contribution, since, as mentioned, 15 N-lysine enrichment in caecotrophes are lower (-18%) than in caecal bacteria.Even applying such a correction, the revised microbial contribution to tissue lysine (0.30) is still greater than values previously reported either with the same method in growing rabbits (0.22; Belenguer et al., 2005) and lactating does (0.23; Abecia et al., 2008), or with the caecotrophes collection technique in lactating does (0.18; Nicodemus et al., 1999).Differences among trials could be explained by different experimental conditions and in the present experiment the limited feed (and N) supply might have also enhanced the ability of the animals to recycle microbial protein.

Microbial N recycling
The average contribution of microbes to total protein intake differed among methodologies (29.3±1.47,33.3±0.94 and 15.3±0.84%for PD, 15 N-lysine and caecotrophe collection procedures, respectively), confirming previous results from our group (Belenguer et al. 2005;Belenguer et al 2008).Similarly, the lowest values of microbial OM recycling (20.1±1.27,22.7±0.90 and 11.3±0.63g/d for PD, 15 N-lysine and caecotrophes collection procedures, respectively) or its contribution to total OM intake (18.7±1.05,21.2±0.54 and 11.6±0.62%,respectively) were observed with the conventional technique.Collar fitting may distress rabbits and reduce feed intake (Belenguer et al., 2002) and caecotrophes excretion (Gidenne and Lebas 1987;Belenguer et al., 2008), as a consequence of the altered nutritional behaviour.The DM intake was more variable and decreased with the collar (-21%), suggesting a lack of adaptation of the animals, which might explain the lower amount of caecotrophes collected (12.7 g DM/d) and the smaller microbial contribution to total N intake (15.3%) than the range reported by Villamide et al. (2010;15-30 g DM/d or 17-29%).
The present study aimed, however, to investigate the effect of diet on microbial N recycling.In this regard, a previous study from our group (Belenguer et al. 2002), using similar diets (fibre/grain sources ratio, 0.80:0.20),showed an improvement of microbial N recycling (g/d) with AH diets in relation to SBP, using the conventional and the PD techniques, in rabbits fed ad libitum.Conversely, in the current study no significant differences were detected in microbial N recycling between treatments and thus the contribution of soft faeces to total DM (or N) was similar in all diets.Some aspects of the PB metabolism are still unclear in rabbits (Belenguer et al., 2008), and as far as the authors are aware, this is the first study of the effect of different sources of fibre on microbial N recycling using the 15 N-lysine incorporation method.With the soft faeces collection, different results on the effect of the type of fibre on microbial N excretion have been reported, with no variations between alfalfa and soya been hulls (Nicodemus et al. 2007) or differences (from 0.34 to 0.83 g microbial N/day) using a range of different fibre sources (paprika meal, olive leaves, alfalfa hay, NaOH-treated straw and hull from soybean and sunflower) at different inclusion levels (NDF varied from 33.1 to 78.7 % of DM; García et al., 2000).In the last case, however, differences were mostly explained by the low values obtained with straw and sunflower (0.34 g DM/d).

Implications:
The utilization of two different sources of structural carbohydrates, AH and SBP, or two types of starch, wheat and maize, did not result in significant variations in microbial protein recycling.Nevertheless, our results suggest that the utilization of SBP, especially in combination with maize, as a dietary source that may provide a greater amount of carbohydrates to the hindgut, might increase microbial biosynthesis in the caecum, indicating a potential to improve microbial protein contribution.Further research would be necessary to elucidate this effect and make dietary recommendations for these ingredients for growing rabbits.Belenguer et al. (2008).‡ Estimated by substraction of dietary PB ingested, assuming digestibility of 0.913 (Chen et al., 1990), from total duodenal flow of PB.

Table 1 .
Ingredients (g/kg) and chemical composition (g/kg DM) of the experimental diets formulated using two sources of fibre (alfalfa hay or sugar beet pulp) and two sources of starch (maize or wheat), either labelled (Lab.; with 15 NH4Cl) or unlabelled (Unlab.).

Table 2 .
Effect of dietary inclusion of alfalfa hay (AH) or sugar beet pulp (SBP), and maize or wheat as sources of fibre and starch respectively on growth rate, dry matter intake, digestible organic matter (OM) intake, and total tract apparent digestibility of SE: Standard error of the treatment means; NS, non-significant; *, P<0.05; **, P<0.01; ***, P<0.001.

Table 4 .
Effect of dietary inclusion of alfalfa hay (AH) or sugar beet pulp (SBP), and maize or wheat as sources of fibre and starch, respectively, on 15 N-enrichments (atom % excess) in non-essential and essential amino acids in caecotrophes and liver in growing rabbits.SE1 and SE2: standard errors of the animal effect and residual term in split-plot design; * NS, non-significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.25

Table 5 .
Effect of dietary inclusion of alfalfa hay (AH) or sugar beet pulp (SBP), and maize or wheat as sources of fibre and starch, respectively, on caecotrophe dry matter (DM) and N production and microbial nitrogen (MN) recycling using the following methods (M): caecotrophy prevention by using neck collar (CC), urinary excretion of purine derivatives (PD), or microbial 15 N-lysine incorporation ( 15 N-lysine) in growing rabbits.
SE: Standard error of the treatment means; SE1 and SE2: standard errors of the animal effect and residual term in split-plot design; M, methodology; NS, non-significant; ***, P < 0.001.