Comparison of endogenous amino acid losses in broilers when offered nitrogen-free diets with differing ratios of dextrose to corn starch

The nitrogen-free diet (NFD) method is widely used to determine the ileal endogenous amino acids (IEAAs) losses in broiler chickens. Starch and dextrose are the main components of NFD, but the effects of their proportion in the NFD on the IEAAs and the digestive physiology of broilers are still unclear. This preliminary study aims to explore the best proportion of glucose and corn starch in NFD to simulate the normal intestinal physiology of broilers, which helps to improve the accuracy of IEAAs determination. For this purpose, 28-day-old broiler chickens were allocated to five treatment groups for a 3-day trial, including a control group and four NFD groups. The ratios of dextrose to corn starch (D/CS) in the four NFD were 1.00, 0.60, 0.33, and 0.14, respectively. Results noted that NFD significantly reduced serum IGF-1, albumin, and uric acid levels compared with the control (P < 0.05), except there was no difference between group D/CS 0.33 and the control for IGF-1. The increased Asp, Thr, Ser, Glu, Gly, Ala, Val, Ile, Leu, His, Tyr, Arg, and Pro contents of IEAAs were detected in broilers fed the NFD with a higher ratio of D/CS (1.00 and 0.60) compared to the lower ratio of D/CS (0.33 and 0.14). Moreover, ileal digestibility of dry matter and activity of digestive enzymes increased as the D/CS elevated (P < 0.001). Further investigation revealed that the number of ileal goblet cells and Mucin-2 expression were higher in the group with D/CS at 1.00 when compared with group D/CS 0.33 and the control (P < 0.05). Microbiota analysis showed that NFD reshaped the gut microbiota, characterized by decreased microbial diversity and lower abundance of Bacteroidetes, and increased Proteobacteria (P < 0.05). Our results indicate that a higher D/CS ratio (1.00 and 0.60) in NFD increases IEAAs by promoting digestive enzymes and mucin secretion. However, the excessive proportion of starch (D/CS = 0.14) in NFD was unsuitable for the chicken to digest. The chickens fed with NFD with the D/CS ratio at 0.33 were closer to the normal digestive physiological state. Thus, the ratio of D/CS in NFD at 0.33 is more appropriate to detect IEAAs of broiler chickens.

Determination of amino acid (AA) digestibility of raw materials is the critical foundation of feed formulation in poultry production, contributing to better protein utilization and minimizing nitrogen losses. The apparent ileal amino acid digestibility (AID) is measured based on the net disappearance of ingested dietary amino acid from the proximal digestive tract to the distal ileum 1 . Dietary formulas based on the AID index have been widely employed in diet formulation. Nevertheless, AID underestimates the actual digestibility of AA in broilers by neglecting the ileal endogenous amino acids (IEAAs) losses, especially for low crude protein (CP) ingredients, such as cereal grains 2 . The IEAAs are considered an inevitable loss, consisting of salivary and gastric secretions, pancreatic and bile secretions, small intestinal secretions, mucus, sloughed epithelial cells, and microbial protein 3,4 . The formula based on standardized ileal digestibility (SID) can correct the AID by accounting for IEAAs losses induced by the protein-free ingredients. Undoubtedly, SID reflect the digestibility of feed protein more accurately, thereby measuring IEAAs losses to determine the SID of AA is necessary for accurate diet formulation.

Results
Serum metabolites. It was found that the serum concentrations of albumin and uric acid were significantly decreased in all NFD groups when compared with the control group (Table 3, P < 0.001). The concentration of IGF-1 was significantly decreased in group D/CS 1.00, group D/CS 0.60, and group D/CS 0.14 when compared to the control (P < 0.05). However, the content of IGF-1was not changed significantly between group D/CS 0.33 and the control. There were no statistical differences in insulin, glucose, glucagon, and TP levels among groups.

IEAAs losses.
The results of IEAAs losses indicated that Glu was the most abundant AA in IEAAs flow among all NFD groups, followed by Ser (Table 4). Other AAs present in relatively high concentrations were Thr, Asp, Leu, and His, but ranking differed among groups, indicating the D/CS ratio changed the composition of IEAAs flow to a certain extent. Additionally, a higher ratio of D/CS (1.00 and 0.60) increased the endogenous losses of most AA than that of the lower ratio of D/CS (0.33 and 0.14), including Asp, Thr, Ser, Glu, Gly, Ala, Val, Ile, Leu, His, Tyr, Arg, and Pro (P < 0.05), suggesting that the NFD with higher dextrose content could increase the IEAAs losses of chickens.
Intestinal morphology and goblet cells abundance. As shown in Table 5, there were no significant differences in the villus height and crypt depth in the duodenum and jejunum among groups. In the ileum, the villus height had an upward trend in NFD groups when compared with the control (0.05 < P < 0.1), but the opposite trend was observed for crypt depth (0.05 < P < 0.1). All NFD groups showed a significant rise in the ileal villus height/crypt depth ratio (V/C) when compared to the control (P < 0.05). Moreover, the number of goblet cells in jejunal and ileal villi of all NFD groups was significantly increased when compared to the control (P < 0.05).
Digestive enzymes. As shown in Table 6, the maltase activity was significantly higher in both D/CS 1.00 and D/CS 0.60 groups compared to the control (P < 0.05). The α-amylase activity showed a significant decrease in all NFD groups when compared with the control (P < 0.05). However, the activity of lipase and sucrase in group D/CS 1.00 was significantly higher than control group (P < 0.05). Group D/CS 0.60 and group D/CS 0.14 showed higher chymotrypsin activity than the control (P < 0.05). These results indicated that increasing the D/CS ratio to 1.00 leads to a higher activity of lipase, sucrase, and maltase, but the activity of most digestive enzymes can be similar to the control group by decreasing the D/CS ratio to 0.33.
The AID of DM. In addition to increasing enzyme activity, group D/CS 1.00 also showed a significant rise in DM digestibility when compared to the control ( Fig. 1, P < 0.05). There was no significant difference among group D/CS 0.60, group D/CS 0.33, and the control group. However, the DM digestibility in group D/CS 0.14 Table 2. Nucleotide sequence of primers for gene expression analysis. SGLT-1: Na( +)-glucose cotransporter 1;GLUT-2: Glucose transporter type 2.    Table 5. The effects of NFD on intestinal morphology of broiler chickens. 1 V/C: The ratio of villus height to crypt depth. 2 The number of goblet cells were quantified by counting the number of stained goblet cells per 100 μm length of villus, and present as the average number of goblet cells per 8 intestinal villi. Values are expressed as the mean and pooled SEM (n = 6 per group). Different superscript letters in the same row mean significant differences (P < 0.05).  Table 6. The effects of NFD on the ileal digestive enzyme activity of broiler chickens. The enzyme activity of maltase and sucrase were detected in the mucous of ileum, and the enzyme activity of lipase, α-amylase, and chymotrypsin were detected in the digesta of ileum. Values are expressed as the mean and pooled SEM (n = 6 per group). Different superscript letters in the same row mean significant differences (P < 0.05). www.nature.com/scientificreports/ was remarkably lower than that in the control group (P < 0.001), indicating the excess starch is unfavorable for NFD digestion.

Gene expression. Mucus secreted by goblet cells is an essential component of endogenous amino acids loss.
Therefore, the goblet cell marker gene Mucin-2 was detected in this study. As shown in Fig. 2, the relative gene expression of Mucin-2 was significantly higher in the D/CS ratio of 1.00 and 0.60 when compared with the control group (P < 0.05). NFD treatments had an upward expression of mucous glucose transporters, with GLUT-2 and SGLT-1 levels were significantly higher in all NFD groups than in the control group (P < 0.05).

Microflora analysis of ileal digesta.
According to the results of 16S sequencing based on operational taxonomic units (OTUs) analysis, the microbiota diversity in the ileum of all NFD treated chickens were reduced markedly at different levels ( Fig. 3a, b). The principal coordinates analysis (PCoA) of the weighted UniFrac distances resulted in significant segregation among groups (Fig. 3c), confirming the presence of compositional microbiota differences in the ileum of NFD treatments and the control. As shown in Fig. 3d, the preponderant bacteria were Proteobacteria (94.29%, 77.87%, 93.63%, 91.18%, respectively) and Firmicutes (4.39%, 16.91%, 5.38%, 6.77%, respectively) in the ileum of four NFD groups at the levels of the phylum. However, Firmicutes (89.11%) was the preponderant bacteria, which was about tenfold higher than Proteobacteria (9.23%) in the control group at the phylum levels. At species level (Fig. 3e), the preponderant bacteria were Escherichia_coli (92.76%, 69.57%, 92.52%, 89.38%, respectively) in the ileum of NFD groups. However, the Lactobacillus_aviarius (19.32%) and Escherichia_coli (8.54%) were the first and second species in ileal microbial communities in the control group, respectively. Considering there were violently disruptive changes of microbiota composition between NFD treatment and the control, it is necessary to analyze the microbial function in this study further. Moreover, the Tax4Fun analysis based on Kyoto Encyclopaedia of Genes and Genomes (KEGG) data predicted differentially expressed functional pathways among groups (Fig. 4). The four NFD groups harbored very close profiles but differed from those predicted in the control. As shown in Fig. 4, a higher abundance of functions reported for transporters, secretion system, ABC transporters, and two-component system were predicted in NFD groups compared to the control, indicating the microorganism in NFD groups plausibly meets its nutrient requirements for adaption. However, metabolic information related  www.nature.com/scientificreports/ to the DNA repair and recombination protein, the metabolism of amino acids, pyruvate, purine, and pyrimidine appears to be increased in the control group.

Discussion
There are many methods to determine the IEAAs losses in broiler chickens, such as the regression method, the enzymatically hydrolyzed casein method, the NFD method, and the 15 N-leucine single-injection method 7,13 . Among these various methods, the NFD method is most widely used for its simplicity, convenience, and low cost. However, one of the fundamental concerns of NFD is its lack of dietary protein, and the animals can be expected to suffer from malnutrition due to the deficiency of essential amino acids. This study suggested that the IEAAs losses evaluated under normal digestive physiology status represent the actual IEAAs loss, and chickens offered the control diet represent the normal digestive physiological state. Therefore, some serum biochemical parameters were determined in this study to indicate the basic physiological status and metabolic functions of broiler chickens. We found that the albumin and uric acid levels were decreased in all of the NFD groups as compared with the control, indicating clearly that the protein deficiency led to malnutrition of animals in NFD treatments because low albumin and uric acid levels in serum indicating a poor nutritional status 14 and efficiency of protein retention 15,16 , respectively. Furthermore, serum IGF-1 concentration in these four NFD groups decreased by varying degrees which could reflect some protein restriction because the previous study has shown that serum IGF-1 concentration was reduced in models of severe energy restriction or protein restriction in young growing rats 17 . However, the levels of glucose, insulin, and glucagon in serum were not influenced by NDF with different ratios of dextrose to starch, suggesting the normal regulation of blood glucose in NDF treatment.
Although changing the ratio of D/CS in NFD did not alleviate the malnutritions of broiler chickens, it did influence the basic IEAAs of broilers in this study. The most abundant amino acids in the ileal endogenous protein of broilers were Glu, Asp, Thr, Pro, Ser, and Gly 18 . Previous research showed that the IEAAs flow depends on the ratio of D/CS in NFD, which increased from 12,779 mg/kg to 17,544 mg/kg of DMI as the proportion of corn starch in NFD decreased from 849.1 to 0 g/kg 10 . Herein, we further narrowed down the ratio of D/CS in the order of 1.00, 0.60, 0.33, and 0.14, and found significant changes when the ratio of D/CS ranged between In the current study, we found that the highest proportion of dextrose (D/CS = 1.00) dramatically increased the activity of lipase, sucrase, and maltase. Excessive digestive enzymes could be reused as either a component of other proteins or recycled after the conservation process 19 , which might be, in part, result in a high level of IEAAs flow. Glucose absorption was mediated by two kinds of transporters, i.e., co-transportation with sodium ions via SGLT-1 and facilitated diffusion via GLUT-2 20 . In this study, we found that NFD rich in glucose and starch significantly increased the gene expression of these two glucose transporters, indicating that the chickens adapted to the absorption of high carbohydrates in NFD. Notably, the digestive enzymatic activity of group D/ CS 0.33 resembled closely with the control group, suggesting that the D/CS at 0.33 was potentially appropriate. However, the digestibility of DM significantly dropped when the D/CS ratio was 0.15, which established its unsuitability for the chicken to digest NFD.
In addition to digestive enzymes, mucins also contribute to the IEAAs loss. The mucus secreted from goblet cells constitutes the interface between the gut lumen and the gut epithelium, which is poorly digested in the small intestine 21 . We observed no significant difference in villus height and crypt depth among the treatment groups, but the number of goblet cells was increased on the villus of jejunum and ileum in NFD treatment groups. Moreover, the gene expression of ileal Mucin-2 was significantly higher in NFD with D/CS at 1.00 at 0.60 than in control. Dietary composition influences the differentiation of intestinal epithelial cells; for instance, diets rich in carbohydrates or AAs lead to different cell differentiation patterns 22 . The nourishment of goblet cells to form mucin depends mainly on the absorbed glutamine and glucose, which is easily provided by the prolamines and www.nature.com/scientificreports/ starch in grain 23 . Previous in vivo studies indicated that feeding starch increased mucus production in pigs and rats 24,25 . As NFD is rich in starch and dextrose, these results potentially indicate that NFD could promote mucin secretion by increasing the number of goblet cells. Apart from mucins, microbial protein in the gut also contributes to IEAAs. A high-glucose diet has been linked to gut microbial diversity losses where the proportion of Bacteroidetes decreased, while the proportion of Proteobacteria was markedly increased 26 . Consistently, in this study, NFD groups with high sugar directly modulate the gut microbiota, characterized by a lower level of Bacteroidetes and a higher level of Proteobacteria. Since Proteobacteria with adherent and invasive properties are considered to be a rich source of lipopolysaccharides (LPS), which has been associated with inflammatory bowel disease and metabolic syndrome 27,28 , we speculate that gut microbiome may be one variable influencing nutrition metabolism function for animals. Therefore, we used Tax4Fun to predict the functional profile of a microbial community and found that NFD increased pathways corresponding to transporters, secretion system, and two-component system, which were closely associated with bacterial adaptation. The adaptation of bacterial species not only relies on the surrounding microenvironment, such as siderophores, exopolysaccharides, protein toxins but also on the two-component system, which allows a pathogen to adapt its gene expression in response to environmental stimuli 29 . Besides, the "secretion systems" facilitate protein toxins transport through the physical barriers that the membranes represent 30 . In this study, the functional prediction of Tax4Fun was consistent with the microbial community changes, indicating that NFD enriched the pathogenic bacteria survival, which represented a considerable health threat to the chicken.
Results of the present study showed that chickens fed with NFD had physiological abnormalities, represented by malnutrition and accumulation of Proteobacteria in the gut, which cannot be effectively ameliorated just by adjusting the proportion of starch and dextrose in NFD. To overcome the limitations of NFD, Moughan et al. 31 proposed an approach that the animal is fed a purified diet containing enzyme-hydrolyzed casein (EHC) as the sole N source to maintain physiologically normal levels of endogenous N flow throughout the intestinal tract. However, the IEAAs losses obtained by this EHC method might be unstable since Ravindran et al. 32 found that increasing dietary EHC concentrations increased the flow of IEAAs at the terminal ileum of broiler chickens in a dose-dependent manner. Due to the restricted scope of the present study, we did not investigate the effects of supplementing EHC or casein to NFD on the IEAAs of broiler chickens. However, it is worth performing more investigations on combining the advantages of different methods to optimize the detection method of IEAAs in the future.

Conclusion
Taken together, the present results demonstrated that a higher proportion of dextrose (D/CS = 1 and 0.6) in NFD increases IEAAs by promoting digestive enzymes and mucin secretion. However, the excessive proportion of starch was unsuitable for the chicken to digest NFD (D/CS = 0.14). The broilers in D/CS 0.33 group were closer to the normal digestive physiological state. Thus, the D/CS in NFD at 0.33 might be more appropriate to detect IEAAs of broiler chickens.

Methods
Experimental design, animals, and animal care. The animal care and experimental procedures described in this experiment were conducted according to the Animal Welfare Committee guidelines and had the approval of Ethics committee of Animal Science and Technology College of China Agricultural University (No.AW11059102-1, Beijing, China). And the experiments were performed in accordance with the ARRIVE guidelines (https:// arriv eguid elines. org).
A total of 210,1-d-old broiler chickens were fed the starter diet up to 27 d. At 28-d-old, chickens with similar body weight were allocated to 5 treatment groups with 6 replicate cages (7 chickens per cage) in each group for a 3-day trial period to estimate IEAAs losses. The five treatment groups comprised a control group (corn-soybean meal), and four NFD groups with different ratios of dextrose to corn starch (D/CS), designated as D/CS 1.00, D/ CS 0.60, D/CS 0.33, and D/CS 0.14. We found that the starch content in the control diet was 38.76%, and it was challenging to make a pelleted diet with starch content less than 30% in NFD after the pretest. Therefore, the variation of starch content in NFD was based on 40% in this study to make it feasible to prepare the experimental diet.
The content of starch in the feed was measured according to the instruction of a commercial kit for starch content detection (A148-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). All the diets were pelleted and contained 0.5% titanium dioxide as an indigestible marker for calculating the IEAAs. The reason why 8.88% zeolite was added to NFD is that the zeolite effectively reduces the viscosity of feed during granulation and prevents molten glucose and starch from sticking together and blocking the granulator. In addition, zeolite does not contain nutrients such as protein, which can meet the experimental requirements.
The dextrose and corn starch were purchased from Qinhuangdao Lihua starch Co., Ltd (Qinhuangdao, China), satisfying the corresponding China National Standard GB/T 8885 and GB/T 20,880, respectively. The ingredient composition and AA levels of NFD and control diets are shown in Table 1.
All chickens were raised in conventional cages, and each group was composed of 6 cages, with 7 chickens in each cage (0.7 m 2 /per cage). The brooding temperature was maintained at 33-35 °C from 1 to 2d, and then the temperature dropped by one degree every two days until 21 °C. The relative humidity was maintained at 65-70% from 1 to 7d, 50-65% from 8 to 31d. The lighting schedule consisted of 24 h from 1 to 2d, 23 h for 3d, 22 h for 4d, 21 h for 5d, and 20 h from 6 to 31d. All birds had free access to feed and water. No animal deaths occurred during the 3-day trial period.
Sample collection. On day 31, one chicken from each replicate was randomly selected to collect blood from the brachial vein and subsequently centrifuged at 1500 × g for 10 min to harvest serum. Afterward, these www.nature.com/scientificreports/ chickens were euthanatized by injecting pentobarbital sodium (50 mg/kg body weight). The intestine was immediately removed and demarcated by the end of the duodenal loop, Meckel's diverticulum, and the ileocecal junction. Approximately 1 cm intestinal segment was excised from the middle portion of the duodenum, jejunum, and ileum and then carefully collected in carnoy fixative (G2312, Solarbio, Japan) for Alcian blue-periodic acidschiff stain (AB-PAS). Another 1 cm ileum segment was collected after washing with ice-cold PBS, snap-frozen in liquid nitrogen, and stored at -80℃ for subsequent analysis of gene expression. The ileal digesta was gently squeezed from the distal ileum and homogenized using a spatula before being collected in two bacteria-free tubes, and then snap-freezing in liquid nitrogen and stored at − 80 °C for further analysis of gut microbiota and digestive enzyme activity, respectively. After the digesta was flushed out with ice-cold PBS, the entire ileal mucosa was scraped off using sterilized microscope slides, collected in tubes, nap-freezing in liquid nitrogen, and stored at − 80 °C for the further analysis of disaccharidase activity. The remaining six chickens in each cage were euthanatized with pentobarbital sodium (50 mg/kg BW), intestine removed, and the digesta of terminal ileum (lower half of the ileum) was collected and pooled within a cage, immediately stored in − 80 °C overnight and then freeze-dried using a vacuum freeze dryer (FD-2, Biocool, Beijing, China).

Calculations of IEAAs and AID.
After freeze-drying, the diet and digesta were ground and sifted through a 40-mesh sieve to ensure homogeneity. The DM concentration of feed and digesta samples was analyzed based on the methods illustrated in AOAC International 33 . The feed and digesta samples were hydrolyzed using 6 mol/L HCl at 105 °C for 24 h under N atmosphere, and then the AA concentration was measured by Amino Acid Analyzer (A-300, Membrapure, Germany). TiO 2 was determined using the procedure described by Myers et al. 34 . Briefly, the homogenized ileal digesta samples were ashed, and then digested using sulphuric acid (7.4 M) and subjected to react with hydrogen peroxide, and the absorbance was measured at 410 nm using a spectrophotometer (Ultrospec 2100 pro, Amersham Biosciences, USA).
The IEAAs were calculated as milligrams of amino acid flow per 1 kg of DM intake using the Eq. (1) by Moughan et al. 35 The apparent digestibility of DM was calculated using Eq. (2). All the data was expressed on DM basis for calculations.
Intestinal morphology. The tissue sections and AB-PAS stain of duodenum, jejunum, and ileum of broiler chickens were prepared by Servicebio Co., Ltd (Beijing, China). The intestinal morphology was measured based on eight representative complete villi in the same AB-PAS stained slide. Mucosal villus height was defined as the length from the tip of the villus to the crypt opening, and the associated crypt depth was determined from the crypt opening to the crypt base. The number of goblet cells was quantified by counting the number of stained goblet cells per 100 um length of villus and present as the average number of goblet cells per 10 villus.
The intestinal digesta and mucosal samples were immediately snap-frozen using liquid nitrogen. The intestinal digesta (0.2 g) was homogenized (3500 rpm, 10 min) with an Ultra-Turrax homogenizer (JIUPIN-92, JIUPIN, WuXi, China) in 9 volumes of saline (4 °C) to collect the homogenate. And then, the amylase, chymotrypsin, lipase activity was determined according to the instruction of the kit (C016-1-1, A080-3-1, A054-2-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). And the data were collected using a spectrophotometer (Ultrospec 2100 pro, Amersham Biosciences, USA). Microflora analysis of ileal digesta. DNA extraction was performed using a QIAampTM Fast DNA Stool Mini Kit (Qiagene, No. 51604). High-throughput sequencing of 16S rDNA gene amplicons was performed by Novogene Biotech Co., Ltd. (Beijing, China) using a NovaSeq PE250 platform (Novogene Biotech Co., Ltd, Beijing, China). The high-quality sequences were clustered into operational taxonomic units (OTUs) at a 97% similarity level, and a total of 1808 OTUs were obtained. And then, the OTUs sequences were annotated with (1) IEAAs flow mg/kg of DM intake = TiO 2diet % TiO 2 ileal digesta % × AA ileal digesta mg/kg of DM www.nature.com/scientificreports/ Silva132 database. According to the species annotation, the α diversity and β diversity were further calculated, and the differences between groups were compared to reveal the different characteristics of microbial community structure under different treatments.
Gene expression. Total RNA was isolated from ileum using RNA Easy Fast Tissue/Cell Kit (DP451, Tiangen Biotech Co., Ltd, Beijing, China), and the concentration and purity of each sample were determined at 260/280 nm. After that, 1ug of total RNA was reversed into the first-strand cDNA using a kit (RR047A, Takara, Kyoto, Japan). Real-time PCR of mRNA was conducted using the ABI 7500 Fluorescent Quantitative PCR system (Applied Biosystems, Bedford, MA). Each RT reaction was carried out by an SYBR Premix ExTaq kit (RR420A, Takara, Kyoto, Japan). The gene-specific primers were commercially manufactured (Table 2; Sangon Biotech, Shanghai, China), and β-actin was chosen as the house-keeping gene. The relative gene expression levels were calculated by the 2 − Ct method 36  Statistical analysis. The data were analyzed by SPSS, version 20.0 (SPSS, IBM, Chicago, IL, USA). Data distribution was checked by Shapiro-Wilk test. Normally distributed data were analyzed by one-way ANOVA for comparisons among groups and then followed by the Dunnett's post hoc test. Values are expressed as the mean and pooled SEM (n = 6 per group). P values less than 0.05 were considered statistically significant, and P values less than 0.01 indicates extremely significant differences.