Effects of Long-Term Dietary Protein Restriction on Intestinal Morphology, Digestive Enzymes, Gut Hormones, and Colonic Microbiota in Pigs

Simple Summary In China, a shortage of protein resources is an important limiting factor to the economic benefit of pig production, and the use of protein-restriction diets balanced with amino acids is an effective strategy to save protein resources. However, long-term protein-restriction diets can impair the growth performance, and the reason is still unknown. This study is to investigate the response of gastrointestinal physiology and gut microbiota to the condition of long-term low-protein diet and to try to provide a theoretical foundation for better use of protein resources in swine production. Results showed that presented with moderate protein-restriction diets, pigs are able to adjust their absorption and consumption of nutrients to maintain growth performance; whereas extremely low-protein diets suppress pigs’ appetite, impair intestinal morphology, decrease Lactobacillus and Streptococcus, and reduce energy expenditure. Thus, moderate reduction of dietary protein is more suitable for pig production than extremely low-protein diets supplemented with essential amino acids, and moderate protein-restriction diets can potentially increase protein utilization in pig production. Abstract Using protein-restriction diets becomes a potential strategy to save the dietary protein resources. However, the mechanism of low-protein diets influencing pigs’ growth performance is still controversial. This study aimed to investigate the effect of protein-restriction diets on gastrointestinal physiology and gut microbiota in pigs. Eighteen weaned piglets were randomly allocated to three groups with different dietary protein levels. After a 16-week trial, the results showed that feeding a low-protein diet to pigs impaired the epithelial morphology of duodenum and jejunum (p < 0.05) and reduced the concentration of many plasma hormones (p < 0.05), such as ghrelin, somatostatin, glucose-dependent insulin-tropic polypeptide, leptin, and gastrin. The relative abundance of Streptococcus and Lactobacillus in colon and microbiota metabolites was also decreased by extreme protein-restriction diets (p < 0.05). These findings suggested that long-term ingestion of a protein-restricted diet could impair intestinal morphology, suppress gut hormone secretion, and change the microbial community and fermentation metabolites in pigs, while the moderately low-protein diet had a minimal effect on gut function and did not impair growth performance.


Introduction
In China, a shortage of protein resources is an important limiting factor to the economic benefit of pig production [1]. In 2016, China's soybean imports were 8391 million tons and accounted for more than 26% of the worldwide production, while high-protein (HP) diets led to excretion of excess nitrogen in feces and urine, resulting in low efficiency of nitrogen utilization and environmental pollution. The

Animals, Experimental Design, and Diet
Eighteen Duroc × Landrace × Yorkshire weaned barrows (35 days of age, average body weight of 9.46 ± 0.61 kg) were randomly assigned using a randomized block design into a normal protein (NP) diet group, a moderately low-protein (MP) diet group, in which protein was reduced by 3% compared to that of the NP group, and a low-protein (LP) diet group, in which protein was reduced by 6% compared to the NP group (Table 1). Each group consisted of six replicates, and the trial lasted 16 weeks. Experimental diets in the MP and LP groups were supplemented with four essential amino acids (L-lysine, L-methionine, L-threonine, and L-tryptophan) to meet the requirements of weaned, growing, and finishing pigs according to NRC (2012). Pigs were maintained individually in metabolic cages with free access to feed and drinking water.

Blood Biochemical Parameters and Hormone Analysis
After a 16-week trial, blood samples were collected by jugular venipuncture after fasted for 12 h, followed by 3000× g at 4 • C for 15 min; serum was separated and immediately stored at −20 • C for further biochemical parameter analysis. Serum biochemical parameters, including total protein, blood urea nitrogen, glucose, cholesterol, and triglycerides, were evaluated using commercial kits according to the manufacturers' instructions (Nanjing Angle Gene Biotechnology, Nanjing, China). Blood hormones of serum were analyzed using commercially-available ELISA kits specific to porcine tissues, according to the manufacturers' instructions (Nanjing Angle Gene Biotechnology, Nanjing, China).

Digestive Enzyme and Intestinal Morphology
Pigs fasted overnight were anesthetized with pentobarbital sodium (50 mg/kg body weight). After jugular exsanguination, the abdomen was incised, and the GIT was immediately removed and rinsed with saline (0.9% NaCl). The digesta of the stomach, jejunum, ileum, and colon were separately collected and stored at −20 • C. The intestinal tissues of the duodenum, jejunum, and ileum (~2 cm long) were obtained and stored in a 10% formaldehyde solution. The villous height and crypt depth were determined. Briefly, each tissue sample was used to prepare five slices of three sections (5 µm thick), which were stained in hematoxylin and eosin with intact and well-oriented crypt-villus units selected for intestinal morphology analysis (Scion Image Software, Version 4.02, 2004, Meyer Instruments, Inc., Houston, TX, USA). The activities of gastric pepsin and trypsin in the digesta were analyzed using commercially-available pepsin and trypsin assay kits according to the manufacturer's instructions (ANG-SH-21041 and ANG-SH-21052, Nanjing Angle Gene Biotechnology, Nanjing, China).

Determination of Short-Chain Fatty Acids, Ammonia-N, and Biogenic Amines
Colonic luminal contents were collected into separate plastic bags for full mixture. One hundred grams of mixed colonic contents were taken and stored in liquid nitrogen for the analysis of the microbial community. The remaining contents were stored at −20 • C to measure the fermentation metabolites, including SCFAs, ammonia-N (NH 3 -N), and biogenic amines.
Three grams of each sample were weighed into a 10-mL tube and mixed with 1 mL of 25% (w/v) metaphosphoric acid and 6 mL of water. Samples were then centrifuged at 17,000× g for 10 min, and the supernatant was removed to analyze SCFA concentration using a capillary column gas chromatograph (GC-14B, Shimadzu, Japan; capillary column: 30 m × 0.32 mm × 0.25 mm film thickness).
The concentrations of biogenic amines in the digesta were determined using the method described by [15]. Briefly, 200 mg of the colonic content was homogenized with 1 mL trichloroacetic acid to precipitate proteins. The mixed solution was centrifuged at 3600 rpm for 10 min, and the supernatant was transferred to a new tube and mixed with an equal volume of n-hexane for fat extraction. This procedure was repeated several times. Then, 20 µL of internal standard, 1.5 mL of saturated sodium bicarbonate solution, 1 mL of dansyl chloride, and 1 mL of NaOH were added to the pretreated sample and vortexed at 60 • C for 45 min. One hundred milliliters of ammonia were added to the mixture to stop the reaction. Finally, the sample was extracted with diethyl ether, dried, and re-dissolved in acetonitrile for injection. The pretreated samples were analyzed by high performance liquid chromatography (HPLC) (Agilent EE00, Palo Alto, CA, USA) equipped with an ultraviolet (UV) detector (Waters 2998, 254 nm, Agilent Technologies, Palo Alto, CA, USA) and a reversed phase column Zorbax Extend-C18 (Agilent Technologies, Palo Alto, CA, USA), 5 mm (250 × 4.6 mm).
For NH 3 -N determination, one gram of colonic digesta was transferred into a 1.5-mL Eppendorf tube, and 1 mL of 0.2 M HCl was added to acidify the samples. After homogenization, the mixed solution was centrifuged at 4000 rpm for 10 min; then, a 40-µL supernatant was removed and acidified with 0.96 mL of 0.2 M HCl and stored in a freezer (20 • C) for NH 3 -N analysis. NH 3 -N concentration was measured using the indophenol method [16].

Illumina MiSeq Sequencing
Universal primers targeting V3 to V4 variable regions of the bacterial 16S rRNA gene were applied for PCR amplification and subsequent Illumina MiSeq sequencing to analyze the microbial community. The forward primer was 341F 5 -barcode-CCTAYGGGRBGCASCAG-3 , and the reverse primer was 806R 5 -GGACTACNNGGGTATCTAAT-3 [17], for which the barcode was an eight-based sequence unique to each sample. The PCR cycling protocol consisted of an initial denaturation at 95 • C for 3 min, followed by 27 cycles at 95 • C for 30 s, 55 • C for 30 s, 72 • C for 45 s, and a final extension at 72 • C for 10 min. A 20-µL mixed reaction was composed of 0.8 µL of 5 µM of each primer, 0.4 µL of Pfu polymerase, 2 µL of 2.5 mM dNTPs, 4 µL of 5-fold Pfu buffer (TransGen Biotech, Shanghai, China), and 10 ng of template DNA. The PCR amplicons were separated by 2% agarose gel electrophoresis and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Shanghai, China). The concentration of PCR products was quantified using QuantiFluor™-ST (Promega, Madison, WI, USA). Then, amplicon pyrosequencing on the IlluminaMiSeq platform was carried out as recommended in the manufacturer's instructions (Shanghai Technology Majorbio Bio-Pharm Co., Ltd., Shanghai, China).

Sequencing Data Analysis
After sequencing, all reads were filtered using QIIME (Version 1.17, Majorbio, Shanghai, China) for quality control. Reads shorter than 50 bp that contained one or more ambiguous bases and two nucleotide mismatches in primer matching were removed. The reads that overlapped longer than 10 bp were assembled. The operational taxonomic units of the effective sequences were carried out based on a 97% similarity cutoff using UPARSE (Version 7.1, Majorbio, Shanghai, China). The chimeric sequences were identified and removed using UCHIME (Majorbio, Shanghai, China). The phylogenetic affiliation of each 16S rRNA gene sequence was analyzed using the RDP Classifier (http://rdp.cme.msu.edu/) with a standard minimum support threshold of 80% [18]. The coverage was calculated to evaluate the sampling effort using Good's method [19]. The richness estimator (using ACE and Chao indices) and the diversity estimator (using Shannon and Simpson indices) were analyzed using the Mothur program [20].

Quantification of Bacterial Populations Using Real-Time PCR
The bacterial groups, including total bacteria and the genes of Lactobacillus, Bacteroides, Ruminococcus, Clostridium Cluster IV, Enterobacteriaceae, and Clostridium Cluster XIV, were quantified using real-time PCR (qPCR) with specific primers (Supplementary Materials Table S1) and SYBR Green PCR Mastermix (Applied Biosystems, Foster City, CA, USA) in the StepOnePlus real-time PCR system (Life Technologies, Carlsbad, CA, USA), as previously described [21].

Statistical Analysis
Statistical analyses were performed using a one-way analysis of variance (ANOVA) with Statistical Software Package (SPSS) 16.0 (IBM, Armonk, NY, USA). The Student-Newman-Keuls multiple range test was employed to compare differences among treatment means. Differences at p < 0.05 were considered significant.

Growth Performance
From weeks 5 to 16, low-protein (LP) diets significantly reduced body weight, average daily gain and feed intake in piglets, growing, and finishing pigs compared with normal protein (NP) diets. The ratio of feed intake to weight gain was decreased in piglets and growing pigs in low-protein (LP) diets. However, moderately low-protein (MP) diets had no effect on body weight, average daily gain, and feed intake during the whole period in pigs (the detailed data belonging to the Institute of Subtropical Agriculture in China).

Blood Biochemical Indices
Reducing dietary protein levels had no effect on the concentration of glucose in the plasma (p > 0.05). Blood urea nitrogen was significantly decreased in pigs fed LP diets compared to those fed NP diets (p < 0.05; Table 2). The concentration of cholesterol in the plasma increased as protein decreased in the LP group (p < 0.05; Table 2). Additionally, the concentration of total protein, albumin, globulin, and triglyceride in the plasma was not affected by different dietary protein levels (p > 0.05; Table 2).

Digestive Enzyme Activity
Stomach pepsin activity increased in the LP group compared to the NP group (p < 0.05), while H + -K + -ATPase enzyme activity in the LP group increased (p < 0.05; Table 3). The dietary protein level had no effect on the activity of jejunal trypsin and ileal trypsin (p > 0.05; Table 3).

Intestinal Morphology
The LP diets reduced the villous height and the crypt depth in the duodenum compared to those in the normal group (p < 0.05; Table 4); however, the ratio of the villous height to crypt depth was not affected in the duodenum. In contrast, there was a reduction of the ratio of the villous height to crypt depth in the jejunum as the dietary protein was decreased in the LP group (p < 0.05; Table 4). Reducing the dietary protein increased the crypt depth in the jejunum (p = 0.082) and decreased the villous height (p = 0.063) and the crypt depth in the ileum (p = 0.091; Table 4).

Gut Hormones
Both MP and LP diets significantly decreased serotonin (5-HT), ghrelin, somatostatin (SS), gastrin, and glucose-dependent insulinotropic polypeptide (GIP) concentrations in plasma compared to the NP diet (p < 0.05; Table 5). Pigs fed an MP diet had significantly increased concentrations of cholecystokinin (CCK) compared to those fed NP and LP diets (p < 0.05; Table 5). Peptide tyrosine tyrosine (PYY) was the same among the three groups. Leptin significantly decreased in the plasma from pigs in the LP group compared to those in the NP and MP groups (p < 0.05; Table 5).

Colonic Microbial Counts
The richness and diversity estimator of microbiota were not affected by dietary protein level in the colonic luminal contents of pigs (Supplementary Materials Table S2). At the genus level, the MP and LP diets significantly decreased the relative abundance of Lactobacillus and Turicibacter compared to those in the colon in the LP group (p < 0.05; Table 6; Figure 1). The relative abundance of Streptococcus was reduced in the LP group compared to the NP group (p < 0.05; Table 6; Figure 1). Significantly higher relative abundance of Prevotella and Lachnospira were observed in the pigs fed the LP and MP diets compared to those fed the NP diet (p < 0.05; Table 6; Figure 1). However, the LP diet rather than the MP diet increased the relative abundance of Dorea, Candidatus, unclassified Clostridiales, and uncultured Peptococcaceae compared to the NP diet (p < 0.05; Table 6; Figure 1).   Real-time PCR quantification revealed that the total bacteria counts were not affected by the CP level in the colon digesta (p > 0.05). However, the counts of Bacteroides, Clostridium Cluster IV, and Lactobacillus showed a significant decrease (p < 0.05; Table 7) in the LP group compared to that in the NP group, and a moderate reduction of protein or extremely low-protein were both related to reduced Ruminococcus counts (p < 0.05; Table 7). Real-time PCR quantification revealed that the total bacteria counts were not affected by the CP level in the colon digesta (p > 0.05). However, the counts of Bacteroides, Clostridium Cluster IV, and Lactobacillus showed a significant decrease (p < 0.05; Table 7) in the LP group compared to that in the NP group, and a moderate reduction of protein or extremely low-protein were both related to reduced Ruminococcus counts (p < 0.05; Table 7).

Microbial Metabolites in Colonic Lumen
In the colon, pigs fed the LP diet had significantly lower concentrations of acetate, isobutyrate, and isovalerate compared with those in the NP diet (p < 0.05; Table 8). Tryptamine and cadaverine were reduced in the MP and LP groups compared to the pigs in the NP group (p < 0.05), and an extremely LP diet decreased the concentration of putrescine (p < 0.05; Table 8). Additionally, significantly reduced concentrations of colonic ammonia were found in the LP group (p < 0.05; Table 8).

Discussion
Reducing dietary CP with free amino acid supplementation is a potential nutritional strategy for saving protein ingredients without impairing growth performance in pigs [22]. Conflicting results were found in the current researchers' previous study: an extremely LP diet reduced growth performance, while moderate restriction of protein had no effect on weight gain and feed intake [5]. Thus, the response to an LP diet could be dependent on the degree of protein restriction. To investigate potential mechanisms underlying protein-dependent regulation of food intake and metabolism, blood parameters, digestive enzymes, intestinal epithelium morphology, intestinal endocrine function, and gut microbiota were tested in this study.

Glycemic Homeostasis
Dietary CP levels reduced by 6% of the NRC (1998), and supplementation with Lys, Met, Thr, and Trp resulted in reduced growth performance in pigs compared to pigs on the control diets [5]. Blood glucose levels remained normal when pigs were fed a low-protein diet for 16 weeks compared to the control group, indicating that glycemic homeostasis was well regulated. GIP is a well-known incretin hormones released from the gut in response to macronutrients and is responsible for amplified insulin release in a glucose-dependent manner [23]. In the present study, protein-restriction decreased plasma GIP levels, which was attributed to reduced feed intake. Ingested macronutrients needed to be digested by digestive enzymes to activate enteroendocrine cells that secrete GIP [24,25]. The current results showed that trypsin activity is not affected by an LP diet, suggesting that enzymatic function may not be the reason for reduced GIP in pigs. Notably, the activity of stomach pepsin and H + -K + -ATPase was increased by an LP diet, potentially promoting protein digestion. However, ingested proteins had little effect on incretin hormone secretion, while carbohydrates and fats potently facilitated gut hormone release [26,27]. The mechanism of protein sensing by enteroendocrine cells is poorly defined [23], and whether dietary protein level participates in blood glucose homeostasis requires further investigation.

Anorectic Effect
Extremely LP diets suppress food intake in humans and rodents [28], which was also shown in pigs in the researchers' previous study [5]. Appetite is partly modulated by gut hormones secreted from the enteroendocrine system after sensing ingested feed. A large increment in PYY, an anorectic hormone, is associated with a high-protein diet compared to isocaloric high-carbohydrate or high-fat diets [29]. These results showed that reducing dietary protein does not decrease plasma PYY levels, despite significantly reduced feed intake induced by protein restriction. Additionally, PYY release can also be stimulated by fats [30,31]. Because dietary protein content was replaced with oil in LP groups, PYY was not reduced significantly by a protein-restricted diet in the present study. Reduced food intake and promoted satiety were also accompanied by an increase in plasma CCK levels with infusion of lipid emulsion into the ileum of humans [32,33], and CCK can also be stimulated by several amino acids and suppress feed intake in rats [34][35][36]. In the present study, although numerous amino acids and oil were included in the LP diet, the concentration of plasma CCK did not increase; thus, pigs in the LP group did not exhibit an obvious sense of hunger, which is also reflected by the decreased concentration of ghrelin in our results, an appetite-stimulating hormone.

Energy Metabolism
The circulating peptide leptin is mainly secreted by adipocytes and reflects the adipose tissue mass of the body [37]; adequate fat stores tend to secrete more leptin to increase energy expenditure; similarly, leptin levels fall after fasting and energy expenditure is reduced; thus, body-weight homeostasis can be controlled via leptin regulation [38]. In the study, plasma leptin levels were reduced in pigs fed an LP diet, suggesting that energy storage was not sufficient and that energy expenditure was likely diminished. In contrast, pigs fed LP diets provided fatter carcasses [39,40], partially due to increased carbohydrates and fats in LP diets [3]. These results showed that cholesterol levels were increased by a protein-restricted diet. Additionally, cholesterol synthesis can be induced by fatty acids [41] and turn into bile acids that facilitate lipid absorption in the gut lumen. However, a previous study showed that a low-protein diet tended to reduce back fat thickness (16.81 ± 2.73) compared to the control group (23.42 ± 2.17) [5]. These inconsistent results may be ascribed to impaired intestinal morphology and reducing energy intake under extremely LP diet conditions [7,42].

Gut Microbiota
Diet modulates the composition of gut microbiota, and long-term dietary habits have a considerable effect on gut microbiota [10]. The present study showed that Streptococcus, Lactobacillus, and Bacteroides are reduced, while Prevotella and Ruminococcus are increased in pigs, probably due to pigs' diet being short of protein and composed predominately of corn and oil. Similar results showed that increased levels of Bacteroides and Prevotella are negatively correlated with energy intake and adiposity [43,44], suggesting that Bacteroidetes may be responsive to calorie intake and that gut microbiota have the potential to maximize energy absorption from a diet rich in carbohydrates. SCFAs are energy substrates for the colonic epithelium (butyrate) and peripheral tissues (acetate and propionate) [45], and the effect of the microbiota on energy deposition is dependent on the SCFA receptor [46]. However, the magnitude of energy produced by gut microbiota fermentation may not be enough to affect the host energy homeostasis, especially when Lactobacillus and SCFA are reduced in the colon, as shown in the present results. Thus, a shift in gut microbiota does not often occur [47][48][49], possibly due to varying experimental conditions, such as age, body weight, and duration of calorie restriction. Further meta-transcriptomic and proteomic studies should be conducted to provide insight into the response of microbial function as a result of dietary change. Apart from their use as energy substrates, SCFAs can also work as regulators, modulating secretion of the hormone GLP-1 by L-cells in the distal small intestine and colon [50], which improves insulin secretion effects. More evidence is needed to support the relationship between gut hormone and microbial metabolites in future research.

Conclusions
In summary, when presented with moderate dietary protein restriction, pigs are able to adjust their absorption and consumption of nutrients to maintain growth performance. However, extremely low-protein diets suppress appetite and reduce energy expenditure, although glucose homeostasis remains stable. Protein-restriction diets affect colonic microbial composition at the genus level, while bacteria diversity showed no significant difference. The production of microbial fermentation was decreased by extremely low-protein diets. However, the large number of pigs and the optimal length of different feeding periods should be considered when the effect of low-protein diets on the growth performance, gut development, and microbiota is investigated in future studies. Moreover, the production scale should also be included.