An activated charcoal-herb extractum complex was obtained from the Fujian Baicaoshaung Biotechnology Co., Ltd. (Nanping, China). The manufacture of CHC was reported in a previous paper (Wang et al., 2019). In brief, CHC is formed by active charcoal sorption of Chinese herbs extractum. Analysis of elemental composition demonstrated that approximately 90% of CHC was carbon. Activated carbon was prepared by crushing and screening cedarwood and pine wood, followed by carbonization and activation at high temperatures (120–130°C). After that, the activated carbon was screened and only the part with the size range of 0.18 to 0.25 mm was retained for the next step. The mixture of traditional Chinese herbs included a raw plant of Pulsatilla chinensis (dried rhizome), Portulaca oleracea L. (dried whole plant), Artemisia argyi Folium (dried whole plant), and Pteris multifida Poir (dried whole plant) in the ratio of 7:8:10:6. The plant mixture was washed and soaked in the five-folds volume of water for 12 h. Subsequently, the plant mixture was decocted twice at 100°C under normal pressure, each for 1 h, then the decoction was filtered and collected, followed by concentration under a low pressure to a small volume (∼0.2-fold volume of initial plant volume) of Chinese herbs extractum. Finally, Chinese herb extractum were mixed with activated carbon in a proportion of 1:9 for 8 h and then dried to make CHC product. AMP was purchased from Qilu Pharmaceutical Co., Ltd. (Shandong, China). Probio (B. subtilis strain C-3102, Calsporin) was purchased from Shanghai Muguan Enterprise Development Co., Ltd. (Shanghai, China). MMT was purchased from Guangzhou Jingmu Feed Co., Ltd. (Guangzhou, China).

Arbor Acres male broilers (n = 864; one day of age; initial BW = 41.83 ± 0.64 g) were allocated randomly to eight treatment groups, with six replicate pens per treatment and 18 broiler chickens per pen. Dietary treatments were: (1) negative control: corn-soybean meal-based diet (NCON), (2) NCON with 200 mg/kg Antibacterial peptide (AMP), (3) NCON with 200 mg/kg calsporin (Probio, equivalent to adding 2.0 × 10⁹ CFU/g of viable B. subtilis), (4) NCON with 500 mg/kg montmorillonite (MMT), and (5) NCON with 250, 500, 750, or 1,000 mg/kg CHC. The optimal supplemental level was selected according to the growth performance in CHC treatments compared to NCON. CHC treatment with optimal dose was used as representative CHC group to analyze intestinal microbial composition with the controls.

The 42-day experimental period included a starter phase (day 0–21) and a grower phase (day 22–42). The basal diet was formulated to meet or exceed estimates of the nutrient requirements of broilers (Table 1) suggested by the (National Research Council [NRC], 1994). Prior to the mixing process, the feed ingredients have been weighed and ground. The ingredients, composed less than 1% (w/w) of the diet (vitamins, minerals, and CHC), were mixed well with soybean meal in advance. Finally, all ingredients were all put into a feed blender and mixed well. The broiler shed contained equipment for ventilation, heating, and lighting. Every wire-mesh cage (90 cm × 90 cm × 75 cm) housed six chicks, each equipped with two automatic teat waterers and a round feeder (diameter: 37 cm). A lighting program with 24-h continuous light was used throughout the trial. All broiler chickens had ad libitum access to feed and water. Environmental temperature for 1-day-old broilers was controlled to about 33°C and then gradually decreased to 24°C at week 3 (the temperature was lowered to 3°C every week), which was maintained at this temperature thereafter. Broilers were vaccinated with Newcastle disease vaccine on day 7 and vaccinated with an inactivated infectious bursal disease vaccine on day 14.

The body weight of birds was recorded on days 1, 21, and 42 of the experiment after 12 h withdrawing of feed but not water. Body weight gain (BWG) was calculated for 1–21, 21–42, and 1–42 days periods. Feed intake (FI) was recorded for the same periods and feed conversion ratio (FCR) was calculated after correcting for mortalities. Two phases of feed samples were collected, and stored at −20°C prior to mycotoxin analysis.

On days 21 and 42, one broiler closest to the average BW of each pen was selected for blood sampling. Blood samples were collected from the jugular vein. Serum was collected after centrifugation at 1,300 × g for 10 min at 4°C and stored (−20°C) for analysis of immune and inflammatory indices. On day 42, after blood collection, birds from each treatment (n = 6) were weighed and then killed by cervical decapitation.

On day 42, the small intestine was divided into three parts: the duodenum (from the pylorus to the ligament of Treitz), the jejunum (from the distal portion of the duodenal loop to Meckel’s diverticulum), and the ileum (anterior portion of the ileocecal junction). The intestine was sliced open lengthwise and gently rinsed in ice-cold PBS. Mucosa from the mid-duodenum, mid-jejunum, and mid-ileum of birds from each treatment (n = 6) was scraped with a glass slide. The glass slide with scrapings was rapidly placed in liquid nitrogen for measurement of intestinal immune and inflammatory indices. Gut contents from the ileum, cecum, and colon of birds from each treatment (n = 5) were collected into sterile plastic bags and immediately chilled on dry ice for analysis of intestinal microbiota.

Mycotoxin content was measured using ultra-high-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) analysis. Briefly, the feed sample was extracted according to the QuEChERS method with slightly modified (Oplatowska-Stachowiak et al., 2015). The analysis was performed using the Acquity UPLC I-class system (Waters, Milford, MA, United States) and mass spectrometer (Xevo TQ-S, Waters) according to Kolawole et al. (2020).

Serum immunoglobulins A, M, and G (IgA, IgM, and IgG) were measured with an automatic biochemical analyzer (Hitachi 7600, Hitachi High-Technologies Corporation, Tokyo, Japan), following the instructions that accompanied commercially available kits for each immunoglobulin (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Concentrations of interleukin-1β (IL-1β) and interferon-γ (IFN-γ) in serum and mucosa of duodenum, jejunum, and ileum were determined by enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Intestinal secretory immunoglobulin A (SIgA) concentrations were assayed using an Sn-69513-type immune counter (Shanghai Nucleus Annular Photoelectric Instrument Co., Ltd., Shanghai, China).

Microbial DNA was extracted from ileal, cecal, and colonic digesta using the E.Z.N.A.^® stool DNA kit (Omega Bio-Tek, Norcross, GA, United States) according to the manufacturer’s instructions. Markers and adaptor-linked universal primers 338F (ACTCCTACGGGAGGCAGCAG) and 806R (GGACTACHVGGGTWTCTAAT) targeting the V3-V4 region were used to amplify microbial 16S rRNA. The PCR amplification of the 16S rRNA gene was performed as follows: 95°C for 3 min, followed by 25 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s and a final extension at 72°C for 10 min. The 2% agarose gel was used for the detection of purity of PCR products, and the products were purified with the AxyPrep DNA Gel Extraction kit (Axygen Biosciences, Union City, CA, United States). Purified amplicons were pooled in equimolar proportions and paired-end sequenced (2 × 300) on an Illumina MiSeq platform (Illumina, San Diego, CA, United States). The processing of sequencing data was conducted as previously described (Wang et al., 2021). Sequenced raw reads were deposited into the NCBI sequence read archive database (Accession Number: PRJNA794039).

After demultiplexing, the resulting sequences were merged with FLASH (v 1.2.11) (Magoč and Salzberg, 2011) and quality filtered with fastp (v 0.19.6) (Chen et al., 2018). Then the high-quality sequences were de-noised using DADA2 (Callahan et al., 2016) plugin in the QIIME 2 (version 2020.2) pipeline with recommended parameters, which obtains single nucleotide resolution based on error profiles within samples (Bolyen et al., 2019). DADA2 denoised sequences are usually called amplicon sequence variants (ASVs). ASVs were aligned with MAFFT and used to construct a phylogeny with FastTree2. Taxonomic assignment of ASVs was performed using the Naive bayes consensus taxonomy classifier implemented in QIIME 2 and the SILVA 16S rRNA database (v138). Analyses of the 16S rRNA microbiome sequencing data was performed using the free online platform of Majorbio Cloud Platform (cloud.majorbio.com).

During the starter phase, compared with the NCON group, supplementation of 500 mg/kg CHC in diets increased BWG (p < 0.01, Table 2). As the level of CHC in the diet increased, BWG tended to increase linearly (p = 0.058) and responded quadratically (p = 0.030). Compared with the AMP and Probio groups, dietary supplementation of 500 mg/kg CHC increased BWG (p < 0.01), and tended to improve FCR (p = 0.062). There was no difference in growth performance among broilers fed positive controls and those fed 0, 250, 750, or 1,000 mg/kg CHC (p > 0.05).

During the grower phase, growth performance was not affected by CHC treatment. Compared with AMP, Probio, and MMT groups, feeding 500 mg/kg CHC tended to improve BWG (p = 0.067) and FCR (p = 0.089) of broilers.

Over the whole experimental period, BWG and FCR quadratically responded with an increasing level of CHC (p < 0.01). Compared with the NCON group, dietary CHC supplementation at a dosage of 500 mg/kg increased BWG (p = 0.024), while other doses showed no effect (p > 0.05). In addition, supplementation of 500 mg/kg CHC tended to improve FCR (p = 0.081), compared with three positive controls.

Concentrations of serum IgA (p < 0.01) on day 21 when feeding 500 mg/kg and 750 mg/kg CHC were higher than that in NCON (Table 3). Concentrations of serum IgA (p < 0.01) on day 42 and IgM (p = 0.018) on day 21 at the level of 500 mg/kg CHC was higher than in the negative control. On day 21 and day 42, serum IgA tended to increase linearly with increased level of CHC from 250 mg/kg to 1,000 mg/kg (p = 0.079 and p = 0.091, respectively). The concentrations of serum IgA (day 21 and day 42) and IgM (day 21) showed quadratic (p < 0.01) responses to increased CHC levels. Serum IgG (p > 0.05) was not affected by CHC on both day 21 and day 42. Immune responses in the 500 mg/kg CHC group were similar to those observed in the three positive controls. Compared with the three positive controls, supplementation of CHC at 500 mg/kg increased serum concentrations of IgA (p < 0.01) both on day 21 and day 42.

Concentrations of serum IL-1β (day 21 and day 42) and IFN-γ (day 21) linearly (p < 0.01) decreased with increasing concentrations of CHC and responded quadratically (p < 0.05) on both day 21 and day 42. Dietary supplementation of 500 mg/kg CHC decreased levels of IL-1β (p < 0.01) and IFN-γ (p < 0.05) in the serum of broilers compared with NCON. Broilers fed 500, 750, and 1,000 mg/kg CHC displayed reduced serum concentrations of IL-1β (p < 0.01) and IFN-γ (p < 0.01) on day 21 compared with NCON. Similarly, 500 and 750 mg/kg CHC groups had lower concentrations of IL-1β (p < 0.01) and IFN-γ (p = 0.017) on day 42 comparing with NCON. Supplementation with 500 mg/kg CHC was more efficient in reducing serum IL-1β (p < 0.01) than AMP and Probio on day 21, and more efficient (p < 0.01) than positive controls on day 42. Supplementation with 500 mg/kg of CHC reduced IFN-γ (p < 0.01) more than AMP and Probio on day 21, while there was no difference between CHC and positive controls on day 42 (p > 0.05).

Supplementation with 500 mg/kg CHC increased concentrations of SIgA in the mucosa of the duodenum and jejunum (p < 0.05) on day 42 compared to that of NCON. CHC at 750 mg/kg increased jejunum mucosa SIgA (p < 0.05) concentration compared with NCON (Figure 1). Duodenum and jejunum mucosa SIgA concentrations were higher (p < 0.05) in the broilers fed 500 mg/kg CHC compared with those in positive controls.

Concentrations of IL-1β in the mucosa of jejunum and ileum in the treatments of 500, 750, and 1,000 mg/kg CHC were lower (p < 0.05) than that in NCON on day 42. No significant differences were observed in the mucosa of duodenal IL-1β and in the mucosa of duodenal, jejunal, and ileal TNF-α concentrations among all treatments (p > 0.05). Broilers fed with 750 mg/kg CHC had lower levels of IL-1β (p < 0.05) in ileum mucosa when compared with the Probio group. Moreover, there was no difference between CHC and positive controls in the mucosa of the duodenum and jejunum (p > 0.05). Furthermore, the level of TNF-α (p > 0.05) in the mucosa of duodenum, jejunum, and ileum did not differ among CHC and positive controls.

After denoising, sequences (4,253,546) from the 75 samples (with an average of 52,513 sequences per sample), representing 19,335 amplicon sequence variants (ASVs), were revealed for the subsequent analyses. Parse curves and species accumulation curves showed that the sampling of each group provided sufficient sequences to reflect the diversity and abundance of bacteria (Supplementary Figure 1). α-Diversity used to measure the distribution of species abundances in a given sample can be indicators of community richness (Sobs, Chao 1, and ACE indexes) and diversity (Shannon index) (Ai et al., 2017; Ji et al., 2017). The α-diversity of the microbiota in the ileum, cecum, and colon were analyzed (Figure 2 and Supplementary Figure 2). In the ileum, CHC treatment decreased community abundance represented by lower Sobs, Chao 1, Ace, and Shannon compared with NCON (p < 0.05). The positive controls of AMP and MMT also decreased the community abundance represented by lower Sobs than NCON (p < 0.05). At this point of decreasing the richness of bacteria, the characteristic of CHC showed more similarity to MMT in that both of them showed similar Sobs, Chao 1, Ace, and Shannon. Shannon index represents the richness and evenness of the bacterial community. The positive control of probiotics (Probio) showed no significant difference from NCON in changing α-diversity of ileal microbiota. There was no statistically significant difference in α-diversity among the five groups in cecum and colon (Supplementary Figure 2). Principal coordinates analysis (PCoA) based on the ASVs showed a clear difference in community composition among groups (NCON, AMP, Probio, MMT, and CHC) in the ileal, cecal, and colonic microbiota of broilers (Figure 3).

At the phylum level, Firmicutes, Bacteroidetes, and Proteobacteria were predominant phyla in the ileal, cecal, and colonic microbiota in all five groups (Figure 4). In the ileum and cecum, the abundances of Firmicutes in CHC treatment and three positive controls were higher than that of NCON, while the abundances of Bacteroidetes in these groups were lower than that of NCON. In the colon, the abundances of Firmicutes and Bacteroidetes were not differently distributed.

A heatmap exhibited similarities and differences in bacterial communities at the genus level (Figure 5). Lactobacillus, Romboutsia, and Candidatus_Arthromitus were the dominant genera in the ileum, whereas Lactobacillus, Romboutsia, and Barnesiella were the dominant genera in the cecum and colon. The abundances of Lactobacillales in the CHC treatment were lower (p < 0.05) than those of the AMP and MMT treatments in the ileum. The abundances of Romboutsia in the ileum in the CHC group were higher (p < 0.05) than those of the three positive controls. The abundances of Candidatus_Arthromitus in the ileum in CHC-fed broilers were lower (p < 0.05) than those of Probio treatment and higher (p < 0.05) than those of NCON treatment. In the ileum, the abundances of Alistipes and Faecalibacterium in CHC treatment were similar to the three positive controls, which were all lower than that of NCON. CHC treatment had lower abundances of Alistipes in the ileum of broilers (p < 0.05) relative to those in the AMP and Probio treatments. The absorption effect of CHC on Candidatus_Arthromitus, Alistipes, and Faecalibacterium was similar to that of MMT. In the cecum and colon, there were no significant differences in the abundances of these dominant genera across treatments.

At the genus level, in the ileum compared with NCON, CHC treatment had significantly lower (p < 0.05) abundances of the pathogenic bacteria Alistipes (Figure 5). The abundance of pathogenic bacteria (Alistipes) in CHC treatment was lower (p < 0.05) than those of the AMP and Probio treatments. Compared with MMT, CHC treatment had a higher (p < 0.05) abundance of Romboutsia compared to NCON, AMP, Probio, and MMT. In addition, compared with NCON, CHC had a higher (p < 0.05) abundance of Candidatus_Arthromitus. However, CHC had no effect on the abundance of Lactobacillus which were lactic acid-producing bacteria that were beneficial for gut health. Compared to NCON, Probio treatment had lower (p < 0.05) abundances of Alistipes; and higher (p < 0.05) abundance of Candidatus_Arthromitus.

Differences in the composition of intestinal microbiota at the genus level were further explored by the LEfSe analysis (Figure 6). In the ileum, we found that 35 bacterial genera including gram-negative bacteria (such as Barnesiella, Alistipes, unclassified_f__Lachnospiraceae) and gram-positive bacteria (such as UCG-005, [Ruminococcus]_torques_group, Faecalibacterium) were enriched in the NCON group. The AMP group was characterized by a higher relative abundance of gram-negative bacteria (Christensenellaceae_R-7_group). In addition, gram-positive bacteria (Romboutsia), gram-negative bacteria (Candidatus_Arthromitus), and gram-positive bacteria (Lactobacillus and Bacillus) were enriched in CHC, Probio, and MMT groups, respectively.

In the cecum, gram-negative bacteria (Alistipes, Barnesiella, norank_f__Desulfovibrionaceae, Parasutterella, Butyricimonas, Intestinimonas, and norank_f__Oscillospiraceae) and gram-positive bacteria (Colidextribacter and Lactobacillus) in the NCON group had the highest richness among the five treatment groups. We also found gram-positive bacteria (Blautia), gram-negative bacteria (Barnesiellaceae, unclassified_f__Lachnospiraceae, Bilophila), and not classified bacteria (CHKCI001, norank_f__norank_o__RF39, and Marvinbryantia) were enriched in the AMP group. Furthermore, gram-positive bacteria (Romboutsia and Turicibacter) were particularly abundant in response to CHC, while only Lactobacillus (gram-positive bacteria) was enriched in the Probio group. The distribution of gram-negative bacteria (Christensenellaceae_R-7_group), gram-positive bacteria (norank_f__Erysipelotrichaceae), and Family_XIII_AD3011_group in group MMT were richer than those in other groups.

In the colon, the NCON group showed enrichment of gram-negative bacteria (Alistipes, Parasutterella, and Anaerotruncus) and gram-positive bacteria (Butyricicoccus, Adlercreutzia, and Colidextribacter). Furthermore, the CHC group was characterized by a higher relative abundance of gram-positive bacteria (norank_f__Clostridiales_vadinBB60_group, Ruminococcaceae_UCG-010, and Senegalimassilia) and not classified bacteria (CHKCI002) than that in other groups.

A Spearman correlation analysis was carried out to determine if any relationship existed among serum inflammatory traits, serum immunological markers, and intestinal microbiota (Figure 7).

In the ileum, the levels of serum IgA were negatively associated with Subdoligranulum and Candidatus_Arthromitus (p < 0.05). In addition, the levels of serum IL-1β and TNF-α were positively associated with the abundance of Erysipelatoclostridium, while negatively associated with the abundance of Enterococcus and Macrococcus. Levels of ileum mucosa SIgA and IL-1β were strongly associated with a series of bacteria (e.g., Unclassified_f__Lachnospiraceae, Barnesiella, Alistipes, Faecalibacterium, and Ruminococcus_torques_group) which indicated an active immune response induced by ileal bacteria.

In the cecum, the immune globulins (higher levels of IgA, IgG, and IgM indicating greater immune functions) exhibited a positive association with the abundance of Lachnoclostridium, NK4A214_group, norank_f__Eubacterium_coprostanoligenes_group, Christensenellaceae_R-7_group, Subdoligranulum, Ruminococcaceae_UCG-005, Streptococcus, unclassified_f__Lachnospiraceae, Blautia, and Barnesiella. In addition, correlation analysis revealed that norank_f__Erysipelotrichaceae, norank_f__Oscillospiraceae, norank_f__Ruminococcaceae, unclassified_f__Oscillospiraceae, and Alistipes had a strong positive association with the level of inflammatory factors (IL-1β and TNF-α) in the serum, while the bacteria Romboutsia and Lactobacillus had a strong negative association with the serum level of inflammatory factors (IL-1β and TNF-α).

In the colon, the abundance of norank_f__norank_o__RF39, Sellimonas, Christensenellaceae_R-7_group, Subdoligranulum, Blautia, and Turicibacter were positively associated with immune globulins (IgA and IgM), while the abundance of Shuttleworthia, norank_f__Clostridiales_vadinBB60_group, and Bacteroides were negatively associated with the levels of IgG. Lower levels of IL-1β and IFN-γ displayed a negative association with the abundance of Butyricicoccus and Alistipes.

To better understand the microbiota dynamic changes in broilers, potential functions of microbiota in the ileum, cecum, and colon were predicted by the PICRUSt algorithm (Supplementary Table 1). In the ileum, the functions related to energy production and conversion; transport and metabolism of amino acid, lipid and inorganic ion transport and metabolism; transcription, replication, recombination, and repair; secondary metabolites biosynthesis, transport, and catabolism; and signal transduction mechanisms in CHC group showed an increase (p < 0.05) compared with MMT group. There was no difference in the functions in the cecum and colon among treatment groups (p > 0.05).