This study was carried out in accordance with the recommendations of the ethical committee of the University of Parma. The protocol was approved by the “Comitato di Etica Università degli Studi di Parma”, Italy. All animal procedures were performed according to national guidelines (Decreto legislativo 26/2014) on the protection of animals used for scientific purposes.

Twenty adult English Setter dogs, reared in the same kennel, were selected to be homogeneous with reference to age (mean age ± SD: 3.50 ± 1.9 years), body weight (mean weight ± SD: 18.83 ± 2.96 kg) and gender (10 males, 10 not pregnant females). Based on age, weight, and sex animals were equally divided into two groups (10 dogs group^-1), individually penned with a rest area inside (2.70 m× 1.40 m) and a paddock outside (4.50 m× 1.40 m). Animals were free of any clinical symptoms indicating gastrointestinal disease and they did not receive medications that are expected to alter the gut microbiota such as antibiotics. Dogs were wormed one month before the start of study. The characteristics of the groups are reported in Table 1. During a period of 28 days, both groups received a diet, based on raw poultry meat (25% crude protein, 24% ether extract, 5% ash, 2% crude fiber, and 18.4 MJ kg^-1 ME, on dry matter). The individual ration, administered at about 25 g dry matter kg^-1 of body weight^0.75, once a day, was supplemented (group Tr) or not (group NTr) with CZ powder at the dose of 5 g day^-1. For each dog, zeolitite was weighed and added to the ration at each meal. Free access to water was provided. During the study, all dogs were daily subjected to an aerobic physical activity characterized by gallop for 20 min, according to the trainer’s practices. Training was performed in two outdoor next areas, at a mean temperature and relative humidity of 24 ± 3°C and of 67 ± 10%, respectively. Inside each group, five pairs of dogs were identified and each of them assigned alternatively to one or to the other of the training areas.

The powder of CZ, was obtained after sterilization at 200°C for 20 min (Chabasite 70^® Verdi S.p.A, Italy). The total zeolitic content was 70 ± 5%, of which 65 ± 3% due to chabazite (Na0.14K1.03Ca1.00Mg0.17) [Al3.46Si8.53O24] × 9.7H₂O and 5 ± 3% to phillipsite (Na0.9Ca0.5K0.6) [Si5.2A12.8O16] × 6H₂O. No traces of clinoptilolite were found. The composition of zeolitic powder was determined by Rietveld-RIR method (Gualtieri, 2000). The cation-exchange capacity and bulk density in relation to particles size were 2.2 ± 0.1 mEq g^-1 and 0.70-0.90 g (cm³)^-1, respectively (Gualtieri, 2000; Cresswell and Hamilton, 2002). Water retention in relation to particles size was about 30-40% (w/w). The granulometry of the powder was less than 100 μm.

Feces consistency was scored using a scale of 1 (hard) to 5 (watery) (Grieshop et al., 2002) at days 0 (Time point 0, T0), 16 (Time point 1, T1) from the beginning of the dietary treatment, and at the end of experimental period (day 29, Time point 2, T2). During the same days, individual fecal samples were collected directly from the rectum, using a sterile glove lubricated with water. The feces were placed in sterile polyethylene bags, immediately transported to the laboratory on ice packs and frozen at -20°C until analysis.

Upon arrival at the laboratory, individual fecal samples were aliquoted and combined with other individual samples from the same treatment to form pooled samples. In fact, in animal health it has been shown recently that pooling stool samples allows a rapid assessment of infection intensity and drug efficacy (Mekonnen et al., 2013). Each individual dog sample was equally represented in the respective pooled sample. DNA was extracted from pooled fecal samples using the QIAamp DNA Stool Mini kit following the manufacturer’s instructions (Qiagen Ltd., Strasse, Germany).

Partial 16S rRNA gene sequences were amplified from extracted DNA using primer pair Probio_Uni and /Probio_Rev, which target the V3 region of the 16S rRNA gene sequence, as previously reported (Milani et al., 2013). Partial ITS sequences were amplified from extracted DNA using the primer pair Probio-bif_Uni/Probiobif_Rev as described by Milani et al. (2014b). The PCR conditions used were 5 min at 95°C and 35 cycles of 30 s at 94°C, 30 s at 55°C, and 90 s at 72°C, followed by 10 min at 72°C. Amplification was carried out using a Veriti Thermocycler (Applied Bio-systems).

16S rRNA gene and ITS sequencing were performed using a MiSeq (Illumina) according to the protocols previously published (Milani et al., 2013, 2014b).

The achieved individual sequence reads were filtered by the Illumina software to remove low quality and polyclonal sequences. All Illumina quality-approved, trimmed, and filtered data were exported as.fastq files. The.fastq files were processed using a custom script based on the QIIME software suite (Caporaso et al., 2010). Paired-end reads pairs were assembled to reconstruct the complete Probio_Uni/Probio_Rev amplicons. Quality control retained sequences with a length between 140 and 400 bp and mean sequence quality score >20 while sequences with homopolymers >7 bp and mismatched primers were omitted. In order to calculate downstream diversity measures (alpha diversity indices, Unifrac analysis), 16S rRNA Operational Taxonomic Units (OTUs) were defined at ≥97 % sequence homology using uclust (Edgar, 2010) and OTUs with less than 10 sequences were filtered. All reads were classified to the lowest possible taxonomic rank using QIIME (Caporaso et al., 2010) and a reference dataset from the SILVA database (Quast et al., 2013). Biodiversity of the samples (alpha-diversity) were calculated with Chao1 index.

For ITS-based microbiota analysis Fastq files obtained by sequencing of the ITS amplicons were analyzed using a custom script, named bif_ITS_analysis.sh script ¹. This script requires QIIME (Caporaso et al., 2010) to be installed (or works in a QIIME virtual machine) and accepts.bam or.fastq input files containing sequencing reads. Input data were processed as previously described (Milani et al., 2014b).

The homogenates fecal specimens were serially diluted with both half-strength Wilkins-Chalgren Anaerobe Broth (WCAB) and Buffered Peptone Water (ThermoScientific-Oxoid, UK). Dilutions in duplicate were plated on MacConkey agar (Merck, Germany) for Enterobacteriaceae, Perfringens agar Base (OPSP) (Oxoid, UK) for C. perfringens, vancomycin and bromocresol green (LAMVAB) agar (Hartemink and Rombouts, 1999) for lactobacilli, and Azide maltose agar (Biolife, Italy) for enterococci counts. MacConkey agar and Azide maltose agar plates were incubated aerobically at 37°C for 24 and 48 h, respectively. Other media were incubated anaerobically at 37°C for 48-72 h. The taxonomy of colonies isolated random on selective media were determined at genus and species level by API System (Bio-Merieux, Italy) to verify the reliability of the media utilized.

The ability of CZ to bind to enteropathogens bacteria was evaluated in pooled feces using two reference strains, i.e., E. coli ATCC 35218 and C. perfringens ATCC 13124. Strains were grown in Mueller-Hinton Broth (Difco, MI, USA) at 37°C for 24 h, then transferred to 10 ml of broth and grown for another 8 h to reach the final exponential phase.

Adsorptive capacity of CZ was evaluated, measuring spectrophotometrically the OD of the samples (An and Friedman, 1997). Twenty-five grams of pooled feces obtained by NTr groups and collected on days 0, 16, and 29 were placed, in triplicate, into flasks containing 225 ml of Buffered Sodium Chloride-Peptone Solution pH 7.0 (Oxoid, UK). CZ was added in different quantities (0, 0.25, 0.5, 1 g). Lastly, C. perfringens ATCC 13124 or E. coli ATCC 35218 strains were added to medium and incubated at 37°C. At 0, 2, 4, 6, and 24 h, 150 μl of the suspension were transferred into a microtiter plate in four replicates and the absorbance was immediately evaluated (VICTOR3, 1420 multilabel counter, PerkinElmer, Italy) at 620 nm.

Data for fecal score and fecal bacteria counts were checked for normality and then analyzed by ANOVA using the GLM procedure in SAS (Version 9.4, SAS Institute Inc., USA). The mixed model included the fixed effects of group (two levels), of sampling time (three levels), the interaction between group and sampling time and the random effect of animal. Values of colony forming units (CFU) have been expressed as log₁₀ g^-1 of feces.

Statistical significance was reached for P ≤ 0.05 as a P-value >0.05 and ≤0.10 was considered as a trend.

Raw sequences of 16S rRNA gene profiling are accessible through SRA study accession number SRP075756. Raw sequences of ITS profiling are accessible through SRA study accession number SRP080281.

Pooled fecal samples from CZ treated (Tr) and no-treated (NTr) dogs were obtained in order to assess the microbiota composition based on 16S rRNA-sequencing analysis as described previously (Milani et al., 2013). The sequencing produced a total of 589784 reads with an average of 98297 reads per sample (Supplementary Table S1).

Assessment of rarefaction curves, based on the Chao1 biodiversity indexes calculated for 10 subsampling of sequenced read pools, indicated that both curves tend to reach a plateau. Therefore, in all cases the obtained sequencing data was deemed adequate to cover the vast majority of the biodiversity contained within the samples (Figure 1A). Moreover, the two curves did not show relevant differences, thus indicating that the analyzed samples have similar biodiversity.

During the study, diarrhea events were not observed in the CZ treated dogs. CZ did not affect the palatability of the feed, which was eaten completely within 30 min after dosing. Fecal scores were not affected by the factors in the statistical model (P > 0.05; Table 2). Fecal microbiota differences were observed in relation to group and sampling time (P < 0.05).

Inspection of predicted taxonomic profiles at phylum level for all NTr samples (T0, NTrT1, NTrT2) highlighted that Firmicutes (average 51.15% ± 11.46%) represented the dominant phylum of the cecal community in dogs, outnumbering the Proteobacteria (average 27.06% ± 15.75%), the Fusobacteria (average 8.54% ± 3.46%) and the Bacteroidetes (average 5.49% ± 2.62%) phyla (Figures 1B,C).

The comparison of the average relative abundance of NTr and Tr samples at time point T1 revealed a decrease of members of the Enterobacteriaceae family (-66.99 %), such as Escherichia (-67.16%), Klebsiella (-94.75%), and Hafnia (-74.87%), in Tr samples (Figure 1D) and an increase of Lactobacillus (205.16%) and Bifidobacterium (75.35%) in CZ treated animals (Figure 1D). At time point T2 in CZ treated animals (Figure 1D), the decrease in Enterobacteriaceae (-15.34%), includes a reduction of the genera Hafnia (-67.85 %), Klebsiella (-77.18%), and Enterobacter (-84.69%), along with an increase in relative abundance of Lactobacillus (861.64%) and Bifidobacterium (157.73%) (Figure 1D).

Notably, data achieved with culture-dependent approaches largely confirmed results obtained with 16S rRNA microbial profiling. In fact, Lactobacillus ssp. and Enterococcus ssp. counts were higher, while Enterobacteriaceae counts were lower in Tr than in NTr group (P < 0.05). Lactobacillus ssp. counts tended to be higher in Tr than in NTr group on day 16 (T1; 7.43 vs. 7.24; P < 0.10) and were higher on day 29 (T2; 8.18 vs. 7.25; P < 0.05). An increase of Enterococcus ssp. concentration (8.10 vs. 7.27) and a decrease of Enterobacteriaceae counts (6.24 vs. 7.14) were found in Tr compared to NTr group on day 29 (T2; P < 0.05). Besides, no change on the fecal C. perfringens counts was reported in relation both to the sampling time and to the treatment (P > 0.05).

Focusing on the contribution of bifidobacteria to the overall dog microbiota, it is worth noticing that at day 0 (T0) and in NTr animals at days 16 and 29 (T1 and T2, respectively), this genus represents 2.32% ± 1.88% of the gut microbiota of hunting dogs. In treated animals (Tr) the presence of the Bifidobacterium genus showed an increase of about 157.73% compared with Tr animals at T2, after 29 days of CZ diet (Figure 1D).

In order to precisely catalog the effects on the bifidobacterial population of dogs after CZ treatment, we performed an ITS profiling of bifidobacterial communities in stool samples of Tr and NTr dogs.

Quality filtering of the sequenced ITS amplicons produced an average of 52468 high-quality and full-length reads per sample (Supplementary Table S2) that were taxonomically attributed reaching the minimal taxonomical rank of species.

The composition of bifidobacterial populations of dogs included in the analysis showed the presence of peculiar species, such as B. pseudolongum (average of 60.70% ± 24.61% for T0 and NTr samples) and B. animalis (average of 7.84% ± 7.50% in T0 and NTr dogs) (Figure 2A), which have been previously described to be typical of the animal GIT (Milani et al., 2014a) and especially of the dog GIT (Gavini et al., 2006). Notably, other bifidobacterial species previously described to be typical of the human gut such B. catenulatum and B. bifidum were detected at a lower extend (Figure 2B).

Furthermore, in untreated animal samples (NTr1 and NTr2), ITS analysis revealed the occurrence of B. longum, B. gallinarum, and B. pseudocatenulatum species, typical human bifidobacterial taxa (Milani et al., 2014a). One possible explanation of the presence of these species in the canine gut microbiota could be a bacterial transmission between animals and trainers as previously reported in literature (Song et al., 2013). However, further investigations will be needed. Notably, a large proportion of the OTUs defined as ‘unclassified’ in T0 dog samples (Figures 2A,B) clusters separately from any current known bifidobacterial taxon, thus putatively representing novel Bifidobacterium taxa. These putative new unclassified bifidobacterial species represents the second most present bifidobacterial taxa in the dog microbiota, in both Tr and NTr animals (Figures 2A,B).

As reported above, at time point T2 in CZ treated animals, there was an increase in relative abundance of the genus Bifidobacterium (Figure 1D). ITS profiling experiments revealed an increase of 803.14 and 54.76% of B. animalis species and B. pseudolongum species, respectively, after the addition of CZ. Moreover, a slight increase was detected also for the here identified putative new bifidobacterial taxa in TrT2 (12.45 %) compared to NTrT2 (Figure 2C).

CZ showed an adsorptive capacity toward E. coli (Figure 3A) and C. perfringens (Figure 3B) strains in a dose- and time-dependent trial. Differences among CZ levels were registered for both strains after 2, 4, 6, and 24 h of incubation (P < 0.05). In particular, higher adsorptive capability against E. coli strain, was observed when CZ was added to the medium at a dose of 0.5 and 1 g rather than of 0 and 0.25 g (P < 0.05). When CZ was added at a dose of 1 g, negative values of OD starting from 0 h of incubation was observed for E. coli. During the first six hours of incubation, the adsorptive effect of CZ on C. perfringens strain was higher for levels of 0.5 and 1 g, than of 0 and 0.25 g (P < 0.05; Figure 3).