Age and Giardia intestinalis Infection Impact Canine Gut Microbiota

Giardia intestinalis is a flagellated protozoan responsible for giardiosis (also called giardiasis in humans), the most prevalent and widespread parasitic infection in humans and mammals worldwide. The intestinal microbiota is highly diverse and any alteration in its composition may impact on the health of the host. While studies on the mouse model of giardiosis described the role of the gut microbiota in host susceptibility to infection by the parasite, little is known about the gut microbiota during natural infections in dogs and particularly in puppies. In this study, we monitored naturally G. intestinalis-infected puppies for 3 months and quantified cyst excretion every 2 weeks. All puppies remained subclinically infected during the sampling period as confirmed by fecal examination. In parallel, we performed 16S Illumina sequencing of fecal samples from the different time points to assess the impact of G. intestinalis infection on gut microbiota development of the puppies, as well as gut health markers of immunity such as fecal IgA and calprotectin. Sequencing results revealed that the canine fecal microbiota of Giardia-infected puppies becomes more complex and less diverse with increasing age. In addition, significant differences in the structure of the microbiota were observed between puppies with high and low Giardia cyst excretion. Chronic subclinical G. intestinalis infection appears to be associated with some detrimental structural changes in the gut microbiota. G. intestinalis-associated dysbiosis is characterized by an enrichment of facultative anaerobic, mucus-degrading, pro-inflammatory species and opportunistic pathogens, as well as a reduction of Lactobacillus johnsonii at specific time points. Calprotectin levels increased with age, suggesting the establishment of chronic low-grade inflammation in puppies. Further work is needed to demonstrate whether these alterations in the canine gut microbiota could lead to a dysbiosis-related disease, such as irritable bowel syndrome (IBS) or inflammatory bowel disease (IBD).


Introduction
Giardia intestinalis (syn. Giardia duodenalis, Giardia lamblia) is a flagellate protozoan responsible for giardiosis (or giardiasis in humans), one of the most prevalent and widespread parasitic infection in humans and mammals worldwide [1,2]. Clinical manifestations of acute giardiosis include intestinal malabsorption, diarrhea, abdominal pain, and weight loss [3]. However, G. intestinalis infection is often asymptomatic and chronic infections are common [4][5][6]. G. intestinalis is classified into eight assemblages that have different host specificities. Assemblages A and B are found in humans and other mammals, C and D in canids, E in ungulates animals, F in cats, G in rodents, and H in pinnipeds. Assemblages A and B are considered zoonotic [7]. The prevalence of G. intestinalis in dogs ranges from 4 to 36.5% in Europe [4,8,9]. The prevalence is higher in young animals (between 9 and 14 weeks of age, 62.4%) compared to puppies between 5 and 8 weeks of age (30.4-32.6%) [10][11][12]. Animals living in communities, for example, in shelters or breeding kennels, are more frequently positive than privately owned pets, due to the high density of animals in these facilities [4,9,13].
The gut microbiota has great diversity and it is well known that any alteration can result in an impact on health [14]. The human and canine gut microbiota are very similar in composition and share similar functions due to a long period of domestication and co-evolution of dogs with humans [15,16]. Similar to humans, the canine microbiota is established with aerobic species at birth and evolves very rapidly during the first weeks of age in favor of anaerobes [17]. In adult dogs, the microbiota is stable over the years. Similar to humans, the composition of the canine gut microbiota is affected by several factors such as diet [16], body weight [18], geographic location [19], genetics [20], age [21,22], and infections [17,23].
Nowadays, it has been known that the gut microbiota can impact the ability of G. intestinalis to colonize the host. Singer and Nash were the first to observe that mice with identical genetic background but from two different breedings showed different susceptibility to G. intestinalis infection [24]. When mice were housed together, resistance to infection was readily transferred to normally susceptible mice. Antibiotic treatment completely reversed the inability of G. intestinalis to infect these mice, demonstrating that the differences in susceptibility to infection are due to changes in the gut microbiota present in these mice [24]. Another study demonstrated that conventional mice were more susceptible to G. intestinalis infection than germ-free and gnotobiotic mice, confirming the stimulatory activity of the gut microbiota on G. intestinalis pathogenicity [25]. More recently, the nematode model Caenorhabditis elegans was used to study the effects of G. intestinalis on the intestinal microbiota [26]. Exposure of human commensal bacteria to G. intestinalis prior to administration in C. elegans induced nematode death, whereas the same bacteria not exposed to the parasite had no effect. In addition, the authors demonstrated that G. intestinalis induced the expression of virulence factors in Escherichia coli. These results suggest that G. intestinalis is capable of modifying the host's commensal gut microbiota and possibly turn on species to opportunistic pathogens [26]. G. intestinalis also induces virulence of gut microbiota towards human intestinal epithelial cells in vitro [27].
Many studies on G. intestinalis and its impact on the gut microbiota involve experimental infections of laboratory animals. Although these studies can be useful in elucidating mechanisms of action, they generally involve laboratory-adapted strains, high infectious doses and specialized animal husbandry practices, that can affect host immunity and microbiota composition and thus lead to misinterpretations. The impact of natural G. intestinalis infection on the gut microbiota is poorly described. Berry et al. sequenced the gut microbiota of dogs naturally infected with one or more eukaryotic parasites and found that parasite infections are associated with significant perturbations of the microbiome and that G. intestinalis is associated with the greatest changes in the canine gut microbiota [5], confirming the observations of Šlapeta et al. on asymptomatic adult dogs [28].
In the present study, puppies naturally infected with G. intestinalis were monitored for 3 months and 16S Illumina sequencing of fecal samples from different time points was performed to assess the impact of G. intestinalis infection on the development of the puppies' gut microbiota, as well as gut health markers such as fecal IgA and calprotectin.

Dogs
Beagle puppies from a French breeding kennel (n = 18; 7 males and 11 females) aged 8 to 10 weeks (mean 8.9 weeks) were included in the study. These puppies were part of a two-arm randomized study in which the dogs presented in this study served as controls. No antibiotics were administered during the experimental protocol. All animals were Microorganisms 2021, 9, 1862 3 of 21 housed in a controlled environment of 4 dogs per box. On days of fecal collection, dogs were single housed for a maximum of 4 h to allow fecal sample collection. Body weight and fecal score were monitored weekly for the full study duration. Fecal samples were collected after spontaneous defecation at 7 time points: 8 days before the start of the study (D−8) and every 14 days after the start (D14, D28, D42, D56, D70, and D85). Fecal samples were stored at 4 and −80 • C immediately after collection until further processing.

Giardia Cysts Count and Genotyping
Giardia cysts were detected and counted by direct immunofluorescence assays (IFA) with MeriFluor ® Cryptosporidium/Giardia (Meridian Bioscience, Milan, Italy). Briefly, 1 g aliquots of fresh feces were diluted in 10 mL of distilled water and filtered through a sieve (mesh size: 63 µm), then IFA assays were performed in duplicate using 20 µL of this solution. Whole slides were examined under a fluorescent microscope with 20× objective (Leica Microsystems, Rueil-Malmaison, France). For genotyping, 7 mL of fecal suspension were added to 3 mL of diethylic ether, homogenized vigorously and centrifuged for 5 min at 2000× g at 20 • C. Pellets were washed twice with Phosphate buffered saline (PBS) and resuspended in 2 mL PBS. Samples were subjected to genomic DNA extraction using the QIAamp Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. To disrupt parasite cyst walls, an initial step of 8 freeze-thaw cycles (freezing in liquid nitrogen for 5 min and thawing at 95 • C for 5 min) was incorporated into the protocol. Semi-nested PCR were performed to amplify glutamate dehydrogenase (gdh), triosephosphate isomerase (tpi), β-giardin, and SSU rRNA genes, as previously described [7,29,30]. The sequence of the primers is shown in Table 1. PCR products were visualized on 2% agarose gels stained with ethidium bromide (0.5 µg/mL) and amplicons were sequenced in both directions.

Quantification of Fecal IgA and Calprotectin
For IgA quantification, frozen fecal samples were thawed and 0.2 g aliquots of feces were diluted in 2 mL PBS. After homogenization with 0.10 to 0.25 mm diameter glass beads (Fisher Scientific) using a Precellys Evolution homogenizer (Bertin Technologies, Montignyle-Bretonneux, France) for 40 s at 6400 rpm, suspensions were centrifuged for 5 min at 13,000× g at 4 • C. Supernatants were collected and stored frozen at −80 • C prior to ELISA analyses following the manufacturer's instructions (Canine IgA, FineTest, Wuhan, China). For calprotectin quantification, fecal extracts were prepared as previously described [36]. Briefly, frozen fecal samples were thawed and 1.3 g aliquots of feces were diluted 1:5 in fecal extraction buffer (20 mM CH 3 CO 2 Na and 3 mM CaCl 2 [pH 7.6]) containing a protease inhibitor cocktail (1 tablet/25 mL, Roche, Basel, Switzerland). After homogenization by vigorous shaking for 30 min at room temperature, the suspensions were centrifuged for 20 min at 2100× g at 4 • C. The supernatants were collected and centrifuged for 30 min at 10,600× g at 4 • C. The final supernatants (fecal extracts) were stored frozen at −80 • C prior to ELISA analyses following the manufacturer's instructions (Canine Calprotectin, MyBiosource, San Diego, CA, USA).

DNA Extraction
A modified version of the protocol by Godon et al. [37] was used for DNA extraction. The tubes were centrifuged for 5 min at 14,000× g rpm at 4 • C and the supernatant was transferred to a deepwell. The plate was transferred on the nucleic acid workstation Chemagic STAR (Hamilton, Perkin Elmer, Waltham, MA, USA) and the extraction protocol was performed with Chemagic STAR DNA BTS kit (Perkin Elmer, Waltham, MA, USA) according to the manufacturer's instructions.

Primer Design and Library Preparation
The V3-V4 hypervariable regions of the 16S rDNA gene were amplified from DNA extracts during the first PCR step using universal primers PCR1F_343 and PCR1_R784 (Table 1). PCR was carried out using 2 U of a DNA-free Taq DNA Polymerase and 1× Taq DNA polymerase buffer (MTP Taq DNA Polymerase, Sigma-Aldrich, St. Louis, MI, USA). The buffer was supplemented with 10 nmol of dNTP mixture (Sigma-Aldrich, St. Louis, MI, USA), 15 nmol of each primer (Eurofins, Luxembourg), and Nuclease-free water (Qiagen, Hilden, Germany) in a final volume of 50 µL. The PCR reaction was carried out in a T100 Thermal cycler (Biorad, USA) as follows: An initial denaturation step (94 • C for 10 min) was followed by 30 cycles of amplification (94 • C for 1 min, 68 • C for 1 min, and 72 • C for 1 min) and a final elongation step at 72 • C for 10 min. Amplicons were then purified using magnetic beads CleanPCR (Clean NA, GC Biotech B.V., Waddinxveen, The Netherlands) in a 96-well format. Sample multiplexing was performed by adding tailor-made 6 bp unique indexes during the second PCR step at the same time as the second part of the P5F/P7R adapters to obtain primers PCR2_P5F and PCR2_P7R (Table 1). This second PCR step was performed on 50-200 ng of purified amplicons from the first PCR using 2.5 U of a DNA free Taq DNA Polymerase and 1× Taq DNA polymerase buffer. The buffer was completed with 10 nmol of dNTP mixture (Sigma-Aldrich, St. Louis, MI, USA), 25 nmol of each primer (Eurofins, Luxembourg), and Nuclease-free water (Qiagen, Hilden, Germany) up to a final volume of 50 µL. The PCR reaction was carried out on a T100 Thermal cycler with an initial denaturation step (94 • C for 10 min), 12 cycles of amplification (94 • C for 1 min, 65 • C for 1 min, and 72 • C for 1 min) and a final elongation step at 72 • C for 10 min. Amplicons were purified as described for the first PCR reaction. All libraries were pooled with equal amounts in order to generate an equivalent number of raw reads for each library. DNA concentration of the pool was quantified on a Qubit Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). The pool at a final concentration between 5 and 20 nM was used for sequencing.

Illumina Sequencing
The library pool (see above) was denatured (NaOH 0.1 N) and diluted to 7 pM. Then, 15% PhiX Control v3 (Illumina, San Diego, CA, USA) was added to the pool as described in the Illumina procedure. Afterwards, 600 µL of this pool and PhiX mix were loaded onto the Illumina MiSeq cartridge according to the manufacturer's instructions using MiSeq Reagent Kit v3 (2 × 300 bp Paired-end reads, 15 Gb output). FastQ files were generated at the end of the run (MiSeq Reporter software, Illumina, San Diego, CA, USA). The run quality was checked internally using PhiX control and each paired-end sequence was assigned to its sample using the multiplexing index. The sequences were processed using FROGS [38]. The quality of the raw sequencing data was checked using FastQC and reads with a Phred quality score <30 were discarded. Chimeras and singletons were removed from the dataset. Quality control retained sequences with a length between 100 and 400 bp. Paired-end reads were merged using Vsearch. 16S rRNA Operational Taxonomic Units (OTUs) were assigned based on at least 99% sequence similarity to the lowest possible taxonomic rank against the SILVA Pintail 100-138 reference database. One hundred percent of the sequences were affiliated. Moreover, 20.8% of sequences were multi-affiliated at the species level and 4.74% of sequences were multi-affiliated at the genus level. Biodiversity of the samples (alpha diversity) was calculated with Chao1, Shannon, and Inverse Simpson indexes, while similarity between samples (beta diversity) was calculated with Jaccard and unweighted Unifrac distances.

Statistical Analyses
For statistical analyses, sequences were rarefied to a uniform depth of 57,698 sequences per sample to account for unequal sequencing depth between samples. Based on rarefaction curves, samples with less than 10,000 sequence reads were excluded from the analysis. Alpha diversity indices were compared using the Mann-Whitney test (GraphPad Prism software version 9.00). The principal coordinates analysis (PCoA) was performed with FROGS to visualize the clustering of samples according to various parameters, using Jaccard and unweighted Unifrac distances. Permutational multivariate analysis of variance (PERMANOVA) was used to test for significant differences in bacterial community composition across samples. All statistical tests were two-sided, and p < 0.05 was considered statistically significant.

G. intestinalis Infection Is Chronic in Puppies
Body weight and fecal score (performed using the Royal Canin Fecal Scoring Chart) were monitored weekly for 93 days and all puppies remained asymptomatic throughout the 93-day experiment (data not shown). G. intestinalis cysts were quantified in fecal samples at seven time points: 8 days before the start of the study (D−8) and every 14 days after the start until day 85 (D85). All animals remained infected at day 85 ( Figure 1), suggesting either that puppies were not able to clear the parasite or that they were re-infected.
Body weight and fecal score (performed using the Royal Canin Fecal Scoring Chart) were monitored weekly for 93 days and all puppies remained asymptomatic throughout the 93-day experiment (data not shown). G. intestinalis cysts were quantified in fecal samples at seven time points: 8 days before the start of the study (D−8) and every 14 days after the start until day 85 (D85). All animals remained infected at day 85 ( Figure 1), suggesting either that puppies were not able to clear the parasite or that they were re-infected. G. intestinalis genotyping may reveal different results depending on the genetic locus sequenced and multiple gene loci should be investigated to increase the specificity of the results [39]. Thus, we determined G. intestinalis assemblages in samples D−8 and D85 by sequencing gdh, tpi, β-giardin, and SSU rRNA genes ( Table 2). All animals were infected by dog-specific assemblages C and/or D. The majority of puppies have mixed infections between assemblages in both D−8 and D85. No zoonotic A or B assemblages were detected in this study.

Age Influences Fecal IgA and Calprotectin Levels
Further analyses focused on three time points: D−8, D42, and D85 to investigate the effects of age on IgA levels and calprotectin concentrations, as non-invasive markers of intestinal health. Fecal IgA levels were measured by ELISA. As shown in Figure 2A, fecal IgA concentrations were significantly lower in D42 and D85 samples compared to the D−8 samples (p = 0.0177 and p = 0.0009, respectively). Next, we investigated whether IgA G. intestinalis genotyping may reveal different results depending on the genetic locus sequenced and multiple gene loci should be investigated to increase the specificity of the results [39]. Thus, we determined G. intestinalis assemblages in samples D−8 and D85 by sequencing gdh, tpi, β-giardin, and SSU rRNA genes ( Table 2). All animals were infected by dog-specific assemblages C and/or D. The majority of puppies have mixed infections between assemblages in both D−8 and D85. No zoonotic A or B assemblages were detected in this study.

Age Influences Fecal IgA and Calprotectin Levels
Further analyses focused on three time points: D−8, D42, and D85 to investigate the effects of age on IgA levels and calprotectin concentrations, as non-invasive markers of intestinal health. Fecal IgA levels were measured by ELISA. As shown in Figure 2A, fecal IgA concentrations were significantly lower in D42 and D85 samples compared to the D−8 samples (p = 0.0177 and p = 0.0009, respectively). Next, we investigated whether IgA variations could be related to G. intestinalis cysts shedding. Fecal IgA concentrations were positively correlated with G. intestinalis cysts shedding in D85 samples (R 2 = 0.2539; p = 0.0330), whereas no significant correlation was observed in D−8 and D42 samples. Fecal calprotectin levels were also measured by ELISA. As shown in Figure 2B variations could be related to G. intestinalis cysts shedding. Fecal IgA concentrations were positively correlated with G. intestinalis cysts shedding in D85 samples (R 2 = 0.2539; p = 0.0330), whereas no significant correlation was observed in D−8 and D42 samples. Fecal calprotectin levels were also measured by ELISA. As shown in Figure 2B, a significant increase in fecal calprotectin was observed in D85 samples compared to D−8 samples (p = 0.0006). Fecal calprotectin concentrations correlate positively with G. intestinalis cysts shedding in D−8 samples (R 2 = 0.6215; p = 0.0002).

Age Influences the Richness and Diversity of the Canine Gut Microbiota
To characterize the evolution of the canine gut microbiota over time in naturally-infected puppies, we used the Illumina MiSeq technology to sequence the V3-V4 region of 16S rDNA obtained from a total of 54 fecal samples (18 puppies at three different time points). Sequencing generated a total of 5.90 million reads with an average of 109,300 reads per sample. The quality and chimera filtering produced a total of 4,475,733 reads with an average of 82,883 filtered reads per sample. Beta diversity was analyzed based on Jaccard and unweighted Unifrac distance matrices and represented by PCoA. The predicted PCoA representation of Jaccard and unweighted Unifrac beta diversity ( Figure 3) and sample clustering by Ward D2 ( Figure S1) revealed separate clusters for samples D−8, D42, and D85. The variable "collection time" explained 38.20% of the variance using Jaccard distance (PER-MANOVA, p < 0.0001) and 46.71% of the variance using Unifrac distance (PERMANOVA, p < 0.0001). PERMANOVA was also applied to assess the similarities of microbial communities as a function of gender, age, and box in both distance matrices. None of these other variables were significantly associated with microbiota composition (gender: p = 0.4453; age: p = 0.2073; box: p = 0.0535), which is consistent with previous studies [13,20].

Age Influences the Richness and Diversity of the Canine Gut Microbiota
To characterize the evolution of the canine gut microbiota over time in naturallyinfected puppies, we used the Illumina MiSeq technology to sequence the V3-V4 region of 16S rDNA obtained from a total of 54 fecal samples (18 puppies at three different time points). Sequencing generated a total of 5.90 million reads with an average of 109,300 reads per sample. The quality and chimera filtering produced a total of 4,475,733 reads with an average of 82,883 filtered reads per sample. Beta diversity was analyzed based on Jaccard and unweighted Unifrac distance matrices and represented by PCoA. The predicted PCoA representation of Jaccard and unweighted Unifrac beta diversity ( Figure 3) and sample clustering by Ward D2 ( Figure S1) revealed separate clusters for samples D−8, D42, and D85. The variable "collection time" explained 38.20% of the variance using Jaccard distance (PERMANOVA, p < 0.0001) and 46.71% of the variance using Unifrac distance (PERMANOVA, p < 0.0001). PERMANOVA was also applied to assess the similarities of microbial communities as a function of gender, age, and box in both distance matrices. None of these other variables were significantly associated with microbiota composition (gender: p = 0.4453; age: p = 0.2073; box: p = 0.0535), which is consistent with previous studies [13,20].
Rarefaction curves generated with these reads revealed that the sequencing depth appropriately covered bacterial diversity in all samples ( Figure 4A). Samples collected in D−8 had a lower level of complexity, with an average of 146 ± 43 OTUs, compared to D42 and D85 samples, which had an average of 218 ± 21 OTUs and 251 ± 11 OTUs, respectively. Observed OTUs, Chao1, Shannon, and Inverse Simpson richness were used to determine taxonomic diversity within samples ( Figure 4B). All diversity measures were significantly increased in D42 samples compared to D−8 samples (ANOVA p < 0.0001; p < 0.0001; p = 0.0002; p = 0.0085, respectively). In D85 samples, observed OTUs and Chao1 alpha diversity increase compared to D42 samples (ANOVA p < 0.0001), while no significant differences were observed for Shannon and Inverse Simpson measures, suggesting that microbiota richness tends to stabilize over time in an adult-like microbiota. Rarefaction curves generated with these reads revealed that the sequencing depth appropriately covered bacterial diversity in all samples ( Figure 4A). Samples collected in D−8 had a lower level of complexity, with an average of 146 ± 43 OTUs, compared to D42 and D85 samples, which had an average of 218 ± 21 OTUs and 251 ± 11 OTUs, respectively. Observed OTUs, Chao1, Shannon, and Inverse Simpson richness were used to determine taxonomic diversity within samples ( Figure 4B). All diversity measures were significantly increased in D42 samples compared to D−8 samples (ANOVA p < 0.0001; p < 0.0001; p = 0.0002; p = 0.0085, respectively). In D85 samples, observed OTUs and Chao1 alpha diversity increase compared to D42 samples (ANOVA p < 0.0001), while no significant differences were observed for Shannon and Inverse Simpson measures, suggesting that microbiota richness tends to stabilize over time in an adult-like microbiota. A total of eight phyla and 81 genera were identified ( Figure 5A). High inter-individual variability was observed in puppies at D−8 and decreased in older puppies at D42 and D85. Despite this high variability, core bacterial groups were identified. At the phylum level, the core microbiota was dominated by taxa belonging to Fusobacteriota (48.7 ± 18.9%), Firmicutes (24.1 ± 17.6%), Bacteriodota (18.8 ± 11.6%), and Proteobacteria (6.8 ± 5.6%). The minor abundant members are phyla Campylobacterota (1.3 ± 2.6%), Spirochaetota (0.1 ± 0.4%), Actinobacteriota (0.2 ± 0.2%), and Deferribacterota (2 × 10 −4 ± 2 × 10 −4 %). The abundance of the four dominant phyla varies greatly with time. Bacteroidota and Actinobacteriota significantly increase in D42 samples (p = 0.0006 and p = 0.0004, respectively) and in D85 samples (p = 0.0006 and p > 0.0001, respectively) compared to the D−8 samples. In contrast, Firmicutes abundance decreased in D85 samples compared to the D−8 samples (p = 0.0447). Fusobacteriota abundance is highly variable in D−8 samples, and tends to stabilise with increasing age in D42 and D85 samples. At the genus level, the abundance of most dominant genera was significantly enriched in D85 samples com-  A total of eight phyla and 81 genera were identified ( Figure 5A). High inter-individual variability was observed in puppies at D−8 and decreased in older puppies at D42 and D85. Despite this high variability, core bacterial groups were identified. At the phylum level, the core microbiota was dominated by taxa belonging to Fusobacteriota (48.7 ± 18.9%), Firmicutes (24.1 ± 17.6%), Bacteriodota (18.8 ± 11.6%), and Proteobacteria (6.8 ± 5.6%). The minor abundant members are phyla Campylobacterota (1.3 ± 2.6%), Spirochaetota (0.1 ± 0.4%), Actinobacteriota (0.2 ± 0.2%), and Deferribacterota (2 × 10 −4 ± 2 × 10 −4 %).  G. intestinalis affects the richness and diversity of the canine gut microbiota. To determine the impact of G. intestinalis on the composition of the fecal microbiota, we classified the samples into three categories according to their parasite load: Low, intermediate, and high. The beta diversity of these three categories was then analyzed using Jaccard and unweighted Unifrac distance matrices and was represented through PCoA ( Figure 6). In samples D−8 and D42, no difference in the structure of the overall microbiota as a function of parasitic load was observed (data not shown). However, in D85 samples, when the parasite load spectrum is broader, the samples cluster significantly with the parasite load (low <150,000 cysts/g of feces, intermediate between 150,000 and 750,000 cysts/g of feces, and high >750,000 cysts/g of feces). The parasite load explains 19.18% of the variance using the Jaccard distance (PERMANOVA p = 0.0404) and 23.18% of the variance using unweighted Unifrac distance (PERMANOVA p = 0.0152). G. intestinalis affects the richness and diversity of the canine gut microbiota. To determine the impact of G. intestinalis on the composition of the fecal microbiota, we classified the samples into three categories according to their parasite load: Low, intermediate, and high. The beta diversity of these three categories was then analyzed using Jaccard and unweighted Unifrac distance matrices and was represented through PCoA ( Figure 6). In samples D−8 and D42, no difference in the structure of the overall microbiota as a function of parasitic load was observed (data not shown). However, in D85 samples, when the parasite load spectrum is broader, the samples cluster significantly with the parasite load (low <150,000 cysts/g of feces, intermediate between 150,000 and 750,000 cysts/g of feces, and high >750,000 cysts/g of feces). The parasite load explains 19.18% of the variance using the Jaccard distance (PERMANOVA p = 0.0404) and 23.18% of the variance using unweighted Unifrac distance (PERMANOVA p = 0.0152).  In addition, a significant correlation was observed between alpha diversity and G. intestinalis cysts excretion at the three time points (Figure 7). Bacterial richness depends on the age of the animal. On D−8, Inverse Simpson richness correlates positively with cysts excretion in young puppies (R 2 = 0.2475; p = 0.0422). A positive trend was also noted in the observed, Chao1, and Shannon richness but without reaching the significance threshold of 0.05. On the contrary, when puppies are older and have a more mature microbiota, cysts excretion correlates with a significant decrease in the observed and Chao1 richness (R 2 = 0.2583; p = 0.0313 and R 2 = 0.3449; p = 0.0104, respectively). At D42, a slight decrease in the observed richness was noted but no significant richness difference, which is consistent with the intermediate stage.
Microorganisms 2021, 9, x FOR PEER REVIEW In addition, a significant correlation was observed between alpha diversity intestinalis cysts excretion at the three time points (Figure 7). Bacterial richness de on the age of the animal. On D−8, Inverse Simpson richness correlates positively wit excretion in young puppies (R 2 = 0.2475; p = 0.0422). A positive trend was also noted observed, Chao1, and Shannon richness but without reaching the significance thr of 0.05. On the contrary, when puppies are older and have a more mature micr cysts excretion correlates with a significant decrease in the observed and Chao1 ri (R 2 = 0.2583; p = 0.0313 and R 2 = 0.3449; p = 0.0104, respectively). At D42, a slight de in the observed richness was noted but no significant richness difference, which sistent with the intermediate stage.

Discussion
G. intestinalis is one of the most common enteric parasites worldwide and is remarkable for its ability to cause an array of clinical phenotypes, ranging from asymptomatic infections to severe acute diarrheal disease to chronic gastrointestinal disease in human

Discussion
G. intestinalis is one of the most common enteric parasites worldwide and is remarkable for its ability to cause an array of clinical phenotypes, ranging from asymptomatic infections to severe acute diarrheal disease to chronic gastrointestinal disease in human and animals. Acute giardiasis is auto-resolutive in 2 to 4 weeks in adults but could become chronic especially in young subjects, resulting in serious growth stunting and long-term health consequences [40,41]. In this study, 18 puppies (9 weeks of age) naturally infected with G. intestinalis remained infected at 21 weeks of age (D85) when no rescue treatment was administered, suggesting that G. intestinalis infection was not auto-resolutive in these animal hosts. In a murine model of infection, animals infected neonatally did not clear the parasite when they reached adulthood, although naive adult animals exposed to G. intestinalis were resistant to infection, suggesting that G. intestinalis manipulated the host during the postnatal period to favor its own persistence [42]. Genotyping results revealed that puppies were infected by assemblages C and D, and that multi-infections were frequent, confirming previous observations [12,43,44].
IgA is the predominant Ig subtype present in host secretions and protects mucosal surface from infectious agents. IgA plays a role in G. intestinalis clearance by targeting the parasite variant-specific surface proteins (VSPs) [45,46]. We demonstrated a decrease of fecal IgA concentration with age, which has been described elsewhere and is explained by the loss of IgA provided by maternal milk after weaning [11,47,48]. Fecal IgA correlated positively with G. intestinalis cysts shedding at day 85. No correlation was observed in earlier time points (Figure 2A). These results are consistent with a previous study in G. muris infected mice where secretions of total and parasite-specific IgA were observed after weaning. This study demonstrated that a protective intestinal immune response started to develop only after weaning, which could explain the prolonged course of infection in neonate mice in comparison to older mice [49]. However, in our study, the IgA immune response at D85 was not sufficient to eradicate the G. intestinalis infection in weaned puppies.
Calprotectin is a protein complex mostly abundant in neutrophiles, monocytes, and reactive macrophages at sites of inflammation. Since it reflects the phagocyte turnover in vivo, calprotectin has been used as a marker of inflammation that correlates with local and systemic signs of disease activity [50,51]. Previous studies have observed a reduction of fecal calprotectin levels with age in healthy puppies [52] and children [53]. In the current study, fecal calprotectin levels increased with age in puppies naturally infected with G. intestinalis. Fecal calprotectin concentration was associated with G. intestinalis cysts shedding in 9-week-old (D−8) puppies. No correlation was observed in later time points ( Figure 2B), in accordance with earlier findings in puppies [11] and asymptomatic children [53,54]. On the contrary, levels of fecal calprotectin are elevated in infected patients presenting clinical signs [55]. Here, all puppies remained asymptomatic during the course of the study, explaining the lack of correlation between cysts shedding and calprotectin levels in later time points.
The canine fecal microbiota was investigated during 12 weeks and revealed a clear gradual shift from a simple and micro-aerobic microbiota in 9-week-old puppies (D−8) to a more complex, homogeneous and anaerobic microbiota in 21-week-old puppies (D85) (Figure 4). An increased richness with age has already been described in dogs [17] and humans [56]. At 21 weeks of age (D85), puppies present an adult-like microbiota dominated by four phyla: Bacteroidota, Firmicutes, Fusobacteriota, and Proteobacteria ( Figure 5A). Although the predominant phyla identified are similar to previous observations in healthy dogs, proportions vary among studies [57][58][59]. The canine breed, environmental conditions, age, weight, infections, sequencing technologies, and sequencing depth are responsible for significant variations reported for gut microbiota composition [18][19][20][21][22][23]. Here, there were more pronounced differences in the composition of fecal microbiota among 9-weekold puppies (D−8) compared to 21-week-old puppies (D85). This high inter-individual variability could be explained by the fact that some puppies are already weaned for a couple of weeks, while others just finished the weaning process. After weaning, the composition of the gut microbiota in puppies seems to vary greatly ( Figure 5B,C), notably by increasing Bacteroides and Prevotella, which is already well described in dogs [17,60] and humans [61,62]. Other genera are enriched after weaning, including Fusobacterium, Alloprevotella, Peptostreptococcus, [Ruminoccocus] gnavus group, Parasuterella, Faecalibacterium, and Prevotellaceae Ga6A1 group. Many of these microbes are efficient degraders of dietary fibers and producers of short chain fatty acids (SCFAs). SCFAs display health properties that promote host intestinal health. For example, they regulate cell differentiation and proliferation, gut endocrine functions, immune response, and promote the intestinal epithelial barrier function. Finally, butyrate (one of the main SCFAs) is the primary source of energy for enterocytes [63,64]. The consumption of a solid diet, which in turn may modify the substrate availability and the physiological conditions of the gastrointestinal tract, was probably the main cause associated with the increased functional capacity for carbohydrates degradation. Surprisingly, genera Bifidobacterium, Lactobacillus, and Enterococcus increased with age, contrasting with findings in humans [65].
Most research of Giardia impact on the gut microbiota focused on animals experimentally infected with laboratory adapted strains and little is known about the canine microbiota during natural infections, in particular in asymptomatic animals. Some report described altered composition [5,28], while others showed modest or no effect in adult dogs [66]. These studies compared Giardia-positive to Giardia-negative individuals. In this study, all animals are infected by the parasite and the effects of the parasite load on the fecal microbiota composition were investigated instead. Results showed that the gut microbiota structure was dependent on the parasite load, as puppies with high cysts shedding displayed altered microbiota compared to puppies with a low cyst load ( Figure 6). Furthermore, G. intestinalis impact on microbiota richness seemed to be related to the age of the host. In 9-week-old puppies with immature gut microbiota (D−8), G. intestinalis was associated with increased bacterial richness, while in older animals with adult-like gut microbiota (D85), the parasite load correlated positively with decreased bacterial richness ( Figure 7). These results are similar to previous findings in asymptomatic children [67] and mice [68]. In either case, an alteration of microbial richness was found to be detrimental for the host. Increased richness in young subjects and lower microbial diversity in adults are often associated with a disease state [69,70].
G. intestinalis correlates positively with pro-inflammatory Prevotella, Suterella, Veillonellaceae, and an unclassified genus of Ruminococcaceae ( Figure 8). Prevotella is a major component of the gut microbiota of mammals and generally associated with gut health [71]. However, when compared with strict commensal bacteria, Prevotella exhibits increased inflammatory properties and might participate in disease by promoting chronic inflammation [72]. Moreover, perturbation of the gut microbiome by Prevotella enhances host susceptibility to mucosal inflammation [73]. Increased levels of Prevotella during G. intestinalis infection have already been reported in adult dogs and children with and without clinical signs [5,67,74,75]. This enrichment of pro-inflammatory species might be responsible for the increased level of fecal calprotectin observed in this study ( Figure 2B), while it normally tends to decrease with age in healthy puppies [52].
In addition, inflammation impairs the ability of the intestinal epithelium to perform β-oxidation leading to oxygen diffusion in the gut lumen thus promoting colonization with facultative anaerobic Proteobacteria and enteric pathogens, as observed in this study [76][77][78][79]. Indeed, G. intestinalis correlates positively with facultative anaerobic Peptostreptococcaceae, Veillonellaceae, and members of Proteobacteria including Suterella, Parasuterella, A. succiniproducens, and A. thomasii (Figure 8), as reported elsewhere [5,80]. Giardia itself could also be responsible for intestinal epithelium disruption due to its ability to induce enterocyte apoptosis and tight junction destabilization [81,82].
G. intestinalis infection is associated with an enrichment of potentially harmful bacteria ( Figure 8) such as Prevotella and A. succiniproducens that have been reported as opportunistic pathogens [83][84][85] and genera Megamonas and Catenibacterium, associated with abdominal pain and aggressive behavioral disorder in dogs [86,87].
Mucus layer plays an important role in protecting intestinal epithelial cells from commensal gut microbiota and pathogenic microorganisms. Mucin inhibited G. intestinalis adhesion to intestinal epithelial cells [88]. In this study, G. intestinalis positively correlates with Peptostreptococcaceae (Figure 8), that is significantly associated with reduction of MUC2 expression and downregulation of tight junction proteins expression [89]. G. intestinalis also positively correlates with Prevotella and Bacteroides (Figure 8), both genera containing members able to degrade host mucins [5,90,91]. These mucolytic bacteria could act in synergy with G. intestinalis cysteine proteases and degrade the protective mucus layer [92], promoting both parasite colonization and invasion of commensal bacteria [93][94][95]. These changes may enable easy contact of microbes with epithelial cells, stimulating an inflammatory response.
L. johnsonii correlates negatively with G. intestinalis in puppies at D85 (Figure 8). L. johnsonii is the most common species among the isolates from the pre-weaning dogs, suggesting that L. johnsonii might be a specific species of infant dogs [60]. L. johnsonii species is known for its probiotic properties and probably plays an important role in the development of a healthy gut in puppies. Allain et al. have demonstrated that L. johnsonii La1 possess anti-giardial activity through the production of bile salt hydrolases in vitro and in vivo in a murine model of infection [114]. Thus, probiotics represent a promising alternative strategy for the control of G. intestinalis infection [115].

Conclusions
The aim of this study was to describe the gut microbiota of growing puppies naturally infected with G. intestinalis. 16S Illumina sequencing revealed that the canine fecal microbiota becomes more complex and less diverse with age. Although some studies have linked the presence of intestinal parasites to gut health [116,117], subclinical chronic G. intestinalis infection in puppies appears to induce some detrimental structural changes in the gut microbiota. G. intestinalis-associated dysbiosis is characterized by an enrichment of facultative anaerobic, mucus-degrading, pro-inflammatory species, and opportunistic pathogens, as well as in a reduction of L. johnsonii at specific time points. These results do not resolve whether infection is responsible for these changes. It is possible that certain compositions of the gut microbiota confer susceptibility or resistance to colonization by the parasite, as has been suggested before [24,25]. On the other hand, G. intestinalis infection is associated with electrolyte, water and nutrient malabsorption, arginine deprivation, brush-border microvilli shortening, and disaccharidase deficiencies [118]. By modulating the availability of resources for commensal bacteria, these pathological changes could explain the alterations in the gut microbiota observed during G. intestinalis infection. Finally, the increased calprotectin levels illustrate the establishment of a chronic low-grade inflammation in puppies. However, further work is needed to demonstrate whether these alterations in the canine gut microbiota could lead to a dysbiosis-related disease, such as IBS or IBD.