Assessment of microbial diversity in the small intestine of dogs remains challenging, because anesthesia is required to obtain a sample, followed by either endoscopic or surgical collection of intestinal samples. Anesthesia may alter intestinal motility, and also repeated endoscopy may lead to perturbations of the intestinal microbiota. Therefore, the response of the jejunal microbiota to tylosin was evaluated in healthy Beagle Dogs each with a pre-existing jejunal fistula . All dogs were accustomed to their fistula for several years and it is, therefore, unlikely that the presence of this fistula has impacted the intestinal microbiota. We collected samples using a sterile cytology brush that was advanced through the fistula. This approach is easier, faster, and more reproducible compared to the aspiration of jejunal content. Furthermore, because an endoscope is too large to advance through the small lumen of the fistula, intestinal biopsies would have to be collected in a blinded fashion, which might have increased the variation in the sampling procedure. In contrast, mucosal brushings are technically easier to obtain and have been shown to be highly reproducible . We speculate that mucosal brushings represent a mixture of luminal content and the mucosa-adherent microbiota .
In this study, massive parallel 16S rRNA gene pyrosequencing proved to be a powerful and sensitive method for the further characterization of canine small intestinal microbiota. In previous studies using a comparative 16S rRNA gene based approach, up to 6 different bacterial phyla have been identified in the canine intestine [2, 24] We have identified 4 additional phyla that were not previously reported in dogs: Tenericutes, Cyanobacteria, Verrucomicrobia, and Chloroflexi. These bacterial phyla were present at low abundance, with less than 1% of all pyrosequencing tags. The ecological significance of these low abundant bacterial phyla in the canine intestine remains to be determined. Furthermore, due to their low abundance, it was not possible to appreciate any significant effect due to tylosin treatment. While the overall composition of the small intestinal microbiota on a phylum through genus level was similar as reported previously in the canine duodenum using 16S rRNA gene analysis [2, 24], the pyrosequencing approach has revealed a much higher richness on a species and strain level (Table 1). Rarefaction curves (Figure 1) revealed that with the number of here obtained sequencing tags per sample (mean ± SD: 3188 ± 1091), we have underestimated the number of OTUs at 1% dissimilarity, but obtained a reasonable coverage at 3% and 5% dissimilarity. Our calculations revealed that the canine jejunum harbors between 32 and 666 (mean: 293) bacterial species and between 183 and 1,789 (mean: 950) bacterial strains. Approximately 38,000 sequence tags would need to be analyzed per jejunal sample to cover 100% of the predicted maximum OTUs present in the canine jejunum. Therefore, future studies evaluating the small intestinal microbiota will need to employ larger sequencing datasets to characterize changes in low abundant bacterial groups.
By altering the intestinal microbiota, antibiotics can exhibit either a deleterious or a beneficial effect on gastrointestinal health. In humans with antibiotic associated diarrhea, a disruption of the intestinal ecosystem may predispose to an overgrowth of pathogenic species (e.g., C. difficile) . However, antimicrobials can also be useful in the treatment of intestinal disorders. The macrolide antibiotic tylosin is commonly used for the treatment of dogs with chronic diarrhea, but the exact mode of action of tylosin remains unclear [11, 12]. Most dogs respond favourably within 3-5 days, and stool consistency remains normal during treatment. However, diarrhea often reappears within weeks after discontinuation of administration . Tylosin belongs to the macrolide class of antibiotics that is characterized by a multi-membered lactone ring . Antibiotics of the macrolide class inhibit bacterial protein synthesis by binding to the L27 protein of the 50S ribosomal subunit. This inhibits the translocation of peptidyl-tRNA from the acceptor to the donor side on the ribosome, as well as the initial steps of assembly of the 50S subunit . Macrolides are more effective in crossing the cell membrane of gram-positive bacteria compared to gram-negatives . Therefore, the proposed antibiotic activity of tylosin is directed against gram-positive bacteria (e.g., Stapylococcus spp., Streptococcus spp., and Clostridium spp.) and also against some Mycoplasma and Chlamydia spp. While tylosin also has an effect against some gram-negative bacteria (e.g., Campylobacter spp., Helicobacter pylori, and Pasteurella spp.) it has no apparent effect against members of the Enterobacteriaceae (e.g., Escherichia coli) .
Antibiotics might exhibit their anti-diarrheal effect by either reducing total bacterial load in the gut or by modulating the proportions of specific bacterial taxa and, therefore, altering bacterial metabolites that affect the gastrointestinal tract. The here used pyrosequencing approach does not allow us to draw conclusions about changes in total bacteria within the intestine, as we did not include any measure for total bacterial load in our mucosal brushing samples. However, our approach shows changes in relative proportions of specific bacterial taxa in response to tylosin in a more comprehensive fashion than previously reported [9, 18]. Recent studies in humans have evaluated the response of intestinal microbiota to a short-course treatment with amoxicillin or ciprofloxacin on fecal microbiota [8, 16]. Similar to our results, antibiotic treatment led to major shifts in the dominant fecal bacterial populations, starting within 24 hours of administration . Furthermore, ciprofloxacin affected the abundance of approximately one third of all bacterial taxa . The human fecal microbiota proved to be generally resilient, and most taxa returned to baseline within 30 days, but some bacterial taxa failed to recover for up to 6 months [8, 16].
In this study evaluating the small intestinal microbiota, we observed significant changes in the canine small intestinal microbiota in response to tylosin. Results of the Unifrac distance metric indicated that the jejunal microbiota of individual dogs were phylogentically more similar during tylosin administration. Samples tended to cluster during tylosin administration, indicating that such changes were due to treatment effect rather than temporal variation. Furthermore, in previous studies, using either bacterial culture or DGGE analysis, it has been shown that the major bacterial groups in the canine jejunum display temporal stability over time [22, 28], further suggesting that the observed changes were indeed caused by tylosin treatment.
In general, the observed microbial shifts occurred in three major patterns: (a) bacterial groups that decreased in their proportions by day 14 and rebounded by day 28, (b) bacterial groups that decreased in their proportions by day 14 and failed to recover by day 28, and (c) bacterial groups that increased in their proportions by day 14 and returned to baseline values by day 28. We also observed unexpected highly individualized responses to tylosin treatment for specific bacterial taxa in some dogs. For dogs with diarrhea it is currently unknown if the effect of tylosin is mediated by a reduction in total bacterial load, by suppression of a single pathogen, or by an immunomodulatory effect . Our findings show that tylosin affects the proportion of various bacterial groups in the intestine. It is, therefore, unlikely that tylosin would have solely an effect on a single pathogen in clinical cases. It can be hypothesized that some of the observed shifts in microbial populations might contribute to the beneficial effect observed in dogs with chronic enteropathies. Examples of the beneficial effect of antibiotics may include altered concentrations of secreted metabolic products, decreased competition for nutrients or vitamins, altered cross-talk with the intestinal immune system, or a modification of cellular metabolism [29–31]. To prove this hypothesis, evaluation of these bacterial groups in clinical studies involving diseased animals are required. Furthermore, changes in bacterial populations will need to be correlated with treatment outcome.
It is interesting that the proportions of Enterococcus-like organisms, which are commonly used in probiotic formulations increased significantly during tylosin treatment. Enterococcus spp. have been reported to be resistant to tylosin in several animal studies [17, 32], and suppression of the commensal microbiota by antibiotic treatment may have allowed the proliferation of this bacterial group. For example, in one study using a continues flow culture model, a tylosin-resistant exogenous E. faecium strain could maintain itself only in the presence of tylosin . These results support the concept that tylosin may promote the growth of potentially beneficial commensal bacteria such as Enterococcus spp., which may have probiotic characteristics. A similar concept has also been suggested for the effect for the antibiotic metronidazole, also commonly used for treatment of dogs with chronic enteropathies. In humans, metronidazole increased the proportions of Bifidobacterium spp. . However, it remains unclear if a mere increase in the proportions of specific bacterial genera is sufficient to exhibit a probiotic effect. It is currently also unknown, if minor changes (i.e., less than 10-fold) as observed have any significant impact on intestinal health. To prove the concept that antibiotics may be able to promote proliferation of probiotic bacteria, it would be useful to isolate native Enterococcus strains and evaluate their functional interactions with other members of the intestinal microbiota and also evaluate their probiotic properties in dogs with gastrointestinal disease.
Tylosin is usually considered safe for long-term use in dogs . However, in this study we observed some unexpected microbial shifts, which may suggest that tylosin, similar to other antibiotics, can lead to a disruption of the intestinal ecosystem and also have potentially deleterious effects on gastrointestinal health. We observed significant increases for Pasteurella spp., E. coli-like organisms, and a dramatic increase in C. perfringens-like organisms in one dog. Tylosin is prescribed for the therapy of upper respiratory infections associated with Pasteurella multocida. However, this group increased significantly during the treatment period. It remains unclear, if Pasteurella multocida has developed resistance to tylosin in the here studied dogs, or if the intestinal phylotypes differ from those isolated from the lung. Tylosin appears to be an appropriate antibiotic for the treatment of C. perfringens-associated diarrhea in canine patients, although resistant strains have been observed . Similarly, in a chicken model of necrotizing enteritis, tylosin quantitatively decreased the proportion of mucolytic C. perfringens . However in this study, the percentage of C. perfringens-like organisms increased from 21.8% on day 0 to 86.7% on day 14 in one dog, suggesting that this dog harbored a resistant strain. Our results also suggest that the proposed mode of action of an antibiotic on different bacterial genera does not necessarily match the in vivo effects, as several bacterial groups that are considered to be sensitive to tylosin increased in their proportions. Because of the nature of an ecosystem, the changes that are induced by an antibiotic on one set of organisms will affect others, and this is not necessarily predicted by in vitro antibiotic sensitivities.
E. coli-like organisms, a bacterial group that has also been associated with a negative impact on gastrointestinal health in dogs [24, 35] increased significantly by day 28. The enrichment of E. coli-like organisms is not surprising, as this group is intrinsically resistant to tylosin, and similar increases have been observed in pigs after tylosin treatment . However, we have no obvious explanation why this effect was observed on day 28 rather than day 14, the last day of tylosin administration. Also, based on the techniques used, it is not possible to determine if a bacterial population proliferated or simply increased in proportion because other bacteria were affected (directly or indirectly) by the antibiotic treatment.
While E. coli-like organisms and C. perfringens increased in some of the dogs, this was not associated with any obvious clinical signs of gastrointestinal disease. We speculate that despite obvious changes in microbial populations, the intestinal ecosystem has enough functional redundancy to maintain gastrointestinal health. Similar findings have also been reported in humans, where short-term courses of antibiotics led to significant shifts in fecal microbiota patterns, yet no obvious gastrointestinal signs were observed [8, 16]. However, all these studies, including the present one, have evaluated healthy individuals, which may harbor a stable intestinal ecosystem that has enough functional redundancy to withstand short-term modulations. It is currently unknown how antibiotics affect dogs with gastrointestinal disease that may be more susceptible to such treatments. Of interest would be also to evaluate the long-term effects of antibiotics on the temporal stability of the intestinal microbiota and their influence on gastrointestinal health. It might be possible that the microbiota in animals undergoing a course of antibiotic treatment is less stable and, therefore, at an increased risk for gastrointestinal disease or infections. Follow up studies over a period of years would be needed to answer this question. In this study we have evaluated healthy dogs, and it is possible that tylosin has a different effect on the microbiota in dogs with signs of gastrointestinal disease. It is suspected that diseased dogs have an altered microbial composition, and it is possible that tylosin results in modulations in microbiota that differ from those observed in the here evaluated healthy animals. Evaluating endoscopically obtained pre- and post treatment samples from dogs with tylosin-responsive diarrhea would be valuable. Future studies will need also to evaluate intestinal contents for changes in bacterial metabolites or gene expression in response to antibiotic treatment as a measure of functional redundancy of the intestinal microbiota.
Studies in humans have shown that the fecal microbiota are generally resilient to short-term modulations by antibiotics, but pervasive effects might last for several months for specific bacterial taxa [8, 16]. The resilience of a microbial community reflects its capability to return to baseline after disturbances to the community (i.e., antibiotic treatment) have ceased. Less is known about the resilience of the small intestinal microbiota. Our results illustrate the complexity of the intestinal microbiota and the challenges associated with evaluating the effect of antibiotic administration on the various bacterial groups and their potential interactions. Our results indicate that tylosin may lead to prolonged effects on the composition and diversity of jejunal microbiota. On day 28, the phylogenetic composition of the microbiota was similar to day 0 in only 2 of 5 dogs. Bacterial diversity as measured by the Shannon-Weaver diversity index resembled the pre-treatment state in 3 of 5 dogs. Several bacterial groups changed in their proportions in response to tylosin. After cessation of tylosin, the phyla Firmicutes and Fusobacteria tended to return to pretreatment values within 14 days. Other phyla, such as Bacteroidetes, Proteobacteria, and Spirochaetes did not return to their pre-treatment proportions. Tylosin had also a pervasive effect on several bacterial groups that failed to recover by day 28 (i.e., 14 days after tylosin therapy had been completed). Those groups included Spirochaetes, Streptomycetaceae, Sphingomonadaceae, and Prevotellaceae. Tylosin has a known activity against Spirochaetes . Spirochaetes have been associated with intestinal disease in chickens and pigs, but their pathogenic role in dogs remains unclear, as they are commonly observed in healthy dogs as well as dogs with diarrhea [2, 24, 38]. The clinical significance of Sphingomonadaceae, Prevotellaceae, and Streptomycetaceae in the small intestine of dogs has, to our knowledge, not been evaluated to date. Furthermore, future studies with longer follow-up periods than 14 days after treatment cessation will be useful to evaluate the long-term effect of tylosin on the jejunal microbiota. Result of such studies may indicate the time needed for the microbiota to return to its pre-treatment state.