Research regarding the use of AGP in agriculture and animal production has predominantly focused on measuring the prevalence of AMR in zoonotic pathogens [4, 5, 23–25]. That giving AGP selects for resistance to antimicrobials in bacteria has promulgated the belief that AGP induce growth promotion in livestock as a result of direct effects on the intestinal microbiota [10, 26]. However, there is a dearth of experimental evidence to support the 'microbiota modulation hypothesis'. Furthermore, it was recently reported that AGP administration did not significantly alter the intestinal microbiota in pigs . Antibiotics are anti-inflammatory and immunomodulatory ; thus, AGP may function by modulating the immune system within the intestine thereby providing a catabolic advantage to the host . To our knowledge this has never been tested. We formulated the 'immunomodulation hypothesis' of AGP action, and we used a CTC-murine-C. rodentium model to test this hypothesis.
An antimicrobial agent is a growth promoter when administered in/on the feed of food animals to promote growth and enhance feed efficiency. Growth promoters are usually administered in relatively low concentrations (i.e. non-therapeutic), ranging from 2.5 to 125 mg kg-1 (ppm), depending on the compound and animal species treated . CTC is commonly used as an AGP for mammalian livestock . In the current study, a simulated non-therapeutic dose of CTC was given to mice with the dose extrapolated from that commonly used in beef cattle. Measurable quantities of ingested CTC (and its epimer, epi-CTC) are excreted in livestock feces . Consistent with this finding, measurable CTC concentrations were excreted in the feces of all mice given CTC. The experimental design required that mice be continuously administered CTC (i.e. to mice assigned to the CTC and CTC+CR treatments), necessitating that the C. rodentium be transformed to have resistance to CTC. Although tetracycline resistance determinants may persist in the absence of selection pressure [31, 32], the plasmid-borne tet O gene transferred into C. rodentium was less persistent in the mice not given CTC. However, there was no significant difference between the CR and CTC+CR treatments in shedding or mucosal colonization by the bacterium at the peak, late, and clearance stages of infection. Reasons for the significantly lower numbers of C. rodentium associated with colonic mucosa in mice not administered CTC at day 3 p.i. is unknown, but is likely due to sampling variability as there was no difference between the treatments in densities of C. rodentium cells shed in feces at day 3 p.i. or any other sample time.
We utilized mice that were reared under conditions in which intestinal inflammation would be expected to be exceptionally low (i.e. in the absence of inducers of immune responses such as C. rodentium), and as such, differential immunomodulation between the control and CTC treatment would be expected to be minimal. However, we observed that the expression of the cytokine genes IL-4 and TGF-β were markedly different in CTC mice as compared to control animals on 21 day p.i. The increase in IL-4 was both statistically and biologically significant (i.e. two times greater than in control animals). Importantly, both CR infected and uninfected mice that received CTC presented with modest to markedly significant higher concentrations of IL-4 at clearance relative to Control mice, thus suggesting that CTC contributes to the differentiation of Th2 cells. Treatment with therapeutic concentrations of tetracycline has been observed to induce changes in IL-4 during both acute and chronic disease by others. For example, Chirgwin et al  reported that IL-4 expression was decreased in tetracycline-treated Mongolian gerbils infected with filarial nematodes. In contrast, IL-4 levels in people infected with Orientia tsutsugamushi (incitant of scrub typhus) and treated with tetracycline did not change relative to background levels . Although therapeutic concentrations of tetracycline were used in these two studies, their study highlighted the varied responses that tetracycline treatment can have on Th2 cytokine expression within intestinal mucosa. The greater than 50% decrease in TGF-β levels that we observed in CTC mice relative to control mice (statistically and biologically significant) also indicates that the intestine of CTC mice reduces numbers of Treg lymphocytes. The observed effects of CTC on IL-4 and TGF-β are consistent with the immunomodulation hypothesis of AGP action, a conclusion that is consistent with the increased weight gain observed (i.e. a biological trend).
C. rodentium was chosen as an activator of inflammation and immunological function because it colonizes the colon of mice and incites acute, self-limited colitis, associated with progressive crypt hyperplasia in the distal colon [19, 35]. Consistent with previous reports, there were characteristic pathologic changes in mice inoculated with C. rodentium, including thickening of the distal colon wall and crypt elongation attributed to mucosal hyperplasia . We measured changes in intestinal morphology as opposed to weighing colons as hypertrophy of the muscularis externa, deposition of submucosal collagen (fibrosis), and infiltration of large numbers of inflammatory cells can all cause marked increases in colonic weight that are independent to changes in the mucosa. In addition to hyperplasia, we observed multifocal transmural inflammation and infiltration of inflammatory cells to the submucosa, muscularis externa, and serosa, as documented previously [22, 37]. C. rodentium induces strong Th1 and Th17 cell-mediated immune responses [36, 38–40], and there were significant increases in transcript levels of both Th1 (INF-γ, TNF-α, IL-2) and Th17 (IL-17A, IL-22, IL-1β, IL-6) cytokines in mice infected with the bacterium. These effector T helper cell responses are important in preventing and eliminating enteric bacterial infections. Th17 cells have a crucial role in the clearance of extracellular bacterial pathogens that are not adequately handled by Th1 or Th2 responses [38, 41], and Th1 cells are needed to eliminate prolonged infections . Moreover, morphological changes in the intestine, namely increased enterocyte proliferation and turnover, were associated with increased expression of the Th17 cytokines, IL-17  and IL-22 , an observation consistent with the present study. One of the key observations of this study was the time dependant activation of Th1 and Th17 of T helper cells. Based on differential expression of Th1 and Th17 cytokine profiles, we inferred there was a temporal relationship in cytokine expression necessary for eliminating infection by enteric pathogens, such as C. rodentium. The initial increase in Th17 cytokines may be needed to induce an effective immune response, but could be inadequate to completely clear an infection; therefore, subsequent induction of Th1 cytokines may be required to clear prolonged infections. This trend has been observed in other enteric pathogen and animal models. As an example, mice infected with Salmonella enterica serovar Typhimurium had a marked Th17 and Th1 response in the early and late stages of infection, respectively .
Consistent with the immunomodulation hypothesis, administration of a non-therapeutic dose of CTC decreased C. rodentium-induced weight loss and mitigated pathologic changes, albeit in a modest manner. For example, at 14 day p.i., mitotic activity was significantly increased for the CR relative to the Control treatment, but not for the CTC+CR relative to the Control treatment. Similarly, crypt height, inflammatory infiltrates, and the cumulative score were significantly increased for the CR treatment, but not for the CTC+CR treatment relative to the Control. Although the reductions in weight loss and histopathologic changes were relatively modest in CTC+CR relative to CR mice, it is apparent that CTC treatment altered immune function as the changes observed in the intestine cannot be explained by CR infection alone. Collectively, these observations are consistent with the immunomodulation hypothesis. We also observed that the administration of CTC (i.e. CTC+CR treatment) altered the expression of cytokine genes involved in the acute phase of inflammation relative to the CR treatment. Concentrations of Th1 cytokines were differentially expressed; IFN-γ mRNA were increased for the CTC+CR treatment at the peak (i.e. day 8 p.i.) and late (i.e. day 14 p.i.) stages of infection, whereas IL-2 mRNA was increased at clearance (i.e. day 21 p.i.), and TNF-α mRNA was increased at the peak stage and decreased at the late stage of infection in mice administered CTC. Concentrations of Th17 cytokines also were differentially expressed between the CTC+CR and CR treatments; IL-17A mRNA was decreased at day 14 p.i., IL-22 mRNA was increased at day 8 p.i. and decreased at day 14 p.i., and IL-1β mRNA was decreased at both days 8 and 14 p.i. An acute-phase response can be triggered by an infectious challenge; it is often characterized by release of the pro-inflammatory cytokines, IL-1β, IL-6, and TNF-α, which orchestrate behavioral, cellular, and metabolic adjustments in the host that alter the partitioning of nutrients away from growth and toward processes that support the immune and inflammatory responses . Furthermore, IFN-γ, IL-1β and IL-6 facilitate TNF-α-induced muscle cachexia . Interestingly, CTC administration only decreased expression of IL-17A mRNA at the late stage of infection in the current study, perhaps due to the concomitant down-regulation of TNF-α and IL-1β mRNA expression. Since the Th17 cell response was amplified by TNF-α and IL-1β , it is therefore likely that a decrease in TNF-α and IL-1β expression will also decrease activation of Th17 cells. Taken together, we conclude that reduced pathological changes at the late stage of infection, as well as at clearance, and the absence of weight loss at late infection in inoculated mice that received CTC were due to up-regulation of IL-22 transcripts (i.e. a cytokine important in mucosal defense) at the peak stage of infection, and down-regulation of TNF-α, IL-1β, and IL-17A transcripts (i.e. cytokines involved in inflammation, innate and adaptive immunity) at the late stage of infection, relative to inoculated mice not given CTC.
Our observations that a non-therapeutic concentration of CTC given to mice modulated C. rodentium-induced cytokine mRNA expression, ameliorated C. rodentium-induced weight loss, and mitigated pathologic changes associated with infection by C. rodentium, provided experimental evidence in support of immunomodulation as a primary mechanism of action of AGP in mammalian livestock. There is additional evidence to support the immunomodulation hypothesis. For example, oral administration of tetracycline reduced mortality rates and inflammatory lesions associated with lipopolysaccharide (LPS)-induced septic shock in mice by decreasing serum concentrations of TNF-α . Furthermore, tetracycline inhibited LPS-induced secretion of TNF-α and IL-1β by human monocytes in vitro. Similarly, non-therapeutic administration of tetracycline to experimentally stressed poultry reduced immunologic stress by decreasing plasma IL-1 concentrations . Also consistent with the immunomodulation hypothesis, Akunda et al.  reported that a low concentration of CTC decreased TNF-α secretion by cultured porcine Kupffer cells that were inoculated with LPS. In addition, intra-peritoneal injection of low doses of tetracycline and doxycycline decreased serum concentrations of IL-1α and TNF-α in mice .
The generally accepted microbiota modulation hypothesis of AGP action postulates that AGP modify the microbiota, thereby reducing the deleterious impacts of enteric bacteria on the host . To measure impacts of CTC on the mucosa-associated microbiota of the distal colon, T-RFLP analysis was used. The T-RFLP method is a medium to high resolution method that is commonly used to compare bacterial communities in mammalian intestines [50, 51]. The composition of the intestinal microbiota in association with mucosal surfaces can be profoundly affected by inflammatory responses, and sequence-based analysis (i.e. examining the partial 16S rRNA gene) of mucosa-associated and luminal bacteria in mice infected with C. rodentium revealed alterations to the microbiota . In contrast, we observed that C. rodentium infection did not impart a profound effect on the mucosa-associated microbiota of the murine distal colon in the current study. Reasons for the discrepancies between our study and that of Hoffmann et al.  are speculative, but may be because they restricted their examination to communities in the cecum and proximal colon, and communities were only resolved at the phylum or order level by pyrosequencing. Furthermore, they did not measure inter-animal variability. Although pyrosequencing of communities has become prominent in recent years, this platform is not a panacea; sequence reads are currently limited to less than 450 bp thereby limiting relevant sequence data to a small number of variable regions of the 16S rRNA gene, which greatly reduces the ability to discriminate amongst taxa, particularly those that are closely related (e.g. at the species level). One of the major advantages of T-RFLP is its ability to comparatively monitor treatment effects on the microbiota in empirical models by obtaining measures of inter-animal variability. It is possible that the inter-animal variability in the current study obscured subtle effects of C. rodentium infection on the colonic microbiota. Alternatively, C. rodentium may not exert substantive impacts on the microbiota of the distal colon, since the bacterium does not colonize this region of the colon throughout the entire infection period .
Although CTC was given at a non-therapeutic concentration in the current study, it affected the composition of the microbiota, albeit inconsistently and in a non-pronounced fashion. It has been reported that administration of non-therapeutic CTC substantially affected the composition of mucosa-associated microbiota in the ilea of piglets . In contrast, it was reported in a recent study that administration of virginiamycin and tylosin to intensively raised swine at the grow/finishing phase had no effect on bacterial communities in feces . Furthermore, there was a high prevalence of bacteria that carried genes encoding resistance to macrolide-lincosamide-streptogramin B (MLSB), from which they attributed to the negligible impact of AGP on the intestinal community . We have similarly observed that CTC and sulfamethazine, an AGP commonly administered to beef cattle, had no affect on the luminal and mucosa-associated microbiota of the distal small intestine and large intestine of beef cattle (unpublished).
It is evident that both host genetics and environmental factors can affect the intestinal microbiota. For example, persons with gene mutations responsible for the autoinflammatory disorder, familial Mediterranean fever, possessed significantly different bacterial community structures than healthy individuals, even when patients were in remission . Also, Altered Schaedler Flora (ASF) mice of the same genetic background exhibited significant variation in relative cell densities of the eight ASF strains throughout the intestine when mice were maintained in separate cages . The inter-animal variability encountered in the current and previous studies [54, 56, 57] further illustrates how difficult it is to quantify the impact of a single variable in such a complex system. Although CTC administration imparted a modest effect on the composition of the microbiota in the current study, it was not possible to definitively conclude whether this effect was direct or indirect. Considering a direct effect of CTC on the microbiota, it is unclear how and to what degree subtle changes in the microbiota caused by CTC would induce growth promotion, especially given the complexity of the mechanisms involved and that AGP exert an effect in animals with highly dissimilar microbiota. Since the indigenous microbiota can influence C. rodentium pathogenesis , even subtle changes in the microbiota may have influenced immune responses to C. rodentium, further illustrating the complexity of the interaction and the difficulty of elucidating specific effects.