ABSTRACT
Ecto-5′-nucleotidase (CD73) is expressed abundantly on the apical surface of intestinal epithelial cells (IECs) and functions as the terminal enzyme in the generation of extracellular adenosine. Previous work demonstrated that adenosine signaling in IECs results in a number of tissue-protective effects during inflammation; however, a rationale for its apical expression has been lacking. We hypothesized that the highly polarized expression of CD73 is indicative of an important role for extracellular adenosine as a mediator of host-microbe interactions. We show that adenosine harbors bacteriostatic activity against Salmonella enterica serovar Typhimurium that is not shared by the related purine metabolite 5′-AMP, inosine, or hypoxanthine. Analysis of Salmonella colonization in IEC-specific CD73 knockout mice (CD73f/fVillinCre) revealed a nearly 10-fold increase in colonization compared to that in controls. Despite the increased luminal colonization by Salmonella, CD73f/fVillinCre mice were protected against Salmonella colitis and showed reduced Salmonella burdens in viscera, suggesting that adenosine promotes dissemination. The knockdown of CD73 expression in cultured IECs resulted in dramatic defects in intraepithelial localization and replication as well as defective transepithelial translocation by Salmonella. In conclusion, we define a novel antimicrobial activity of adenosine in the gastrointestinal tract and unveil an important role for adenosine as a regulator of host-microbe interactions. These findings have broad implications for the development of new therapeutic agents for infectious disease.
INTRODUCTION
The intestinal microbiota has emerged as a source of an untold number of interactions that significantly affect both host and bacterial physiologies (1). Microbe-derived factors, such as lipopolysaccharide (2), short-chain fatty acids (3), toxins (4), and amino acid metabolites (5), among others, have been shown to alter host physiology and therefore contribute to both gastrointestinal health and disease. Likewise, host-derived factors affect the ability of intestinal flora to colonize and, in the case of enteric pathogens, infect the host (6, 7). The intestinal epithelium is uniquely situated at the interface between the host and the intestinal microbiota. Thus, a crucial function of the intestinal epithelium is to mediate interactions between the host and commensal and pathogenic microbes in the gastrointestinal lumen. Mucin production (8), antimicrobial peptide secretion (9), the activation of adaptive and innate immune responses (10), and tight junction formation (11) are among the mechanisms that the intestinal epithelium employs to maintain host-microbe homeostasis. In this study, we implicate extracellular adenosine generated through the activity of ecto-5′-nucleotidase (CD73) as an important factor present on the intestinal epithelium that influences multiple aspects of infection by the enteric pathogen Salmonella enterica serovar Typhimurium.
Extracellular adenosine has been recognized as an important regulator of inflammation in the intestinal mucosa. It has been shown that tissue-protective effects of adenosine are mediated via signaling through specific adenosine receptors expressed on intestinal epithelial cells (IECs) to suppress the production of proinflammatory cytokines and induce the production of cytokines that promote the resolution of inflammation (12). Extracellular adenosine is formed through the metabolism of extracellular ATP through the sequential activities of CD39 and CD73 (13, 14). CD73 is widely expressed on multiple cell types but is more highly expressed on the colonic intestinal epithelium than any other tissue. Furthermore, it is expressed almost exclusively on the apical surface of IECs. We believe that this highly polarized expression is particularly suggestive of a novel function of extracellular adenosine in the intestinal lumen.
Crane et al. previously examined the role of extracellular adenosine in enteropathogenic Escherichia coli (EPEC) infection (15–17). Those studies demonstrated that during infection by a number of enteric pathogens, extracellular ATP is produced and subsequently metabolized to adenosine. Furthermore, adenosine appeared to enhance the growth of EPEC and alter the pattern of virulence factor expression. Together, data from those studies suggested that host-derived adenosine and its metabolites are used as signals to the bacteria during infection. Interestingly, a large amount of ATP is released during S. Typhimurium infection, similar to the amount released during EPEC infection. However, the role of extracellular adenosine in the intestinal lumen has not been examined for Salmonella infection.
In this study, we demonstrate a novel role of IEC-derived CD73 in host-pathogen interactions during infection by Salmonella. We show that adenosine, the product of CD73 activity, potently inhibits the growth of Salmonella in a dose-dependent manner. Next, we demonstrate that CD73 expression is important for transepithelial translocation and intracellular replication by Salmonellain vitro. Using a novel mouse model, we show that the conditional knockout (KO) of CD73 in IECs, which reduces the amount of luminal adenosine, results in increased colonization of the intestinal lumen by Salmonella in murine Salmonella colitis and that CD73 expression plays a key role in the systemic dissemination of Salmonellain vivo. These findings clearly demonstrate the critical role of CD73 in the regulation of infection by the enteric pathogen Salmonella and implicate adenosine as an important host-derived factor that mediates host-microbe interactions.
RESULTS
Adenosine inhibits Salmonella growth in vitro.CD73 is expressed at higher levels in the colonic epithelium than in any other tissue (18), and its expression is restricted almost exclusively to the apical surface of IECs (13). Furthermore, the only known function of CD73 is in catalyzing the hydrolysis of 5′-AMP to adenosine. The explanation for the high apical expression level of CD73 has remained unclear. Given the interface of the intestinal lumen with that of the microbiota, we hypothesized that CD73-derived extracellular adenosine contributes to host-microbe interactions. To test this hypothesis, we examined the influence of adenosine on the growth of the enteric pathogen Salmonella. Bacteria grown in M9 minimal medium (with glucose and histidine) supplemented with increasing concentrations of adenosine revealed a dose-dependent delay in growth (Fig. 1A and B). Although the exact concentration of adenosine at the mucosal surface has not been measured during infection by Salmonella, the concentration of adenosine in cecal fluid from EPEC-infected animals was previously estimated to be 10 to 40 μM (15). Our data demonstrate that the transition from lag phase to exponential phase was delayed by as much as 2-fold in the presence of physiological concentrations of adenosine and was significantly different at concentrations as low as 1 μM (Fig. 1B). Once initiated, bacterial growth proceeded through the exponential phase at normal rates (Fig. 1A). Such a pattern of concentration-dependent bacteriostatic action is suggestive of a metabolic response to extracellular adenosine.
Inhibition of Salmonella SL1344 by adenosine. (A) Growth curves of bacteria in M9 minimal medium without adenosine (blue) and with an initial concentration of 1 μM (black), 10 μM (orange), 100 μM (magenta), or 1 mM (green) adenosine. (B) Growth delay of Salmonella bacteria with adenosine in M9 minimal medium is dose dependent. Salmonella bacteria were grown with the indicated initial concentrations of adenosine, and the duration of the lag phase was determined. (C) HPLC chromatograms demonstrating adenosine (100 μM initial concentration) metabolism to inosine and hypoxanthine by Salmonella in cell-free growth medium from 0 to 24 h after inoculation. (D) Adenosine and adenine inhibit Salmonella growth, but other purine metabolites do not inhibit growth. Bacteria were cultured in M9 minimal medium in the presence of each of the specified purine metabolites at an initial concentration of 100 μM. (E) Growth of an adenosine deaminase mutant strain (add mutant) and a purine nucleoside phosphorylase mutant strain (deoD mutant) is not sensitive to 100 μM adenosine in liquid culture, but growth of the parent strain (wild type [WT]) and a hypoxanthine phosphoribosyltransferase mutant strain (hpt mutant) is sensitive to 100 μM adenosine in liquid culture. Data represent the means and standard deviations of results from at least three independent experiments. **, P < 0.005; ***, P < 0.0005; ****, P < 0.00005; n.s., not significant (as determined by Student's t test).
Given the pattern of concentration-dependent growth retardation, we examined whether other purine metabolites exert a similar bacteriostatic effect on Salmonella. In both prokaryotes and eukaryotes, adenosine is metabolized in multistep catalytic reactions. In the host, AMP is metabolized to adenosine by CD73 and subsequently to inosine by adenosine deaminase. Inosine is then metabolized to xanthine and hypoxanthine by xanthine oxidase. To address whether such bacteriostasis might be explained by adenosine metabolism, we developed an assay to examine the metabolism of adenosine by Salmonella in liquid culture. As shown in Fig. 1C, by using reversed-phase high-performance liquid chromatography (RP-HPLC), this approach revealed that during lag phase, adenosine is metabolized to inosine and hypoxanthine by Salmonella. Once adenosine becomes undetectable in the culture medium, bacterial growth rapidly transitions to the exponential phase. During exponential growth, both inosine and hypoxanthine become undetectable at >20 h. These results raised the possibility that a downstream metabolite of adenosine could account for the bacteriostatic effect that was observed.
To determine whether adenosine or one of its downstream metabolites is bacteriostatic, we examined the influence of each of these purine metabolites on Salmonella growth in vitro. As shown in Fig. 1D, adenosine exerted a greater bacteriostatic effect on Salmonella than did its upstream and downstream metabolites: ATP, AMP, inosine, and hypoxanthine. In Salmonella, adenosine is metabolized to inosine and adenine by adenosine deaminase (add) and purine-nucleoside phosphorylase (deoD), respectively. To investigate the role of adenosine metabolism in the bacteriostatic activity of adenosine, deletions of either add or deoD were generated. Unlike the parent strain, the deoD mutant strain was not sensitive to adenosine (Fig. 1E), suggesting that the metabolism of adenosine to adenine is important for the bacteriostatic effect of adenosine. This is supported by the observation that adenine, the product of adenosine metabolism by deoD, retains the antimicrobial activity of adenosine (Fig. 1D). In contrast, while the add mutant strain was also not sensitive to adenosine (Fig. 1E), inosine did not exhibit bacteriostatic activity (Fig. 1D). Finally, a hypoxanthine phosphoribosyltransferase (hpt)-deficient strain remained sensitive to adenosine, suggesting that the metabolism of hypoxanthine to IMP is not required for the bacteriostatic activity of adenosine. These findings indicate that extracellular adenosine serves as a critical link to exponential-phase initiation, and as such, high concentrations of adenosine function as a bacteriostatic mediator in the local microenvironment.
Salmonella virulence is attenuated after knockdown of CD73 in IECs.As previously reported, extracellular adenosine is produced during Salmonella infection (16), even though we have shown that adenosine exerts antimicrobial effects on Salmonella. We hypothesized that extracellular adenosine produced through the activity of CD73 promotes Salmonella virulence to overcome the bacteriostatic effect that may otherwise hinder the infectious process. To investigate the relative influence of IEC CD73 expression on Salmonella virulence, we turned to a well-established in vitro model of Salmonella infection utilizing T84 intestinal epithelial cells (19, 20). To define the specificity of CD73, we utilized a lentiviral short hairpin RNA (shRNA) approach to knock down CD73 expression in T84 cells. As shown in Fig. 2A, this strategy achieved 91% ± 5.6% knockdown (KD) (P < 0.005 by Student's t test) relative to a short hairpin nontargeting control (shNTC), as determined by reverse transcription-PCR (RT-PCR). Functional analysis of CD73 activity indicated an 85% ± 7% decrease in the ability of CD73 KD cells to convert a 5′-AMP analogue, 1,N6-etheno-AMP, to its corresponding adenosine analogue, 1,N6-etheno-adenosine, as we described previously (21). This is comparable to the influence of adenosine 5′-(α,β-methylene)diphosphate (APCP), a specific inhibitor of CD73 activity (P < 0.0005 by Student's t test for each comparison) (Fig. 2B).
Knockdown (KD) of CD73 expression in T84 intestinal epithelial cells impairs intracellular replication of and translocation by Salmonella. (A) Lentiviral shRNA-mediated KD of CD73 (shCD73) in T84 intestinal epithelial cells results in diminished CD73 mRNA relative to the short hairpin nontargeting control (shNTC). **, P < 0.005 as determined by Student's t test. (B) CD73 KD results in diminished phosphohydrolysis of 1,N6-etheno-5′-AMP to 1,N6-ethenoadenosine, similar to the effect of the specific CD73 inhibitor adenosine 5′-(α,β-methylene)diphosphate (APCP). ***, P < 0.0005 for conversion by APCP-treated monolayers as well as CD73 KD monolayers relative to the control, as determined by Student's t test. (C) Initial invasions of control and CD73 KD monolayers by Salmonella are equivalent, as determined by a gentamicin protection assay. Monolayers were harvested at 1 h postinfection, as described in Materials and Methods. (D) Intracellular replication of Salmonella is decreased after CD73 KD in T84 cells at 8 h postinfection compared to that in control cells in an in vitro gentamicin protection assay. Recovered CFU were normalized to the CFU recovered at 1 h postinfection. **, P < 0.005 as determined by two-way ANOVA. (E) Translocation of Salmonella bacteria across polarized T84 monolayers on semipermeable supports is reduced 100-fold after CD73 KD. Data represent the means and standard deviations of results from at least three independent experiments. ***, P < 0.0005 as determined by Student's t test.
Using CD73 KD cells, we examined how the loss of CD73 might the impact the invasion and intracellular replication of Salmonella in IECs. Using a gentamicin protection assay, we found that the initial invasion of IECs was not impacted by the loss of CD73 expression (Fig. 2C). Conversely, the intracellular replication of Salmonella in CD73 KD cells was significantly compromised. A time course of intracellular replication following initial invasion revealed a nearly complete loss of intracellular replication in CD73 KD cells relative to cells with the shNTC (P < 0.005 by two-way analysis of variance [ANOVA]) (Fig. 2D). These results suggest that Salmonella virulence is decreased with the loss of CD73 expression.
To further examine the role of CD73 in Salmonella virulence, we examined the transepithelial translocation of Salmonella across polarized T84 monolayers on semipermeable inserts. Monolayers were infected with Salmonella apically at a multiplicity of infection (MOI) of 100. Translocated bacteria were detectable in the basolateral compartment starting at 3 h postinfection (p.i.). At 5 h postinfection, the CD73 KD monolayers demonstrated <1% of the number of translocated bacteria relative to control monolayers (123-fold decrease; P < 0.0005 by Student's t test) (Fig. 2E), again showing that Salmonella virulence is significantly attenuated with the loss of CD73 expression in IECs.
We next sought to examine the mechanism by which CD73 expression modulates Salmonella virulence. Using confocal immunofluorescence microscopy, we determined the intracellular distribution of Salmonella during the gentamicin protection assay. It was previously shown that following invasion of epithelial cells, Salmonella bacteria traffic within the Salmonella-containing vacuole (SCV) to a juxtanuclear locale (22, 23), after which the bacteria may escape the SCV into the cytosol and become hyperreplicative (24). Two hours after infection of control cells, Salmonella bacteria were observed in an even distribution pattern along the apical to basolateral axes intracellularly (Fig. 3A to H). In stark contrast, infected CD73 KD cells showed clusters of bacteria localized only at the apical pole of cells and almost no bacteria at the basolateral aspect of the cell. Furthermore, the basolateral localization defect in the knockdown cells persisted to at least 8 h postinfection (Fig. 3I to P).
CD73 KD results in altered localization of intracellular Salmonella bacteria. Shown are confocal immunofluorescence microscopy images of control (A, B, E, F, I, J, M, and N) and CD73 KD (C, D, G, H, K, L, O, and P) T84 cells 2 h (A to H) and 8 h (I to P) after infection with Salmonella expressing the mCherry plasmid. Shown are reconstructions of sections through the apical (A to D and I to L) and basolateral (E to H and M to P) aspects of the monolayers. Images showing mCherry alone (A, C, E, and G) and corresponding merged images with mCherry (red) and F-actin (blue) (B, D, F, and H) highlight the intracellular distribution of bacteria in the monolayers. Bars, 10 μm.
CD73 is essential for the production of luminal adenosine in the gut in vivo.Based on our observation that adenosine inhibits Salmonella growth in vitro, we extended these studies to determine whether adenosine produced through the activity of CD73 would decrease the colonization of the bowel by Salmonellain vivo (25). To examine the role of CD73 in vivo, we selectively deleted CD73 expression in the intestinal epithelium with the goal of reducing luminal adenosine production. CD73flox/flox mice were bred with a transgenic strain expressing Cre recombinase under the control of the murine villin promoter (CD73f/fVillinCre) (Fig. 4A). CD73f/fVillinCre mice showed a >65% decrease in the expression level of CD73 mRNA in luminal scrapings of the colonic epithelium by RT-PCR (P < 0.005 by Student's t test) (Fig. 4B) and unchanged expression levels of CD73 in other tissues (not shown). Since lymphocytes express CD73 (26), the residual CD73 expression likely represents that of intraepithelial lymphocytes associated with mucosal scrapings. Of note, CD73f/fVillinCre mice bred and developed normally and did not develop spontaneous colitis or other overt disease.
Selective knockout (KO) of CD73 in intestinal epithelial cells (IECs) results in reduced severity of Salmonella colitis. (A) Breeding scheme to generate IEC-specific CD73 knockout (CD73f/fVillinCre) and control (CD73fl/fl) mice. (B) Reduced CD73 mRNA expression levels in epithelial scrapings of CD73−/− mice compared to controls. **, P < 0.005 by Student's t test. (C) Colon lengths of CD73f/fVillinCre and control mice 72 h after induction of Salmonella colitis. Data represent the means and standard deviations of results from at least three independent experiments. *, P < 0.05 by Student's t test. (D) Body weight curves of CD73f/fVillinCre and control mice after induction of Salmonella colitis following streptomycin treatment. *, P = 0.05 by two-way ANOVA. (E to H) Histological sections of control (E and F) and CD73f/fVillinCre (G and H) mice at baseline (E and G) and 48 h postinfection (F and H). Bars, 100 μm. (I) Pathological scoring of cecal sections from streptomycin-treated control mice and Salmonella-infected mice according to the scoring scheme reported previously by Barthel et al. (25). No statistically significant differences between control and infected mice were observed (P > 0.05). S. Tm, S. Typhimurium.
Using these CD73f/fVillinCre mice, we initially examined responses to Salmonella-induced colitis. As shown in Fig. 4C and D, CD73f/fVillinCre mice were significantly protected from colitis, as demonstrated by the less pronounced weight loss and decreased colon shortening at 7 days postinfection. Despite these clear differences, however, the histology of Salmonella-infected CD73f/fVillinCre mice was not significantly different from that of Salmonella-infected control mice, as shown in Fig. 4E to H and reflected in pathological scores (Fig. 4I).
We next looked at how the loss of CD73 expression by IECs affects the colonization of the intestinal lumen by Salmonella. The loss of IEC expression of CD73 resulted in a nearly 10-fold increase in Salmonella colonization of the cecum and colon by 48 h compared to Cre-negative control animals (P < 0.05 by Student's t test) (Fig. 5A and B). These observations suggest that luminal adenosine, generated through the activity of apically expressed CD73, results in the suppression of colonization by Salmonella. This is supported by our in vitro observations that extracellular adenosine has bacteriostatic activity toward Salmonella.
Knockout of CD73 in IECs increases luminal Salmonella burden and decreases dissemination in Salmonella colitis. (A and B) Salmonella bacteria in the cecum (A) and colon (B) contents of CD73f/fVillinCre and control mice 48 h after induction of Salmonella colitis. *, P < 0.05 by Student's t test. (C to E) Salmonella bacteria in the mesenteric lymph node (C), spleen (D), and liver (E) tissues from CD73−/− and control mice 48 h after induction of Salmonella colitis. (F) Intracellular bacteria from colon tissues of CD73f/fVillinCre and control mice 48 h after induction of Salmonella colitis after extracellular bacteria were killed with gentamicin. Data represent the means and standard deviations of results from at least three independent experiments. **, P < 0.005 by Student's t test.
Dissemination of Salmonella is dependent on IEC expression of CD73 in Salmonella colitis.Based on the above-described observations, we next sought to determine whether systemic Salmonella dissemination was influenced by the loss of IEC CD73. Salmonella colitis was induced as described above, and the dissemination of bacteria to mesenteric lymph nodes (MLN), spleen, and liver was examined at 48 h postinfection. As shown in Fig. 5, the dissemination of Salmonella to the MLN, spleen, and liver was attenuated by as much as 35-fold in CD73f/fVillinCre mice. These findings corroborate our in vitro observations that the loss of IEC CD73 results in abnormal intracellular localization and decreased bacterial translocation, demonstrating that the IEC-specific expression of CD73 is a major determinant of bacterial translocation and systemic dissemination during enteric Salmonella infection. An ex vivo gentamicin protection assay in colon tissues also demonstrated a decrease in the level of intracellular Salmonella bacteria in CD73f/fVillinCre mice (P < 0.005 by Student's t test) (Fig. 5F), suggesting that the loss of CD73 results in attenuated intracellular replication, which is also consistent with our in vitro observations.
Taken together, our in vitro and in vivo data reveal that IEC CD73 is central to the localization of Salmonella across the intestinal epithelium. In the absence of CD73 expression on IECs, bacterial translocation is compromised. This results in attenuated Salmonella virulence through decreased systemic dissemination.
DISCUSSION
The intestinal epithelium serves as the initial interface for host-microbe interactions. Given their anatomic location between the intestinal lumen and the submucosal immune system, IECs form the ultimate physical barrier between the host and the intestinal microbiota. This barrier is crucial for the maintenance of intestinal homeostasis, as intestinal barrier dysfunction has been clearly linked to gastrointestinal disorders such as inflammatory bowel disease (27). This barrier is multifactorial; tight junction formation and mucin secretion contribute to the physical barrier, while secreted factors such as antimicrobial peptides represent a soluble barrier that acts directly upon luminal and surface-associated microbes to compartmentalize the intestinal microbiota (9).
Here, we present a new role for IEC CD73 in the regulation of infection by the enteric pathogen S. Typhimurium. Adenosine, the product of the only known activity of CD73, showed bacteriostatic actions on the growth initiation of Salmonella. This activity has striking specificity. Indeed, other than adenine, none of the proximal upstream or downstream metabolites, including AMP, inosine, and hypoxanthine, were similarly active in growth inhibition at equimolar concentrations. It is also notable that adenosine metabolism by Salmonella coincided with the initiation of exponential growth. Salmonella metabolizes adenosine by one of two enzymes, namely, to inosine through add or to adenine via deoD. The HPLC analysis done here shows that extracellular adenosine is metabolized, and the major purine metabolites are ultimately depleted by Salmonella in liquid culture. Interestingly, mutants of the two major enzymes responsible for adenosine metabolism abrogated the bacteriostatic effect of adenosine. Consistent with these observations, adenine, the product of adenosine metabolism by deoD, was also bacteriostatic, which raises the possibility that the bacteriostatic activity of adenosine requires metabolism to adenine. Surprisingly, the add mutant also lost susceptibility to the bacteriostatic effect of adenosine; however, inosine, the product of adenosine metabolism by add, lacked the bacteriostatic activity of its upstream metabolite. While unclear at present, it is possible that the deletion of add could have secondary effects on purine transport, metabolism, or sensitivity. Further investigation into the metabolism and transport of adenosine, including the precise locations of adenosine metabolism in Salmonella, will be necessary to elucidate the specific mechanism by which adenosine exerts its bacteriostatic effect.
Interestingly, although CD73 expression appears to suppress colonization by Salmonella, there is an opposite effect on Salmonella virulence. This suggests a second important role of CD73 in Salmonella infection, in which the virulence of the pathogen is promoted. In the absence of CD73 expression on the intestinal epithelium, the severity of Salmonella colitis is attenuated in CD73flx/flxVillinCre mice, as reflected by the decreased weight loss and attenuated colon shortening in response to Salmonella colitis in these animals. This observation is supported by our in vitro findings that intracellular replication and bacterial translocation across the epithelium are compromised after KD of CD73 expression. The concept that extracellular adenosine or its downstream metabolites can enhance the virulence of enteric pathogens was suggested previously for EPEC, Clostridium difficile, and Pseudomonas aeruginosa (28–30), although in each case, different mechanisms have been proposed. Our data suggest that the intracellular localization of Salmonella is altered in the absence of CD73 expression, which results in defective intracellular replication and translocation. However, the specific mechanism of this localization defect is unknown.
The relevance of the antibacterial activity of adenosine was demonstrated by using both a CD73 knockdown approach in vitro and a conditional IEC knockout model in vivo. In the absence of IEC CD73 expression, colonization of the intestinal lumen by Salmonella was dramatically increased following oral infection. The in vitro studies using IECs lacking CD73 expression revealed that while Salmonella invasion is intact, CD73 appears to significantly influence intracellular localization. In particular, the absence of IEC CD73 attenuated intracellular Salmonella movement from the apical to basal poles and prohibited effective translocation. While the mechanism of such attenuated intracellular movement is not known at present, we suspect that this defect in trafficking prevents localization to the juxtanuclear location and SCV maturation. We speculate that in the absence of normal CD73 expression, Salmonella bacteria may be less likely to lyse the vacuole and consequently will replicate more slowly than cytosolic bacteria (24). Furthermore, we predict that the replication of cytosolic bacteria is not restricted by adenosine because cytosolic adenosine concentrations have been reported to be <1 μM (31). SCV-containing bacteria in the apical aspect of the cell are unable to translocate intracellularly to the basolateral membrane, explaining the defect in transepithelial translocation seen in CD73 KD T84 cells. These observations suggest that intraluminal adenosine functions as an endogenous antimicrobial in vivo and support an expanding role for adenosine as an innate immune mediator. These data support the hypothesis that luminal adenosine is important for regulating the growth of luminal microbiota.
Alam et al. recently reported the anti-inflammatory role of CD73 in murine Salmonella infection (32). Those authors showed that S. Typhimurium suppresses CD73 expression in a number of tissues, including the intestine and lymphocytes. Mice with a global knockout of CD73 demonstrated increased expression levels of proinflammatory cytokines in response to S. Typhimurium infection. Similar to our findings, the severity of Salmonella infection was attenuated in CD73−/− mice, as demonstrated by decreased weight loss following infection and decreased dissemination of bacteria to the spleen and liver. Those authors concluded that the augmented inflammatory response that occurs in the absence of CD73 expression allows more efficient clearance of the pathogen via adaptive and innate immune mechanisms. A fundamental difference between the present study and that by Alam et al. is our use of an IEC conditional CD73 knockout rather than a whole-body knockout model. Our results show that much of the in vivo phenotype that is seen in both the present study and the study by Alam et al. can be explained by the absence of CD73 expression in the intestinal epithelium. Augmented proinflammatory cytokine expression in dextran sulfate sodium salt-induced colitis in whole-body CD73−/− mice was reported previously (33), similar to what was observed in the study by Alam et al. It remains to be seen how this enhanced proinflammatory cytokine expression contributes to the mechanism of attenuated bacterial dissemination. In fact, our data argue that the most significant CD73-dependent host-pathogen interactions occur at the intestinal epithelium and that these interactions largely determine the course of infection in either IEC-specific or whole-body CD73−/− mice. This is consistent with the observation that the increased colonization of the liver and spleen in conditional KO mice mimics the colonization differences seen in whole-body KO mice. Our in vitro translocation data give this argument plausibility by suggesting that the absence of CD73 in the intestinal epithelium greatly reduces bacterial translocation, which leads to decreased systemic dissemination and, thus, decreased colonization of distant sites. Therefore, our findings significantly advance the understanding of the role of CD73 in Salmonella infection by identifying the intestinal epithelium as the pivotal site of CD73 expression that determines the course of local and systemic infection.
In conclusion, we have shown that the expression of CD73, the key enzyme in the production of extracellular adenosine, by intestinal epithelial cells is an important regulator of the intestinal colonization and virulence of Salmonella and has a significant effect on the dissemination of bacteria to extraintestinal sites. These findings suggest a role for adenosine as an endogenous regulator of the intestinal microbiota and a potential therapeutic target. Further investigation into the mechanism by which adenosine exerts antimicrobial actions and the modulation of Salmonella virulence is under way.
MATERIALS AND METHODS
Bacterial cultures and growth conditions.Wild-type Salmonella enterica subsp. enterica serovar Typhimurium strain SL1344 was obtained as a gift from C. Detweiler, University of Colorado (Boulder, CO). Wild-type SL1344 bacteria were transformed with the pFPV-mCherry plasmid, which was a gift from O. Steele-Mortimer (Addgene plasmid 20956) (34). Mutations were constructed by using the λ Red recombinase system (35). For growth curves, cultures grown overnight in Luria-Bertani-Miller broth (LB) at 37°C with shaking and 100 μg/ml ampicillin (Sigma-Aldrich, St. Louis, MO, USA) were washed and diluted in fresh M9 minimal medium with 0.4% glucose and 40 μg/ml l-histidine (Sigma-Aldrich) to an optical density at 600 nm (OD600) of 0.10 and then diluted again 1:20. Five microliters of this solution was used to inoculate each 150 μl culture. Growth curves were performed in 96-well flat-bottom tissue culture plates (Corning, Corning, NY, USA) at 37°C with periodic shaking, and growth was monitored by measuring the OD600 every 15 to 30 min for up to 60 h. Growth curves were analyzed with BioTek Gen 5 software (BioTek Instruments, Inc., Winooski, VT, USA), using kinetic analysis to determine the initial lag time. For animal experiments and in vitro experiments, cultures grown overnight in LB were diluted 1:33 in fresh LB and grown to mid-log phase. The bacterial density was determined by the OD600 and enumeration on agar plates. Cultures were pelleted, resuspended in cold phosphate-buffered saline (PBS), and used immediately.
Chemicals and reagents.Adenosine, 5′-AMP, 5′-ATP, inosine, adenine, hypoxanthine, and streptomycin were obtained from Sigma-Aldrich (St. Louis, MO). 1,N6-Etheno-AMP was obtained from Molecular Probes (Eugene, OR).
Measurement of purine metabolites.Samples of bacterial cultures were collected at specified times and centrifuged at 10,000 × g for 10 min, and the supernatant was collected. The supernatant was passed through a 5,000-molecular-weight-cutoff (MWCO) spin column prior to analysis to remove higher-molecular-weight compounds (Vivaspin 500; Sartorius Stedim, Göttingen, Germany). Samples were analyzed on an Agilent Technologies 1260 liquid chromatography system (Agilent Technologies, Santa Clara, CA, USA) by using a Phenomenex Luna C18(2) 150-by-4.6-mm-internal-diameter (ID), 5-μm-particle-size, 100-Å-pore-size column (Phenomenex, Torrance, CA, USA). Samples were analyzed by using an isocratic method consisting of a mobile phase of ultrapure water-acetonitrile (96:4) and 0.5 mM tetrabutylammonium bisulfate (Sigma-Aldrich) at a flow rate of 1 ml/min (25°C). Eluted compounds were detected at 260 nm by using an inline UV detector.
Animals.CD73 conditional knockout mice (Mus musculus) were generated by crossbreeding CD73 floxed (CD73fl/fl) mice (36) containing loxP sites flanking exon 2 of the CD73 gene to mice harboring the Cre recombinase under the control of the villin promoter (villin-Cre mice; Jackson Laboratories, Bar Harbor, ME, USA).
Induction of Salmonella colitis was performed as described previously (25). Briefly, food and water were withdrawn 4 h prior to oral gavage with 20 mg streptomycin in 75 μl water, after which food and water were returned. Twenty hours after streptomycin gavage, food and water were again withdrawn, and 4 h later, animals were gavaged with 108 CFU Salmonella in 50 μl PBS. Water was returned immediately, and food was returned at 2 h p.i. Animals were monitored daily and sacrificed at 24 to 72 h p.i. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado.
Cell culture.Human T84 intestinal epithelial cells were cultured in 95% air with 5% CO2 at 37°C. Lentiviral particles were acquired from the Functional Genomics Facility, University of Colorado at Boulder (Boulder, CO, USA) and transduced into T84 IECs by using established protocols.
Transcriptional analysis.TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was used to isolate RNA from confluent T84 monolayers. RNeasy (Qiagen, Hilden, Germany) was used to isolate RNA from mouse tissues. cDNA was reverse transcribed by using the iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA, USA). PCR analysis was performed by using SYBR green (Applied Biosystems, Carlsbad, CA, USA), using primers targeting human CD73 mRNA (forward primer 5′-ATTGCAAAGTGGTTCAAAGT and reverse primer 5′-ACACTTGGCCAGTAAAATA) and mouse CD73 mRNA (forward primer 5′-CTGGGGCACTCTGGTTTTGA and reverse primer 5′-TCCCCGCAGGCACTTCTTTG). Relative gene expression was calculated by using the method of Pfaffl (37) and the 2−ΔΔCT method (38) for human and mouse samples, respectively, in comparison to the β-actin housekeeping gene.
CD73 activity assay.CD73 activity was determined by monitoring the hydrolysis of 1,N6-etheno-5′-AMP (Molecular Probes) to 1,N6-etheno-5′-adenosine as previously described (21).
In vitro epithelial infection.For bacterial invasion assays, T84 cells were plated onto 6-well dishes and grown to confluence. All incubation steps were performed at 37°C with 5% CO2 unless otherwise noted. One day prior to infection, monolayers were washed once and then incubated overnight with antibiotic-free growth medium. On the day of infection, bacteria were grown to mid-log phase and enumerated, as described above. Monolayers were infected in Hanks' balanced saline solution with 10 mM HEPES (HBSS) at an MOI of 25 CFU per epithelial cell for 10 min. Monolayers were washed three times with HBSS and then incubated in antibiotic-free medium for 20 min. Gentamicin was added to a final concentration of 50 μg/ml, and the mixture was incubated for 30 min and then changed to 5 μg/ml gentamicin for the remainder of the assay. Monolayers were solubilized in 1% Triton X-100 and 0.1% sodium dodecyl sulfate at 1 to 8 h p.i. Bacteria were enumerated by plating serial dilutions onto LB agar plates supplemented with streptomycin at 50 μg/ml and ampicillin at 100 μg/ml. For intracellular replication assays, the number of invading bacteria was normalized to the number of invading bacteria recovered at 1 h p.i.
For transepithelial bacterial translocation assays, T84 cells were seeded onto 0.33-cm2, 3.0-μm-pore-size polyester Transwell inserts (Corning). The confluence of monolayers was monitored by transepithelial electrical resistance (TER), and monolayers were used approximately 7 days after seeding, once TERs were stable. Monolayers were placed into antibiotic-free medium 1 day prior to infection. On the day of infection, monolayers were washed with HBSS. Bacteria were prepared as described above, and monolayers were infected apically at an MOI of 100 in HBSS. Translocation was monitored by plating serial dilutions of the basolateral compartment onto dual-selection medium at times points from 1 to 6 h p.i.
Pathological scoring of Salmonella colitis histology.Hematoxylin-and-eosin-stained sections of paraffin-embedded cecal sections were scored in a blind manner according to a scheme described previously (25).
Immunofluorescence.T84 cells were grown to confluence on collagen-coated glass coverslips (Thermo Fisher Scientific, Waltham, MA, USA), fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and blocked with 10% normal goat serum (NGS). Retinoblastoma (Rb) monoclonal antibody (MAb) against CD73 (Cell Signaling Technology, Danvers, MA, USA) was used at a 1:100 dilution in 5% NGS. The secondary antibody was Alexa Fluor 568 goat anti-rabbit (Invitrogen, Carlsbad, CA, USA), used at a 1:500 dilution in 5% NGS. F-actin was concurrently stained with Alexa-Phalloidin 546 (Life Technologies, Carlsbad, CA, USA).
Images were acquired on a Zeiss LSM 510 confocal microscope (Zeiss, Thornwood, NY) equipped with a temperature-controlled incubation chamber (Solent Scientific, Fareham, UK). Imaging settings were defined empirically to maximize the signal-to-noise ratio and to avoid saturation. Image processing was performed by using Zeiss ZEN 2009 software.
Statistical analysis.GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA) was used to generate figures and perform statistical analysis, including paired and unpaired t tests.
ACKNOWLEDGMENTS
This work was supported by NIH grants DK099452, DK50189, and DK095491 and by the Crohn's and Colitis Foundation of America.
We declare no financial interests in any of the work submitted here.
FOOTNOTES
- Received 14 December 2016.
- Returned for modification 19 January 2017.
- Accepted 3 July 2017.
- Accepted manuscript posted online 17 July 2017.
- Copyright © 2017 American Society for Microbiology.
