Specialized Proresolving Mediators Rescue Infant Mice from Lethal Citrobacter rodentium Infection and Promote Immunity against Reinfection

ABSTRACT

Infants are generally highly susceptible to oral pathogens. Intestinal infection and the associated diarrhea are significant global causes of morbidity and mortality in infants. Among the enteric pathogens, enteropathogenic Escherichia coli (EPEC) stands out as showing the highest risk for infection-induced death in infants ≤12 months old. We have developed an experimental model of infant infection with EPEC, using the mouse-specific pathogen Citrobacter rodentium. Our murine infant model is similar to EPEC infection in human infants since infant mice are much more susceptible to C. rodentium infection than adult mice; infants infected with 50-fold fewer bacteria than the standard adult dose uniformly succumbed to the infection. Infant infection is characterized by high early and sustained bacterial titers and profound intestinal inflammation associated with extensive necrosis and systemic dissemination of the bacteria. Therefore, it seems likely that infant deaths result from sepsis secondary to intestinal damage. Recently, specialized proresolving mediators (SPM) have been found to exert profound beneficial effects in adult models of infection. Thus, we investigated the actions of two proresolving lipid mediators, resolvin D1 (RvD1) and resolvin D5 (RvD5), on the course of infection in infants. Strikingly, postinfection treatment with RvD1 and RvD5 reduced bacterial loads, mitigated inflammation, and rescued the infants from death. Furthermore, postinfection treatment with RvD1 and RvD5 led to protection from reinfection associated with C. rodentium-specific IgG responses comparable to those in adults. These results indicate that SPM may provide novel therapeutic tools for the treatment of pathological intestinal infections in infants.

INTRODUCTION

Intestinal disease in infants infected with enteropathogens constitutes a substantial global health burden. Infants infected with oral pathogens can develop pathological diarrhea and experience significant morbidity and mortality. Why infants are so susceptible to oral pathogens is poorly understood, but the reasons may be multifactorial. For example, poor colonization resistance (1, 2) is likely to be an important contributing factor, but developmental disparities in the immune system almost certainly play a role. In particular, it is known that many of these pathogens, including the bacteria enteropathogenic Escherichia coli (EPEC) (3–5) and Shigella (5–7), rotavirus (8), and the parasite Cryptosporidium (9, 10), induce profound innate intestinal inflammation in infants. The relative nonspecificity of innate function may lead to damage of the intestinal barrier and contribute importantly to disease pathology. Additionally, excessive innate inflammation may fail to adequately promote the development of adaptive immunity (11, 12) and thereby further contribute to susceptibility. At present, however, the relationship between inflammation and immune-mediated susceptibility to pathological disease in infants is not known.

In adult animals, active resolution of innate inflammation is critical for healing and the return to homeostasis (11–16). A number of specialized proresolving mediators (SPM) have been identified and shown to regulate the phagocytosis and killing of infecting microbes, the curtailment of neutrophil recruitment, apoptosis of neutrophils, phagocytosis of apoptotic neutrophils by macrophages (efferocytosis), and a shift in the overall cytokine and chemokine milieu from an inflammatory to a resolving phenotype (11, 12). In various adult models of bacterial infection, SPM have been found to promote bacterial clearance and diminish the duration of the inflammatory response (15, 17). However, whether and how SPM may impact infection in early life is largely unknown.

We have recently established in our laboratory an infant mouse model of infection with the EPEC-related Citrobacter rodentium. EPEC is a particularly important pediatric enteropathogen since it shows the highest risk for infection-induced death in infants ≤12 months old (18). Our murine model resembles human infant EPEC infection because C. rodentium-infected infant mice are also highly susceptible to infection. Moreover, infected murine infants demonstrate profound intestinal inflammation and damage and are severely compromised in the development of specific antibody responses. Here we present compelling evidence that the SPM resolvin D1 (RvD1) and resolvin D5 (RvD5) exert substantial protective effects in infant C. rodentium infection. RvD1 and RvD5 administered 2 days postinfection (p.i.) diminished bacterial loads, ameliorated inflammation, led to the development of protective memory responses associated with mature or nearly mature B cell adaptive responses, and rescued infants from death. These findings raise the intriguing possibility that SPM may represent powerful new tools for not only the mitigation of acute disease in early life but also the prevention of recurrent childhood infection.

RESULTS

Murine infants are highly susceptible to C. rodentium infection.

EPEC is an attaching and effacing (A/E) bacterial enteropathogen that causes severe disease burden in human infants (19–22). Thus, it would be valuable to have a pediatric animal model that mimics the human disease condition. It has recently been reported (23) that murine infants are susceptible to EPEC infection. However, normal adult mice do not support EPEC infection, making comparative developmental studies impossible. Thus, we chose to study 15-day-old mouse infants, roughly corresponding to human infants during the first year of life, and infect them with the EPEC-related C. rodentium, an excellent and well-accepted experimental model for EPEC infection in adult mice (24–26). Others have reported (27, 28) that 14-day-old infant mice are more susceptible than adults to a single, relatively high infectious dose of C. rodentium. We extended these studies to determine the sensitivity of infant mice to a broad range of doses (Fig. 1A). As previously reported by many other laboratories, adult C57BL/6 mice all survived a dose of 5 × 10⁸ CFU. In striking contrast, it was necessary to reduce the inoculum to 10³ CFU before most infant mice survived (Fig. 1A). It is noteworthy that mice of this age weigh approximately 5 g, or ∼4-fold less than adults, but the reduction in dose necessary for survival was over 5 logs! These results indicate that mouse infants, like human infants, are highly susceptible to infections with A/E enterobacteria. The susceptibility of infant mice to the high lethal dose (10⁷ CFU) was associated with early high and sustained bacterial burdens in the colons of infants (Fig. 1B).

FIG 1.

Infants are highly sensitive to primary infection with C. rodentium; association with high early and sustained bacterial titers. (A) Survival of 2-week-old infant and adult C57BL/6 mice orally infected with the indicated doses of C. rodentium (n ≥ 11 mice per group in 2 independent experiments). *, P < 0.01 by log rank (Mantel-Cox) test, compared to adult survival curves. (B) CFU in the colons at the indicated days after oral infection of adults with 5 × 10⁸ CFU or infants with 1 × 10⁷ CFU. Each symbol represents an individual animal. *, P ≤ 0.01 by Mann-Whitney test.

Lethal infection of infants is associated with severe intestinal pathology and systemic dissemination.

Our earlier studies (29, 30) showed that 7-day-old neonatal mice infected with the oral pathogen Yersinia enterocolitica mounted profound innate inflammatory responses in the mesenteric lymph nodes (MLN) that exceeded those of adults. We reasoned that the high bacterial load in infants infected with C. rodentium may similarly elicit strong inflammation in the intestines. Indeed, by 10 days postinfection (p.i.), infants displayed marked inflammation in the colon, dominated by neutrophils (Fig. 2A). This inflammatory response was associated with extensive necrosis (Fig. 2B) and dissemination to systemic tissues (Fig. 2C).

FIG 2.

Infant susceptibility to C. rodentium infection is associated with late severe inflammation in the colon and systemic dissemination. Infants (INF) and adults (AD) were orally infected with 10⁷ and 5 × 10⁸ CFU, respectively, of C. rodentium and examined 10 days p.i. (A) Colon sections were stained with hematoxylin & eosin and examined for inflammation; 4 mice in each group were examined. Scores were determined on a scale of 0 to 3 (0, none; 1, mild; 2, moderate; and 3, severe). Of note, the pathologist was given only numbers for each sample and was unaware of any specifics of the experiments. INFLAM, inflammasomes; LYMPH, lymphocytes; MONO, monocytes; NEUT, neutrophils. (B) Representative sections of colons. AD, mild inflammation limited to the mucosal surface; INF, severe inflammation with necrosis extending from the mucosa to the submucosa and muscularis. Magnification, ×50. (C) Organs were dissected and homogenized for CFU counts. COL, colon; MLN, mesenteric lymph nodes; SP, spleen; LIV, liver. Each symbol represents an individual mouse. *, P ≤ 0.01 by Mann-Whitney analysis. ns, not significant.

These results indicate that infants manifest profound late inflammation that may contribute importantly to damage of the intestinal barrier, systemic dissemination, and death by sepsis.

SPM reduce bacterial loads and inflammation, rescue infants from death, and lead to adaptive responses comparable to those in adults.

RvD1 and RvD5 have been previously found to decrease bacteremia and increase survival in an adult murine peritonitis model (31). Thus, we investigated whether RvD1 and RvD5 could affect the course of C. rodentium infection in infants. As a first test of this idea, infants were pretreated with RvD1 and RvD5 2 days prior to infection and then infected with a threshold lethal dose (10⁴ CFU) or a high lethal dose (10⁷ CFU). C. rodentium titers in the colons were evaluated 2 days p.i. (Fig. 3A). Remarkably, we found no colonies whatsoever in the colons of infants infected with 10⁴ CFU and significantly reduced titers in those infected with 10⁷ CFU (Fig. 3B).

FIG 3.

Pretreatment with RvD1 and RvD5 diminishes early bacterial loads in C. rodentium-infected infants. Infant mice were left untreated or treated with a vehicle (PBS) or 100 ng each of RvD1 and RvD5 i.p. Two days later, the mice were infected with the indicated doses of C. rodentium, and CFU in the colons were assessed 2 days after that. (A) Experimental scheme. (B) Colon CFU. Symbols represent individual animals. The dashed line indicates the limit of detection of the assay. *, P < 0.01 by Mann-Whitney analysis.

To test the therapeutic potential of these resolvins, we infected infants with the lethal dose (10⁷ CFU) and treated them with a vehicle (phosphate-buffered saline [PBS]) or RvD1 and RvD5 2 days after infection. Colonic C. rodentium loads were measured at 4 and 10 days p.i. (Fig. 4A). Small but statistically significant decreases in titers were observed as early as 2 days after RvD1 and RvD5 treatment (4 days p.i.), and substantial and significant decreases were seen 10 days p.i. (Fig. 4B). Strikingly, dramatically reduced neutrophilic inflammation was detected 10 days p.i. following this single RvD1/5 treatment at 2 days p.i. (Fig. 4C and D). Moreover, and remarkably, this treatment rescued approximately 33.3% of the infected infants from death (Fig. 4E). These results indicate that SPM improve survival of infants infected with a lethal dose of C. rodentium. This may result, in part, from reduced bacterial burdens in the colons, similar to related earlier reports for infected adult mice (31–33), but also from a dramatic reduction in pathological inflammation.

FIG 4.

Treatment with RvD1 and RvD5 2 days after infection with a lethal dose diminishes bacterial loads and inflammation and improves survival in C. rodentium-infected infants. Infant mice were left untreated or treated with a vehicle (PBS) or 100 ng each of RvD1 and RvD5 i.p. 2 days after infection with 10⁷ CFU of C. rodentium. Colon CFU, inflammation, and survival were scored. (A) Scheme of the experiment. (B) Colon titers 4 and 10 days p.i. Symbols represent individual animals. #, P < 0.04; *, P < 0.01; ***, P < 0.0001 (Mann-Whitney analysis). (C) Colon sections 10 days p.i. were stained with hematoxylin and eosin and assessed as described in the legend for Fig. 2. Age-matched control infants (uninfected and untreated) showed no score (i.e., scores = 0) (D) Representative sections of colons 10 days p.i. from the indicated groups. Magnification, ×50. (E) Survival of similarly treated infants (n ≥ 12 mice per group in 2 independent experiments). *, P < 0.01 by log rank (Mantel-Cox) test.

We next tested the impact of RvD1 and RvD5 on infection with a threshold lethal dose (10⁴ CFU) of C. rodentium. Infants infected with 10⁴ CFU and treated with RvD1 and RvD5 2 days p.i. showed undetectable bacterial loads 10 days p.i. (Fig. 5B). Remarkably, 100% of infected infants treated with RvD1 and RvD5 survived, in contrast to only ∼50 to 60% survival in the untreated or vehicle-treated group (Fig. 5C). Thus, RvD1 and RvD5 treatment of infants infected with a threshold lethal dose of C. rodentium abolished bacterial burdens and promoted survival of all infants.

FIG 5.

Treatment with RvD1 and RvD5 2 days after infection with a threshold lethal dose leads to vastly diminished bacterial loads 10 days p.i. and complete survival. Infant mice were left untreated or treated with a vehicle (PBS) or 100 ng each of RvD1 and RvD5 i.p. 2 days after infection with 10⁴ CFU of C. rodentium. Colonic titers were assessed at 10 days p.i. Survival was monitored in parallel groups of animals. (A) Experimental scheme. (B) Colon CFU. Symbols represent individual animals. *, P < 0.01 by Mann-Whitney analysis. (C) Survival curves (n ≥ 8 mice per group in 2 independent experiments). *, P < 0.01 by log rank (Mantel-Cox) test.

Strikingly, early administration of RvD1 and RvD5 in infected infants also appeared to lead to the development of immunological memory, as assessed by (i) the development of vigorous adaptive responses and (ii) resistance to colonization on reinfection. First, adults were infected with 5 × 10⁸ CFU and infants were infected with the high lethal dose of 10⁷ CFU. The infected mice were treated 2 days p.i. with the vehicle or 100 ng (infants) or 400 ng (adults; to account for weight differences) each of RvD1 and RvD5. Thirty days later, all surviving animals were challenged with 5 × 10⁸ CFU; 10 days later, sera were collected and anti-C. rodentium IgG responses were tested by enzyme-linked immunosorbent assay (ELISA). Strikingly, surviving infants infected with a lethal dose (10⁷ CFU) and treated with RvDs developed serum IgG responses comparable to those of adults (Fig. 6B). Adult IgG titers in the groups treated with the vehicle and RvD1 plus RvD5 were completely overlapping; for the sake of clarity, only the titers for adults treated with RvD are shown. Second, parallel groups of treated infant mice showed significantly reduced colon bacterial loads, relative to naive (i.e., never previously infected or treated) adult mice, upon challenge 30 days following initial infection (Fig. 6C). Note that we could not obtain data for vehicle-treated infants infected with 10⁷ CFU at this late time point because they had all died by 20 days p.i. (Fig. 4E). However, this comparison can be obtained with infants infected with a threshold lethal dose (10⁴ CFU), at which some vehicle-treated animals survive. Under these conditions, significant protection against reinfection developed in the infants treated with RvD1 plus RvD5 (Fig. 6C). Moreover, memory IgG titers in the infants infected with 10⁴ CFU and treated with RvD1 plus RvD5 were markedly increased over those in vehicle-treated infants, largely achieving the levels seen in adults (Fig. 6D). It is well established that B cells and IgG contribute importantly to immunity against C. rodentium in adult mice (34–37). Thus, a synthesis of these observations indicates that the promotion of mature IgG levels by SPM in infected infant mice may markedly enhance protective immunity against reexposure.

FIG 6.

Treatment of infant mice with RvD1 and RvD5 after infection leads to vigorous memory responses. Adult mice were infected with 5 × 10⁸ CFU; infant mice were infected with 10⁷ or 10⁴ CFU. Two days p.i., mice were treated with a vehicle or 100 ng (infants) or 400 ng (adults) each of RvD1 and RvD5 i.p. Thirty days later, all mice were challenged with 5 × 10⁸ CFU, and 10 days later, sera were collected and colon CFU measured. Serum anti-C. rodentium IgG responses were measured by ELISA. (A) Experimental scheme. (B) Serum IgG titers in infants (inf) infected with 10⁷ CFU and in infected adults (Ad). (C) Colon titers in the indicated mice. *, P < 0.01 by Mann-Whitney analysis. (D) Serum IgG titers in infants infected with 10⁴ CFU and in infected adults. In panels B and D, IgG titers for vehicle-treated adult mice are not shown due to complete overlap with the IgG titers in RvD1/5-treated adult mice. Symbols represent individual animals.

DISCUSSION

We have described a murine infant model of infection with the intestinal pathogen EPEC, using the related mouse-specific bacterium C. rodentium. As with human infants and EPEC infection, murine infants were highly susceptible to C. rodentium infection and developed severe inflammation dominated by neutrophils. At late time points postinfection, infant intestines showed extensive necrosis, and this was associated with systemic dissemination of the bacteria. It seems likely that the infected infants ultimately succumbed to sepsis due to damage to the intestinal barrier and bacterial spread throughout the body. Using this system, we investigated the effects of exogenous treatment with two SPM, RvD1 and RvD5, on the course of infection in infants. Strikingly, postinfection treatment with RvD1 and RvD5 reduced the bacterial burdens, alleviated inflammation, and rescued the infants from death. In addition, infants treated early postinfection with RvD1 and RvD5 were protected against reinfection, and this was linked to the development of C. rodentium-specific IgG responses comparable to those in adults. These results identify SPM as outstanding candidates for treating infectious intestinal diseases in human infants.

SPM have been shown to accelerate the clearance of pathogens and increase survival in numerous infection models in adult mice (13, 15). Bacterially mediated diseases include pneumonia (38–40), peritonitis (31, 32, 41, 42), skin infections (43), and cecal ligation and puncture-induced sepsis (33, 44, 45). However, this is the first demonstration that these SPM have potent activity in a setting of intestinal infection with a bacterial enteropathogen. Furthermore, this is the first description of the ability of resolvins to relieve disease burden and dramatically improve survival in infant intestinal disease. Importantly, this approach has strong potential applicability to human infants, as supported by observations that infants treated with topical SPM (46) or their oral precursors (47) show, respectively, reduced severity of skin disease or reduced incidence of respiratory or diarrheal illnesses.

The dramatic effects of RvD1 and RvD5 on survival in infant infection may arise through multiple processes. First, as in infected adult animals (31, 33, 38, 41–43), SPM treatment postinfection reduced the bacterial load in infected infants. This is unlikely to be due to a direct antibacterial effect, since our earlier studies demonstrated that SPM lack microbicidal activity (45). Thus, it is possible that the resolvins may act in infants as they do in adults—to increase phagocytosis of bacteria by innate immune cells. Interestingly, the majority of findings support the idea that under physiological conditions, infant and adult phagocytes have comparable capacities for phagocytosing and killing various species of bacteria (48–53). Thus, infant phagocytes appear to be able to respond to resolvins and increase their phagocytic capacity, in a manner similar to that of adults. Together, these findings indicate that the receptors for RvD1 and RvD5 and their downstream signaling pathways may be fully mature by mid-infancy. Of interest, human breast milk contains RvD1 and RvD5 (54, 55), which may function in the newborn. Second, the major reduction in intestinal inflammation induced by RvD1 and RvD5 treatment almost certainly contributes to the enhanced survival of infants. Indeed, resolvin treatment appears to allow maintenance of the intestinal barrier, as systemic dissemination of the bacteria was completely abolished in treated infants (see Fig. S1 in the supplemental material). How inflammation is so substantially reduced is currently unclear. Although there are a number of potential mechanisms, the possibility that SPM mediators enhance suboptimal efferocytosis in infants is compelling. This idea is supported by observations that infants with protracted bacterial bronchitis show dysfunction in alveolar macrophage efferocytosis (56). This idea is under investigation.

Antibody responses in early life are generally reduced in both quality and quantity compared with those in adults (reviewed in references 57–62). Achieving robust IgG response is the major goal of many pediatric vaccines, and the importance of B cells and IgG responses in protection against extracellular microbial pathogens, such as C. rodentium (34–37), is well established. Thus, our observations that RvD1 and RvD5 support the development of resistance against reinfection associated with nearly mature IgG responses have major implications for infant vaccine responses. How these mature responses are elicited is not currently known, but it is tempting to speculate that resolvins may support the development of mature germinal-center (GC) reactions in the intestinal lymphoid tissues. This possibility is based on observations (63, 64) by Mastelic and colleagues that the poor IgG responses of neonatal mice to systemic vaccination with inert antigens are linked with poor GC development and that T follicular helper cell and GC responses can be substantially boosted with specific adjuvant. Thus, it will be of considerable interest to examine the effects of SPM on GC formation during infant intestinal bacterial infection.

It is well appreciated that bacterial and viral infections of infants often lead to profound inflammation (65–67). Yet, most in vitro tests of innate immune function in infancy show depressed or deficient proinflammatory function (68, 69). Of course, in vitro tests largely read out the initiation of inflammation and are not designed to measure the accumulation of inflammation. Perhaps our results provide insights into a possible reconciliation of these observations. That is, that the severe inflammation often seen in infant infections in vivo is largely independent of early events in inflammation and, instead, occurs due to immaturity in the later phases of inflammation.

MATERIALS AND METHODS

Mice.

Adult C57BL/6 were purchased from Harlan Laboratories. All mice were bred and housed under barrier conditions in the Division of Veterinary Resources of the University of Miami Miller School of Medicine, Miami, FL. Mice were regularly screened for specific common pathogens. Adult mice (6 to 10 weeks of age) and infant mice (15 days of age) were used in experiments. Infant mice were weaned at 3 1/2 weeks of age. All mice were handled in compliance with the Institutional Animal Care and Use Committee (IACUC) of the University of Miami Miller School of Medicine.

Bacterial infections and SPM treatments.

GPM1831a (gift from G. P. Munson, University of Miami) is a kanamycin-resistant derivative of C. rodentium DBS100 (ATCC 51459). It was constructed by lambda Red-mediated recombination of a kanamycin cassette into the noncoding region ca. 400 bp upstream of an unnamed gene encoding a putative membrane protein (NCBI database accession number KIQ50635). The bacteria were grown at 37°C in Luria-Bertani broth (LB) medium or MacConkey agar (Sigma-Aldrich, USA) plates containing 25 μg/ml of kanamycin. Adults were inoculated orogastrically with 5 × 10⁸ CFU using a 22-gauge, round-tipped feeding needle (Fine Science Tools, Foster City, CA) attached to a 1-ml syringe (Becton Dickinson, Franklin Lakes, NJ). Infants were inoculated orogastrically with the desired doses using PE-10 tubing (polyethylene tubing with an outside diameter of 0.61 mm [Clay Adams, Sparks, MD]) attached to a 30-gauge needle and Hamilton syringe (70). The actual administered dose was determined by plating serial dilutions of the suspensions on Luria broth plates and incubating them for 24 h at 37°C.

RvD1 and RvD5 were obtained from Cayman Chemical Company (Ann Arbor, MI). The RvDs were stored in undiluted aliquots in the dark at −80°C. Upon thawing, the RvDs were diluted in cold sterile PBS to 5 mg/ml and used immediately. Infant mice were injected intraperitoneally (i.p.) with 20 μl containing 100 ng each of RvD1 and RvD5 using a Hamilton 25-μl gas-tight syringe with a 30-gauge attachable needle; adult mice were injected i.p. with 100 μl containing 400 ng each of RvD1 and RvD5 using a 1.0-ml syringe and a 25-gauge needle. The times of injection relative to the day of infection are indicated elsewhere in the text and figure legends.

Bacterial enumeration from organs of infected mice.

To measure C. rodentium titers, colons, livers, or spleens were weighed and homogenized in Hanks balanced salt solution (HBSS) using a Seward Biomaster 80 stomacher (Brinkman, Westbury, NY) for 4 min at high speed. Individual mesenteric lymph nodes were homogenized in 400 μl (neonates) or 500 μl (adults) of HBSS using a VWR disposable pellet mixer with cordless motor (VWR International). C. rodentium titers were enumerated by plating dilutions of homogenates on MacConkey agar containing 25 μg/ml of kanamycin.

Histology and inflammation.

Sections of colon were fixed in 10% neutral buffered formalin, sectioned, and stained with hematoxylin and eosin. The slides were examined for histological changes by a board-certified veterinary pathologist. Slides were assessed blindly; inflammation scores were determined on a scale of 0 to 3 (0, none; 1, mild; 2, moderate; and 3, severe).

ELISA.

Overnight (o.n.) cultures of C. rodentium were homogenized in bicarbonate buffer, pH 9.5, using the VWR disposable pellet mixer. ELISA plates were coated o.n. at room temperature with 100 μl of 20-μg/ml bacterial lysate in bicarbonate buffer. The wells were blocked with PBS containing 2% bovine serum albumin (BSA) for 1 h at room temperature and washed, and serum dilutions in PBS were incubated o.n. at room temperature. Rabbit anti-mouse IgG-peroxidase (Sigma; A9044) was added for 2 h at room temperature, and the wells were developed with TMB solution (Life Technologies) for 30 min.

Statistical analyses.

All experiments were performed at least two times. Statistical tests were performed using GraphPad Prism software, as follows: Mann-Whitney test for the bacterial colonization experiments and log rank (Mantel-Cox) test for the survival experiments. The significance threshold was a P value of ≤0.05.