Streptococcus pneumoniae–Induced Pneumonia and Citrobacter rodentium–Induced Gut Infection Differentially Alter Vitamin A Concentrations in the Lung and Liver of Mice^,

Abstract

In the developing world, vitamin A (VA) deficiency is endemic in populations that are also at great risk of morbidity and mortality because of pneumococcal pneumonia and enteric infections. To better understand how lung and gastrointestinal pathogens affect VA status, we assessed VA concentrations in serum, lung, and liver during an invasive pneumonia infection induced by Streptococcus pneumoniae serotype 3, and a noninvasive gut infection induced by Citrobacter rodentium, in vitamin A–adequate (VAA) and vitamin A–deficient (VAD) mice. For pneumonia infection, mice were immunized with pneumococcal polysaccharide serotype 3 (PPS3), or not (infected-control), 5 d prior to intranasal inoculation with S. pneumoniae. Two days post-inoculation, immunization was protective against systemic infection regardless of VA status as PPS3 immunization decreased bacteremia compared with infected-control mice (P < 0.05). Retinol concentrations in the lung were higher in infected-control VAA mice (15.7 nmol/g: P < 0.05) compared with PPS3-immunized mice (8.23 nmol/g), but this was not associated with increased lung bacterial burden. VAA mice had reduced severity of C. rodentium–induced gut infection as measured by fecal bacterial shedding compared with VAD mice (P < 0.05). Liver retinol and retinyl ester concentrations in VAA mice decreased at the peak of infection (retinol, 8.1 nmol/g; retinyl esters, 985 nmol/g; P < 0.05, compared with uninfected mice; retinol, 29.5 nmol/g; retinyl esters, 1730 nmol/g), whereas tissue VA concentrations were low in VAD mice during both infections. Colonic mucin gene expression was also depressed at peak infection compared with uninfected mice (P < 0.05). Overall, pneumonia had less effect on VA status than gastrointestinal infection, predominantly owing to reduced hepatic VA storage at the peak of gut infection.

Introduction

Pneumonia, enteric infections, and vitamin A (VA)⁸ deficiency are all major public health problems that annually affect millions, globally increasing morbidity and mortality (1–5). Streptococcus pneumoniae is the main causative pathogen of bacterial pneumonia, and S. pneumoniae serotype 3 (ST3) is a virulent serotype that has been predicted to become increasingly common in human populations (6, 7). Intranasal inoculation of mice with ST3 causes invasive disease characterized by bacteremia and, in nonimmunized mice, a 100% mortality rate by 3 d post-inoculation (dpi) (8). Citrobacter rodentium is a gram-negative bacterium and is used as a murine model of attaching and effacing lesions that exhibits a pathogenesis similar to the human enteropathogenic Escherichia coli (4, 5). Colonization of the apical surface of the colon occurs by 7 dpi with C. rodentium, causing hyperplasia and epithelial thickening similar to inflammatory bowel disease (4, 5, 9). Bacterial clearance is mediated by both the innate, i.e., microbiota, antimicrobial defensins, and adaptive, i.e., the T helper (Th)/Th17 proinflammatory response immune mechanisms Citrobacter clearance occurs around 28 dpi in most mice (5, 9–11).

VA deficiency causes immune system abnormalities and decreased response to infectious diseases (2, 12). Adequate VA status allows for normal lymphocyte responses such as homing to a site of inflammation and maintenance of the mucosal epithelial barrier (13–15). Serum VA in the form of retinol has been previously demonstrated to decline in response to inflammation induced by LPS and IL-6 in rats (16, 17). The liver is the predominant storage organ for VA in the form of retinyl esters, and the lung is a secondary storage site. Hepatic mobilization of VA is reduced in response to inflammation induced by treatment with LPS or recombinant human (rh)IL-6 (17).

The mouse is the most commonly used experimental model for immunology research, yet relatively little is known concerning VA status in response to infection in this model. Previous studies have not been performed to determine VA status in mouse tissues during bacterial mucosal infections. In our present studies, we used mouse models of an invasive lung pathogen, ST3, and a noninvasive gastrointestinal pathogen, C. rodentium, to examine tissue VA status during infection. We hypothesized that there would be a decline in serum retinol concentrations because of the increased need of VA at the tissue sites of infection. Therefore, we characterized VA in serum, liver, and lung during pneumonia and in serum, liver, lung, and colon during C. rodentium infection.

Methods

Animals and generation of vitamin A–deficient mice.

Mice were housed in a pathogen-free environment and exposed to a 12-h light, 12-h dark cycle with ad libitum access to food and water. All mouse experiments were performed in agreement with the Pennsylvania State University’s Institutional Animal Care and Use Committee. Vitamin A–deficient (VAD) mice were bred as described previously (18). Pregnant dams were fed an AIN-93–modified purified diet (19) adequate in vitamin A (VAA), containing 4 μg retinol/g diet, or the same diet without vitamin A (Research Diets Inc.). Both dams and pups were fed the VAA or VAD diet until the time of weaning, after which dams were fed a normal chow diet (Purina 4.45 μg retinol/g) for 2 wk to replenish vitamin A stores before rebreeding; weanlings were maintained on their respective VAA or VAD diets. At 7 wk of age, blood was collected from the mandibular facial vein and pooled serum samples were analyzed by ultra-high performance liquid chromatography (UPLC) to ensure VA deficiency. Deficient status was defined as serum retinol < 0.35 μmol/L, whereas >1.05 μmol/L was considered VA–adequate. Eight- to 9-wk-old male and female BALB/c mice were used to study VA status changes during pneumococcal infection, and 8-wk-old C57/BL6 mice of both sexes were used to study C. rodentium infection.

Pneumococcal pneumonia infection model.

ST3 (ATCC 6303) was reconstituted with 1 mL Todd Hewitt broth (BD Difco) and grown for 24 h to log phase in Todd Hewitt broth at 37°C with an atmosphere of 5% CO₂ (8). A loop full of bacteria was plated on trypicase soy agar plates containing 5% sheep blood (BD Diagnostic Systems) for overnight growth. In the morning, a single colony of bacteria was selected, grown to mid-log phase in Todd Hewitt broth for 6 h, and frozen at −80°C in a final concentration of 25% glycerol (20).

As previously described, to prevent mortality from pneumonia, mice in the pneumonia studies were immunized i.p. with 0.5 μg of pneumococcal polysaccharide serotype 3 (PPS3) in 0.5 mL PBS 5 d prior to infection (8). Mice mock-immunized with vehicle (PBS) served as infected controls. An aliquot of frozen bacteria at mid-log phase was thawed, washed, and centrifuged at 1500 × g for 15 min twice in PBS. Mice were briefly anesthetized with isoflurane and intranasally inoculated by applying an infectious dose of 5 × 10⁴ CFU in 50 μL of PBS to the nares. Uninfected mice were mock-infected with 50 μL of PBS. Pneumonia severity was characterized at 2 dpi; mice were killed by CO₂ asphyxiation, and the lungs, liver, and blood from the vena cava were collected. The left lobe of the lung and liver were snap frozen in liquid nitrogen and stored at −80°C for later analysis. The right lobes were homogenized in 1 mL of PBS, and serial dilutions were plated overnight at 37°C on trypticase soy agar plates containing 5% sheep blood (BD Biosciences). To determine septicemia, 10 μL of whole blood was diluted in 90 μL of PBS and 3 serial dilutions were plated and grown for 12 h at 37°C (20), after which bacterial colonies were counted.

Serum PPS3–specific IgM ELISA.

An ELISA, described previously (8), was used to measure IgM specific to PPS3 in serum samples collected from mock-infected and ST3-infected mice (PPS3-immunized or not) at 2 dpi.

C. rodentium gut infection model.

VAA and VAD mice were used in the kinetic study in which mice were killed before infection (day 0, uninfected), at the peak of infection (9 dpi), and after clearance of infection (30 dpi). Uninfected VAA mice were used as age-matched controls to compare the age effects of VA status at clearance time point. C. rodentium was grown overnight in Luria-Bertani medium at 37°C to stationary phase, and anesthetized mice were administered an oral gavage dose of 5 × 10⁹ CFU in 200 μL PBS. An aliquot of the infectious dose was serially diluted in PBS and plated on Luria-Bertani agar containing 50 g/L naladixic acid (21). Mice were housed 1 per cage throughout the duration of infection, and bacterial shedding was evaluated by collecting fecal pellets, which were weighed and emulsified in PBS to 100 g/L. Serial dilutions were plated on Luria-Bertani media with 50 g/L naladixic acid, and plates were counted after overnight incubation at 37°C. Mice were killed using CO₂, and blood, lungs, liver, and distal colon were collected for further analyses.

Determination of vitamin A status.

Blood was collected from uninfected controls and at the peak and clearance time point of infection with C. rodentium, as well as from age-matched uninfected controls at the time of bacterial clearance. Total retinol concentration in serum and colon was determined after saponification by UPLC using trimethylmethoxyphenyl-retinol as an internal standard (16). Liver and lung retinol and retinyl esters from infected tissue homogenates were determined by UPLC using trimethylmethoxyphenyl-retinol as an internal standard (16).

qPCR expression of mucin in colon.

Briefly, RNA was isolated from 0.5 mg of colon tissue with TRiZOL (Invitrogen) and reverse transcribed into cDNA with reverse transcriptase (Promega), and qPCR was performed using iQ SYBR Green supermix (BioRad). Primer sets melting temperatures (Tm) and product size (bp) for mucin 2 and 3 (Muc2, Muc3) are listed as follows. For Muc2, the 5′-primer 5′-GCTGACGAGTGGTTGGTGAATG-3′ and 3′-primer 5′- GATGAGGTGGCAGACAGGAGAC-3′ were used for 40 cycles of qPCR with a Tm of 60°C, to yield a 175-bp product. For Muc3, the 5′ -primer 5′ –CGTGGTCAACTGCGAGAATGG-3′ and 3′ -primer 5′- CGGCTCTATCTCTACGCTCTCC-3′ were used for 40 cycles of qPCR with a Tm of 60°C, to yield a 200-bp product. mRNA expression is shown as ΔΔCT or change in cycle threshold values as compared with controls normalized to expression of the reference gene 18s rRNA (22, 23).

Statistical Analysis.

Statistical analyses were conducted using Prism 5.0 software (GraphPad). A Student’s t test was used to analyze serum and colon retinol values between uninfected VAA and VAD mice at 7 wk of age. One-factor ANOVA with Bonferroni’s multiple comparison post-hoc test was used to analyze serum, lung, liver, and colon retinol and lung and liver retinyl ester concentrations, as well as colonic Muc2, Muc3 gene expression values. Two-factor ANOVA with Bonferroni’s multiple comparison post-hoc test was used to analyze C. rodentium fecal shedding, as well as pneumonia lung and blood bacterial counts and serum IgM concentrations. A P value of < 0.05 was considered significant. Values in the text are means ± SDs.

Results

Duration and severity of a noninvasive gut infection induced by C. rodentium and an invasive lung infection caused by ST3 in VAA and VAD mice.

Bacterial fecal shedding and body weight were monitored throughout C. rodentium infection. Fig. 1A depicts a typical bacterial shedding curve for mice infected with C. rodentium. Shedding peaked on days 9–14, and the infection was cleared by 30 dpi. VAA mice cleared C. rodentium infection more quickly than VAD mice and had significantly less fecal shedding at 18, 21, 23 (P < 0.0001), and 25 dpi (P < 0.05, in comparison with VAD mice, Fig. 1A). In the pneumonia model, an invasive infection with virulent ST3, studies were limited to 2 d and mice were immunized with pneumococcal polysaccharide, PPS3, prior to infection to allow for survival (8). Immunization with PPS3 decreased bacteremia in both VAA and VAD mice (P < 0.05, compared with infected control, Fig. 1B), although there was no significant difference in lung CFU (Fig. 1C). Immunization increased serum IgM specific to PPS3 (Fig. 1D), although it decreased bacteremia rates independently of VA status (Fig. 1B).

FIGURE 1.

Duration and severity of C. rodentium gut infection (A), pneumococcal infection in lung (B) and blood (C), and serum antibodies (D) in VAA and VAD mice. Bacterial shedding of C. rodentium in feces from VAA (n = 7) and VAD (n = 5) mice throughout primary infection (A). The curve is representative of a thrice-repeated experiment. The interaction of diet and infection was significant, P < 0.001. For the pneumonia infection model, mice were immunized with vehicle or PPS3 5 d prior to infection with ST3. Bacterial burden at 2 dpi with ST3 in infected control and PPS3-immunized VAA (n = 15) and VAD (n = 20) mice in blood (B) and lung (C). Each symbol represents 1 mouse, the horizontal line represents the mean, and the error bars denote ± SEMs. Graphs show data from 3 combined experiments. Serum PPS3-specific IgM in uninfected mice and at 2 dpi after ST3 infection in infected control and PPS3-immunized mice (n = 7 per group) (D). Absorbance (OD) values were normalized to VAA mock-infected controls set at 1.0. Error bars denote ± SEMs. Significant results from 1- or 2-factor ANOVA are indicated for each panel, and within each panel means without a common letter differed significantly, P < 0.05, by post-hoc analysis. dpi, days post-inoculation; PPS3, pneumococcal polysaccharide serotype 3; ST3, Streptococcus pneumoniae serotype 3; VAA, vitamin A–adequate; VAD, vitamin A–deficient.

Serum retinol reflected diet but was not affected by either infection.

To confirm VA status, serum retinol was measured in 7-wk-old mice. Serum retinol concentrations preinfection were determined for both mouse strains (Fig. 2). Serum retinol concentrations in 7-wk-old VAA mice were significantly greater than in VAD mice (P < 0.001, Fig. 2A for C. rodentium infection in C57BL/6 mice and Fig. 2B for pneumonia infection in BALB/c mice). In both strains, VAD mice had low serum retinol (<0.35 μmol/L), characteristic of VA deficiency, confirming our dietary models. Serum retinol did not differ throughout C. rodentium infection in VAA mice (Fig. 2C), or by infection or immunization status during the acute phase of invasive pneumonia infection in VAA mice (Fig. 2D). Although VAD mice did not have detectable liver or lung VA stores, they did not exhibit any outward manifestations of VA deficiency such as weak limbs or ocular abnormalities by 16 wk of age, even though serum retinol on average was depressed (0.14 ± 0.05 μmol/L) at this time. Serum retinol did not differ by infection status of the lung or the gastrointestinal tract in either VAD or VAA mice.

FIGURE 2.

Serum retinol concentrations before (A,B) and during gut (C) and lung (D) infections in mice. Serum retinol concentrations at 7 wk of age in VAA and VAD mice (n = 7 per group) (A). Error bars denote ± SEMs, and different letters indicate a significant difference in retinol content, P < 0.0001. Serum retinol concentrations in VAA C57/BL6 mice are shown before infection (uninfected, n = 5), at the peak of infection (9 dpi, n = 7), and after clearance (30 dpi, n = 7) of C. rodentium infection, and in age-matched uninfected controls at the same time point as for clearance in infected mice (n = 4) (B). Serum retinol concentrations in 7-wk-old VAA and VAD BALB/c mice (C). Serum retinol in VAA BALB/c mice is shown for mock-infected controls, at 2 dpi in infected controls or PPS3-immunized, infected mice (n = 7 per group) (D); mice were immunized with vehicle or PPS3 5 d prior to infection with ST3. The horizontal line is the mean and error bars denote ± SEMs. Data shown are representative of 3 separate in vivo experiments. Age-match uninf., age-matched uninfected controls; dpi, days post-inoculation; PPS3, pneumococcal polysaccharide serotype 3; ST3, Streptococcus pneumoniae serotype 3; VAA, vitamin A–adequate; VAD, vitamin A–deficient.

Lung retinol was lower during the peak of C. rodentium infection, but increased at 2 dpi with S. pneumoniae infection.

To further characterize the VA status of mice during pneumonia and C. rodentium infections, retinol and retinyl esters in the lung and liver were determined. With C. rodentium infection, lung retinol was lower at the infection peak (9 dpi) and at the time of clearance (30 dpi) in comparison with age-matched uninfected controls at clearance time point (P < 0.05, Fig. 3A), which implies an age effect related to diet rather than infection. Lung retinyl esters accumulated with age as values were lower in uninfected and peak-infection lung tissue than in age-matched uninfected controls at the clearance time point (P < 0.05, Fig. 3B). Pneumonia infection increased lung retinol in VAA-infected control mice at 2 dpi compared with mock-infected mice (P < 0.05, Fig. 3C), although retinyl ester concentrations did not differ as a result of infection or immunization (Fig. 3D).

FIGURE 3.

Lung retinol (A,C) and retinyl ester (B,D) concentrations during gut (A,B) or lung (C,D) infections in mice. In VAA C57/BL6 mice, lung retinol (A) and retinyl ester (B) were determined before infection (uninfected, n = 5), and at peak (n = 8), and clearance (n = 7) of C. rodentium infection and from age-matched uninfected controls at the clearance time point (n = 4). In VAA BALB/c mice, lung retinol (C) and retinyl ester (D) were determined in mock-infected mice (n = 7), at 2 dpi with ST3 in infected controls (n = 7), or in PPS3-immunized mice (n = 7); mice were immunized with vehicle or PPS3 5 d prior to infection with ST3. Symbols represent individual mice, and data are representative of 3 separate in vivo experiments. Horizontal line is the mean and error bars denote ± SEMs. Within each graph means without a common letter differ significantly, P < 0.05. Age-match uninf., age-matched uninfected controls; dpi, days post-inoculation; PPS3, pneumococcal polysaccharide serotype 3; ST3, Streptococcus pneumoniae serotype 3; VAA, vitamin A–adequate.

Liver retinol decreased at the peak and remained low after clearance, whereas retinyl esters declined at the peak but returned to preinfection concentrations after C. rodentium clearance.

Liver retinol decreased significantly at peak and clearance times after C. rodentium infection in comparison with uninfected control mice (P < 0.01, Fig. 4A), although no change in liver retinol was observed during pneumonia infection (Fig. 4B). However, liver retinol did not return to the concentration of uninfected controls after clearance and was also lower in age-matched uninfected controls [P < 0.05 compared with uninfected (and younger) controls, Fig. 4A]. Liver retinyl esters were lower at the peak of C. rodentium infection versus at clearance and in uninfected age-matched controls (P < 0.05, Fig. 4C), but did not differ in the S. pneumoniae infection model (Fig. 4D).

FIGURE 4.

Liver retinol (A,C) and retinyl ester (B,D) concentrations during gut (A,B) or lung (C,D) infections in mice. In VAA C57/BL6 mice, liver retinol (A) and retinyl ester (B) were determined before infection (uninfected, n = 5), and at peak (n = 8), and clearance (n = 7) of C. rodentium infection and from age-matched uninfected controls at the clearance time point (n = 4). In VAA BALB/c mice, liver retinol (C) and retinyl ester (D) were determined in mock-infected mice (n = 7), at 2 dpi with ST3 in infected controls (n = 8), or in PPS3-immunized mice (n = 8); mice were immunized with vehicle or PPS3 5 d prior to infection with ST3. Symbols represent individual mice, horizontal line is the mean, error bars denote ± SEMs, and data shown are representative of 3 separate in vivo experiments. Different letters indicate a mean significant difference, P < 0.05. Age-match uninf., age-matched uninfected controls; dpi, days post-inoculation; PPS3, pneumococcal polysaccharide serotype 3; ST3, Streptococcus pneumoniae serotype 3; VAA, vitamin A–adequate.

Total colon retinol remained constant in VAA mice throughout infection, whereas Muc2 and Muc3 gene expression was lower at the peak of C. rodentium infection.

To determine if VAD mice had significantly less VA in the distal colon, the site of C. rodentium colonization, we measured total retinol in the distal colon from 7-wk-old VAD and VAA mice. Total retinol was nearly undetectable in the colon of VAD mice (P < 0.05 compared with VAA, Fig. 5A). We also assessed total retinol in the colon before, at peak, and after clearance of C. rodentium infection and found no significant changes in VAA mice (Fig. 5B). The expression of colonic mucin was assessed to determine if alteration in VA status from gut infection also perturbs mucosal barrier function at the site of infection. Muc2 and Muc3 are the transcript variants of mucin most highly expressed in rodents (24). In VAA mice, Muc2 (Fig. 5C) and Muc3 (Fig. 5D) gene expression decreased at the peak of C. rodentium infection (P < 0.05, compared with uninfected controls); however, assessment of H&E-stained colon sections determined no difference in severity of inflammatory infiltrate or edema (data not shown).

FIGURE 5.

Colon retinol concentrations (A,B) and mucin Muc2 (C) and Muc3 (D) gene expression levels during gut infection in mice. Colon total retinol was determined in 7-wk-old VAA and VAD C57/BL6 mice (n = 3 per group) prior to infection (A). Colon total retinol in mice before (uninfected, n = 3), at peak (n = 8), and at the time of clearance (n = 7) of C. rodentium (B). Colonic expression of Muc2 (C) and Muc3 (D) mRNA before (uninfected, n = 6), at peak (n = 7), and after clearance (n = 8) of C. rodentium in VAA mice. Data are the fold change in comparison with baseline uninfected values. The horizontal line is the mean, error bars denote ± SEMs, and different letters indicate a significant difference, P < 0.05. VAA, vitamin A–adequate; VAD, vitamin A–deficient; Muc2, mucin2; Muc3, mucin3.

Discussion

To our knowledge, bacterial pneumonia and gut infection have not been previously characterized in mice with varying VA statuses. We observed no protective or detrimental effect due to diet alone on pneumococcal pneumonia infection in the acute phase, but an increase in fecal bacteria shedding, and thus in disease severity, in VAD mice during C. rodentium infection. Other studies that have examined responses to enteric infection in rat models of VA deficiency have reported a more drastic loss of immune function (25, 26). VAD rats monocolonized with E. coli exhibited increased diarrhea and bacterial translocation at day 3 dpi in part due to decreased function of the intestinal brush border enzymes lactase, sucrase, γ-glutamyl-transpeptidase, and dipeptidyl peptidase IV, in comparison with VAA rats (25). Furthermore, VAD rats inoculated with Salmonella typhimurium had increased intestinal damage from a loss of both humoral and cellular immune function (26). These discrepancies in infection severity using E. coli and salmonella in rats vs. our findings conducted in mice may be due to differences in VA metabolism and catabolism rates between different rodents, but could also be due to differences in the infection models.

Although we observed a decrease in colonic mucin expression at the gene level in VAA mice at the peak of infection, we did not observe dramatic changes in colon histopathology at the tissue level (data not shown). A previous study that characterized the differences in the gut microbiota between VAD and VAA rats, without additional active pathogenic gut infection, observed decreased Muc2 mRNA expression in the jejunum, ileum, and colon as a result of deficiency, but, conversely, increased Muc3 gene expression was measured in the ileum and colon (24). We also noted differences in goblet cell location and shape, because deficiency caused cells to be more scattered along the colon instead of located solely in the lower crypt, as well as an enlargement in the “cup” area of the goblet cell (24). We did not detect any differences in Muc2, Muc3 mRNA expression as a result of diet between adult VAD and VAA mice throughout C. rodentium infection (data not shown).

Serum retinol concentrations, the most simple method of VA status assessment, were not reduced in VAA mice in our studies during either lung or gut infections. This is in contrast to previous investigations, in which others have reported serum retinol’s diminishing in response to the immunostimulatory effects of LPS or rhIL-6 treatment in rats (16, 17), or in humans who have active bacterial or viral infections (14, 20, 27). By using a model-based compartmental analysis approach to examine whole body VA kinetics during inflammation in response to LPS infusion for 3 d or rhIL-6 treatment for 7 d in rats, Gieng et al. (17) determined that serum retinol declined because of decreased mobilization of VA from the liver, which may be in part due to reduced retinol-binding protein (RBP) synthesis (16). However, an active bacterial infection is more complex and multifaceted than inflammation from endotoxemia. It should be noted that a hepatic acute phase response was induced in our pneumonia model, as shown by elevations in serum amyloid A protein, SAA, and hepatic Saa1 gene expression (8). It is possible that the mouse is less susceptible to inflammation-induced hyporetinolemia than other species. Alternatively, VA may be mobilized to immune organs for use during infection and retinol may constantly recycle to the serum from a storage site or “pool” other than the liver or lung (e.g., adipose tissue, bone, etc.), such that changes in serum VA are undetectable. Although the aforementioned idea is speculative, further studies are warranted using VA tracer kinetics to characterize alterations in VA throughout the body during bacterial infection.

An interesting observation from this research is that a fatal pneumonia infection did not alter VA status as greatly as a nonfatal gastrointestinal infection. More drastic changes in VA status in storage organs were observed as a result of C. rodentium–induced gut infection, which is most proximal to the site of intestinal VA absorption. Alterations in VA status in the liver during C. rodentium infection may be due to acute liver inflammation. Raczynski et al. (28) reported an increase in liver pathology, serum liver transaminases, and hepatic cytokines, including IL-1β, granulocyte-colony stimulating factor, keratinocyte chemoattractant, monocyte chemotactic protein-1, macrophage inflammatory protein-1 α, and regulated on activation, normal T-cell expressed and secreted at 3 dpi with C. rodentium in mice. Cytochrome P450 family members have been previously reported to be altered in the liver and kidney as a consequence of C. rodentium infection (30), but hepatic cytochrome P (Cyp)26 enzymes that oxidize retinoic acid to control cellular concentration (29) have not been characterized throughout infection. Further experiments could define the mechanisms by which VA status is altered during enteric C. rodentium infection by examining expression and function of RBP, lecithin:retinol acyltransferase, or Cyp450 family members. We have investigated these (RBP, lecithin:retinol acyltransferase, and Cyp26a1) at the mRNA level in liver tissue during C. rodentium infection because the liver is the main tissue storage site of VA and also the organ responsible for RBP production (29), but did not observe a difference in expression in VAA mice (data not shown). Perhaps regulatory molecules responsible for VA metabolism and catabolism are altered earlier during the colonization period (<7 dpi) of C. rodentium infection as the abovementioned report characterized inflammation at 3 dpi (28).

An assumption of the pneumococcal pneumonia infection model is that infected control mice should have greater disease severity as measured by increased lung bacterial burden and bacteremia than PPS3-immunized, infected mice (8). However, the slight increase in lung retinol in infected control mice cannot be attributed to more severe bronchopneumonia because significantly decreased lung CFU counts were not observed in all PPS3-immunized groups compared with infected control mice (8). ST3 may be too virulent a serotype to study changes in VA metabolism as infected control mice are very sick by 2 dpi and succumb to infection by 3 dpi (8). In future studies that aim to examine the effects of VA status on pneumococcal pneumonia outcome in mice, a less virulent serotype of S. pneumoniae, such as serotype 11A (M10 strain), 14 (DW14 strain), or 19 may provide a less severe model (20, 31, 32) that is more conducive to longer-term studies of VA metabolism.

We previously demonstrated that IgM binding directly to ST3 is one mechanism by which PPS3-immunized mice were protected from infection with this deadly serotype of S. pneumoniae (8). Conversely, a prior study (33) reported the inability of VAD rats to make primary IgM in response to immunization with PPS3, which is a TI-2 antigen, whereas in the current study PPS3-immunized VAD mice produced IgM. Moreover, in a mouse model of influenza A virus, a high level of dietary VA resulted in increased Th2 cytokines and IgA, but this was not helpful for controlling the viral infection, which requires a Th1 response (34). It is possible that a diet high in VA may not be helpful in controlling pneumonia infection because of increased immunosuppressive T-regulatory cells or Th2 cells, when, instead, a mixed Th1/Th17 response is necessary for a favorable clinical outcome (35). However, we characterized pneumonia infection only during the acute phase as death was observed in all infected control (nonimmunized) mice by 3 dpi, which is too early to determine changes in the adaptive immune response.

In summary, VA deficiency increased the severity of gut infection, but VAD mice did not exhibit worse pneumonia symptoms than VAA mice. Although pneumonia infection was fatal to nonimmunized mice, VA status was less affected than in a model of nonfatal gut infection. In the lung, retinol decreased at peak and after clearance of a noninvasive gut pathogen but increased during the acute phase of invasive, virulent pneumonia infection. Although VA was reduced in the liver during peak C. rodentium infection, the concentration of VA in the colon, which is the site of infection, was not affected. Interestingly, serum retinol in VAA mice was not reduced in either infection. Although the exact mechanisms responsible for the alterations in tissue VA during these 2 infections, 1 chronic and 1 acute, remain to be elucidated, our assessment of VA status during 2 active bacterial infections points to complex differences in VA distribution due to specific types of bacterial mucosal infections.