Altered Expression and Localization of Ion Transporters Contribute to Diarrhea in Mice With Salmonella-Induced Enteritis

Abstract

BACKGROUND & AIMS

Salmonella enterica serovar Typhimurium is an enteropathogen that causes self-limiting diarrhea in healthy individuals, but poses a significant health threat to vulnerable populations. Our understanding of the pathogenesis of Salmonella-induced diarrhea has been hampered by the lack of a suitable mouse model. After a dose of oral kanamycin, Salmonella-infected congenic BALB/c.D2^NrampG169 mice, which carry a wild-type Nramp1 gene, develop clear manifestations of diarrhea. We used this model to elucidate the pathophysiology of Salmonella-induced diarrhea.

METHODS

BALB /c.D2^NrampG169 mice were treated with kanamycin and then infected with wild-type or mutant Salmonella by oral gavage. Colon tissues were isolated and Ussing chambers, quantitative polymerase chain reaction, immunoblot, and confocal microscopy analyses were used to study function and expression of ion transporters and cell proliferation.

RESULTS

Studies with Ussing chambers demonstrated reduced basal and/or adenosine 3′,5′-cyclic monophosphate–mediated electrogenic ion transport in infected colonic tissues, attributable to changes in chloride or sodium transport, depending on the segment studied. The effects of infection were mediated, at least in part, by effector proteins secreted by the bacterial Salmonella pathogenicity island 1– and Salmonella pathogenicity island-2–encoded virulence systems. Infected tissue showed reduced expression of the chloride–bicarbonate exchanger down-regulated in adenoma in surface colonic epithelial cells. Cystic fibrosis transmembrane conductance regulator was internalized in colonic crypt epithelial cells without a change in overall expression levels. Confocal analyses, densitometry, and quantitative polymerase chain reaction revealed that expression of epithelial sodium channel β was reduced in distal colons of Salmonella-infected mice. The changes in transporter expression, localization, and/or function were accompanied by crypt hyperplasia in Salmonella-infected mice.

CONCLUSIONS

Salmonella infection induces diarrhea by altering expression and/or function of transporters that mediate water absorption in the colon, likely reflecting the fact that epithelial cells have less time to differentiate into surface cells when proliferation rates are increased by infection.

Infection with Salmonella enterica serovar Typhimurium (henceforth Salmonella) is a world-wide health problem common to both industrialized and nonindustrialized countries.^1-3 Salmonella-induced diarrhea contributes to the spread of this pathogen, secondary to contamination of food and water supplies.⁴ However, little is known about how this pathogen induces diarrhea. In vitro studies with infected colonic epithelial cell lines have suggested modulation of ion transport. However, in vitro models cannot fully recapitulate the effects of infection on physiological responses that produce diarrhea in vivo.⁵

Salmonella relies on 2 type-3 protein secretion systems (T3SS) to invade and spread systemically in the host. Salmonella pathogenicity island (SPI)-1 encodes a needle-like protein appendage through which soluble effectors are injected into the host-cell cytosol when in close proximity. These effectors facilitate bacterial uptake into host epithelial cells through manipulation of the cytoskeleton and a variety of signal transduction pathways.⁶ The second T3SS, SPI-2, is necessary for maintenance of the integrity of the Salmonella-containing vacuole that is essential for intracellular growth in macrophages and systemic spread of the bacteria. SPI-2 translocates bacterial effectors through the phagosomal membrane, which then alter cytoskeletal structures, disrupt membrane trafficking, and affect signal transduction.⁶

The study of Salmonella-induced diarrhea is hindered by the fact that oral infection of rodents does not cause enteritis or diarrhea, but if mice are pretreated orally with a nonabsorbed aminoglycoside antibiotic, they will develop colitis.⁷ However, many common inbred mouse strains carry a mutant Nramp1 (Scl11a1) allele, and after oral infection these mutant mice typically die from systemic salmonellosis without manifesting diarrheal symptoms.⁸ The role of Nramp1 is to efflux divalent ions from the phagolysosomes,⁹ thereby controlling infection by reducing essential Fe²⁺ and Mn²⁺ required for Salmonella growth in the phagolysosome or Salmonella-containing vacuole.¹⁰ Without functional Nramp1, mice cannot limit Salmonella propagation within macrophages and organisms spread systemically, quickly overwhelming the host without the manifestation of diarrhea.⁶ Therefore, we developed a mouse model that exploits congenic BALB/c mice expressing the wild-type (wt) Nramp1^G169 allele from DBA/2 mice. BALB/c.D2^NrampG169 mice develop clear manifestations of diarrhea when orally infected with Salmonella Typhimurium strain 14028s after pretreatment with kanamycin (Kana), making this mouse an appropriate model for Salmonella-induced diarrhea.¹¹ Within 3 days post infection, stool water content is significantly increased and, by 7–8 days post infection, mice have severe pan-colitis with marked weight loss, colonic shortening, high stool water content, and diarrhea. A high proportion of mice die by 10 days post infection, although the systemic bacterial counts in the liver and spleen are still at sublethal levels, suggesting that the mechanism of death is related to the severe intestinal disease rather than sepsis.⁸ Both the SPI-1 and SPI-2 T3SS are required for full enteric virulence in this BALB/c.D2 model of Salmonella enteritis.^8,11 A loss of function SPI-1 mutant, invA, causes significantly less diarrhea and fecal water content, as well as an absence of inflammatory changes distal to the cecum. A SPI-2 mutant (ssaV) also causes lower fecal water content and markedly reduced colonic inflammation, despite comparable levels of cecal infection compared with wt Salmonella.¹¹

Therefore, we sought to determine whether Salmonella can alter ion transporter function and expression before development of severe colitis and whether these changes contribute to the etiology of diarrhea by altering water absorption. Because SPI-1 and SPI-2 mutants cause attenuated diarrhea and colitis in this model, we used these isogenic mutant strains as specific controls for the virulence effects of Salmonella on epithelial physiology and transporter expression.

Materials and Methods

Mouse Model

Ussing chamber experiments used tissue from male and female BALB/c.D2^NrampG169 congenic mice,¹² and gene expression studies were done with female mice. Mice were bred in-house under specific pathogen–free conditions and allowed free access to food and water. Mice were administered a single dose of Kana (20 mg) by gastric gavage and infected by gavage 48 hours later (5 × 10³ colony-forming units of Salmonella Typhimurium 14028s in 0.1 mL of 0.1M NaHCO₃). Mice were sacrificed by cervical dislocation at 3 days post infection, and all studies were approved by the San Diego VA Institutional Animal Care and Use Committee.

Bacterial Stocks

We constructed a derivative of Salmonella Typhimurium 14028s containing a Kana-resistance cassette inserted in a genetically silent region downstream from the spv operon.¹³ This strain was used as the wt background strain. SPI-1 (invA) and SPI-2 (ssaV) mutants of 14028s are Kana-resistant and were as described previously.^11,13

Ussing Chamber Studies

Segments of proximal (first one third from cecum) or distal (last one third to rectum) colon obtained from uninfected and infected mice were stripped of their muscle layers and mounted in Ussing chambers (window area = 0.6 cm²) according to protocol.^14,15 Tissues were voltage-clamped to zero potential difference by the application of short-circuit current (I_sc), and baseline was established. Tissues were treated with adenosine 3′,5′-cyclic monophosphate (forskolin [FSK], 20 μM]) and Ca²⁺ (carbachol, 300 μM]) agonists and the I_sc response measured. Under these conditions, changes in I_sc (ΔI_sc) in response to agonists are largely reflective of electrogenic chloride secretion and/or sodium absorption. For ion substitution studies, sodium isethionate, CaSO₄, and MgSO₄ were used in place of Cl⁻ and choline (choline bicarbonate and choline chloride) was used as a substitute for Na⁺. Both buffers were supplemented with 10 mmol glucose.

Real-Time Polymerase Chain Reaction

Full-thickness proximal and distal colons were isolated from Kana-treated controls or Salmonella (wt and mutants)-infected mice, homogenized using a tissue lyser, and processed using Trizol (Sigma, St Louis, MO) following manufacturer’s instructions. The homogenate was then placed on an RNeasy column (Qiagen, Valencia, CA) and RNA isolated and stored at −80°C. RNA was incubated with Turbo DNA-free (Ambion, Austin, TX) before reverse transcription using a high capacity kit (Applied Biosystems, Carlsbad, CA) according to manufacturer’s protocol. Primers (IDT, Coralville, IA) used are listed in Supplementary Table 1. All reactions were performed using the Step-One plus system (Applied Bio-systems) under the following conditions: 95°C, 5 minutes; 40 cycles and 95°C, 3 seconds; 60°C 30 seconds. Target genes were normalized to villin. The resulting Ct values were calculated using single delta Ct as 2^{-(Target–Villin)} and expressed in arbitrary units.

Western Blot

Full-thickness colonic tissues were excised, placed in RIPA buffer, and lysed using a mini bead beater (BioSpec Products, Inc, Bartlesville, OK). The resulting lysate was spun down and the supernatant removed. Equivalent amounts of protein for each sample were run on a 4%–15% gradient polyacrylamide gel (mini-protean TGX; Bio-Rad, Hercules, CA). After semi-dry transfer, the blot was blocked (5% nonfat milk or 5% bovine serum albumin in phosphate-buffered saline; 1 hour) and incubated with rabbit anti–down-regulated in adenoma (DRA) (kindly provided by Dr Pradeep Dudeja), anti–epithelial sodium channel-β (ENaCβ), anti–cystic fibrosis transmembrane conductance regulator (CFTR), anti– β1ATPase (Genetex, Irvine, CA), anti-Salmonella LPS 0-4 [1E6], or anti-NKCC1 (Cell Signaling, MA); and processed according to standard protocols.^16,17 Densitometry was performed using ImageJ software (National Institutes of Health, Bethesda, MD).

Immunofluorescence

Antibodies to DRA (SLC26a3; Sigma; ENaCβ, CFTR, β1ATPase, Salmonella lipopolysaccharide 0–4, and NKCC1; Abcam, Cambridge, MA), as well as conjugated secondary antibodies (Alexa Fluor 568 donkey anti-mouse, Alexa Fluor 488 donkey anti-rabbit, and Hoechst 33258; Invitrogen, Carlsbad, CA) were utilized. Colons were removed, rinsed of contents, infused with 4% paraformaldehyde, and “Swiss rolled” (rectum to cecum). The roll was placed in 4% paraformaldehyde, embedded in paraffin, and sectioned (5 μM) onto glass slides. Sections were de-waxed using xylene, hydrated with graded ethanol and heated for antigen retrieval (10mM sodium citrate buffer). The sections were then treated with proteinase K (0.2 μg/mL) Tris-EDTA buffer and blocked (5% bovine serum albumin/2% donkey serum in phosphate-buffered saline). The sections were further blocked with donkey anti-mouse unconjugated Fab fragment (1:10) before immunostaining with primary antibody. Imaging and line scanning were performed using the Zeiss LSM 510 confocal imaging system. Antibodies to NHE3 were generously provided by Dr Mark Donowitz and NHE3 immunostaining was kindly performed by Dr Nicholas Zachos.

Immunohistochemistry

Sections were prepared, cut, and processed as described for immunofluorescence studies. Ki67 antibody (Abcam) was coupled with ABC detection (Vector, Burlingame, CA) followed by AEC chromagen (Vector), and hematoxylin counterstained for visualization. Crypt length was assessed measuring 20 well-oriented, nonadjacent crypts (Nanozoom Digital Image Viewer, Hamamatsu, Japan).

Statistics

All values are presented as means ± SEM for the number of experiments noted. Student’s unpaired t test or one-way analysis of variance were used as appropriate followed by post-hoc analysis using Bonferroni (quantitative polymerase chain reaction [qPCR]) or Newman-Keuls tests (I_sc and ΔI_sc). Values of P < .05 were considered significant. All analyses were done using GraphPad Prism 4 (La Jolla, CA).

Results

Electrogenic Ion Transport Was Suppressed by Salmonella Infection

Proximal colon was isolated from untreated (Con), Kana-treated, Salmonella (wt)-infected, or Salmonella mutant–infected (invA or ssaV) mice and mounted in Ussing chambers. Baseline resistance (data not shown) showed no difference between treatment groups and histology indicated no significant ulcerations (Supplementary Figure 1). In contrast, baseline I_sc (Figure 1) showed a significant decrease between Kana and wt-Salmonella–infected groups. Tissue from wt-infected animals had reduced ion transport responses to the adenylate cyclase agonist FSK compared with tissues from control or Kana-treated mice. Neither mutant Salmonella decreased sensitivity to FSK (Figure 1B). In contrast, transport responses of proximal colon from wt-infected mice to the muscarinic agonist carbachol were not suppressed compared with uninfected control tissues (Figure 1C). In the distal colon, baseline transport was also reduced after wt-Salmonella infection when compared with Kana-treated, although not significantly (Figure 1D), and FSK-mediated ΔI_sc responses were significantly reduced (Figure 1B). Carbachol-mediated responses were not significantly different between Kana-treated and wt-infected tissues (Figure 1E).

Figure 1.

wt Salmonella inhibits adenosine 3′,5′-cyclic monophosphate–dependent ion transport in proximal and distal colon. In the proximal colon, baseline I_sc (A), and I_sc responses to FSK (B) and carbachol (CCh [C]) were determined in control (Con), Kana-treated or Salmonella-infected (wt, invA, ssaV mutants) mice (*P < .05; analysis of variance [ANOVA]). Similarly, in the distal colon, baseline I_sc (D) and I_sc responses to FSK (E) and CCh (F) stimulation were compared for wt and Kana-treated mice (*P < .05; Student’s t test, with data was normalized to the Kana control).

Suppression of Electroneutral Transporter Expression in Salmonella-Infected Proximal Colon

Confocal microscopy was used to assess whether infection with Salmonella alters localization of the major electroneutral anion transporter, DRA. Confocal microscopy showed that DRA (green) remained sharply colocalized with the epithelial intermediate cytoskeletal filament, villin (purple) at the apical membrane of proximal colonic surface epithelial cells in uninfected Kana-treated mice (Figure 2A). Line scans across the apical membrane of the surface epithelium (arrows) showed a decrease in DRA/villin coincident peaks in wt-infected mice corresponding to the reduction in pixel intensity of DRA when compared with Kana-treated and mutant-infected mice. Corresponding findings were obtained when full-thickness proximal colon isolated from Kana-treated and wt-infected mice was analyzed by Western blot (Figure 2B). Densitometry showed a significant (n = 6; P < .05) decrease in DRA in the wt-infected compared with Kana-treated mice. We next assessed whether infection alters messenger RNA (mRNA) levels for the major electroneutral absorptive ion transporters (NHE3 and DRA). qPCR revealed no changes in NHE3 mRNA due to Salmonella infection (Supplementary Figure 2), and DRA mRNA was significantly reduced in wt infection compared with Kana-treated tissues or those from mice infected with invA or ssaV mutants (Figure 2C). NHE3 expression in epithelial cells was also not altered by Salmonella infection, as determined by microscopy (Supplementary Figure 2). These observations indicate Salmonella can reduce the capacity for chloride absorption in the proximal colon and that functional SPI-1 and SPI-2 are necessary for the full suppression of DRA expression.

Figure 2.

Apical localization of DRA in surface epithelial cells of the proximal colon is reduced by wt Salmonella infection. Confocal imaging of DRA (green) and villin (purple) in proximal colon from control (Con), kanamycin-treated (Kana) or Salmonella-infected (wt, invA, or ssaV mutants) mice (A). Line scans (red) quantify DRA colocalization with villin (white) at the apical membrane of surface epithelial cells (arrows). All images are representative of 6–10 fields from n = 2–3 experiments (scale bar = approximately 20 μm); Hoechst-stained nuclei (blue). Western blot (inset) and densitometry (B) for DRA protein (*P < .05; Student’s t test). mRNA levels (C) for DRA (*P < .05; analysis of variance).

Altered Trafficking of the Electrogenic Transporter, CFTR, in the Proximal Colon

In Ussing chamber studies, FSK-stimulated electrogenic ion transport (most likely chloride secretion) was diminished by Salmonella infection. Using confocal microscopy, CFTR (green) was confirmed by line scan to be colocalized with villin (purple) to the apical membrane of crypt epithelial cells (red line, white box) in control Kana-treated mice (Figure 3A). In contrast, CFTR appeared to redistribute to the cytosol in wt-infected mice, as evidenced by the reduction of green peaks in the intensity profile. Conversely, the apical localization of CFTR was largely maintained after infection with ssaV and invA mutants (Figure 3A), although there might be partial mislocalization of CFTR in ssaV mutant–infected animals. In addition, to eliminate the possibility of modulation of expression or an enhancement of degradation, densitometry (Figure 3B) and qPCR (Figure 3C) were performed. We observed no difference in CFTR mRNA or protein levels in wt-infected mice vs Kana-treated controls. The reduction of basal and FSK-stimulated I_sc seen in wt-infected mice might result from altered CFTR trafficking rather than a reduction in overall expression or degradation.

Figure 3.

CFTR is internalized in proximal colon from wt-infected mice. Confocal imaging of CFTR (green) and villin (purple) in proximal colon from control (Con), Kana-treated or Salmonella-infected (wt, invA, or ssaV mutants) mice (A). CFTR colocalization with the apical membrane marker, villin, in crypt epithelial cells (box) was determined by line scans in selected crypts (white box). All images are representative of 6–10 fields from n = 2–3 experiments (scale bar = approximately 20 μm); Hoechst-stained nuclei (blue). Western blot (inset) and densitometry (B) for CFTR protein and qPCR (C) for CFTR mRNA.

Reduced ENaC Apical Localization in Wild-Type Salmonella Challenged Distal Colon

The distal colon is an important site for salvage fluid absorption, driven by the Na⁺ channel, ENaC. Confocal microscopy of the distal colon from Kana-treated mice showed that ENaCβ (green) was localized to a distinct population of surface cells (Figure 4, arrow). Staining for ENaCβ was greatly diminished in tissues from wt-infected mice, but not in tissues from mice infected with either the invA or ssaV mutant (Figure 4). Western blot and densitometry show a significant decrease in ENaCβ, supporting the confocal microscopy. Overall, these findings are consistent with the observed reduction in ion transport in the distal colon from wt-infected mice, which is likely reflective, in part, of reduced electrogenic Na⁺ absorption.

Figure 4.

ENaCβ-enriched cells are reduced in wt-infected mice. Confocal imaging of ENaCβ (green) and the apical membrane marker, villin (purple) in distal colon from control (Con), Kana-treated or Salmonella-infected (wt, invA, or ssaV mutants) mice (A). Arrows represent epithelial cells enriched in ENaCβ. All images are representative of 6–10 fields from n = 2–3 experiments; Hoechst-stained nuclei (blue). Western blot (inset) and densitometry (B) for ENaCβ protein (*P < .05; Student’s t test). mRNA levels (C) for ENaCα, β, and γ (*P < .05; analysis of variance).

ENaC subunits were assessed by qPCR in specimens from mice infected with either wt Salmonella or the invA and ssaV mutants and compared with the Kana-treated control. Wild-type Salmonella, but not the InvA or ssaV mutant, reduced mRNA for ENaCβ (Figure 4C). In contrast, mRNA for ENaCα and ENaCγ was not significantly affected by infection with any of the Salmonella strains when compared with Kana (Figure 4C).

Salmonella-Infected Colons Demonstrate Reduced Electrogenic Cl⁻ and Na⁺ Transport

Ion substitution buffers were employed to establish whether the functional electrogenic ion transport changes we observed in Ussing chambers could be attributed to the altered expression and/or localization of specific ion transporters after Salmonella infection. Removal of Cl⁻ significantly decreased baseline I_sc in both kana-treated and wt-infected tissues from the proximal colon (Figure 5A). However, removal of Cl⁻ significantly decreased the ΔI_sc response to FSK in the Kana control only, but not in wt-infected tissues (Figure 5B). We interpret these data as consistent with the model that the reduced response to FSK in the infected proximal colon can be accounted for, in large part, by the redistribution of CFTR described earlier, because it could not be significantly reduced further by chloride removal. In the distal colon, removal of Na⁺ reduced baseline I_sc in both Kana-treated and wt-infected tissues, but the effect was more pronounced in uninfected controls (Figure 5C). In contrast, the FSK response, whether at control levels in Kana tissues or at reduced levels in tissues from wt-infected mice, was unaffected by removal of Na⁺ (Figure 5D). We interpret these data to suggest that the reduction in baseline ion transport in infected distal colon is attributable in part to a reduction in ENaC-mediated sodium transport, because Na⁺ removal was less effective in reducing I_sc values in wt-infected tissues vs Kana controls. By extension, responses to FSK in these tissues do not appear to reflect electrogenic sodium absorption under either condition.

Figure 5.

Ion substitution reveals reduced Cl⁻ and Na⁺ transport in proximal and distal colon, respectively, from mice infected with wt Salmonella. Baseline I_sc (A, C) and FSK-stimulated I_sc responses (B, D) were measured in Ringer’s solution, or in the absence of Cl⁻ (proximal, Kana [Cl⁻] or wt [Cl⁻]) or Na⁺ (distal, Kana [Na⁺] or wt [Na⁺]) ions (*P < .05; analysis of variance).

Salmonella-Infected Proximal Colon Demonstrates Aberrant Basolateral Transport

The ion transporter responsible for the cellular loading of Cl⁻, NKCC1 (green), appeared to be enriched in crypt epithelial cells in Kana-treated proximal colon (Supplementary Figure 3A) with exclusive basolateral localization when referenced to villin (red: apical). The NKCC1-enriched cell population appears to expand in wt-infected colon (Supplementary Figure 3A). This increase in NKCC1-enriched cells was not observed in SPI-1 or SPI-2 mutant–infected mice (Supplementary Figure 3A). In addition, Western blot and qPCR did not show a significant increase in NKCC1 protein (Supplementary Figure 3B) or gene expression (Supplementary Figure 3C) in wt-infected mice compared with Kana-treated controls (when normalized for cellular proteins), probably because the amount of NKCC1 per cell was unchanged. In summary, this suggests that the increase in NKCC1 seen in tissue sections is not due to changes in gene or protein expression but rather to an increase in NKCC1-enriched crypt cells (described in findings of epithelial crypt hyperplasia).

Mice Infected With Wild-Type Salmonella Have Reduced Expression of β1ATPase

Confocal microscopy reveals that mice infected with wt Salmonella show a reduced basolateral localization β1ATPase (green) as referenced to the apical localization of villin (purple) and intracellular nuclei (blue; Supplementary Figure 4). In addition, in wt-infected tissues, β1ATPase and villin appeared inappropriately co-localized in surface cells, perhaps suggesting that basolateral and apical membrane domains are no longer fully segregated. Densitometry also supported confocal observations of reduced β1ATPase protein levels in wt-infected versus Kana-treated mice.

Salmonella Causes Epithelial Crypt Hyperplasia

In the proximal colon, an increased number of epithelial cells stained positively for the proliferation marker, Ki67, in wt Salmonella–infected tissues compared with Kana-treated controls (Figure 6A). This increase was not observed in either invA or ssaV mutant–infected tissues. An increase in colonic crypt depth was also observed in Salmonella-infected tissues compared with Kana controls (Figure 6B and C). Together, this suggests that crypt hyperplasia is driven by cell proliferation in response to wt-Salmonella infection.

Figure 6.

Epithelial cell proliferation and colonic hyperplasia after Salmonella infection. Staining for the proliferation marker Ki67 was performed, followed by quantification of positive cells from control (Con), Kana-treated, or Salmonella-infected (wt, invA, or ssaV mutants) mouse proximal colonic tissue (A). Colonic epithelial cell hyperplasia (B) was assessed by measuring crypt depth. Representative images (C) (*P < .05; analysis of variance, n = 3–4 from 2 experiments).

Functional Effects of Salmonella Infection Are Not Restricted to Infected Cells

Confocal microscopy was utilized to determine whether the effects of Salmonella infection were restricted to infected epithelial cells or whether they were generalized. Wild-type Salmonella were clearly localized within a minor population of surface epithelial cells, but none were detected in the crypts (Figure 7A and D). In contrast, fewer ssaV mutant bacteria were detected in the surface epithelium and none were seen in crypts (Figure 7B and E), and invA mutants were exclusively localized to the intestinal lumen (Figure 7C and F). Because wt Salmonella infected only a small percentage of epithelial cells, ion transport modulation by wt Salmonella infection is not restricted to epithelial cells that have been directly invaded by Salmonella, but rather represents a more generalized response of host colonocytes to this infection.

Figure 7.

Translocation of invA and ssaV mutant Salmonella is attenuated compared with wt Salmonella. Confocal imaging of Salmonella (red) in proximal colon from infected (wt, invA, or ssaV mutants) mice (A–F). Lower panels (D–F) are magnified images of the corresponding boxed areas. Bacteria (white arrows); Hoechst-stained nuclei (blue). All images are representative of 6–10 fields from n = 2–3 experiments.

Discussion

Functional analysis of electrogenic ion transport and transporter expression and localization studies suggest that Salmonella modulates colonic ion transport and this requires the bacterial T3SS systems encoded by SPI-1 and SPI-2 virulence loci. Ussing chamber studies of wt Salmonella–infected mice showed aberrations in ion transport function when compared with uninfected or mutant-infected mice in both proximal and distal colon. Neither segment showed barrier dysfunction or evidence of ulcerations (Supplementary Figure 1). Salmonella-induced modulation of ion transport could induce diarrhea in the affected host, through either decreased absorption or increased secretion of water and electrolytes.¹⁸ In fact, enteropathogens have been reported to modulate ion transporters in epithelial cell lines via effectors injected through their T3SS.^17,19,20 Similarly, Citrobacter rodentium infection of mouse colons decreased expression of DRA.¹⁴ Loss-of-function mutations in DRA cause congenital chloride-mediated diarrhea.²¹ In addition, the Salmonella SPI-1 Sop E effector family has homology with the EPEC T3SS effector Map, which has been implicated in indirect regulation of DRA, CFTR, and NHE3.^22-25 In addition, the major transporter responsible for the uptake of Na⁺ in the distal colon, ENaC, is down-regulated by lipopolysaccharide via nuclear factor-κB– mediated inhibition of SGK1.²⁶ Finally, in the lung, Pseudomonas aeruginosa decreases mRNA for ENaC subunits.²⁷ Comparisons of proximal and distal colon from mice treated with Kana only or subsequently infected with either wt or mutant Salmonella revealed that there was profound modulation of ion transport that likely underlies, at least in part, the diarrhea associated with wt Salmonella infection.^28,29 In particular, the reduced ability of sodium removal to abrogate baseline I_sc in wt-infected vs Kana-treated distal colon argues that tonic salvage fluid absorption in infected tissues is likely reduced secondary to the loss of ENaC expression.

Confocal analysis also revealed a reduction in DRA (but not NHE3) apical localization in surface cells of the proximal colon from wt-infected mice compared with Kana-treated or mutant-infected mice. Both DRA protein and mRNA are reduced in wt-infected mice. However, because this transporter is electroneutral, the decrease in DRA expression does not explain the defect in electrogenic transport responses observed in colonic epithelium from wt-infected mice. Nevertheless, the decrease in DRA in wt-infected mice could increase luminal Cl⁻ concentrations, retaining water in the luminal space and contributing to diarrhea.

Expression of CFTR appeared to be mislocalized in crypt cells in the proximal colon of wt-infected mice compared with controls. Because CFTR mediates electrogenic Cl⁻ transport, this aberrant localization might underpin the reduction in basal I_sc, because CFTR likely contributes to baseline secretion.³⁰ In addition, CFTR internalization could explain the reduction in FSK-stimulated ΔI_sc in wt-infected mice. The fact that qPCR and densitometry revealed no change in overall CFTR expression supports that aberrant CFTR trafficking, and not lower expression, explains the loss of function. Reduced I_sc responses to FSK in wt-infected tissues are also likely attributable to the loss of apical CFTR because they were no longer sensitive to chloride removal. Nevertheless, redistribution of CFTR and a subsequent decrease in Cl⁻ secretion cannot directly account for Salmonella-induced diarrhea, although a relative failure of crypt secretory function might ultimately be expected to compromise crypt sterility and barrier function.

Wild-type Salmonella infection was associated with hyperplasia of crypt cells, similar to that observed in the setting of C rodentium infection in mice.³¹ The proliferative response may represent an attempt by the host to shed bacterially infected cells at the expense of cellular differentiation across the length of the crypt. Although NKCC1 was highly enriched in crypt cells of the proximal colon in controls, in wt-infected mice, NKCC1-positive cells extended further up the crypt when compared with the Kana-treated and mutant Salmonella controls. However, neither protein nor mRNA analyses revealed a change in normalized NKCC1 expression between Kana-treated and wt-infected mice, suggesting that the confocal findings reflected an increased number of NKCC1-positive cells rather than increased protein expression per cell. The increased epithelial proliferation in wt-infected mice further supports our model of an immature epithelium that contributes to the diarrhea. In addition, NKCC1 was diffusely localized in infected tissues compared with the sharp basolateral localization in controls. In fact, a reduction in basolateral NKCC1, as well as β1ATPase, could contribute to the suppression of basal electrogenic transport by wt Salmonella in the proximal colon.

Salmonella infection resulted in bacterial colonization throughout the colon by day 3 post infection, but the distal colon had 10-fold fewer bacteria/gram of tissue than did the proximal colon (data not shown).¹¹ By confocal microscopy, the ssaV mutant appeared to be present in epithelial tissue to a lesser extent than the wt, and the invA mutant was not observed to translocate from the lumen. These findings agree with earlier observations in the Salmonella-infected rat colon demonstrating distal effects of Salmonella on transport function.³²

Overall, our results support a model in which Salmonella infection of the colon produces early inflammatory changes that drive compensatory epithelial proliferation and thereby alter the physiology of ion transporters via expression and functional localization. Events occurring solely in epithelial cells directly infected by Salmonella are unlikely to account for all of the early physiologic manifestations of diarrhea; rather, our data suggest the importance of diffusible factors that link Salmonella infection to overall epithelial proliferation and ion transport physiology. Salmonella has been shown to induce mediators of intestinal epithelial cell proliferation through specific inflammatory pathways that include production of interleukin-17/interleukin-22, interleukin-18, and granulocyte-macrophage colony-stimulating factor.^33-35 In addition, Salmonella infection activates the β-catenin/Wnt pathway.³⁶ Salmonella-induced intestinal inflammation can promote the colonization phase of enteric infection through the modulation of specific conditions, including antimicrobial peptides and metabolic factors, favoring its growth over the resident microbiota.³⁵ Our results indicate that Salmonella-induced inflammation likely also induces diarrhea through changes in epithelial cell proliferation and the expression and function of ion transporters that lead to decreased absorptive function in the colon. The combined effect of these virulence mechanisms on host inflammatory responses ensures both intestinal colonization and increased transmission of this enteric pathogen.

In conclusion, Salmonella infection modulates colonic ion transport at an early stage in experimental colitis, in part due to increased epithelial cell proliferation influencing absorptive epithelial function. Although the physiological effects of infection are not limited to absorptive transport, the suppression of absorption would be expected to cause an accumulation of luminal fluid as an important contributor to Salmonella-induced diarrhea that precedes frank dysentery in this model. Many patients with Salmonella enteritis present mainly with diarrhea. Therefore, strategies to restore absorptive transport would be expected to be beneficial in Salmonella gastroenteritis.

Abbreviation used in this paper

CFTR

cystic fibrosis transmembrane conductance regulator

DRA

down-regulated in adenoma

ENaC

epithelial sodium channel

FSK

forskolin

Kana

kanamycin

mRNA

messenger RNA

qPCR

quantitative polymerase chain reaction

SPI

Salmonella pathogenicity island

T3SS

type-3 protein secretion system

wt

wild-type