Commensal Escherichia coli Reduces Epithelial Apoptosis through IFN-αA–Mediated Induction of Guanylate Binding Protein-1 in Human and Murine Models of Developing Intestine Free

As the largest interface between the organism and the environment, the intestine interacts with a vast microflora, and appropriate colonization is critical for intestinal function and development (1). Bacterial colonization is a dynamic process, with the intestine of the term newborn specifically set to foster tolerance of initial colonizing bacteria (2). Through recognition of complex bacterial patterns, the term intestine senses the presence of luminal bacteria and distinguishes between pathogenic and commensal organisms, regulating tissue responses through a balance of inflammatory and apoptotic signaling pathways (1–6). Furthermore, maladaptive responses to colonizing organisms have been implicated in pathogenic inflammatory responses in both mature (7–9) and developing intestine (10).

Both the nature of colonizing flora and the timing of colonization are likely to be critical in the immature intestine of premature infants who are uniquely vulnerable to unchecked inflammatory responses, sometimes culminating in intestinal necrosis and death due to necrotizing enterocolitis (NEC). The onset of NEC is developmentally dependent with peak onset occurring at 30–32 wk postconceptual age, regardless of gestational age at the time of birth (11). Therefore, it is likely that the elaboration of disease requires a developmental trigger or event. Furthermore, NEC does not occur in utero and appears to require bacterial colonization prior to the manifestation of disease. Growing evidence indicates that the immature human intestine is developmentally unprepared for bacterial colonization, elaborating an inappropriately exuberant response to proinflammatory stimuli (12). In addition, preterm infants, subject to frequent antibiotic use, instrumentation, and the hospital environment, are vulnerable to colonization with abnormal flora relative to the term infant (13–15). Thus, the premature human intestine is predisposed to NEC by colonization with atypical flora at a time when it is developmentally unprepared to respond in an appropriate manner (16).

In experimental models of NEC, increased intestinal epithelial apoptosis precedes the progression to frank necrosis (17). Thus the regulation of apoptosis has been implicated as a critical event in the early evolution of NEC. These findings, in combination with the tendency of immature intestine to elaborate exaggerated responses to proinflammatory stimuli, have led to the hypothesis that epithelial apoptosis contributes to barrier compromise, leading to bacterial translocation and an exuberant inflammatory response which, when inappropriately regulated, leads to necrosis in the naive and immature intestine (16).

Among the first colonizing organisms evident in the term newborn human intestine are strains of Escherichia coli derived from the maternal gastrointestinal tract (1, 2, 6). Specific commensal strains of E. coli have been shown to reduce intestinal epithelial inflammatory signaling, in vitro, through repression of NF-κB signaling (18). Although pathogenic, and more specifically enteroinvasive E. coli is known to be proapoptotic due to the elaboration of virulence factors activating caspase 1 (19), the effects of commensal strains on apoptosis remain to be determined.

TLRs are critical for the specific detection of microbe-associated patterns, allowing differentially regulated responses to commensal versus pathogenic flora (20). Independent activation of specific TLRs, such as TLR2 and TLR4, has been linked through downstream activation of MyD88, to modulation of flora-dependent intestinal inflammation and immune responses (21). Additional evidence argues that beneficial effects of commensal and probiotic bacteria in the intestine are mediated at least in part by TLR9 activation, specifically through recognition of bacterial DNA (22). Furthermore, reciprocal TLR4 and TLR9 signaling are important in the developing intestine in that increased TLR4 and decreased TLR9 expression occur in human and murine NEC, and attenuation of TLR4 signaling both increases TLR9 expression and decreases disease severity in a model of NEC in 2-wk-old mice (23). The protective effects of TLR9 signaling are mediated by the downstream induction of type I IFNs (24), such as IFN-αA. In these studies, we sought to examine the influence of a commensal strain of E. coli on IFN-αA signaling and downstream effects on apoptosis in developing intestinal epithelium in a well-established murine model of immature intestine.

Several lines of evidence identify the large GTPase, guanylate binding protein-1 (GBP-1) as a candidate mediator of IFN-mediated changes in apoptosis. The mouse GBP-1 promoter has been well characterized in embryonic fibroblasts as a target of both type I (α, β) and type II (γ) IFNs, which transcriptionally activate GBP-1 expression through Stat-independent activation of IFN regulator factor 1 (25). Short-term IFN-α treatment of endothelial cells inhibits angiogenesis through GBP-1–mediated prevention of apoptosis (26). And recently, we have shown that GBP-1 promotes intestinal epithelial barrier integrity through the prevention of apoptosis in model intestinal epithelia, in vitro (27). Furthermore, GBP-1 is upregulated both in response to exogenous IFN-γ, in vitro, and in the colonic epithelia of patients with inflammatory bowel disease (IBD) (27). The role of GBP-1 and its regulation by IFN-αA in developing intestine has not been previously studied.

In the studies presented in this study, we optimized a model of immature mouse intestine to examine mechanisms regulating apoptosis during an interval where apoptosis could be most easily induced prior to or in the absence of associated inflammation. We then examined the influence of exposure to commensal E. coli on both inducible apoptosis and the expression of candidate mediators for its regulation. IFN-αA and GBP-1 were specifically identified as molecules associated with the prevention of apoptosis, which were induced after exposure to commensal E. coli in immature mouse intestine. Findings in the murine model system were further extended to mature and immature human model epithelia, demonstrating that IFN-αA exerted an antiapoptotic effect that was specifically GBP-1 dependent, in vitro. Finally, we confirmed that exogenous IFN-αA not only induced GBP-1 expression, but also prevented inducible apoptosis in an ex vivo murine model of developing intestine. These findings represent the first evidence of a role for GBP-1 in mediating the antiapoptotic effects of type I IFNs induced by commensal flora in the developing intestine. Furthermore, prevention of apoptosis through induction of GBP-1 may provide a critical target for the promotion of epithelial barrier integrity and the prevention of bacterial translocation and the exaggerated mucosal inflammatory responses leading to NEC.

T84 colonic epithelial cells were grown as previously described (28). FHs 74 intermediate (FHs 74 Int) fetal small intestinal epithelial cells were obtained from American Type Culture Collection (ATCC; Manassas, VA) and cultured in ATCC complete growth medium according to the provided protocol. Cells were grown directly on glass cover slips for TUNEL staining and activated caspase-3 detection. T84 cells were grown on inserts and IFN-αA was applied basolaterally. FHs 74 Int cells were grown directly on 6-well plates.

A strain of naturally occurring commensal E. coli, which was isolated from human intestine during colonoscopy of healthy donors, was generously provided by Dr. Andrew Neish (Emory University, Atlanta, GA) and prepared overnight in LB broth at 37°C, as previously described (18, 29). E. coli cultures were washed, concentrated in DMEM and gavage fed to 2-wk-old mice at 1–2 × 10⁷ CFUs/ml.

All animal experiments were conducted according to protocols approved by the Institutional Animal Care and Use Committee at Emory University. C57BL/6J mice were bred and maintained within the animal facility at Emory University.

Two-week-old mice were orally gavage fed 0.2 ml DMEM with or without 1–2 × 10⁷ CFUs/ml commensal E. coli. Four hours after feeding, mice were sacrificed, colons were harvested, opened lengthwise, and stool was removed and resuspended in sterile HBSS. The stool suspension was plated on blood agar plates and incubated at 37°C for 16 h. Equal areas of bacterial colonies were recovered and resuspended in sterile HBSS. Bacterial DNA was recovered using a modification of the QIAmp DNA stool mini kit (Qiagen, Valencia, CA). Bacteria were lysed by incubation for 10 min at 70°C, then excess proteins were digested with proteinase K for 10 min at 56°C. Bacterial DNA was then isolated using QIAmp spin columns according to the manufacturer’s protocol for washing and elution. Bacterial DNA concentrations were measured using a NanoDrop 1000 (Thermo Scientific, Waltham, MA).

E. coli-specific DNA was quantitated by real-time PCR using previously published primers specific for the E. coli uid A gene (30) (sense primer: 5′-AGC CAA AAG CCA GAC AGA GT-3′ and reverse primer: 5′-CAT GAC GAC CAA AGC CAG TA-3′) and normalized according to the total bacterial DNA in each sample as determined by PCR for universal 16s rDNA (sense primer: 5′-TCC TAC GGG AGG CAG CAG T-3′ and reverse primer: 5′-GGA CTA CCA GGG TAT CTA ATC CTG TT-3′) using primers and conditions previously validated and published by Nadkarni et al. (31).

For additional experiments, 2-wk-old mice were gavage fed 0.2 ml DMEM with or without 1–2 × 10⁷ CFUs/ml commensal E. coli, then returned to their mother for 24 h. Mice were then anesthetized using CO₂ inhalation and euthanized by cervical dislocation. Whole colons were isolated and immediately frozen in TRIzol for RNA isolation or assayed, ex vivo, for vulnerability to apoptosis as described later.

For analysis of the influence of exogenous IFN-αA on immature murine intestine, 2-wk-old mice were weighed to establish a baseline weight of ~6 g per mouse. Mice were injected with 40 U IFN-αA per gram of baseline body weight by i.p. injection. After 5 h, the mice were anesthetized, euthanized, and RNA isolation (32) and ex vivo analyses of staurosporine (STS)-induced apoptosis (33) were performed as previously published and briefly described.

Excised mouse colons were homogenized in TRIzol (Invitrogen Life Technologies, San Diego, CA), then subjected to phenol-chloroform extraction according to the manufacturer’s protocol and as previously described (34).

For analysis of the effects of IFN-αA, in vitro, cultured T84 and FHs 74 Int cells (ATCC) were pretreated with 10 or 100 U/ml IFN-αA for 24 and 48 h. Cells were then lysed in TRIzol, followed by phenol-chloroform extraction. RNA was digested with DNase I (Ambion, Austin, TX) to remove contamination with genomic DNA, and then cDNA was synthesized by reverse transcription using oligo(dT12–18) primers and superscript II reverse transcriptase (Invitrogen). Real-time PCR was performed using a MyIQ real-time PCR machine and SYBR Green supermix (BioRad, Hercules, CA). Forward and reverse primers recognized different exons for each gene product with the exception of IFN-αA, which contains no introns. Primer sequences were as follows: human signal recognition particle (SRP)-14 sense, 5′-AGCACTGTGGTGAGCTCCAAG-3′ and antisense, 5′-TGAGCCCATCCATGTTAGCTCTA-3′; human GBP-1 sense, 5′-GGTCCAGTTGCTGAAAGAGC-3′ and antisense, 5′-TGACAGGAAGGCTCTGGTCT-3′; murine SRP-14 sense, 5′-AAGTGTCTGTTGAGAGCCACGGAT-3′ and antisense 5′-CTGTCACTGTGCTGGTTTGCTCTT-3′; murine IFN-γ sense, 5′-TCAAGTGGCATAGATGTGGAAGAA-3′ and antisense, 5′-TGGCTCTGCAGGATTTTCATG-3′; murine TNF-α sense, 5′-CCACCACGCTCTTCTGTCTAC-3′ and antisense, 5′-TGGGCTACAGGCTTGTCACT-3′; murine IL-10 sense, 5′-ATGCTGCCTGCTCTTACTGACTG-3′, and antisense, 5′-CCCAAGTAACCCTTAAAGTCCTGC-3′; murine IFN-αA sense, 5′-CTGACCCAGGAAGACTACCT-3′ and antisense, 5′-GGCTGAGGAAGACATGGCTCT-3′; and murine GBP-1 sense, 5′-AAAAACTTCGGGGACAGCTT-3′ and antisense, 5′-CTGAGTCACCTCATAAGCCAAA-3′. Data were analyzed by the ΔΔC_t threshold cycle method and normalized to the housekeeping gene SRP-14 as previously described (35).

For in vitro analysis, T84 and FHs 74 Int cells were grown directly on glass coverslips and pretreated in media with or without 10 U/ml or 100 U/ml of IFN-αA for 24 and 48 h in a 5% CO₂ incubator at 37°C. Media was changed and apoptosis was then induced using STS (1 μg/ml) or serum-free media control for 2 h. Numbers of TUNEL-positive nuclei were counted per 10 high-power fields (HPFs).

For ex vivo analysis, intestines from 2-wk-old mice were surgically excised and opened lengthwise, exposing the intestinal epithelia. Intestines were then maintained in ex vivo organ culture, in HBSS in 24-well cell culture plates at 37°C. Tissue was incubated either in DMEM alone without serum, or with DMEM plus STS (1 μg/ml) for indicated time intervals, and then analyzed for number of apoptotic cells. Intestines were subsequently washed in PBS and frozen in embedding medium (Sakura Finetek, Torrance, CA) for histologic staining. Previous studies have reported successful maintenance of murine intestines in organ culture for the induction of apoptosis and the study of inflammatory signaling (33, 36).

After experimental treatment, T84 and FHs 74 Int cells grown on coverslips were washed and fixed in 4% paraformaldehyde (Fisher Scientific). Alternatively, intestinal tissue was fixed, paraffin embedded, and sections mounted on slides.

Apoptotic cells were analyzed by TUNEL using an INSitu Cell Death Detection Kit (Roche Diagnostic Systems, Somerville, NJ), according to the manufacturer’s guidelines.

Cells containing activated caspase 3 were detected using ApoLogix Detection Kit according to the manufacturer’s instructions (Bachem, Torrance, CA). The procedure detects apoptosis through the inhibition of activated caspase 3 via emitted fluorescence.

Equal amounts of protein were resolved by PAGE and subjected to electrophoretic transfer to activated polyvinylidene fluoride membranes. Blocking and probing were performed using SNAP-ID (Millipore, Bedford, MA) with 0.5% skim milk for blocking and with rat anti-human GBP-1 Ab (Calbiochem, San Diego, CA) for 10 min, washed three times in TBS containing 0.05% Tween 20, and bound Ab was detected by probing with species-specific peroxidase conjugated secondary Abs, followed by visualization using BM chemiluminescence substrate (Roche Diagnostic Systems).

Attenuation of GBP-1 expression was performed using sequence specific small interfering RNA (siRNA), as previously described (37). FHs 74 Int cells were grown to 60–80% confluence and then loaded with GBP-1 siRNA (27). (GBP-1_3, 5′-AAGGCATGTACCATAAGCTAA-3′ and GBP-1_6, 5′-ATGGCATGTACCATAAGCTA-3′.) Transfection with a validated siRNA against cyclophilin B (CyB) obtained from Dharmacon (Lafayette, CO) was used to control for off-target siRNA treatment. Transfections were performed using HiPerfect (Qiagen) in Opti-MEM I medium (Invitrogen) according to the manufacturer’s protocol. Twenty-four hours after loading with siRNA, cells were then treated with vehicle or indicated concentrations of IFN-αA and incubated for an additional 24 or 48 h, followed by analysis of apoptosis by staining for activated casapase-3 as described previously.

Statistical differences were analyzed by ANOVA using Prism 5 for Mac OS X, version 5.0a, 1992–2008 GraphPad Software (San Diego, CA).

In initial experiments, we characterized the developmental time frame of apoptosis in immature mouse intestine to assess whether changes in vulnerability to apoptosis might be an early step contributing to changes in epithelial barrier function, increased bacterial translocation, and finally to exaggerated inflammation and necrosis in the immature intestine.

For assessment of apoptosis at antenatal time points, the pups of timed pregnant C57BL/6 mice were delivered by cesarian section and immediately sacrificed at embryonic day 18. In contrast, for postnatal time points, dams were allowed to deliver spontaneously, and pups were maintained under conventional conditions then sacrificed at day 2, 8, 14, or 21. To assess sensitivity to apoptosis, sectioned intestine was treated ex vivo under serum-free conditions alone or in the presence of STS, a broad spectrum inducer of apoptosis, known to compete for ATP-binding, inhibiting multiple kinase signaling pathways, and causing apoptosis through both caspase-dependent and independent mechanisms (38). STS and serum-starvation were specifically chosen as proapoptotic stimuli with the goal of studying apoptosis in the absence of concomitant intestinal inflammation.

The extent of epithelial apoptosis was assessed in small intestine by TUNEL, analyzed by confocal microscopy (Fig. 1A), and TUNEL-positive cells were counted. Neither treatment had measurable effects on apoptosis during the first few postnatal days. Apoptosis in association with serum starvation was first evident in mice sacrificed at 1 wk of age only after prolonged incubation for 2 h (Fig. 1B), whereas STS-induced apoptosis was present in 1-wk-old mouse colon within 1 h of exposure to the drug (Fig. 1C). Apoptotic responses to STS serum starvation alone, at least at the shorter time interval, appeared to peak at 2 wk postnatal age (Fig. 1B). Based on these results, we concluded that vulnerability to induction of apoptosis emerges in immature mouse intestinal epithelium between 1 and 2 wk after birth. Based on these results, we selected 2 h of combined serum starvation and STS treatment of 2-wk-old mouse intestines to model mechanisms of regulation of apoptosis in immature intestine.

We have previously shown that the probiotic Lactobacillus rhamnosus GG protects against STS-induced apoptosis in the small intestine of preweaned mice (33). However, it remains unclear how early colonizing commensal flora protect against apoptosis and promote barrier integrity in the developing intestine. For this work, we initially examined whether a naturally occurring strain of commensal E. coli isolated from healthy human colon and previously characterized as anti-inflammatory (18, 29) would protect against apoptosis in immature murine intestine. In addition, we used this model to further examine the mechanisms by which commensals exert their antiapoptotic effects.

To first determine whether gavage feeding with commensal E. coli would result in a measurable change in the colonic E. coli population of developing mice, real-time PCR was used to compare colonic E. coli DNA content in 2-wk-old mice E. coli-fed mice, relative to media-fed controls. Results revealed that, when normalized for total bacterial DNA based on 16s rDNA, E. coli-fed mice showed a 1 × 10⁶ ± 0.3-fold increase in colonic E. coli uid A DNA, relative to media-fed controls (data not shown).

Given that gavage feeding with E. coli demonstrably altered intestinal flora, we next examined the influence of this E. coli exposure on intestinal apoptosis. For these experiments, 2-wk-old mice were fed media with or without 1–2 × 10⁷ CFU of commensal E. coli by oral gavage, and sacrificed 24 h later. Then, their intestines were excised and analyzed for apoptosis in response to STS, ex vivo. TUNEL staining of controls revealed a significantly lower baseline rate of apoptosis as well as a more robust induction of apoptosis in response to STS in the small intestine (Fig. 2A) than in colon (Fig. 2B). These results are in contrast to previously published comparisons of baseline apoptotic rates in the adult murine intestine showing either similar rates of apoptosis in small intestine and colon (39) or lower baseline apoptosis in colon relative to distal small intestine (40, 41). However, gnotobiotic pigs demonstrate increased TLR-mediated intestinal apoptosis during conventionalization with commensal flora, including E. coli (42), and furthermore, that these rates of apoptosis were dependent on the composition of the colonizing flora. These increased rates of baseline of apoptosis in colon relative to small intestine may represent early responses to a greater burden of bacteria during colonization of the immature colon. In our model, E. coli feeding had no significant effect on baseline apoptosis, but attenuated the response to STS, reducing STS-induced apoptosis by nearly 50% in both the small intestine (Fig. 2A) and colon (Fig. 2B) of E. coli-fed mice relative to STS-treated, media-fed controls.

In an effort to better understand the mechanisms by which commensal E. coli protects against STS-induced apoptosis, we next profiled the influence of E. coli exposure on the expression of a panel of cytokines in the colon, the site exposed to the greatest burden of colonizing bacteria. RNA was isolated from 2-wk-old mouse colon 24 h after media or E. coli feeding, and mRNA expression of IFN-αA, IFN-γ, IL-10, and MIP-2 was analyzed by real-time PCR (Fig. 3A). IFN-αA mRNA was selectively induced by 10.2 ± 2.8-fold after feeding with commensal E. coli, whereas levels of MIP-2, IFN-γ, and IL-10 were not significantly altered, relative to media-fed controls. This selective induction of IFN-αA by commensal E. coli, supports a role for IFN-αA in mediating the downstream effects of E. coli in immature intestine and is consistent with the previously published work implicating TLR9-mediated induction of IFN-αA in probiotic-mediated cellular signaling (22).

Members of the IFN-α family are unique among the type I IFNs in that they prevent rather than promote apoptosis through the specific activation of antiapoptotic target genes (43). One such target is the gene that encodes GBP-1, a large GTPase, which promotes cell survival and prevents apoptosis in multiple cell types. GBP-1 has been identified in endothelial cells as preventing apoptosis in response to both IFN-α (26) and IFN-γ (44). Our group has recently reported that GBP-1 can preserve intestinal epithelial barrier function by blocking apoptosis both at baseline and in response to the proinflammatory cytokine, IFN-γ (27). However, GBP-1 has not been studied in the developing intestinal epithelium and nothing is known about its contributions to responses to bacterial colonization or response to microflora. To evaluate GBP-1 as a candidate mediator of the antiapoptotic effects of E. coli in our murine model of immature intestine, we analyzed colonic mRNA expression of GBP-1 in E. coli-fed 2-wk-old mice by real-time PCR. Twenty-four hours after E. coli administration, GBP-1 mRNA expression was induced by 2.8 ± 0.8-fold relative to media-fed controls (Fig. 3B).

To confirm the role of IFN-αA in GBP-1 induction by commensal E. coli, we next examined the influence of IFN-αA treatment on GBP-1 expression and apoptosis in immature intestinal epithelium. As an in vitro model of immature epithelia, FHs 74 Int cells, derived from human fetal small intestine, were treated for 4 or 24 h with either 10 or 100 U/ml IFN-αA and then analyzed for GBP-1 mRNA expression (Fig. 4A). Treatment with either 10 or 100 U/ml IFN-αA resulted in an induction of GBP-1 mRNA, with expression peaking at a 16.1 ± 5.3-fold increase after 4 h of treatment with 100 U/ml IFN-αA. and declining by 24 h to a 4.9 ± 1.8-fold increase over controls. The effects of IFN-α on GBP-1 were dose dependent, with more modest induction seen in response to 10 U/ml IFN-αA. Western blot analysis revealed a rapid induction in GBP-1 protein expression in response to 100 U/ml IFN-αA (Fig. 4B). This increase over baseline GBP-1 protein expression was first evident after 4 h of IFN-αA exposure and peaked by 8 h, but remained elevated throughout the 24 h assayed. A similar, although more rapid and robust dose- and time-dependent induction of GBP-1 mRNA and protein was seen in response to IFN-αA treatment in T84 cells, a well-established model for mature intestinal epithelium (data not shown).

To evaluate the influence of IFN-αA on apoptosis in model epithelia, in vitro, we treated FHs Int 74 cells with IFN-αA for 24 or 48 h prior to induction of apoptosis with STS. Epithelial cell apoptosis was quantitated by TUNEL and activated caspase-3 staining. Even at the lower dose of 10 U/ml, IFN-αA significantly reduced the number of TUNEL positive cells by 2.6-fold in FHs 74 Int cells (Fig. 5A). The influence of IFN-αA on STS-induced apoptosis in FHs 74 Int cells was also assayed through staining for and quantitation of the number of FHs 74 Int cells positive for activated caspase-3 (Fig. 5B), with statistically significant reductions in STS-induced apoptosis observed after 4 h of treatment with low-dose IFN-αA. Similar reductions in apoptosis were observed after both treatment with higher doses and longer incubations with IFN-αA. Similar doses of IFN-αA also protected T84 cells from STS-induced apoptosis (data not shown).

To determine whether this antiapoptotic effect of IFN-αA was dependent on GBP-1, GBP-1 expression was attenuated using previously validated, GBP-1-specific, siRNA (27). FHs 74 Int cells were loaded with either GBP-1 specific siRNA or CyB siRNA, as an off target siRNA control, then treated with IFN-αA for 24 h, and analyzed for GBP-1 protein expression by Western blot (Fig. 6A). GBP-1 siRNA treatment attenuated GBP-1 expression both at baseline and in response to IFN-αA treatment. FHs 74 Int cells treated with either GBP-1 siRNA or with control siRNA control were then exposed to STS and analyzed for resultant apoptosis by staining for activated caspase-3. Control siRNA (CyB)-treated FHs 74 Int cells, still demonstrated significant IFN-αA–dependent protection against STS-induced apoptosis, relative to vehicle-treated controls (Fig. 6B). Interestingly, in the absence of both STS and IFN-αA, siRNA-mediated attenuation of GBP-1 expression resulted in a significant increase in baseline apoptosis (Fig. 6C) relative to CyB siRNA-treated controls (Fig. 6B, p < 0.05). In addition, siRNA-dependent attenuation of GBP-1 expression completely eliminated the antiapoptotic effects of IFN-αA (Fig. 6C). Thus GBP-1 expression is required both for IFN-αA–mediated prevention of STS-induced apoptosis and for prevention of baseline apoptosis in FHs 74 Int cells.

Given that E. coli exposure was associated with both prevention of STS-induced apoptosis and induction of IFN-αA expression in 2-wk-old mouse colon, we sought to determine whether exogenous IFN-αA would be similarly protective against apoptosis in our ex vivo model of immature intestines. The 2-wk-old mouse pups were treated with IFN-αA at a dose of 40 U/g body weight by i.p. injection, based on previously published doses shown to ameliorate colitis in IFN-αA–deficient animals in adult murine models of colitis (32). Based on our in vitro studies indicating that exogenous IFN-αA–induced GBP-1 protein expression in intestinal epithelial cells within 4 h of treatment, mice were injected i.p.with either IFN-αA, 40 U/g body weight or an equal volume of sterile PBS . Mice were then returned to their mothers for 5 h to allow adequate time for absorption and distribution of IFN-αA and subsequent GBP-1 induction, then sacrificed and analyzed for both for colonic GBP-1 mRNA expression and vulnerability to apoptosis in both small and large intestine. At this dose and interval, IFN-αA treatment had no measurable effect on apoptosis relative to PBS-injected controls at baseline. However, exogenous IFN-αA effectively blocked STS-induced apoptosis in immature mouse small intestine (Fig. 7A) with a statistically significant decrease in the number of TUNEL positive cells down to the level of non-STS–treated controls. STS-induced apoptosis was also significantly reduced in colon of IFN-αA–treated 2-wk-old mice (Fig. 7B). Furthermore, real-time PCR analysis of colonic mRNA expression confirmed a corresponding significant 2.6-fold induction in GBP-1 mRNA in colon from 2-wk-old IFN-αA–treated mice relative to controls (Fig. 7C).

In this study, we present evidence that apoptotic responses either to serum starvation or STS emerge between 1 and 2 wk in the intestine of developing mice, implicating this window as an optimal interval for the study of mechanisms regulating inducible apoptosis during intestinal development. Previous work has clearly shown that this developmental interval is associated with increased epithelial turnover in rat small intestine (45, 46). It is unclear whether this process is associated with increased baseline apoptosis in mice. Interestingly, we found little or no significant change in baseline apoptosis in mice at this age, with the primary finding, ex vivo, being an increased response to injury, indicating that the immediate tissue response to the same apoptotic stimulus is accentuated at this age.

In addition, we confirm our earlier findings that probiotic or commensal flora, in this case a commensal strain of E. coli, can modulate intestinal epithelial apoptosis in the developing murine gut (33). As novel findings, we further demonstrate that commensal E. coli exposure modifies cytokine expression in immature murine intestine, selectively inducing IFN-αA and the subsequent expression of the antiapoptotic protein, GBP-1. Additional work using both mature and fetal intestinal epithelia reveals that IFN-αA administration also induces GBP-1 and protects against STS-induced apoptosis in these model human systems. Furthermore, studies in fetal intestinal epithelial cells reveal that the antiapoptotic properties of IFN-αA require GBP-1 expression, in vitro. Finally, the effects of E. coli both on intestinal GBP-1 expression and epithelial apoptosis are recapitulated by the administration of exogenous murine IFN-αA alone.

The importance of the resident microbiota in the developing intestine has become increasingly evident. It is known that neonates are born with a sterile gut and become colonized after birth. In the preterm infant, this process can be tainted because of the immaturity of the gut, the lack of breastfeeding, use of antibiotics, and exposure to iatrogenic lines and tubes for nutrition (47, 48). Specifically, disturbance of normal colonization through the use of broad-spectrum antibiotics significantly increases the incidence of NEC and related death in premature infants (14). Furthermore, the establishment of appropriate commensal flora and the use of probiotic organisms have shown promise in the prevention of NEC in clinical trials (49–54). However, therapy with commensal or probiotic organisms has been complicated by case reports of documented sepsis in immunocompromised patients, including premature infants (55–57). Therefore, it has become increasingly important to understand the mechanisms by which appropriate intestinal colonization with nonpathogenic flora protects the developing intestine. A better understanding of these beneficial effects will allow more specific and safer targeting of these protective pathways with the hopes of preventing NEC.

In our study, we examined potential protective mechanisms mediated by commensal E. coli, one of the earliest colonizers of the newborn human intestine. Commensal organisms are known to signal via TLRs, and more specifically TLR-9, a known inducer of type I IFN signaling (22, 24). It is also known that type I IFNs have anti-inflammatory properties and beneficial effects in experimental models of IBD (32) and NEC (58), as well as in clinical trials for the treatment of IBD (59, 60). In this study, we demonstrate that commensal E. coli selectively induces IFN-αA and protects against apoptosis in developing intestine.

The role of type I IFNs in the regulation of apoptosis is determined by the nature of activated target genes and is therefore both cell-type specific and dependent on the class of type I IFN. Although IFN-β is known to have potent proapoptotic effects (43), promoting apoptosis in several cancer (61, 62) and immune cells (63, 64), IFN-α activates different downstream targets to prevent apoptosis in endothelial cells (26) and neuronal tissue (65). IFN-α has also previously been reported to protect against apoptosis and promote barrier integrity in polarized intestinal epithelial cell monolayers (66). Furthermore, the protective effects of TLR-9 signaling in experimental colitis appear to result from cell-type specific effects of type I IFNs, which promote barrier function through the prevention of epithelial apoptosis while inhibiting the inflammatory responses of activated macrophages (66).

IFN-αA has been found to attenuate apoptosis in nonepithelial cell lines through the induction of GBP-1 (26). Recently our group has shown that GBP-1 localizes to epithelial tight junctions, is induced in the intestinal epithelium of patients with IBD, and promotes intestinal epithelial barrier integrity through the prevention of apoptosis, in vitro (27). In this study, we provide further specific evidence that the IFN-αA–mediated induction of GBP-1 in response to commensal flora attenuates apoptosis in models of immature intestine, both in vitro and ex vivo.

Previous work has demonstrated that baseline fluctuations in epithelial apoptosis do not appear to alter barrier integrity. However, increased epithelial apoptosis in response to exogenous toxic stimuli, such as doxorubicin, is associated with measurable bidirectional changes in permeability in adult rats (67). Therefore, the impact of proapoptotic stimuli on epithelial barrier is likely to be dependent on the magnitude of resultant apoptosis. Furthermore, the antiapoptotic effects of commensals, such as nonpathogenic E. coli, may help prevent disruption of intestinal epithelial barrier integrity in the developing gut. However, the current existing models for neonatal human epithelial cells require growth conditions, which are not amenable to the assay of barrier function. Therefore, the contributions of the regulation of apoptosis to barrier function in the immature intestine could not be assessed in our in vitro model system.

However, our studies identify a novel and specific role for GBP-1 as a regulator of epithelial apoptosis in response to commensal E. coli in developing intestine. Members of the murine family of GBPs have been implicated as critical elements of host defense activated in both macrophages and fibroblasts in response to TLR signaling and proinflammatory cytokines as well as in response to infection with both bacteria and parasites, in vivo (68). These studies represent the first characterization of GBP-1 regulation and its importance in the protection of the immature intestinal epithelium against apoptosis in response to microflora. Additional studies are needed to determine whether this IFN-αA–mediated induction of GBP-1 is specific to nonpathogenic E. coli, commensal organisms in general, or a more global response to microflora, independent of the nature of the organism.

The induction of GBP-1 has previously been shown to prevent apoptosis, protecting epithelial barrier in the presence of proinflammatory mediators (IFN-γ, TNF-α) both in vitro and in the setting of IBD (27), and here we demonstrate an important protective role for GBP-1 in response to commensal flora in the developing intestine. Recently, murine GBP-1 was identified by combined quantitative trait loci and microarray analysis as once of three genes to be differentially upregulated in conventionally raised, colitis-sensitive strains of IL-10^−/− mice, relative to more colitis-resistant strains. Specifically, GBP-1 expression was upregulated in IL-10^−/− C3H/HeJBir mice, which develop severe, spontaneous, commensal-dependent colitis relative to IL10^−/− C57BL/6 mice, which develop only mild spontaneous intestinal inflammation (69). Given the antiapoptotic and barrier protective roles of GBP-1, this strain-specific induction of GBP-1 most likely represents a compensatory and protective mucosal response in the intestines of mice, which are otherwise more vulnerable to inflammatory responses to colonizing bacteria in the setting of IL-10 deficiency.

Although our ex vivo and in vivo systems are intended to mimic responses to commensal bacteria in developing intestine, the exact relevance of our findings to NEC remains uncertain. However, these findings are particularly intriguing given that the developing human intestine is vulnerable to barrier compromise, bacterial translocation, and unchecked inflammatory responses culminating in NEC (16). We reveal in this study that developing murine intestine is deficient both in baseline and commensal-driven IL-10 expression. Our findings argue a novel role for GBP-1 in acute responses to commensal flora in the immature intestine and may indicate a specific role for GBP-1 in response to colonization at a developmental interval when IL-10 is relatively deficient. Whether GBP-1 is critical for initial colonization or more generally significant for defense of developing intestine against existing microflora has yet to be determined. Further characterization of the regulation of GBP-1 during intestinal development may better inform our understanding of the maintenance of epithelial barrier in immature intestine and its failure in NEC.

Disclosures The authors have no financial conflicts of interest.