In Vivo Pattern of Lipopolysaccharide and Anti-CD3-Induced NF-κB Activation Using a Novel Gene-Targeted Enhanced GFP Reporter Gene Mouse¹

Nuclear factor κB is an inducible family of heterodimeric transcription factors comprised of cRel, RelA, RelB, p50, and p52 (synthesized as p105 and p100 precursors, respectively) (1), which drives the expression of a set of diverse genes involved in numerous biological process such as inflammation, innate/adaptive immunity, cancer, and development (2).

NF-κB association with the cytoplasmic inhibitor IκB prevents the transcription factor from binding to consensus cis elements and thus inhibits transcriptional activity. The most commonly recognized pathway to trigger NF-κB activation involves signal-induced proteasome-mediated IκB degradation (3). This mechanism requires inducible serine phosphorylation of IκB, which occurs at positions 32 and 36 for IκBα (4). The kinase responsible for inducible IκB phosphorylation is the IκB kinase (IKK)³ complex, composed of the catalytic IKKα and IKKβ subunits, and the regulatory element IKKγ (5, 6). Phosphorylation is followed by the activation of a complex enzymatic system (E1, E2, E3) that adds multiple ubiquitin proteins at lysine residues 21 and 22 of phosphorylated IκBα (3). Ubiquinated IκBα is then selectively and rapidly degraded via the nonlysosomal, ATP-dependent, 26S proteolytic complex (3). Destruction of IκBα reveals the NF-κB nuclear localization signal and allows nuclear translocation, binding to κB-cis elements, and induction of gene transcription. In addition to subcellular localization, posttranslational modification such as phosphorylation and acetylation of some NF-κB subunits strongly modulate NF-κB transcriptional activity (7, 8, 9, 10, 11).

Because of its pivotal role in driving expression of multiple genes involved in adaptive and innate immunity, NF-κB activation has been extensively measured in various inflammatory disorders. For example, tissues from patients with rheumatoid arthritis, asthma, atherosclerosis, and inflammatory bowel diseases showed enhanced p65 nuclear localization and DNA-binding activity as measured by immunohistochemistry or EMSAs, respectively, indicating that NF-κB is associated with the inflammatory process (12, 13). Although informative, these techniques do not measure the dynamic nature of NF-κB activation and transactivation ability in vivo and are usually insufficient to identify specific cell types within tissues.

Recent findings suggest that NF-κB may have a protective role during the course of inflammation. We demonstrated that transgenic mice expressing an IκBα superrepressor restricted to intestinal epithelial cells (IEC) display an increased sensitivity to dextran sulfate sodium-mediated colitis (14), whereas mice with an intestine-specific knockout of IKK are more sensitive to intestinal ischemia/reperfusion injury (15). In related studies, mice lacking both p50 and one p65 allele are more susceptible to Helicobacter hepaticus-induced typhlocolitis compared with wild-type (WT) mice (16). Finally, the Crohn’s disease mutation leading to truncation of Nod2 results in decreased NF-κB activation in response to LPS (17). In addition, NF-κB-induced genes have been associated with the resolution phase of inflammation in vivo (18), implicating again that this transcription factor may have beneficial effects during selective phases of inflammation.

Although NF-κB has been implicated in acute and chronic inflammation as well as in tissue repair, the exact function of NF-κB is still unclear. The above-mentioned studies highlight the need for a more detailed characterization of NF-κB activation in vivo during acute and chronic phases of inflammation and repair, and demand a thorough investigation of the positive and negative impacts of manipulating signaling pathways that modulate NF-κB activation. However, limited approaches are available to directly measure NF-κB activity in vivo. In this regard, two transgenic animal models using NF-κB-driven lacZ (19, 20) or luciferase gene expression (21) have been generated to study NF-κB activation in vivo. These two models represent a clear improvement over the pre-existing in vitro approaches to study NF-κB activity; however, these models require exogenous substrate and/or posthumous analysis of prepared tissues or cells and therefore fail to address the instantaneous state of NF-κB activation in different organs. To this date, no animal models have allowed the precise measurement and location of NF-κB-activated cells during various physiological situations following treatment.

In this study, we report the generation and characterization of a novel transgenic mouse strain which utilizes NF-κB-dependent enhanced GFP (EGFP) reporter gene expression. We show that reporter gene expression is specifically induced by NF-κB, correlates with p65 localization to the transgene promoter, and responds in vivo to stimulation with LPS and anti-CD3. In addition, we use this transgenic mouse as a source of dendritic cells, splenocytes, and hepatic stellate cells (HSCs) to demonstrate the feasibility of conducting reporter assays in primary cells. Moreover, we demonstrate in vivo differences in regional and cellular NF-κB activation in the intestine following challenge with LPS. Thus, this strain can be used to study the spatial and temporal activation of NF-κB in vivo and ex vivo in response to physiological inducers.

To generate the cis-NF-κB^EGFP targeting construct, a chimeric promoter containing three HIV NF-κB cis elements, 5′ of the minimal c-fos promoter was isolated from Δ56FosdE-Lux (22 ; Fig. 1). The NF-κB promoter fragment was then ligated into the targeting vector (hypoxanthine phosphoribosyltransferase (HPRT)/EGFP) (23). The preceding cloning steps were also followed to generate a mutant (MT) NF-κB targeting vector (the MT NF-κB chimeric promoter (22) contains three mutated NF-κB cis elements: GCGGATTCCC). Fifteen micrograms of the targeting constructs, cis-NF-κB^EGFP or MT cis-NF-κB^EGFP, was linearized with PvuI, suspended in 0.5 ml of PBS containing 1 × 10⁷ HPRT-deficient BK4 (24, 25) ES cells, and pulsed (350 V, 50 μF for 1 s) in a 4-mm gap electroporation cuvette. The mouse ES cell line BK4 (a subclone of E14TG2a) was maintained in an undifferentiated state on embryonic fibroblast feeder cells in DMEM high glucose, supplemented with 15% FBS (ES qualified; Life Technologies, Grand Island, NY), 0.1 mM 2-ME, 2 mM l-glutamine, 100 U/ml penicillin, and 1 mg/ml streptomycin. ES cells were grown at 37°C in a humidified atmosphere containing 5% CO₂. Twenty-four hours after electroporation, the growth medium was replaced with new medium containing hypoxanthine/aminopterin/thymidine (HAT; final concentration of 0.1 mM hypoxanthine/0.4 μM aminopterin/16 μM thymidine). HAT-resistant clones were isolated and expanded 10 days after initiation of selection. Individual clonal populations were used in all experiments unless otherwise indicated.

Targeted ES cells were injected into day 3 C57BL6 strain blastocysts. Chimeric mice were crossed to C57BL6 mice and agouti F₁ offspring were tested for the cis-NF-κB^EGFP transgene by PCR amplification of EGFP contained in the construct (EGFPFOR, 5′-GAG CTG AAG GGC ATC GAC TTC AAG-3′; EGFPREV, 5′-GGA CTG GGT GCT CAG GTA GTG G-3′: amplicon: 246 bp). For all experiments, F₁ EGFP-positive heterozygotes were crossed to C57BL6 strain mice. WT littermates were used as controls for all experiments. F₁ cis-NF-κB^EGFP mice were genotyped by collecting a drop of blood (∼5 μl) from a tail vein nick (using heparinized capillaries); placing the drop of blood on a glass microscope slide with a glass coverslip, and using fluorescence microscopy to identify EGFP-positive mononuclear cells. To conclude that “phenotype” (i.e., EGFP⁺ cells in blood) correlated with “genotype,” we amplified EGPF cDNA from genomic DNA isolated from the tails. Using this genotyping method, we were able to determine that the cis-NF-κB^EGFP transgene was inherited as a Mendelian sex-linked trait and, furthermore, genotype could be directly correlated with phenotype.

For whole organ analysis, animals were sacrificed and dissected followed immediately by image analysis. Live tissues and organs were imaged using either a Magnafire SP charge-coupled device camera (Optronics, Goleta, CA) in a light-tight imaging box with a filtered light source and emission filters specific for EGFP (LT-9900, excitation: 470 nm/40 nm; emission: 515 nm, Lightools Research, Encinitas, CA), or alternatively, an EGFP-specific filter (XF116-2; Omega Optical, Brattleboro, VT) was connected to a fiber optic light source and images were captured using a Nikon Cool-pix 990 digital camera (Nikon, Melville, NY) attached to a dissecting microscope (Lieder, Ludwigsburg, Germany). Identical exposure times were used to capture images within each experiment.

Epifluorescence microscopy was used to detect EGFP in live cells and tissue sections from cis-NF-κB^EGFP mice. For tissue sections, liver and spleens were resected from mice following treatment with the NF-κB stimulus described in the text, fixed in 4% paraformaldehyde (PFA) for 10–24 h, washed twice in PBS, and transferred to vials containing 30% sucrose (made in PBS) for 24 h. Five- to 7-μm sections were cut on a cryostat and mounted in glycerol-based mounting medium before microscopic analysis. EGFP expression was imaged using an Olympus IX70 (Olympus, Melville, NY) fitted with EGFP-specific filters (XF116-2; Omega Optical). Images were captured using a digital SPOT^M camera (Diagnostic Instruments, McHenry, IL). Identical exposure times were used for each data point within an individual experiment.

RBC-depleted splenocytes from cis-NF-κB^EGFP mice were cultured in ultralow adherence six-well plates (Costar, Corning, NY) in RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine, 50 μM 2-ME, 50 μg/ml gentamicin, 10 ng/ml murine rGM-CSF (PeproTech, Rocky Hill, NJ), and 1 ng/ml human rTGF-β1 (R&D Systems, Minneapolis, MN). After 7 days of culture, cells were collected, washed, and replated with fresh medium. Cultures were then monitored daily and split as the cells became dense, usually every 2–3 days. After 3 wk of culture, the resulting population was >93% CD11c⁺ and CD11b⁺ as determined by flow cytometry. For in vitro stimulation, cultured dendritic cells were harvested, washed, and resuspended in complete medium without additional cytokines. Cells were stimulated with cecal bacteria lysate (10 μg/ml) for 12 h to assess EGFP expression.

Flow cytometry was used to detect EGFP in individual live cells following in vitro or in vivo stimulation. Peripheral blood was isolated by thoracic puncture and placed into sodium-heparin vials. RBC from whole blood were lysed using RBC lysis buffer (0.15 M NH₄Cl, 10 mM KHCO₃, and 0.1 mM Na₂EDTA, pH 7.2), and PBMC were pelleted at 800 × g for 10 min, washed once with PBS, placed into 35-mm tissue culture dishes and stimulated with TNF-α (30 ng/ml).

For anti-CD3/anti-CD28 treatment, 200 μl (10 μg/ml) of anti-CD3 (BD PharMingen, San Diego, CA) was added to a 48-well plate and allowed to coat the plate for 2 h at room temperature. The plates were then washed twice with PBS/5% FCS and 0.5 ml of complete culture medium (RPMI 1640, 5% heat-inactivated FCS, 5 × 10⁻⁵ M 2-ME, 2 mM glutamine, and 1 mM pyruvate) was added for 30 min. PBMC were plated at ∼50,000 cells/well and anti-CD28 (BD PharMingen) was added to a concentration of 2 μg/ml. At specified times following stimulation, PBMC were washed once in PBS, pelleted, and resuspended in FACS analysis buffer (0.1% BSA/0.01% azide in PBS) just before FACS analysis.

HSCs cells were isolated as described previously (26). Twenty-four hours after isolation, HSCs were pretreated for 2 h with WT or MT NEMO-binding peptide (NBP; 200 μg/ml) and then stimulated with TNF-α (30 ng/ml) for 18 h. HSCs were trypsinized, washed once with PBS, pelleted at 800 × g, and resuspended in FACS analysis buffer before quantification.

Splenocytes were isolated as described above and pretreated with the inhibitor of IκB phosphorylation, BAY 11-0782 (25 μM), for 20 min and then plated in 48-well tissue culture plates coated with anti-CD3/CD28 (4 and 2 μg/ml, respectively). Eighteen hours after anti-CD3/CD28 stimulation, splenocytes were washed once with PBS, pelleted at 800 × g, and resuspended in FACS analysis buffer before analysis.

Conventionally housed cis-NF-κB^EGFP mice or WT littermate control mice were injected i.p. with LPS (5 mg/kg) or anti-CD3 (0.2 mg). At specified times, mice were euthanized using halothane and organs were resected. Splenocytes and thymocytes were isolated by mechanical mincing followed by separation through a 70-μm cell strainer (Falcon, Franklin Lakes, NJ). RBC were lysed and PBMC were isolated as described above and resuspended in FACS analysis buffer before analysis.

An established method to detect Kupffer cells (resident liver macrophages) is to elicit phagocytosis of fluorescent tracer spheres (27). One-micrometer red fluorescent spheres (F13083; Molecular Probes, Eugene, OR) were opsonized with normal rabbit serum for 30 min and the spheres were then washed five times with 1.5 ml of PBS. Fluorescent spheres (2 × 10⁹) were suspended in 200 μl of PBS and injected into the tail vein. Two to 3 h postinjection, mice were perfused with PBS for 10 min to remove unphagocytosed spheres in capillaries. The livers were then resected and fixed in 4% PFA overnight at 4°C. The livers were then washed twice in PBS and placed in 30% sucrose for 24 h before cryostat sectioning. Spheres were imaged by fluorescent microscopy using standard rhodamine excitation and emission filters.

Small bowel was resected 16 h after LPS treatment (as described above) and fixed in 4% PFA for 8 h at room temperature. The tissue sections were washed twice in PBS and then placed in 30% sucrose at 4°C for at least 16 h before sectioning. For immunostaining, transverse cryosections (10 μm) of the small bowl were placed on glass slides and blocked for 1 h in “blocking solution” (1% normal goat serum (anti-CD64) or 1% FCS (anti-CD3ε), 1% Triton X-100 in PBS). The sections were then washed two times for 5 min at room temperature using 0.1× blocking buffer. Polyclonal Abs, anti-CD3ε M-20, and anti-CD64 H-250 (Santa Cruz Biotechnology, Santa Cruz, CA) were diluted 1/500 in 0.1× blocking buffer, applied to sections, and allowed to incubate for 16 h at 4°C in a humidified chamber. The sections were then washed three times with 0.1× blocking buffer at room temperature for 10 min. Either anti-goat-Cy3-conjugated (for anti-CD3ε) or anti-donkey-Cy3-conjugated (for anti-CD64) secondary Abs were diluted 1/200 in 0.1× blocking buffer, applied to sections, and allowed to incubate at room temperature for 1 h in a humidified chamber. The sections were then washed five times with 0.1× blocking buffer at room temperature for 5 min each. Immunostaining was visualized using confocal imaging as described earlier.

Flow cytometry determination was performed as previously described (23). Briefly, 5 × 10⁵ RBC-depleted splenocytes or PBMC were washed with PBS containing 0.1% BSA and 0.01% azide and EGFP expression was measured on a FACscan (BD Biosciences, Mountain View, CA) using the FL1 channel to detect EGFP fluorescence. Coexpression of the early activation marker CD69 was measured by surface staining with anti-CD69-PE (Caltag Laboratories, Burlingame, CA) for 20 min on ice followed by washing in PBS containing 0.1% BSA and 0.01% azide before FACScan analysis. All flow cytometric analyses were conducted on living cells within 30 min of isolation.

Mice were administered either 2.5 mg/kg body weight LPS or PBS i.p. vehicle for 2 h and splenocytes were isolated as described above. Splenocytes were resuspended in 1 ml of PBS and fixed by addition of 100 μl of formaldehyde solution (11% formaldehyde, 100 mM NaCl, 1 mM EDTA (pH 8.0), 0.5 mM EGTA, and 50 mM Tris-Cl (pH8.0)). DNA was sheared and ChIP was performed as described previously (28).

Real-time PCR was used to semiquantify RelA binding to various gene promoters. The following promoter-specific oligonucleotides were used to amplify ChIP DNA: κBChIPFOR, 5′-CGAATTCTGCAGGTCGACGGAAAG-3′; κBChIPREV, 5′-CGGTGAACAGCTCCTCGCCCTTC–3′; RANTESFOR, 5′-GTGAAGACCAATGGCTTGACC-3′; RANTESREV, 5′-CATGTGCTGTCTCAGAGTCCTC-3′; MIP-2FOR, 5′-CAACAGTGTACTTACGCAGACG-3′; MIP-2REV, 5′-CTAGCTGCCTGCCTCATTCTAC-3′; IL-6FOR, 5′-GACATGCTCAAGTGCTGAGTCAC-3′; IL-6REV, 5′-AGATTGCACAATGTGACGTCG-3′; MCP-1FOR, 5′-CAGCATCTGGAGCTCACATTCC-3′; and MCP-1REV, 5′-GCATGAACAAGTTGAGAGATGCC-3′.

One microliter of immunoprecipitated DNA or total DNA input control was used in a standard real-time PCR using a LightCycler real-time PCR analyzer and LightCycler cyber green reagents (Roche, Basel, Switzerland). Three millimolar MgCl₂ was required to produce specific amplification products with no primer-dimer formation. The PCR was cycled at 97°C, 10 s; 58°C, 5 s; and 72°C 30 s for 40 cycles. Linear amplification ranges were determined using LightCycler analysis software and reaction crossing points were applied to the formula described in the study of Pfaffl (29) to determine increases of RelA binding in control vs treated animals. Identical crossing points in “input” DNA from control and treated mice indicated that amounts of DNA put into the immunoprecipitation reaction were equivalent.

The adenoviral vector (Ad5NIK) encoding a WT NF-κB-inducing kinase, a strong inducer of NF-κB activity, and the control virus Ad5LacZ were amplified and purified as described elsewhere (30, 31). The viruses were dialyzed twice against PBS for 8 h and once overnight in 5% sucrose (made in PBS). Mice were injected i.v. with 200 μl of PBS or 1 × 10⁷ or 1 × 10⁹ PFU of virus. EGFP expression was assessed 7 days following Ad5NIK infection.

To study NF-κB-dependent gene transcription in a spatial and temporal manner, we generated a transgenic mouse expressing the EGFP under the transcriptional control of NF-κB. Since NF-κB is a ubiquitous transcription factor involved in many homeostatic biological processes, it was necessary to minimize transgene expression due to basal NF-κB activation to be able to observe increases in transgene expression following inductive stimuli. Therefore, we used a gene-targeting approach to integrate a single copy of the NF-κB reporter construct, in a single locus, 5′ of the mouse HPRT gene. Gene targeting was chosen over standard pronuclear injection to generate the transgenic mouse to assure the strain would harbor only a single copy of the transgene, integrated in a well-characterized locus that has permitted the appropriate expression of other transgene constructs (25, 32). The NF-κB-dependent reporter consisted of three HIV NF-κB sites, 5′ of the minimal c-fos promoter, driving expression of the EGFP reporter gene, which was flanked by HPRT gene homology (Fig. 1,A). Following homologous recombination into a male-derived, HPRT-deficient ES cell line (25), the transgene is integrated 5′ of the promoter with concomitant rescue of the HPRT gene defect (Fig. 1 A). A successful homologous recombination event is permissive for ES cell growth in HAT medium; conversely, random integration of the targeting vector does not restore HPRT gene expression, resulting in cell death in HAT medium.

Following gene targeting and HAT selection, ES cell colonies proliferated and were cloned for analysis of cis-NF-κB^EGFP transgene activity. Spontaneously differentiated clonal lines of the cis-NF-κB^EGFP ES cells displayed induced expression of EGFP, whereas the undifferentiated, pluripotent ES cells exhibited no EGFP expression demonstrating the inductive capability of the cis-NF-κB^EGFP transgene at the cellular level (Fig. 1 B). ES colonies containing mutant NF-κB sites exhibited no EGFP expression (data not shown).

Chimeric mice were generated from a single clone of cis-NF-κB^EGFP ES cells and bred to C57BL6 mice to produce heterozygous female mice. A second breeding step of transgene-positive, F₁ generation females to C57BL6 males was required to produce a hemizygous male. Mice were easily genotyped by the presence of EGFP-positive mononuclear cells in peripheral blood and demonstrated Mendelian sex-linked segregation of the transgene. Upon gross analysis of whole organs from cis-NF-κB^EGFP mice, basal levels of EGFP expression were observed in the lymph nodes, Peyer’s patches, thymus, and epididymis, which are known to exhibit high basal levels of NF-κB activation (33, 34) (Fig. 1 C).

Although the transgene was targeted to an X-linked locus, we did not observe any significant differences in the number of EGFP-expressing PBMC in males (n = 5) and heterozygous females (n = 5), suggesting this transgene is resistant to X chromosome inactivation (data not shown). We have initiated breeding to generate homozygous females to investigate the impact of transgene copy number on EGFP expression levels.

Recruitment of NF-κB subunits, especially RelA, to a defined promoter region is prerequisite for transcriptional activation. To selectively assess the recruitment of the potent transcriptional activator RelA to the cis-NF-κB^EGFP promoter in the transgenic mice, we performed ChIP analysis on various well-defined NF-κB responsive genes (2) following in vivo challenge with LPS. LPS has been demonstrated to be a strong inducer of NF-κB activation by signaling through the TLR4 (35, 36, 37). Two hours after injection of LPS into cis-NF-κB^EGFP-transgenic mice, splenocytes were isolated and the ChIP assay was performed. As predicted, we found enhanced RelA loading on the promoter of cis-NF-κB^EGFP mice treated with LPS compared with PBS-treated mice (11.7-fold; Fig. 2), indicating the promoter of the cis-NF-κB^EGFP transgene functionally binds the endogenous RelA subunit upon stimulation. Moreover, there was enhanced RelA recruitment to the endogenous promoters of the NF-κB responsive genes, IL-6, MIP-2, RANTES, and MCP-1 in LPS-treated mice compared with control-treated mice (Fig. 2). This demonstrates an effective recruitment of RelA to the cis-NF-κB^EGFP promoter as well as to the promoters of other NF-κB-responsive genes in vivo.

To determine the involvement of NF-κB signal transduction in stimuli-induced cis-NF-κB^EGFP transgene expression, we specifically blocked NF-κB activation using NBP (38), an inhibitor of IKK activity, and BAY 11-0782, an inhibitor of IκBα phosphorylation (39) in two cell types isolated from cis-NF-κB^EGFP mice. Primary HSCs were isolated from cis-NF-κB^EGFP mice, cultured for 10 days, pretreated with the NBP, and then stimulated with the well-characterized NF-κB inducer, TNF-α, for 12 h. TNF-α-induced EGFP expression was completely blocked in NBP-treated cis-NF-κB^EGFP HSCs compared with those treated with the mutated NBP control peptide (Fig. 3,A and B). Likewise, the pharmacological inhibitor, BAY 11-0782, completely blocked anti-CD3-induced EGFP expression in primary splenocytes isolated from cis-NF-κB^EGFP mice (Fig. 3 C).

To assess the ability of other transgenic mouse cells to respond to NF-κB activators, embryonic fibroblasts, PBMC, or dendritic cells were isolated from the cis-NF-κB^EGFP mice and stimulated with the classic NF-κB activators TNF-α and/or LPS. Both TNF-α and LPS induced EGFP expression in embryonic fibroblasts as determined by fluorescence microscopy (Fig. 4,A). In addition, immature dendritic cells derived from the spleen of cis-NF-κB^EGFP mice displayed a clear increase in EGFP expression following incubation with endotoxin-containing cecal bacterial lysates (24 h, 100 μg/ml; Fig. 4 B). A time course was conducted to determine the earliest time point that EGFP could be visualized using fluorescence microscopy. We determined that cis-NF-κB^EGFP transgene expression could be first observed in PBMC as early as 4 h following TNF-α stimulation (data not shown).

To assess the ability of the cis-NF-κB^EGFP transgene to respond to stimuli in vivo, cis-NF-κB^EGFP mice were treated with either anti-CD3, LPS or Ad5NIK and expression was analyzed in the lymph node cells, splenocytes, and PBMC using FACS analysis. The data indicate that the cis-NF-κB^EGFP transgene was induced in all three cell types as shown by significant increases in the number of cells expressing EGFP (Fig. 5,A). The induction of the cis-NF-κB^EGFP transgene in splenocytes was restricted to cells that expressed the early activation marker CD69 (Fig. 5 B), consistent with a NF-κB-activated cell type. Interestingly, only a subset of CD69-positive cells displayed enhanced EGFP expression, suggesting that in vivo NF-κB activation is heterogeneous in these cells.

To characterize the in vivo time course of anti-CD3 induction of the cis-NF-κB^EGFP transgene, cis-NF-κB^EGFP mice were injected with anti-CD3, and EGFP expression was assessed using FACS analysis and epifluorescence microscopy at 4, 8, 12, 24, 48, and 72 h following treatment. In PBMC and splenocytes, EGFP expression was first observed at 4 h, peaked at 8 h, and was absent by 72 h (Fig. 6,A and B). In the livers of LPS-treated cis-NF-κB^EGFP mice, this observation was essentially recapitulated in cells with morphologies consistent with hepatocytes and Kupffer cells (Fig. 6,C) with peak EGFP expression at 12 h following i.p. LPS injection. NF-κB activation in Kupffer cells was further confirmed by the colocalization of EGFP with phagocytosed 1-μm red fluorescent spheres (Fig. 6 D).

To further document cis-NF-κB^EGFP transgene expression in whole organs, in vivo, a specific EGFP imaging camera was used to visualize EGFP expression in the intestine and liver following challenge with inductive stimuli. First, cis-NF-κB^EGFP mice were injected with LPS (25 μg/g body weight) for 0, 4, and 24 h, after which mice were euthanized and the intestinal tract was dissected and analyzed for EGFP expression. Although EGFP expression was unchanged in the colon and ileum, a strong induction was observed in the duodenum and proximal jejunum (Fig. 7 A), suggesting regional specificity in NF-κB responsiveness in the intestine.

To gain more insight into the tissue distribution of EGFP-expressing cells following in vivo LPS injection, we performed confocal microscopy analysis on sections of the duodenum and proximal jejunum in cis-NF-κB^EGFP mice. Interestingly, EGFP-positive cells were mainly located in the lamina propria of the intestine and displayed morphologies that were consistent with intraepithelial lymphocytes and dendritic cells (Fig. 7, B and C). To further characterize the EGFP-positive cells, we stained these sections with a T cell marker (CD3ε) or a general monocyte marker (CD64). Confocal microscopy shows colocalization of EGFP with both CD3ε and CD64 (Fig. 7,D). Although sparse, EGFP-expressing IECs were detected in the intestine of LPS-injected cis-NF-κB^EGFP mice (Fig. 7 C). These data demonstrate the heterogeneity of the cellular NF-κB response to LPS and suggests that IECs are not the primary target of LPS stimulation under i.p. injection conditions.

Second, to selectively induce NF-κB activity, in vivo, we used an adenoviral vector encoding the NF-κB-inducing kinase (Ad5NIK). This viral vector strongly induces IKK kinase activity, IκBα serine phosphorylation, and NF-κB-dependent gene transcription (31). EGFP expression is markedly induced in a dose-dependent manner in the livers of Ad5NIK-infected cis-NF-κB^EGFP mice compared with PBS-injected or Ad5LacZ control virus-injected cis-NF-κB^EGFP mice (Fig. 8). Together, these data show that the cis-NF-κB^EGFP transgene responds to various stimuli both in vivo and in vitro by inducing EGFP expression in organs and cells in a NF-κB-dependent manner.

The transcription factor NF-κB has been extensively studied in various biological processes such as cellular proliferation, differentiation, and apoptosis. Using biochemical and pharmacological approaches, numerous studies demonstrate that induced NF-κB activity is associated with various inflammatory disorders and carcinogenesis. For example, the induction of NF-κB-dependent gene expression in various animal models of diseases such as arthritis, asthma, and inflammatory bowel and neurodegenerative diseases (13, 40, 41) directly correlates with the activity index of these diseases. Furthermore, NF-κB DNA-binding activity and nuclear translocation of some NF-κB subunits is increased during inflammatory bouts (13). However, previous methods to detect NF-κB activation fail to capture the dynamic aspect of NF-κB activity and provide limited physiological information on this complex signaling pathway.

In this study, we report the generation of the first NF-κB-responsive, enhanced GFP (EGFP)-reporter gene mouse using the single copy HPRT gene-targeting system. This transgenic mouse demonstrated increased EGFP expression in PBMC, splenic-derived dendritic cells, thymus, spleen, liver, and small intestine of anti-CD3- and LPS-, or Ad5NIK-treated transgenic mice. Using three different approaches, we demonstrated that stimuli-induced EGFP expression was specifically dependent on NF-κB activation. First, we performed ChIP assays on in vivo-stimulated cells to ascertain the recruitment/binding of RelA to the NF-κB-transgenic promoter. Using the ChIP technique, we found a strong induction of RelA loading not only on the cis-NF-κB^EGFP-transgenic promoter, but also on the promoters of the NF-κB-responsive genes RANTES, MCP-1, MIP-2, and IL-6, suggesting that increased EGFP expression correlates with induction of physiologically relevant endogenous genes. Second, we used two different strategies to pharmacologically ablate NF-κB activation and assess the effects on the cis-NF-κB^EGFP transgene expression. We used the NBP to selectively block IKK activity and the BAY 11-0782 to prevent IκBα phosphorylation. These two inhibitory approaches clearly prevented stimuli-induced EGFP expression in vivo and in vitro, demonstrating the essential role of NF-κB signaling in driving expression of the cis-NF-κB^EGFP transgene. We also induced, in vivo, the NF-κB signaling cascade using adenoviral delivery of NIK, a kinase that increases IKK activity, and, thus, NF-κB-dependent gene transcription. Using this strategy, we observed a strong induction of EGFP in the livers of infected cis-NF-κB^EGFP mice. These data clearly show the NF-κB-specific inducible nature of this transgenic mouse.

This study also demonstrates that dynamic expression of the cis-NF-κB^EGFP transgene in whole tissues and organs is easily documented using digital fluorescence photography. Using this approach, we found that LPS selectively induced EGFP expression in the duodenum and proximal jejunum, but not in the ileum and colon. This differential responsiveness of the gut to LPS may reflect the differential exposure of the intestine to endogenous bacteria (or bacterial products) leading to a differential zone of susceptibility within the gastrointestinal tract. Bacterial distribution varies along the gastrointestinal tract with low amounts of bacteria in the small bowel (10³) and, by contrast, very high levels in the distal ileum and colon (10¹²). Therefore, LPS challenge may initiate a stronger induction of the cis-NF-κB^EGFP transgene in a section of the intestine not accustomed to high bacterial loads. Derivation of the cis-NF-κB^EGFP mice in germfree conditions will help assess the differential intestinal response to endogenous bacteria.

At the cellular level, we found that in LPS-injected cis-NF-κB^EGFP mice, EGFP-expressing cells were mostly located in the lamina propria of the intestine with minimal activation of IECs. This indicates that i.p. administered LPS primarily activates intestinal lamina propria mononuclear cells (LPMNC) and subsequently IECs. Whether these events are directly mediated by LPS, or are due to soluble mediators, remain to be determined. Of note, we reported that Bacterioides vulgatus colonization of germfree rats triggers NF-κB activation in IECs, but not in the LPMNC (28). This differential effect of bacteria compared with exogenously administered LPS on the NF-κB signaling pathway in the intestine likely reflects the delivery mode since bacteria engage IECs through the lumen, whereas exogenously administered LPS accesses the intestine (and thus LPMNC) through the microvascular circulation. As proof-of-concept, these studies demonstrate that similar macroscopic analysis of EGFP expression following various in vivo treatments may be useful to identify regional and cellular differences in EGFP expression.

The participation of NF-κB in the maintenance of host homeostasis in the face of various inflammatory and stress-induced challenges are well documented in the liver and intestine (42). Recently, tissue-specific deletion of IKKβ in enterocytes enhanced susceptibility of ischemia-reperfusion-induced intestinal injury (15). Although an important observation alone, a more thorough analysis is required to elucidate the role of NF-κB signaling in specific cells, in vivo, during disease states. Cross-breeding various tissue-specific NF-κB gene-deleted mice or various inflammatory gene-engineered murine models of inflammation with the cis-NF-κB^EGFP mouse will provide critical information on the role of this pathway in various inflammatory conditions. For example, we are currently crossing the IL-10 gene-deficient mouse with the cis-NF-κB^EGFP mouse to study the kinetics and cellular localization of NF-κB activation in the context of spontaneous colitis that results in IL-10^−/− mice.

A possible limitation of the cis-NF-κB^EGFP-transgenic mouse could be the inability to detect EGFP due to weak NF-κB activation leading to minimal EGFP accumulation that is beyond the detectable threshold. Conversely, it is predicted that the ability to detect EGFP during the chronic phase of inflammation will mainly be dependent on the nature of NF-κB transcriptional activity and on the stability of EGFP. Previous reports have demonstrated increased NF-κB activity in mucosal intestinal tissue from patients with chronic inflammatory diseases (12, 13, 43); therefore, one could anticipate that NF-κB activity will remain present in the chronic phase of inflammation in this model and that EGFP expression will be continuously induced since EGFP is under the transcriptional control of NF-κB. In addition, a bile duct ligation model of liver injury showed a clear increase of EGFP-positive cells in the liver 14 days after operation (S.T.M. and D.A.B., unpublished data). Together, these data strongly suggest that EGFP is expressed and can be measured during the course of chronic inflammation.

An attractive feature of this gene-targeting approach is the generation of cis-NF-κB^EGFP-stable ES cells expressing EGFP. These cells could be used in a high-throughput random mutagenesis analysis to identify new genes involved in the regulation of NF-κB activity. Identification of cells showing loss or gain of NF-κB function could be used to generate a transgenic mouse strain to further study the functional impact of the mutation in the context of the whole organism. An additional unique feature of the cis-NF-κB^EGFP mouse is the ability to FACS distinct populations of live, EGFP-positive cells that could be further studied at the molecular/biochemical level or used as donor cells in transplantation studies.

In summary, the cis-NF-κB^EGFP mouse will allow for the first time the visualization of NF-κB transcriptional activation in the whole animal or cultured cells. We believe that this transgenic mouse will help address multiple questions regarding the state of NF-κB activation in embryonic development, inflammation, neurodegenerative disorders, and carcinogenesis and, additionally, will help to determine the effect of various in vivo treatments on this transcription factor. Moreover, pharmacokinetic and toxicology studies in these transgenic mice could be linked to in vivo NF-κB activation during diseased states to provide an accurate picture of the involvement of this transcription factor in this important cell fate pathway.

We thank Dr. Carol Albright and Brigitte Allard for expert assistance with dendritic cell isolation, Chad Torrice for help with imaging EGFP in various organs, David Detweiler for animal husbandry, and Balfour Sartor for stimulating discussion.