De Novo Synthesized RelB Mediates TNF-Induced Up-Regulation of the Human Polymeric Ig Receptor¹

In an adult human, the mucosal surface area amounts to ∼400 m², which is constantly exposed to potentially harmful agents from food, drink, and inhaled air. The mucosae are generally covered by a single layer of epithelial cells and are thus quite vulnerable. However, an efficient immunologic surface protection is provided by local secretory Ig (SIg)⁴ production in cooperation with various innate defense mechanisms (1). SIgs result from the joint efforts of mucosal J chain-expressing plasma cells that produce polymeric IgA (pIgA; mainly as dimers) and pentameric IgM (2), and secretory epithelial cells that express the polymeric Ig receptor (pIgR) (3, 4). This receptor is also known as membrane secretory component (SC) and binds the pIgs at the basolateral face of the epithelial cells. The endocytosed pIgR-pIg complexes as well as unoccupied receptor are transcytosed through the epithelial cell and exocytosed at the apical surface after specific cleavage of pIgR (3, 4). Thus, SIgs that contain the extracellular portion of pIgR bound to pIg (bound SC) and similarly cleaved free SC are released to the secretions. The essential role of pIgR for the formation of SIgs as well as for maintenance of the mucosal barrier integrity has recently been demonstrated through studies of pIgR knockout mice (5, 6).

Due to cleavage of pIgR at the apical epithelial surface, this specific transport pathway requires a relatively high level of constitutive expression for continuous generation of SIgs. In addition, pIgR expression and thereby the pIg transport capacity to form protective SIgs can be modulated by several factors, such as the nutritional status of the host, bacterial products, hormones, and cytokines (reviewed in Refs.4 and 7). In situ studies have demonstrated up-regulated levels of pIgR in several chronic inflammatory disorders, probably due to the locally produced cytokines (reviewed in Refs.4 and 7).

The proinflammatory cytokine TNF is a key mediator of host protection against infection (8, 9) and is known to up-regulate the transcription of pIgR by a delayed and protein synthesis-dependent mechanism (10). The best-characterized mediator of TNF action is the NF-κB family of latent, rapidly activated transcription factors (TFs) (11, 12). Up-regulation of pIgR mRNA by TNF is sensitive to NF-κB inhibitors (13). We have recently demonstrated that TNF-mediated up-regulation of pIgR in the human intestinal adenocarcinoma cell line HT-29.m3 required cooperation between DNA elements located in both an enhancer region of intron 1 and the proximal promoter (14). A region of the human pIgR promoter between −177 and −83, which contains two putative IFN-stimulated response elements (ISREs) important for IFN-γ induction (15), was also important for TNF responsiveness (14). An NF-κB element within the intronic enhancer likewise mediated TNF responsiveness, and stimulation with TNF induced binding of NF-κB p65- and NF-κB p50-containing complexes to this element, as revealed in vitro by EMSA (14).

We and others have reported that NF-κB can also bind to a target site at position −450 in the human pIgR promoter (13, 14), but deletion or mutation of this site did not reduce TNF responsiveness of the pIgR-derived reporter constructs (14). An ISRE in exon 1 has furthermore been shown by deletional and mutational analysis to contribute positively to TNF responsiveness (14, 15), particularly when the intronic NF-κB site is missing or mutated (14). It was recently reported that NF-κB activation also plays a role in pIgR induction by IFN-γ and IL-4, cytokines that are known to activate STATs rather than Rel/NF-κB family members (16). Surprisingly, the promoter NF-κB site was more important for IFN-γ and IL-4 induction of pIgR than TNF induction (16). In agreement with this study, we have recently characterized an IL-4-responsive enhancer that partially overlaps the TNF-responsive region in intron 1 and found that a mutation within the NF-κB site also reduced IL-4 responsiveness (17).

In this study we have analyzed the role of NF-κB in the TNF-mediated induction of pIgR. We show that both the core sequence and the context of the intronic NF-κB site are important for proper function. Furthermore, we identify RelB as a slowly activated factor that binds to this site and show that RelB overexpression is sufficient to activate pIgR-derived reporter genes. Finally, we show by small interfering RNA (siRNA) that RelB is required for optimal TNF-mediated induction of pIgR.

The human colonic adenocarcinoma cell line HT-29.m3, previously selected for high expression of pIgR (18), was maintained in RPMI 1640 medium supplemented with 50 μg/ml gentamicin, 2 mM l-glutamine, and 10% FCS. Transient transfections were performed with FuGene 6 reagent (Roche, Indianapolis, IN) as previously described (19). Transfected cells were either left untreated or stimulated with 10 ng/ml recombinant human TNF for 12 h when indicated. The luciferase activity of both the reporter gene (firefly luciferase) and the internal control plasmid pRL-PGK (Renilla luciferase), was measured in a luminometer (Victor (Wallace, Turku, Finland) or Luminoskan Ascent (Thermo Labsystems, Helsinki, Finland)) with the Dual Luciferase Reporter Assay System (Promega, Madison, WI). The data presented in Figs. 2, 3,B, 6, and 7 show the mean ± 1 SEM of three or more independent experiments.

Total RNA was isolated from untreated and treated HT-29.m3 cells with the RNAwiz reagent (Ambion, Austin, TX) according to the manufacturer’s protocol. One microgram of RNA, primed with oligo(dT) and reverse transcribed with Superscript II or III (Invitrogen, Leek, The Netherlands) was used for a 20-μl cDNA reaction. Gene-specific primers were designed with Primer3 software (20) as described previously (19). Specific cDNA was quantified by real-time PCR with the Light Cycler (Roche) and LC-Fast Start DNA Master SYBR Green I reagents (Roche) or with the QuantiTect SYBR Green PCR kit (Qiagen, Valencia, CA), according to the manufacturer’s protocol. PCR primer sequences are given in Table I. Quantification of specific mRNA levels was performed as previously described (19) or using dilution curves of exogenous standards. Levels of pIgR and RelB mRNA were normalized to GAPDH before fold induction was calculated. The PCR products were analyzed by melting curve analysis (LightCycler) and visualized on an agarose gel to confirm the expected size of the amplified cDNA.

Nucleotide numbering is given relative to the transcription start site (21). The plasmids pRL-PGK, pSC1, pSC2, pSC11, pSC15, pSC27, pSC29, and pSC53 have been described previously (14, 19). The 5′ and 3′ deletions introduced in plasmids denoted pSC31–pSC44, and pSC77–pSC80 were constructed with a PCR that introduced restriction sites for either KpnI (for 5′ deletions) or BglII (for 3′ deletions), and subsequent restriction enzyme digestion and ligation into KpnI- and BglII-digested pSC27 vector. The mutations in pSC62, pSC63, and pSC67 were introduced into the pSC1 vector either with the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) or by conventional PCR-based splicing by overlap extension technique. All mutations as well as the integrity of the vector-insert boundary of all deletions were confirmed by DNA sequencing (Medigenomix, Martinsried, Germany).

The expression plasmids for murine NF-κB p65 (RelA), RelB, and p50 were provided by Dr. G. Natoli (Institute for Research in Biomedicine, Bellinzona, Switzerland). The pCDNA3.1 vector (Invitrogen Life Technologies, Carlsbad, CA) was used as an empty vector control in coexpression experiments. pRelB-siRNA and pRelB-sense were constructed by ligating targeting and control oligonucleotides (Table I) into pSIREN-RetroQ (BD Biosciences, Palo Alto, CA).

Preparation of nuclear extracts from HT-29.m3 cells was performed essentially as previously described (22) with certain modifications detailed in an earlier report (19). Approximately 5 μg of nuclear proteins were incubated with ³²P end-labeled, double-stranded oligonucleotide probe (Table I; 0.25 pmol/reaction). The EMSA reactions were performed in buffer containing 1 mM EDTA, 50 or 70 mM KCl, 1 mM DTT, 0.1 μg/μl ssDNA (or dI/dC), 0.05% Nonidet P-40, 12.5 mM Tris (pH 7.9), and 6% glycerol for 30 min at room temperature. Bound and free probes were separated by electrophoresis in a 4.5% (w/v) polyacrylamide gel (0.25× Tris/borate/EDTA) at 150 V for ∼1.5 h at room temperature, dried, and visualized on x-ray film overnight. Cold competitors were added in a 100-fold excess before addition of labeled probe when indicated. For supershift experiments, 2 μl of specific Ab, with or without 2 μl of blocking peptide, was added to the reaction mixture and incubated at room temperature for 20 min. The labeled probe was then added, and the reaction was incubated for another 30 min before electrophoresis. Abs specific for NF-κB p65 (RelA), c-Rel or NF-κB p50 were purchased from Geneka Biotechnology (Montreal, Canada), whereas Abs specific for RelB (C-19), NF-κB p52 (K-27), or NF-κB p65 (C-20) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). An additional Ab specific for NF-κB p50 was purchased from Upstate (Lake Placid, NY).

Analysis of DNA sequences was performed with the Genetics Computer Groups package (Genetics Computer Group, Madison, WI) and MatInspector (http://genomatix.de/cgi-bin/matinspector/matinspector.pl) (23) or TFSEARCH (24) software (http://molsun1.cbrc.aist.go.jp/research/db/TFSEARCH.html).

To investigate the kinetics of TNF-mediated up-regulation of pIgR transcription in HT-29.m3 cells, we isolated RNA from cells stimulated with TNF for various time periods and performed quantitative real-time RT-PCR. The level of mRNA for the housekeeping gene GAPDH was analyzed in each sample as an internal control for mRNA integrity and yield. Significant up-regulation of pIgR mRNA was shown after 4–8 h of TNF stimulation (Fig. 1,A; increased 2- to 6-fold), and the steady state level of pIgR mRNA peaked after 8–12 h and persisted for the longest period investigated (48 h; Fig. 1,A). The magnitude of mRNA up-regulation by TNF varied somewhat between experiments, but the delayed response (after 4 h) and the timing of the steady state plateau were reproducible. Importantly, the results were in accordance with our previous report (10). The kinetics of TNF-mediated up-regulation of the pIgR-derived reporter gene pSC1 was investigated by transient transfection in HT-29.m3 cells and stimulation with TNF for various time periods as described in Materials and Methods (Fig. 1,B). The results were in accordance with the kinetics for up-regulation of the endogenous pIgR gene as analyzed by RT-PCR (Fig. 1, compare A with B). This result substantiated that our reporter gene represented a valid model system to study the delayed TNF-mediated up-regulation of pIgR expression.

We have previously identified a 1.3-kb TNF-responsive region in intron 1 (from position +3528 to +4872 relative to transcription start) of the pIgR gene, which also conferred responsiveness when located upstream of the transcription start site (14). Within this region, we identified an NF-κB site that was required for full TNF responsiveness (14). To identify the minimal functional region of the intronic enhancer, we made sequential deletions from the 5′ or 3′ end of the 1.3-kb intronic fragment and subcloned these upstream of the weakly TNF-responsive pIgR promoter. These new reporter constructs were analyzed in transient transfection as described above. After 12 h of TNF treatment, the minimal pIgR promoter alone (pSC27) or with the 1.3-kb intronic fragment subcloned upstream (pSC29) showed ∼2.0- and 3.7-fold induction, respectively. Deletion of up to 686 bp from the 5′ end did not result in altered TNF responsiveness, but further deletions reduced TNF responsiveness to a similar level as the promoter alone (Fig. 2). From the 3′ end of the intronic enhancer, deletions of up to 455 bp could be tolerated without a loss of TNF responsiveness, whereas a deletion of 516 bp or more reduced TNF responsiveness to that of the promoter alone (Fig. 2).

The intronic NF-κB site mediated TNF responsiveness also when the intronic fragment was subcloned upstream of the pIgR promoter (14) (Fig. 2). In contrast, an NF-κB site located at −450 in the human pIgR promoter was found to be dispensable for TNF responsiveness, although it has been demonstrated to bind inducible p50- and p65 (RelA)-containing factors in EMSA experiments (14). To test whether the disparate behaviors of these two NF-κB sites were due to core sequence differences and/or differences in flanking sequences, we introduced substitution mutations in the pSC1 reporter gene to convert each of these distinct NF-κB sites into the other. First, we altered the intronic NF-κB site and the immediate flanking sequence into an NF-κB element equivalent to that found in the promoter (pSC63; Fig. 3,A). When analyzed in transient transfection assays, as described above, this conversion resulted in a loss of TNF responsiveness comparable to that of abolishing the intronic NF-κB site altogether (pSC53; Fig. 3,B). However, a 3-bp mutation flanking the intronic NF-κB site on the 3′ side (pSC62) did not reduce TNF responsiveness (Fig. 3,B). Next, we made a reporter construct in which the core and flanking sequences of the promoter NF-κB site were converted to those of the intronic NF-κB site (Fig. 3,A). This mutation was introduced on the background of pSC63 that had the intronic NF-κB site mutated into a promoter-like NF-κB site, and this new reporter construct (pSC67) was tested in transient transfections as described above. The presence of the intronic NF-κB element in the promoter at position −450 did not restore TNF responsiveness (Fig. 3,B). Thus, sequences surrounding the intronic NF-κB site, but more distant than 5 bp from the actual binding site (Fig. 3 A), were required for the function of this site.

We have previously demonstrated TNF-induced binding of p65- and p50-containing nuclear complexes to both the intronic and the promoter NF-κB sites (14). To test for the induction of other members of the NF-κB family and their ability to bind the intronic NF-κB site, we performed in vitro EMSA experiments with nuclear extracts isolated from unstimulated and TNF-stimulated HT-29.m3 cells and performed supershift analysis with specific Abs against all five NF-κB family members. Abs specific for c-Rel or p52 did not shift any of the TNF-induced complexes, whereas Abs specific for p65 or p50 (Geneka Biotechnology) each shifted different complexes (Fig. 4,A and data not shown). A different p50-specific Ab (Upstate) also partially supershifted the p65-containing complex, suggesting the presence of the classical p65-p50 heterodimer (Fig. 4,B). Furthermore, we found that RelB-specific Ab diminished a weak complex induced at later time points (Fig. 4,A, lane 6, and data not shown) that was not present after 30 min of TNF stimulation (data not shown). Addition of a blocking peptide specific for the RelB antiserum restored this complex (Fig. 4,A, lane 7). To confirm the presence of RelB in this faint complex, which migrated only slightly faster than the stronger p65-containing complex, we performed an EMSA with the p65 and RelB supershifts conducted in adjacent lanes and with an additional reaction containing both Abs (p65 and RelB; Fig. 4,C). The results confirmed the presence of distinct p65- and RelB-containing complexes (Fig. 4,C, lanes 1 and 2). The p65-specific Ab shifted the slower migrating complex, whereas RelB-specific Ab shifted the faster migrating complex (Fig. 4,C). Furthermore, addition of Abs specific for both p65 and RelB resulted in the disappearance of both these complexes and the appearance of two new supershifted complexes (Fig. 4 C, lane 3).

To determine whether the appearance of TNF-induced RelB-containing EMSA complexes was due to up-regulation of RelB at the transcriptional level, we performed RT-PCR with RNA isolated after different times of TNF stimulation of HT-29 cells (Fig. 4,D). Real-time RT-PCR demonstrated that RelB mRNA was rapidly up-regulated, reaching a peak of induction after 3–6 h. The RelB response diminished after 6 h, but remained above basal levels 12–18 h after TNF induction. Furthermore, a second wave of RelB mRNA increase appeared after 24 h, consistent with the sustained RelB DNA binding activity 24 h after TNF treatment (Fig. 4 D).

To test whether the presence of active NF-κB was sufficient to induce transcription from the pIgR-derived reporter gene, we cotransfected the full-length pSC1 reporter construct with expression vectors for either NF-κB p65, RelB, p50, p65 and p50, or RelB and p50. The level of reporter gene expression was compared with that of pSC1 cotransfected with empty expression vector (pCDNA3.1). Preliminary experiments demonstrated a dose-dependent induction of pSC1 reporter gene expression when cotransfected with the p50, p65, or RelB expression vector for 24 h (data not shown). This NF-κB-dependent induction reached a plateau with 0.2 μg of p50 or p65 expression vector and 0.05 μg of RelB expression vector (data not shown). We therefore used these amounts of expression vectors to cotransfect with pSC1 and assayed luciferase activity at different time points after transfection (Fig. 5). A reproducible, although weak, induction was seen 24 h after cotransfection with p65, but this induction was abolished after 36–48 h (Fig. 5). In contrast, cotransfection of p50 displayed more delayed kinetics of induction, which continued to increase up to 54 h after transfection (Fig. 5). Cotransfection of both p65 and p50 resulted in a high level of induction at the earliest time point (24 h), whereas the later effect (36–48 h) was comparable to that of p50 alone (Fig. 5). RelB was by far the most potent NF-κB family member in activating pIgR transcription, resulting in ∼5-fold induction 24 h after transfection, which increased to ∼10-fold induction 54 h after transfection. The combination of p50 and RelB was even more powerful at the 24 h point, but by 42 h the luciferase activity of pSC1 cotransfected with p50 and RelB was similar to that seen with RelB cotransfection alone.

It became clear that RelB was a strong activator of pIgR transcription. To determine through which sequence elements in the reporter gene this transcriptional activator was working, we cotransfected RelB expression vector with or without p50 with several different pIgR-derived reporter genes (Fig. 6). Expression of RelB (or RelB and p50) resulted in a 4- to 5-fold activation of the full-length reporter gene (pSC1), but only ∼2.5-fold induction of the reporter gene lacking intron 1 (pSC2) 24 h after transfection (Fig. 6). A similar dependence on intron 1 for RelB responsiveness was seen when most of the promoter (from −2.7 to −0.54 kb) was deleted (Fig. 6; pSC15 and pSC11). Finally, deleting the NF-κB site at position −450 did not affect induction by RelB (or RelB and p50; Fig. 6, compare pSC11 and pSC27). These results agreed with our previous findings regarding the sequence requirements in the pIgR gene for TNF responsiveness (14).

To determine whether endogenous RelB protein was required for optimal TNF induction of pIgR, we targeted it for removal by transfecting HT-29 cells with an expression vector encoding an siRNA specific for RelB mRNA. TNF was added 48 h after transfection to allow siRNA for RelB to accumulate before stimulation. Cotransfection of empty expression vector (pCDNA3.1) or expression vector encoding the sense strand of the RelB siRNA (RelB-sense), resulted in similar levels of TNF induction of luciferase activity from the pSC1 reporter plasmid (4.2- and 3.9-fold, respectively; Fig. 7). However, we found that expressing siRNA specific for RelB reduced TNF responsiveness of pSC1 from 4.2- to 2.3-fold (Fig. 7).

In this study we investigated the role of NF-κB in TNF-mediated up-regulation of human pIgR expression. Proper function of an NF-κB site located in an intronic cytokine-responsive enhancer required both its specific core sequence and surrounding sequences. Overexpression of NF-κB p50, p65, or RelB was sufficient to induce the expression of a pIgR-derived reporter construct, providing support for the central role of NF-κB in TNF-mediated up-regulation of pIgR. EMSA experiments with nuclear extracts from TNF-treated HT-29 cells revealed rapid induction of NF-κB p65 DNA binding and a more delayed induction of RelB, suggesting that these factors might have different roles in the initial activation and the sustained expression of pIgR. Finally, we demonstrated by the use of a vector-based siRNA strategy that RelB was required for maximal TNF induction of the human pIgR-derived reporter gene.

We have previously shown that an NF-κB site in the first intron of the human pIgR gene is required for its TNF responsiveness, and that this intronic element needs to cooperate with promoter proximal elements for full induction to occur (14). To identify the required sequence elements for TNF responsiveness within the intronic enhancer region, we performed deletion analysis of the 1.3-kb intronic fragment fused upstream of the minimal TNF-responsive pIgR promoter. We found that deletions to within 181 bp of the NF-κB site from the 5′ side did not cause a loss of function, but the next deletion, which still left 82 bp upstream of the NF-κB site, no longer promoted TNF responsiveness. This result suggested that another factor(s) binding within 180 bp upstream of the intronic NF-κB site is required for full TNF responsiveness. Conversely, we found that deletions up to within 11 bp of the NF-κB site from the 3′ end were tolerated. Further deletions that removed the intronic NF-κB site resulted in a significant loss of TNF responsiveness. This was in accordance with our previous observation that a specific mutation of this intronic NF-κB site reduced TNF responsiveness to the level of the pIgR promoter alone (14). Thus, the intronic TNF-responsive region is contained within a 204-bp DNA fragment that is largely overlapping the IL-4-responsive enhancer (17). As this region contains several TF binding sites, it will be interesting to determine whether up-regulation of pIgR by these two cytokines shares TFs other than NF-κB, as was shown by Ackermann et al. (16).

In our previous report on TNF induction of the pIgR gene, we demonstrated binding of the same nuclear factors to the intronic NF-κB site that were able to bind a promoter NF-κB site (at −450), although the latter showed a relative increase in affinity toward p50 homodimers compared with p65-containing complexes (14). In reporter gene assays, the intronic NF-κB site was required, whereas the promoter NF-κB site was dispensable for TNF responsiveness of the human pIgR gene (14). To determine whether the disparate behavior of these two NF-κB sites in reporter gene assays was due to different core sequences and immediate flanking bases or was caused by the larger context of these sites, we altered the core and flanking sequences of each site into that of the other. We found that changing the intronic NF-κB site to the sequence of the promoter site was as deleterious as abolishing the NF-κB site altogether. Furthermore, replacement of the promoter NF-κB element with the intronic element and flanking sequences did not substitute for the mutated site in intron 1 to restore TNF responsiveness. These results strongly implied that the nature of the complex forming at these two sites in vivo differs. Thus, the proper function of the intronic NF-κB element appeared to depend on both the nature of the element itself (the core and immediate flanking sequences) and the specific sequences in the surrounding DNA. It is tempting to speculate that this might reflect a requirement for activation and binding by a specific NF-κB dimer in cooperation with TFs that bind to neighboring DNA elements in the intronic enhancer. Such target gene specificity of the NF-κB family members has been suggested by studies of NF-κB knockout mice (reviewed in Refs. (11, 12 , and 25) as well as studies of specific target genes (26, 27). Although the mechanisms underlying this target gene specificity of the NF-κB family members remain poorly understood, they could in part be explained by differential binding affinities and core sequence preferences displayed by the different homo- and heterodimers (11, 28).

We have previously demonstrated that TNF treatment of HT-29 cells activates the formation of p65- and p50-containing nuclear complexes (14). In this study we demonstrated the presence of the classical p65-p50 heterodimer, not previously detected, after only 30 min of TNF treatment. In addition, supershift experiments with RelB-specific Abs demonstrated a delayed induction of this NF-κB family member in nuclear extracts from HT-29 cells. The RelB-positive complex was also affected by Ab specific for p50, suggesting the presence of a RelB-p50 heterodimer, in accordance with the previously reported dimerization pattern of RelB (reviewed in Refs.29 and 30). Although the relative intensity of the RelB-containing complex was low compared with that of the p65- and p50-containing complexes, it remains unknown which of these factors actually binds to the endogenous pIgR gene and thus regulates the transcription of the gene in vivo. It has been suggested that RelB-mediated activation in some cell types might be regulated by de novo protein synthesis as well as by post-translational modifications (Ref.31 and references therein). In agreement with this idea, we found that RelB mRNA was greatly increased after 3 h of TNF exposure of HT-29 cells. Furthermore, the kinetics of the appearance of the RelB-containing complex in HT-29.m3 cells was in better accordance with the slow kinetics of pIgR up-regulation by TNF than the immediately induced p65-containing complexes. Thus, RelB could be a de novo synthesized intermediate between TNF activation and pIgR transcription.

In cotransfection experiments, we found that p65 resulted in a relatively early, but weak and transient, effect on our pIgR-derived reporter genes, whereas p50, and in particular RelB, resulted in sustained transcriptional activation. These results suggested that p65 and RelB have different roles in the activation of pIgR transcription. Direct evidence for a requirement of RelB in TNF-mediated transcriptional activation of pIgR was obtained from experiments in which endogenous RelB mRNA was targeted for removal by cotransfection of siRNA expression vector with the pIgR-derived reporter gene. Inhibiting RelB mRNA significantly reduced the TNF responsiveness of the full-length reporter gene.

Engagement of TNF receptor initiates a signaling cascade that rapidly leads to the release of NF-κB dimers from their cytoplasmic inhibitors (IκBs), followed by translocation to the nucleus (11). Contrasting with this classical paradigm, recent experiments have demonstrated a continuous shuttling of NF-κB between the cytoplasm and the nucleus, and the presence of a nuclear pool of NF-κB in some unstimulated cells (reviewed in Ref.12). Furthermore, RelB appears to be restricted to the cytoplasm or inhibited in the nucleus exclusively by binding to p100 (NF-κB2, the precursor of p52) (32, 33). However, upon new synthesis of RelB, it may dimerize with available p50 and migrate directly into the nucleus because it has low affinity to IκB. Although TNF might also activate p100 expression, an excess of RelB (to p100) may accumulate over time (34). The role of RelB as an activator of pIgR transcription differentiates the regulation of this gene with that of classical proinflammatory genes expressed upon TNF stimulation of epithelial cells such as IL-8 (35). However, activation of p65/p50 is presumably also critical for pIgR activation because it is required for activation of the RelB gene itself (13, 31). It is tempting to speculate that this dimer also binds to target elements in the pIgR gene at early time points after TNF stimulation and is exchanged for RelB-containing dimers at a later time point to achieve a sustained activation.

The role of the pIgR is to transport pIgA Abs to the mucosal surface where they perform several noninflammatory defense functions (36, 37, 38, 39). TNF stimulation of epithelial cells activates a proinflammatory gene repertoire, exemplified by IL-8 (35), as well as pIgR transcription (7). Detailed understanding of the different mechanisms involved in regulation of proinflammatory genes and the noninflammatory pIgR gene may therefore reveal strategies that would inhibit inflammation without suppressing pIgR expression and the formation of secretory Abs.

We thank the technical staff at Laboratory for Immunohistochemistry and Immunopathology for excellent laboratory assistance. The expression plasmids for murine NF-κB p65 (RelA), RelB, and p50 were kindly provided by Dr. G. Natoli (Institute for Research in Biomedicine, Bellinzona, Switzerland).