Phosphorylation of RelA/p65 on Serine 536 Defines an IκBα-independent NF-κB Pathway*

The association of the NF-κB p65/p50 dimer with IκBα plays a pivotal role in regulating its nuclear translocation and gene transcription. In addition, serine phosphorylation at various sites of the p65 subunit has been shown to be important in initiating transcription. Here we demonstrate that the regulation of nuclear translocation of p65 phosphorylated at serine 536 is not dependent on IκBα. Stimulation of either Jurkat or normal human T cells resulted in the nuclear translocation of phospho-p65 (Ser536). In addition, the phospho-p65 (Ser536) was not associated with either IκBα or p50, and the nuclear translocation of phospho-p65 (Ser536), but not total p65, was unaffected by the proteosome inhibitor MG-132, which blocks IκB protein degradation and prevents p65/p50 dimer nuclear translocation. Accordingly, the co-expression of a dominant negative mutant of IκBα blocked the transcriptional activity mediated by wild type but not the dominant positive p65 mutant (S536D). Furthermore, the transfection of the S536D form of p65 led to the induction of interleukin-8 transcription following stimulation, whereas the S536A form, which cannot be phosphorylated at this site, did not. Together, the findings suggest that p65 phosphorylated on serine 536 is not associated with or regulated by IκBα, that it has a distinct set of target genes, and that it may represent a noncanonical NF-κB pathway that is independent of IκBα regulation.

The association of the NF-B p65/p50 dimer with IB␣ plays a pivotal role in regulating its nuclear translocation and gene transcription. In addition, serine phosphorylation at various sites of the p65 subunit has been shown to be important in initiating transcription. Here we demonstrate that the regulation of nuclear translocation of p65 phosphorylated at serine 536 is not dependent on IB␣. Stimulation of either Jurkat or normal human T cells resulted in the nuclear translocation of phospho-p65 (Ser 536 ). In addition, the phospho-p65 (Ser 536 ) was not associated with either IB␣ or p50, and the nuclear translocation of phospho-p65 (Ser 536 ), but not total p65, was unaffected by the proteosome inhibitor MG-132, which blocks IB protein degradation and prevents p65/p50 dimer nuclear translocation. Accordingly, the co-expression of a dominant negative mutant of IB␣ blocked the transcriptional activity mediated by wild type but not the dominant positive p65 mutant (S536D). Furthermore, the transfection of the S536D form of p65 led to the induction of interleukin-8 transcription following stimulation, whereas the S536A form, which cannot be phosphorylated at this site, did not. Together, the findings suggest that p65 phosphorylated on serine 536 is not associated with or regulated by IB␣, that it has a distinct set of target genes, and that it may represent a noncanonical NF-B pathway that is independent of IB␣ regulation.
The NF-B signaling pathway responds rapidly to a wide range of stimuli (1). Activation leads to the translocation of the transcription factors from the cytoplasm to the nucleus. The NF-B transcription factor consists of two subunits of either homo-or heterodimers of RelA/ p65, c-Rel, and p50. The complexes are held in the cytoplasm and prevented from activating transcription by a class of proteins referred to as inhibitors of NF-B or IB proteins. Upon stimulation, the IB proteins are phosphorylated by one of a number of IB kinases (IKK-␣, -␤, and -␥), ubiquitinylated, and degraded, which thereby allows the NF-B complex to translocate into the nucleus (2). However, recent findings have demonstrated the shuttling of the NF-B complex in and out of the nucleus in the absence of stimulation (3,4). In addition to nuclear translocation of the NF-B complex, several studies have shown that the NF-B proteins are modified post-translationally, and those changes influence transcriptional activity. Examples of activation-induced posttranslational modifications include the acetylation of p65 to facilitate the retention of the NF-B complex in the nucleus (5,6). In addition, the S-nitrosylation of cysteine 62 of p50 has been shown to affect the NF-B binding to DNA (7,8).
We have previously described the phosphorylation of various Rel proteins following the stimulation of T cells, and the phosphorylation of p50 increased the DNA binding capacity (9). Several studies have demonstrated the phosphorylation of p65 in response to various stimuli (10). Serine 276 of p65 is phosphorylated by protein kinase A during IB degradation and is necessary for the recruitment of CREB-binding protein/p300 to p65 for active transcription (11)(12)(13). The mitogen-and stress-activated protein kinase 1 also phosphorylates serine 276 of nuclear p65 following tumor necrosis factor (TNF) 2 stimulation (14). In addition, the recruitment of CREB-binding protein and RNA polymerase II to the IL-6 promoter requires the phosphorylation of p65 at serine 311 by protein kinase C- (15). The phosphorylation of serine 529 by casein kinase II (16,17) and serine 536 by IB kinases following TNF (18 -20) and lipopolysaccharide (21) stimulation increased transcriptional activity. Serine 276 is within the Rel homology domain, which is involved in dimerization, whereas the serines 529 and 536 are in the C-terminal transactivation domain. Moreover, threonine phosphorylation of p65 at residue 254 has been shown to be required for Pin1 association and the increase in p65 stability (22). It is not known if the phosphorylations are mutually exclusive or all three are required for active transcription.
The role of phosphorylation of the serine residues 529 and 536 in the second transactivational domain has been in question with the recent data from Okazaki et al. (23). Site-directed mutants of the serines either individually or together did not result in a loss of transcriptional activity, which suggested that the phosphorylation of the serine residues was not necessary for p65 transactivational activity. The data herein describes the nuclear translocation of p65 phosphorylated on serine 536 following phorbol 12-myristate 13-acetate (PMA) and ionomycin (PI) or TNF stimulation in Jurkat cells. The phosphorylation of p65 did not increase following activation in either Jurkat or primary T cells. Moreover, the phospho-p65 was not associated with either IB␣ or p50, and the proteosome inhibitor MG-132 reduced nuclear translocation of total p65 but not phospho-p65. The overexpression of the dominant negative IB␣ (DN-IB␣) mutant suppressed p65 transcriptional activity; however, the transcriptional activity of the phosphomimetic form of serine 536 (S536D) of p65 was not affected. The findings suggested that p65 phosphorylated on serine 536 translocated to the nucleus following activation, and this nuclear translocation is not regulated by IB␣. Furthermore, the data suggest that at least some genes are transcribed selectively by an NF-B complex containing a form of p65 phosphorylated on serine 536. and Dulbecco's modified Eagle's medium, respectively, supplemented with 10% fetal bovine serum, 2 mM glutamine, and 100 units of penicillin, 100 g of streptomycin (QBS). Human peripheral blood T cells were collected from healthy donors who provided informed consent and enriched as previously described (24). Stock solutions of PMA (Sigma) and ionomycin (Calbiochem) were prepared in ethanol and used at 10 ng/ml and 1 g/ml, respectively. MG-132 was purchased from Biomol (Plymouth Meeting, PA) and prepared in Me 2 SO (Sigma).
Plasmids-The p65 cDNA was subcloned from pCMV-p65 (a generous gift from Dr. H. A. Young) into pEF6/V5-His (Invitrogen), and the following primers (Sigma) were used with the QuikChange II site-directed mutagenesis kit (Stratagene, La Jolla CA) to generate the sitedirected mutant: S536A forward (GATGAAGACTTCTCCGCCATT-GCGGACATGGAC) and reverse (GTCCATGTCCGCAATGGCGG-AGAAGTCTTCATC); S536D forward (GGAGATGAAGACTTCTC-CGACATTGCGGACATGGACTTC) and reverse (GAAGTCCATG-TCCGCAATGTCGGAGAAGTCTTCATCTCC). Site-directed mutagenesis was confirmed by DNA sequencing (Lofstrand Labs Ltd., Gaithersburg, MD). The wild type and mutant p65 constructs were transiently transfected into relA Ϫ/Ϫ knockout mouse embryonic fibroblasts (25) (a generous gift from Dr. David Baltimore), and lysates of the transfectants were processed for immunoblot analysis. The wild type pCMV-IB␣ and dominant negative pCMV-IB␣ M expression vectors were purchased from BD Biosciences.
Immunoprecipitation and Immunoblotting-Whole cell lysates were prepared with radioimmune precipitation buffer (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) supplemented with the complete protease inhibitor mixture (Roche Applied Science), 1 M sodium fluoride, 1 M sodium orthovanadate. Cytoplasmic and nuclear extracts were prepared as described (26) with the modification of adding the aforementioned supplements. Total protein concentration in the samples was determined with the BCA protein assay (Pierce). 5-10 g of sample were loaded into each well, separated by gel electrophoresis, transferred to polyvinylidene difluoride, and probed with the antibodies indicated in the figure legends. The antibodies against phospho-p65 serine 536 (catalog number 3031) and phospho-IB␣ serine 32 were purchased from Cell Signal (Beverly, MA) and used at the indicated concentrations as described by the manufacturer. The rabbit antibodies against p65 (40 ng/ml; C-20), Oct-1 (200 ng/ml; C-21), and IB␣ (200 ng/ml; C-21) were from Santa Cruz Biotechnology (Santa Cruz, CA). The concentrations of the monoclonal antibodies against V5 (Invitrogen) and ␤-tubulin (Sigma) were used as indicated by the manufacturer. The rabbit antisera against p50 (1157) and c-Rel (265) were a generous gift from Dr. Nancy Rice and diluted 10,000-fold. The antibodies were detected by chemifluorescence and scanned with the Typhoon scanner (Amersham Biosciences). Quantitation of the band intensity was done with Image-QuaNT software (Amersham Biosciences).
For immunoprecipitations, samples were precleared with 2 g of normal goat immunoglobulin and 20 l of Protein A/G beads (Santa Cruz Biotechnology) for 1 h, and the precleared samples were incubated with the following antibodies from Santa Cruz Biotechnology at the concentrations indicated in the figure legends: goat normal immunoglobulin, goat anti-p65 (C-20-g), goat anti-IB␣ (C-20-g), and goat anti-p50 (C-20-g). An aliquot of the precleared sample (one-twentieth of the volume) before the immunoprecipitation reaction was designated as the input sample. After incubating for 2 h, 20 l of Protein A/G beads were added for an additional 1 h. The beads were washed three times with lysing buffer and then eluted with sample buffer. For the immunodepletion experiments, the unbound samples in the supernatant from the bound beads were subsequently immunoprecipitated with the antibod-ies indicated in the figure legends. In some experiments, this process was repeated to immunodeplete multiple proteins from a single sample.
NF-B Reporter Assay-The NF-B reporter assay was performed as previously described (27) with the following modifications. Cells were transfected with the reporter plasmids NF-B/␤-galactosidase and pGL3-luciferase (Promega, Madison, WI) and the transfection reagent FuGENE 6 (Roche Applied Science). As indicated in the figure legends, expression plasmids of p65, p65 mutants, IB␣, and IB␣ mutant were co-transfected with the reporter plasmids. After 48 h of incubation, ␤-galactosidase and luciferase activity were assessed with the Beta-glo and Bright-glo (Promega, Madison, WI) assay according to the instructions of the manufacturer. NF-B transcriptional activity, as detected as ␤-galactosidase activity, was normalized with the luciferase activity. The relative light units and the error bars represent the average and S.D. of triplicate samples.
Chromatin Immunoprecipitation-Chromatin immunoprecipitation was conducted as described with the following modifications (28). Jurkat T cells, 2 ϫ 10 7 cells, were either untreated or treated with PI for the indicated time periods. The cells were fixed with formaldehyde and washed twice with PBS. The cell pellet was lysed and sonicated twice with 10 pulses on ice. After centrifugation, 1% of the extract was aliquoted and used for the total input control. The remaining extracts were precleared with protein A-agarose containing salmon sheared DNA (Upstate, Waltham, MA). The precleared samples were evenly aliquoted into four tubes containing agarose-conjugated normal rabbit Ig, anti-acetylated H4 chromatin immunoprecipitation (ChIP) grade (Upstate), agarose-conjugated anti-p50 (Santa Cruz Biotechnology), and agarose-conjugated anti-p65 (Santa Cruz) (designated total p65) and incubated overnight at 4°C with rotation. Protein A-agarose salmon sheared DNA was added to the anti-acetylated H4 and anti-RNA polymerase II samples and incubated for 1 h at 4°C with rotation. The unbound fraction of the anti-p50 immunoprecipitation reaction was collected and incubated with agarose-conjugated anti-p65 (Santa Cruz Biotechnology) (designated as p50-independent p65) with salmon sheared DNA (Ambion, Austin, TX) overnight for the sequential chromatin immunoprecipitation. The subsequent steps of washing, eluting, and the purification of DNA were conducted with the ChIP assay kit (Upstate). The purified DNA was suspended in 40 l of TE. Ten microliters were used for each PCR and amplified through 40 cycles. The PCR products were separated through an agarose gel and stained with ethidium bromide. Samples from the anti-p65 precipitation reaction were subjected to 45 cycles of PCR, and the gels were stained with SYPRO gold (Molecular Probes, Inc., Eugene, OR).
Quantitative Real-time PCR-Total RNA (Qiagen, Valencia, CA) was isolated from Jurkat cells that were transfected with 1 g of the various p65 constructs alone or with 1 g of DN-IB␣. The complementary DNA was generated from 1 g of RNA using the SuperScript III reverse transcription kit and random hexamers (Invitrogen). Onehundredth of the sample was used in a real-time PCR reaction containing a 0.2 M concentration of both forward and reverse primers (29) and SYBR green PCR master mix (Applied Biosystems, Foster City, CA). Quantitation of -fold induction was analyzed by the method described by Schmittigen et al. (30).

RESULTS
The role of serine phosphorylation at residue 536 in the regulation of the transactivation potential of p65 is not clearly defined. We investigated the status of serine 536 phosphorylation in human T cells and its role in activation-induced transcriptional activity. As seen in Fig. 1A, the activation of HeLa cells, but not Jurkat cells, resulted in an increase in the phosphorylation of p65 at serine 536 (henceforth referred to as phospho-p65). In addition, the activation of Jurkat cells by TNF did not induce an increase in p65 phosphorylation. Jurkat cells constitutively expressed a serine 536 phosphorylated form of p65, and this form appeared to be present mainly in the cytoplasm of unstimulated cells (Fig. 1B). Upon activation, the phospho-p65 appeared in the nucleus as early as 15 min after stimulation, peaked at 30 min, and stayed in the nucleus up to 2 h. As expected, stimulation of Jurkat cells resulted in the nuclear translocation of both p65 and p50 (Fig. 1B). Regarding IB␣, the level was decreased due to the early degradation upon PI stimulation and then increased as the protein was subsequently re-expressed. The success of the fractionation process and equivalent sample loading was confirmed by detecting equal amounts of p105 and Oct-1 in the cytoplasmic and nuclear extracts, respectively. These data suggested that a serine 536 phosphorylated form of p65 was present exclusively in the cytoplasm of unstimulated Jurkat cells, and upon activation, this phospho-p65 translocated to nucleus with similar kinetics to total p65.
It is conceivable that the constitutive phosphorylation of p65 on serine 536 was unique to the immortal Jurkat T cell line and was not a feature of primary human T cells. To investigate this question, human peripheral blood T cells were stimulated with PI for various time points, cytoplasmic and nuclear extracts were collected, and phospho-p65 levels were analyzed. As shown in Fig. 1B, phospho-p65 was constitutively expressed in the cytoplasm of normal human T cells. Furthermore, the FIGURE 1. Increase in nuclear phospho-p65 following PI treatment. A, induction of p65 phosphorylation at serine 536 in HeLa but not in Jurkat cells. Whole cell lysates from HeLa cells treated with PMA for 20 min and Jurkat cells treated with either PMA and ionomycin (PI) or 50 ng/ml TNF for 20 and 60 min were analyzed. B, increase in phospho-p65 levels in nuclear extracts after PI treatment. Jurkat cells were stimulated with PI for the indicated time period (h) as indicated at the top, and cytoplasmic and nuclear extracts were prepared as described under "Materials and Methods." C, increase in nuclear phospho-p65 after PI treatment in T cells. Enriched T cells from healthy donors were prepared as described under "Materials and Methods." T cells were stimulated with PI for the indicated time periods (h) as indicated at the top, and cytoplasmic and nuclear extracts were prepared. Five micrograms of total protein were loaded into each well, separated by gel electrophoresis, and transferred to a polyvinylidene difluoride membrane. The membrane was probed with the indicated antiserum and detected by chemifluorescence. Membranes were subsequently stripped and reprobed. Each experiment was repeated at least three times. levels of nuclear phospho-p65 increased following PI stimulation, and the levels of phospho-p65 were maintained throughout the 4 h of stimulation. A similar increase in nuclear p65 was observed following PI stimulation. Although a slight decrease in phospho-p65 was observed after 0.25 h of PI stimulation, the cytoplasmic levels of phospho-and total p65, in general, did not change with PI treatment. In contrast, increases in both cytoplasmic and nuclear c-Rel levels were detected following PI stimulation as described by others (31). To verify the subcellular fractionation process, the membranes were probed for p105 and Oct-1. As seen in the middle and bottom of Fig. 2B, p105 and Oct-1 were detected in their respective cytoplasmic and nuclear fractions through-out the 4 h of PI stimulation. These results were also observed in at least in five donors, whose cells were stimulated with either PI or PMA and anti-CD28. Taken together, these findings suggested that, in Jurkat and human T cells, at least a portion of the total pool of p65 is constitutively phosphorylated on serine 536, and phospho-p65 translocates to the nucleus following stimulation.
The canonical NF-B complex is composed of a dimer of p50 and p65 and predominantly resides in the cytoplasm bound to IB␣. Upon activation, IB␣ is phosphorylated, ubiquitinylated, and degraded resulting in nuclear translocation of p50 and p65 (2). To assess the relationship between the previously shown nuclear translocation of phospho-p65  6 -9). B, co-immunoprecipitation of unphosphorylated p65 with p50. Whole cell lysates were immunoprecipitated with the indicated amounts of normal or anti-p50 antibodies. C, detection of phospho-p65 in samples immunodepleted of p50. Whole cell lysates were incubated with normal goat immunoglobulin for 1 h. The remaining lysates were then incubated with anti-p50, and this process was repeated for a total of three times to completely immunodeplete p50 (lanes [3][4][5]. The p50-immunodepleted samples were subsequently immunoprecipitated with anti-p65 (lanes 6 and 7). D, phospho-p65 was not associated with IB␣ or p50 in peripheral blood T cell lysates. Whole cell lysates from peripheral blood T cells were immunoprecipitated with the indicated antibodies for 2 h. For all of the immunoprecipitation experiments, the input and output samples refer to the sample prior to and after immunoprecipitation analysis. The eluted samples were separated by gel electrophoresis, transferred to polyvinylidene difluoride membranes, and then probed for phospho-p65. The membranes were stripped and reprobed with the indicated antibodies to determine the success of the immunoprecipitation.
upon cell activation and the IB pathway, cytoplasmic lysates were used for immunoprecipitation with increasing amounts of antibody to IB␣, and the unbound proteins from the first immunoprecipitations were reimmunoprecipitated with anti-p65 antiserum. Immunocomplexes from both steps were analyzed by immunoblotting with total p65 and phospho-p65. As shown in Fig. 2A, p65 was co-immunoprecipitated with anti-IB␣ antiserum in a dose-dependent manner (see lanes 3-5 at the top), and this interaction was specific, since immunoprecipitation with normal immunoglobulin did not produce any IB␣ or p65 (see lane 2). At higher doses of anti-IB␣, immunoprecipitation of IB␣-associated p65 was complete, since no IB␣ was detected in the subsequent precipitates with anti-p65 antiserum (see lanes [7][8][9]. Surprisingly, a substantial amount of p65 was recovered from the unbound fractions by immunoprecipitation with anti-p65, indicating that only a small portion of total p65 was associated with IB␣ (compare lanes 3-5 with lanes 7-9). Interestingly, reprobing the membrane with anti-phospho-p65 antiserum revealed the presence of phospho-p65 exclusively in the fractions that were not associated with IB␣ (see lanes 7-9 of Fig. 2A). The presence of both IB␣ and phospho-p65 in lane 6 of the top and bottom of Fig. 2A, where the unbound sample of the normal immunoglobulin control immunoprecipitation reaction was subsequently immunoprecipitated with anti-p65 antibody, demonstrated that the normal immunoglobulin did not react with either IB␣ or p65. These data indicated that a portion of the total pool of p65 was phosphorylated at serine 536, and this phosphorylated p65 was not associated with IB␣.
Since phospho-p65 was not associated with IB␣ in the cytoplasm, we wanted to know whether phospho-p65 was able to form complexes with p50. Total cell lysates were subjected to immunoprecipitation with different amounts of anti-p50 antiserum, and the immunocomplexes were analyzed by immunoblotting using anti-phospho-p65 and anti-p65 antisera. As shown in Fig. 2B, anti-p50 antiserum was unable to co-precipitate phospho-p65, regardless of the amount of antiserum used in the immunoprecipitation (see lanes 3-5 at the top). This lack of interaction between p50 and phospho-p65 was not due to the inefficient immunoprecipitation by anti-p50 antiserum, since this antiserum was able to co-precipitate total p65 (see lanes 3-5 of both the middle and bottom of Fig. 2B). To further investigate the lack of interaction between p50 and phospho-p65, a series of sequential immunodepletions were performed with anti-p50 antiserum for three successive cycles under conditions that should preserve p50 associations with the other proteins. After the third cycle of immunodepletion, the unbound fraction was immunoprecipitated with anti-p65 antiserum for two more cycles. The immunocomplexes obtained from each step were analyzed by immunoblotting using either anti-phospho p65, anti-p65, or anti-p50 antiserum (Fig. 2C). Although anti-p50 antiserum was able to significantly deplete p50-p65 complexes (see lanes 3-5 of the bottom panel), a significant amount of p65 was recovered after the third cycle of immunoprecipitation by anti-p65 antiserum (lanes 6). Interestingly, phospho-p65 was detected in the fraction that was not associated with p50, indicating that the phospho-p65 was unable to form a complex with p50.
To determine whether the lack of interaction between phospho-p65 and p50 or IB␣ was also seen in human peripheral blood T cells, whole cell lysates from peripheral blood T cells were subjected to immunoprecipitation with either anti-IB␣, anti-p50, or anti-p65 antiserum, and the immunocomplexes were analyzed by immunoblotting with antiphospho-p65, anti-p65, anti-IB␣, and anti-p50 antisera (Fig. 2D). Although immunoprecipitation by both anti-IB␣ and anti-p50 antisera co-precipitated IB␣, p50, p105, and p65 (lanes 3-5 of Fig. 2D,  bottom), phospho-p65 was detected only in the p65-containing immunocomplex (lane 5 of top). Collectively, the data demonstrated that the phospho-p65 was not associated with either IB␣ or p50.
To investigate the functional consequences of the lack of interaction  4 -9) for 30 min and stimulated with PI for the indicated time periods (h) as indicated at the top. Cytoplasmic and nuclear extracts were prepared as before and analyzed by immunoblots.
A, MG-132 inhibited total p65, but not phospho-p65, nuclear translocation. Five micrograms of nuclear proteins were blotted initially for phospho-p65 (top panel) and then stripped and blotted for total p65 (middle panel). B, the membranes were restripped and blotted for Oct-1 to demonstrate equal loading (bottom panel). The band intensity of the phospho-p65 and total p65 blots were quantitated by the ImageQuaNT software and represented as a graph of phospho-p65 to total p65 ratio versus time. The experiment has been repeated several times with the same results. C, accumulation of phospho-IB␣ following PI and MG-132 treatment. Five micrograms of cytoplasmic protein were blotted for phospho-IB␣ levels.
The membranes were stripped and blotted for p105 to demonstrate equal loading (bottom panel).
between phospho-p65 and IB␣, we tested the effect of the proteosome inhibitor MG-132, which has been shown to block the degradation of IB␣ and the nuclear translocation of p50-p65 heterodimer (32). Jurkat cells were treated with MG-132 for 30 min before PI stimulation, and the nuclear translocation of phospho-p65 and total p65 were analyzed by Western blot analysis along with the status of IB␣ (Fig. 3A). The activation-induced nuclear translocation of phospho-p65 was not affected by the pretreatment with MG-132 (compare lanes 1-3 with lanes 7-9), whereas this pretreatment inhibited the nuclear translocation of total p65 in a dose-dependent manner (compare lanes 1-3 with lanes 7-9 of the middle of Fig. 3A). The densitometric measurements of the bands of Fig. 3A were graphically represented in Fig. 3B as a ratio of phospho-p65 to total p65. These data showed that the ratio of phospho-p65 to total p65 increased with MG-132 treatment, whereas the ratio decreased in the samples treated with the vehicle Me 2 SO. To assess the effectiveness of MG-132 treatment, phospho-IB␣ levels were analyzed in the cytoplasmic fractions. As shown in Fig. 3C, an increase in the phospho-IB␣ level was detected in the vehicle-treated samples 15 min after PI treatment, and this level returned to the basal level by 30 min (see lanes 2 and 3). In contrast, the samples pretreated with MG-132 possessed higher basal levels of phospho-IB␣, which were further increased upon PI stimulation and stayed up for the rest of the time they were treated (lanes 4 -9). These results suggested that the activationinduced nuclear translocation of phospho-p65 was independent of IB␣-mediated regulation, a finding that reinforced the conclusion that phospho-p65 was not associated with IB␣.
To investigate the contribution of serine 536 in the transactivation potential of p65, two p65 mutants were generated by site-directed mutagenesis where serine 536 was substituted with either alanine (S536A) or aspartic acid (S536D). S536A cannot be phosphorylated at serine 536, and S536D behaves like phospho-p65. The wild type and the mutant constructs were transiently transfected into relA Ϫ/Ϫ mouse embryo fibroblasts to determine whether the phospho-p65 antiserum could distinguish the two forms. The lysates from the transfectants were immunoprecipitated with an antibody against p65, and the immunoprecipitated samples were probed with the phospho-p65 antiserum. As shown in Fig. 4A, the phospho-p65 antiserum recognized only the S536D, but not the wild type and S536A, p65 construct, indicating the specificity of the phospho-p65 antiserum. The lack of reactivity of the phospho-p65 antiserum to the immunoprecipitated S536A p65 construct was not due to the lack of expression of the mutant construct, since the expressions of all of the constructs were detected by the anti-V5 antibody (Fig. 4A, bottom).
Next, we transfected the different constructs into Jurkat cells and tested the transactivation potential of the constructs in a reporter assay using a reporter construct where multicopy canonical NF-B binding sites from the immunoglobulin gene were fused to the ␤-galactosidase gene. As shown in Fig. 4B, the substitution of serine 536 with alanine resulted in the loss of transcriptional activity of p65, whereas the transcriptional activity was not inhibited by the substitution of serine with aspartic acid. Interestingly, at a lower level of S536D construct, an increase (2-fold) in reporter activity was observed. Since it was possible that the substitution of alanine for serine 536 could affect the nuclear translocation of the S536A mutant construct, cytoplasmic and nuclear extracts were collected from Jurkat cells transfected with the different mutant constructs before and after PI stimulation (Fig. 4C). The transiently transfected p65 constructs were present in the nucleus before PI stimulation. After PI stimulation, an increase in both endogenous and exogenously transfected p65 was detected in the nucleus. These findings indicated that the loss of transcriptional activity observed with S536A was not due to the inability of the mutant to translocate into the nucleus. To investigate whether the transcriptional activity of the S536D construct was unaffected by IB␣-mediated regulation, DN-IB␣ was included in the reporter assay. As shown in Fig. 4D, the co-transfection of the DN-IB␣ inhibited the transcriptional activity of the wild type p65 construct in a dose-dependent manner, whereas the DN-IB␣ construct had a minimal effect on the transcriptional activity of the S536D construct. It is conceivable that the minimal inhibitory effect of DN-IB␣ on NF-B transcriptional activity of Jurkat cells expressing the S536D construct was directed against the endogenous p65. Collectively, the data suggested that the serine 536 phosphorylated p65 was a unique entity, whose nuclear translocation mechanism was independent of IB␣-mediated regulation.
The evidence suggested that there are at least two populations of p65 that can be distinguished based on their phosphorylation status at serine 536 and their association with p50. It is conceivable that the two populations of p65 may influence the transcription of distinct sets of NF-B responsive genes. To address this question, we have performed a modified ChIP assay. The antibody against phospho-p65 is incapable of immunoprecipitation. Therefore, samples prepared for the ChIP assay from Jurkat cells treated with PI were initially incubated with antibodies against p50, and then the remaining p50-depleted sample was incubated with antibodies against p65 (p50-independent p65 panel of Fig. 5). In addition, ICAM-1 and IL-8 promoters were selected, because previous reports have demonstrated the binding of the p65 homodimer to these sites (33,34). As seen in Fig. 5, PI stimulation resulted in the association of acetylated histone H4 at the ICAM-1 and IL-8 promoter, which is associated with chromatin rearrangement (lanes 4 and 5 of the Acetyl H4 panel). Similarly p65 was recruited to the ICAM-1 and IL-8 promoter following stimulation (lanes 4 and 5 of the Total p65 panel). A distal upstream region of the IL-8 promoter, which does not contain a NF-B response element, was used as a control to monitor specificity (see lanes 3 and 6 of all panels except Input). The modified ChIP analysis showed the association of the p50-independent p65 only with the IL-8 promoter following PI stimulation (see lane 5 of p65 of the p50-independent p65 panel). This interaction was specific to the IL-8 promoter, since no such recruitment was observed at the ICAM-1 promoter (compare lane 4 with lane 5 of the p50-independent p65 panel). Furthermore, the absence of p65 following the immunodepletion of p50 only at the ICAM-1 promoter indicated that there was a selective recruitment of either the p50-p65 or p50-independent p65 dimer to the ICAM-1 or IL-8 promoter (compare lane 4 of Total p65 with the p50-independent p65 panel). It is not clear if the dimer composition or promoter sequence determines the selective recruitment following stimulation. Taken together, PI stimulation resulted in histone H4 acetylation and recruitment of the p50-independent p65 to the IL-8, but not ICAM-1, promoter. Thus, the p50-independent p65 may be involved in the regulation of IL-8 transcription.
Although the p50-independent p65 was selectively recruited to the IL-8 promoter following PI stimulation, it was not certain whether the recruited p65 was phosphorylated at serine 536 and whether the recruitment of p65 to the promoter resulted in the induction of IL-8 transcription. To investigate this question, Jurkat cells were transfected with the various p65 constructs alone or with the dominant negative IB␣ construct and stimulated with PI, and the levels of IL-8 transcripts were analyzed. Upon PI stimulation for 2 h, the levels of IL-8 transcripts increased ϳ80-fold with the cells transfected with the control vector, pEF (Fig. 6A). The increase in IL-8 mRNA levels could be attributed to the activation of endogenous p65 along with other transcriptional factors (35). The increase in IL-8 transcript levels following activation was . Phosphomimetic p65 mutant S536D was not inhibited by IB␣. A, immunoreactivity of phospho-p65 antibody to serine 536 site-directed mutants. Whole cell lysates from relA Ϫ/Ϫ mouse embryo fibroblasts transiently transfected with the indicated V5-tagged p65 mutant constructs (control (pEF), wild type (p65), serine 536 to alanine (S536A), and serine to aspartic acid (S536D)) were immunoprecipitated with goat anti-p65 antibodies and probed with anti-phospho-p65 antibody. The membranes were stripped and reprobed with alkaline phosphatase-conjugated anti-V5 antibody to detect p65 expression. B, phosphorylation of serine 536 is required for p65 transcriptional activity. Jurkat cells were transiently transfected with either the vector (pEF), p65, S536A, or S536D at the indicated amounts along with the NF-B/␤-galactosidase reporter plasmid. After a 48-h incubation, the lysates were assessed for ␤-galactosidase activity for p65 transactivational activity. The bottom panel represents the expression control where transfectant lysates were probed for V5 and ␤-tubulin expression. C, substitution of alanine for serine 536 does not prevent nuclear translocation of p65. Cytoplasmic (CE) and nuclear (NE) extracts were prepared from Jurkat cells transiently transfected with the different mutant constructs before and after PI stimulation for 1 h. The levels of both endogenous and exogenous (p65-V5) p65 were analyzed by immunoblot to assess nuclear translocation. ␤-tubulin and Oct1 served as cytoplasmic and nuclear loading controls. D, S536D mutant is partially resistant to IB␣ inhibition. As described earlier, Jurkat cells were transiently transfected with 1 g of either the wild type p65 or S536D mutant expression construct along with varying amounts of DN-IB␣. The bottom panels represent the expression control, where transfectant lysates were probed for V5, ␤-tubulin, and IB␣ expression. The data points and error bars in the graphs represent the average and S.D. of at least three independent experiments. partially inhibited with the co-transfection of the DN-IB␣. This partial inhibition could be due to the presence of endogenous phosphorylated p65. The transfection of the wild type p65 resulted in an additional increase in IL-8 induction following PI stimulation. This induction was abrogated with the co-transfection of DN-IB␣. In contrast, the transfection of the S536A p65 mutant did not result in an additional IL-8 induction compared with the control vector alone. However, an increase in IL-8 transcript levels was detected in cells transfected with the phosphomimetic p65 S536D that was equal to the amount observed with the wild type p65. Furthermore, co-transfection of the DN-IB␣ had a minimal effect on S536D-mediated induction of IL-8 transcription. In contrast to the increase in IL-8 transcript levels, ICAM-1 transcript levels were not induced with the expression of the S536D p65 mutant. An increase in ICAM-1 levels was detected following the activation of cells transfected with wild type and the S536A p65 constructs. Moreover, the induction of ICAM-1 transcript was inhibited with the co-transfection of the DN-IB␣. These findings suggest that the S536D p65 is resistant to the inhibitory activity of DN-IB␣ and that the transcription of IL-8 involves the phosphorylation of Ser 536 . Taken together, although the phosphorylation status of the p50-independent p65 recruited to the IL-8 promoter is unclear, the data collectively suggested that phospho-p65 induced IL-8 transcription following PI stimulation based on the fact that 1) the phospho-p65 was not associated with either p50 and IB␣, 2) the p50-independent p65 was recruited to the IL-8 promoter following PI stimulation, and 3) the phosphomimetic form of p65 S536D was resistant to IB␣-mediated regulation.

DISCUSSION
In the present study, we have shown that a portion of total NF-B p65 was constitutively phosphorylated at serine 536, and this phosphorylated form was present in the cytoplasm of both Jurkat and human peripheral blood T cells. Activation of the cells caused nuclear translocation of the phospho-p65 with similar kinetics as with total p65 without affecting the phosphorylation status. A unique feature of this form of p65 was that the phospho-p65 was not associated with either p50 or IB␣, and as a result, the activation-induced nuclear translocation of the phospho-p65 was independent of IB␣-mediated regulation. Furthermore, ChIP analysis revealed that upon activation, the p50-independent p65 was selectively recruited to the NF-B site of the IL-8 promoter but not the ICAM-1 promoter. Furthermore, the transfection of the phosphomimetic form S536D resulted in an increase in IL-8 transcription following PI stimulation, whereas the nonphosphorylated p65 S536A did not, and IB␣ minimally inhibited the induction.
The inducible phosphorylation of p65 at serine 536 following stimulation has been described by several groups (21, 36 -39). In agreement with the published reports, we have detected the inducible phosphorylation of p65 in HeLa cells following stimulation with either PMA or TNF-␣ (data not shown). On the other hand, we observed the constitutive phosphorylation of p65 at serine 536 in T cells, and the phosphorylation status was unchanged upon activation, which is in contrast to the report by Mattioli et al. (40), who have shown the inducible phosphorylation of p65 in Jurkat cells stimulated through T cell receptors. The discrepancy between our results could be due to undefined differences in the Jurkat cell lines. We have also observed the constitutively phosphorylated form of p65 at serine 536 in human peripheral blood T cells, and like Jurkat, the phosphorylation status was unchanged upon activation.
The inability to co-immunoprecipitate phospho-p65 with either p50 or IB␣ in both cytoplasmic and whole cell lysates of either Jurkat or human peripheral blood T cells suggested that the phospho-p65 did not exist in the canonical NF-B complex of p50 and p65. In support of our results, Bohuslav et al. (41) have shown that the phosphorylation of p65 at serine 536 by the ribosomal S6 kinase 1 resulted in the loss of association between phospho-p65 and IB␣ in vitro. Moreover, the nuclear translocation of phospho-p65 was not affected by the proteosome inhibitor MG-132, whereas this inhibitor blocked the degradation of IB␣ and the nuclear translocation of unphosphorylated p65 (Fig. 4). Interestingly, in the report by Mattioli et al. (40), the activation-induced nuclear translocation of phospho-p65 was also unaffected by the pretreatment of MG-132 (Fig. 5B), but this point was not addressed in their work. The role of phosphoserine at 536 in IB␣-independent regulation of nuclear translocation of phospho-p65 was further demonstrated in the reporter assay where p65 (S536D)-mediated transactivation was minimally affected by a dominant negative construct of IB␣, whereas this dominant negative construct substantially blocked wild type p65mediated transactivation of the reporter gene (Fig. 5C). The inability of MG-132 and the dominant negative IB␣ to prevent the nuclear translocation of phospho-p65 was consistent with phospho-p65 not being associated with IB␣, indicating the possibility of the involvement of another IB-like protein that is not regulated by proteosomal degradation. It is important to note that we could not detect the association of either IB␤ or IB⑀ with the cytoplasmic phospho-p65 (data not shown). Although the factors regulating the nuclear translocation of phospho-p65 are not known, the data demonstrated that the phospho-p65 was retained in the cytoplasm without the association with IB␣ and p50. The cytoplasmic localization could be explained by both the action of the nuclear export sequence of p65 (42) and the lack of p50 association, since the nuclear localization sequence of p50 has been shown to dominate the nuclear export sequence of p65 (4). It is possible that the nuclear translocation of phospho-p65 after stimulation is facilitated by another protein not involved in the canonical NF-B pathway. Akiyama et al. (43) have demonstrated the nucleocytoplasmic shuttling of the human telomerase reverse transcriptase by phospho-p65. However, the association of human telomerase reverse transcriptase and FIGURE 5. Recruitment of p50-independent p65 to the IL-8 promoter following PI stimulation. Selective recruitment of p65/p50 heterodimer and p50-independent p65 dimers to the ICAM-1 and IL-8 promoter following PI stimulation. Jurkat cells were treated with PI for the indicated times and processed for chromatin immunoprecipitation analyses as described under "Materials and Methods." Input, 1% of the total sample. The indicated antibodies were used for the chromatin immunoprecipitation experiment. The p50-independent p65 refers to the anti-p65 ChIP analysis of the p50-immunodepleted sample as described under "Materials and Methods." phospho-p65 was not detected in Jurkat T cells (data not shown). The identification of a phospho-p65-interacting protein by current proteomic approaches will be fruitful in understanding the regulation of the nuclear translocation of phospho-p65. It is important to note that the lack of availability of an antiserum capable of immunoprecipitating endogenous phospho-p65 makes the identification of phospho-p65-interacting protein(s) difficult.
Although the IL-8 and ICAM-1 promoters were selected based on previous reports describing the binding of the p65 homodimer to these promoters by gel shift assays, p65 ChIP analyses of PI-stimulated Jurkat cells demonstrated the association of the p50-independent phospho-p65 only to the IL-8 promoter. The lack of recruitment of the p50independent p65 to the ICAM-1 promoter could be attributed to the difference in cell type, since the binding of p65 homodimers to the IL-8 promoter has been described in endothelial cells following TNF stimulation (34). Furthermore, the selective recruitment of p65 to either the ICAM-1 or IL-8 promoter was dependent on whether p65 was associated with p50. A recent study by Leung et al. (44) has demonstrated that a single nucleotide can influence the recruitment of specific NF-B dimers and the required cofactors for efficient gene transcription. Although it is not certain if the IL-8 or ICAM-1 promoter sequence influences the recruitment of the p50-independent phospho-p65, it is conceivable that the constitutive phosphorylation of p65 at serine 536 could affect p65 dimer composition and thereby confer the selective recruitment of phospho-p65 to specific promoters (45). It is also possible that the recruitment of this specific form of p65 in context with other transcription co-factors and modifiers of chromatin at various gene promoters will determine whether the phospho-p65 will activate gene transcription. Furthermore, it is not known what role the other described phosphorylated serine may play in the regulation of transcription of specific genes or whether there is a cooperative or antagonistic relationship shared among the various phosphoserines of p65 in gene transcription.
The loss of p65 transcriptional activity as a result of substituting serine with alanine at residue 536 has been unclear. It has been reported that the substitution of serine 536 with alanine abrogated Akt-mediated (38), IKK␤-mediated (21), TNF-mediated (39), lymphotoxin-␤-mediated (37), and Gram-negative bacteria-mediated (36) p65 transcriptional activity. However, Okazaki et al. (23) and Buss et al. (46) have shown that the substitution of alanine for serine 536 did not have any effect on TNF-and IL-1-mediated NF-B signaling, in particular IL-6 production and IL-8 transcription. Our findings are in agreement with both situations where the substitution of alanine for serine 536 resulted in a loss of p65 transcriptional activity, as seen in the NF-B reporter assay, but the induction of ICAM-1 was not affected. The critical role of serine 536 was also shown in the ChIP analysis, where the p50-independent p65 (presumably the phosphorylated form) was recruited to the IL-8 promoter, but not the ICAM-1 promoter, and the functional consequence of this recruitment was an increase in IL-8 transcription by the S536D construct but not by the S536A construct. Taken together, the conflicting data on whether the substitution of serine 536 with alanine of p65 is required for transcriptional activity may be dependent on the promoter, the activation signal, and the cell type that is being analyzed. The role of the Rel homology domain of p65 in regulating dimerization has been well documented (13) and would suggest that the phosphorylation of serine 536 in the second transactivation domain does not influence dimerization. Interestingly, the data presented here suggest that two mutually exclusive populations of p65 that can be distinguished based on their phosphorylation status at serine 536 exist in the cytoplasm of resting T cells. Upon activation, these two populations translocate to the nucleus independently and appear to induce transcription of a different set of genes. The identification of the kinase(s) responsible for the phosphorylation at serine 536 will enhance our understanding of the regulation of the phospho-p65 population. We propose that the pathway leading to the activation of phospho-p65 represents a noncanonical NF-B pathway, and the role of serine phosphorylation at 536 might be more than the regulation of p65 transactivation.