Characterization and Quantitation of NF-kB Nuclear Translocation Induced by Interleukin-1 and Tumor Necrosis Factor-a DEVELOPMENT AND USE OF A HIGH CAPACITY FLUORESCENCE CYTOMETRIC SYSTEM*

A new quantitative cytometric technique, termed the ArrayScanTM, is described and used to measure NF-kB nuclear translocation induced by interleukin (IL)-1 and tumor necrosis factor-a (TNFa). The amount of p65 staining is measured in both the nuclei defined by Hoechst 33342 labeling and in the surrounding cytoplasmic area within a preselected number of cells/well in 96-well plates. Using this technique in synchronously activated human chondrocytes or HeLa cells, NF-kB was found to move to the nucleus with a half-time of 7–8 min for HeLa and 12–13 min for chondrocytes, a rate in each case about 4–5 min slower than that of IkBa degradation. IL-1 receptor antagonist and anti-TypeI IL-1 receptor antiserum on the one hand and anti-TNFa and monoclonal anti-TNF receptor 1 antibodies on the other hand could be shown to respectively inhibit IL-1 and TNFa stimulation in both cell types. In contrast, a polyclonal anti-TNF receptor 1 antiserum exhibited both a 50% agonism and a 50% antagonism to a TNFa stimulation in a dose-dependent fashion, indicating that subtle functional responses to complex agonist and antagonist stimuli could be measured. The effects of different proteasome inhibitors to prevent IkBa degradation and subsequent NF-kB translocation could also be discriminated; LeuLeu-Leu aldehyde was only a partial inhibitor with an IC50 of 2 mM, while clastolactacystin b-lactone was a complete inhibitor with an IC50 of 10 mM. The nonselective kinase inhibitor K252a completely inhibited both IL-1 and TNFa stimulation in both cell types with an IC50 of 0.4 mM. This concentration, determined after a 20-min stimulation, was shown to be comparable with that obtained for inhibition of IL-6 production induced by a 100-fold lower IL-1 and TNFa concentration measured after 17 h of stimulation. These results suggest that the ArrayScanTM technology provides a rapid, sensitive, quantitative technique for measuring early events in the signal transduction of NF-kB.

IL-1 1 and TNF␣ are two master cytokines that induce an almost identical proinflammatory response, including the pro-duction of chemotactic cytokines, adhesion molecules, and enzymes such as cyclooxygenase, nitric-oxide synthetase, and matrix metalloproteinases (1,2). Many of these effects are a result of the activation by both IL-1 and TNF␣ of the NF-B transcription factor pathway, which is associated with the activation of many cellular defense genes (3,4). Composed of p65 (RelA) and p50 proteins, NF-B is normally present in the cytoplasm in an inactive state in a complex with members of the IB inhibitor protein family, chiefly the 37-kDa IB␣ form. In this complexed form, a nuclear localization sequence found on NF-B is masked by the IB␣, preventing nuclear translocation of NF-B, DNA binding, and subsequent transcriptional activation (5)(6)(7)(8)(9)(10)(11)(12). IL-1 or TNF␣ receptor activation induces within several minutes the specific phosphorylation of Ser 32 and Ser 36 on IB␣, the destruction of the phosphorylated IB␣ protein by proteasomes, and the translocation of NF-B to the nucleus (13)(14)(15)(16)(17). Recent reports have identified an IL-1-and TNF␣-activated Ser/Thr kinase cascade containing at least four kinases that serially phosphorylate each other prior to the phosphorylation of IB␣ (reviewed in Refs. 18 and 19). The inhibition of proteasome activity by specific inhibitors as LLL-H (MG132) prevented both the destruction of IB␣ as well as the subsequent activation of NF-B as measured by the production of NF-B-dependent proteins such as leukocyte adhesion proteins (20,21). Cellular assays of the early signaling events leading up to NF-B activation are difficult because of the rapidity and complexity of the protein interactions. To show activation of NF-B, electrophoretic mobility shift assays are typically performed to look at the specific binding of activated NF-B to DNA (5,7,22), but this technique requires relatively large numbers of cells and is laboriously quantitative, and the assay is not performed in intact cells. In contrast, the use of gene reporter constructs in transfected cells measures a response occurring hours after cell activation, and the resulting gene transcription is influenced by other transcription factors that act cooperatively to activate individual genes (12,23). Because protein translocation from the cytoplasm to nucleus can be readily visualized by immunocytolocalization (see e.g. Refs. 9, 10, 24, and 25), a computerized cytometric fluorescence system, termed the ArrayScan™, has been developed to analyze translocation of cytoplasmic proteins in cells grown in 96-well plates (26). We have used this ArrayScan™ system to quantitate the rate and extent of NF-B translocation following stimulation of varying IL-1 and TNF␣ concentrations. We show that in human chondrocytes and HeLa cells complete translocation of NF-B occurs within 10 -20 min, with a half-time several minutes following that of IB␣ destruction. Furthermore, the effects of receptor agonists and antagonists and kinase and proteasome inhibitors can be differentially quantitated.

EXPERIMENTAL PROCEDURES
Cells-Human THC-igfI chondrocyte cells were obtained from M. B. Goldring (27) and passaged weekly in Dulbecco's modified Eagle's medium (low glucose; Life Technologies, Inc.) with 10% FBS. HeLa cells were obtained from the ATCC collection and passaged weekly in Dulbecco's modified Eagle's medium with high glucose. For experiments, cells were removed with 0.25% trypsin/EDTA, plated in 96-well plates (Polyfiltronics, Rockland, MA) at 10,000 cells/75 l/well, and grown for 20 h.
Translocation Experiments-The plates were washed once in Dulbecco's modified Eagle's medium without FBS, and 90 l of fresh medium was added. Various compounds were added in Me 2 SO, and up to 1% Me 2 SO could be added without any effects on NF-B translocation (data not shown). The compounds used together with their source included K252a (Alexis Biochemicals, San Diego, CA), carbobenzoxy-Leu-Leu-Leu-aldehyde (Peptide Institute, Osaka, Japan), and clastolactacystin ␤-lactone (Affiniti Research Products, Mamhead, Exeter, United Kingdom). Protein antagonists were added in Dulbecco's modified Eagle's medium and included IL-1 receptor antagonist (IL-1RA; prepared at Merck), neutralizing anti-TNF␣ and monoclonal anti-TNFR1 and anti-TNFR2 antibodies (R & D Systems, Minneapolis, MN). Polyclonal IL-1R1, anti-TNFR1, and anti-TNFR2 antibodies were prepared at Covance (Denver, PA) from soluble IL-1R1, TNFR1, and TNFR2 (R & D Systems). All samples were performed at least in duplicate. The cells were preincubated for 20 min at 37°C, and then 10 l of stimulator (typically 10 ng/ml (final concentration) IL-1␣ or TNF␣; R&D Systems) was added for each well. The cells were repipetted twice for mixing and then incubated for typically 20 min more at 37°C. Experiments were ended by washing the plates twice in ice-cold phosphate-buffered saline followed by fixation.
Cell Fixation and Staining-The cells were fixed with 100 l of 4% formaldehyde in phosphate-buffered saline for 20 min at room temperature, permeabilized with 100 l of 0.1% Triton X-100 in phosphatebuffered saline for 5 min at room temperature, and then washed twice with 300 l of 0.1 M Tris-HCl buffer, pH 7.8. To block nonspecific antigenic sites, the wells were incubated for 20 min with 100 l of 5% nonfat dry milk in 0.1 M phosphate buffer, pH 7.8, at room temperature. After washing two times in 0.1 M Tris wash buffer, the cells were incubated for 1 h with 100 l of rabbit anti-p65 NF-B antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) diluted 1:2000 in 0.1 M phosphate buffer, pH 7.8, with 0.1% bovine serum albumin (fraction V; Sigma). The plates were washed three times in Tris wash buffer and incubated 30 min, room temperature, with 100 l of a 10 g/ml solution in water of biotinylated antirabbit IgG (Vector Laboratories, Burlingame, CA). The plates were washed three times in Tris wash buffer and incubated 30 min, room temperature, with 100 l of a 2.5 g/ml solution of Texas Red avidin (Vector) in the phosphate/bovine serum albumin buffer. The cells were washed three times in Tris wash buffer and stored in 100 l of 0.1 M Tris. Two hours prior to analysis 100 l of a 1 g/ml solution of Hoechst 33342 (Molecular Probes, Inc., Eugene, OR) in phosphate-buffered saline was added to each well at room temperature, and the wells were scanned in the ArrayScan™ instrument.
Data Acquisition and Analysis-The ArrayScan™ cytometer (Cellomics, Inc., Pittsburgh, PA) is an automated fluorescent imaging microscope for extracting information about the spatial and temporal distribution of fluorescently labeled components in cells grown in the microtiter plates (26). The system was used to scan multiple fields from well to well and to acquire and analyze each of the cells in the images according to the algorithm described in Fig. 2. Within each well, multiple cellular images/well were acquired by moving the position of the plate the width of one image field (350 m) in a square pattern of locations centered on the center of the well. In each well, images were acquired until a preselected number of cells had been imaged and analyzed. The ArrayScan™ system consists of an optical system with a spatial resolution of 0.68 m (Carl Zeiss, Inc., Thornwood, NY); a triple band fluorescence emission filter set with matched single band excitation filters for selectively imaging Hoechst, FITC, or Texas Red (model XF57, Omega Optical, Brattleboro, VT); a CCD camera with frame grabber; and a Pentium PC computer and applications software.
Immunoblotting and Densitometry-Cells grown on 96-well plates were lysed in 30 l of 2ϫ SDS-PAGE sample buffer, combining eight wells into one sample (2.4 ϫ 10 5 cells). Alternatively, cells were grown in 35-mm clusters at the same density (10 6 cells/well) and solubilized in 50-l sample buffer. SDS-PAGE was performed on 10% Novex gels (San Diego, CA) using 6.4 ϫ 10 4 cell equivalents/lane. Immunoblotting was performed with rabbit anti-IB␣ antibody (Santa Cruz Biotechnology) with ECL visualization as described earlier (28). Quantitative densitometry was performed using PDI Quantity One software (Bio-Rad) as previously reported (29).

Cytometric Measurements of p65 NF-B Localization in Cytoplasmic and Nuclear Cellular Compartments-Comparable
with previous reports (10,24), immunofluorescence staining of p65 in unstimulated HeLa cells and human chondrocytes shows largely a cytoplasmic location with significantly less nuclear fluorescence (Fig. 1, A and D, respectively). Following a 20-min stimulation with either IL-1 (Fig. 1, B and E) or TNF␣ ( Fig. 1, C and F), large amounts of the p65 are found in nuclei, although significant amounts of p65 still remain in the cytoplasm. To quantitate this shift in p65 immunofluorescence, a technology needed to be developed to measure fluorescence separately in both the cytoplasm as well as the nucleus and to be able to do so in large numbers of cells in multiple wells. This was achieved with the development of the ArrayScan™ cytometer as is outlined in Fig. 2. On an individual cell basis, the cells were stained not only with anti-p65 antibodies to identify NF-B but also with Hoechst 33342 to identify the nucleus of all of the cells in the well. The ArrayScan™ cytometer focused on the stained nucleus and determined the nuclear boundary by the application of a dynamic thresholding method that identifies a change in fluorescence above the background (Fig. 2, Step 2). To delineate a representative nuclear region relatively free of contaminating cytosol, the nuclear boundary to be analyzed was eroded by 2 pixels (Fig. 2, Step 3). The resultant nuclear ring is shown in Fig. 3, where in unstimulated and IL-1-stimulated HeLa cells (Fig. 3, A and C, respectively) it is superimposed on the p65 immunofluorescence image (Fig. 3, B and D). The cytoplasm was identified by the area lying between a 2-pixel-wide annular ring drawn around the nuclear boundary and the nuclear boundary itself (Fig. 2, Step 3). In images of unstimulated (Fig. 4, A-D) and IL-1-stimulated HeLa cells (Fig. 4, E-H), the cytoplasmic boundary is denoted by the two red rings.
Quantitation of the amount of p65 found in each cell is shown in 16 representative unstimulated and IL-1-stimulated HeLa cells (Fig. 5). The amount of p65 fluorescence found in the nucleus is a composite of background fluorescence plus a small amount of cytoplasm surrounding the nucleus, the result averaged across all of the pixels within the nuclear area (see Fig. 2, Side View). As is seen visually in Figs. 1, 3, and 4, the calculated p65 nuclear fluorescence increases after stimulation (Fig.  5A). The cytoplasmic fluorescence is a composite of background fluorescence within the cell as well as some background fluorescence observed in the annular ring that might fall outside of the cells (see Fig. 4). The occasional placement of the cytoplasmic ring in part outside of the cells often led to a reduced mean cytoplasmic intensity because of averaging in noncytoplasmic regions. After stimulation, the cytoplasmic staining was only partially reduced, suggesting only a partial translocation of NF-B (Fig. 5B). Because of the low signal/noise (stimulated versus unstimulated cell values) observed in both the nuclear (Fig. 5A) and cytoplasmic regions (Fig. 5B) following stimulation, the individual cellular mean cytoplasmic staining was subtracted from its corresponding nuclear staining. This nuclear/cytoplasmic difference yielded a much more sensitive signalto-noise ratio (Fig. 5C). Consequently the nuclear/cytoplasmic difference was typically used as a measure of NF-B translocation.
Time Course of NF-B Translocation-To determine how long the assays needed to be run to achieve maximal NF-B translocation, HeLa and chondrocyte cells were incubated for up to 45 min following stimulation by a 25-ng/ml dose of IL-1 or TNF␣. As is shown in Fig. 6, the HeLa cells responded slightly faster than the chondrocytes with a time for 50% maximal response being reached by 7-8 min for TNF␣ and IL-1␣ stimulation, whereas the chondrocytes were slightly slower with 50% maximal responses at 12-13 min. For both cells, maximal response was achieved by 20 min. Hence, for the translocation assays, a 20-min incubation was typically used for both cell types.
Because NF-B translocation follows IB␣ phosphorylation and degradation, the corresponding effects on IB␣ in the cell types were also studied using IL-1␣ as a stimulus. As is shown in Fig. 7, IB␣ was almost totally destroyed by 10 min after stimulation. The time for 50% maximal response was about 4 min for the HeLa cells and about 7.5 min for the chondrocytes, indicating that IB␣ proteolysis preceded NF-B nuclear translocation by 4 -5 min.
IL-1 and TNF␣ Dose Response-Titration of the concentration of IL-1 and TNF␣ used for cell stimulation indicated that maximal stimulation of each cytokine occurred at about 10 ng/ml in each cell type (Fig. 8). The dose response for IL-1 was steeper than that of TNF␣ in both cell types; for IL-␣ stimulation the ED 50 for both cell types was about 1-2 g/ml, whereas for TNF␣ stimulation the ED 50 was about 0.03-0.05 g/ml.
Cell Numbers Required for Statistical Significance-To determine the minimal number of cells required for the assay to Hoechst 33342 staining defines the nucleus and enables focusing on the cells within the field (Step 1). Each cell completely within the field of view is identified by number, and the nuclear boundary is drawn (Step 2). To reduce cytoplasmic contamination within the nuclear area, a nuclear ring 2 pixels inner to the nuclear boundary is drawn to define the nuclear area used for quantitation. The outermost cytoplasmic boundary is drawn 2 pixels beyond the nuclear boundary, and the cytoplasmic area used for quantitation is that which lies between nuclear boundary and this outer cytoplasmic annular ring. In both the nuclear and cytoplasmic areas, the intensity of fluorescence of a given protein is averaged throughout the measured area. The typical results for a cytoplasmic protein stained in green are shown at the bottom. Prior to stimulation, the cytoplasm is intensely green while the nucleus without the protein is white as is seen from the side view (viewing from the top shows some nuclear staining because of cytoplasmic overlap above and below the nucleus). Hence, the nuclear minus cytoplasmic (Nuc-Cyt Difference) fluorescence is low. In the stimulated cell, the amount of cytoplasmic fluorescence is decreased, while the nuclear fluorescence markedly increases, producing a much darker green in the nucleus due to the concentration of the cytoplasmic protein.
Hence the nuclear minus cytoplasmic fluorescence is high. assure statistical significance, a series of plates containing unstimulated and IL-1␣-stimulated chondrocytes were analyzed by two-way analysis of variance. In these experiments, the mean difference between nuclear and cytoplasmic p65 staining in stimulated cells was about 55 units more than in unstimulated cells. The number of cells needed for analysis to be assured that a change of 90% had occurred (i.e. a nuclear/ cytoplasmic difference of 49.5 (0.9 ϫ 55)) was 56 at the 99% confidence level (i.e. a 1% false positive, type I error; see Table  I). This was true regardless of whether the change was due to the addition of varying amounts of a cytokine stimulator or to varying amounts of an inhibitor added to a 10-ng/ml cytokine stimulation. For 95% confidence that the 90% inhibition was achieved, a total of only 37 cells was required to be analyzed. For statistically significant smaller changes in the nuclear/ cytoplasmic difference, correspondingly larger numbers of cells needed to be analyzed to achieve comparable confidence (Table  I). If only the change in nuclear p65 staining were analyzed, rather than the nuclear/cytoplasmic difference, then about 9% more cells were required to be analyzed to give the same statistical significance (data not shown). Based upon these data, a cell number of 75 was typically analyzed in the experiments described in this paper.
Assay Variability-Several replicate plates of chondrocytes stimulated with IL-1␣ in the presence or absence of an IL-1RA antagonist were analyzed by the ArrayScan™ system. Reproducibility of this assay for screening purposes was assessed by calculating the coefficient of variation between IC 50 (with IL-1RA) or EC 50 (without IL-1RA) values generated from 11 point curves in each row (n ϭ 8) of a 96-well microtiter plate. In this way, row to row, plate to plate, and day to day variability was measured and did not exceed 20, 30, and 22%, respectively ( Table II). The reproducibility of the detection method was evaluated by repeatedly scanning the same plate five times, yielding after analysis a coefficient of variation of 6.8% (Table II).

Effects of Cytokine and Receptor Antagonists on NF-B
Translocation-IL-1RA (30) inhibited the NF-B translocation with an IC 50 of 20 -60 nM, a concentration about 100-fold higher than the stimulatory IL-1 concentration (Fig. 9, top), but it had no effect on the activation of NF-B translocation induced by TNF␣ (not shown). This IC 50 concentration of IL-1RA determined by the ArrayScan™ method was comparable with that shown earlier to be required to inhibit IL-1-driven cellular responses (31). For complete IL-1 antagonism, 1000-fold higher IL-1RA concentrations were required to prevent NF-B translocation (Fig. 9, top), just as were seen in other bioassays (32). In contrast, a neutralizing anti-TNF␣ antibody blocked TNF␣ stimulation of NF-B translocation (Fig. 9, bottom) but had no effect on IL-1␣ stimulation (not shown). The blockage of IL-1 stimulation of NF-B translocation by IL-1RA occurred at the same concentration as that necessary to block IB␣ degradation. In an experiment in which IB␣ degradation was quantitated in a similar plate in which NF-B translocation was measured, both assays yielded the same 24 nM IL-1RA concentration for 50% inhibition (Fig. 10). This gives further evidence of the close correlation of NF-B translocation to the degradation of IB␣.
A similar inhibition was performed with antibodies specific for individual cytokine receptors. IL-1 stimulation of chrondrocyte or HeLa cell NF-B translocation could be prevented by serial dilutions of an anti-IL-1R1 antiserum but not by a preimmune bleed (Fig. 11, A and C, respectively). The antiserum on its own had no agonist activity, indicating that it was a strict antagonist. Similarly, a neutralizing monoclonal antibody against the TNFR1 blocked TNF␣-induced NF-B translocation, but it had no agonism of its own, and a monoclonal anti-TNFR2 antibody was ineffective (Fig. 11, B and D). This indicated that all of the functional TNF receptors in these cell types were TNFR1. In contrast, when a polyclonal anti-TNFR1 antibody was added with chondrocytes stimulated with TNF␣, it showed both partial antagonism and partial agonism (Fig.   11E). This antiserum in the absence of TNF␣ gave roughly a maximal 50% stimulation of the NF-B translocation, while in the presence of the TNF␣ stimulus, the same 50% of NF-B translocation was seen. Titration of the antiserum down re-

FIG. 9. Inhibition of NF-B translocation by IL-1RA and anti-TNF␣ antibody in chondrocytes (q) and HeLa cells (OE).
Top, cells were stimulated with 10 ng/ml IL-1␣ for 20 min at 37°C in the presence of the indicated concentration of IL-1RA. Bottom, cells were stimulated for 20 min at 37°C with 10 ng/ml TNF␣ that had been preincubated 30 min at room temperature with the indicated concentration of a neutralizing anti-TNF␣ antibody. Nuc Cyt, nuclear minus cytoplasmic. sulted in either an increasing loss of NF-B translocation (without TNF␣ stimulation) or a comparable increase in NF-B translocation (samples with TNF␣ present ; Fig. 11E). The IC 50 values were the same for stimulation and inhibition, suggesting that the same antibodies that could block a TNF␣-induced activation could on their own activate the cells.
Effect of Proteasome and Kinase Inhibitors on NF-B Translocation-Because proteasome inhibitors prevent IB␣ degradation (16,33), the potent proteasome inhibitors LLL-H (20) and clastolactacystin ␤-lactone (34) were tested for their potency as NF-B translocation inhibitors in chondrocytes and HeLa cells stimulated with IL-1 and TNF␣. As is shown in Fig.  12, clastolactacystin ␤-lactone gave complete inhibition of translocation with an IC 50 of about 10 M. In contrast, LLL-H only inhibited maximally about 2 ⁄3 of the total translocation, and this occurred with an IC 50 of about 2 M. The only partial inhibitory effect of LLL-H on NF-B translocation was also seen correspondingly on IB␣ degradation; concentrations as high as 25 M LLL-H only delayed rather than blocked IB␣ degradation (data not shown; see Ref. 28).
The nonspecific kinase inhibitor K252b has been shown to inhibit the phosphorylation of both the IL-1 receptor association kinase and IB␣ (28). The compound K252a (which is more cell-permeable than K252b) was tested for its effects on NF-B translocation in chondrocytes and HeLa cells stimulated with IL-1 or TNF␣ and analyzed by the ArrayScan™. As is shown in Fig. 13A, the IC 50 for both cell types and both stimuli in the ArrayScan™ assay was about 0.4 M. To compare the inhibition of a compound such as K252a in the ArrayScan™ assay to its effects in a traditional assay measuring inhibition of an NF-B-up-regulated and -secreted protein, the effects of K252a on the IL-1-and TNF␣-induced production of IL-6 in MRC-5 cells were determined in an overnight assay. This normal human fibroblast MRC-5 cell line is highly sensitive to both cytokines, IB␣ is rapidly phosphorylated and degraded (28), and the MRC-5 cells behaved comparably with chondrocytes when tested in the ArrayScan™ technique (data not shown), factors making them an excellent choice for comparison with chondrocytes and HeLa cells. K252a was found to inhibit the MRC-5 cells at an IC 50 of approximately 0.2-0.3 M (Fig. 13B), about the same as the chondrocytes and HeLa cells (Fig. 13A). DISCUSSION In the present paper, we describe a new technique for quantitation of early events in IL-1 and TNF␣ signal transduction in intact cells by measuring translocation of NF-B from cytoplasm to the nucleus. Cellular analysis of NF-B translocation is performed within 20 min following the synchronized stimulation of cells with IL-1 and TNF␣. This rapid response focuses on just the process leading up to the appearance of NF-B in the nucleus prior to any nuclear binding and expression of its transcription activity. The assay thus reduces the number of postreceptor, intracellular signaling steps affecting the measurement to those of the kinase and adaptor protein cascade producing IB␣ phosphorylation, the IB␣ destruction by proteasomes, and the unknown factors enabling nuclear entry of NF-B. As is shown in this report, ArrayScan™ analysis of NF-B translocation is highly quantitative, sensitive, and reproducible. The low coefficient of variation (6.8%) for repeatedly scanning the same wells (but different fields of cells) indicates the ability of the system to extract statistically similar data from a number of fields in the same well. The analysis of variance indicated that this nuclear translocation technique was robust and produced only a minimal variation between rows and between plates regardless of whether they were analyzed on the same or successive days. Assay, plate, and day variability (Table II) were similar to those obtained for other cell-based assay formats currently used in compound screening. ArrayScan™ analysis is also rapid, taking less than 1 h to scan 75 cells/well in a 96-well plate. Furthermore, determination of protein distribution can be made on native, untransfected cells using fixation and staining such as that described here, or dynamic analysis can be performed in living cells in which proteins of interest are labeled with a fluorescent probe such as green fluorescent protein attached to receptors like the glucocorticoid receptor (26,35) or to other transcription factors such as NF-AT (25).
Previously, NF-B translocation was chiefly identified indirectly by electrophoretic mobility shift assays where measurement could only occur following subsequent NF-B binding to DNA, and quantitation could only be made following cellular lysis, fractionation, and separation by gel electrophoresis (5,7,22). Such a technique is tedious, multistep, and subject to artifacts of protein partitioning and separation (36). Analysis with the present ArrayScan™ technique preserves the spatial distribution of proteins within the cell without disruption. Electrophoretic mobility shift assay measurement is also subject to the influences on other components of the transcription apparatus that are necessary for NF-B binding to the DNA (37). Hence, results of electrophoretic mobility shift assay analysis may reflect the influences of a number of signaling pathways rather than that of NF-B alone. On the other hand, with the ArrayScan™ measurement of NF-B at an earlier stage in signaling, the comparison of results by ArrayScan™ analysis with that of measurement of transcriptional activity of NF-B may enable a clearer understanding of those events activating NF-B in the nucleus such as phosphorylation (38).
As is shown here, IL-1-or TNF␣-stimulated nuclear NF-B translocation begins within 2-4 min following the phosphorylation of IB␣, which is initiated 1-2 min after stimulation (see also Ref. 28). The half-time for nuclear NF-B translocation occurs at about 7-12 min, depending on the cell type, and occurs about 4 -5 min following IB␣ degradation. This time course is comparable for both IL-1 and TNF␣ activation despite the differences in the nature of the receptors and the different adaptor proteins involved (see Ref. 18). The NF-B translocation occurs gradually and uniformly in the entire cell population during the course of the cytokine stimulation. Only a small amount of the entire pool of NF-B in the cytoplasm translocates despite the proteolysis of essentially all of the associated IB␣; the cytoplasmic fluorescence decreases by only 20% following stimulation (Fig. 5). This observation is comparable with earlier observations in which only 10 -20% of the cytoplasmic NF-B was observed to translocate (11). One might consider an alternative scenario producing the same average nuclear/cytoplasmic difference in which say only 50% of the cells responded but where the amount of the NF-B translocation in those cells was twice as much. If this were the case, however, then the S.D. in the nuclear/cytoplasmic difference values would be increased over the course of the translocation. This increase is not seen; the S.D. between cells is constant during the entire translocation (data not shown).
Both IL-1 and TNF␣ produce a maximal response when incubated at about 10 ng/ml (0.6 nM), but they show substantial differences in the concentrations for half-maximal response; whereas IL-1 titrates over about 1.5 logs with an ED 50 of about 2 ng/ml, the TNF␣ titrates over 3 logs with an ED 50 of about 0.15 ng/ml (Fig. 8). These titrations in NF-B nuclear translo- cation are comparable with other cellular responses induced by both IL-1 (see Fig. 13) and TNF␣ (see Ref. 39) and are similar to that observed for cytokine receptor binding for both IL-1 (40, 41) 2 and TNF␣ (42). Precisely why TNF␣ is effective in receptor ligation and cell stimulation over such an extended concentration range is unknown but is perhaps a function of its trimeric nature (43)(44)(45). The antagonism of the IL-1 response by IL-1RA as measured in the ArrayScan™ faithfully replicates its antagonism of IL-1 receptor binding, where in both cases roughly a 100-fold excess of IL-1RA is needed for antagonism. The agonism induced by the aggregation of the TNFR1 molecules by the polyclonal antiTNFR1 mimics the TNF␣ response as reported earlier (39), but quantitation by the ArrayScan™ shows that this agonism is only partial and is mixed with partial antagonism.
In this paper, we discuss the effects of inhibitors of NF-B translocation at two measurable stages following receptor activation: at the level of the kinase signaling cascade and at the level of proteosome involvement in IB␣ destruction and liberation of NF-B. The nonspecific kinase inhibitor K252a (46) has been shown to inhibit NF-B activation (47), and the closely related compound K252b inhibits both the IL-1 receptor association kinase and the phosphorylation of IB␣ (28). As is shown in Fig. 13, K252a inhibits TNF␣ signaling equally well as IL-1 signaling, indicating that it must inhibit other kinases in the NF-B activation cascade that are common to both stimulators. As expected from the necessary role for proteasomes in the degradation of phospho-IB␣ prior to NF-B nuclear translocation, proteasome inhibitors such as LLL-H (20) and lactacystin (48) equally prevent both IL-1 and TNF␣-induced cell activation (Fig. 12). Because of the ease of quantitation of the NF-B translocation in the ArrayScan™, however, subtle differences between the two proteasome inhibitors can be seen such that LLL-H gives only partial inhibition of NF-B translocation, while the lactacystin analog clastolactacystin ␤-lactone gives complete inhibition in both HeLa and chondrocyte cells. This suggests that there may be at least two subsets of proteasomes operating in these cells with only one of them inhibited by LLL-H. Such an interpretation is supported by recent evidence that cells resistant to LLL-based proteasome inhibitors, lacking enzymatic activity associated with this class of proteasomes, can still process substrates such as peptides for assembly of class I histocompatibility antigens. These resistant cells contain a different high molecular weight proteasome-like complex with a different substrate specificity (49).