- Email: [email protected]
Pneumococcal cell wall activates NF-κB in human monocytes: aspects distinct from endotoxin
Microbial Pathogenesis 1996; 20: 309–317
Pneumococcal cell wall activates NF-jB in human monocytes: aspects distinct from endotoxin Barbara Spellerberg,1 Carsten Rosenow, William Sha,2 and Elaine I. Tuomanen1∗ 1
Laboratory of Molecular Infectious Diseases, 1230 York Avenue, The Rockefeller University, New York, NY 10021, U.S.A., 2Department of Biology, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, U.S.A.
(Received September 20, 1995; accepted in revised form December 22, 1995)
Spellerberg, B. (Laboratory of Molecular Infectious Diseases, 1230 York Avenue, The Rockefeller University, New York, NY 10021, U.S.A.), C. Rosenow, W. Sha and E. I. Tuomanen. Pneumococcal cell wall activates NF-jB in human monocytes: aspects distinct from endotoxin. Microbial Pathogenesis 1996; 20: 309–317. The transcription factor NF-jB plays a central role in inflammation by controlling the transcription of multiple genes which participate in the acute phase response. Mice with a targetted disruption of the p50 subunit of NF-jB are hyper-susceptible to challenge with pneumococci but not endotoxin. We sought to clarify the role of NF-jB in the host response to the critical inflammatory component of pneumococci, the cell wall. Activation of NF-jB was monitored by expression of luciferase from cells transfected with an NF-jB dependent luciferase reporter construct. 70Z/3 murine pre-B cells and U937 human monocytes failed to produce luciferase in response to 107 pneumococci or 10 lg cell wall; strong responses were obtained with 10 lg of LPS. In contrast, THP-1 human monocytes showed strong luciferase production with all three stimuli: LPS, intact pneumococci and cell wall. The response was time and dose dependent. Cell wall activity was retained despite alteration of the choline of the teichoic acid or protease treatment suggesting the glycopeptide backbone to be a critical determinant of bioactivity. We conclude that activation of NF-jB by pneumococci is restricted to certain cells and that this proinflammatory activity may be a specific feature of the pneumococcal cell wall glycopeptide backbone. 1996 Academic Press Limited
Key words: NF-jB, pneumococcus, acute phase response, cell wall, Gram positive bacteria.
Introduction The heterodimeric transcription factor NF-jB is involved in the transcriptional control of multiple genes that participate in inflammation, lymphocyte activation, and the acute phase response. Originally identified in B-lymphocytes as a nuclear factor that bound to a specific DNA fragment of the j-light chain enhancer, NFjB has been demonstrated in many mammalian cells.1–4 In its inactive, cytosolic form, NF-jB is bound to the inhibitor IjB. Dissociation from the inhibitor, nuclear translocation and resultant transcription of target genes can occur in response to a variety of inflammatory stimuli including bacterial endotoxin, cytokines and engagement of antigen receptors. Cellular target genes include the cytokines IL1b, IL-2, IL-6 and TNF-a, MHC surface molecules and the Ig j-light chain. ∗ Corresponding author. 0882–4010/96/050309+09 $18.00/0
1996 Academic Press Limited
310
B. Spellerberg et al.
The acute phase response is initiated in response to the presence of bacterial components in a normally sterile body site. In the case of endotoxin, the transcription factor NF-jB plays a central role in generating the acute phase response.5 However, the cascade that initiates inflammation in response to Gram positive bacteria is much less well defined than its endotoxin-induced counterpart. Studies initiated a decade ago using highly defined subfractions of various Gram positive pathogens uniformly suggested that inflammatory capability was attributable to the Gram positive cell wall.6–11 The wall is a macromolecule consisting of a peptidoglycan network decorated with teichoic acids.12 The major building blocks of the peptidoglycan for all Gram positive bacteria are a disaccharide tetrapeptide (N-acetylglucosaminyl-N-acetylmuramyl-L-alanyl-Dglutaminyl-L-Lysyl-D-alanine, >30% of the total wall) and its dimer (another >30% of the total wall). However, the remaining components of the wall are structurally variable from bacteria to bacteria creating a unique composition for each pathogen. Differences in the inflammatory capabilites of various bacterial cell walls have been attributed to these structural differences.9,13 Several observations suggest that substantial differences exist in the acute phase response to the major surface components on the two classes of bacteria, endotoxin and cell wall. Cells from C3H HeJ mice are resistant to activation by endotoxin but remain sensitive to cell walls.14 Similarly, mice lacking the receptor for tumor necrosis factor (TNF)15 or intracellular adhesion molecule-1 (ICAM-1)16 succumb to Gram positive infection but are resistant to Gram negative sepsis. Conversely, gene targetting of the p50 subunit of NF-jB resulted in mice hypersusceptible to pneumococcal sepsis but not that due to several Gram negative pathogens.17 This latter finding suggests that the acute phase responses to cell walls and endotoxin might be distinguishable at the level of the transcriptional activator NF-jB. This study sought to characterize the activation of NF-jB in monocytes in response to cell wall from the major Gram positive pathogen, Streptococcus pneumoniae.
Results Activation of NF-jB by S. pneumoniae cell wall Induction of NF-jB/Rel complexes by pneumococci was demonstrated in primary murine macrophages using an EMSA of nuclear extracts (Fig. 1). To further quantitate and characterize this response, the human monocyte cell line THP-1 was transfected with a luciferase reporter gene construct and then challenged with S. pneumoniae and its purified cell wall. Transfected cells showed a strong induction of NF-jB-linked luciferase activity at 4 h compared to saline controls upon stimulation with endotoxin, purified pneumococcal cell wall and intact bacteria (Fig. 2). The response to cell wall demonstrated a time dependence with values well above controls at 3 and 4 h and a peak at 6 h (Fig. 3). The response of THP-1 cells to pneumococcal cell wall was dose dependent with a threshold concentration for activation of 10 ng/ml, an amount equivalent to 105 bacteria/ml. Responses were quantitatively similar between LPS and cell wall when compared on a bacterial cell equivalent basis. The amount of cell wall equivalent to 108 bacteria/ml (10 lg/ml) elicited an NF-jB response similar in magnitude to the amount of endotoxin from the same number of bacteria (1 lg/ ml) (Fig. 2). Incubation with polymixin B decreased the activity of endotoxin by >90% while activity was retained in cell wall preparations.
Pneumococcal cell wall activates NF-jB in human monocytes
311
Fig. 1. Induction of NF-jB/Rel complexes by LPS and S. pneumoniae in primary resting peritoneal macrophages (RPMs). Electrophoretic mobility shift assay (EMSA) of 5 lg of nuclear extracts from RPMs prepared in parallel after 4 h in media alone (solid line) or treatment with LPS (100 ng/ml) or S. pneumoniae strain R6 (5×106/ml). The arrow indicates the location of specific NF-jB complexes.
Both of the major subcomponents of the Gram positive cell wall, the peptidoglycan and the teichoic acid, have been shown to strongly induce cytokine production in human monocytes.8,18 This activity of the teichoic acid is known to be critically dependent on the presence of phosphorylcholine since substitution of this constituent with ethanolamine greatly reduces the bioactivity of pneumococcal cell wall preparations.8 Incubation of THP-1 cells with 1 or 10 lg/ ml of ethanolamine-containing cell wall triggered high levels of luciferase production at least equivalent to that of the choline-containing cell wall, suggesting that the activation of NF-jB is not dependent on this component of the teichoic acid (Table 1). Gram negative peptidoglycan shares many structural features of the non-teichoicated portion of the cell wall of Gram positive bacteria. Both contain alternating N-acetylglucosamine and N-acetylmuramic acid in the glycan backbone but Gram positive bacteria harbor stem peptides of L-alanyl-Dglutaminyl-L-lysyl-D-alanyl-D-alanine while Gram negative bacteria substitute meso-diaminopimelic acid for the lysine. Gram positive walls are also thicker and more highly crosslinked. These differences appeared to be important for induction of NF-jB in that peptidoglycan preparations from H. influenzae in which endotoxin was neutralized with polymixin B failed to induce NF-jB (Table 1).
B. Spellerberg et al.
312
Index of light units
2.0
1.5
1.0
0.5
Pn cell wall (µg/ml)
Pn (log)
1 0. 0. 1 01
20 10
1 0. 0. 1 0 0. 1 00 1
20 10
sa lin
e 7. 5 7
0
LPS (µg/ml)
Fig. 2. Activation of NF-jB in human THP-1 monocytes after stimulation by pneumococcal cell wall. THP-1 cells (106) were exposed to endotoxin (LPS), pneumococcal cell wall (Pn cell wall), intact S. pneumoniae (Pn), or saline. Activation of NF-jB was quantitated at 4 h by measuring luciferase activity arising from the NF-jB luciferase reporter construct. Values (mean±standard deviation; n=3) are expressed in relation to the activity generated by 1 lg/ml endotoxin arbitrarily set to 1.
1.0
Index of light units
0.8
0.5
0.2
0 3
4
5
6
8
24
Hours Fig. 3. Time course of activation of NF-jB in human THP-1 monocytes after stimulation by pneumococcal cell wall. THP-1 cells (106) were exposed to pneumococcal cell wall (1 lg/ml). Activation of NF-jB was quantitated at the times indicated by measuring luciferase activity arising from the NFjB luciferase reporter construct. Values (mean±standard deviation; n=3) are expressed in relation to the activity generated by 1 lg/ml endotoxin at 4 h arbitrarily set to 1. Control value is indicated by the horizontal line (0.02±0.03).
Pneumococcal cell wall activates NF-jB in human monocytes
313
Table 1 Effect of variation of cell wall structure on NF-jB stimulation Stimulus (10 lg/ml)a
Index of light units
Pn wall Pn ethanolamine wall Hi wall with LPS Hi wall LPS Tetra
1.1±0.4 1.6±0.5 1.3±0.4 0.2±0.1 0.9±0.3 0.1±0.1
a THP-1 cells were exposed to the indicated stimulus for 4 h. Quantitation of activation was determined by luciferase activity standardized to a value of 1 light unit for 1 lg/ml endotoxin. Pn: pneumococcal; Hi: H. influenzae; LPS: endotoxin; Tetra: disaccharide tetrapeptide component of Hi peptidoglycan. Hi wall is composed of the Hi wall+LPS preparation incubated with polymixin B (4 lg/ml) so as to neutralize endotoxin.
60
50
pg/ml
40
30
20
10
0
2
4
6
8
24
Hours Fig. 4. Cytokine production by THP-1 cells following stimulation of NF-jB by cell walls. THP-1 cells were stimulated with 1 lg of cell wall and the supernatant fluid was harvested at the indicated times and analysed by ELISA for the presence of TNF (Φ) or IL-1b (Χ). Data is representative of two experiments with duplicate samples. The threshold of detection was 10 pg/ml for TNF and 5 pg/ml for IL-1b.
Further differences in the response of the THP-1 cells to cell wall and endotoxin were noted in experiments in the presence of serum. Cell walls given at 0.1 or 1 lg/ml elicited only slightly greater luciferase activity in the presence of serum (114±15 and 128±17% respectively). Values for endotoxin (0.03 and 1 lg/ml) were increased 154±22 and 228±29% respectively.
Cytokine profile arising from NF-jB stimulation by cell walls THP-1 cells secreted both TNF and IL-1b following stimulation with cell walls (Fig. 4). When compared with 1 lg endotoxin, 1 lg of cell wall induced 10 times less
B. Spellerberg et al.
314
Index of light units
1.5
1.0
0.5
0
100 10 1 Hib LPS
100 10 1 Hib cell wall
100 10 1 PN cell wall
(µg/ml) Fig. 5. Failure of NF-jB stimulation by cell walls in undifferentiated cell lines. U937 [human premonocytic (solid bars)] and 70Z/3 [mouse pre-B (hatched bars)] cells bearing the NF-jB luciferase reporter construct (106/ml/well) were incubated with pneumococcal (Pn) cell wall, H. influenzae (Hi) endotoxin/cell wall complex (LPS) and Hi cell wall alone. Endotoxin activity was neutralized in all cell wall preparations by preincubation with polymixin B (4 lg/ml). Luciferase activity was determined at 4 h. Values are expressed in relation to the activity generated by 1 lg/ml E. coli endotoxin arbitrarily set to 1 (representative of three experiments).
TNF (350–400 pg/ml vs 35–40 pg/ml respectively at 4 and 6 h) and half the amount of IL-1b (95 pg/ml vs 45 pg/ml respectively at 8 h). The relatively poor ability of cell wall to induce TNF as compared to IL-1 has been demonstrated previously.8
Differential induction of NF-jB by cell walls in different cell types To investigate if a variety of undifferentiated monocytes undergo NF-jB activation in response to Gram positive cell walls, the monocytoid cell line U937 and the pre-B murine cell line 70Z/3 were transfected with the NF-jB luciferase reporter construct. In contrast to the THP-1 cells, challenge of 70Z/3 cells with S. pneumoniae purified cell wall failed to generate increased luciferase levels (Fig. 5). U937 cells demonstrated a very low but detectable activity which was not enhanced even after extension of the incubation period from 4 to 8 h. Endotoxin and the endotoxin/cell wall complex from H. influenzae induced several fold increases in luciferase production in both types of transfected cells (Fig. 5). This activity was greatly reduced after neutralization of endotoxin by preincubation with polymixin B, suggesting the cell wall was only minimally active. Activity slightly above saline control levels (0.15 light units) required 100 lg/ ml of the polymixin B treated preparation, supporting the finding that Gram negative-derived peptidoglycan components were not able to activate NF-jB, similar to the results for the THP-1 cells. To test if bacterial peptidoglycan could prevent the activation of NF-jB, U937 cells were simultaneously incubated with 10 lg/ml of pneumococcal cell wall and known activators of NF-jB, of endotoxin (0.2 or 20 lg/ml) or H. influenzae cell surface extract (0.1 or 10 lg/ml). High levels of luciferase production could be
Pneumococcal cell wall activates NF-jB in human monocytes
315
Table 2 Activation of NF-jB by bacterial surface components Bacterial component Gram positive bacteria Teichoic acid Peptidoglycan (Lys)a Gram negative bacteria Endotoxin Peptidoglycan (Dap)a
NF-jB activation
No Yes Yes No
a Difference in the stem peptide of Gram positive and Gram negative peptidoglycans is in position three: lysine (Lys) vs diaminopimelic acid (Dap).
observed in all cases (data not shown) leaving no indication that NF-jB activation was negatively affected by the presence of peptidoglycan components.
Conclusions We sought to determine whether Gram positive bacteria could activate NF-jB and, if so, could the cell wall engender this activity. The enhanced mortality of pneumococcal sepsis in mice with a targetted genetic disruption of the p50 subunit of NF-jB suggested that the acute phase response required to maintain survival in the face of Gram positive infection is critically dependent on NF-jB.17 Our results demonstrate directly that pneumococci activate NF-jB in undifferentiated human and mature murine monocytes. However, unlike the NF-jB response to endotoxin, this response was restricted to only some monocytic cell lines. The NF-jB response to intact pneumococci was quantitatively matched by purified cell wall, the component of the pneumococcal surface previously determined to cause the host inflammatory response. The specific activity of pneumococcal cell wall was equivalent to endotoxin when compared on a bacterial cell equivalent basis. The cell wall is a heteropolymer of teichoic acid and peptidoglycan and different pro-inflammatory activities have been ascribed to structurally distinct subcomponents of the cell wall complex.10,13 For instance, the presence of teichoic acid, a component restricted to Gram positive bacteria, greatly augments the ability of cell walls to induce cytokine secretion. In particular, the phosphorylcholine determinant of the pneumococcal teichoic acid (Cpolysaccharide) strongly induces IL-1b secretion from macrophages.8 Yet, when purified cell wall components were compared for the ability to induce NF-jB, phosphorylcholine-containing teichoic acid was not required for a response as evidenced by the retention of full activity by cell walls containing biosynthetically modified teichoic acid. This suggested that the ability to activate NF-jB might be a property of the peptidoglycan fraction of the pneumococcal cell wall. Confirmation of this hypothesis awaits the ability to purify teichoic acid free disaccharide peptides from Gram positive peptidoglycan. However, polymeric Gram negative peptidoglycan was not able to induce NF-jB suggesting either the composition of the glycopeptide or its thickness and crosslinking are structural constraints on this inflammatory activity. These results suggest the hypothesis that Gram positive bacteria harbor two classes of inflammatory surface components which differ in the requirement for the participation of NF-jB in the induction of the host response (Table 2). The
316
B. Spellerberg et al.
strong inflammatory response attributed to the phosphorylcholine in the teichoic acid may arise without directly activating NF-jB. This latter activity may result from the ability of the phosphorylcholine to bind to the G-protein coupled, platelet activating factor receptor.19,20 Conversely, Gram positive cell wall strongly induces activation of NF-jB independent of the phosphorylcholine in the teichoic acid indicating the activity may reside in the peptidoglycan. If so, the ability to induce NF-jB is apparently not shared by the variant of peptidoglycan which characterizes the cell walls of Gram negative bacteria. These differences may contribute to the different patterns of participation of acute phase reactants and cytokines in Gram positive vs Gram negative infection.
Materials and methods Bacterial components. Streptococcus pneumoniae strain R6 (unencapsulated derivative of Type 2) was grown to midlogarithmic phase in semi-synthetic medium.10 To obtain a preparation of cells for experiments to be reproduced over days, bacteria were incubated in a 60°C waterbath for 10 min to inactivate the autolytic enzyme. This has been shown to result in a highly inflammatory, readily reproducible bacterial preparation.10 Bacteria were washed and resuspended in phosphate buffered saline (PBS) for exposure to cells. Purified cell wall was prepared as described elsewhere.10 Briefly, crude cell wall was extracted in 5% sodium dodecyl sulfate (SDS) at 100°C for 30 min. After the detergent was removed by washing, the cell wall preparation was treated with DNase, RNase, and trypsin. Purified cell wall was reprecipitated in SDS, washed, lyophilized, and stored at −20°C. Ethanolaminecontaining cell wall was prepared in the same manner except that pneumococci were grown in medium with ethanolamine substituting for choline and the lack of bioactivity of this preparation was demonstrated in a pneumococcal adherence assay with endothelial cells (data not shown).10,20 H. influenzae muropeptides and cell surface extracts were prepared as previously described.3 Briefly, crude cell wall was precipitated in boiling 10% SDS, digested with alpha amylase and pronase and reprecipitated in boiling SDS. This material, containing peptidoglycan and endotoxin, was lyophilized and stored at −20°C. Prior to incubation with eucaryotic cells, the peptidoglycan/endotoxin preparation was dissolved in PBS and sonicated. Endotoxin from E. coli (serotype 26:B6) and polymixin B sulfate were purchased from Sigma (St. Louis, MO). Transfection procedures and reporter expression assays. The murine pre B cell line 70Z/ 3 and the human monocytic cell lines U937 and THP-1 (American Type Culture Collection, Rockville, MD) were cultured as described elsewhere.21 Activation of NF-jB was monitored by expression of luciferase from U937, 70Z/3 or THP-1 cells transfected with the pBIIX reporter plasmid in which transcription of the Photinus pyralis luciferase gene is stimulated by two copies of the murine Igj NF-jB binding site in front of a minimal c fos promoter.21 Transfection of 2–4×107 cells per experiment was done using the DEAE dextran method. After transfection, the cells were transferred to 24 well Falcon plates at a density of 106/ well. Twenty-four hours later, cells were stimulated with bacterial components and luciferase activity was measured by a luminometer with the Luciferase Assay System (Promega, Madison, WI). Cytokine determination. The supernatant of transfected human cells was harvested after incubation with bacterial components and stored at −70°C. Cytokine determinations were performed with the Quantikine ELISA kits (R&D Systems) for IL-1b and TNF-a according to the manufacturer’s instructions. Electrophoretic mobility shift assays (EMSA). Resting peritoneal macrophages (RPM) pooled from six specific-pathogen-free mice were isolated by peritoneal lavage with 5 ml of ice-cold phosphate buffered saline. Isolated cells were washed and >107 cells plated per condition in 60 mm plates. After several hours at 37°C, non-adherent cells were removed by two washes with warm saline, and adherent cells were incubated for 4 h in 3 ml of media as follows: media alone, LPS (100 ng/ml) or heat killed S. pneumoniae strain R6
Pneumococcal cell wall activates NF-jB in human monocytes
317
(5×106/ml). Preparation of nuclear and cytoplasmic extracts and EMSA analysis (5 lg of nuclear extract) were performed as previously described.22 A loop oligonucleotide probe containing the class I MHC jB site was used.23
References 1. Grilli MJC, Lenardo M. NF-kappa B and Rel: participants in a multiform transcriptional regulatory system. Int Rev Cyt 1993; 143: 1–62. 2. Nolan G, Baltimore D. The inhibitory ankyrin and activator Rel proteins. Curr Opin Gen Devel 1992; 2: 211–20. 3. Sen R, Baltimore D. Multiple nuclear factors interact with the immunoglobulin enhancer sequences. Cell 1986; 46: 705–16. 4. Sen R, Baltimore D. Inducibility of kappa immunoglobulin enhancer-binding protein NF-kappa B by a post-translational mechanism. Cell 1986; 47: 921–8. 5. Muller J, Ziegler-Heitbrock H, Baeuerle P. Nuclear factor kappa B, a mediator of lipopolysaccharide effects. Immunobiology 1993; 187: 233–56. 6. Geelen S, Battacharyya C, Tuomanen E. Cell wall mediates pneumococcal attachment and cytopathology to human endothelial cells. Infect Immun 1993; 61: 1538–43. 7. Geelen S, Bhattacharyya C, Tuomanen E. Induction of procoagulant activity on human endothelial cells by Streptococcus pneumoniae. Infect Immun 1992; 60: 4179–83. 8. Riesenfeld-Orn I, Wolpe S, Garcia-Bustos JF, Hoffman MK, Tuomanen E. Production of interleukin1 but not tumor necrosis factor by human monocytes stimulated with pneumococcal cell surface components. Infect Immun 1989; 57: 1890–3. 9. Tuomanen E, Hengstler B, Zak O, Tomasz A. Induction of meningeal inflammation by diverse bacterial cell walls. Eur J Clin Microbiol 1987; 5: 682–4. 10. Tuomanen E, Liu H, Hengstler B, Zak O, Tomasz A. The induction of meningeal inflammation by components of the pneumococcal cell wall. J Infect Dis 1985; 151: 859–68. 11. Tuomanen E, Rich R, Zak O. Induction of pulmonary inflammation by components of the pneumococcal cell surface. Amer Rev Respir Dis 1987; 135: 869–74. 12. Garcia-Bustos J, Tomasz A. A biological price of antibiotic resistance: Major changes in the peptidoglycan structure of penicillin-resistant pneumococci. Proc Natl Acad Sci USA 1990; 87: 5415–19. 13. Burroughs M, Rozdzinski E, Geelen S, Tuomanen E. A structure-activity relationship for induction of meningeal inflammation by muramyl peptides. J Clin Invest 1993; 92: 297–302. 14. Pugin J, Heumann D, Tomasz A, Kravchenki V, Akamatsu Y, Nishijima M, Lauser M, Tobias P, Ulevitch R. CD14 is a pattern recognition receptor. Immunity 1994; 1: 509–16. 15. Pfeffer K, Matsuyama T, Kundig T, Mak T. Mice deficient for the 55 kD tumor necrosis factor receptor are resistant to endotoxic shock, yet succumb to L. monocytogenese infection. Cell 1993; 73: 457–67. 16. Tan T, Smith C, Hawkins E, Mason E, Kaplan S. Hematogenous bacterial meningitis in an ICAM1 deficient infant mouse model. J Infect Dis 1995; 171: 342–9. 17. Sha W, Liou H, Tuomanen E, Baltimore D. Targeted disruption of the p50 subunit of NF-kB leads to multifocal defects in immune responses. Cell 1995; 80: 321–30. 18. Heumann D, Barras C, Severin A, Glauser M, Tomasz A. Gram positive cell walls stimulate synthesis of tumor necrosis factor alpha and interleukin-6 by human monocytes. Infect Immun 1994; 62: 2715–21. 19. Cabellos C, MacIntyre DE, Forrest M, Burroughs M, Prasad S, Tuomanen E. Differing roles of platelet-activating factor during inflammation of the lung and subarachnoid space. J Clin Invest 1992; 90: 612–18. 20. Cundell D, Gerard N, Gerard C, Idanpaan-Heikkila I, Tuomanen E. Streptococcus pneumoniae anchors to activated eukaryotic cells by the receptor for platelet activating factor. Nature 1995; 377: 435–8. 21. Saksela K, Baltimore D. Negative regulation of immunoglobulin kappa light chain gene transcription by a short sequence homologous to the murine B1 repetitive element. Molec Cell Biol 1993; 13: 3698–705. 22. Liou H, Sha W, Scott M, Baltimore D. Sequential induction of NF-kappa B/Rel family proteins during B cell terminal differentiation. Molec Cell Biol 1994; 14: 5349–59. 23. Fugita T, Nolan G, Ghosh S, Baltimore D. Independent modes of transcriptional activation by the p50 and p65 subunits of NF-kappa B. Genes Dev 1992; 6: 775–87.
Pneumococcal cell wall activates NF-jB in human monocytes: aspects distinct from endotoxin Barbara Spellerberg,1 Carsten Rosenow, William Sha,2 and Elaine I. Tuomanen1∗ 1
Laboratory of Molecular Infectious Diseases, 1230 York Avenue, The Rockefeller University, New York, NY 10021, U.S.A., 2Department of Biology, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, U.S.A.
(Received September 20, 1995; accepted in revised form December 22, 1995)
Spellerberg, B. (Laboratory of Molecular Infectious Diseases, 1230 York Avenue, The Rockefeller University, New York, NY 10021, U.S.A.), C. Rosenow, W. Sha and E. I. Tuomanen. Pneumococcal cell wall activates NF-jB in human monocytes: aspects distinct from endotoxin. Microbial Pathogenesis 1996; 20: 309–317. The transcription factor NF-jB plays a central role in inflammation by controlling the transcription of multiple genes which participate in the acute phase response. Mice with a targetted disruption of the p50 subunit of NF-jB are hyper-susceptible to challenge with pneumococci but not endotoxin. We sought to clarify the role of NF-jB in the host response to the critical inflammatory component of pneumococci, the cell wall. Activation of NF-jB was monitored by expression of luciferase from cells transfected with an NF-jB dependent luciferase reporter construct. 70Z/3 murine pre-B cells and U937 human monocytes failed to produce luciferase in response to 107 pneumococci or 10 lg cell wall; strong responses were obtained with 10 lg of LPS. In contrast, THP-1 human monocytes showed strong luciferase production with all three stimuli: LPS, intact pneumococci and cell wall. The response was time and dose dependent. Cell wall activity was retained despite alteration of the choline of the teichoic acid or protease treatment suggesting the glycopeptide backbone to be a critical determinant of bioactivity. We conclude that activation of NF-jB by pneumococci is restricted to certain cells and that this proinflammatory activity may be a specific feature of the pneumococcal cell wall glycopeptide backbone. 1996 Academic Press Limited
Key words: NF-jB, pneumococcus, acute phase response, cell wall, Gram positive bacteria.
Introduction The heterodimeric transcription factor NF-jB is involved in the transcriptional control of multiple genes that participate in inflammation, lymphocyte activation, and the acute phase response. Originally identified in B-lymphocytes as a nuclear factor that bound to a specific DNA fragment of the j-light chain enhancer, NFjB has been demonstrated in many mammalian cells.1–4 In its inactive, cytosolic form, NF-jB is bound to the inhibitor IjB. Dissociation from the inhibitor, nuclear translocation and resultant transcription of target genes can occur in response to a variety of inflammatory stimuli including bacterial endotoxin, cytokines and engagement of antigen receptors. Cellular target genes include the cytokines IL1b, IL-2, IL-6 and TNF-a, MHC surface molecules and the Ig j-light chain. ∗ Corresponding author. 0882–4010/96/050309+09 $18.00/0
1996 Academic Press Limited
310
B. Spellerberg et al.
The acute phase response is initiated in response to the presence of bacterial components in a normally sterile body site. In the case of endotoxin, the transcription factor NF-jB plays a central role in generating the acute phase response.5 However, the cascade that initiates inflammation in response to Gram positive bacteria is much less well defined than its endotoxin-induced counterpart. Studies initiated a decade ago using highly defined subfractions of various Gram positive pathogens uniformly suggested that inflammatory capability was attributable to the Gram positive cell wall.6–11 The wall is a macromolecule consisting of a peptidoglycan network decorated with teichoic acids.12 The major building blocks of the peptidoglycan for all Gram positive bacteria are a disaccharide tetrapeptide (N-acetylglucosaminyl-N-acetylmuramyl-L-alanyl-Dglutaminyl-L-Lysyl-D-alanine, >30% of the total wall) and its dimer (another >30% of the total wall). However, the remaining components of the wall are structurally variable from bacteria to bacteria creating a unique composition for each pathogen. Differences in the inflammatory capabilites of various bacterial cell walls have been attributed to these structural differences.9,13 Several observations suggest that substantial differences exist in the acute phase response to the major surface components on the two classes of bacteria, endotoxin and cell wall. Cells from C3H HeJ mice are resistant to activation by endotoxin but remain sensitive to cell walls.14 Similarly, mice lacking the receptor for tumor necrosis factor (TNF)15 or intracellular adhesion molecule-1 (ICAM-1)16 succumb to Gram positive infection but are resistant to Gram negative sepsis. Conversely, gene targetting of the p50 subunit of NF-jB resulted in mice hypersusceptible to pneumococcal sepsis but not that due to several Gram negative pathogens.17 This latter finding suggests that the acute phase responses to cell walls and endotoxin might be distinguishable at the level of the transcriptional activator NF-jB. This study sought to characterize the activation of NF-jB in monocytes in response to cell wall from the major Gram positive pathogen, Streptococcus pneumoniae.
Results Activation of NF-jB by S. pneumoniae cell wall Induction of NF-jB/Rel complexes by pneumococci was demonstrated in primary murine macrophages using an EMSA of nuclear extracts (Fig. 1). To further quantitate and characterize this response, the human monocyte cell line THP-1 was transfected with a luciferase reporter gene construct and then challenged with S. pneumoniae and its purified cell wall. Transfected cells showed a strong induction of NF-jB-linked luciferase activity at 4 h compared to saline controls upon stimulation with endotoxin, purified pneumococcal cell wall and intact bacteria (Fig. 2). The response to cell wall demonstrated a time dependence with values well above controls at 3 and 4 h and a peak at 6 h (Fig. 3). The response of THP-1 cells to pneumococcal cell wall was dose dependent with a threshold concentration for activation of 10 ng/ml, an amount equivalent to 105 bacteria/ml. Responses were quantitatively similar between LPS and cell wall when compared on a bacterial cell equivalent basis. The amount of cell wall equivalent to 108 bacteria/ml (10 lg/ml) elicited an NF-jB response similar in magnitude to the amount of endotoxin from the same number of bacteria (1 lg/ ml) (Fig. 2). Incubation with polymixin B decreased the activity of endotoxin by >90% while activity was retained in cell wall preparations.
Pneumococcal cell wall activates NF-jB in human monocytes
311
Fig. 1. Induction of NF-jB/Rel complexes by LPS and S. pneumoniae in primary resting peritoneal macrophages (RPMs). Electrophoretic mobility shift assay (EMSA) of 5 lg of nuclear extracts from RPMs prepared in parallel after 4 h in media alone (solid line) or treatment with LPS (100 ng/ml) or S. pneumoniae strain R6 (5×106/ml). The arrow indicates the location of specific NF-jB complexes.
Both of the major subcomponents of the Gram positive cell wall, the peptidoglycan and the teichoic acid, have been shown to strongly induce cytokine production in human monocytes.8,18 This activity of the teichoic acid is known to be critically dependent on the presence of phosphorylcholine since substitution of this constituent with ethanolamine greatly reduces the bioactivity of pneumococcal cell wall preparations.8 Incubation of THP-1 cells with 1 or 10 lg/ ml of ethanolamine-containing cell wall triggered high levels of luciferase production at least equivalent to that of the choline-containing cell wall, suggesting that the activation of NF-jB is not dependent on this component of the teichoic acid (Table 1). Gram negative peptidoglycan shares many structural features of the non-teichoicated portion of the cell wall of Gram positive bacteria. Both contain alternating N-acetylglucosamine and N-acetylmuramic acid in the glycan backbone but Gram positive bacteria harbor stem peptides of L-alanyl-Dglutaminyl-L-lysyl-D-alanyl-D-alanine while Gram negative bacteria substitute meso-diaminopimelic acid for the lysine. Gram positive walls are also thicker and more highly crosslinked. These differences appeared to be important for induction of NF-jB in that peptidoglycan preparations from H. influenzae in which endotoxin was neutralized with polymixin B failed to induce NF-jB (Table 1).
B. Spellerberg et al.
312
Index of light units
2.0
1.5
1.0
0.5
Pn cell wall (µg/ml)
Pn (log)
1 0. 0. 1 01
20 10
1 0. 0. 1 0 0. 1 00 1
20 10
sa lin
e 7. 5 7
0
LPS (µg/ml)
Fig. 2. Activation of NF-jB in human THP-1 monocytes after stimulation by pneumococcal cell wall. THP-1 cells (106) were exposed to endotoxin (LPS), pneumococcal cell wall (Pn cell wall), intact S. pneumoniae (Pn), or saline. Activation of NF-jB was quantitated at 4 h by measuring luciferase activity arising from the NF-jB luciferase reporter construct. Values (mean±standard deviation; n=3) are expressed in relation to the activity generated by 1 lg/ml endotoxin arbitrarily set to 1.
1.0
Index of light units
0.8
0.5
0.2
0 3
4
5
6
8
24
Hours Fig. 3. Time course of activation of NF-jB in human THP-1 monocytes after stimulation by pneumococcal cell wall. THP-1 cells (106) were exposed to pneumococcal cell wall (1 lg/ml). Activation of NF-jB was quantitated at the times indicated by measuring luciferase activity arising from the NFjB luciferase reporter construct. Values (mean±standard deviation; n=3) are expressed in relation to the activity generated by 1 lg/ml endotoxin at 4 h arbitrarily set to 1. Control value is indicated by the horizontal line (0.02±0.03).
Pneumococcal cell wall activates NF-jB in human monocytes
313
Table 1 Effect of variation of cell wall structure on NF-jB stimulation Stimulus (10 lg/ml)a
Index of light units
Pn wall Pn ethanolamine wall Hi wall with LPS Hi wall LPS Tetra
1.1±0.4 1.6±0.5 1.3±0.4 0.2±0.1 0.9±0.3 0.1±0.1
a THP-1 cells were exposed to the indicated stimulus for 4 h. Quantitation of activation was determined by luciferase activity standardized to a value of 1 light unit for 1 lg/ml endotoxin. Pn: pneumococcal; Hi: H. influenzae; LPS: endotoxin; Tetra: disaccharide tetrapeptide component of Hi peptidoglycan. Hi wall is composed of the Hi wall+LPS preparation incubated with polymixin B (4 lg/ml) so as to neutralize endotoxin.
60
50
pg/ml
40
30
20
10
0
2
4
6
8
24
Hours Fig. 4. Cytokine production by THP-1 cells following stimulation of NF-jB by cell walls. THP-1 cells were stimulated with 1 lg of cell wall and the supernatant fluid was harvested at the indicated times and analysed by ELISA for the presence of TNF (Φ) or IL-1b (Χ). Data is representative of two experiments with duplicate samples. The threshold of detection was 10 pg/ml for TNF and 5 pg/ml for IL-1b.
Further differences in the response of the THP-1 cells to cell wall and endotoxin were noted in experiments in the presence of serum. Cell walls given at 0.1 or 1 lg/ml elicited only slightly greater luciferase activity in the presence of serum (114±15 and 128±17% respectively). Values for endotoxin (0.03 and 1 lg/ml) were increased 154±22 and 228±29% respectively.
Cytokine profile arising from NF-jB stimulation by cell walls THP-1 cells secreted both TNF and IL-1b following stimulation with cell walls (Fig. 4). When compared with 1 lg endotoxin, 1 lg of cell wall induced 10 times less
B. Spellerberg et al.
314
Index of light units
1.5
1.0
0.5
0
100 10 1 Hib LPS
100 10 1 Hib cell wall
100 10 1 PN cell wall
(µg/ml) Fig. 5. Failure of NF-jB stimulation by cell walls in undifferentiated cell lines. U937 [human premonocytic (solid bars)] and 70Z/3 [mouse pre-B (hatched bars)] cells bearing the NF-jB luciferase reporter construct (106/ml/well) were incubated with pneumococcal (Pn) cell wall, H. influenzae (Hi) endotoxin/cell wall complex (LPS) and Hi cell wall alone. Endotoxin activity was neutralized in all cell wall preparations by preincubation with polymixin B (4 lg/ml). Luciferase activity was determined at 4 h. Values are expressed in relation to the activity generated by 1 lg/ml E. coli endotoxin arbitrarily set to 1 (representative of three experiments).
TNF (350–400 pg/ml vs 35–40 pg/ml respectively at 4 and 6 h) and half the amount of IL-1b (95 pg/ml vs 45 pg/ml respectively at 8 h). The relatively poor ability of cell wall to induce TNF as compared to IL-1 has been demonstrated previously.8
Differential induction of NF-jB by cell walls in different cell types To investigate if a variety of undifferentiated monocytes undergo NF-jB activation in response to Gram positive cell walls, the monocytoid cell line U937 and the pre-B murine cell line 70Z/3 were transfected with the NF-jB luciferase reporter construct. In contrast to the THP-1 cells, challenge of 70Z/3 cells with S. pneumoniae purified cell wall failed to generate increased luciferase levels (Fig. 5). U937 cells demonstrated a very low but detectable activity which was not enhanced even after extension of the incubation period from 4 to 8 h. Endotoxin and the endotoxin/cell wall complex from H. influenzae induced several fold increases in luciferase production in both types of transfected cells (Fig. 5). This activity was greatly reduced after neutralization of endotoxin by preincubation with polymixin B, suggesting the cell wall was only minimally active. Activity slightly above saline control levels (0.15 light units) required 100 lg/ ml of the polymixin B treated preparation, supporting the finding that Gram negative-derived peptidoglycan components were not able to activate NF-jB, similar to the results for the THP-1 cells. To test if bacterial peptidoglycan could prevent the activation of NF-jB, U937 cells were simultaneously incubated with 10 lg/ml of pneumococcal cell wall and known activators of NF-jB, of endotoxin (0.2 or 20 lg/ml) or H. influenzae cell surface extract (0.1 or 10 lg/ml). High levels of luciferase production could be
Pneumococcal cell wall activates NF-jB in human monocytes
315
Table 2 Activation of NF-jB by bacterial surface components Bacterial component Gram positive bacteria Teichoic acid Peptidoglycan (Lys)a Gram negative bacteria Endotoxin Peptidoglycan (Dap)a
NF-jB activation
No Yes Yes No
a Difference in the stem peptide of Gram positive and Gram negative peptidoglycans is in position three: lysine (Lys) vs diaminopimelic acid (Dap).
observed in all cases (data not shown) leaving no indication that NF-jB activation was negatively affected by the presence of peptidoglycan components.
Conclusions We sought to determine whether Gram positive bacteria could activate NF-jB and, if so, could the cell wall engender this activity. The enhanced mortality of pneumococcal sepsis in mice with a targetted genetic disruption of the p50 subunit of NF-jB suggested that the acute phase response required to maintain survival in the face of Gram positive infection is critically dependent on NF-jB.17 Our results demonstrate directly that pneumococci activate NF-jB in undifferentiated human and mature murine monocytes. However, unlike the NF-jB response to endotoxin, this response was restricted to only some monocytic cell lines. The NF-jB response to intact pneumococci was quantitatively matched by purified cell wall, the component of the pneumococcal surface previously determined to cause the host inflammatory response. The specific activity of pneumococcal cell wall was equivalent to endotoxin when compared on a bacterial cell equivalent basis. The cell wall is a heteropolymer of teichoic acid and peptidoglycan and different pro-inflammatory activities have been ascribed to structurally distinct subcomponents of the cell wall complex.10,13 For instance, the presence of teichoic acid, a component restricted to Gram positive bacteria, greatly augments the ability of cell walls to induce cytokine secretion. In particular, the phosphorylcholine determinant of the pneumococcal teichoic acid (Cpolysaccharide) strongly induces IL-1b secretion from macrophages.8 Yet, when purified cell wall components were compared for the ability to induce NF-jB, phosphorylcholine-containing teichoic acid was not required for a response as evidenced by the retention of full activity by cell walls containing biosynthetically modified teichoic acid. This suggested that the ability to activate NF-jB might be a property of the peptidoglycan fraction of the pneumococcal cell wall. Confirmation of this hypothesis awaits the ability to purify teichoic acid free disaccharide peptides from Gram positive peptidoglycan. However, polymeric Gram negative peptidoglycan was not able to induce NF-jB suggesting either the composition of the glycopeptide or its thickness and crosslinking are structural constraints on this inflammatory activity. These results suggest the hypothesis that Gram positive bacteria harbor two classes of inflammatory surface components which differ in the requirement for the participation of NF-jB in the induction of the host response (Table 2). The
316
B. Spellerberg et al.
strong inflammatory response attributed to the phosphorylcholine in the teichoic acid may arise without directly activating NF-jB. This latter activity may result from the ability of the phosphorylcholine to bind to the G-protein coupled, platelet activating factor receptor.19,20 Conversely, Gram positive cell wall strongly induces activation of NF-jB independent of the phosphorylcholine in the teichoic acid indicating the activity may reside in the peptidoglycan. If so, the ability to induce NF-jB is apparently not shared by the variant of peptidoglycan which characterizes the cell walls of Gram negative bacteria. These differences may contribute to the different patterns of participation of acute phase reactants and cytokines in Gram positive vs Gram negative infection.
Materials and methods Bacterial components. Streptococcus pneumoniae strain R6 (unencapsulated derivative of Type 2) was grown to midlogarithmic phase in semi-synthetic medium.10 To obtain a preparation of cells for experiments to be reproduced over days, bacteria were incubated in a 60°C waterbath for 10 min to inactivate the autolytic enzyme. This has been shown to result in a highly inflammatory, readily reproducible bacterial preparation.10 Bacteria were washed and resuspended in phosphate buffered saline (PBS) for exposure to cells. Purified cell wall was prepared as described elsewhere.10 Briefly, crude cell wall was extracted in 5% sodium dodecyl sulfate (SDS) at 100°C for 30 min. After the detergent was removed by washing, the cell wall preparation was treated with DNase, RNase, and trypsin. Purified cell wall was reprecipitated in SDS, washed, lyophilized, and stored at −20°C. Ethanolaminecontaining cell wall was prepared in the same manner except that pneumococci were grown in medium with ethanolamine substituting for choline and the lack of bioactivity of this preparation was demonstrated in a pneumococcal adherence assay with endothelial cells (data not shown).10,20 H. influenzae muropeptides and cell surface extracts were prepared as previously described.3 Briefly, crude cell wall was precipitated in boiling 10% SDS, digested with alpha amylase and pronase and reprecipitated in boiling SDS. This material, containing peptidoglycan and endotoxin, was lyophilized and stored at −20°C. Prior to incubation with eucaryotic cells, the peptidoglycan/endotoxin preparation was dissolved in PBS and sonicated. Endotoxin from E. coli (serotype 26:B6) and polymixin B sulfate were purchased from Sigma (St. Louis, MO). Transfection procedures and reporter expression assays. The murine pre B cell line 70Z/ 3 and the human monocytic cell lines U937 and THP-1 (American Type Culture Collection, Rockville, MD) were cultured as described elsewhere.21 Activation of NF-jB was monitored by expression of luciferase from U937, 70Z/3 or THP-1 cells transfected with the pBIIX reporter plasmid in which transcription of the Photinus pyralis luciferase gene is stimulated by two copies of the murine Igj NF-jB binding site in front of a minimal c fos promoter.21 Transfection of 2–4×107 cells per experiment was done using the DEAE dextran method. After transfection, the cells were transferred to 24 well Falcon plates at a density of 106/ well. Twenty-four hours later, cells were stimulated with bacterial components and luciferase activity was measured by a luminometer with the Luciferase Assay System (Promega, Madison, WI). Cytokine determination. The supernatant of transfected human cells was harvested after incubation with bacterial components and stored at −70°C. Cytokine determinations were performed with the Quantikine ELISA kits (R&D Systems) for IL-1b and TNF-a according to the manufacturer’s instructions. Electrophoretic mobility shift assays (EMSA). Resting peritoneal macrophages (RPM) pooled from six specific-pathogen-free mice were isolated by peritoneal lavage with 5 ml of ice-cold phosphate buffered saline. Isolated cells were washed and >107 cells plated per condition in 60 mm plates. After several hours at 37°C, non-adherent cells were removed by two washes with warm saline, and adherent cells were incubated for 4 h in 3 ml of media as follows: media alone, LPS (100 ng/ml) or heat killed S. pneumoniae strain R6
Pneumococcal cell wall activates NF-jB in human monocytes
317
(5×106/ml). Preparation of nuclear and cytoplasmic extracts and EMSA analysis (5 lg of nuclear extract) were performed as previously described.22 A loop oligonucleotide probe containing the class I MHC jB site was used.23
References 1. Grilli MJC, Lenardo M. NF-kappa B and Rel: participants in a multiform transcriptional regulatory system. Int Rev Cyt 1993; 143: 1–62. 2. Nolan G, Baltimore D. The inhibitory ankyrin and activator Rel proteins. Curr Opin Gen Devel 1992; 2: 211–20. 3. Sen R, Baltimore D. Multiple nuclear factors interact with the immunoglobulin enhancer sequences. Cell 1986; 46: 705–16. 4. Sen R, Baltimore D. Inducibility of kappa immunoglobulin enhancer-binding protein NF-kappa B by a post-translational mechanism. Cell 1986; 47: 921–8. 5. Muller J, Ziegler-Heitbrock H, Baeuerle P. Nuclear factor kappa B, a mediator of lipopolysaccharide effects. Immunobiology 1993; 187: 233–56. 6. Geelen S, Battacharyya C, Tuomanen E. Cell wall mediates pneumococcal attachment and cytopathology to human endothelial cells. Infect Immun 1993; 61: 1538–43. 7. Geelen S, Bhattacharyya C, Tuomanen E. Induction of procoagulant activity on human endothelial cells by Streptococcus pneumoniae. Infect Immun 1992; 60: 4179–83. 8. Riesenfeld-Orn I, Wolpe S, Garcia-Bustos JF, Hoffman MK, Tuomanen E. Production of interleukin1 but not tumor necrosis factor by human monocytes stimulated with pneumococcal cell surface components. Infect Immun 1989; 57: 1890–3. 9. Tuomanen E, Hengstler B, Zak O, Tomasz A. Induction of meningeal inflammation by diverse bacterial cell walls. Eur J Clin Microbiol 1987; 5: 682–4. 10. Tuomanen E, Liu H, Hengstler B, Zak O, Tomasz A. The induction of meningeal inflammation by components of the pneumococcal cell wall. J Infect Dis 1985; 151: 859–68. 11. Tuomanen E, Rich R, Zak O. Induction of pulmonary inflammation by components of the pneumococcal cell surface. Amer Rev Respir Dis 1987; 135: 869–74. 12. Garcia-Bustos J, Tomasz A. A biological price of antibiotic resistance: Major changes in the peptidoglycan structure of penicillin-resistant pneumococci. Proc Natl Acad Sci USA 1990; 87: 5415–19. 13. Burroughs M, Rozdzinski E, Geelen S, Tuomanen E. A structure-activity relationship for induction of meningeal inflammation by muramyl peptides. J Clin Invest 1993; 92: 297–302. 14. Pugin J, Heumann D, Tomasz A, Kravchenki V, Akamatsu Y, Nishijima M, Lauser M, Tobias P, Ulevitch R. CD14 is a pattern recognition receptor. Immunity 1994; 1: 509–16. 15. Pfeffer K, Matsuyama T, Kundig T, Mak T. Mice deficient for the 55 kD tumor necrosis factor receptor are resistant to endotoxic shock, yet succumb to L. monocytogenese infection. Cell 1993; 73: 457–67. 16. Tan T, Smith C, Hawkins E, Mason E, Kaplan S. Hematogenous bacterial meningitis in an ICAM1 deficient infant mouse model. J Infect Dis 1995; 171: 342–9. 17. Sha W, Liou H, Tuomanen E, Baltimore D. Targeted disruption of the p50 subunit of NF-kB leads to multifocal defects in immune responses. Cell 1995; 80: 321–30. 18. Heumann D, Barras C, Severin A, Glauser M, Tomasz A. Gram positive cell walls stimulate synthesis of tumor necrosis factor alpha and interleukin-6 by human monocytes. Infect Immun 1994; 62: 2715–21. 19. Cabellos C, MacIntyre DE, Forrest M, Burroughs M, Prasad S, Tuomanen E. Differing roles of platelet-activating factor during inflammation of the lung and subarachnoid space. J Clin Invest 1992; 90: 612–18. 20. Cundell D, Gerard N, Gerard C, Idanpaan-Heikkila I, Tuomanen E. Streptococcus pneumoniae anchors to activated eukaryotic cells by the receptor for platelet activating factor. Nature 1995; 377: 435–8. 21. Saksela K, Baltimore D. Negative regulation of immunoglobulin kappa light chain gene transcription by a short sequence homologous to the murine B1 repetitive element. Molec Cell Biol 1993; 13: 3698–705. 22. Liou H, Sha W, Scott M, Baltimore D. Sequential induction of NF-kappa B/Rel family proteins during B cell terminal differentiation. Molec Cell Biol 1994; 14: 5349–59. 23. Fugita T, Nolan G, Ghosh S, Baltimore D. Independent modes of transcriptional activation by the p50 and p65 subunits of NF-kappa B. Genes Dev 1992; 6: 775–87.