Persistent Activation of RelA by Respiratory Syncytial Virus Involves Protein Kinase C, Underphosphorylated IκBβ, and Sequestration of Protein Phosphatase 2A by the Viral Phosphoprotein

Abstract

Respiratory syncytial virus (RSV) activated the RelA (p65) subunit of nuclear factor kappa B (NF-κB) over many hours postinfection. The initial activation coincided with phosphorylation and degradation of IκBα, the cytoplasmic inhibitor of RelA. During persistent activation of NF-κB at later times in infection, syntheses of inhibitors IκBα as well as IκBβ were restored. However, the resynthesized IκBβ was in an underphosphorylated state, which apparently prevented inhibition of NF-κB. Use of specific inhibitors suggested that the pathway leading to the persistent—but not the initial—activation of NF-κB involved signaling through protein kinase C (PKC) and reactive oxygen intermediates of nonmitochondrial origin, whereas phospholipase C or D played little or no role. Thus, RSV infection led to the activation of NF-κB by a biphasic mechanism: a transient or early activation involving phosphorylation of the inhibitor IκB polypeptides, and a persistent or long-term activation requiring PKC and the generation of hypophosphorylated IκBβ. At least a part of the activation was through a novel mechanism in which the viral phosphoprotein P associated with but was not dephosphorylated by protein phosphatase 2A and thus sequestered and inhibited the latter. We postulate that this led to a net increase in the phosphorylation state of signaling proteins that are responsible for RelA activation.

Human respiratory syncytial virus (RSV) is the leading cause of respiratory illness and death in young infants worldwide (3, 29). It is the prototype member of the genus Pneumovirus in the family Paramyxoviridae and contains a nonsegmented negative-strand RNA genome about 15 kb long (6, 17). Due to the profound clinical importance of the virus and the lack of a reliable vaccine, new lines of investigation have placed much emphasis on host-virus interactions in relation to the immunopathology of the infection process. Recent studies have demonstrated the elaboration of a number of cytokines and other immunoregulatory molecules following RSV infection of a variety of susceptible host cells of the respiratory tract. These products include but are not limited to leukotrienes (2), intracellular adhesion molecule-1 (43, 51), major histocompatibility class I molecule (25), soluble tumor necrosis factor (TNF) receptor (5), and a battery of interleukins and chemokines, such as interleukin-1 (IL-1), IL-6, IL-8, IL-10, and IL-11 (11–13, 20–23, 26, 38, 42, 45). In order to address the mechanism underlying the activation of these cytokines, such studies have been extended to RSV infection of defined and established cell lines of lung origin. RSV infection of A549 cells, in particular, has been shown to result in the induction of essentially all of the interleukins mentioned above (13, 23, 25, 26, 38, 43).

Others and we have recently demonstrated that RSV infection leads to the activation of cellular transcription factor NF-κB, which is in turn responsible for transcriptional activation of a number of interleukin promoters (13, 22, 26, 38). RSV infection was shown to induce nuclear translocation of the existing RelA subunit and to a lesser extent the p50 subunit of NF-κB over many hours postinfection (p.i.). A large body of recent literature has established a relatively detailed mechanism of NF-κB induction that can occur in response to various extracellular signals (reviewed in references 10, 37, 48, and 50). In the uninduced cell, NF-κB is retained in the cytoplasm in complex with its inhibitory subunit, IκBα, which is believed to mask the nuclear localization sequence (NLS) of NF-κB. One of the earliest discernible biochemical reactions in the NF-κB activation pathway is the phosphorylation of IκBα by a novel multisubunit kinase complex (27, 37, 50), followed by their degradation, most likely by the ubiquitin-proteasome pathway (15). This leads to nuclear translocation of NF-κB, which then activates a variety of cellular genes, including those of many interleukins and IκB. Following the initial activation, NF-κB therefore produces new rounds of IκBα which restores inhibition, thus generating an autoregulatory loop, which explains the transient induction of NF-κB by signals such as TNF-α and phorbol esters. This mechanism, however, fails to explain the persistent induction of NF-κB by lipopolysaccharides and IL-1, which lasts for many hours following stimulation. Recent studies have suggested a role for another inhibitor, IκBβ, in this process (44, 54). The newly synthesized IκBβ, which was found to be underphosphorylated, was shown to complex with NF-κB; however, unlike IκBα, it apparently did not mask the NLS of NF-κB. Additionally, it prevented IκBα from binding to NF-κB. Thus, the transcriptionally competent NF-κB–IκBβ complex entered the nucleus and functioned essentially like activated NF-κB. Although the kinetic details of the interaction between the two forms of IκB with the various subunits of NF-κB remain to be elucidated, this model offers a plausible mechanism for persistent induction over long periods.

As alluded to earlier, activation of the NF-κB RelA (p65) subunit by RSV was found to be clearly persistent in nature (13, 26) (Fig. 1). Here, we report a potential mechanism for this induction and show that the signal transduction pathway leading to it involves protein kinase C (PKC) enzyme(s) that apparently generates reactive oxygen of potentially nonmitochondrial origin. This in turn leads to phosphorylation and degradation of IκB proteins, and eventual generation of newly synthesized underphosphorylated IκBβ, thus leading to persistent induction. Interestingly, the viral phosphoprotein P alone caused substantial activation of RelA through an apparently novel mechanism. It was not a substrate of cellular phosphatases; nevertheless, it did bind protein phosphatase 2A (PP2A) and thus sequestered, and essentially inhibited, the latter. We propose that the inhibition of PP2A leads to increased phosphorylation of specific cellular proteins, some of which are signaling molecules in the RelA activation pathway.

FIG. 1.

Kinetics of RelA activation by RSV. Infection of A549 cells by RSV, preparation of nuclear (Nucl) and cytoplasmic (Cyto) extracts, and determination of RelA levels by immunoblotting were performed as described in Materials and Methods. At various times p.i., cells were harvested for lysis: lane 1, 0 h; lane 2, 2 h; lane 3, 4 h; lane 4, 6 h; lane 5, 8 h; lane 6, 10 h; lane 7, 20 h; lane 8, 30 h; lane 9, 10 h; lane 10, 30 h. ECL immunoblot analyses of cytosolic and nuclear fractions (as indicated) were carried out with anti-p65 antibody.

MATERIALS AND METHODS

Antibodies and inhibitors.

Rabbit antipeptide antibodies made against epitopes of RelA (p65), IκBα, and IκBβ were purchased from Santa Cruz Biotechnology (Santa Cruz, Calif.). Antibodies against the catalytic subunits of PP1 and PP2A were purchased from Transduction Laboratories (Lexington, Ky.). Calyculin A, rotenone, sodium orthovanadate, sphingosine, and pyrrolidone dithiocarbamate (PDTC) were from Sigma Chemical Co. (St. Louis, Mo.); D609 and U73122 were from BIOMOL Research Laboratories (Plymouth Meeting, Pa.); MG132 (Z-Leu-Leu-Leu-CHO) was from Peptide Institute (Osaka, Japan); and Myr-ψPKC (myr.Arg-Phe-Ala-Arg-Lys-Gly-Ala-Leu-Arg-Gln-Lys-Asn-Val), a pseudosubstrate peptide that is cell permeable due to its myristyl (myr.) group and inhibits PKC, was from Promega Corp. (Madison, Wis.). The lambda phosphatase (PPλ) was purified as described previously (8); the catalytic subunit of PP2A and cyclic AMP-dependent protein kinase (PKA) were purchased from Promega. Stock solutions (200× to 500×) of the inhibitor chemicals (for example, okadaic acid [OA]) were made in water or dimethyl sulfoxide as instructed by the manufacturer. Unless otherwise mentioned, the inhibitors were added to RSV-infected cells at 3 h after addition of the virus and at the final concentrations indicated for each experiment. Anti-RSV antibody was purchased from Chemicon International, Inc. (Temecula, Calif.).

Estimation of RSV growth.

The growth of RSV (Long) in A549 cells in the presence or absence of various drugs was quantitated by either of the following, as and where mentioned: determination of viral titer on HEp-2 monolayers by standard procedures, immunoprecipitation of [³⁵S]methionine-labeled RSV proteins as described elsewhere (7), and immunoblotting (Western blotting) of the total infected cell extract with an anti-RSV antibody.

Measurements of NF-κB and IκB.

RSV (Long) inoculum was grown in HEp-2 cells and purified as described earlier (13). For the NF-κB experiments, monolayers of A549 cells were infected with purified RSV at a multiplicity of infection (MOI) of 3.0. Inhibitor drugs at appropriate concentrations were added 1.5 h after the addition of the virus. The time of addition of the virus was considered as 0 h; at specified times afterward, the infected cells were processed for analysis essentially as described above. Briefly, the infected monolayer (and the uninfected control) was washed twice with ice-cold phosphate-buffered saline. Then 500 μl of lysis buffer (50 mM Tris-Cl [pH 8.0], 50 mM NaCl, 0.1 mM EDTA, 1% Tween 20, 1 mM dithiothreitol, leupeptin, aprotinin, phenylmethylsulfonyl fluoride) was added per 150-cm² T flask, and the mixture was incubated for 10 min at room temperature. The lysed cells were scraped off, and the extract was centrifuged at 2,000 × g for 5 min. The supernatant was further clarified by centrifugation at 15,000 × g for 15 min in cold and was used as cytosolic extract in immunoblot analysis where mentioned. The pellet, containing nuclei, was washed twice with 500 μl of ice-cold lysis buffer. Both fractions were either immediately used in immunoblot analysis or stored frozen at −80°C until use.

Immunoblot analyses were carried out with antibodies against RelA (p65), IκBα, and IκBβ, purchased from Santa Cruz Biotechnology; 15 μg of the nuclear (for RelA) extract or 30 μg of cytosolic extract (for IκB and P proteins) was analyzed on 14% denaturing polyacrylamide gels. To achieve a good separation of the phosphorylated and basal forms of IκBβ, such gels were made in a 10-in.-tall gel apparatus, and electrophoresis was carried out at a constant 100 V overnight. Following electrophoresis, the proteins were electroblotted to Immobilon-P (Millipore) membranes. Blocking and probing of the membranes and development of the bands by enhanced chemiluminescence (ECL; Amersham) were performed essentially as described previously (39).

Electrophoretic mobility shift assays (EMSA) using nuclear extracts (19) of RSV-infected cells and double-stranded ³²P-labeled NF-κB oligonucleotide (5′ GGGGAATTTCCCC 3′) were carried out as described earlier (13).

Functional assays of NF-κB activity were performed in A549 cells transfected with the NF-κB monitor plasmid pBIIxluc followed by infection with RSV and luciferase assays, essentially as described previously (13, 33).

Immunoprecipitation.

A549 cells, grown in 10-cm-diameter dishes to a confluency of 70%, were transfected with pcDNA-3 clones of wild-type or mutant P genes of RSV as described earlier (9). Preparation of cell extracts and immunoprecipitation using either anti-P or anti-PP2A antibodies and protein A-coupled Sepharose beads were performed as described previously (9). The anti-P antibody was raised in rabbits against peptide CSDNPFSKLYKETIETFD (residues 94 to 110 of the P protein of Long strain with a cysteine added to the N terminus) by conjugation with keyhole limpet hemocyanin according to standard procedures.

Phosphatase assays.

Phosphatase activities were assayed in standard 20-μl reaction mixtures containing 50 mM Tris-Cl (pH 7.5), 50 mM NaCl, 5 mM MgCl₂, 5 mM MnCl₂, 100 μM ATP (to reduce the specific activity of any [γ-³²P]ATP that might be carried over from the previous phosphorylation reaction), approximately 10 ng of recombinant human PP2A, and either 20 mM p-nitrophenyl phosphate (pNPP) or 0.5 μg of ³²P-labeled casein or histone as the substrate, essentially as described previously (4). The hydrolysis of pNPP was quantitated by a colorimetric assay; that of casein or histone was quantitated by measuring the liberated inorganic phosphate (4). Casein and histone were phosphorylated by the catalytic subunit of PKA, using [γ-³²P]ATP as the phosphate donor (4), followed by removal of the ATP by filtration chromatography through Sephadex G-50. Where mentioned, a variable amount of the phosphorylated or phosphate-free P protein was also included in the PP2A reaction. Phosphorylated P protein was prepared in a similar manner by phosphorylating the bacterially expressed phosphate-free P protein with casein kinase 2 (CK2) in vitro (4, 9, 39).

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis was carried out by the Laemmli procedure (34) as described earlier (39).

RESULTS

Persistent activation of RelA by RSV.

To investigate the mechanism of induction of NF-κB by RSV, we needed to first determine the kinetics of the induction and the relationship between the cytoplasmic and the nuclear concentrations of NF-κB. To accomplish this, we isolated the nuclei and cytosolic fractions of A549 cells infected with RSV (MOI of 2.0) at different times p.i. and probed them with anti-RelA antibody in immunoblot analysis. Results presented in Fig. 1 demonstrate the following major features of the activation process. (i) An increase in the nuclear RelA was accompanied by a parallel decrease in cytoplasmic RelA concentrations, confirming earlier results (26). Thus, the total quantity of RelA polypeptides in the infected cell remained nearly unchanged during induction, except for a small increase beyond 10 h p.i. (ii) This finding also suggested that at least at the earlier times, activation of RelA probably involved the classic mechanism of translocation of preexisting cytoplasmic pool of RelA into the nucleus, pointing to a posttranslational modification. A continuous turnover of both cytoplasmic and nuclear pools at later times cannot be excluded at this point. (iii) Translocation seemed to begin as early as 6 h p.i. and reached its maximal level by 10 to 12 h under the conditions of this experiment. (iv) Perhaps most interestingly, the elevated nuclear RelA concentration was maintained essentially through the whole phase of intracellular multiplication of the virus, tested up to at least 30 h, which confirmed and extended earlier findings (13) and clearly established the persistent nature of RSV-mediated RelA induction. As expected, translocation of RelA was not seen in A549 cells mock infected with virus-free supernatants.

RSV-mediated phosphorylation and degradation of IκB subunits.

To further elucidate the steps in RSV-mediated NF-κB induction, we decided to determine the quantity as well as the phosphorylation status of the two IκB proteins during induction. Increased phosphorylation of the IκBs has been shown to result in their slower mobility in denaturing polyacrylamide gels (54). Thus, to determine whether phosphorylation of the IκBs is induced by RSV, the total-cell extracts of RSV-infected cells were electrophoresed such that the phosphorylated forms of the IκB proteins were resolved from their nonphosphorylated (or underphosphorylated) forms. The gel was then subjected to standard immunoblot analysis using anti-IκB antibodies as described in Materials and Methods. Results presented in Fig. 2 clearly show that both IκBα and IκBβ were found in uninfected A549 cells. Soon after RSV infection, both IκBs rapidly disappeared (lanes 3 and 4). For IκBα, an intermediate, more slowly migrating form could be detected, suggesting a causal relationship between RSV-induced phosphorylation and degradation. The more highly phosphorylated nature of the more slowly migrating forms of both IκBs could be judged by their disappearance (Fig. 2, lane W) upon treatment with PPλ, a Ser/Thr protein phosphatase of broad specificity (4, 8, 9). The dephosphorylation was not seen in the presence of sodium orthovanadate, a potent inhibitor of PPλ (data not shown), or when a phosphatase-defective mutant of PPλ (Asp20 to Asn [4]) was used (Fig. 2, lane M), demonstrating that the disappearance of the more slowly migrating band upon phosphatase treatment was not due to a nonspecific protease present in the phosphatase preparation. In contrast to IκBα, IκBβ clearly existed in a constitutively phosphorylated state in uninfected cells (Fig. 2B, lane 1). In our gel conditions, we could not detect its additional phosphorylation; thus, further studies are needed to ascertain a possible RSV-induced phosphorylation of IκBβ that would serve as a signal for its subsequent degradation.

FIG. 2.

Time course of phosphorylation and degradation of IκBα (A) and IκBβ (B). ECL immunoblot analyses of total extracts of RSV-infected (and control, mock-infected) A549 cells were carried out with anti-IκBα and -IκBβ antibodies as described in Materials and Methods. The time points of harvest were as in Fig. 1. Cell extracts corresponding to lane 3 of panel A or lane 2 of panel B were incubated with 1 μg of wild-type (lane W) or mutant (lane M) PPλ in the presence of 4 mM MnCl₂ at room temperature for 2 min before being subjected to SDS-PAGE; 20 μg protein was analyzed in each lane, except lane M in panel B, for which 40 μg was used.

Regardless of its exact mechanism, the onset of the degradation of IκBs (about 6 h p.i.) correlated well with the nuclear translocation of RelA, as observed in Fig. 1. At later times in RSV infection, however, much of the IκBα and even more of IκBβ reappeared (about 12 h p.i. onward) and continued to be maintained throughout the rest of the infection, which is in agreement with the fact that transcription of both genes is activated by NF-κB (10, 48). The newly synthesized IκBβ was clearly unphosphorylated, as judged by its faster mobility on SDS-PAGE (Fig. 2B, lanes 4 through 8). Thus, it appears that RSV, like cellular agonists, induces the degradation of the IκBs through a net increase in their phosphorylation (10, 48) and leads to persistent activation NF-κB through resynthesized hypophosphorylated IκBβ (44, 54).

Requirement of PKC and unphosphorylated IκBβ in RSV-mediated activation of NF-κB.

A number of studies have recently attempted to delineate the signal transduction pathway leading to the activation of NF-κB. Briefly, various extracellular signals seemed to activate a PKC or a phospholipase pathway, either of which eventually produced reactive oxygen intermediates (ROIs) as a common intracellular messenger (summarized in references 10 and 48). In support of this view, inhibitors of either PKC or phospholipase C (PLC) inhibited activation of NF-κB in different systems. Compounds that inhibited mitochondrial ROI synthesis, such as amytol and rotenone, also inhibited NF-κB activation. Finally, a great variety of antioxidants were reported to suppress the activation of NF-κB; notable among these were glutathione, 2-mercaptoethanol, dithiocarbamates, and vitamin E (10). To determine whether similar pathways are operative in RSV-mediated activation of NF-κB, we added various inhibitors to RSV-infected A549 cells at 2 h p.i. and, at various times afterward, examined the nuclear translocation of RelA, the ability of nuclear RelA to bind NF-κB DNA elements, and the status of the IκBα and β polypeptides.

To start, we tested the effects of three PLC inhibitors, D609, U73122, and sphingosine. As shown in Fig. 3A, D609 treatment strongly inhibited the NF-κB activity as measured in EMSA in vitro, while U73122 or sphingosine (data not shown) had no effect. None of the PLC inhibitors, including D609, inhibited the nuclear translocation of RelA (Fig. 3B), measured at either 8 or 30 h p.i.; therefore, only the D609 data are shown as representative. The two PKC inhibitors, staurosporine and the pseudosubstrate peptide (Myr-ψPKC), both strongly inhibited DNA-binding activity (Fig. 3A) as well as nuclear translocation (Fig. 3B) of RelA at 30 h p.i. only. Rotenone, an inhibitor of mitochondrial synthesis of reactive oxygen, had very little effect on either the quantity or activity of nuclear RelA. However, the antioxidant PDTC did inhibit RelA activation by RSV at both time points, and so did MG132, a proteasome inhibitor, although the effect of the latter was more modest. Based on these results, it appears that the drugs that inhibited RelA are of two kinds: the PKC inhibitors affected mainly the persistent or long-term activation of RelA, whereas MG132 and PDTC inhibited the persistent as well as the early phases.

FIG. 3.

Effect of signal transduction inhibitors on RSV-mediated activation of NF-κB. The inhibitors were added to RSV-infected A549 cells at the following final concentrations: D609, 20 μM; staurosporine (Stau.), 20 ng/ml; pseudosubstrate peptide (Pept.), 100 μM; PDTC, 40 μM; rotenone (Roten.), 40 μM; MG132, 80 μM; and U73122, 30 μM. Numbers above the lanes indicate the hours p.i. at which the cells were harvested for preparing extracts as described in Materials and Methods. (A) EMSA for NF-κB-binding activity in nuclear extracts; (B to D) immunoblots of nuclear RelA (B), IκBα (C), and IκBβ (D). The phosphorylated and nonphosphorylated forms of IκB polypeptides are indicated. To study the effect of inhibitors on NF-κB activity ex vivo, A549 cells were transfected with plasmid pBIIxluc, using Lipofectamine (Gibco-BRL), and then infected with RSV (or mock infected as a control) essentially as described elsewhere (13). The inhibitors were added at 2 h after addition of the virus. At 20 h thereafter, cells were processed for luciferase assay (13). Luciferase activities, expressed as percentages of the untreated activity (E), represent averages of three measurements.

To explore the mechanism of RelA activation further, the effects of these same inhibitors on the status of the IκB-α and -β polypeptides were tested by using specific antibodies. Results presented in Fig. 3C and D reveal that following the initial degradation at 8 h p.i., substantial quantities of IκBα reappeared by 30 h p.i. in essentially all cases regardless of the nature of the inhibitor. It is notable that in the case of PDTC and MG132, in contrast to the other inhibitors, more IκBα escaped degradation at 8 h p.i., which may explain the inhibitory effect of these two drugs on the early activation of RelA as well (Fig. 3A and B). Rotenone, which did not inhibit RelA activation, was included as a control in these experiments. Based on their slower mobility, the large fraction of undegraded IκBα and IκBβ found in MG132-treated cells was judged to be phosphorylated. As observed in Fig. 3, MG132 was only partially effective in inhibiting the proteolysis of the IκBs, which correlates with its modest inhibitory effect on RelA activation. The most obvious mechanistic commonality of the drugs that did abrogate RelA activation (viz., PDTC, MG132, and the PKC inhibitors) was that they all strongly inhibited the reappearance of IκBβ, thus reinforcing a critical role of the underphosphorylated resynthesized IκBβ in the persistent activation of RelA by RSV.

To determine whether the DNA-binding activities determined in EMSA reflect true NF-κB transcription activity ex vivo, we tested the effect of these drugs on the NF-κB-dependent expression of reporter luciferase enzyme from the plasmid pBIIxluc as described earlier (33). The ex vivo activities of NF-κB under the effect of these drugs (Fig. 3E) closely paralleled the DNA-binding activity in EMSA (Fig. 3A).

Taken together, these results lead to the following immediate conclusions. First, RSV effects the persistent activation of RelA through PKC and reactive oxygen; PLC is not involved in this activation, although a D609-sensitive isozyme may play an indirect role in RelA function. Second, reappearance of IκBα required active NF-κB; D609 and PDTC inhibited both. Finally, persistent activation by RSV generally correlated with the reappearance of IκBβ, which most likely served to protect RelA from the newly synthesized IκBα.

Intracellular expression of RSV phosphoprotein activates RelA.

In a preliminary attempt to identify potential RSV gene products that may be responsible for the activation of NF-κB, we decided to test if intracellular expression of recombinant P protein of RSV might activate RelA. A549 cells were transfected with the available pcDNA3 clone of P (Long strain), and the status of RelA in the transfected cells was determined. The P clone was earlier shown to produce P protein in transfected cells, which was phosphorylated mostly at Ser²³² and to a smaller extent at Ser²³⁷ (9, 39, 46). Results in Fig. 4A show that the recombinant P protein produced a modest but appreciable increase in nuclear RelA protein. The biological activity of RelA was ascertained by two criteria that we have used earlier: EMSA (Fig. 4B) and production of luciferase from the NF-κB-dependent reporter plasmid pBIIxluc (Fig. 4C). The authenticity of the role of NF-κB was further confirmed by inhibition of luciferase synthesis by sodium salicylate (13, 33). Interestingly, expression of a deletion mutant of P missing the C-terminal 39 amino acids that included the phosphorylation sites Ser²³² and Ser²³⁷ failed to activate RelA in all these experiments. Two P clones (T1 and T2), in which the third and fourth codons, respectively, of the P gene were mutated to termination codon TAA, also failed to activate RelA (Fig. 4C), suggesting the activation was likely a property of the P protein itself and not due to its mRNA or DNA sequence. Cells transfected with control pcDNA3 vector did not show any increase in nuclear RelA concentration or activity (Fig. 4B and C). To eliminate the possibility that salicylate had affected the expression of recombinant P protein, we monitored the P protein levels by immunoblot analysis. As shown in Fig. 4D, P protein levels in salicylate-treated cells were comparable to those in untreated cells. Thus, we conclude that the recombinant P protein can activate RelA in the absence of any other viral gene product(s). It appeared, however, that the degree of activation was lower than that seen in RSV-infected cells.

FIG. 4.

Activation of RelA by recombinant RSV P protein. A549 cells at 75% confluency were transfected with pcDNA3-P alone (A, B, and D) or together with pBIIxluc (C) in multiple dishes and processed as indicated, using procedures described in Materials and Methods. The various pcDNA-3 plasmids are vector (V), full-length P clone (P), ΔC39 P clone (ΔP), and P clones with premature termination codons (T1 and T2). RelA and P protein levels were measured by immunoblotting of nuclear and cytoplasmic fractions, respectively (A and D). NF-κB activities in nuclear extracts were assayed by EMSA (B). Ex vivo activities of NF-κB were determined by reporter luciferase assay (C); average values from three experiments are shown with standard deviation bars. Where indicated, sodium salicylate (15 mM in panels A and B; 2 and 8 mM in panel C) was added to the transfected cells 4 h after addition of the DNA.

Inhibition of PP2A by RSV phosphoprotein.

We have shown earlier that the RSV phosphoprotein is resistant to dephosphorylation by cellular phosphatases in vitro as well as ex vivo (9). However, it could be dephosphorylated in vitro by the more promiscuous Ser/Thr phosphatase encoded by bacteriophage lambda (4), suggesting that the phosphate groups of the P protein probably fail to interact with the catalytic pocket of the eukaryotic phosphatases in a proper manner. Thus, to explain activation of RelA by P, we entertained the scenario that the P protein may still bind to eukaryotic phosphatases and thus inactivate the latter, which in turn would lead to higher levels of cellular phosphorylation leading to activation of RelA. Marine toxins such as OA and calyculin have been indeed shown to activate RelA when added to cell cultures at concentrations that inhibit PP2A more strongly than PP1 or PP2B (30, 49, 52). To test whether P protein might function through a similar mechanism, we tested whether P can inhibit purified human PP2A in vitro. First, we confirmed that ³²P-labeled P protein is indeed resistant to dephosphorylation by purified PP2A in vitro (Fig. 5A) although it could be dephosphorylated by PPλ. The effect of unlabeled phosphorylated P was then tested on the activity of PP2A, using ³²P-labeled casein as the substrate. The results show that phosphorylated P protein did inhibit PP2A activity and that phosphate-free P protein at similar concentrations was considerably less inhibitory (Fig. 5B and C). Essentially similar conclusions were reached in assays using phosphohistone as the substrate (data not shown).

FIG. 5.

(A) Resistance of RSV P protein to PP2A. Bacterially expressed RSV P protein was phosphorylated by CK2 in the presence of [γ-³²P]ATP (9, 39); 2 μg of the labeled P protein was incubated with 20 ng of purified enzyme (Enz.) PPλ or PP2A in 20-μl reaction mixtures under standard conditions (4). At the indicated time points, 4 μl of the reaction mixtures was removed and added to 20 μl of Laemmli sample buffer (34). All samples were heated in a boiling water bath for 5 min and analyzed by SDS-PAGE (39) followed by autoradiography as shown. (B and C) Inhibition of PP2A by phosphorylated P protein. Standard phosphatase reactions (20 μl) were carried out with 2 μg of ³²P-labeled casein as the substrate and 4 μg of phosphate-free (P) or phosphorylated (P-PO₄) RSV P protein. At indicated times, 4 μl of the reaction mixture was removed. Reactions were analyzed by SDS-PAGE (34) and autoradiography (B). The multiple bands of ³²P-casein in each lane were scanned by densitometry, and their total intensities were plotted as percentages of the starting intensities (at 0 min) (C). Results are shown for assays with no P added (□), unphosphorylated P (○), and phosphorylated P (•).

To determine whether P may inhibit PP2A in RSV-infected cells, we took advantage of the facts that the peptide RRREEETEEEAA, phosphorylated at the Thr residue (by CK2 in vitro), is a relatively specific substrate of PP2A and that OA at a concentration of 2 nM preferentially inhibits the PP2A class of phosphatases. Thus, the specific activity of PP2A in RSV-infected A549 extracts was quantitated by assaying the OA-sensitive phosphatase activity and comparing it with similar activity in extracts of uninfected cells. These results revealed a 50 to 60% drop of PP2A activity in the infected cells (data not shown), although it was not possible to confirm that this was specifically due to the P protein and not due to any other viral gene product.

RSV phosphoprotein associates with PP2A.

To test directly whether RSV P indeed binds to PP2A ex vivo, we transfected A549 cells with the pcDNA3-P clone, immunoprecipitated the expressed P protein with a specific anti-P antibody, and then tested for the presence of phosphatases in the precipitate by immunoblot analysis using antibodies against PP1, PP2A, and PP2B. While no PP1 or PP2B could be detected (data not shown), substantial amounts of PP2A were found in the immunoprecipitates (Fig. 6A). In the reciprocal experiment, immunoprecipitates obtained with anti-PP2A also contained P protein (Fig. 6A). The C-terminal deletion mutant of P, lacking the phosphorylation domain, associated with PP2A to a much lesser extent, although it was expressed as abundantly as the wild type. Addition of sodium salicylate to the cells had no effect on the association (data not shown). Association of P with PP2A, perhaps to a somewhat lesser extent, was also observed in RSV-infected cells (Fig. 6B). Together, these results suggest that P protein specifically associates with PP2A and that the phosphate groups of P may have a role in stabilizing the association.

FIG. 6.

Association of recombinant (A) and viral (B) P proteins with PP2A. A549 cells were transfected with plasmid pcDNA3-P (lane P), pcDNA3 clone of a ΔC39 deletion mutant of P (lane ΔP), or pcDNA3 vector without P gene (lane V) (A) or infected with RSV (Long) at an MOI of 3.0 (B). At 48 h posttransfection (or p.i.), the cells were processed for immunoprecipitation (IP) using antibodies against either PP2A or RSV P protein as indicated. The samples were analyzed by SDS-PAGE in which the immunoprecipitate from about 10⁶ cells was applied in each lane. The proteins were transferred to a membrane and probed with the same antibodies in Western blot analysis as shown.

DISCUSSION

In this report, we have attempted to delineate the pathway leading to the persistent activation of NF-κB by RSV. The principal participants in the activation of the RelA subunit of NF-κB were (i) viral gene products acting as the proximal signal(s) of activation, at least one of which is the phosphoprotein P functioning through the sequestration and inhibition of cellular PP2A and thus likely increasing the net phosphorylation of phosphoproteins involved in the RelA activation pathway; (ii) a D609-sensitive enzyme—perhaps some form of phospholipase—that may positively regulate gene transcription by RelA rather than its nuclear translocation; (iii) PKC that signals to promote phosphorylation and degradation of both IκBα and IκBβ in the initial stages of the infection; (iv) ROI generated by a nonmitochondrial pathway; (v) phosphorylation and proteasomal degradation of the IκB proteins, the former event appearing to be more important than the latter; and finally (vi) new IκBβ, synthesized in a nonphosphorylated form in the later stages of infection. These results are integrated in the signal transduction pathway in Fig. 7. In what follows, we discuss the various details and implications of these findings.

FIG. 7.

Postulated model for NF-κB activation by RSV. Unidentified RSV gene product(s) “X” intracellularly activates protein kinases such as PKC that may in turn activate other downstream kinases. Raf, members of the MAPK cascade, and ribosomal protein S6 kinase (pp90^rsk) are some postulated examples (see text for details). Inhibition of PP2A by P may lead to a higher steady-state level of the phosphorylated product of any of these kinases; however, it is also possible that the kinases themselves are regulated by PP2A.

Although the recombinant P protein activated RelA, the low levels of activation suggested that there must be other viral gene products that play additional or more important roles in an RSV-infected cell. It remains to be determined whether these entities are RNA or protein in nature and how exactly they interact with the cellular signal transduction machinery. At least for the P gene of RSV, our evidence suggests that its protein product and not the mRNA is the primary signal, since introduction of premature translation stop codons in the pcDNA3-P clone abolished P protein expression as well as RelA activation (Fig. 4C). The xanthate compound D609 was previously shown to inhibit the growth of at least two nonsegmented negative-strand RNA viruses, vesicular stomatitis virus (41) and RSV (55). Subsequent studies revealed that it inhibited the synthesis of a number of viral proteins, including a 78-fold reduction of the viral N protein and a 7-fold reduction of the P protein. Nonetheless, replication of the RSV genome remained unaffected (55), which was somewhat surprising, since de novo synthesis of N proteins of negative-strand RNA viruses, including those of vesicular stomatitis virus and RSV, is essential for viral RNA replication (reviewed in reference 6). The phosphorylation of the P protein of RSV appeared to be affected by D609 even more drastically (55). It was therefore hypothesized that D609 inhibited a viral morphogenetic step, perhaps requiring phosphorylation of the P protein (55). Based on the results presented here, it is possible that any or some of these viral targets of D609 may be involved in the function, rather than the nuclear translocation, of RelA. Clearly, identification of the exact viral target(s) of D609 should clarify this issue.

On the other hand, activation of NF-κB by a number of cellular agonists, such as TNF, has also been shown to be inhibited by D609 (47). It was long thought that this is due to the fact that D609 is a direct and specific inhibitor of phosphotidylcholine-specific PLC (1, 16); however, PLD was also shown to be inhibited recently (32). Since D609 failed to inhibit the nuclear translocation of RelA (Fig. 3B), neither PLC nor PLD may be required for the inactivation of IκB by RSV. In accordance with our results, a recent study has indeed postulated that D609 may inhibit the DNA-binding activity of nuclear NF-κB by inhibiting a putative accessory factor required for NF-κB function (16); the identity of the factor and the effect of D609 on it remain to be determined.

It was quite obvious that in the early phase of NF-κB activation by RSV, both the IκB proteins were destroyed and essentially disappeared by 4 to 5 h p.i. (Fig. 3C and D). Both inhibitors, however, reappeared at around 8 h p.i. As has been shown for other NF-κB agonists, this is most certainly due to the transcriptional induction of the IκB genes by activated NF-κB, since all of the drugs that inhibited NF-κB activation also blocked the reappearance of the IκBs (Fig. 3). In contrast, the newly synthesized IκBs appeared to escape phosphorylation and degradation during the persistent activation of NF-κB by RSV, suggesting that the increased synthesis probably overwhelmed the activation signals and that the nonphosphorylated form of IκBβ may in fact protect RelA from inhibition by IκBα (44, 54). Two major lines of recent evidence lend precedence to our findings. The first is derived from studies of the Tax protein of human T-lymphotrophic virus type 1, which remains the most extensively studied viral activator of NF-κB to date. Tax is known to persistently activated NF-κB, and a recent study of its mechanism has revealed that expression of the protein resulted in the degradation of both species of IκB, at least at the early times (40). Interestingly, activation of PKC by Tax was also shown to be essential for Tax-mediated activation of NF-κB (36). Second, as mentioned earlier, underphosphorylated IκBβ, synthesized during persistent activation of NF-κB by TNF or phorbol esters, was shown to form complexes with RelA (p65) that were transcriptionally active and insensitive to IκBβ (54). Thus, RSV seems to employ a combination of these two mechanisms to persistently activate RelA. It remains to be seen whether IκBβ also reappears in the persistent activation of NF-κB by Tax.

The inhibitory effects of a variety of PKC inhibitors strongly implicated a role of PKC enzymes in the activation of RelA by RSV. Chelerythrine A, another specific PKC inhibitor, had similar effects (data not shown). The pseudosubstrate peptide Myr-ψPKC is known to inhibit at least three PKC isozymes: α, β, and γ. Use of more specific inhibitors and antisense oligonucleotides against specific isozymes should allow us to identify the particular isozyme(s) of PKC that is involved in NF-κB activation by RSV. We want to note here that the atypical isozyme PKC-ζ has recently been implicated to play a role in the activation of NF-κB by simian virus (SV40) small t antigen (49) (see below) and by human immunodeficiency virus (24).

PDTC is an efficient scavenger of reactive oxygen; thus, its strong inhibitory effect on nuclear translocation of RelA by RSV suggests a major role of ROIs in this pathway. PDTC prevented the phosphorylation of IκBα (Fig. 3), indicating that the ROIs work upstream of the phosphorylation step. The inability of rotenone to inhibit the process suggested that the reactive oxygens are probably of nonmitochondrial origin. As expected, the specific antiproteasome drug MG132 must act at the distal end of the signaling pathway, since unlike the other drugs, it allowed phosphorylation of both IκB polypeptides but prevented the degradation of the phosphorylated end products. Curiously, both IκBα and IκBβ were affected in a comparable manner by the inhibitor drugs, suggesting that the mechanisms of their inactivation as well as resynthesis may share similar, if not identical, features. Similar conclusions have recently been derived from studies of a mutant cell line exhibiting a lack of degradation of all kinds of IκB proteins (18). In contrast to IκBα, the mechanism of regulation of IκBβ is virtually unknown, although a role of phosphorylation and proteolysis is strongly implicated (30, 57). It was, therefore, concluded that phosphatases rather than kinases may be the major regulators of IκBβ phosphorylation. It is tempting to speculate that the phosphatase in question is PP2A and that its inhibition by the RSV P protein may be a major contributor of increased IκBβ phosphorylation as observed by us. The stabilizing effect of MG132 on phosphorylated IκBβ is also in agreement with our finding and indicates a role of proteasomes in IκBβ degradation as well (30, 57). It is noteworthy that none of the inhibitors tested here seemed to inhibit the early activation of RelA by RSV. Recent studies have indicated that at least a part of the early activation is due to the binding of the virus to its receptors and occurs even in the absence of viral gene expression (23). This mechanism may, therefore, employ membrane-bound signaling molecules and a signal transduction pathway fundamentally different from that of persistent activation.

The activation of a signaling pathway through the sequestration and inhibition of a phosphatase by a nonsubstrate phosphoprotein, as suggested by our studies, is without a direct precedent. However, a number of earlier findings provide indirect support for this novel mechanism. First, whereas a variety of phosphorylated proteins (such as casein, histone, and lysozyme) are excellent substrates for all phosphatases, the corresponding thiol-phosphorylated proteins containing nonhydrolyzable thiophosphate groups, produced in vitro by using γ-thio-ATP, are phosphatase resistant, as expected. However, they still retain the capacity to bind phosphatases and have in fact been traditionally used to affinity purify phosphatases (28). While a substrate phosphoprotein must also bind the phosphatase, the catalytic role of the enzyme demands that the binding be reversible. In other words, the substrate must dissociate from the enzyme following the removal of the phosphate group so that the enzyme can then act on the next substrate molecule. In principle, therefore, a stable and irreversible sequestration of the phosphatase can be achieved by a nonsubstrate phosphoprotein which would bind to the substrate-binding domain of the phosphatase but not interact with the enzyme’s catalytic domain, perhaps due to the lack of a proper sequence or conformation around the phosphate groups. We postulate the P protein of RSV satisfies these criteria. Interestingly, the inhibition of PP2A by P protein was not detectable when pNPP was used as the substrate. While we do not know the exact reason behind this, perhaps the simplest scenario that may explain all of these findings is that PP2A contains a substrate-binding domain which is distinct from the catalytic site of the enzyme. Binding of the large P protein to this domain may prevent binding of other macromolecular substrates by steric hindrance. However, the small molecular substrate pNPP may not require the protein substrate-binding site and can directly access the catalytic pocket even when P protein is bound to the enzyme.

Second, as indicated earlier, potent phosphatase inhibitors, such as OA and calyculin, activate RelA in cells of various origins (30, 52, 53). We have made essentially similar observations when using OA (at 3 nM) on A549 cells (data not shown). RSV P, like these toxins, strongly inhibits PP2A (Fig. 5) and therefore may activate RelA through similar mechanisms. The lack of association of P with PP1 or PP2B points to the specificity and importance of the P-PP2A interaction. We predict that the steady-state levels of a variety of cellular phosphoproteins that are otherwise substrates for PP2A may also undergo an increase in cells expressing the P protein. As a corollary, RSV infection may affect cellular metabolism in many different ways that are yet uncharacterized. This is currently being tested.

Among the best-studied examples of PP2A regulation are those accomplished by SV40 T antigens and CK2. Detailed studies by Walter and Mumby have shown that the SV40 small t antigen inhibits PP2A by directly associating with the catalytic subunit C, although the exact amino acid residues involved in the association remain to be identified (reviewed in reference 56). Curiously, this led to activation of the mitogen-activated protein kinase (MAPK) cascade and NF-κB, most likely through the stimulation of the atypical PKC isozyme ζ (49). An intriguing relationship between a kinase and a phosphatase has been revealed by the recent demonstration that the catalytic α subunit of CK2 also binds to PP2A in vitro and in mitogen-starved cells and that the overexpression of CK2α resulted in deactivation of the MAPK pathway and suppression of cell growth (31). These results suggested that PP2A serves an important role in cell signaling and that its sequestration by CK2α abrogates cell growth. Interestingly, the binding of CK2α to PP2A required the sequence motif HENRKL, common to both CK2α and small t antigen (31). This sequence motif, however, is absent in RSV P protein (39). Thus, RSV P may use a novel sequence motif to interact with the catalytic subunit of PP2A, or it may associate with one of the regulatory subunits of PP2A. A detailed knowledge of which subunits of the PP2A holoenzyme are present in the RSV P immunoprecipitate will be a starting point in resolving the nature of P-PP2A interaction.

It is interesting to us that an essentially cytoplasmic negative-strand RNA virus exploits some of the downstream signaling molecules that are normally used by physiological agonists to activate cellular NF-κB. The chain of signaling events must be triggered by specific RSV gene products that need to be characterized, although our results suggest that the phosphoprotein P is likely to be one of them. It will be important to determine whether the continued presence of these gene products is needed to maintain the persistent activation of NF-κB. Nevertheless, based on the foregoing, we propose that these gene products activate intracellular PKC, which in turn may signal through a variety of potential kinases including members of the Raf/MAPK pathway (Fig. 7). In addition to human immunodeficiency virus and SV40 as mentioned earlier (24, 49), adenovirus was recently shown to stimulate of the Raf/MAPK signaling pathway, and this was required for the induction of IL-8 gene expression in adenovirus-infected cells (14). Finally, recent studies suggested that MEKK1, a member of the MAPK cascade, could indeed activate the multisubunit IκBα kinase complex (35). At least some of the MAPK appeared to activate the ribosomal protein S6 kinase (pp90^rsk), which could in turn directly phosphorylate IκBα (27, 37). The role of these kinases in RSV-mediated phosphorylation of IκB proteins is currently under investigation.