Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Cell. Author manuscript; available in PMC 2008 October 24.
Published in final edited form as:
PMCID: PMC2572259

Phosphatase RNAi screen in mouse astrocytes identifies new regulators of NF-κB signaling


Regulation of NF-κB activation is controlled by a series of kinases, however the roles of phosphatases in regulating this pathway are poorly understood. We report a systematic RNAi screen of phosphatases that modulate NF-κB activity. Nineteen of 250 phosphatase genes were identified as regulators of NF-κB signaling in astrocytes. RNAi selectively regulates endogenous chemokine and cytokine expression. Co-immunoprecipitation identified associations of distinct protein phosphatase 2A core or holoenzymes with the IKK, NF-κB, and TRAF2 complexes. Dephosphorylation of these complexes leads to modulation of NF-κB transcriptional activity. In contrast to IKK and NF-κB, TRAF2 phosphorylation has not been well elucidated. We show the Thr117 residue in TRAF2 is phosphorylated following TNFα stimulation. This phosphorylation process is modulated by PP2A and is required for TRAF2 functional activity. These results provide direct evidence for TNF-induced TRAF2 phosphorylation and demonstrate that phosphorylation is regulated at multiple levels in the NF-κB pathway.


Astrocytes are the most abundant glial cell type in the central nervous system (CNS). They contribute to homeostasis of the CNS by participation in neurogenesis (Song et al., 2002), synapse formation (Mauch et al., 2001), synaptic transmission (Kang et al., 1998), brain repair (Garcia-Segura et al., 1999), and maintenance of the blood-brain barrier (Bush et al., 1999; Prat et al., 2001). Astrocytes also play a role in the pathophysiology of inflammatory and neurodegenerative diseases (Volterra and Meldolesi, 2005). Brain lesions observed in Alzheimer Disease, ischemic damage, autoimmune responses, infections (e.g. HIV), and tumors are rapidly bordered by hypertrophic astrocytes. These reactive astrocytes can produce a variety of proinflammatory mediators which amplify the inflammatory response.

Tumor necrosis factor-α (TNFα) plays a critical role in the induction and perpetuation of innate, immune, and inflammatory responses. TNFα signaling occurs through specific receptors which induce activation of NF-κB along with other transcription factors (MacEwan, 2002). NF-κB plays an essential role in inflammation, immunity, development, cell proliferation and apoptosis (Hayden and Ghosh, 2004). The activity of NF-κB is tightly regulated by association with an inhibitor of NF-κB (IκB). NF-κB bound to IκB is found in the cytoplasm as an inactive complex. However, following TNFα treatment the IκB-kinase (IKK) is activated resulting in phosphorylation of IκB proteins. This signal-induced phosphorylation targets IκB for polyubiquitination and subsequent degradation allowing the freed NF-κB molecules to translocate to the nucleus and modulate specific gene transcription. Phosphorylation has been shown to regulate the various steps in NF-κB signaling (Hayden and Ghosh, 2004; Viatour et al., 2005), a process that is controlled by kinases and phosphatases with opposing roles. Dozens of kinases have been demonstrated to be involved in the phosphorylation of IκB, NF-κB and other components in the NF-κB pathway (Hayden and Ghosh, 2004; Viatour et al., 2005). In contrast to the extensive analysis of kinase function, the roles of phosphatases in NF-κB signaling remain poorly understood.

In this study, a large-scale RNAi screen was adopted to elucidate the roles of phosphatases in the NF-κB pathway. After two rounds of screening, 19 phosphatases were identified as regulators of NF-κB signaling either activating or suppressing NF-κB transcriptional activity and binding ability. Distinct protein phosphatase 2A (PP2A) enzymes were associated with the IKK, NF-κB, and TRAF2 complexes. Dephosphorylation of these complexes led to inhibition of NF-κB transcriptional activity and regulation of endogenous chemokine or cytokine expression in astrocytes.


Primary screen using siRNA library

To identify which phosphatases were involved in the NF-κB pathway, a large-scale RNAi approach was adopted to characterize the role of individual phosphatase genes. A siRNA library comprising 250 phosphatase or putative phosphatase genes was prepared based on a bidirectional siRNA vector transcribing siRNAs from convergent opposing promoters (Kaykas and Moon, 2004; Zheng et al., 2004) (Fig. 1A). Astrocytes were transfected with the pNF-κB-Luc and Renilla luciferase reporters plus a pair of siRNA constructs for each gene. The Renilla luciferase vector was used as a control of transfection efficiency.

Figure 1
RNAi screens of phosphatase genes that regulate NF-κB transcriptional activity. (A) The design of the pBabe-Dual vectors. The position of the puromycin resistance gene and the U6 and H1 promoters are indicated. (B) Data from representative experiments ...

TNFα treated or untreated astrocytes were used to screen phosphatases involved in NF-κB activation. Genes that satisfied the following 4 criteria of activity and specificity were categorized as positive candidates: (1) Genes scoring 2 standard deviations (SD) above or below the median were considered potential hits. Two SD roughly equals a 4 fold increase over media controls, a 3 fold increase over the TNFα treated control, or a 70% reduction in NF-κB activity as measured by changes of NF-κB reporter activity (Table 1). (2) Candidate genes exhibited reporter specificity by demonstrating <2 fold activity changes with a mutant NF-κB reporter on the same vector backbone. (3) The expression and activity of each phosphatase gene was confirmed in primary astrocytes; (4) Candidate genes identified by reporter assay were confirmed by use of an independent assay of NF-κB activity. Those genes that scored negative by any of these imposed conditions were excluded from further study.

Table 1
Identification of phosphatase genes associated with the NF-κB pathway

Twenty-five candidate phosphatase genes were identified according to criteria (1) (Fig. 1B and 1C). These siRNA pairs were further tested using a mutant NF-κB reporter containing a scrambled NF-κB sequence to exclude effects of the vector backbone. One phosphatase was excluded after RNAi showed >2 fold increases on the mutated NF-κB reporter. Phosphatase expression in astrocytes was confirmed for all but one gene by RT-PCR.

Secondary screen

To confirm positive hits from the reporter assay, we investigated the remaining candidates by a chemiluminescent transcription factor binding assay (CTFA) which detects active nuclear NF-κB from astrocytes and its binding ability. Down-regulation of 19 phosphatases consistently showed ≥2 fold changes consistent with the results of the reporter assay. After all screening we identified 13 NF-κB suppressing phosphatases (Fig. 1D) and 6 NF-κB activating phosphatases (Fig. 1E). These genes included 13 components of the serine/threonine phosphatases, 4 tyrosine phosphatases and 2 lipid phosphatases (Table 1 and Supplementary Fig. 1).

Individual siRNA constructs to each of the 19 “hits” were specific (Supplementary Fig. 2A) and generally reduced target protein or mRNA levels by more than 60% except one PP2Cζ siRNA construct that demonstrated only 35% efficiency (Supplementary Fig. 2). Several phosphatases previously known to be involved in NF-κB signaling were identified, including PPP2CA (Yang et al., 2001), PPP4C (Hu et al., 1998), PTPN2 (Ibarra-Sanchez et al., 2001) and PTEN (Mayo et al., 2002), thus underscoring the validity and robustness of our two rounds of screening (Fig. 1B and 1C). Moreover, the present study identified 8 catalytic and regulatory subunits of protein phosphatase 2A (PP2A) or protein phosphatase 1 (PP1) that modulated NF-κB activity (Table 1). With the exception of PPP1R7 down-regulation of all these phosphatase components activated NF-κB transcriptional activity, which is consistent with the activation of NF-κB and nuclear translocation observed with specific pharmacologic inhibitors of PP1 and PP2A (Sun et al., 1995) (Supplementary Fig. 3A and 3B).

We next determined cell specificity by examining the activity of all 19 hits on mouse NIH3T3 fibroblasts (Fig. 1F). RNAi to 16 phosphatase genes demonstrated similar effects on both astrocytes and fibroblasts. However, 3 genes lacked activity on fibroblasts (INPP4A, PTPRN, and PTPRJ). Equally important, was the reciprocal observation that PPM1B RNAi enhanced TNF-induced reporter activity in fibroblasts (Fig. 1F). This finding is consistent with a previous report that PPM1B associates with the IKK complex in 293T kidney cells causing dephosphorylation of IKKβ and reducing kinase activity (Prajapati et al., 2004). In contrast, PPM1B RNAi was consistently inactive in astrocytes (Table 1). The combined results demonstrate that distinct phosphatase genes can selectively modulate NF-κB responses in various cell types.

Protein phosphatase 2A associates with and dephosphorylates IKK or NF-κB

To examine the interactions of the suppressing phosphatases (PPP2CA, PPP2CB, PPP2R1A, PPP2R1B, PPP2R2D, PPP2R5C, G4.1, and PPM1L) or NF-κB activating phosphatases (PP2Cδ and PPP1R7) the genes were tagged with the myc- or Flag-epitope. Localization of over-expressed phosphatase components was examined by anti-epitope staining. With the exception of PP2Cδ the phosphatases were localized in both the cytoplasm and nucleus of astrocytes. PP2Cδ was localized exclusively in nucleus (data not shown).

Over-expressed phosphatase genes were next examined for interactions with IKK, NF-κB, and TRAF2 by co-immunoprecipitation. Among the 10 phosphatase components tested, only PPP2CB and PPP2R1A pulled down endogenous IKK complexes containing IKKβ, IKKγ and IKKα (Fig. 2A and data not shown). IKKβ was not co-precipitated with the other eight candidates following immunoprecipitation with anti-IKKβ antibody (Fig. 2A and data not shown). Using antibodies against IKKβ or the PP2A catalytic subunit, we detected endogenous IKK complexes containing endogenous PP2A (Fig. 2B). Since PPP2CB and PPP2R1A are catalytic and structural subunits of PP2A, respectively, the combined data indicate that a PP2A core enzyme containing PPP2CB and PPP2R1A is associated with the IKK complex. Activation of the IKK complex depends on phosphorylation of its two catalytic subunits, IKKα and IKKβ (Delhase et al., 1999). Therefore, we tested whether PPP2CB complexes dephosphorylated IKKβ. To examine this, a PP2A core complex containing myc-PPP2R1A was co-transfected along with Flag-PPP2CB or other phosphatase components and the complexes were eluted from anti-Flag beads with Flag peptide. In vitro studies showed dephosphorylation of IKKβ after treatment with purified Flag-PPP2CB plus PPP2R1A (Fig. 2C). To further test the specificity of PP2A on IKKβ phosphorylation, we co-transfected PPP2CB RNAi constructs with IKKβ or the IKKβ constitutively active mutant, IKKβ S177E/S181E (IKKβ SSEE) (Mercurio et al., 1997). We expected that PPP2CB RNAi would synergize with IKKβ but not with the IKKβSSEE mutant in which the phosphorylated serine residues were replaced by glutamic acid. Indeed as shown in Fig. 2D, PPP2CB RNAi synergized with IKKβ to increase NF-κB reporter activity but PPP2CB RNAi failed to synergize with IKKβSSEE. Furthermore, PPP2CB RNAi enhanced basal and TNF-induced IKKβ phosphorylation (Fig. 2E), suggesting one mechanism of PPP2CB suppression of NF-κB signaling is through dephosphorylation of the IKK complex by this PP2A phosphatase.

Figure 2
PP2A catalytic and structural components selectively form complexes with IKK and p65 NF-κB. (A) The indicated myc-tagged phosphatases were transfected into astrocytes and immunoprecipitated with either anti-myc or anti-IKKβ antibody. The ...

Several phosphorylation sites have been identified in NF-κB p65 (Hayden and Ghosh, 2004; Viatour et al., 2005), thus it was of interest to determine whether phosphatases also controlled NF-κB phosphorylation. The same set of 10 phosphatase components were tested by reciprocal co-immunoprecipitation; PPP2CA and PPP2R1B were associated with NF-κB, either by immunoprecipitation with myc antibody or by reverse immunoprecipitation with NF-κB antibody (Fig. 2F). PPP2CA showed stronger associations with NF-κB than PPP2R1B (Fig. 2F). Endogenous PP2A also formed a complex with endogenous NF-κB p65 (Fig. 2G). Phosphorylation of the Ser536 or Ser276 residues in NF-κB p65 is one sign of NF-κB activation (Sakurai et al., 2003; Vermeulen et al., 2003; Zhong et al., 1997). Using an in vitro dephosphorylation assay with phospho-specific antibodies, we found purified Flag-epitoped PPP2CA/PPP2R1B core enzyme dramatically dephosphorylated the p65 residue Ser536, but failed to dephosphorylate residue Ser276 (Fig. 2H). Purified PPP1R7, PP2Cδ and G4.1 had no visible impact on dephosphorylation of either phosphorylation site (Fig. 2H). The combined results demonstrate the in vitro specificity of PP2A enzymatic activity on selected residues.

Dephosphorylation of TRAF2 by PP2A holoenzyme inhibits NF-κB activity

TRAF2 plays an important role in the TNFα-mediated NF-κB signaling pathway. Although TRAF2 is a phosphorylated protein (Chaudhuri et al., 1999; Pomerantz and Baltimore, 1999), the mechanism of TRAF2 phosphorylation and the potential effects of TRAF2 dephosphorylation on NF-κB activity are poorly understood. To address these issues, the association of 10 selected phosphatases with TRAF2 was tested. As shown in Fig. 3A, PPP2CA and PPP2R1A were associated with TRAF2 as noted by immunoprecipitation with anti-myc antibody or by reverse immunoprecipitation with anti-TRAF2 antibody. Endogenous PP2A also formed a complex with endogenous TRAF2 (Fig. 3B). We also found one PP2A regulatory subunit, PPP2R5C, associated with the TRAF2 complex (Fig. 3C). Further mapping of TRAF2 to evaluate the roles of various functional domains found that both the TRAF-N and TRAF-C domains were required for binding of PPP2R5C (Fig. 3D). As reported previously, over-expression of TRAF2 induces NF-κB activation presumably because it induces TRAF2 trimerization thereby mimicking the effects of ligand stimulation on the TNF receptor (Takeuchi et al., 1996). Co-overexpression of PPP2R5C dramatically inhibited TRAF2 induced NF-κB reporter activity while another PP2A regulatory subunit (PPP2R2D) and other phosphatases displayed little or no inhibition (Fig. 3E). The IL-1 and TNF signal pathways use different TRAF molecules to transduce signals but the signaling pathways converge further downstream to activate NF-κB. Thus, we tested the effect of PPP2R5C RNAi on IL-1 mediated NF-κB reporter activity. As shown in Fig. 3F, there was no apparent effect of PPP2R5C RNAi on IL-1 stimulated reporter activity although PPP2CA, PPP2CB, and PPP2R1B RNAi which affect the IKK and p65 complexes common to both the TNF and IL-1 signaling pathways were enhanced (Fig. 3F). Weak responses were noted with PPP2R1A; a 2.2 fold increase with IL-1 (Fig. 3F) and 1.8 fold with TNF (Table 1). Thus, the functional in vitro data support the physical association of PP2A with TRAF2 and suggest that the PPP2CA/PPP2R1A/PPP2R5C holoenzyme suppresses NF-κB activity by dephosphorylating TRAF2.

Figure 3
Dephosphorylation of TRAF2 by PP2A holoenzyme inhibits NF-κB activity. (A) The indicated myc-tagged phosphatases were transfected into astrocytes and immunoprecipitated with either anti-myc or anti-TRAF2 antibody. (B) Cell lysates were immunoprecipitated ...

Previous evidence for TRAF2 phosphorylation was based on a two-dimensional phosphoamino acid separation which provided little mechanistic insight (Chaudhuri et al., 1999; Pomerantz and Baltimore, 1999). To determine the critical TRAF2 phosphorylation site and corresponding function, we first compared the NF-κB reporter activity of different TRAF2 truncation mutants (Supplementary Fig. 4A and 4B). Consistent with a previous study (Takeuchi et al., 1996), the ring and finger domains were important for TRAF2 activity. To further define the TRAF2 phosphorylation site, 21 conserved serines or threonines were mutated to alanine. Most of these sites were located in the ring and finger domains of TRAF2 (Fig. 4A and data not shown). After transfection of these mutants into 293T cells, two mutants Ser102Ala and Thr117Ala showed the lowest NF-κB reporter activity (Supplementary Fig. 4 C). These two point mutants also showed dramatically reduced NF-κB reporter activity in astrocytes (Fig. 4B) and TRAF2-/- MEFs (Fig. 4C). To investigate phosphorylation, we noted that the finger domain (residues 99-271) showed two distinct bands by electrophoretic mobility in a 4-20% SDS-PAGE gel (Fig. 4D). The upper band was sensitive to CIP phosphatase treatment (Fig. 4D). This suggested that the finger domain of TRAF2 was phosphorylated. Therefore we generated several finger domain mutants and found only the Thr117Ala mutation abolished the upper band (Fig. 4E). Finally, we generated antibody against a phospho-Thr117 peptide which specifically recognized phosphorylated Thr117 in TRAF2 (Fig. 4F and Supplementary 4D). Using anti-phospho Thr117 antibody, we noted increased TRAF2 Thr117 phosphorylation 15 min after TNFα stimulation (Fig. 4G). TNF-induced Thr117 phosphorylation of TRAF2 was inhibited by PPP2R5C over-expression while neither IKKβ Ser181 nor NF-κB p65 Ser536 phosphorylation were affected (Fig. 4H). In addition, PPP2R5C RNAi enhanced Thr117 phosphorylation of TRAF2 while control RNAi had no effect on TRAF2 phosphorylation (Fig. 4I).

Figure 4
TNF induced TRAF2 Thr117 phosphorylation. (A) Amino acid sequence alignment of a portion of the first zinc finger domain of TRAF2. (B) The fold induction of NF-κB reporter activity in astrocytes transfected with TRAF2 and different Ser/Thr to ...

Phosphatases regulate chemokine and cytokine transcription in astrocytes

Since NF-κB regulates the production of proinflammatory chemokines in astrocytes (Kim et al., 2005; Li et al., 2001; Zhai et al., 2004) the effects of inhibition of the selected 10 phosphatases were investigated on chemokine and cytokine transcription. In resting astrocytes, silencing 6 NF-κB suppressing phosphatases (PPP2CA, PPP2CB, PPP2R1B, PPP2R2D, PPP2R5C, and PPM1L) enhanced expression of the monocyte chemoattractant MCP-1 and the neutrophil chemoattractant KC, although not always by the 4-fold level used to identify hits in our initial screens (Fig. 5A). In contrast, PPP2R1A RNAi increased KC but displayed minimal effects on MCP-1 expression while G4.1 or PPP1R7 RNAi failed to modulate chemokine levels in resting astrocytes (Fig. 5A). In contrast, silencing PP2Cδ resulted in reduction of basal MCP-1 and KC mRNA levels (Fig. 5A).

Figure 5
Phosphatases regulate chemokine and cytokine expression. (A) The relative mRNA levels of MCP-1 (open bars), KC (shaded bars), and IL-6 (solid bars) in RNAi transfected astrocytes without stimulation. Astrocytes were transfected with phosphatase RNAi vectors ...

Silencing of the NF-κB suppressing phosphatases (PPP2CA, PPP2CB, PPP2R1B, PPP2R2D, PPP2R5C) also synergized with TNFα for enhanced expression of MCP-1 and KC by >3 fold (Fig. 5B). In contrast, G4.1 RNAi selectively enhanced KC expression and PPP1R7 RNAi inhibited TNFα induced expression of MCP-1 by >70% (Fig. 5B).

IL-6 expression is tightly regulated and transcription is dependent on both NF-κB and C/EBP in astrocytes (Schwaninger et al., 2000; Van Wagoner and Benveniste, 1999). Silencing the NF-κB suppressing phosphatases, PPP2CA, PPP2R2D, PPP2R5C, G4.1 and PPM1L increased mRNA levels for IL-6 by >4 fold in resting astrocytes and >3 fold in TNFα stimulated cells, but IL-6 mRNA was not dramatically enhanced in cells transfected with RNAi to PPP2CB, PPP2R1B and PPP2R1A (Fig. 5A and 5B). In contrast, inhibition of the NF-κB activating phosphatase PPP1R7 resulted in >70% reduction of IL-6 mRNA in resting astrocytes while silencing PP2Cδ failed to significantly modulate IL-6 expression in untreated astrocytes (Fig. 5A). In summary, silencing of various phosphatase genes resulted in differential patterns of chemokine and/or cytokine regulation; all phosphatase genes examined significantly modulated expression of at least one endogenous chemokine or cytokine.


Reversible protein phosphorylation is an essential regulatory mechanism in many cellular processes. Cells use this post-translational modification to alter the activity or localization of key regulatory proteins. Tyrosine and serine/threonine protein phosphatases are highly abundant proteins present in many cellular compartments in mammalian cells. Together with kinases, they set the phosphorylation state of signaling and effector proteins and thereby play a large role in controlling cellular responses. Inappropriate or defective phosphatase or kinase activity leads to aberrant patterns of phosphorylation. Dramatic changes in phosphorylation of many proteins were demonstrated during global ischemia, including enriched phosphatase activity in reactive astrocytes (Hasegawa et al., 2000). To date there has not been a systematic examination of phosphatase activity in astrocytes.

Here we report a large-scale classification of phosphatases focused on their control of NF-κB-mediated transcriptional activity. Nineteen phosphatases were identified to participate in either up- or down-regulation of NF-κB activity in astrocytes. Most of these phosphatases were not previously known to associate with this pathway. The involvement of additional phosphatases can not be excluded as rigid criteria and a high threshold of NF-κB activity were used to identify candidate genes. Stimulus and cell specificity, compensatory or redundant pathways, and the presence of non-functional siRNAs may cause additional underestimates of the number of phosphatase genes involved in NF-κB transcriptional activity.

At least 13 phosphatases were previously implicated in NF-κB signaling, including PPP2CA (Yang et al., 2001), PPM1B (Prajapati et al., 2004), PPM1L (Li et al., 2003; Takaesu et al., 2003), INPP4A (Franke et al., 1997; Romashkova and Makarov, 1999), PTEN (Mayo et al., 2002), PTPN2 (Ibarra-Sanchez et al., 2001), PPP4C (Hu et al., 1998), CDC25B (Zheng et al., 2004), PPP6C (Bouwmeester et al., 2004), PPP2R1A (Zheng et al., 2004), PPP2R1B (Zheng et al., 2004), PPP2R5C (Moreno et al., 2004), and DUSP5 (Zheng et al., 2004). Nine of these genes were also identified by the present analysis although the mechanisms by which most of these phosphatase genes impact NF-κB signaling are poorly understood. The four genes missed in our screen include DUSP5, however, the murine homolog of DUSP5 has not been identified. Silencing CDC25 phosphatases, which are critical to mitotic entry, markedly inhibited Renilla luciferase activity suggesting damage to the target cells; therefore, analysis of CDC25B was not pursued. RNAi to PPP6C inhibited basal NF-κB reporter activity but failed to meet the threshold established for our screening. PPM1B (also termed PP2Cβ) bound and dephosphorylated IKK in human HeLa and 293 embryonic kidney cells. However, we failed to detect any activity of PPM1B on NF-κB activity in mouse astrocytes (Table 1), even though the RNAi constructs effectively inhibited mRNA levels (Supplementary Fig. 2Z) and modulated NF-κB reporter activity in mouse fibroblasts (Fig. 1F). Reciprocally, 3 phosphatases that regulated NF-κB activity in astrocytes failed to modulate NF-κB reporter activity in fibroblasts. These results suggest potential cell type specificity in the activity of phosphatases on NF-κB signaling, an observation with potential implications for controlling inflammation in various clinical conditions.

PP2A enzymes regulate at least three different steps in the NF-κB pathway including TRAF2, IKK, and NF-κB p65 (Fig. 5C). Previous studies showed that the activity of IKK on IκB kinase was associated with PP2A and down regulated by the PP2A catalytic subunit (DiDonato et al., 1997; Fu et al., 2003). We observed selective non-redundant utilization of specific catalytic and structural chains in the core enzyme complexes, i.e. PPP2CB/PPP2R1A were selectively coupled to the IKK complex while PPP2CA/ PPP2R1B were physically and functionally associated with the p65 NF-κB complex (Fig. 2). Although the PP2A complex was shown to bind and dephosphorylate the p65 chain of NF-κB (Yang et al., 2001), there was no description of the composition of the PP2A enzyme. The present report functionally extends these observations by identifying PPP2CA and PPP2R1B as the NF-κB interactive chains (Fig. 2) and demonstrates the selective dephosphorylation of the Ser536 residue in the NF-κB p65 subunit. Our data suggest the potential of multiple corresponding site specific-phosphatases for NF-κB p65.

In addition, we identified a PP2A holoenzyme associated with TRAF2 (Fig. 3). Analysis of this interaction demonstrated Thr117 in the first TRAF2 zinc finger domain is a phosphorylation site and phosphorylation of Thr117 is required for TRAF2-mediated-NF-κB activity (Fig. 4). The present data also demonstrate ligand-induced phosphorylation of TRAF2 and suggest TRAF2 may be the target of the PP2A holoenzyme (Fig. 4). Future experiments will address the mechanisms involved in TRAF2 phosphorylation.

The PP2A chains combine in different combinations to form core enzymes and holoenzymes. In mice the PPP2CA and PPP2CB catalytic chains are 97% identical and the structural chains are 86% identical. However, PPP2CA null mutant mice were embryonic lethal, demonstrating that PPP2CA is an essential non-redundant gene (Gotz et al., 1998). In the present study co-immunoprecipitation showed non-redundant PP2A catalytic and structural chains were preferentially associated with their substrate. This suggests selective combinations of non-redundant PP2A catalytic and structural chains may be critical for substrate targeting.

Several phosphatases regulated basal NF-κB activity suggesting that NF-κB activity is tightly regulated and may be required for cellular homeostasis. Indeed, phosphorylation of p65 and its shuttling in and out of the nucleus have been observed in several cell types including astrocytes (Zhai et al., 2004). Basal NF-κB activity was reported to be critical for protecting cells from apoptosis (Bureau et al., 2002). Constitutive NF-κB activity has also been detected in glioblastomas and other tumors. The molecular mechanisms responsible for altered regulation of the NF-κB pathway in cancer cells remain largely unknown but some phosphatase genes (eg. PTPRJ and PPP2CB) identified in this report sensitize or promote cell death (MacKeigan et al., 2005) and therefore hold potential roles as tumor suppressors.

Astrocytes are implicated in the pathophysiology of neurodegenerative and inflammatory diseases including Alzheimer's disease and multiple sclerosis (Miller, 2005). These diseases are characterized by scarring lesions containing reactive hypertrophic astrocytes. These reactive astrocytes are a major source of chemokines that orchestrate migration and activation of leukocytes and microglial cells into neuronal lesions. The knowledge that phosphatases identified in this report can selectively regulate chemokine and cytokine expression (Fig. 6) offers new therapeutic targets with the potential of regulating inflammatory diseases.

Experimental Procedures

Mice and astrocyte isolation

BALB/cByJ mice (Jackson Laboratory, Bar Harbor, ME) were maintained in accordance with the guidelines of the Committee on Animals of the Harvard Medical School. Astrocytes were prepared from neonatal (<24 h) mice, as previously described (Luo et al., 2002). The purity of the primary astrocyte cultures was >95%, as determined by indirect immunofluorescence with anti-GFAP antibody (Dako, Carpinteria, CA) (Supplementary Fig. 3B).

NF-κB reporter gene design

The RNAi screen was based on the pLuc-MCS reporter (Stratagene, La Jolla, CA), which consists of a basic TATA element driving expression of a cDNA encoding the firefly luciferase gene. To optimize this assay in a 96-well plate format, we designed six pNF-κB-Luc-like reporters containing 1, 2, 6, 12, 18, or 24 NF-κB p65 binding sites, respectively. Although these reporters showed different basal activities, the reporter containing 6 binding sites exhibited the highest signal to noise ratio and was used in reporter screens.

siRNA vector design

We modified the dual siRNA retrovirus-based expression vector named pBabe-Dual (pBabe-puro with dual RNA polymerase III promoter). It contains two opposing RNA polymerase III promoters to drive expression of both strands of a template DNA cloned between the promoters. Both the H1 and U6 promoters were modified to contain a five Thymidine pol. III termination sequence at the -5 to -1 position and two Bbs I sites in the insertion (Fig. 1A). The target sequence for any mRNA can be cloned into pBabe-Dual and the DNA will be transcribed from both strands to form a double stranded RNA with two 3′ Uridine overhangs. The efficiency of inhibition of this siRNA vector was determined by RNAi experiments in which firefly luciferase and GFP were inhibited. Using this system RNAi-mediated knockdown of a positive regulator (NF-κB p65) suppressed TNFα-stimulated reporter activity while RNAi-knockdown of a negative regulator (IκBα) activated the reporter in the absence of stimulus or synergistically activated the reporter when induced by TNFα (data not shown).

siRNA library construction and screen

All 250 mouse phosphatase genes and putative phosphatase genes were chosen from the public UniGene library. siRNA target sequences in the gene were chosen using the siRNA design program from the Whitehead Institute web page Two siRNA target sequences were designed for each phosphatase gene using the following criteria: 1) the selected siRNA sequences for a given gene should not have more than 85% similarity to any other gene in mouse UniGene database using the Blast program; 2) Sequences were selected with 40-65% GC content. 3) No sequence containing four or more sequential bases of the same nucleotide was allowed; 4) No thermodynamically stable secondary structure (< 0 Kcal/mol). 5) A 5′-terminus on the anti-sense strand that is more AT-rich than the 3′-terminus. The siRNA sequences are available on request. A p6XNF-κB-Luc reporter plasmid was selected to screen for regulators of NF-κB transcriptional activation. A Renilla luciferase plasmid and two siRNA constructs for each gene target were combined and cotransfected into mouse astrocyte cells for screening. 48 hours after transfection, the cells were starved overnight and then stimulated for 6-8h with 10 ng/ml TNFα and luciferase activity was subsequently measured. To screen for regulators of basal NF-κB transcriptional activity, cells were not stimulated and luciferase activity was measured 72 h after transfection.

Data analysis

Normalized values (N), where N = (Firefly luciferase value)/(Renilla luciferase value) were calculated as described elsewhere (DasGupta et al., 2005). We chose this log transformation analysis because the data fit in a linear progression for both increases and decreases with respect to the plate average. Genes scoring >2 SD from the average [log(N)] were considered potential hits.

Cells and reagents

NIH3T3 cells were purchased from American Type Culture Collection (ATCC) (Manassas, VA). TRAF2-/- MEFs were kindly provided by Dr. Tak Mak (University of Toronto). Recombinant mouse and human TNFα were purchased from R&D Systems (Minneapolis, MN). Okadaic acid and Calyculin A, were obtained from Calbiochem (La Jolla, CA). DAPI was purchased from Sigma Chemical Co. (St. Louis, MO).

Anti-phospho-TRAF2 (Thr117) antibody was generated in rabbits (Convance, Denver, PA) using synthetic phospho-peptide CTWKGT*LKEYE (T*: phospho-T) conjugated to keyhole limpet hemocyanin as immunogen. Immune serum was passed through a phosphopeptide affinity column and washed with 0.1 M Tris (pH 8.0). Bound antibodies were eluted with 0.2 M glycine (pH 2.5) and neutralized with 1 M Tris (pH 8.0). Antibodies directed to IκBα, phospho-IκBα (Ser32/Ser36), IKKα, IKKβ, IKKγ, phospho-IKKα (Ser180)/β (Ser181), TRAF2, phospho-p65 (Ser276), and phospho-p65 (Ser536) were bought from Cell Signaling (Beverly, MA). Antibodies specific for myc or p65 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies specific for Flag were purchased from Sigma Chemical Co.

All phosphatase cDNAs used for over-expression were ordered from ATCC and tagged with the myc epitope in the pcDNA3.1-myc-His vector. IKKβ-Flag was kindly provided by Dr. R. B. Gaynor (Lilly Research Laboratories). IKKβ SSEE (Mercurio et al., 1997) was purchased from Addgene (Cambridge, MA).

Real time RT-PCR

mRNA was quantified using SYBR Green based real-time PCR. Total RNA was prepared using TRIzol Reagent (Invitrogen, Carlsbad, CA). Two μg RNA were transcribed into cDNA using 200 U Superscript II (Invitrogen Life Technologies). For one real-time reaction a 20 μl SYBR Green PCR reaction Mix (Roche Applied Science) was supplemented with 1/40 of the synthesized cDNA plus an appropriate oligonucleotide primer pair and run on the LightCycler II (Roche). Reverse transcriptase controls were done in parallel without adding enzyme. The comparative Cτ method was used to determine relative mRNA expression of examined genes as normalized by the β-glucuronidase housekeeping gene.

Cell transfection and luciferase activity assay

Astrocytes were transiently transfected with Lipofectamine 2000 (Invitrogen). 48 h later, the cells were starved overnight and then stimulated for the indicated time. Luciferase activity was determined as recommended by the manufacturer (Promega, Madison, WI). Values are expressed as mean ± SD of three experiments. Luciferase assays were performed using the Dual Luciferase reporter system (Promega). Relative luciferase units (RLU) were measured and normalized against Renilla luciferase activity 72 hr after transfection.

Immunoblotting, immunoprecipitation (IP) and immunocytochemistry

Cells were harvested and analyzed by Western blot. Protein concentration was determined by BCA protein assay kit (Pierce, Rockford, IL). 10 μg samples were loaded per lane. Blots were probed with the indicated antibody and immune complexes were detected by enhanced chemiluminescence (ECL) Plus (Amersham Pharmacia Biotech, Piscataway, NJ). For the NF-κB nuclear translocation assay, astrocytes were grown on glass slide chambers for 2 days after transfection. After TNFα treatment, cells were fixed in 4% paraformaldehyde and permeablized in phosphate-buffered saline (PBS) containing 0.1% Triton X-100. After blocking with 5% normal goat serum in PBS, cells were incubated with anti-p65 antibody, followed by incubation with Cy2 or Cy3 conjugated goat anti-rabbit IgG (Chemicon). Nuclei were stained with 100 ng/ml DAPI in PBS for 5 min. Immunoprecipitation kits (protein G) were purchased from Roche and immunoprecipitation was performed according to the manufacture's protocol.

Chemiluminescent transcription factor assays (CTFA)

EZ-Detect Transcription Factor Kit for NF-κB p65 was purchased from Pierce Biotechnology (Rockford, IL). Assays were performed according to the manufacturer's protocol.

Supplementary Material



This work was supported by NIH grant 1 RO1 NS42900 and NMSS grant RG2989B3/1. We thank Dr. T. Mak (University of Toronto) and Dr. R. B. Gaynor (Lilly Research Laboratories) for their generous gift of reagents.


  • Bouwmeester T, Bauch A, Ruffner H, Angrand PO, Bergamini G, Croughton K, Cruciat C, Eberhard D, Gagneur J, Ghidelli S, et al. A physical and functional map of the human TNF-alpha/NF-kappa B signal transduction pathway. Nat Cell Biol. 2004;6:97–105. [PubMed]
  • Bureau F, Vanderplasschen A, Jaspar F, Minner F, Pastoret PP, Merville MP, Bours V, Lekeux P. Constitutive nuclear factor-kappaB activity preserves homeostasis of quiescent mature lymphocytes and granulocytes by controlling the expression of distinct Bcl-2 family proteins. Blood. 2002;99:3683–3691. [PubMed]
  • Bush TG, Puvanachandra N, Horner CH, Polito A, Ostenfeld T, Svendsen CN, Mucke L, Johnson MH, Sofroniew MV. Leukocyte infiltration, neuronal degeneration, and neurite outgrowth after ablation of scar-forming, reactive astrocytes in adult transgenic mice. Neuron. 1999;23:297–308. [PubMed]
  • Chaudhuri A, Orme S, Vo T, Wang W, Cherayil BJ. Phosphorylation of TRAF2 inhibits binding to the CD40 cytoplasmic domain. Biochem Biophys Res Commun. 1999;256:620–625. [PubMed]
  • DasGupta R, Kaykas A, Moon RT, Perrimon N. Functional genomic analysis of the Wnt-wingless signaling pathway. Science. 2005;308:826–833. [PubMed]
  • Delhase M, Hayakawa M, Chen Y, Karin M. Positive and negative regulation of IkappaB kinase activity through IKKbeta subunit phosphorylation. Science. 1999;284:309–313. [PubMed]
  • DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E, Karin M. A cytokine-responsive IkappaB kinase that activates the transcription factor NF-kappaB. Nature. 1997;388:548–554. [PubMed]
  • Franke TF, Kaplan DR, Cantley LC, Toker A. Direct regulation of the Akt proto-oncogene product by phosphatidylinositol-3,4-bisphosphate. Science. 1997;275:665–668. [PubMed]
  • Fu DX, Kuo YL, Liu BY, Jeang KT, Giam CZ. Human T-lymphotropic virus type I tax activates I-kappa B kinase by inhibiting I-kappa B kinase-associated serine/threonine protein phosphatase 2A. J Biol Chem. 2003;278:1487–1493. [PubMed]
  • Garcia-Segura LM, Naftolin F, Hutchison JB, Azcoitia I, Chowen JA. Role of astroglia in estrogen regulation of synaptic plasticity and brain repair. J Neurobiol. 1999;40:574–584. [PubMed]
  • Gotz J, Probst A, Ehler E, Hemmings B, Kues W. Delayed embryonic lethality in mice lacking protein phosphatase 2A catalytic subunit Calpha. Proc Natl Acad Sci U S A. 1998;95:12370–12375. [PubMed]
  • Hasegawa S, Morioka M, Goto S, Korematsu K, Okamura A, Yano S, Kai Y, Hamada JI, Ushio Y. Expression of neuron specific phosphatase, striatal enriched phosphatase (STEP) in reactive astrocytes after transient forebrain ischemia. Glia. 2000;29:316–329. [PubMed]
  • Hayden MS, Ghosh S. Signaling to NF-kappaB. Genes Dev. 2004;18:2195–2224. [PubMed]
  • Hu MC, Tang-Oxley Q, Qiu WR, Wang YP, Mihindukulasuriya KA, Afshar R, Tan TH. Protein phosphatase X interacts with c-Rel and stimulates c-Rel/nuclear factor kappaB activity. J Biol Chem. 1998;273:33561–33565. [PubMed]
  • Ibarra-Sanchez MJ, Wagner J, Ong MT, Lampron C, Tremblay ML. Murine embryonic fibroblasts lacking TC-PTP display delayed G1 phase through defective NF-kappaB activation. Oncogene. 2001;20:4728–4739. [PubMed]
  • Kang J, Jiang L, Goldman SA, Nedergaard M. Astrocyte-mediated potentiation of inhibitory synaptic transmission. Nat Neurosci. 1998;1:683–692. [PubMed]
  • Kaykas A, Moon RT. A plasmid-based system for expressing small interfering RNA libraries in mammalian cells. BMC Cell Biol. 2004;5:16. [PMC free article] [PubMed]
  • Kim JM, Oh YK, Lee JH, Im DY, Kim YJ, Youn J, Lee CH, Son H, Lee YS, Park JY, Choi IH. Induction of proinflammatory mediators requires activation of the TRAF, NIK, IKK and NF-kappaB signal transduction pathway in astrocytes infected with Escherichia coli. Clin Exp Immunol. 2005;140:450–460. [PubMed]
  • Li MG, Katsura K, Nomiyama H, Komaki K, Ninomiya-Tsuji J, Matsumoto K, Kobayashi T, Tamura S. Regulation of the interleukin-1-induced signaling pathways by a novel member of the protein phosphatase 2C family (PP2Cepsilon) J Biol Chem. 2003;278:12013–12021. [PubMed]
  • Li QQ, Bever CT, Burt DR, Judge SI, Trisler GD. Induction of RANTES chemokine expression in human astrocytic cells is dependent upon activation of NF-kappaB transcription factor. Int J Mol Med. 2001;7:527–533. [PubMed]
  • Luo Y, Berman MA, Zhai Q, Fischer FR, Abromson-Leeman SR, Zhang Y, Kuziel WA, Gerard C, Dorf ME. RANTES stimulates inflammatory cascades and receptor modulation in murine astrocytes. Glia. 2002;39:19–30. [PubMed]
  • MacEwan DJ. TNF receptor subtype signalling: differences and cellular consequences. Cell Signal. 2002;14:477–492. [PubMed]
  • MacKeigan JP, Murphy LO, Blenis J. Sensitized RNAi screen of human kinases and phosphatases identifies new regulators of apoptosis and chemoresistance. Nat Cell Biol. 2005;7:591–600. [PubMed]
  • Mauch DH, Nagler K, Schumacher S, Goritz C, Muller EC, Otto A, Pfrieger FW. CNS synaptogenesis promoted by glia-derived cholesterol. Science. 2001;294:1354–1357. [PubMed]
  • Mayo MW, Madrid LV, Westerheide SD, Jones DR, Yuan XJ, Baldwin AS, Jr, Whang YE. PTEN blocks tumor necrosis factor-induced NF-kappa B-dependent transcription by inhibiting the transactivation potential of the p65 subunit. J Biol Chem. 2002;277:11116–11125. [PubMed]
  • Mercurio F, Zhu H, Murray BW, Shevchenko A, Bennett BL, Li J, Young DB, Barbosa M, Mann M, Manning A, Rao A. IKK-1 and IKK-2: cytokine-activated IkappaB kinases essential for NF-kappaB activation. Science. 1997;278:860–866. [PubMed]
  • Miller G. Neuroscience. The dark side of glia. Science. 2005;308:778–781. [PubMed]
  • Moreno CS, Ramachandran S, Ashby DG, Laycock N, Plattner CA, Chen W, Hahn WC, Pallas DC. Signaling and transcriptional changes critical for transformation of human cells by simian virus 40 small tumor antigen or protein phosphatase 2A B56gamma knockdown. Cancer Res. 2004;64:6978–6988. [PubMed]
  • Pomerantz JL, Baltimore D. NF-kappaB activation by a signaling complex containing TRAF2, TANK and TBK1, a novel IKK-related kinase. Embo J. 1999;18:6694–6704. [PubMed]
  • Prajapati S, Verma U, Yamamoto Y, Kwak YT, Gaynor RB. Protein phosphatase 2Cbeta association with the IkappaB kinase complex is involved in regulating NF-kappaB activity. J Biol Chem. 2004;279:1739–1746. [PubMed]
  • Prat A, Biernacki K, Wosik K, Antel JP. Glial cell influence on the human blood-brain barrier. Glia. 2001;36:145–155. [PubMed]
  • Romashkova JA, Makarov SS. NF-kappaB is a target of AKT in anti-apoptotic PDGF signalling. Nature. 1999;401:86–90. [PubMed]
  • Sakurai H, Suzuki S, Kawasaki N, Nakano H, Okazaki T, Chino A, Doi T, Saiki I. Tumor necrosis factor-alpha-induced IKK phosphorylation of NF-kappaB p65 on serine 536 is mediated through the TRAF2, TRAF5, and TAK1 signaling pathway. J Biol Chem. 2003;278:36916–36923. [PubMed]
  • Schwaninger M, Petersen N, Prinz S, Sallmann S, Neher M, Spranger M. Adenosine-induced expression of interleukin-6 in astrocytes through protein kinase A and NF-IL-6. Glia. 2000;31:51–58. [PubMed]
  • Song H, Stevens CF, Gage FH. Astroglia induce neurogenesis from adult neural stem cells. Nature. 2002;417:39–44. [PubMed]
  • Sun SC, Maggirwar SB, Harhaj E. Activation of NF-kappa B by phosphatase inhibitors involves the phosphorylation of I kappa B alpha at phosphatase 2A-sensitive sites. J Biol Chem. 1995;270:18347–18351. [PubMed]
  • Takaesu G, Surabhi RM, Park KJ, Ninomiya-Tsuji J, Matsumoto K, Gaynor RB. TAK1 is critical for IkappaB kinase-mediated activation of the NF-kappaB pathway. J Mol Biol. 2003;326:105–115. [PubMed]
  • Takeuchi M, Rothe M, Goeddel DV. Anatomy of TRAF2. Distinct domains for nuclear factor-kappaB activation and association with tumor necrosis factor signaling proteins. J Biol Chem. 1996;271:19935–19942. [PubMed]
  • Van Wagoner NJ, Benveniste EN. Interleukin-6 expression and regulation in astrocytes. J Neuroimmunol. 1999;100:124–139. [PubMed]
  • Vermeulen L, De Wilde G, Van Damme P, Vanden Berghe W, Haegeman G. Transcriptional activation of the NF-kappaB p65 subunit by mitogen- and stress-activated protein kinase-1 (MSK1) Embo J. 2003;22:1313–1324. [PubMed]
  • Viatour P, Merville MP, Bours V, Chariot A. Phosphorylation of NF-kappaB and IkappaB proteins: implications in cancer and inflammation. Trends Biochem Sci. 2005;30:43–52. [PubMed]
  • Volterra A, Meldolesi J. Astrocytes, from brain glue to communication elements: the revolution continues. Nat Rev Neurosci. 2005;6:626–640. [PubMed]
  • Yang J, Fan GH, Wadzinski BE, Sakurai H, Richmond A. Protein phosphatase 2A interacts with and directly dephosphorylates RelA. J Biol Chem. 2001;276:47828–47833. [PMC free article] [PubMed]
  • Zhai Q, Luo Y, Zhang Y, Berman MA, Dorf ME. Low nuclear levels of nuclear factor-kappa B are essential for KC self-induction in astrocytes: requirements for shuttling and phosphorylation. Glia. 2004;48:327–336. [PubMed]
  • Zheng L, Liu J, Batalov S, Zhou D, Orth A, Ding S, Schultz PG. An approach to genomewide screens of expressed small interfering RNAs in mammalian cells. Proc Natl Acad Sci U S A. 2004;101:135–140. [PubMed]
  • Zhong H, SuYang H, Erdjument-Bromage H, Tempst P, Ghosh S. The transcriptional activity of NF-kappaB is regulated by the IkappaB-associated PKAc subunit through a cyclic AMP-independent mechanism. Cell. 1997;89:413–424. [PubMed]