A novel interaction between the juxtamembrane region of the p55 tumor necrosis factor receptor and phosphatidylinositol-4-phosphate 5-kinase.

Tumor necrosis factor-α (TNF-α) binding to its receptors leads to a diversity of biological responses. The actions of TNF are the result of the interaction of cytoplasmic proteins that bind directly to the intracellular domains of the two TNF receptors, p55 and p75. Here we report a novel interaction between the juxtamembrane region of the p55 TNF receptor and a newly discovered 47-kDa isoform of phosphatidylinositol-4-phosphate 5-kinase (PIP5K), a member of the enzyme family that generates the key signaling messenger, phosphatidylinositol 4,5-bisphosphate. The interaction was found to be specific for the p55 TNF receptor and was not observed with the p75 TNF receptor, the Fas antigen, or the p75 neurotrophin receptor, which are other members of the TNF receptor superfamily. In vitro experiments using recombinant fusion proteins verify the authenticity of the interaction between the p55 receptor and PIP5KIIβ, a new isoform of PIP5K, but not the previously identified 53-kDa PIP5KIIα. Treatment of HeLa cells with TNF-α resulted in an increased PIP5K activity. These results indicate that phosphatidylinositol turnover may be linked to stimulation of the p55 TNF receptor and suggest that a subset of TNF responses may result from the direct association of PIP5KIIβ with the p55 TNF receptor.

Tumor necrosis factor-␣ (TNF-␣) 1 initiates its proliferative, differentiative, or cytotoxic actions on mammalian cells by binding to two transmembrane molecules, the p55 and p75 TNF receptors (1)(2)(3). The p55 receptor is responsible for many of the biological effects of TNF, including programmed cell death, cell differentiation, and cell proliferation (4 -6). A major step in understanding the mechanism of the p55 receptor has been the identification of the interacting protein, TRADD, which accounts for signals leading to apoptosis and increased gene expression through NF-B-mediated events (7). Likewise, the p75 receptor is capable of signal transduction through the association of ring finger proteins, such as TRAF1 and TRAF2, with the cytoplasmic domain of p75 (8,9).
The interaction of TRADD with the p55 TNF receptor has revealed the importance of protein-protein interactions via a region of homology called the "death domain." This sequence has been found in a variety of transmembrane and cytosolic molecules and is usually localized at the C-terminal region of each protein. The functional significance of this domain has been demonstrated in studies with the p55 TNF receptor and the Fas antigen, which contain similar functional death domain sequences (10). The binding of p55 receptors to TRADD, a cytoplasmic protein containing a death domain, and the binding of the Fas antigen to FADD, an analogous protein, have been localized to an 80-amino acid region at the C terminus of both receptors. While overexpression of TRADD or FADD in heterologous cells leads to cell death (7,11), deletions or mutations in the death domains abolish the ability of these molecules to participate in the initiation of apoptosis (6,12).
A link to the interleukin converting enzyme/ced-3 protease family was made recently with the identification of a cysteine protease, FLICE/MACH, which interacts with the death effector domain of FADD (13,14). The connection of the Fas antigen and the p55 TNF receptor with a member of the interleukin converting enzyme protease family provides a mechanism for how cell death signals are initiated by ligand-receptor interactions.
While cell death can be instigated by TNF-␣ and represents the major signaling function of the Fas antigen, TNF-␣ can participate in many other diverse activities, including the synthesis of proinflammatory mediators and cell proliferation and differentiation (15), neuroprotection (16), and synaptic transmission (17). The striking interactions of the C-terminal half of the intracellular domain of the p55 receptor raise the issue of the functional significance of the juxtamembrane region of the p55 receptor. This region is ϳ100 amino acids in length, is rich in proline residues, and is not required for TNF-mediated cellular cytotoxicity. Here we report the interaction of the juxtamembrane region of the p55 TNF receptor with phosphatidylinositol-4-phosphate 5-kinase (PIP5K), an enzyme that produces phosphatidylinositol 4,5-bisphosphate (PIP 2 ). PIP5K is a significant enzymatic activity to be linked to TNF signaling since its product, PIP 2 , is a critical second messenger intermediate that also has many direct modulatory effects. Moreover, this study has identified a new member of the PIP5K family, PIP5KII␤. The association of the p55 receptor with PIP5K indicates how diverse functions are encoded in the TNF receptor structure to generate multiple signal transduction events after ligand binding.

EXPERIMENTAL PROCEDURES
Cell Culture-HeLa, U373, and MCF7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. In experiments using TNF, the cells were transferred to serumfree medium when an 80% confluency was reached. The cells were serum-starved for 12-24 h before TNF was added at a final concentration of 100 ng/ml. Cells (10-cm dish) were harvested at appropriate time intervals, washed three times in cold phosphate-buffered saline, and lysed in 500 l of cold lysis buffer (20 mM Tris, pH 7.5, 25 mM ␤-glycerophosphate, 137 mM NaCl, 100 mM EDTA, 1% Triton X-100, 2 mM sodium pyrophosphate, 100 M sodium vanadate, 0.5 mM LiCl, 1 g/ml leupeptin, 2 g/ml aprotinin, and 25 g/ml phenylmethylsulfonyl fluoride). The cell debris was pelleted after a 20-min incubation on ice. The lysates were stored frozen at Ϫ70°C.
For each PCR fragment, the reaction mixture (100 l) contained the primers at 1 ng/l; the DNA template at 0.1 ng/l; 200 nM each dATP, dCTP, dGTP, and dTTP; 1 ϫ Pfu buffer; and 1.25 units of cloned Pfu polymerase (Stratagene). The PCR conditions for Y1, Y2, and Y3 were as follows: 1 cycle at 85°C for 8 min followed by 5 cycles at 94°C for 2 min, annealing temperature of 30°C for 1.5 min, and extension temperature of 72°C for 1 min and 25 cycles at 94°C for 2 min, 60°C for 1.5 min, and 72°C for 1 min. This was followed by a final cycle of 72°C for 4 min. For the Fas antigen, the PCR conditions were similar, except the annealing temperature was 50°C. For the p75 TNF receptor, the annealing temperature was 55°C. The DNA templates were pCMVp55 and pCMVp75 (19) and pCMVmFas (20). These conditions yielded PCR fragments of the expected sizes. The PCR fragments were phenol/ chloroform-extracted, ethanol-precipitated, and digested with NotI and SpeI before ligation in the lexA DNA-binding yeast expression vector. Transformants were screened by restriction digestion and verified by DNA sequencing.
Yeast Growth and Transformation-Saccharomyces cerevisiae strain S260, which contains a lexA operator-lacZ reporter gene, was maintained as described (18). The baits were transformed into S. cerevisiae (21), and the transformants were selected and maintained on Trp Ϫ plates and Trp Ϫ medium. Log phase cells (10 10 ) were cotransformed with 150 g of murine cerebellar library DNA constructed in the VP16 activation domain according to published procedures (21). The cotransformants were plated on Amersham Hybond-N filters laid on Ura Ϫ Trp Ϫ plates containing 2% glucose. After incubation at 30°C for 48 h, the filters were transferred to similar plates containing 2% galactose and further incubated for 24 h to induce expression of the VP16-cDNA fusion proteins encoded by the library plasmids. Interacting proteins were detected by a qualitative colony ␤-galactosidase activity assay as described (18). Colonies positive for ␤-galactosidase were streaked on Ura Ϫ Trp Ϫ plates. The yeast plasmids were rescued by transforming into Escherichia coli and were subjected to DNA sequencing analysis (Sequenase).
Generation and Purification of GST Fusion Proteins-The cyto-plasmic (residues 204 -426), juxtamembrane (residues 204 -337), and death domain (residues 340 -426) regions of human p55 were amplified using PCR. The primers used were as follows: for the cytoplasmic domain, 5Ј-primer (5Ј-AAAGGATCCATGTATCGCTAC-3Ј) and 3Јprimer (5Ј-GAGTCGACCTCTGAGAAGACT-3Ј); for the juxtamembrane region, 5Ј-primer (5Ј-AAAGGATCCTGTATCGCTAC-3Ј) and 3Јprimer (5Ј-GAGTCGACCTCACACGTTCTC-3Ј); and for for the death domain, 5Ј-primer (5Ј-AAAGGATCCTTGCGCTGGAAG-3Ј) and 3Јprimer (5Ј-GAGTCGACCTCTGAGAAGACT-3Ј). A BamHI linker was incorporated into the 5Ј-primers, and a SalI linker was incorporated into the 3Ј-primers. PCRs were carried as described above using 30 cycles with a melting temperature of 94°C for 2 min, an annealing temperature of 48°C for 1 min, and an extension temperature of 72°C for 2 min. The PCR fragments were digested with BamHI and SalI and ligated into pGSTag (22). Large-scale GST fusion proteins expressed from the various constructs were purified according to published procedures (23). For making GST and MBP fusion proteins of clone 11 (c11), the yeast library plasmid corresponding to c11 was used as a template to amplify the library insert. In both cases, the 5Ј-primer (5Ј-AAAG-GATCCGGTGGGGAATTCCCA-3Ј) and the 3Ј-primer (5Ј-TATCTA-GACTAGTCTAATCGATCTCGAGCCA-3Ј) had a BamHI linker incorporated in the 5Ј-primer. The fragment was PCR-amplified as described above using the conditions described for the p75 TNF receptor. The fragment was digested with BamHI and XbaI and ligated separately into pGSTag and pMAL-c2 (New England Biolabs Inc.). The positive clones were tested for their ability to yield fusion proteins of the expected sizes after isopropyl-1-thio-␤-D-galactopyranoside induction (0.5 mM). The GST fusion protein was purified as described (23), and the MBP fusion protein was purified using the protocol provided by New England Biolabs Inc.
Generation of Anti-GST-c11 and Anti-MBP-c11 Antibodies-The GST-c11 and MBP-c11 fusion proteins were denatured in SDS and used as immunogens to generate polyclonal antibodies (Pocono Farms).
Isolation of PIP5KII␤ cDNA-In a data base search, the PIP5KII␣ sequence (GenBank TM /EMBL accession number U14957) matched the HFBEP52 (GenBank TM /EMBL accession number T07883) expressed sequence tag from the human infant brain cDNA library with 70.5% identity. This 2.7-kb clone was obtained and sequenced (Sequenase Version 2.0, U. S. Biochemical Corp.). A 2.0-kb EcoRI fragment corresponding largely to the 3Ј-untranslated region of this clone was used to screen 8 ϫ 10 5 plaques from a human fetal brain DR2 cDNA library (CLONTECH). The plaques were transferred to nitrocellulose filters and hybridized with the probe at 42°C in 5 ϫ SSPE, 5 ϫ Denhardt's solution, 50% formamide, 0.1% SDS, and 100 g/ml salmon sperm DNA. These filters were washed at high stringency (65°C, 0.01 ϫ SSC and 0.01% SDS) for 20 min. Eleven positive clones were obtained, plaque-purified, and converted to their respective phagemids. Upon sequencing, two clones were found to extend the sequence of HFBEP52 in both the 5Ј-and 3Ј-directions. Additional sequencing of the 5Ј-region was done on an Applied Biosystems Model 373 sequencer at the University of Wisconsin Biotechnology Center. This revealed a putative start site within the context of a Kozak consensus sequence (24) that was preceded by upstream stop codons in all three reading frames. This evidence demonstrated that a full-length open reading frame for PIP5KII␤ had been obtained.
Northern Blotting-A human multiple tissue Northern blot (CLON-TECH) was probed with a 1.7-kb KpnI fragment of HFBEP52, which lies entirely within the 3Ј-untranslated region of PIP5KII␤. The DNA fragment was gel-purified following restriction enzyme digestion and labeled by random priming with [␣-32 P]ATP and Klenow DNA polymerase. Hybridization and washing were performed as outlined by the manufacturer, and the membrane was exposed to film overnight.
Expression of PIP5KII␤ in E. coli-The 1.
2-kb open reading frame of PIP5KII␤ was amplified by PCR (30 cycles at 94°C for 1 min, 55°C for 1 min, and 72°C for 3 min followed by an extension temperature of 72°C for 10 min) using native Pfu polymerase (Stratagene) and the following primers: 5Ј-GCCGCCGCGGATCCCATGTCGTCC-3Ј (forward) and 5Ј-GCTGAAGGTACTCGAGAACTACG-3Ј (reverse). The forward primer was designed to anneal over the start codon (in boldface) and to introduce a BamHI site (underlined) just upstream of the start site, while the reverse primer was designed to anneal over the stop codon (in boldface) and to introduce a XhoI site (underlined) just downstream of the termination codon. These PCR products were sucloned into the BamHI and XhoI sites of pBluescript SK(Ϫ) (Stratagene). The entire insert was sequenced to confirm that no errors had been introduced by PCR. The BamHI/XhoI fragment was then subcloned into pET28b (Novagen) for the purpose of expressing a hexahistidine fusion version of PIP5KII␤ (His-PIP5KII␤).
E. coli strain BL21(DE3) was transformed with either pET28b containing the PIP5KII␤ open reading frame or just pET28b. For the initial expression, these strains were grown at 37°C to A 600 ϭ 0.6, induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside, and incubated at 37°C for an additional 3 h. The cells were pelleted and resuspended in 50 mM Tris-HCl, pH 8.0, and 2 mM EDTA. Next, 4000 units of Ready-lyse solution (Epicentre Technologies Corp.) was added to the resuspended cells, and Triton X-100 was added to a final concentration of 0.1%. After incubating the cells at 30°C for 15 min, they were sonicated at 70 watts twice for 10 s. These lysates were centrifuged at 12,000 ϫ g for 15 min at 4°C, and the supernatant was decanted. The pellet was dissolved in 1 ϫ SDS-PAGE sample buffer, while an aliquot of the supernatant was mixed with an equal volume of 2 ϫ sample buffer. All samples were boiled for 10 min and analyzed by SDS-PAGE in the presence of 250 mM ␤-mercaptoethanol.
His-PIP5KII␤ was purified by Ni 2ϩ chelate chromatography. The strain containing the His-PIP5KII␤ open reading frame was grown at 30°C to A 600 ϭ 0.6, induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside, and incubated at 30°C for an additional 3 h. The cells were centrifuged and resuspended in 8 ml of 1 ϫ binding buffer (20 mM Tris-HCl, pH 7.9, 500 mM NaCl, and 5 mM imidazole). After sonication with three 20-s pulses at 70 watts, the extract was centrifuged at ϳ39,000 ϫ g for 20 min at 4°C and passed through a 0.45-m syringe filter. An aliquot of this extract was set aside, and the rest was purified on a 2.5-ml gravity-packed column of His-Bind resin (Novagen). Bound proteins were eluted in three 5-ml fractions. Each of these fractions was dialyzed overnight at 4°C against 2.0 liters of phosphate-buffered saline.
Kinase Activity Assays-The supernatants of the crude fraction of His-PIP5KII␤ and of partially purified His-PIP5KII␤ were assayed for phosphatidylinositol-4-phosphate 5-kinase activity as described previously (25) with some modifications. Kinase activity was assayed in 50-l reactions done for 20 min at room temperature in a final concentration of 50 mM Tris-HCl, pH 7.6, 0.5 mM EGTA, 10 mM MgCl 2 , 160 M phosphatidylinositol 4-phosphate (Sigma), 50 M ATP, and 10 Ci of [␥-32 P]ATP. The lipids were extracted, and the labeled products were separated by thin layer chromatography and detected by autoradiography.
Western Blotting-Proteins were transferred to nitrocellulose (Micron Separations, Inc.) following SDS-PAGE. Antibodies were incubated with the membrane for 1 h at room temperature following blocking with 5% powdered milk in phosphate-buffered saline containing 0.2% Tween 20. The primary antibody was raised against the MBP-c11 fusion protein and was detected by chemiluminescence using a horseradish peroxidase-conjugated antibody (Santa Cruz Biotechnology, Inc.) and LumiGLO substrates (Kierkegaard and Perry Laboratories, Inc.).
In Vitro Binding-GST fusion proteins containing the cytoplasmic, the juxtamembrane, or the death domain of the p55 fusion protein were purified on glutathione-Sepharose beads, and 100 l of the fusion protein on the glutathione beads was incubated with 3 g of purified MBP-c11. The reaction volume was brought up to 500 l with 50 mM Tris, pH 7.5, 200 mM NaCl, 20 mM EDTA, and 0.1% Nonidet P-40. After incubation at 4°C for 1 h, the slurry was washed three times with the same buffer. The slurry was then boiled in SDS-PAGE buffer, and the samples were loaded on a 10% SDS-polyacrylamide gel and electrophoresed at 100 V until the dye front reached the bottom of the gel. The gel was then transferred to an Immobilon-P transfer membrane (Millipore Corp.), and immunoblotting was carried out using chemiluminescence. The primary antibody was the rabbit anti-MBP antibody (New England Biolabs Inc.) used at a 1:1000 dilution. The secondary antibody was a peroxidase-conjugated goat anti-rabbit IgG (Sigma) used at a 1:15,000 dilution.
Immunoprecipitation-Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Near confluency, the cells were washed three times with phosphate-buffered saline, scraped, and centrifuged. The cells were lysed in 500 l of radioimmune precipitation assay buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton, and 0.4% SDS) supplemented with protease inhibitors. Lysates were incubated on ice for 30 min; the cell debris was removed by centrifugation; and an aliquot of the lysate was used to determine protein concentration using the Bio-Rad reagent. Lysates (4 mg) were subjected to immunoprecipitation in a final volume of 500 l of radioimmune precipitation assay buffer. Antibodies against PIP5K (protein A-purified) that were coupled to Affi-Gel-10 beads (Bio-Rad) were added to the lysate, and the mixture was incubated at 4°C for 1 h. In separate reactions, lysates were incubated either with 100 l of protein A-purified anti-p55 antibodies (Genzyme Corp.) coupled to Affi-Gel-10 beads or with nonspecific goat serum. The immunoprecipitate was washed three times in radioimmune precipitation assay buffer and boiled for 30 s in SDS-PAGE buffer. The sample was subjected to electrophoresis on a 10% SDS-polyacrylamide gel. Chemiluminescence was carried out using the ECL procedure (Amersham Corp.).

RESULTS
The p55 TNF receptor is a transmembrane protein with four extracellular cysteine-rich repeats and an intracellular domain of 222 amino acids. While the sequences responsible for mediating cytotoxicity (6) and aggregation of the p55 receptor (26) have been localized to the death domain in the C-terminal half of the cytoplasmic tail, the functional significance of the juxtamembrane region has not been fully defined.
To identify cellular proteins that bound specifically to the juxtamembrane domain of the p55 TNF receptor, a recombinant LexA fusion protein was generated that contained amino acids 204 -280 (Y1), beginning at the end of the transmembrane domain and extending to the middle of the cytoplasmic domain. Y1 did not contain sequences representing the death domain of p55. The Y1 domain was then used as the bait in the yeast two-hybrid screen.
A mouse cerebellar cDNA library was used to screen for proteins that interacted specifically with the Y1 region. The library was engineered in a yeast expression vector containing VP16 as the activation domain (18). Colonies were obtained following cotransformation of the S260 yeast strain with the Y1-LexA construct, and the cDNA library colonies were analyzed by ␤-galactosidase activity measurements. Of 1.5 ϫ 10 7

FIG. 1. Specificity of the TNF receptor interaction.
Proteins representing the cytoplasmic domain of p55 were tested with truncated PIP5K (clone 11) in the yeast two-hybrid system. Y1, Y2, Y3, Y4, and Y5 represent different cytoplasmic regions of the p55 TNF receptor (p75-TNFR). The colony assay for ␤-galactosidase activity of the cotransformants indicates the interaction between the bait, cloned in the lexA DNA-binding domain, and c11, one of the library clones constructed in the VP16 activation domain. The entire cytoplasmic domains of other receptors in the same family were also used in the specificity tests. WT, wild-type; TM, transmembrane domain; p75-NGFR, p75 nerve growth factor receptor.
cotransformants, 20 cDNAs were found to be positive using ␤-galactosidase activity as a measure of the activation of the Y1-LexA construct. After tests for specificity and DNA sequence analysis were conducted, one positive clone of 0.7 kb in size (c11) was pursued for further analysis.
The yeast two-hybrid assay was further used to test the specificity of the interaction. For this purpose, four other LexA fusion proteins containing different regions of p55 were generated. These proteins are schematically represented in Fig. 1. Each construct (called Y2, Y3, Y4, and Y5) encoded different segments of the intracellular domain of p55, including the juxtamembrane region, the C-terminal death domain (Y3), or both regions (Y4, Y5). The interaction of c11 with the p55 receptor was found to be confined to sequences representing the juxtamembrane region. Cotransformation of yeast strain S260 with each of these constructs and the plasmid harboring c11 rescued from the initial screening indicates that p55 con-structs containing the death domain (Y3) or LexA alone did not yield any ␤-galactosidase activity, whereas other constructs containing p55 juxtamembrane sequences (Y1, Y4, and Y5) gave positive ␤-galactosidase activity (Fig. 1).
This analysis was also extended to the cytoplasmic regions of the p75 TNF receptor, the Fas antigen, and the p75 neurotrophin receptor. No interactions were detected between c11 and LexA constructs containing these receptor sequences (Fig. 1). In support of this observation is the fact that there is little similarity in sequence between these family members in the juxtamembrane region. Therefore, the yeast two-hybrid results indicate that the protein expressed from the c11 cDNA specifically associates with the juxtamembrane domain of the p55 TNF receptor.
The cDNA Encodes Phosphatidylinositol-4-phosphate 5-Kinase-Sequence analysis revealed that the mouse c11 cDNA clone encoded a protein that was highly homologous to human  (24) and is preceded by an in-frame stop codon. The open reading frame is 1248 base pairs, encoding a protein of 416 residues. The in-frame stop codons are in boldface. The underlined sequence was used for Northern blotting, while the sequence in italics was used to screen the DR2 cDNA library. The shaded sequence corresponds to c11, detected via the yeast two-hybrid screen. B, the translated amino acid sequences of PIP5KII␤ and PIP5KII␣ are compared with the 202-amino acid protein generated from the c11 clone. The alignment was generated using the PileUp program (Genetics Computer Group). Residues conserved in at least two sequences are in boxed in black.

FIG. 2-continued
PIP5KII cloned previously (25). The c11 cDNA sequence showed 77% (protein level: 83%) identity to the published sequence of PIP5K cloned from a human placental cDNA library (Fig. 2B). The PIP5K enzymes have been defined as types I and II on the basis of their elution from a phosphocellulose ion exchange column. These kinases differ in size, kinetic properties, and differential sensitivity to heparin, spermine, and phosphatidic acid (27,28). The placental PIP5K sequence, encoded by a 4.1-kb mRNA, may now be identified as the type II␣ isoform (25), which was previously isolated as a 53-kDa enzyme (28,29).
A search of the sequence data bases with the DNA sequence of PIP5KII␣ revealed a putatively transcribed sequence that was highly identical to the query sequence. This clone, HF-BEP52, was obtained, sequenced, and used to screen a human fetal brain DR2 cDNA library. Several different cDNAs were obtained, and the composite sequences from these partial cDNAs predicted an open reading frame of 1248 base pairs, encoding a 416-amino acid protein with a calculated molecular mass of 47,378 Da ( Fig. 2A). Based on this composite sequence, the full-length open reading frame was amplified by PCR from the human fetal brain DR2 cDNA library. The open reading frame was found to be 77.8% identical to PIP5KII␣ at both the nucleotide and protein levels.
Based on this similarity, this new clone has been designated as PIP5KII␤. The murine clone c11 isolated in the yeast twohybrid screen is 90.3% identical to PIP5KII␤ at the nucleotide level and 99.5% identical at the amino acid level. Because of the level of identity, the c11 cDNA corresponds to a newly identified isoform of PIP5K, referred to as PIP5KII␤. A comparison of the amino acid sequences of PIP5KII␣, PIP5KII␤, and c11 appears in Fig. 2B.
Using a probe representing the 3Ј-untranslated region of HFBEP52, a discrete 6.3-kb mRNA, distinct from the ␣-isoform 4.1-kb mRNA (25), was detected (Fig. 3). Identical results were obtained by hybridization with the c11 cDNA (data not shown). Hence, PIP5KII␤ is encoded by a 6.3-kb mRNA, while PIP5KII␣ is represented by a 4.1-kb mRNA. Northern blot analysis also revealed that a probe derived from the 5Ј-coding region of PIP5KII␣ also detected the 6.3-kb transcript at lower stringency (data not shown). This observation was expected, considering the degree of identity between the two isoforms. Abundant levels of messages were found in heart, placenta, kidney, and pancreas, whereas lung and liver displayed lower levels. Although both the 4.1-and 6.3-kb transcripts are highly expressed in brain, a major difference is that skeletal muscle is more enriched for ␤-isoform mRNAs than the ␣-isoform.
Enzymatic Activity-To verify the enzymatic activity of PIP5KII␤, a cDNA containing the coding region of PIP5KII␤ was expressed as a hexahistidine fusion protein (Fig. 4A). The recombinant His-PIP5KII␤ protein was partially purified by Ni 2ϩ affinity chromatography and found to be ϳ51 kDa, as measured by SDS-PAGE (Fig. 4B). This molecular mass is ϳ3 kDa larger than the native enzyme due to the hexahistidine tag and linker sequence, which would make the size of the wildtype protein 47 kDa. This is consistent with the calculated molecular mass of 47.4 kDa. The purified His-PIP5KII␤ protein exhibited phosphatidylinositol-4-phosphate 5-kinase activity (Fig. 4C). As a control, lysates from cells containing the empty vector lacked kinase activity (Fig. 4C), whereas lysates expressing PIP5KII␤ displayed enzymatic activity (data not shown).
In Vitro Interaction-To verify the interaction of PIP5KII␤ with the p55 receptor in vitro, two separate approaches were taken. In the first case, the truncated PIP5KII␤ cDNA (c11) isolated from the yeast two-hybrid screen was fused in frame with the coding sequences for MBP. The resulting fusion protein was expressed and purified from E. coli. To test for binding in vitro, the purified MBP-PIP5KII␤ fusion protein (Fig. 5A, right panel, lane 4) was incubated with glutathione-Sepharose beads and GST fusion proteins representing different regions of the p55 TNF receptor. After incubation, the reaction was extensively washed, separated by SDS-PAGE, and analyzed by Western blotting using antibodies against MBP. These antibodies do not cross-react with the GST protein alone, PIP5K, or p55 receptor proteins (data not shown). The PIP5KII␤ fusion protein interacted with GST fusion proteins containing the juxtamembrane region (residues 204 -337) and the entire cytoplasmic region (residues 204 -426) of the p55 receptor (Fig. 5A). The full-length p55 cytoplasmic fusion protein was somewhat less reactive. However, the fusion protein containing the p55 death domain did not exhibit any binding to MBP-PIP5KII␤. This provides independent evidence that the interaction between PIP5KII␤ and p55 TNF receptor is restricted to the juxtamembrane region.
In a separate approach, an analogous experiment was undertaken with recombinant His-PIP5KII␤ or His-PIP5KII␣ proteins (Fig. 5B, lanes 5 and 6). The immunoblot indicates that PIP5KII␤ binds to the cytoplasmic region, and not to the death domain. No interaction was detected between p55 cytoplasmic sequences and PIP5KII␣. Taken together, these results suggest that the juxtamembrane region of the p55 TNF receptor interacts specifically with PIP5KII␤.
Co-immunoprecipitation-To determine if the p55 TNF receptor interacts with PIP5KII␤ in vivo, extracts of two different cell lines, MCF7 and HeLa cells, were prepared and immunoprecipitated either with antibodies against the p55 receptor or with antibodies directed against MBP-c11 coupled to Affi-Gel-10 beads. Following immunoprecipitation with anti-p55 antibodies, Western blot analysis using anti-PIP5KII␤ (c11) antibodies was carried out. A 47-kDa protein, representing PIP5KII␤, was detected after immunoprecipitation with anti-p55 antibodies, indicating that the p55 TNF receptor directly associated with PIP5KII␤ in HeLa and MCF7 cells (Fig. 6A). Further immunoblotting verified that this 47-kDa isoform was expressed in the cell lines (Fig. 6B). Mock immunoprecipitation reactions with nonspecific goat serum (Fig. 6C) indicated the co-immunoprecipitation of PIP5K and p55 was specific. These results indicate that PIP5KII␤ is associated with the p55 TNF receptor in two TNF-responsive cell lines.
Activation of PIP5K-PIP5K is a pivotal enzyme in phosphoinositide metabolism since it gives rise to PIP 2 , the parent molecule for the production of 1,2-diacylglycerol, inositol 1,4,5trisphosphate, and phosphatidylinositol 3,4,5-trisphosphate. These lipid second messengers are involved in mitogenic responses to polypeptide growth factors and G proteins through phospholipase C␤ (30,31). To investigate whether the enzymatic activity of PIP5K is relevant to TNF-mediated signaling, the activity of PIP5K was determined in cell lysates of TNF responsive cells, HeLa, MCF7, and U373.
Cells were treated with 100 ng/ml TNF-␣, and lysates were prepared as described under "Experimental Procedures." To determine the activity of the ␤-isoform, cell lysates were first immunodepleted of the PIP5KII␣ isoform using antibody sc-1330 (Santa Cruz Biotechnology, Inc.) raised against the Nterminal peptide of PIP5KII␣. Western blot analysis indicated that the immunodepleted lysates contained the ␤-isoform (data not shown). Using phosphatidylinositol 4-phosphate as a substrate in the presence of [␥-32 P]ATP, the activity of PIP5K was measured (Fig. 7). After 30 min of TNF treatment, a significant increase in the level of PIP 2 was detected. The ligand-dependent activation of PIP5KII␤ in TNF-responsive cells demonstrates that one potential signaling mechanism for TNF-␣ may be in the induction of the phosphatidylinositol pathway. DISCUSSION Multiple signaling pathways have been characterized for TNF-␣, a prominent cytokine produced by macrophages. Since TNF mediates divergent responses ranging from inflammatory, cytotoxic, and metabolic functions, it is likely that many second messengers are responsible for TNF-dependent signaling. These may include increased tyrosine phosphorylation, production of ceramide from the hydrolysis of sphingomyelin, activation of the mitogen-activated kinase cascade and p38 stress-activated kinase, production of arachidonic acid, and activation of protein kinase C and phosphatidylcholine-specific phospholipase C (3,15). The mechanism of TNF in initiating cell death has been clarified by the identification of receptor adaptor proteins, such as TRADD, which are required for apoptotic signal transduction. The binding of the p55 receptor to TRADD occurs via domains located at the C terminus of both proteins, which can lead to the recruitment of FLICE/MACH proteases (13,14).
Here we have identified a distinctive protein interaction between the juxtamembrane region of the p55 TNF receptor and a phosphatidylinositol lipid kinase, PIP5KII␤. This is the first identified enzymatic activity directly associated with the TNF receptor. The PIP5K enzyme is responsible for the phosphorylation of phosphatidylinositol 4-phosphate, giving rise to PIP 2 . PIP 2 can be further hydrolyzed to inositol 1,4,5-trisphosphate and 1,2-diacylglycerol by phospholipase C, or it can be phosphorylated by phosphatidylinositol 3-kinases to generate another lipid second messenger, phosphatidylinositol 3,4,5trisphosphate. The interaction of the TNF receptor with PIP5K therefore reveals a potential link to several potent second messenger systems and cellular activities, including vesicular transport, endocytosis, and lysosomal function (32).
In addition to the activation of protein kinase C by diacylglycerol and the mobilization of intracellular calcium by inositol 1,4,5-trisphosphate, phosphoinositide metabolites contribute to other cellular activities that may be relevant to TNF signaling. The binding of PIP 2 to pleckstrin homology domains may influence localization of pleckstrin homology domain-containing proteins to the membrane. PIP 2 binds to cytoskeletal proteins such as gelsolin and profilin that can influence cytoskeletal remodeling (33). Also, clathrin coat-associated proteins, such as AP-2, bind avidly to inositol polyphosphates (34). In the case of AP-2, the binding of inositol phosphate blocks its clathrin coat assembly properties.
Other growth factor receptor systems have been linked to PIP5K activities. EGF stimulation of A431 cells results in an increase in the activity of lipid kinases associated with the cytoskeleton (35). Indeed, co-immunoprecipitation experiments using antibodies against the EGF receptor indicate that phosphatidylinositol 4-kinase and PIP5K activities are directly associated with the EGF receptor. Interestingly, these lipid kinases also interact in the juxtamembrane region of the EGF receptor near the ATP-binding site of the catalytic tyrosine kinase domain (36). Treatment with EGF increases the activities of these phosphoinositide kinases, phosphatidylinositol 4-kinase and PIP5K. One potential function of this association may be to aid the mitogenic responses to EGF stimulation (37).
A mitogenic role for PIP5K is also supported by evidence showing that monoclonal antibodies against PIP 2 block cell proliferation in response to platelet-derived growth factor and bombesin (38) and are growth inhibitory for S. cerevisiae (39). Moreover, PIP5K activity has been directly linked to proliferation and malignancy (40). The increase in PIP5K activity observed following TNF treatment suggests that the known mitogenic effects of TNF upon cells (41) may be facilitated by increased PIP 2 production through the association of PIP5K with the p55 receptor.
How can the same receptor system give rise to such divergent biological responses such as cell death and cell proliferation? Through its interaction with the p55 and p75 receptors, FIG. 6. Coimmunoprecipitation of PIP5KII␤ with the p55 TNF receceptor. Lysates from HeLa and MCF7 cells were prepared, and 2 mg of lysate was immunoprecipitated with anti-p55 TNF receptor antibodies (Genzyme Corp.) (A), with anti-PIP5KII␤ antibodies generated from the MBP-c11 fusion protein coupled to Affi-Gel-10 beads (B), or with nonspecific serum (C). The immunoprecipitates (IP) were run on an SDSpolyacrylamide gel, transferred to polyvinylidene difluoride membrane, and immunoblotted with anti-PIP5KII␤ antibodies.
FIG. 7. PIP5K activity of HeLa, MCF7, and U373 cells treated with TNF-␣. Cells were treated with 100 ng/ml TNF-␣ for 0, 5, 10, 15, 30, and 60 min, and lysates were prepared (total). The lysates were immunodepleted of PIP5KII␣ and subjected to PIP5K measurements using phosphatidylinositol 4-phosphate as a substrate. PIP 2 generated from the reaction (see "Experimental Procedures") was visualized by autoradiography. The values for the total lysates versus the depleted lysates were normalized relative to their respective zero time points. Similar results were obtained in duplicate assays. TNF can potentially generate second messengers, such as ceramide (42) and proinflammatory metabolites (15). The activity of PIP5K versus other signaling activities may determine whether cells respond to TNF by cell proliferation or apoptosis. Discrete actions by TNF may be dictated by the presence or absence of cellular proteins, such as TRADD or PIP5K. Other proteins, such as FADD, RIP, and TRAF2, can be recruited as a result of these initial interactions with TRADD (43,44). When multiple proteins are present, the affinity of binding with the receptor will undoubtedly dictate the choice of the signaling pathway. It is important to note that the binding of PIP5KII␤ to the p55 receptor was not observed with other family members, such as Fas and the p75 neurotrophin receptor, and that PIP5KII␤ interacted preferentially over PIP5KII␣ (Fig. 5C). These interactions would be expected to take place only in specific cell types, thereby providing the diversity in signaling potential by this cytokine family.
Additionally, phospholipase C␥ (45,46) and phosphatidylinositol 3-kinase (47) contain intrinsic SH3 domains or are associated with other SH3 domain-containing proteins. These SH3 domains interact with proline-rich regions and are thought to form specific protein-protein interactions. Both PIP5KII␣ (amino acids 307-329) and PIP5KII␤ (amino acids 323-344) contain proline-rich clusters that are highly homologous to the consensus sequence of SH3 domain targets (48,49). Such an interaction between phospholipase C␥ or phosphatidylinositol 3-kinase and PIP5Ks could account for the observed coupling of PIP 2 production to its utilization by these enzymes. Similar interactions involving signal-generating enzymes have been observed for many mitogenic receptors (50,51).
The PIP5K enzymes exist in multiple isoforms, which are found in brain, placenta, and erythrocytes. These isoforms have various immunoreactivities, kinetic properties, and molecular masses (28,52,53). Moreover, PIP5Ks are unique in that they possess almost no homology to kinase motifs present in other phosphatidylinositol, protein, and lipid kinases (25). Two yeast genes, MSS4 and FAB1, display a similar catalytic domain as PIP5KII. Interestingly, these genes are involved in cell cycle progression and chromosome segregation in S. cerevisiae. PIP5K is likely to play a central role in phosphoinositide signal transduction for a variety of growth conditions. Recently, it has been shown that another phosphatidylinositol kinase is indirectly involved in TNF-promoted signaling (54). Stimulation of 3T3-L1 adipocytes with TNF promoted the phosphorylation of the IRS-1 insulin receptor substrate, which can bind to phosphatidylinositol 3-kinase. The results reported here demonstrate a direct interaction of the p55 receptor with a specific member of the PIP5K family, which is involved in many signaling pathways (55).
The finding of a specific association between PIP5KII␤ and the p55 TNF receptor and the stimulation of PIP 2 production following TNF production suggest that this receptor system may be closely linked to the control of protein kinase C and calcium-dependent enzymatic activities. Moreover, the involvement of phospholipase C␥ and phosphatidylinositol 3-kinase in the breakdown and modification of PIP 2 suggests that receptor tyrosine kinase-linked pathways may be biochemically and physically linked to the action of cytokine receptors, such as those for TNF-␣. The regulated synthesis and breakdown of PIP 2 may serve as an important crossroad to merge several seemingly diverse signal transduction pathways. Further investigation into the cellular functions regulated by activation of phosphatidylinositol kinases will likely provide insights into the mechanisms of how receptor-generated signals are transduced and integrated.