Intracellular Redistribution of Nucleolin upon Interaction with the CD3ε Chain of the T Cell Receptor Complex

T cell activation through the antigen receptor (TCR) involves the cytoplasmic tails of the CD3 subunits CD3γ, CD3δ, CD3ε, and CD3ζ. Whereas the biological significance of the cytoplasmic tails of these molecules is suggested, in part, by their evolutionarily conserved sequences, their interactions with signal transduction molecules are not completely understood. We used affinity chromatography columns of glutathione S-transferase fused to the CD3ε cytoplasmic tail to isolate proteins that specifically interact with this subunit. In this way, we identified the shuttling protein nucleolin as a specific CD3ε-interacting molecule. Using competition studies and affinity chromatography on peptide columns, we were able to identify a central proline-rich sequence as the nucleolin-interacting sequence in CD3ε. Transfection in COS cells of wild type CD3ε, but not of nonbinding mutants of CD3ε, resulted in redistribution of nucleolin from the nucleus and nucleoli to the cytoplasm. This property was transferred to a CD8 protein chimera by appending the cytoplasmic tail of CD3ε. We also found that nucleolin associated with the TCR complex. This association was increased upon TCR engagement, suggesting that the CD3ε/nucleolin interaction may have a role in T cell activation.

T cells respond to antigen via a polypeptide complex composed of ligand-binding T cell receptor (TCR) 1 ␣ and ␤ chains (or ␥ and ␦ in ␥␦ T cells) and the CD3 subunits CD3␥, CD3␦, CD3⑀, and CD3 (1,2). Unlike the TCR chains, the CD3 components have long cytoplasmic tails that associate with cytoplasmic signal transduction molecules. This association is mediated at least in part by a double tyrosine-based motif present in a single copy in the CD3␥, CD3␦, and CD3⑀ chains and in three copies in CD3 (3). This motif, named immune-receptor tyrosine-based activation motif (ITAM), becomes tyrosine phosphorylated during T cell activation by the Src family proteintyrosine kinases Lck and/or Fyn (4 -6). Tyrosine phosphoryl-ated ITAM become docking sites for the Syk family proteintyrosine kinase ZAP70 and other signal-transducing molecules. It is well established that antibody-mediated engagement of protein chimeras containing the cytoplasmic tail of either CD3 or CD3⑀ results in T cell activation (7)(8)(9)(10). These data indicate that the cytoplasmic tail of one of these subunits can be sufficient to induce T cell activation. Regarding the role of CD3 subunits in T cell activation, most of the attention has been focused on the ITAM. However, the cytoplasmic tails of the CD3 subunits contain other evolutionarily conserved features that suggest ITAM-independent roles for them.
The CD3⑀ cytoplasmic tail, highly conserved (11,12), can be tentatively subdivided into three regions; the N-terminal region contains a basic amino acid cluster, the central region contains a proline-rich sequence, and the C-terminal region contains the ITAM (13). The proline-rich sequence contains the SH3-binding consensus motif XPPXP, and the C-terminal region contains the YXXLXXR endoplasmic reticulum (ER) retention sequence, which partially overlaps the ITAM (14,15). Previous attempts to identify proteins that associate with the cytoplasmic tail of CD3⑀ have shown the specific interaction of a nuclear protein, topoisomerase II␤, and a tyrosine-phosphorylated protein, CAST, with the N-terminal region of the CD3⑀ tail (13,16).
Nucleolin is a major nucleolar protein of exponentially growing eukaryotic cells that is directly involved in the regulation of ribosome biogenesis and maturation (17,18). Nucleolin has a molecular mass of 100 -110 kDa and is mainly found in the fibrillar components of the nucleoli where it associates with nascent preribosomal RNA. Numerous reports have implicated the involvement of nucleolin either directly or indirectly in the regulation of cell proliferation and growth, cytokinesis, replication, embryogenesis, and nucleogenesis (17,18). Although predominantly localized in the nucleolus, nucleolin has also been found in the cytoplasm and at the plasma membrane, where it can function as a cell surface receptor for ligands as different as coxsackie B viruses and the complement inhibitor factor J (18 -20). Because nucleolin acts as a shuttling protein between the cytoplasm and the nucleus, it might provide a mechanism for extracellular regulation of nuclear events.
Nucleolin activity is regulated by proteolysis, methylation, ADP-ribosylation, and phosphorylation by casein kinase II, Cdc2, PKC, cyclic AMP-dependent protein kinase, and ectoprotein kinase (17,18). Nucleolin is cleaved by a leupeptinsensitive protease that is tightly associated with it. It has been also suggested that nucleolin itself may possess a self-cleaving activity.
In an attempt to identify novel CD3⑀ tail-interacting proteins, we have utilized affinity chromatography using glutathione S-transferase (GST) ⑀ columns. In this way, we were able to characterize nucleolin as a major CD3⑀-interacting protein that associates with the central proline-rich region. We also show herein that expression of CD3⑀ in a heterologous cell system results in loss of both nucleolin localization in the nucleolus and redistribution to the cytoplasm. A possible role of nucleolin/CD3⑀ interaction in TCR-mediated T cell activation is proposed.

EXPERIMENTAL PROCEDURES
Cells and Reagents-The COS-7 African green monkey cell line was grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum (Sigma). The human leukemic T cell line Jurkat was grown in RPMI 1640 medium supplemented with 5% fetal bovine serum.
The mouse monoclonal anti-human nucleolin antibody D3 (21) used in this study was a gift from Dr. B. Ballou (University of Pittsburgh, PA). The mouse monoclonal anti-CD8␣ B9.4 was donated by Dr. B. Malissen (Center d'Immunologie, Marseille-Luminy, France). The mouse monoclonal antibody SP34, specific for the extracellular domain of human CD3⑀ (22), was a gift from Dr. C. Terhorst (Beth Israel Deaconess Hospital, Boston, MA). The mouse monoclonal anti-human CD3 antibody UCHT1 was donated by Dr. P. Beverley (The Edward Jenner Institute for Vaccine Research, Berkshire, UK). Peptides 7 and 8, corresponding to amino acids 170 -185 and amino acids 150 -166 of human CD3⑀ respectively, were synthesized by the N-(9-fluorenyl)methoxycarbonyl (Fmoc) method and purified by HPLC.
DNA Constructs-To generate the GST⑀ fusion protein, a 165-base pair cDNA fragment, corresponding to the whole cytoplasmic tail of human CD3⑀ (amino acids 131-185), was generated by polymerase chain reaction. This fragment was digested and inserted into the XhoI and NotI sites of the plasmid pGEX4T3 (Amersham Pharmacia Biotech). The truncated CD8 construct was prepared by polymerase chain reaction by introducing a stop codon after the second cytoplasmic amino acid of human CD8␣. The polymerase chain reaction product was cloned into the XhoI and BamHI sites of the pSR␣ expression vector. The CD8/⑀ construct expressing the extracellular and transmembrane domains of human CD8␣ fused to the cytoplasmic tail of CD3⑀ has been previously described (23) and was a gift from Dr. C. Terhorst.
Affinity Chromatography-To characterize proteins that interact with the cytoplasmic tail of CD3⑀, a GST⑀ column was generated by absorbing 20 ml of GST⑀-producing Escherichia coli lysate (resulting from a 1-liter culture) to 1 ml of glutathione-Sepharose (Amersham Pharmacia Biotech). A similar preabsorb column was prepared from GST-producing E. coli. A total of 4 ϫ 10 9 Jurkat cells were lysed in 1% Nonidet P-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 20 mM Tris-HCl, pH 7.8, 10 mM iodoacetamide, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin). A postnuclear supernatant of the Jurkat cell lysate was serially passed first through the GST and then through the GST⑀ columns. Both columns were washed with 100 ml of lysis buffer. GST-and GST⑀-bound proteins were eluted in lysis buffer containing 10 mM glutathione (eluate 1). The eluates were dialyzed against lysis buffer and passed through new GST and GST⑀ columns. These second columns were first eluted with 20 mM triethylamine, pH 11.0 (eluate 2), and then with 10 mM glutathione in lysis buffer (eluate 3). The three eluates were subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore). Protein bands on the membrane were visualized by Coomassie Blue staining, excised and subjected to trypsin digestion in situ. The resulting peptides were purified by HPLC and were sequenced either by the Edman degradation method or by electrospray mass spectrometry.
For affinity chromatography on peptide columns, 10 mg of peptides 7 or 8 were coupled to 1-ml Hi-Trap N-hydroxysuccinimide-activated agarose columns (Amersham Pharmacia Biotech). A postnuclear cell lysate from 4 ϫ 10 9 Jurkat cells in lysis buffer was passed first through the peptide 7 column and subsequently through the peptide 8 column. After washing with 2 column volumes of lysis buffer, bound proteins were eluted in 50 mM triethylamine, pH 11.0. The eluates were neutralized with 100 mM Tris-HCl, pH 7.4, and subjected to SDS-PAGE.
COS Cell Transfections and Immunoprecipitation-COS cell transfections and immunoprecipitation were performed as previously described (12).
Immunofluorescence and Microscopy-COS cells were fixed and stained for immunofluorescence as previously described (12).
Immunoselection-CD8 ϩ COS cells transfected with either CD8/⑀ or truncated CD8 were detached from the culture plates using phosphatebuffered saline and repeated pipetting. After incubating the cells with 4 g/ml B9.4 antibody on ice for 1 h, they were washed and incubated with goat anti-mouse IgG-coated magnetic beads (Dynal) at a 3:1 bead to cell ratio for 30 min. The CD8 ϩ cells were isolated with a magnet (Dynal), washed with phosphate-buffered saline, and plated on culture dishes.

Identification of Nucleolin as a Specific CD3⑀-binding Pro-
tein-To characterize proteins that specifically associate with the cytoplasmic tail of CD3⑀, a construct for the expression of a GST⑀ fusion protein containing the whole cytoplasmic tail of CD3⑀ was made. GST⑀ protein purified from E. coli extracts was absorbed to glutathione-Sepharose columns. A cell lysate from the human T cell line Jurkat was first preabsorbed to a GST column and then passed through the GST⑀ column. Bound proteins were eluted with glutathione, dialyzed, and absorbed again to GST and GST⑀ columns to increase specificity. Proteins bound to the second columns were eluted first with a high pH buffer and finally with glutathione. A number of proteins were absorbed to GST⑀ but not to GST (Fig. 1A, lanes 1 and 2). These protein bands of 110, 97, 90, 68, and 47 kDa (indicated as arrows a, b, c, d, and e) were excised, eluted, and fragmented, and partial amino acid sequences were obtained. The fact that proteins d and e reassociated with the GST⑀ column and resisted extraction with 50 mM triethylamine, pH 11 (Fig. 1A, lane 6), suggests that their interaction with the cytoplasmic tail of CD3⑀ is quite strong. Proteins a, c, and d produced sequences that were identical with nucleolin and products of its partial proteolysis (Fig. 1A). Protein bands b and e yielded no sequence. To confirm that nucleolin associates to the cytoplasmic tail of CD3⑀, the eluates from the GST and GST⑀ columns were immunoblotted with a specific monoclonal anti-nucleolin antibody. This antibody reacted with a 110-kDa protein band (protein a) present in the GST⑀ eluate but not in the GST eluate (Fig. 1B). These results showed that nucleolin and its partial proteolytic fragments specifically associate with the cytoplasmic tail of CD3⑀ in vitro.
Nucleolin Binds to the Central Proline-rich Region of the Cytoplasmic Tail of CD3⑀-To map the binding site of nucleolin on the cytoplasmic tail of CD3⑀, we used the monoclonal antibody APA1/1, which binds a well defined sequence within this tail. Using a GST⑀ column to pull down CD3⑀-interacting proteins from [ 35 S]methionine-labeled Jurkat cell lysates, we detected several major protein bands, including nucleolin and its partial degradation product of 68 kDa, and actin ( Fig. 2A,  mock). Incubation of the cell lysate with APA1/1 completely inhibited the binding to GST⑀ of the 110-and 68-kDa nucleolin forms but not the binding of actin or other contaminant proteins ( Fig. 2A, APA1/1). A 75-kDa GST⑀-associated protein was also completely displaced by APA1/1, but this protein probably represents a partial proteolysis product of nucleolin. This result indicated that the antibody APA1/1 specifically inhibits the association of nucleolin with the cytoplasmic tail of CD3⑀. In previous work (12), we mapped the binding site of APA1/1 to a 10-amino acid region in the central proline-rich region of the tail of CD3⑀ (Fig. 2B). This suggested that the binding site of nucleolin maps to the central region of the tail of CD3⑀. To confirm this, a competition experiment was set up using a 17-mer synthetic peptide (peptide 8) that expands the APA1/1binding site. Peptide 8 but not a control peptide of the same length expanding the C-terminal region of CD3⑀ (peptide 7) inhibited binding of nucleolin forms to GST⑀ (Fig. 2A). Additional evidence that nucleolin binds the central proline-rich region of the tail of CD3⑀ was the finding that nucleolin binds to a column of immobilized peptide 8 but not of peptide 7 (Fig.  2C).
Expression of CD3⑀ in an Heterologous Cell System Promotes Intracellular Redistribution of Nucleolin-To determine whether the observed interaction of nucleolin with the cytoplasmic tail of CD3⑀ resulted in a change in the intracellular distribution of these proteins, CD3⑀ was transfected into COS cells, and the localization of transfected CD3⑀ and endogenous nucleolin was assessed by two-color immunofluorescence. Nucleolin was found in the nucleus and the nucleolus (Fig. 3A, red staining), whereas CD3⑀ was detected in the cytoplasm and nuclear membrane in an ER-characteristic pattern (Fig. 3A, green staining). Interestingly, both stainings were mutually exclusive, i.e. nucleolin staining was not observed in CD3⑀expressing cells. To determine whether the effect of CD3⑀ expression on nucleolin distribution correlated with its capacity to interact with nucleolin, different CD3⑀ mutants were assayed. Unlike wild type CD3⑀, transfection with a truncated (tail-less) CD3⑀ construct did not alter nucleolin distribution (Fig. 3B). This inferred that expression of the cytoplasmic tail of CD3⑀ was necessary for nucleolin redistribution and suggested that the effect of CD3⑀ on nucleolin is mediated by its ability to interact with it. However, because the deletion of CD3⑀ tail resulted in a loss of its ER retention (Fig. 3B), it is also possible that the effect on nucleolin requires the localization of CD3⑀ in the ER rather than direct binding to nucleolin. To discriminate between these possibilities, mutant 9, a deletion mutant that lacks 10 amino acids of the central, prolinerich region of CD3⑀ (Fig. 2B) was assayed. This mutant has lost the capacity to interact with the antibody APA1/1 (12). Like wild type CD3⑀, mutant 9 was also located in the ER (Fig. 3B). However, mutant 9 did not alter the nucleolar location of nucleolin, strongly suggesting that redistribution of nucleolin requires direct binding to CD3⑀. Indeed, in cells that overexpress CD3⑀, nucleolin was found to colocalize with CD3⑀ in the ER (Fig. 3B). These results indicate that CD3⑀ requires nucleolin binding capacity for its effect on nucleolin distribution. Nevertheless, localization of CD3⑀ to the ER seems to be re-quired as well, because transfection of two C-terminal deletion mutants of CD3⑀ that result in loss of ER retention (14) did not cause CD3⑀ redistribution (data not shown).
Expression of a Protein Chimera Containing the Cytoplasmic Tail of CD3⑀ Results in Redistribution of Nucleolin-To determine whether expression of the cytoplasmic tail of CD3⑀ is sufficient to enable nucleolin relocalization, a protein chimera consisting of the cytoplasmic tail of CD3⑀ appended to the transmembrane and extracellular domains of CD8␣ (CD8/⑀) was obtained (23). As a control, a truncated mutant of CD8␣ lacking the cytoplasmic tail was used (CD8t). Although the CD8/⑀ construct is in part retained in the ER because the cytoplasmic tail of CD3⑀ contains an ER retention sequence (14,15), both CD8/⑀ and CD8t were expressed on the cell surface. This allowed separation of transfected COS cells from untransfected cells by immunoselection with antibody-coated magnetic beads. As anticipated, the expression of the CD8/⑀ chimera in the magnetic bead-selected COS cell population (65% CD8/⑀ ϩ ) led to the localization of nucleolin in the cytoplasm (Fig. 4). In contrast, in the nonselected population (95% CD8/⑀ Ϫ ) nucleolin was located in the nucleus and nucleolus. Expression of CD8t did not alter the nuclear localization of nucleolin (Fig. 4). This indicated that the cytoplasmic tail of CD3⑀ was sufficient to promote the intracellular redistribution of nucleolin.
Antibody-mediated TCR Cross-linking Increases Nucleolin Recruitment to the TCR-To determine whether the CD3⑀/ nucleolin interaction takes place in T cells and whether the interaction changes upon TCR engagement, the human T cell line Jurkat was stimulated with the anti-CD3 antibody UCHT1 followed by cross-linking with a secondary antibody. Mockstimulated and stimulated cells were lysed, and immunopre- cipitation was carried out with the anti-CD3⑀ antibody SP34. Immunoblotting of the SP34 immunoprecipitates with antinucleolin antibody showed that nucleolin is associated to the TCR complex in nonstimulated T cells (Fig. 5). The association was increased in Jurkat cells stimulated with anti-CD3 antibody. These results show that the TCR complex, probably via CD3⑀, interacts with nucleolin in T cells and that more nucleolin is recruited to the TCR complex when this is cross-linked with antibodies, suggesting that the CD3⑀/nucleolin interaction may have a role in T cell activation. DISCUSSION These results show that the cytoplasmic tail of human CD3⑀ interacts with nucleolin in vitro. We mapped the site of interaction to a central 10 -17-amino acid proline-rich sequence within the cytoplasmic tail. Recently, Saito's group (13,16) has described the interaction of the cytoplasmic tail of CD3⑀ with two other proteins, topoisomerase II␤ and CAST, a tyrosine phosphorylated protein. Both proteins interact with the N-terminal part of the cytoplasmic tail of CD3⑀, a region rich in basic amino acids. Therefore, the cytoplasmic tail of CD3⑀ appears to interact with different proteins along its sequence: with topoisomerase II␤ and CAST in the N-terminal region, ]methionine-labeled Jurkat cells was incubated with GST⑀-Sepharose beads in the absence (mock) or presence of 100 g/ml APA1/1 antibody, 100 g/ml peptide 7, or 100 g/ml peptide 8. The samples were subjected to SDS-PAGE, and the gel was dried and exposed to the PhosphorImager. The positions of nucleolin and the 68-kDa nucleolin fragment as well as the position of the contaminant protein actin are indicated. Compared with mock-treated beads, incubation with APA1/1 reduced nucleolin binding by 95%. Competition by peptide 8 resulted in a 67% inhibition, whereas peptide 7 was noninhibitory (22%). B, sequence of the cytoplasmic tail of human CD3⑀ indicated in the one-letter code. The sequence deleted in mutant 9 as well as the APA1/1 binding site is shown in bold type. The sequences corresponding to peptides 7 and 8 are also indicated. C, binding of nucleolin to peptide columns. A postnuclear Jurkat cell lysate was incubated with peptides 7 and 8 immobilized on agarose columns. Bound and eluted proteins were resolved by SDS-PAGE and immunoblotted with the anti-nucleolin antibody.

FIG. 3. Relocalization of nucleolin in CD3⑀-transfected COS cells.
A, COS cells transfected with wild type CD3⑀ were stained with the anti-nucleolin antibody (red fluorescence) and an anti-CD3⑀ antibody (green fluorescence). The image shows a 0.5-m-thick optical section taken at mid-distance from the coverslip in the confocal microscope. Note that CD3⑀-expressing cells show no staining of nucleolin. B, effect of mutations in the cytoplasmic tail of CD3⑀ on nucleolin distribution. COS cells were transfected with a tail-less, truncated, CD3⑀ mutant, with a CD3⑀ deletion mutant lacking the nucleolin-binding site (del 9), or with wild type CD3⑀. Cells were stained for nucleolin and CD3⑀ and examined under fluorescence microscopy. Notice that the expression of truncated (del 9) CD3⑀ did not alter the nuclear distribution of nucleolin, whereas expression of wild type CD3⑀ resulted in redistribution of nucleolin to the cytoplasm. with nucleolin in the central portion, and via the ITAM (C terminus) with ZAP70 and probably other SH2-containing proteins (23,24).
Although at first glance it would seem unlikely that the cytoplasmic tail of CD3⑀ interacts with two nuclear proteins, topoisomerase II␤ and nucleolin, the interaction with both proteins might be facilitated by a possible location of CD3⑀ in the inner nuclear membrane (13). CD3⑀ contains a sequence at the N-terminal portion of the cytoplasmic tail reminiscent of a nuclear localization signal. Moreover, CD3⑀ has a double arginine sequence in the central portion of its cytoplasmic tail that is reminiscent of the signal sequence responsible for the localization of glycoprotein B of human cytomegalovirus (a transmembrane protein) in the inner nuclear membrane (25). Indeed, CD3⑀ has been located in the nucleus (13), although the roles of the nuclear localization signal and the presence of a nuclear inner membrane localization signal have not yet been demonstrated. Therefore, the intracellular location site of CD3⑀ for its interaction with nucleolin, and topoisomerase II␤, could conceivably be the nucleus.
A second possible location for the interaction of CD3⑀ with nucleolin could be the cytoplasm. Nucleolin has been shown to shuttle between the cytoplasm and the nucleus. Indeed, the interaction of nucleolin with several cytoplasmic proteins and even plasma membrane proteins has been reported (18 -20). We have shown in this study that nucleolin interacts with the TCR complex in T cells, probably through CD3⑀. Although we cannot discriminate whether nucleolin interacts with the TCR at the plasma membrane or with the intracellular pool of TCR, the fact that antibody engagement of the TCR results in increased recruitment of nucleolin suggests that nucleolin interacts with the TCR at the plasma membrane. In addition, our present results reveal that expression of CD3⑀ in COS cells results in the loss of nuclear localization of nucleolin and, in some cases, in relocalization to the cytoplasm. This effect is dependent on the expression of the nucleolin-interacting sequence in the central portion of CD3⑀ and can be transferred by appending the CD3⑀ tail to the extracellular and transmembrane domains of an irrelevant protein. Thus, the correlation between nucleolin-binding capacity and nucleolin redistribution induced by CD3⑀ indicates that CD3⑀ promotes nucleolin relocalization by directly binding nucleolin.
Similar to the effect of CD3⑀ expression, it has been described that infection by poliovirus causes a relocalization of nucleolin to the cytoplasm, perhaps by binding of nucleolin to the 3Ј noncoding region of poliovirus RNA (26). However, although the role of nucleolin binding to poliovirus RNA seems to be that of promoting assembly of new virions, the role of CD3⑀induced relocalization of nucleolin is not well understood yet. For the topoisomerase II␤-CD3⑀ interaction, it has been proposed that it could be involved in signal transduction because topoisomerase II inhibitors up-regulate IL-2 production and apoptosis (13). Therefore, by binding topoisomerase II␤, CD3⑀ could participate in TCR-induced growth arrest and apoptosis of T cells. It has also been described that nucleolin binds specifically to a Jun N-terminal kinase response element and that this binding is required for interleukin-2 mRNA stabilization induced by T cell activation signals (27). Our observation that the association of nucleolin to the TCR is increased upon antibody-mediated cross-linking of the TCR suggests that nucleolin/CD3⑀ interaction may have a role in T cell activation. Although a positive effect on T cell activation cannot be excluded, we favor the hypothesis that recruitment of nucleolin to the TCR through CD3⑀ and redistribution of nucleolin to the cytoplasm may have roles in TCR-induced growth arrest, given the important roles for cell survival and proliferation that nucleolin plays at the nucleus (17,18). FIG. 4. Effect of CD3⑀ tail expression on nucleolin levels and localization. Expression of a CD8/⑀ chimera but not of truncated CD8 resulted in redistribution of nucleolin to the cytoplasm. COS cells transfected with the CD8/⑀ chimera were immunoselected. The selected and nonselected populations were stained with anti-CD8 and anti-nucleolin antibodies and examined under fluorescence microscopy. Notice that in the selected CD8/⑀-expressing COS cells nucleolin is distributed to the cytoplasm, whereas in the nonselected population, nucleolin is located in the nucleus. In COS cells expressing truncated CD8 nucleolin remained in the nucleus.
FIG. 5. TCR engagement increases nucleolin association to the TCR complex. Jurkat cells were stimulated with a combination of the anti-CD3 antibody UCHT1 and a cross-linking second antibody for 5 min (ϩ stimulus) or left untreated (Ϫ stimulus). The cells were then lysed and immunoprecipitation (Ip) was carried out with anti-CD3⑀ antibody SP34. Immunoprecipitates were resolved by SDS-PAGE and immunoblotting was performed with the anti-nucleolin antibody. A sample of the total lysate was run in parallel as a control. NIS, precipitation with nonimmune serum. H, immunoglobulin heavy chain.