The T-Cell Receptor ζ Chain Contains Two Homologous Domains with Which Simian Immunodeficiency Virus Nef Interacts and Mediates Down-Modulation

Abstract

We have recently demonstrated that simian immunodeficiency virus (SIV) Nef binds to the ζ chain of the T-cell receptor (TCR), leading to its down-modulation from T-cell surfaces (I. Bell, C. Ashman, J. Maughan, E. Hooker, F. Cook, and T. A. Reinhart, J. Gen. Virol. 79:2717–2727, 1998). Using a panel of human as well as rhesus macaque TCR ζ cytoplasmic domain mutants, we have identified in this report two linear peptides in the cytoplasmic domain of TCR ζ which independently interact with SIV Nef. Each SIV Nef interaction domain was sufficient in the absence of the other for interaction with SIV Nef in a yeast two-hybrid assay. In parallel, we demonstrated that Nef down-modulation of CD8-TCR ζ fusion proteins containing full-length or truncated portions of the TCR ζ cytoplasmic domain occurs in transiently transfected 293T cells. Furthermore, using proteins expressed in Escherichia coli, a glutathione S-transferase–Nef fusion protein coprecipitated histidine-tagged portions of the TCR ζ cytoplasmic domain which contained SNID-1 or SNID-2. The peptides targeted by SIV Nef, YNELNL and YSEIGMKGERRR, are portions of the first and second of three immunoreceptor tyrosine-based activation motifs which are important in signal transduction, thymocyte development, and TCR biogenesis. These results demonstrate that SIV Nef binds to two distinct domains on TCR ζ in the absence of other T-cell-specific factors, and that interaction with either domain is sufficient to cause down-modulation of TCR ζ.

The nef gene is conserved among the primate lentiviruses, Simian immunodeficiency virus (SIV) and Human immunodeficiency virus types 1 and 2 (HIV-1 and HIV-2, respectively). Although nef is not required for replication in established T-cell lines in vitro, it is a major contributor to the maintenance of high viral loads and induction of immunodeficiency in adult rhesus macaques (23). Nef has no identified enzymatic activity and therefore likely exerts its function via interaction with cellular proteins. Many diverse functions have been attributed to Nef, including down-modulation of CD4 and major histocompatibility complex class I (MHC I) molecules from the cell surface (8, 25, 33) and enhancement of virion infectivity (38). Nef has also been shown to interfere with signal transduction and protein trafficking pathways by binding to cellular proteins such as Lck (7, 13), Hck (30), mitogen-activated protein kinase (13), protein kinase C theta (37), clathrin-associated adapter proteins (11, 25, 29), β-COP (3), a p21-activated kinase-related kinase (18, 31, 32), and a vacuolar ATPase (28).

Another critical protein bound by SIV Nef is the invariant T-cell receptor (TCR) ζ chain, as demonstrated independently by us (2) and others (17). We also demonstrated that the direct interaction between SIV Nef and TCR ζ leads to the down-modulation of the complex formed by the TCR and CD3 (TCR/CD3 complex) and therefore reduced availability of CD3 for stimulation by cross-linking with anti-CD3ɛ antibodies (2).

The TCR complex consists of the clonotypic TCR α and β chains and the invariant CD3 γ, δ, ɛ, and ζ chains which are assembled in the endoplasmic reticulum and transported to the cell surface. As part of the TCR complex, the ζ chain exists in large part as a disulfide-linked homodimer. The ζ chain is synthesized in restricted amounts compared to the other TCR chains and is required for efficient transport of assembled TCR complexes to the cell surface (24, 40). Failure of the ζ chain to associate with the pre-TCR leads to lysosomal degradation of the pre-TCR (40). In addition, TCR ζ plays a crucial role in signal transduction. TCR ζ contains three immunoreceptor tyrosine-based activation motifs (ITAMs) with the consensus sequence YxxI/Lx_6-8YxxI/L, where X represents any amino acid. The sequential phosphorylation of the six tyrosines collectively present in the three ITAMs of TCR ζ following ligation by MHC and antigenic peptide or by CD3 cross-linking is a crucial step in initiating activation of a T lymphocyte (22). TCR ζ is also important in the process of thymic selection, and although no individual ITAM is required, there is a direct relationship between the number of ζ-chain ITAMs within the TCR/CD3 complex and the efficiency of both positive and negative selection (35).

In this report we demonstrate that SIV Nef binds to TCR ζ at two different sites on the cytoplasmic domain, interaction with either being capable of conferring Nef-mediated susceptibility to down-modulation in a mammalian cell assay. Alanine-scanning mutagenesis of these short peptide domains, which overlap ITAMs 1 and 2, demonstrated that the signature tyrosines and the +2 positions could not be replaced by alanine and still allow SIV Nef binding.

MATERIALS AND METHODS

Generation and analysis of yeast strains.

All manipulations of the Y190 strain of Saccharomyces cerevisiae (15) followed standard yeast genetic methods (21). Recombinant yeast strains stably expressing fusion proteins of the Gal4 DNA binding domain (BD) and Nef (Nef/BD) were generated by targeted integration of plasmid vector into the trp1 locus of strain Y190 as described elsewhere (11). The Gal4 BD fusion protein expression vector was pYTH9 (11).

Total yeast cellular extracts used for Western blot analyses were prepared by harvesting appropriate yeast cultures in mid-log phase, pelleting 3 ml of culture, and resuspending in 200 μl of gel loading buffer (50 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate [SDS], 10% glycerol, 0.1% bromophenol blue, 100 mM dithiothreitol). Cellular lysates were boiled for 5 min, chromosomal DNA was sheared with a Hamilton syringe, and debris was pelleted by microcentrifugation. Cellular extracts were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotting using an antibody directed to the hemagglutinin (HA) tag present in Gal4 BD and Gal4 transcriptional activation domain (AD) fusion proteins (12CA5; Boehringer Mannheim) as described elsewhere (2).

Yeast two-hybrid assay.

Interactions between Nef/BD fusion protein and mutant forms of TCR ζ expressed as Gal4 AD fusion proteins from pACT2 were examined in S. cerevisiae Y190 expressing the SIVmacJ5 (for brevity referred to as J5) Nef/BD as described elsewhere (11). Transformants were grown on minimal medium agar plates supplemented with the appropriate amino acids but lacking tryptophan and/or leucine. The selected transformants were stained for β-galactosidase using the substrate 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) in a colony lift, freeze-thaw fracture technique (11).

PCR and subcloning.

All PCR amplifications of the TCR ζ cytoplasmic domain were performed with Pfu DNA polymerase (Stratagene) according to the manufacturer's recommendations. The full-length human TCR ζ cytoplasmic domain (Hum ζ 52-164) was amplified using primers Tz5′ (5′-GCGGAATTCTAAGAGTGAAGTTCAGCAGG-3′; nucleotides [nt] 228 to 245 in the published TCR ζ sequence [42]) and Tz3 (5′-GCGCTCGAGTGTCTCATAATCTGGGCGTCT-3′; nt 604 to 624), using as template the pACT2 cDNA clone 1.11 (2) obtained from an H9 T-cell cDNA library. All TCR ζ PCR products were restriction digested with EcoRI and XhoI and ligated to identically prepared pACT2, pCD8-pcDNA3.1(+), or pET28c vector; DNA was sequenced using established protocols for automated or manual sequencing.

The generation of N-terminal, C-terminal, and combined N- and C-terminal truncation mutants of TCR ζ was accomplished by PCR using Pfu polymerase. The generation of hexa-alanine replacement and alanine substitution mutants of TCR ζ was accomplished by overlap extension PCR in which the mutations were specified within the overlapping sequences. Complete details regarding these PCRs and the respective oligonucleotide primers can be obtained from the corresponding author.

Plasmid expression vectors with full-length and truncated sequences encoding the TCR ζ cytoplasmic domain containing phenylalanine substitutions for tyrosines were generated by PCR amplification of plasmid template pGEX-ZT4, which contains Y→F substitutions in ITAMs 1 and 2 (kindly provided by Martin Sims, Glaxo Wellcome), restriction digestion with EcoRI and XhoI, and ligation to similarly prepared pACT2.

For reverse transcription (RT)-PCR amplification of the rhesus macaque TCR ζ cytoplasmic domain, mRNA was isolated from 2 × 10⁶ Histopaque (Sigma)-purified, rhesus macaque peripheral blood mononuclear cells (PBMCs) using the PolyATract 1000 system (Promega). The rhesus macaque TCR ζ cytoplasmic domain was then amplified by RT-PCR with primers Tz5′ and Tz3 after RT using avian myeloblastosis virus reverse transcriptase (Promega). The resulting subcloned products were DNA sequenced.

To generate a mammalian cell expression vector which encoded fusion proteins containing the extracellular (EX) and transmembrane (TM) domains of CD8, and which contained multiple, unique restriction sites just after the TM sequences, we PCR amplified the CD8α signal peptide (SP), EX, and TM domain-encoding portions of plasmid pTFneo-CD8/ζ (kindly provided by Art Weiss, University of California, San Diego). Primers 5′CD8/TM (5′-GCAAGCTTACCATGGCCTTACCAGTGACC-3′) and 3′CD8/TM (5′-GCCTCGAGGTACCGGATCCGAATTCCGTGGTTGCAGTAAAGGGTGA-3′) were used with Pfu polymerase and pTFneo-CD8/ζ as the template for PCR. The blunt-ended products were ligated directly to pCR-Blunt (Invitrogen), and DNA was sequenced. The HindIII-XhoI insert was moved to the mammalian expression vector pcDNA3.1(+) (Clontech). This resulted in an expression vector encoding the CD8α SP, EX, and TM domains, with unique EcoRI, BamHI, and KpnI sites immediately 3′ to the sequences encoding the TM domain. EcoRI/XhoI inserts encoding portions of the human and rhesus macaque TCR ζ cytoplasmic domain were ligated in frame to identically prepared pcDNA3.1/CD8.

Expression of recombinant proteins in Escherichia coli and use in coprecipitations.

J5 Nef-(glutathione S-transferase) GST was generated by inoculating 20 ml of fresh Luria-Bertani (LB) medium with 2 ml of an overnight culture of E. coli B834 transformed with pGEX-3X-J5-4.2 and grown for 8 h at 37°C. The cells were pelleted and resuspended in 2 ml of lysis buffer (1% Triton X-100, 0.1% N-lauryl sarcosine, 10 μg each of chymostatin, leupeptin, and pepstatin per ml, 800 μM phenylmethylsulfonyl fluoride, 0.2 mM EDTA, and 1 mM iodoacetamine, all in phosphate-buffered saline [PBS]). Cells were lysed by sonication followed by centrifugation at 10,000 × g for 15 min at 4°C. The supernatant was collected and incubated with 50 μl of glutathione-Sepharose beads (Pharmacia) for 1 h at 4°C. The beads were then washed three times (15 min each) with lysis buffer, resuspended in 100 μl of lysis buffer, and stored at 4°C.

To generate His fusion proteins, 20 ml of fresh LB medium was inoculated with one colony of E. coli B834(DE3) transformed with either the parental pET28c(+) vector (Novagen) or the same vector containing TCR ζ sequences and grown for 4 h at 37°C before induction with 1 mM isopropyl-β-d-thiogalactopyranoside. Following an additional 5 h of incubation, the cells were pelleted and stored at −80°C. Prior to coprecipitation, the pellet was thawed on ice and resuspended in 2 ml of lysis buffer. Cells were lysed by sonication followed by centrifugation at 10,000 × g for 15 min at 4°C.

Twenty microliters of GST (1:15 dilution)- or J5 Nef-GST-coated glutathione-Sepharose beads was added to 1 ml of His-tagged TCR ζ protein lysate and incubated for 1 h at 4°C. The lysate-bound GST- and J5 Nef-GST-coated glutathione-Sepharose beads were washed five times (15 min each) with lysis buffer, resuspended in 50 μl of loading buffer, and boiled for 5 min. The coprecipitated proteins were separated by SDS-PAGE and electroblotted onto an Immobilon-P transfer membrane (Millipore). Immunodetection was performed with a 1:10,000 dilution of anti-T7 tag monoclonal antibody (Novagen) followed by a 1:10,000 dilution of goat anti-mouse antibody (Boehringer Mannheim) and developed by the Amersham ECL system.

RT-PCR amplification and subcloning of the CD2 cDNA.

Poly(A)⁺ RNA was prepared from the JJK Jurkat T-cell line using the PolyATract kit (Promega). Full-length CD2 cDNAs were generated by RT-PCR (Promega) using the primers CD2R.3PCS (5′-ACACGAATTCTTAATTAGAGGAAGGGG-3′) and CD2F.FL (5′-ACACGGATCCTGATGAGCTTTCCTGTAAATTTG-3′). The resulting products were cloned into the TA cloning vector pGemT (Promega), and the entire CD2 coding sequence was sequenced and compared to the published reference sequence (34) (accession no. M16445). Insert-containing clones were restriction digested with BamHI and EcoRI and ligated to identically prepared pcDNA-3.1(−).

Transfection of 293T cells and flow cytometry.

For analysis of Nef-mediated down-modulation of CD8/TCR ζ fusion proteins in mammalian cells, confluent cultures of 293T cells (American Type Culture Collection) were split 1:3 on the day prior to electroporation. Cells were prepared for electroporation by washing cells adhered to culture flasks with PBS, incubation with trypsin-EDTA for 3 to 5 min, addition of 5 ml of Dulbecco modified Eagle medium containing 5% fetal bovine serum (FBS), and dilution with PBS to a total volume of 20 ml. The cells were washed twice with PBS and resuspended in RPMI 1640–20% FBS at a cell density of 10⁷ per ml; 500 μl of the cell suspension was electroporated in 0.4-cm cuvettes containing the appropriate expression plasmids. Cells were electroporated with a Gene Pulser II (300 V, 975 μF; BioRad), and transferred to a 25-cm² flask, and cultured at 37°C in 5% CO₂ in a humidified incubator. At 48 h posttransfection, cells were washed with PBS and gently lifted using trypsin-EDTA at room temperature for 3 to 5 min. Trypsin action was then inhibited by addition of 5 ml of DMEM Dulbecco-modified Eagle medium 5% FBS. The cells were washed in 10 ml of PBS, and repelleted cells were resuspended at a density of 10⁶ cells per ml in PBS containing 2% bovine serum albumin (Sigma) plus 2 mM sodium azide; 10⁵ cells were stained with saturating quantities of fluorescein isothiocyanate-conjugated mouse anti-human CD2 (Leu-5b, clone S5.2; Becton Dickinson) and phycoerythrin-conjugated mouse anti-human CD8 (Leu-2a, clone SK1; Becton Dickinson) monoclonal antibodies. Cells were incubated with antibodies for 1 h on ice, washed in PBS, and resuspended in 0.5 ml of PBS. Based on forward/side scatter profiles, 10⁵ viable cells were analyzed on a Coulter Elite flow cytometer. Isotype-matched, fluorochrome-conjugated antibodies were used as controls to establish gates for fluorescein isothiocyanate- and phycoerythrin-positive cells. Gates were set based on the fluorescence profiles obtained with the isotype control antibodies such that 98% of these cells were excluded from the gate.

Nucleotide sequence accession numbers.

The sequences reported here have been deposited in the GenBank nucleotide sequence database (accession no. AF204109–AF204113).

RESULTS

SIV Nef interacts with two nonoverlapping, homologous domains on TCR ζ.

The invariant TCR ζ chain has been demonstrated previously by us (2) and by Howe et al. (17) to interact with SIV Nef. In addition, we have demonstrated that this interaction leads to the down-modulation of the TCR/CD3 complex from T-cell surfaces (2). To map the determinants on TCR ζ for interaction with SIVmac Nef, we used a yeast two-hybrid system in which Nef was stably expressed in S. cerevisiae Y190 as a Gal4 BD fusion protein (2). Expression plasmids encoding TCR ζ cytoplasmic domain mutants fused to the Gal4 AD were generated (Fig. 1A) and introduced into a strain of Y190 expressing the SIVmacJ5 Nef/BD. When successful interactions occur between J5 Nef/BD and a Gal4 AD fusion protein, transcription of the lacZ gene is induced and the transformants can be stained for β-galactosidase (Fig. 1B).

FIG. 1.

Identification of two independent domains on TCR ζ for interaction with J5 Nef via a yeast two-hybrid assay. (A) Portions of the TCR ζ cytoplasmic domain are indicated by the gray boxes and were expressed as Gal4 AD fusion proteins in S. cerevisiae Y190 stably expressing the J5 Nef/BD fusion protein. Replacement of specific regions of TCR ζ with a stretch of six alanines is indicated by Δ. Interaction with J5 Nef/BD was determined in a filter-lift, freeze-fracture assay for β-galactosidase. +, interaction with J5 Nef/BD; −, no interaction with J5 Nef/BD. Each TCR ζ/AD fusion protein was examined in the yeast two-hybrid system three to five times. Amino acid numbering follows that of Weissman et al. (42). (B) Representative set of filter β-galactosidase assays in which the indicated Gal4 AD fusion constructs were expressed in the J5 Nef/BD-expressing strain of Y190. (C) Western blotting was performed on total S. cerevisiae lysates by immunoblotting filter-immobilized proteins with an anti-HA antibody, which detects the HA tag present in all AD and BD fusion proteins used in these studies.

Within the cytoplasmic domain of TCR ζ are the three ITAMs, which are involved in antigen-specific signal transduction (18), TCR/CD3 biogenesis (27, 35), and thymocyte selection (36). To determine if the individual ITAMs are involved in the interaction with J5 Nef, we generated mutant forms of TCR ζ in which each of the three ITAMs was independently replaced with six alanines. Removal of each ITAM did not abrogate interaction with J5 Nef/BD, as demonstrated with Gal4 AD fusions 52-164Δ1/AD, 52-164Δ2/AD, and 52-164Δ3/AD (Fig. 1A), indicating that the domain responsible for interaction with SIV Nef was not solely composed of any single ITAM. To further delineate the region on TCR ζ with which SIV Nef interacted, N-terminal and C-terminal truncation mutants of TCR ζ were generated using ITAM boundaries as sites of truncation. Progressive truncation of the cytoplasmic domain of TCR ζ from the C terminus up to amino acid (aa) 86 did not abrogate interaction with J5 Nef/BD (Fig. 1A, clone 52-86/AD). However, upon the removal of ITAM 1 (aa 72 to 86), interaction with J5 Nef/BD was lost (Fig. 1A, mutant 52-71/AD). Interestingly, progressive removal of N-terminal portions of the TCR ζ cytoplasmic domain demonstrated that aa 52 to 111 were dispensable for interaction with J5 Nef/BD, but upon removal of ITAM 2 (aa 111 to 126), J5 Nef/BD was no longer bound (Fig. 1A, mutants 111-164/AD and 127-164/AD). These data indicated, therefore, that SIV Nef interacts with two separate domains on TCR ζ, one domain between aa 52 and 86 and the second domain between aa 111 and 164, and indicate that ITAMs 1 and 2 are components of the two respective domains.

To further define the minimal interaction domains on TCR ζ, additional mutants combining N-terminal and C-terminal truncations were generated. An AD fusion construct consisting of ITAM 1 alone interacted with J5 Nef/BD (Fig. 1A, mutant 72-86), demonstrating that the first interaction domain lies entirely within ITAM 1. Because ITAMs consist of two tyrosine-containing YxxL/I motifs, we generated mutants lacking either the first YxxL (YNEL, aa 72 to 75) or the second YxxL (YDVL, aa 83 to 86) of ITAM 1 (Fig. 1A, mutants 76-110 and 52-82, respectively) to determine if both YxxL/I motifs are necessary for interaction with J5 Nef/BD. Removal of aa 72 to 75 (YNEL) abrogated interaction with J5 Nef/BD (AD clone 76-110) while removal of aa 83 to 86 (YDVL) did not (AD clone 52-82), thus identifying aa 72 to 82 (YNELNLGRREE) within ITAM 1 of TCR ζ as the most N-terminal interaction domain. Expression of an AD fusion protein containing only ITAM 2 did not interact with J5 Nef/BD (AD clone 111-126), but interaction was restored with the addition of aa 127 to 141 (Fig. 1A, AD clone 111-141). Further truncation of the region between aa 111 to 142 demonstrated that the second YxxL/I motif in ITAM 2 along with the immediately C-terminal 8 aa are sufficient to allow interaction with J5 Nef/BD (AD clone 123-134). The second interaction domain, therefore, consisted of aa 123 to 134 (YSEIGMKGERRR). These data demonstrate that SIV Nef interacts with two short, nonoverlapping linear peptides on TCR ζ which are contained within ITAM 1 and overlap ITAM 2, respectively.

In control assays performed in parallel, TCR ζ/AD fusion proteins were expressed in strains of Y190 expressing no Gal4 BD or stably expressing Gal4 BD alone. CD2/AD was included in all experiments as a negative control for interaction with J5 Nef/BD. In all instances, no β-galactosidase was detected in a filter-lift, freeze-fracture assay, indicating that up-regulation of β-galactosidase expression was Nef dependent. Western blot analyses of total cellular lysates from Y190 transformants indicated that there were differing levels of protein expression directed by the pACT2 plasmids (Fig. 1C and data not shown), although the levels of expression did not correlate with interaction with J5 Nef/BD (Fig. 1A).

Alanine-scanning mutagenesis of TCR ζ identifies amino acids critical for interaction with Nef.

Having defined the most N-terminal interaction domain as a motif of 11 or fewer aa, we used alanine-scanning mutagenesis to determine which amino acids are essential for interaction with J5 Nef/BD. All 11 aa were replaced individually by alanine, using AD clone 52-110 as the backbone, and each mutant TCR ζ/AD clone was examined for its ability to interact with J5 Nef/BD in S. cerevisiae. These analyses demonstrated that replacement of amino acid Y72, E74, L75, or L77 with alanine abrogated interaction with J5 Nef/BD (Fig. 2). To determine whether the Gal4 AD/TCR ζ fusion protein could be truncated upstream of E82, further truncations were generated after L77, G78, and R79, all of which retained the ability to interact with J5 Nef/BD (Fig. 2). Therefore, the interaction of J5 Nef with the membrane-proximal interaction domain is conferred by a 6-aa peptide, YNELNL (aa 72 to 77) within ITAM 1, which we have named SNID-1 (SIV Nef interaction domain 1).

FIG. 2.

Alanine substitution mutagenesis of TCR ζ identifies amino acids critical for interaction with SIV Nef. The N- and C-terminal domains with which SIV Nef interacts, aa 72 to 82 and 123 to 134, respectively, were mutated to replace individual amino acids with alanine. These mutations were placed in the context of clones 52-110/AD and 111-141/AD, respectively, which in their nonmutated forms expressed Gal4 AD fusion proteins that interacted with SIV Nef BD (Fig. 1). Interaction with J5 Nef/BD was examined using the yeast two-hybrid assay by a freeze-fracture, colony lift assay for β-galactosidase. Successful interaction is indicated by +. All mutant TCR ζ/AD fusion proteins were assayed in the yeast two-hybrid system a minimum of two times. Shown schematically in the lower portion are the TCR ζ/AD fusion proteins in which the indicated tyrosines in ITAMs 1 and 2 were mutated to phenylalanine.

Having demonstrated that Y72 in SNID-1 could not be replaced by alanine without disrupting interaction with SIV Nef, we generated constructs in which tyrosines were substituted more conservatively with phenylalanine. When examined in the context of only SNID-1 (clone 52-82/Y72F/AD) or SNID-1 and SNID-2 (clone 52-164/Y72-123[4]/AD), replacement of the tyrosines in ITAMs 1 and 2 did not abrogate interaction with J5 Nef/BD (Fig. 2).

Mutagenesis of the corresponding positions in SNID-2 (123YSEIGM) also demonstrated a requirement for the tyrosine and isoleucine in the YxxL/I motif in order to interact with SIV Nef. As for SNID-1, interaction with SIV Nef/BD occurred when the +1 position (S124) was replaced by alanine, but it did not occur when the +2 position (E125) was replaced by alanine (Fig. 2). In contrast to SNID-1, substitution for alanine in the +4 position (G127) abrogated interaction with SIV Nef/BD. For all of AD fusion constructs, Western blot analyses of the steady-state levels of the Gal4 TCR ζ/AD fusion proteins in the yeast strain Y190 demonstrated roughly equivalent amounts of protein (data not shown).

TCR ζ SNID-1 and SNID-2 confer susceptibility to Nef-mediated down-modulation in mammalian cells.

A common property attributed to SIV Nef is the modulation of cell surface proteins including CD4 (4, 14), MHC I (8, 25, 33), and the TCR/CD3 complex (2). To determine whether subdomains of the TCR ζ cytoplasmic domain containing either SNID-1 or SNID-2 conferred susceptibility to Nef-mediated down-modulation in mammalian cells, we constructed expression vectors encoding chimeric proteins containing the CD8 EX and TM domains fused to cytoplasmic portions of TCR ζ. These expression vectors were cotransfected into the 293T cell line along with a CD2 expression vector to allow quantitative analysis of cell surface CD8 on the successfully transfected population which are CD2 positive, either in the presence or in the absence of SIV Nef expression. As shown in Fig. 3, in the absence of J5 Nef expression, the majority of transiently transfected 293T cells were surface antigen positive for both CD8 and CD2, with values typically >90%. When J5 Nef was coexpressed, however, CD8/TCR ζ fusion proteins containing either SNID-1 (construct CD8/Hum ζ 52-110), SNID-2 (constructs CD8/Hum ζ 111-141 and CD8/Hum ζ 87-141), or both (construct CD8/Hum ζ 52-164) were down-modulated from the cell surface (Fig. 3). The CD8/TCR ζ fusion protein containing the N-terminal 59 aa in which ITAM 1 was replaced with six alanines (construct CD8/Hum ζ 52-110Δ1) was not appreciably down-modulated. Similarly, when the cytoplasmic portion of the CD8 fusion protein contained only the C-terminal 38 aa (construct 127-164) which does not contain either SNID-1 or SNID-2, the CD8 fusion protein was not down-modulated from the cell surface. Unexpectedly, the CD8 fusion protein containing aa 87-141 was not down-modulated to the same extent as the CD8 fusion protein containing only aa 111 to 141, although they both contain SNID-2. In this context, therefore, SNID-2 is likely presented in a structurally distinct form when the region between ITAMs 1 and 2 is present.

FIG. 3.

Analysis of the ability of SNID-1 and SNID-2 to confer susceptibility to Nef-mediated down-modulation in mammalian cells. The 293T cell line was transiently cotransfected with plasmids encoding the indicated CD8/TCR ζ fusion proteins and a plasmid encoding CD2, either with or without a plasmid encoding J5 Nef. The cells were stained 2 days posttransfection for cell surface CD2 and CD8. The portion of the TCR ζ cytoplasmic domain contained in each CD8/TCR ζ fusion protein is indicated schematically in bold at the right. The percentages in the right quadrants of the scatter plots represent the percentage of CD2⁺ cells that are either CD8⁺ or CD8⁻. These data are from one representative experiment repeated a minimum of two additional times with similar results.

These data demonstrated, therefore, that SNID-1 and SNID-2 not only are sufficient for interaction with J5 Nef, as demonstrated in the yeast two-hybrid system, but more importantly confer susceptibility to Nef-mediated surface modulation in mammalian cells.

SIVmac Nef interacts with the cytoplasmic domain of the rhesus macaque TCR ζ.

Thus far, studies of the interactions between SIV and HIV-1 Nef and cellular proteins have focused on human binding partners. Differences have been demonstrated, however, in the repertoire of cellular partners which interact with SIV or HIV-1 Nef (17, 26, 29). We therefore sought to determine whether SIV Nef interacted with the rhesus macaque TCR ζ. Partial cDNAs encoding the cytoplasmic domain of the rhesus macaque TCR ζ were obtained by RT-PCR amplification using polyadenylated mRNA extracted from purified rhesus macaque PBMCs. Among the 10 subclones sequenced, we identified five different isoforms, all of which had signature rhesus macaque-specific amino acid substitutions (E115A and A160T) and a 2-aa insertion (NQ between aa 131 and 132) (Fig. 4A). Interestingly, the E115A and A160T substitutions are also present in the mouse (41) and sheep (16) TCR ζ, while the NQ insertion between aa 131 and 132 is present only in the sheep TCR ζ (16). Also, 9 of the 10 rhesus macaque TCR ζ clones contained a single amino acid deletion of Q100, as observed in two of the three published human TCR ζ cytoplasmic domain sequences (2, 17, 42) (Fig. 4A). It is interesting that both this amino acid insertion and the NQ insertion after E131 are located at exact exon/exon junctions (19) and the observed polymorphisms are likely due to alternative splice site utilization, although we cannot entirely rule out the possibility that RT or PCR errors have contributed to the observed differences.

FIG. 4.

Cloning the rhesus macaque TCR ζ cytoplasmic domain and demonstration of interaction with SIV Nef. (A) The sequences encoding the cytoplasmic domain of the rhesus macaque TCR ζ were obtained by RT-PCR amplification of mRNA obtained from Ficoll-purified PBMCs. The putative amino acid sequences of 10 different subcloned RT-PCR products are shown in alignment with published human, mouse, and sheep sequences. Dots represent amino acid identity; dashes represent gaps. ITAMs are underlined, and SNID-1 and SNID-2 are shown in bold. ∗, premature stop codon. (B) Portions of the rhesus macaque TCR ζ cytoplasmic domain are indicated by the gray boxes and were expressed as Gal4 AD fusion proteins in S. cerevisiae Y190 stably expressing the J5 Nef/BD fusion protein. Interaction with J5 Nef/BD was determined in a filter-lift, freeze-fracture assay for β-galactosidase. +, interaction with J5 Nef/BD; −, no interaction with J5 Nef/BD. Amino acid numbering follows that of Weissmann et al. (42). The macaque TCR ζ-Gal4/AD fusion proteins were examined in the yeast two-hybrid system three to five times.

The RhMac-2, -3, -4, and -5 isoforms of the rhesus macaque TCR ζ cytoplasmic domain were subcloned into the pACT2 Gal4 AD expression vector and introduced into the J5 Nef/BD-expressing strain of Y190. The RhMac-3 52-165/AD clone (Fig. 4B) and all other isoforms (data not shown) interacted with J5 Nef/BD in the yeast two-hybrid system, demonstrating for the first time, to the best of our knowledge, an interaction between a rhesus macaque cellular protein and SIVmac Nef. As seen in Fig. 4A, the amino acid sequence of SNID-1 in the deduced rhesus macaque TCR ζ sequence is identical to those in human TCR ζ, whereas there is an NQ within or near rhesus macaque SNID-2. To determine if this insertion abrogates interaction with J5 Nef/BD, we generated two AD fusion protein constructs which contained truncated forms of the RhMac-3 isoform, RhMac 52-86/AD, and RhMac 122-142/AD. As expected, RhMac 52-86/AD interacted with J5 Nef/BD, as did RhMac 122-142/AD (Fig. 4B). Similarly, as observed with the human TCR ζ cytoplasmic domain, a CD8 fusion protein containing the rhesus macaque TCR ζ cytoplasmic domain, a CD8 fusion protein containing the rhesus macaque TCR ζ cytoplasmic domain was susceptible to SIVmacJ5 Nef-mediated down-modulation in 293T cells (Fig. 3). These data demonstrate that SIVmacJ5 Nef interacts with the rhesus macaque TCR ζ cytoplasmic domain and can modulate its surface expression levels in mammalian cells and that as with human TCR ζ, there are two interaction domains.

SIVmac Nef interacts with TCR ζ in the absence of any additional eukaryotic cellular proteins.

The interaction between SIVmac Nef and TCR ζ might be facilitated or augmented by other eukaryotic proteins. To determine whether SIV Nef and TCR ζ interact in the absence of additional eukaryotic proteins, we expressed SIV Nef as a GST fusion protein, and the cytoplasmic domain of TCR ζ, or portions thereof, as hexahistidine-tagged fusion proteins, in E. coli. As shown in Fig. 5, Nef-GST immobilized on Glutathione-sepharose beads coprecipitated the full-length human TCR ζ cytoplasmic domain from bacterial lysates (construct Human 52-164/His), whereas immobilized GST did not. Similar to the yeast two-hybrid and 293T cotransfection results, His-tagged subdomains containing either SNID-1 (construct 52-110/His) or SNID-2 (construct 87-141/His) were coprecipitated by Nef-GST but not GST alone (Fig. 5). Because the results from the yeast two-hybrid and 293T cell assays demonstrated that the context in which SNID-2 was presented was an important determinant of the extent of interaction with SIV Nef, we examined multiple SNID-2-containing constructs. Construct 111-141/His, which interacted with SIV Nef in S. cerevisiae (Fig. 1) and 293T cells (Fig. 3), was not coprecipitated by Nef-GST (Fig. 5). To further determine whether presentation of SNID-2 in an alternate context would allow coprecipitation by Nef-GST, the full-length cytoplasmic domain in which ITAM 1, and therefore SNID-1, was replaced with six alanines (construct 52-164Δ1/His), was expressed as a His-tagged fusion protein. As shown in Fig. 5B, in this context, Nef-GST coprecipitated this SNID-2-containing fusion protein. This was also observed when ITAM 2 was replaced with six alanines, leaving SNID-1 intact (construct 52-164Δ2/His).

FIG. 5.

Interaction between Nef and TCR ζ recombinant proteins expressed in E. coli. (A) Hexahistidine-tagged portions of the TCR ζ cytoplasmic domain expressed in E. coli. (B) GST and J5 Nef-GST proteins were expressed in E. coli and purified using glutathione-Sepharose beads. The immobilized GST proteins were incubated with extracts of His-T7-tagged TCR ζ cytoplasmic domains, also expressed in E. coli. After thorough washing, proteins associated with the GST (lanes designated by G)- and GST-Nef (lanes designated by N)-coated beads were boiled, and the resulting supernatant was analyzed by immunoblotting with an anti-T7 monoclonal antibody which detects an epitope present in all His-tagged fusion proteins used in these studies. Shown are representative data from one experiment repeated twice with identical results.

Similar again to results for the yeast two-hybrid and 293T cell cotransfection assays, the rhesus macaque cytoplasmic domain was coprecipitated by SIV Nef-GST (Fig. 5B). These results therefore demonstrate that both the rhesus macaque and human TCR ζ cytoplasmic domains can interact with SIV Nef in the absence of other eukaryotic proteins and that interaction with Nef can be conferred by either the SNID-1 or SNID-2 interaction domain on TCR ζ.

DISCUSSION

The exact mechanisms by which Nef contributes to immunodeficiency have not been completely elucidated, whether considering systemic or subcellular aspects of the virus-host interactions, such as increased viral replication or decreased levels of cell surface CD4, respectively. Obtaining a more complete understanding of Nef function could prove to be extremely useful in understanding HIV-1/SIV pathogenesis and in developing strategies for educating the immune system to successfully hold viral replication in check. Toward this end, we have examined in this report the specific interactions between SIV Nef and the cytoplasmic domain of the TCR ζ chain, an interaction which we previously demonstrated (2) to lead to the down-modulation of TCR/CD3 complexes from T-cell surfaces.

Interestingly, we found that there are two domains on TCR ζ which can interact with SIV Nef and confer susceptibility to Nef-mediated down-modulation in mammalian cells. These domains, named here SNID-1 and SNID-2, are conserved between human and rhesus macaque TCR ζ; therefore, not unexpectedly, SIV Nef interacted with the rhesus macaque TCR ζ cytoplasmic domain and down-modulated a CD8/rhesus macaque TCR ζ fusion protein. It is not clear whether there has been selective pressure on SIV Nef to interact with multiple domains on TCR ζ or whether evolutionary selective pressures on TCR ζ structure and function have fortuitously resulted in a molecule that happens to contain two structurally related domains that are bound by Nef.

It is interesting that studies of ordered pattern of phosphorylation of the tyrosines present in TCR ζ ITAMs demonstrated that the first tyrosine of ITAM 2 and the second tyrosine of ITAM 3 are constitutively phosphorylated in resting T lymphocytes (22), neither of which are contained in SNID-1 or SNID-2. Therefore, Nef would be able to bind to TCR ζ in resting T lymphocytes. In addition, Kersh et al. (22) demonstrated that mutation of the first or second tyrosine residue in ITAM 1 to phenylalanine not only abrogated phosphorylation at that specific position but also prevented ITAM phosphorylation events further downstream in the sequential pattern. If Nef interaction with TCR ζ has effects on signal transduction in addition to modulated surface TCR/CD3 levels, then binding to ITAM 1 is likely to prevent early phosphorylation events required for appropriate TCR signal transduction (12, 20, 39).

Alanine-scanning mutagenesis of SNID-1 and SNID-2 indicated that the Y and I/L at position +3 are critical determinants for interaction with SIV Nef, since substitution of either of these amino acids with alanine abrogates interaction (Fig. 2). However, substitution of the tyrosines in SNID-1 or SNID-2 with phenylalanine did not prevent interaction, suggesting that the bulky aromatic group is important for SIV Nef binding, but the oxygen present on tyrosine is not required. It is not unreasonable to expect that phosphorylation of the tyrosines in SNID-1 or -2 would abrogate interaction with SIV Nef because substitution of tyrosine with phenylalanine still allowed binding to SIV Nef (Fig. 2), suggesting that the amino acid in this position is involved in critical hydrophobic or planar interactions, though this will require further investigation. Because both SNID-1 and SNID-2 are comprised in large part of YxxL/I motifs present in ITAMs, it will be interesting to determine whether SIV Nef interacts with and modulates other proteins, whether in T lymphocytes or monocytes/macrophages, which contain tyrosine-based activation and endocytic motifs.

Both SNIDs contain canonical YxxL/I motifs and could potentially serve as endocytic signals. There is evidence demonstrating the presence and differing availability of YxxL endocytic signals in the cytoplasmic portions of the TCR/CD3 complex (5, 10). TCR ζ has been demonstrated to cycle to and from the cell surface more rapidly than the other components of the TCR/CD3 complex (6) and therefore must have signals that are efficiently recognized by the endocytic machinery. It has recently been demonstrated for CD3ɛ that a YxxL endocytic signal is present in the cytoplasmic domain (5). In addition, the CD3γ YxxL endocytic signal is masked until a signal is delivered via antigen/MHC (10). Nef interacts with components of the protein trafficking machinery, including the μ chains of the AP-1 and AP-2 complexes (25, 29), and it also contains a dileucine protein trafficking motif (1, 9). Nef might therefore be decreasing the cell surface levels of TCR/CD3 complexes by binding to one or both of the YxxL/I motifs present in the SNIDs on TCR ζ and then using its dileucine sorting properties to shunt the TCR/CD3 complex from the cell surface. The mechanism of down-modulation of the TCR/CD3 complex requires further investigation to determine whether SIV Nef increases the endocytosis of this complex or whether SIV Nef interacts with TCR ζ earlier in its biogenesis and thereby prevents the proper assembly and transit of complete TCR/CD3 complexes.

Interaction with the TCR ζ chain is a conserved feature of SIVmac Nef proteins, having been demonstrated for both SIVmacJ5 Nef (this report and reference 2) and SIVmac239 (2, 17), as well as HIV-2 (17). Regarding HIV-1 Nef, our preliminary studies indicate that it does not bind to TCR ζ (unpublished observations), which is in agreement with published findings by Howe et al. (17). However, a recent report suggests that HIV-1 Nef expressed in stably transfected and selected cell clones as a CD8/Nef fusion protein does bind to TCR ζ (43). These differences will require further examination and are likely due in part to the specific systems used to study interaction, as well as possible phenotypic variation of the nef alleles examined.

TCR ζ is crucial for TCR assembly and biogenesis, signal transduction, and T-cell selection and development (18, 27, 35, 36). Interference with any of these properties is likely to impair the recognition and effector function of the T lymphocyte in which this occurs. The down-modulation of TCR/CD3 complexes from T-lymphocyte surfaces in the host would have profound consequences on the collective immune capability of the host. Down-modulation of the TCR/CD3 complex during the period between early transcription of integrated provirus and death of the infected cell, regardless of the mechanism by which the cell dies, will clearly reduce the ability of the host to mount immune responses. Infection of CD4⁺ T lymphocytes would render them unresponsive to antigen/MHC stimulation as a result of Nef expression, and ongoing inhibition of this sort in spleen and lymph nodes throughout the entire course of infection could eventually result in a diminution or erosion of the ability to mount immune responses to pathogens. We have demonstrated that SIV Nef interacts with rhesus macaque TCR ζ and down-modulates a CD8/RhMac TCR ζ fusion protein from the cell surface, and we fully expect that Nef exhibits this activity in productively infected cells in the host, though this remains to be demonstrated. Down-modulation of a surface molecule so profoundly important in mounting immune responses might be a contributing factor in the more rapid disease progression observed in SIVmac-infected rhesus macaques than in HIV-1-infected humans.