key: cord-0009890-74zvvig0
authors: Knight, Andrew M.; Harrison, Georgina B.; Pease, Richard J.; Robinson, Peter J.; Julian Dyson, P.
title: Biochemical analysis of the mouse mammary tumor virus long terminal repeat product. Evidence for the molecular structure of an endogenous superantigen
date: 2005-11-17
journal: Eur J Immunol
DOI: 10.1002/eji.1830220339
sha: 821a702c78c8084247ef6206b3c2fb2c45a24fa7
doc_id: 9890
cord_uid: 74zvvig0

Recent reports have shown that both exogenous and endogenous mouse mammary tumor viruses (MMTV) can encode superantigens. Transfection and transgenic studies have identified the open reading frame (ORF) present in the 3′ long terminal repeat (LTR) as encoding superantigen function. In this study, we have used an in vitro translation system in an attempt to characterize the molecular nature of the protein encoded by the 3′ ORF of Mtv‐8. Using various constructs encoding full‐length and truncated versions of the ORF product, we report that the hydrophobic region close to the amino terminus of the 36‐kDa protein can function as a transmembrane domain. Protease digestion experiments also demonstrate that the protein has a type‐II transmembrane conformation with an extra‐cytoplasmic carboxy terminus. Since this hydrophobic region is conserved between all known MMTV, we speculate that LTR ORF, including those proposed to encode the minor lymphocyte stimulatory antigens, are also capable of encoding type‐II transmembrane glycoproteins. The polymorphism between MMTV LTR ORF products, which correlates with deletion phenotypes, is predominantly in the carboxy‐terminal extracellular region, consistent with a major role in interaction with the T cell receptor.

Recent reports have shown that both exogenous and endogenous mouse mammary tumor viruses (MMTV) can encode superantigens. Transfection and transgenic studies have identified the open reading frame (ORF) present in the 3' long terminal repeat (LTR) as encoding superantigen function. In this study, we have used an in vitro translation system in an attempt to characterize the molecular nature of the protein encoded by the 3' ORF of Mtv-8. Using various constructs encoding full-length and truncated versions of the ORF product, we report that the hydrophobic region close to the amino terminus of the 36-kDa protein can function as a transmembrane domain. Protease digestion experiments also demonstrate that the protein has a type-I1 transmembrane conformation with an extra-cytoplasmic carboxy terminus. Since this hydrophobic region is conserved between all known MMTV, we speculate that LTR ORF, including those proposed to encode the minor lymphocyte stimulatory antigens, are also capable of encoding type-I1 transmembrane glycoproteins. The polymorphism between MMTV LTR ORF products, which correlates with deletion phenotypes, is predominantly in the carboxy-terminal extracellular region, consistent with a major role in interaction with the T cell receptor.

gested interaction of toxins with a region distinct from the peptide binding pocket [8] . Additionally, the cellular response of T cells can differ between "superantigen" and T cells involves a processed form of the antigen being presented in the antigen-binding pocket of a major histocompatibility molecule [l]. In conventional antigen recognition, all variable germ-line segments (Va, J,, Vp, Dg and Jp), as well as N-region additions, contribute to the specificity of the TcR [2] . In contrast to this, Tcells stimulated by certain bacterial toxins [2] and the recently identified endogenous "superantigens" encoded by germ-line copies of mouse mammary tumor virus (MMTV; [3] [4] [5] [6] ) have a particular TcR Vp usage. The major effect of endogenous superantigens is seen as thymic deletion of T cells expressing particular Vp regions. In contrast, in the periphery, superantigen interaction leads to polyclonal activation of T cells expressing such TcR Vg regions with limited contribution from other TcR components [2] .

Evidence suggests that superantigens are quite distinct from peptides in their interaction with both MHC and TcR molecules. First, exogenous superantigens do not require cellular processing for presentation to T cells [7] and, second, mutation studies of class I1 molecules have sug-conventional antigen stimulation [9] .

While bacterial toxins have been characterized at the molecular level, no such information is available for endogenous superantigens. Following recent reports identifying the MMTV 3' long terminal repeat (LTR) as the region required for superantigen function [lo, 111, we have investigated the biochemical properties of the protein encoded by the 3' LTR open reading frame (ORF) of Mtv-8 (Mtv-8 ORF) which is responsible for partial thymic deletion of Vpll+ T cells [6] . Previous in vitro translation studies have shown that the 3' LTR ORF encodes four overlapping polypeptides of 36,24,21 and 18 kDa by the use of alternative initiation sites [12, 131. In vivo identification of MMTV ORF has been limited to the product of the partially deleted LTR present in the lymphoma cell line EL4 [14] . Using a number of truncated constructs from Mtv-8 ORF, we have translated in vitro the resulting transcripts and examined their protein products.

r I ioooii Mtv-8 ORF inserts for translation studies were constructed as follows (see Fig. 1 ): using the polymerase chain reaction (PCR), Mtv-8 3' Eco RI fragment [15] DNA (a kind gift from Dr. Gordon Peters, ICRF, London, GB) was ampli- Construct 3 (see Fig. 1D and I; Mtv-8 ORF truncated at the carboxy terminus, aa 1-316) 5' primer: same as primers used for construct 1 and 2, and 3' primer: ATA AGT CGA CTTAAA AAG TAT CGT CAG AAAT. Construct 4 (see Fig. 1E and J; Mtv-8 ORF truncated at the amino terminus, so as to cxclude region from aa nos. 1-122) 5' primer: AAT ACG ACT CAC TAT AGG GCG AAT TGC GCC ATG GTTand 3' primer as used for construct 1. All 5' primers are mismatched to introduce an Nco I restriction site at the first ATG start codon. All 3' primers include an in-frame stop codon and a Sal I restriction site. PCR conditions were as described [16] with 25 cycles each consisting of 1 min 94 "C, 1 min 55 "C, 30 s 72 "C followed by an additional 9 min at 72 "C, using 2 mM Mg2+. PCR products were gel purified (1 % low-melting point agarose, BRL, Uxbridge, GB), digested with Nco I and Sal 1 according to manufacturer's conditions (C.P. Biolabs, Bishops Stortford, GB) and ligated into the Nco I, Sal I linearized transcription vector (a kind gift from Drs. Mike Howell and Tim Hunt, University of Cambridge). Efficient cap-independent translation is obtained using this vector as it contains nucleotides 259-837 of the 5' untranslated region (UTR) from mouse encephalomyocarditis virus (EMCV) upstream of its cloning site [17] . Following linearization with Sal I, plasmids containing constructs 1,2,3 and 4 were transcribed using T7 RNA polymerase (Boehringer-Mannheim, Mannheim, FRG). Transcripts encode Met-Ala followed by residues encoded by the OW. For carbonate treatment, post-translational rat liver microsomes were added to mixtures and ultracentrifuged in a Beckman airfuge for 15 min at 30 psi. Membrane fractions were resuspended in sodium carbonate (pH 11.5) for 30 min at 0 "C. After another centrifugation for 15 min at 30 psi. the protein content of membrane fractions and supernatant were analyzed. All proteins were analyzed by SDS-PAGE on 8% to 15% gradient gels.

Previous reports have identified a hydrophobic region close to the amino terminus of an MMTV ORF protein sequence [13] and suggested this may encode a membrane transfer sequence (signal sequence) [ 101. Hydrophobicity plot analysis of the predicted amino acid sequence of Mtv-8 ORF did not reveal any other significant non-polar regions (see codons. Open box = hydrophobic region, Open kDa, respectively (lanes 1, 3,5 and 7). On addition of canine microsomes to the translation mixtures, the molecular weight of the protein products from constructs 1 , 2 and 3 increased by approximately 10 kDa (Fig. 2 lanes 2, 4  and 6 ). This increase in size would be accounted for by the post-translational addition of carbohydrate side chains to the protein backbone, in agreement with the usage of the consensus N-linked glycosylation sites at Asn 80, 90, 94, 132 and 147. It is clear, however, that the protein product from construct 4 does not change in molecular weight in the presence of microsomal membranes (Fig. 2,  lanes 7, 8; even though the truncated sequence retains two potential glycosylation sites at Asn 132 and 147, see Fig. 1J ). These results indicate that the domain directing membrane translocation would be expected to be located within residues 1-122.

To determine whether the products of the Mtv-8 ORF constructs are membrane anchored, we used alkaline sodium carbonate on microsomal translations to rupture the membranes [20] . Using this treatment, we have confirmed absolute discrimination between in vitro translated soluble proteins (yeast ct mating factor, E.coli lactamase) and a transmembrane protein (coronavirus El glycoprotein; data not shown). Fig. 3 shows the distribution of protein products after carbonate treatment of the translation mixtures. For the in vitro translation products of constructs 1, 2 and 3, exclusive association with the membrane fraction is seen (lanes 1-6) consistent with their being membrane-associated glycoproteins. In contrast, the product of construct 4 is clearly not tightly associated with the membrane fraction, as it is seen in both membrane and supernatant fractions (lanes 7 and 8). The weak membrane association observed is not understood but may reflect limited solubility.

These results are consistant with the utilization of the hydrophobic region from residues 38 to 63 (See Fig. 1F) [12, 131. To improve the efficiency of initiation at the first methionine residue, MMTV fragments were cloned into a transcription vector (see Sect. 2.1) containing the 5' UTR from EMCVupstream of its cloning site [17] . Fig. 2 shows that constructs 1,2,3 and 4 are all capable of producing major protein products of predicted size, 36, 31,36 and 22 transmembrane domain. Trypsin treatment in the presence of microsomes was, therefore, used to define, by the degree of protection, the extent to which the protein has translocated across the microsomal membrane, and hence its membrane orientation. To simplify the interpretation of these experiments, the competitive N-glycosylation inhibitor N-acetyl-Asn-Tyr-Thr-carboxyamide was included. Fig. 4A illustrates that for the product of constuct 1, a significant decrease of approximately 3 to 5 kDa is seen after trypsin treatment (lanes 1 and 2) , suggesting that the majority of the protein is protected by the microsomal membranes. (B) shows that in the absence of microsomes no such protection is observed. These results are consistant with the hydrophobic domain (residues 38 to 63) anchoring the protein in a type-I1 membrane configuration. The product of construct 4 does not show evidence of microso-ma1 protection from trypsin cleavage (data not shown) consistant with its lack of a membrane translocation domain.The product from construct 1, (Fig. 4A, lane 1) in the presence of the glycosylation inhibitor is the same size as that seen in the absence of microsomes (Fig. 2 ). The correlation between carboxy sequence and Vg deletion phenotype has suggested a role for t h e carboxy terminus in determination of Vg deletion specificity; for a type-I1 membrane protein this region would be able to interact with the TcR without requirement for cellular processing. One possibility, currently under investigation, is that the type-I1 glycoproteins encoded by the 3'0RF of MMTV may bind either intracellularly or at the plasma membrane with class I1 and that this complex is what is recognized by endogenous superantigen-responsive T cells either thymically (leading to deletion) or peripherally (leading to expansion).

Cold Spring Harbor Symp

Proc. Natl. Acad. Sci

The authors would like to thank Ms. I.! Tikerpae for typing the manuscript.