Alternative Proteolytic Processing of Mouse Mammary Tumor Virus Superantigens

Abstract

Mouse mammary tumor viruses express a superantigen essential for their life cycle. It has been proposed that viral superantigens (vSags) require processing by prohormone convertases (PCs) for activity. We now observe, using a panel of mutant forms of potential PC cleavage sites and in vitro cleavage assays, that only the CS1 (position 68 to 71) and CS2 (position 169 to 172) sites are utilized by furin and PC5. Other members of the convertase family that are expressed in lymphocytes are not endowed with this activity. Furthermore, mutant forms of two different viral superantigens, vSag7 and vSag9, which completely abrogated in vitro processing by convertases, were efficient in functional presentation to responsive T-cell hybridomas. This effect was observed in both endogenous presentation and paracrine transfer of the vSag. Processing by convertases thus appears not to be essential for vSag function. Finally, we have identified the purified endosomal protease cathepsin L as another protease that is able to cleave convertase mutant vSag in vitro, yielding fragments similar to those detected in vivo, thus suggesting that proteases other than convertases are involved in the activation of vSags.

Mouse mammary tumor virus (MMTV) is a type B retrovirus responsible for the induction of mammary carcinomas in mice when viral integration occurs near a cellular oncogene (34). MMTVs can be transmitted either horizontally as infectious milk-borne particles, or vertically as integrants into the germ line (35). Following B-cell infection (14), MMTVs express a viral superantigen (vSag), encoded in the 3′ long terminal repeat, to expand the pool of infected B cells. vSags can stimulate a large proportion of T cells by interacting with the variable region of the T-cell receptor β chain (Vβ) (1). Infected B cells expressing the vSag induce T cells to proliferate and secrete lymphokines, driving B cells into proliferation. As a result of this cognate B-cell–T-cell interaction, T cells become infected and deliver the virus to the mammary gland, where viral progeny secreted in milk initiates another round of infection (14). In mice harboring integrated endogenous MMTVs, vSag expression leads to deletion of responsive T cells (1), protecting mice from infectious MMTVs sharing the same Vβ specificity and, hence, from mammary carcinomas (11).

MMTV vSags are type II integral glycoproteins having an extracellular carboxy terminus (7, 18). Their primary structure shows a high degree of sequence conservation, except for a carboxy-terminal polymorphic region that imparts Vβ specificity (4, 53). They carry five potential N-linked glycosylation sites, four of which are used (26). It has been shown that vSag activity could be transferred from class II⁻ to class II⁺ cells in vivo (47, 48), and we have shown that vSags could be transferred between cells separated by a semipermeable membrane in vitro (10). Clearly, this phenomenon requires proteolytic processing of vSags to generate a soluble fragment that can be transferred. The exact transferred fragment has not yet been identified.

Members of the mammalian subtilase family of endoproteases have been proposed as candidates for vSag cleavage, and vSags possess conserved potential dibasic sites (4). Proprotein convertases (PC) of the subtilase family are responsible for cleaving prohormones and proproteins at pairs of basic amino acid residues and have an absolute requirement for an arginine in the P1 position (31, 42). Two types of convertases exist. PC1, PC2, and possibly PC5A are located in secretory granules and are specific to neural and endocrine tissues. Furin, PACE4, PC5, and PC7 are broadly distributed, while PC4 is localized to testicular germ cells, and are responsible as a group for processing of proteins along the default constitutive secretory pathway (42). Characterization of furin specificity has shown that the consensus sequence recognized is RXXR, with a preference for an RX(K/R)R motif (29, 31), the latter motif being conserved at two positions in most of the vSags known (4).

The perfect conservation of the putative convertase-processing sites in vSags points toward a functional importance of these motifs. It has been shown that vSag processing could occur in transfected B cells (49, 50), but the exact position of cleavage was not determined. Mutagenesis studies suggested a requirement for furin-mediated processing to generate the active form of the superantigen (27, 37), but it remains unclear whether the mutations introduced in these studies perturbed the vSag structure. It has been shown that vSags are very sensitive to mutations, resulting in intracellular retention and degradation with loss of biological activity (25). Transfection of vSag7 into furin-deficient CHO cells resulted in much-reduced presentation, but some residual vSag activity remained (27). This activity could be abrogated by treatment with the arginine-specific protease inhibitor leupeptin (27). This inhibitor is inefficient toward convertases (28, 30), raising the possibility that alternate proteases are involved in vSag activation. To ascertain the importance of vSag processing by convertases, mutagenesis of endoprotease sites was performed on two different vSags. Using in vitro cleavage assays, we demonstrate that only certain convertases can cleave vSags. Furthermore, we show that while convertases can cleave vSags at the two conserved endoprotease sites, this processing is not essential for functional activity. Finally, alternate proteases such as cathepsin L can substitute for convertases to process vSags to fragments of sizes similar to those detected in vivo.

MATERIALS AND METHODS

Cell lines, transfections, and T-cell stimulation assays.

DAP-DR1 cells are DAP-3 murine fibroblasts transfected with the human DR1 class II molecule (22). CH12, a murine B-cell lymphoma line expressing IE^k was grown as described previously (43). BJAB, a human B-lymphoma cell line, was grown in RPMI medium (GIBCO-BRL, Burlington, Ontario, Canada) supplemented with 5% fetal calf serum (GIBCO-BRL). The murine T-cell hybridoma cell lines used were Kmls 13.11 (Vβ6.1; vSag7 reactive) and Vβ5#11 (Vβ5; vSag9 reactive). Hybridoma and DAP-DR1 cells were grown as described previously (45). AtT-20 is a murine pituitary cell line expressing all of the known convertases except PC5, and AtT-20-PC5 is AtT-20 stably transfected with murine PC5B (8). BSC40 cells were used to produce recombinant convertases by vaccinia virus infection.

DAP-DR1 cells were transfected with 10 μg of either wild-type (WT) or mutant vSag DNA using calcium phosphate coprecipitation (45) and selected with G-418 (GIBCO-BRL) at 1 mg/ml. CH12 cells were transfected by electroporation (250 mV, 960 μF, 10 ms) using a Gene-Pulser (Bio-Rad Laboratories, Inc., Hercules, Calif.) with 25 μg of vSag DNA. Stable transfectants were selected in RPMI 1640 medium supplemented with 5% fetal calf serum and G-418 at 1.3 mg/ml. Presentation of stably transfected vSags was performed as follows: DAP-DR1 or CH12 cells were cocultured in the presence of 6 × 10⁴ vSag-reactive hybridoma cells at different stimulator-to-effector ratios (1:1, 1:3, and 1:10). Expression of MHC class II molecules was monitored by flow cytometry with a FACScan (Becton Dickinson Immunocytometry Systems, San Jose, Calif.) using the anti-DR1 antibody L243 (ATCC HB55) or the anti-IE^k antibody 14-4-4S (36). For endogenous and transfer presentation of the CS12X triple mutant, DAP or DAP-DR1 cells were transiently transfected using DEAE-dextran as described previously (45). Transfection efficiency was evaluated by cotransfection of the human CD4 molecule and fluorescence-activated cell sorter analysis using the anti-CD4 antibody OKT4 (American Type Culture Collection). Transfer presentation assays were performed as described previously (10); in brief, 3 × 10⁴ BJAB cells were cocultured overnight with 3 × 10⁴ Kmls 13.11 cells together with various numbers, ranging from 3 × 10⁴ to 3 × 10³, of DAP cells transiently transfected with either WT or cleavage site mutant molecules. All cocultures for both direct and transfer presentations were performed in flat-bottom 96-well plates for 18 h at 37°C in a final volume of 200 μl, and interleukin-2 (IL-2) release was determined using the CTLL hexoaminidase assay (21).

Mutagenesis and cloning.

Site-directed mutagenesis of vSag7 (20) and vSag9 (43) was performed using the PCR overlap extension technique (15) and the oligonucleotides listed in Table 1. Mutants were cloned in pBlueScriptKS+ (Stratagene, La Jolla, Calif.) for sequencing and subcloned in the expression vector pHβ-Apr1-neo (13). For in vitro production of radiolabeled vSags, vSag7 cloned in pBlueScriptKS+ was amplified by PCR using the Universal primer (Pharmacia Biotech, Baie d'Urfé, Quebec, Canada) and the vSagHIS oligonucleotide, introducing an NheI site at the boundary of the transmembrane domain. PCR products were cloned between the NheI and BamHI sites of pRSETb (Invitrogen, Carlsbad, Calif.), introducing an N-terminal His tag.

TABLE 1.

Oligonucleotides used for mutagenesis and RT-PCR analysis

Oligonucleotide	Sequence
5′ CS1	GGTTCGTGCTGCCATGGCTCTCACCC
3′ CS1	GGGTGAGAGCCATGGCAGCACGAACC
5′ CS2	GAAAATAGCAAGAGCCGGTCGACCGCAGTC
3′ CS2	GACTGCGGTCGACCGGCTCTTGCTATTTTC
5′ CSX	GAAGGAAAAAAGAGTGTGTTTGTC
3′ CSX	GACAAACACACTCTTTTTTCCTTC
vSagHis	CTTTGCCTGGGGGCTAGCGGGAAGTTGCG
5′ Furin	TGAGCCATTCGTATGGCTACG
3′ Furin	GGACACAGCTTTTCTGGTGCA
5′ PACE4	GCATAGAAAGGAATCACCCAG
3′ PACE4	TGTAGCCATCACAGGAGCAG
5′ PC5	GTGTGGGCATCTGGCAATGGTGGA
3′ PC5	TTTGTCGGTCTGTGCTTTCCAC
5′ PC7	CCCACCCTGATGAGGAGAATG
3′ PC7	AAAGGCATCCGTCCCTCCTCA

RT-PCR analysis.

Determination of the pattern of convertase expression in cells used for functional presentation of vSags was performed by reverse transcription (RT)-PCR analysis. Total RNA was isolated using TRIZOL (Roche Diagnostics, Laval, Quebec, Canada), and first-strand cDNA was synthesized as follows. Total RNA (5 μg) was reverse transcribed in the presence of 5 μg of oligo(dT)_12–18 (Pharmacia Biotech), 20 U of RNAguard (Pharmacia Biotech), 2 mM dithiothreitol, 1 mM deoxynucleoside triphosphates, and 200 U of Moloney murine leukemia virus reverse transcriptase (GIBCO-BRL) for 2 h at 37°C. A 100-ng sample of total RNA was used for PCR amplification using 200 M each pair of primers under the following conditions: 1 min of denaturation at 94°C, 1.5 min of annealing at 55°C, and 1.5 min of extension at 72°C for 40 cycles. PCR products were fractionated on 1.2% agarose gels.

Recombinant convertase production.

Recombinant vaccinia virus constructs for convertase production have been described previously (8, 30). Enzymes were produced as follows. BSC40 monolayers at 70 to 80% confluence were washed three times with phosphate-buffered saline (PBS), and recombinant vaccinia virus infections (16) were carried out for 30 min at room temperature. Cells were incubated at 37°C for 18 h, culture supernatants were centrifuged to remove debris, concentrated 20-fold on Centricon-30 cartridges (Millipore Corp., Bedford, Mass.), and stored at −20°C in 40% glycerol until use. Enzymatic activity was determined by cleavage of the fluorogenic peptide substrate pERTKR-MCA (Peptides International, Louisville, Ky.), and fluorescence was monitored on an LS50B spectrofluorometer (Perkin-Elmer Corp., Norwalk, Conn.).

In vitro transcription-translation and cleavage assays.

In vitro transcription and translation of vSag7 were carried out using an Escherichia coli S30 system (Promega Corp., Madison, Wis.). In brief, 50-μl reaction mixtures containing 1 μg of DNA and 20 μCi of l-[³⁵S]methionine (1,200 Ci/mmol; New England Nuclear) were incubated at 37°C for 60 min. Products were batch purified using 20 μl of Ni-nitrilotriacetic acid agarose (Qiagen Inc., Mississauga, Ontario, Canada) and eluted with 50 μl of 1× PBS–200 mM imidazole. For in vitro cleavage with convertases, 10 μl of purified vSag7 was incubated overnight with 1 U of each convertase in 50 mM Tris · HCl (pH 7.0)–2 mM CaCl₂–0.1 mM 2-mercaptoethanol–0.01% Triton X-100. For cathepsin L cleavage, the imidazole was removed by overnight dialysis against 1× PBS on Slide-A-Lyzer microdialysis cassettes with a 10-kDa cutoff (Pierce Chemical Co., Rockford, Ill.). Ten microliters of labeled vSags was incubated with 10 ng of cathepsin L (Calbiochem, La Jolla, Calif.) in 85 mM sodium acetate–15 mM acetic acid–1 mM EDTA–2 mM dithiothreitol (pH 5.5) at 25°C (34). Products were fractionated by sodium dodecyl sulfate (SDS)–15% polyacrylamide gel electrophoresis (PAGE), transferred to polyvinylidene difluoride (PVDF) membranes (Roche Diagnostics), and exposed overnight on PhosphorImager screens (Molecular Dynamics, Sunnyvale, Calif.).

RESULTS

Only furin and PC7 are expressed in the cells used for vSag presentation.

We have previously reported presentation of vSags by different MHC class II-positive antigen-presenting cells (APCs) to their responsive hybridoma cells bearing specific Vβ elements (20, 45). Furin, PACE4, PC5, and PC7 are broadly distributed convertases responsible for processing of proproteins along the constitutive secretory pathway (8, 30, 42). In order to assess the potential involvement of each of these convertases in vSag activation, their expression was verified by RT-PCR in cells used in our vSag functional presentation assays. Positive controls included AtT-20 cells, which express all of the known convertases except PC5, and AtT-20-PC5 cells, which are AtT-20 cells transfected with mouse PC5 (8). vSag7-responsive hybridoma Kmls 13.11 cells were tested since they can provide a source of convertases in the medium, given the knowledge that both furin and PC7 can recycle between the cell surface and Golgi (28, 42). Hence, it was plausible that vSag processing could occur in trans at the APC surface. Figure 1 shows that cells used for vSag presentation (DAP-DR1 and CH12) and the vSag7-responsive hybridoma cells only express furin and PC7. Given that the AtT-20-PC5 cells have been transfected with PC5 and express nonphysiological levels of PC5 (8), we cannot rule out PC5 expression below the limit of detection of the RT-PCR assay used.

FIG. 1.

RT-PCR analysis of convertase expression. Using 100 ng of reverse-transcribed total cellular RNA, convertase expression was analyzed by PCR using the primers listed in Table 1. Lanes: 1, AtT-20-PC5; 2, AtT-20; 3, CH12; 4, DAP-DR1; 5, Kmls 13.11. The values on the left are molecular sizes in base pairs.

Generation and biochemical characterization of vSag cleavage site mutants.

The structure of vSags and the nomenclature used for convertase-processing sites are shown in Fig. 2. While the CS1 (68 to 71) and CS2 (169 to 172) endoprotease sites are conserved in most vSags, the CSX (194 and 195) site is not present in all MMTV sequences (4) and lacks the P4 arginine required for furin-like convertase recognition (29, 42). These convertase cleavage sites were subjected to in vitro mutagenesis (Fig. 2) for biochemical characterization and functional presentation to T cells. The two R→S mutations introduced at CS2 were chosen because a similar change present in the insulin receptor results in extreme insulin resistance (54). This mutated protein is refractory to furin cleavage but can still bind insulin, indicating that protein structure is preserved (54). All of the mutants referred to in this paper are identified by the mutated residues unless otherwise indicated (e.g., in CS12, the CS1 and CS2 sites are mutated while CSX is not).

FIG. 2.

Schematic representation of vSag structure. The endoprotease sites and mutations introduced are shown. TM, transmembrane region. At the bottom are approximate amino acid numbers.

Biochemical characterization of vSags has always proven difficult due to the low protein levels (12, 49, 50). In order to characterize the exact cleavage sites utilized by the different convertases and demonstrate that the mutations introduced abrogated convertase processing, we resorted to an in vitro approach. The extracellular portion of the mutants was cloned for expression in a coupled transcription-translation system. [³⁵S]methionine-radiolabeled material was subjected to cleavage by recombinant furin, PC5, and PC7 produced using a vaccinia virus expression system. Convertase activity was verified by fluorescent peptide substrate cleavage, and vSag cleavage was performed under conditions of maximal digestion. PACE4 was not tested, since it is not expressed in the cells used (Fig. 1). PC5 was included because of the presence of a sequence located at amino acids 253 to 260 (RERLAR↓AR), similar to one present in pro-Mullerian inhibitory substance (RGRAGR↓SK), a PC5 substrate (31).

Using radiolabeled double mutants with a single accessible convertase site, cleavage with recombinant convertases was performed, and Fig. 3A shows the expected molecular weights of cleavage products. Despite some spontaneous degradation in the uncleaved control, Fig. 3B shows that the CS2X double mutant, having a CS1 site exposed, can be cleaved by furin and PC5, yielding a 29-kDa fragment, removing the N-terminal His tag. The CS1X double mutant, having a free CS2 site, was cleaved to give two fragments of 18 and 14 kDa (Fig. 3C), proving that furin and PC5 can cleave at that position; the difference in intensity between the 18- and 14-kDa bands is most probably due to the number of methionines present in each fragment. Figure 3D shows a WT vSag7 digestion. The 29- and 14-kDa bands correspond to a partial digestion at the CS1 and CS2 sites, respectively, and the 18- and 11-kDa bands are derived from total digestion at both sites (Fig. 3A). It is clear from Fig. 3E that the CSX site is not cleaved by any of the convertases tested and that the mutations introduced at either CS1 or CS2 totally abrogate convertase processing. It is also apparent that the putative PC5 site located at amino acids 253 to 260 is not a substrate for PC5 and that PC7 has very little activity toward vSags.

FIG. 3.

In vitro cleavage of vSag7 mutants by recombinant convertases. vSags were produced with a coupled in vitro transcription-and-translation system as [³⁵S]methionine-labeled proteins. Equal amounts of Ni-nitrilotriacetic acid-purified material were subjected to cleavage with equal units of convertase activity. The mutated and WT vSags were digested overnight. Cleavage products were separated by SDS–15% PAGE, transferred to PVDF membranes, and exposed on PhosphorImager screens. (A) Expected molecular weights of cleavage products. (B) Cleavage of the CS2X mutant. (C) Cleavage of the CS1X mutant. (D) Cleavage of WT vSag7. (E) Cleavage of the CS12 mutant. Lanes: 1, furin; 2, PC5; 3, PC7; 4, furin plus EDTA; 5, uncleaved control. These data are representative of three independent experiments. The values shown are the molecular masses in kilodaltons.

Convertase site mutations do not affect functional presentation to T cells.

Both vSag7 and vSag9 were stably transfected into DAP-DR1, a murine fibroblastic cell line transfected with the DR1 class II molecule (22), and vSag7 was stably transfected in CH12, a murine B-cell lymphoma line. The mutant vSag9s were not introduced into CH12, as these cells express endogenous vSag9 (43). The efficiency of the WT and cleavage site mutants was tested in functional presentation to responsive T-cell hybridomas. Two different hybridoma lines were used: Kmls 13.11, which is responsive to vSag7 but not to vSag9, and Vβ5#11, which is responsive to vSag9 but not to vSag7. T-cell hybridoma stimulation was assayed by monitoring IL-2 production. The optimal APC-to-hybridoma cell ratio for presentation of WT vSags was determined to be 1 to 3 (data not shown). Figure 4 clearly shows that neither the CS1 nor the CS2 single mutations had a major effect on functional presentation of vSag9 in DAP-DR1 cells (Fig. 4A) with levels of IL-2 production by the T-cell hybridoma comparable to those obtained following stimulation with transfectants expressing WT molecules. Similar results were obtained with vSag7 in two different cell lines: DAP-DR1 (Fig. 4B) and CH12 (Fig. 4C). To eliminate the possibility that cleavage at either one of the free convertase sites was activating vSag7 or vSag9, the CS12 double mutants were tested and stimulation was comparable to that obtained with the WT molecules at the optimal ratio for stimulation (Fig. 4A, B, and C).

FIG. 4.

Convertase site mutants are efficiently presented to T cells. Different ratios of DAP-DR1 or CH12 stably transfected with the WT or cleavage mutant form of vSag7 or vSag9 were used to stimulate 6 × 10⁴ vSag-responsive T-cell hybridoma cells (Kmls 13.11 for vSag7 or Vβ5#11 for vSag9) for 18 h. Supernatants were harvested and tested for IL-2 activity. MFV, mean fluorescence value for DR1 (A and B) or IE^k (C). These data are representative of three independent experiments.

Mutants encompassing all putative convertase-processing sites can still be presented to T cells.

Biochemical analysis using B cells transfected with vSag7 has revealed a 16-kDa fragment that might correspond to processing at CSX (position 194 and 195) (49, 50). This raised the possibility that the presentation seen with the CS12 double mutants might be attributed to cleavage at the CSX position. However, the CSX site is not conserved in all MMTV isolates (4) and lacks a P4 arginine found in typical convertase sites (33), arguing that convertase-mediated processing at this position would be unlikely. Indeed, our data show that this position is not cleaved in vitro by either furin, PC5, or PC7 (Fig. 3E), showing that this motif is not a convertase substrate. To rule out the possibility that the presentation observed with vSag7 and vSag9 CS12 double mutants was due to cleavage at the CSX position, a nonclassical convertase site, it was mutated to a site previously shown to be uncleavable (5). A representative experiment is illustrated in Fig. 5A and shows that DAP-DR1 cells expressing vSag7 CS12X triple mutants, after transient transfection, were only 30% less efficient than the WT in endogenous presentation assays. This effect was observed at all effector-to-target cell ratios.

FIG. 5.

vSag7 CS12X triple mutant is efficiently presented to responsive T-cell hybridoma cells. (A) Endogenous presentation. Kmls 13.11 T-cell hybridoma cells (25 × 10³) were cocultured for 18 to 20 h with various numbers of DAP-DR1 cells transiently transfected with either WT vSag7 or the CS12X triple mutant. (B) Intercellular transfer of two populations of vSag7 CS12X mutants transiently transfected into DAP cells. BJAB cells (3 × 10⁴) were cocultured for 18 to 20 h with 3 × 10⁴ Kmls 13.11 T-cell hybridoma cells and various numbers of two different DAP-vSag7 cell clones (either the WT or the CS12X triple mutant). This figure is representative of two independent experiments.

It is known that vSags can be transferred from class II⁻ to class II⁺ cells both in vivo (47, 48) and in vitro (10). Clearly, vSags must be cleaved for such a transfer to occur, so we used the transfer assay, as outlined in Materials and Methods, to verify whether the CS12X triple mutant could still be shed in vitro. Figure 5B shows a representative experiment for transfer presentation between DAP cells transiently transfected with the vSag7 CS12X triple mutant and BJAB. Similar to what was observed with endogenous presentation (Fig. 5A), the CS12X triple mutant was 30% less efficient than the WT in transfer presentation (Fig. 5B). The presentation seen in the transfer assay shows that vSag processing must occur, either because the mutations introduced did not abrogate convertase processing or alternate proteases can cleave vSags. Since the biochemical analysis shows that the mutations introduced abrogate convertase cleavage (Fig. 3E) and the functional data show that convertase mutants can still stimulate T cells, it is clear that convertase processing is not required for vSag activity. The stimulation observed in the transfer assay with the CS12X triple mutant (Fig. 5B) suggests that alternate proteases could be involved in vSag processing.

Cathepsin L can cleave vSag7 into discrete fragments.

Given that the convertase site mutations abrogate in vitro processing of vSag7 and the mutant molecules are still efficient in functional presentation, we investigated whether other proteases might cleave vSag7. Localization studies have shown that vSags are present in the MIIC endosomal compartments enriched in H2-M and MHC class II molecules (12). Cathepsins are the most abundant lysosomal and endosomal proteases responsible for antigen generation and invariant chain (Ii) degradation. In addition, they are present in MIIC compartments (2, 6). Thus, we investigated whether cathepsins could cleave vSags. Cathepsin L is a ubiquitously expressed cysteine protease that has tryptase activity (2, 6) and, hence, would be expected to cleave at the dibasic motif conserved among vSags (4). Furthermore, it is inhibited by leupeptin (2), which was previously shown by Mix and Winslow to inhibit the residual vSag activity observed with vSag cleavage mutants. The radiolabeled WT vSag7 and the CS12X triple mutant were subjected to in vitro cleavage with purified cathepsin L, and a fragment of about 27 kDa appeared after cathepsin L digestion (Fig. 6). This 27-kDa fragment should result from cleavage between the CS1 and CS2 sites. Interestingly, a predominant 27-kDa vSag COOH fragment has been previously detected biochemically (12, 26, 49, 50). It is also clear from Fig. 6 that the convertase site mutations did not abrogate cathepsin L processing, raising the possibility that the efficient presentation observed with convertase site mutants might be due to alternate processing by cathepsin L or other related lysosomal or endosomal enzymes.

FIG. 6.

In vitro cleavage of WT vSag7 or the CS12X triple mutant by cathepsin L. Equal amounts of purified [³⁵S]methionine-labeled proteins produced with a coupled in vitro transcription-translation system were subjected to cleavage with 10 ng of purified cathepsin L, fractionated by SDS–15% PAGE, transferred to PVDF membranes, and exposed to a PhosphorImager screen. Time of cleavage is in minutes. These data are representative of three independent experiments. The values on the left are the molecular masses in kilodaltons.

DISCUSSION

We have provided a rigorous analysis of the processing requirements of vSags using two complementary approaches: first, by performing in vitro processing of the cleavage site mutants by recombinant convertases to determine the exact sites utilized by each convertase (Fig. 3); second, by studying the functional presentation of two different vSags using two different types of APCs, including B lymphocytes, the natural host cells for vSag presentation (Fig. 4 and 5). Given that all of the convertase site mutants tested were efficient in functional presentation, alterations of vSag structure can be easily ruled out.

It is evident from the results presented here that cleavage at the CS1 proximal convertase site is not required for vSag activity, in agreement with results obtained by other groups (49, 50). Furthermore, exogenous MMTV-SIM has superantigenic activity (24) but possesses a CS1 site lacking the canonical P4 arginine required for convertase recognition. Given the high level of phylogenetic conservation of this position, it appears likely that this region serves a function different from the one relevant to superantigenic activity.

It is clear from our in vitro cleavage assays that the dibasic CSX site is not a substrate for convertases (Fig. 3E). This confirms and extends a previous report that showed that furin could not cleave at the CSX position (37). The 16-kDa COOH-terminal fragment previously detected biochemically (49, 50) and assumed to arise from cleavage at the CSX position would thus be derived from cleavage by another protease. This is supported by our data showing that the CS12X triple mutant, which cannot be cleaved by convertases, can stimulate T cells in the transfer assay (Fig. 5B), where cleavage is expected to be required.

The data presented here show that convertase-mediated processing at the CS2 site is not essential for presentation of vSag9 and vSag7 to T cells. While the CS2 mutants were efficient in functional presentation (Fig. 4), the mutations introduced completely abrogated convertase cleavage (Fig. 3E). This is in sharp contrast to a report suggesting that processing at CS2 is essential for presentation (37). It is likely that the mutations introduced in that study (RKRR→GEEF) have altered the structure of the superantigen, leading to intracellular retention or degradation since this mutant was not expressed at the cell surface (37). In support of this possibility is a report that showed that several different point mutations abolished vSag cell surface expression and presentation because of retention in the endoplasmic reticulum (25). Processing at the CS2 position appears to occur in vivo, given that the 18-kDa fragment detected by Western blot analysis corresponds to cleavage at that position (49, 50), and Fig. 3C shows that an 18-kDa fragment is generated by both furin and PC5. However, the presence of such a fragment does not prove that processing at that position is required for vSag activity.

Using a mutant furin-deficient CHO cell line transfected with vSag7 and the murine MHC class II molecule IE^k, Mix and Winslow (27) have shown that reintroduction of furin could increase vSag presentation to T cells, arguing that furin participates in generating active vSag. Nevertheless, the residual vSag activity in these furin-deficient cells could only be inhibited by leupeptin (27), an inhibitor inefficient toward the subtilase-type convertases (28, 30). This suggested that alternate proteases might contribute to the activation of vSags. Recently, more conservative mutations of the putative CS2 and CSX sites have been introduced by inserting sequences naturally occurring in MMTV isolates (51). Using WT CHO cells as APCs, these authors showed that removal of the CS2 site (while keeping a WT CSX site) abolished presentation and that cell surface expression was readily detectable (51). The discrepancy between these reports proposing an essential requirement for convertase-mediated vSag processing and ours might be due to differences in the cell lines used. While cathepsins are ubiquitously expressed (38), their function in antigen processing is tissue dependent (32).

We do not exclude the possibility that convertase-mediated processing generates the active superantigen, but we propose that alternative proteases like cathepsins can substitute for this set of proteases. Alternatively, it is possible that dibasic sequence-specific convertases might activate a cell surface enzyme which is critical for cleavage of vSag. Possible candidates include surface mammalian proteins containing a disintegrin and metalloprotease domains (ADAMs) (52), such as ADAM-17 (tumor necrosis factor alpha-converting enzyme) (3) and ADAM-10 (Kuzbanian) (40). Another argument for alternate processing resides in experiments aimed at biochemical characterization of vSags. The major vSag COOH-terminal cleavage product detected in B cells, after N-glycanase treatment to remove glycosylation, has a molecular mass of 27 kDa (12, 26, 49, 50). This 27-kDa COOH-terminal product would correspond to cleavage between the CS1 and CS2 positions. Given that in vitro processing by convertases was not observed in this region, alternate proteases must be involved in vSag processing.

Little is known about vSag intracellular trafficking, mainly because of major technical difficulties in detecting the protein. The protein appears to be highly unstable (19) and targeted for degradation in the endoplasmic reticulum if mutations are introduced (25) or glycosylation is perturbed (26). It has been proposed that vSags and MHC class II molecules might interact in the endoplasmic reticulum and traffic together to the cell surface (49, 50). Such a hypothesis was also supported by the fact that increasing MHC class II levels has a much more profound effect on T-cell stimulation than increasing vSag levels (23). However, it has been reported that vSags traffic independently of MHC class II molecules using the default exocytic pathway, but increased concentrations of vSags have been detected in specialized MIIC compartments that are enriched in MHC class II molecules and cathepsins (12).

Given the high concentration of cathepsins in MIIC compartments (2, 6), cathepsin-mediated proteolytic processing is a possibility that must be considered. Cathepsins are ubiquitously expressed (2, 6, 38) but are mostly active in professional APCs like B cells. While the fibroblastic DAP cells used in functional presentation are not professional APCs, they appear to possess efficient cathepsin-like proteolytic machinery since they are competent in Ii processing (17; unpublished observation), a cathepsin-mediated event (6, 32, 39, 46). Hence, expression of cathepsins in the cells used in our presentation assays cannot be considered a limiting factor.

We focused on cathepsin L because it has tryptase activity (2) and it could have processed vSags at the same dibasic motifs recognized by convertases. However, it is apparent that alternate sites are recognized given the in vitro cleavage of the CS12X triple mutant by cathepsin L and the efficient vSag cell-to-cell transfer observed with that mutant in functional presentation. Since there are no specific inhibitors of cathepsin L, it is difficult to test its contribution in an in vivo setting. In addition, attempts to inhibit cathepsin activity would have a drastic effect on MHC class II function (6, 32, 39, 44, 46), a parameter that is absolutely essential for vSag presentation (1, 20, 23). Recent reports of cathepsin L knockout mice (32) indicated that although cathepsin L is expressed in various tissues, its activity is more restricted to the thymic environment. However, antigen processing and Ii degradation in the periphery are more dependent on a newly identified cathepsin, cathepsin S (32, 39, 44, 46). It is important to consider that exposure to vSag occurs in the periphery (in the case of horizontal transfer through milk of infected mothers) and/or the thymus (in the case of vertical transfer through germ line integrants). Given the overlap in substrate specificity (6) and tissue distribution (38) of active cathepsins, the possible involvement of other members of this family, like cathepsin S, in vSag processing cannot be excluded.

We have shown that convertases can cleave vSags at two of their consensus processing sites. While convertase-mediated cleavage probably occurs in vivo, this does not appear to be an absolute prerequisite for biological activity of two different vSags. The possible involvement of different proteases in vSag activation indicates that MMTV is able to use alternative pathways to manipulate the immune system and ensure its own survival.