MHC class I antigen processing distinguishes endogenous antigens based on their translation from cellular vs. viral mRNA

Abstract

To better understand the generation of MHC class I-associated peptides, we used a model antigenic protein whose proteasome-mediated degradation is rapidly and reversibly controlled by Shield-1, a cell-permeant drug. When expressed from a stably transfected gene, the efficiency of antigen presentation is ∼2%, that is, one cell-surface MHC class I–peptide complex is generated for every 50 folded source proteins degraded upon Shield-1 withdrawal. By contrast, when the same protein is expressed by vaccinia virus, its antigen presentation efficiency is reduced ∼10-fold to values similar to those reported for other vaccinia virus-encoded model antigens. Virus infection per se does not modify the efficiency of antigen processing. Rather, the efficiency difference between cellular and virus-encoded antigens is based on whether the antigen is synthesized from transgene- vs. virus-encoded mRNA. Thus, class I antigen-processing machinery can distinguish folded proteins based on the precise details of their synthesis to modulate antigen presentation efficiency.

CD8⁺ T lymphocytes play a key role in immunosurveillance by eradicating tumor cells and cells harboring viruses and other intracellular pathogens. Recognition is based on the interaction of the clonally restricted T-cell receptor with MHC class I molecules bearing oligopeptides. Peptides are predominantly generated from defective ribosomal products (DRiPs), nascent translation products that are degraded rapidly either by design (i.e., dedicated for antigen presentation) or necessity (defective forms of proteins that can interfere with cell function) (1). The use of DRiPs greatly speeds recognition of virus-infected cells, because many peptides derive from otherwise highly stable proteins. For viral infections, the efficiency of recognition afforded by DRiPs is of the essence, because viruses can replicate within hours, and killing must be fast to be effective. Time is not limiting for tumor cells but efficiency is critical, because T cells often recognize antigens derived from gene products expressed at extremely low levels. Indeed, T cells may be the only means of detecting expression of the source antigen (2).

Only a handful of studies have addressed the efficiency of peptide generation, a critical issue with broad implications for the efficiency of protein synthesis and cytosolic compartmentalization of antigen processing (3). Pamer and colleagues seminally reported that class I–peptide complexes are generated with an efficiency of 3–25% (i.e., 3–25 complexes generated per 100 proteins degraded) from Listeria monocytogenes proteins secreted into the cytosol of mouse macrophage-like cells (4, 5). Using recombinant vaccinia viruses (rVVs) to express a rapidly degraded full-length chimeric protein, we reported that the efficiency of complex (K^b–SIINFEKL) generation was much lower in a variety of mouse fibroblast and macrophage/dendritic cell (DC)-like cell lines, 0.25%–0.05% (1/400–1/2,000), with an efficiency of ∼2% for SIINFEKL synthesized as the MSIINFEKL minigene product (6). Fruci et al. devised a clever method for quantitating minimal peptides released into the cytosol based on liberation from chimeric GFP-ubiquitin fusion proteins by ubiquitin hydrolases, reporting an efficiency of 0.2% (1/500) for EBV-transformed human B cells expressing permanently transfected genes (7). Extending this approach, Wolf and Princiotta demonstrated that ubiquitin hydrolase-liberated SIINFEKL is presented at high (15–20%) efficiency in a macrophage-like cell line whether synthesized by VV or intracellular Listeria (8), showing that two very different systems of antigen introduction into the cytosol need not differ widely in processing efficiency.

Taken together, these findings demonstrate system-dependent-wide variation in the efficiencies of generating complexes from full-length proteins and oligopeptides. To better understand these differences, here we measure the context dependence of antigen-processing efficiency using a number of antigens, cells, and expression vectors.

Results

K^b–SIINFEKL Is Generated from SCRAP at Remarkably High Efficiency in EL4 Cells.

Building on the work of the Wandless laboratory (9), we recently described SCRAP, a chimeric antigen whose stability is precisely controlled by the cell-permeant drug Shield-1 (10). SCRAP consists of a Shield-1 interaction domain NH₂-terminally fused to SIINFEKL and GFP (10) (Fig. 1A). In the absence of Shield-1, nascent SCRAP is degraded with a t_1/2 of 16 min. In the presence of Shield-1, a stable pool of SCRAP accumulates that is degraded with a t_1/2 of 30 min upon removal of Shield-1 (10).

Fig. 1.

Model antigens and a schematic of SCRAP presentation and quantification. (A) The various constructs used in this study are depicted. The red “S” box denotes the SIINFEKL peptide, influenza A virus (IAV) nucleoprotein is in orange, and the green “Ub” box represents ubiquitin. (B) EL4/SCRAP cells were washed in a mild citric acid buffer (pH 3.0) to remove existing K^b–SIINFEKL complexes and cultured in the presence or absence of 5 μM Shield-1. At the indicated times, K^b–SIINFEKL complexes and GFP levels were determined by flow cytometry. (C) Same as in A, except Shield-1 was removed following an initial 3.5-h incubation and, following a second acid wash, cells were cultured in the presence of cycloheximide (CHX). (D) An example of a Western blot for GFP using recombinant GFP standards and lysates from EL4/SCRAP cells treated with or without Shield-1 for 3 h.

Fig. 1 B and C illustrate the behavior of SCRAP when expressed as a permanently transfected gene in EL4 cells, an H-2^b T-cell lymphoma. Via flow cytometry, properly folded fluorescent SCRAP and K^b–SIINFEKL are measured simultaneously in live cells based on, respectively, GFP and directly conjugated 25-D1.16 mAb (11). Because GFP fluorescence requires proper folding and not all SCRAP is properly folded, only measuring fluorescent SCRAP will underestimate the amount of SCRAP present in cells. Via immunoblotting, however, we can measure all SCRAP forms in total cell extracts that interact with a mAb specific for GFP (Fig. 1D). Using purified GFP of known concentration to generate a standard curve, we could calculate the average number of immunoreactive SCRAP molecules expressed per cell (Fig. 1D). We simultaneously determined absolute numbers of directly conjugated 25-D1.16 bound to cells using flow cytometry standardization beads as described in Materials and Methods. At the start of the efficiency measurement period, we treated cells with mild acid (pH 3) to destroy preexisting K^b–SIINFEKL complexes, thereby improving the signal-to-noise ratio.

The SCRAP system enabled us to determine the efficiency of peptide generation from nascent proteins vs. retirees (12). Retirees are folded, functional proteins that are selected for degradation either stochastically according to the classical view (13) or, in this case, due to unfolding based on Shield-1 withdrawal. For nascent proteins, we determined the amount of SCRAP rescued over a 4-h period by Shield-1 and the concomitant decrease in K^b–SIINFEKL expression (Fig. 1B). For retirees, we accumulated a pool of SCRAP by treating for 4 h with Shield-1, and then induced degradation by removing Shield-1 while measuring the increase in K^b–SIINFEKL expression concomitant with SCRAP degradation (Fig. 1C) in the presence of cycloheximide to prevent newly synthesized and rapidly degraded SCRAP from entering the antigen presentation pathway.

The efficiency of generating K^b–SIINFEKL complexes was statistically indistinguishable between Shield-1–sensitive nascent and retiree pools (Tables 1 and 2), and was calculated to be ∼2%. Remarkably, this is 40-fold more efficient than generation of K^b–SIINFEKL complexes from VV-expressed rapidly degraded nucleoprotein (NP)- SIINFEKL (S)-GFP [ubiquitin (Ub)-R-S-GFP in Fig. 1A], and 10-fold more efficient than misfolded NP-S-GFP due to the insertion of a KEKE sequence in the NP domain (6, 14).

Table 1.

Efficient antigen presentation of peptides derived from rapidly degraded self-proteins

Nascent	Experiment	Kb–SIINFEKL complexes	GFP molecules	Calculated efficiency (%)
	1	5,445	2.54 × 105	2.1
	2	2,958	5.62 × 105	0.5
	3	3,104	6.83 × 104	4.5
	Average			2.4

EL4/SCRAP cells were treated as in Fig. 1. Peptide–MHC complexes were determined by quantitative flow cytometry. Shield-1–sensitive nascent SCRAP molecules were quantitated via immunoblotting. Efficiencies were calculated at 4 h after addition of Shield-1.

Table 2.

Retired SCRAP is presented with similar efficiency as nascent, rapidly degraded SCRAP

Retired	Experiment	Kb–SIINFEKL complexes	GFP-degraded	Calculated efficiency (%)
	1	297	15,556	1.9
	2	793	33,106	2.4
	3	1,156	51,454	2.2
	Average			2.2

EL4/SCRAP cells were treated as in Fig. 1C. Peptide–MHC complexes were determined by quantitative flow cytometry. Retired SCRAP molecules were quantitated via immunoblotting. Efficiencies were calculated 2 h following the removal of Shield-1.

These large differences could be related to differences in the cell type studied (EL4- vs. L-K^b), antigen (SCRAP vs. NP-S-GFP), or expression system (transfection vs. VV infection). To distinguish between the types of antigens, we measured the efficiency of antigen presentation from VVs expressing SCRAP vs. NP-S-GFP (Fig. 2). L-K^b cells infected with VV SCRAP in the presence of Shield-1 expressed nearly identical amounts of GFP as detected by flow cytometry as cells infected with rVV expressing NP-S-GFP, and generated essentially identical amounts of K^b–SIINFEKL. In the absence of Shield-1, nearly identical amounts of fluorescent GFP were detected compared with Ub-R-S-GFP (6), and peptides were generated at half the rate from SCRAP vs. rapidly degraded NP. Therefore, an intrinsic difference between SCRAP and NP-S-GFP (or Ova, also reported in the original study) does not account for the high efficiency of peptide generation from SCRAP in EL4 transfectants.

Fig. 2.

SCRAP protein expressed by rVV is similar to other model antigens. rVV expressing either SCRAP, NP-S-GFP, or Ub-R-NP-S-GFP was used to infect L-K^b cells. Cells were cultured with or without Shield-1 (for SCRAP infection) and analyzed at the indicated times for GFP (Left) or K^b–SIINFEKL (Right) by FACS analysis.

Are there cell type-related differences in antigen-processing efficiency? Because EL4 cells are highly resistant to VV infection, we compared the antigen-processing efficiencies of EL4 vs. L-K^b cells using recombinant vesicular stomatitis viruses (rVSVs) expressing Venus fluorescent protein (VFP) fused to ubiquitin followed by the SIINFEKL peptide (VFP-Ub-S). Following infection with VSV-VFP-Ub-S, the cell lines expressed nearly identical amounts of K^b–SIINFEKL (Fig. 3A, red trace), although EL4 cells expressed almost 50% more fluorescent protein signal (Fig. 3A, black trace). As peptides are in excess under these conditions (3), this demonstrates that EL4 and L-K^b cells have a similar overall capacity to generate surface K^b–SIINFEKL complexes from a cytosolic SIINFEKL pool. This implies similar functional levels of the transporter associated with antigen processing (TAP), K^b, and intracellular trafficking machinery in EL4 and L-K^b cells.

Fig. 3.

Similar antigen presentation kinetics between different cell types and viruses. (A) L-K^b and EL4 cells were infected with rVSV expressing either Venus-Ub-SIINFEKL (Left) or NP-S-GFP proteins (Right), and antigen presentation as well as fluorescent protein expression were monitored by FACS. (B) L-K^b cells were infected with rVV and rVSV viruses expressing NP-S-GFP and monitored as in A.

We next compared the abilities of EL4 and L-K^b cells to generate K^b–SIINFEKL complexes from a full-length stable protein expressed by VSV. Following infection with rVSV-NP-S-GFP, EL4 and L-K^b cells expressed nearly identical levels of fluorescent NP-S-GFP between 1 and 3 h postinfection (hpi) (Fig. 3A). Levels of K^b–SIINFEKL were similar, although L-K^b cells demonstrated a more complicated pattern, with increased immediate presentation and a 50% decrease in complex generation between 2 and 3 hpi Overall, however, the efficiency of generating K^b–SIINFEKL complexes from DRiPs derived from this full-length stable protein was similar between the two cell types.

We also examined the efficiency of K^b–SIINFEKL generation in L-K^b cells infected with rVV-NP-S-GFP or rVSV-NP-S-GFP with virus dose adjusted to yield equivalent levels of NP-S-GFP fluorescence. Generation of K^b–SIINFEKL complexes was superimposable (Fig. 3B), demonstrating that antigen presentation can occur at similar efficiency from two very different viral vectors.

Antigen Processing Efficiency is Dependent on the mRNA Source of the Antigen.

The findings demonstrate that the efficiency of generating K^b–SIINFEKL complexes is similar between EL4 and L-K^b cells and between rVV and rVSV expressing NP-S-GFP. Could the specifics of gene expression related to host- vs. virus-directed expression influence the efficiency of antigen processing?

To explore this possibility, we transiently transfected SCRAP in HeLa K^b cells and compared their antigen-processing efficiency to HeLa K^b cells infected with rVV SCRAP at a dose titrated to match the rate of SCRAP synthesis in the transfected cells (Fig. 4 A and B). Although the cells were synthesizing near-identical amounts of SCRAP from host vs. VV mRNA, transfectants generated K^b–SIINFEKL complexes at a much higher rate from the Shield-1–sensitive pool (Fig. 4 A and B). In five separate experiments (Table 3), we determined that the average fold increase in antigen presentation efficiency of transfect-encoded protein compared with VV-encoded protein was 8.3 (±1.6 SD units, P < 0.05).

Fig. 4.

Virus-expressed antigens are presented at lower efficiencies than self-antigens. (A) HeLa K^b cells were either infected with rVV SCRAP or transfected with SCRAP DNA constructs, and antigen presentation as well as GFP expression were determined 4 hpi or post-acid wash. (B) Representative histograms after 4 h of Shield-1 treatment of SCRAP-expressing cells (blue trace) compared with non-SCRAP-expressing cells (red trace). K^b–SIINFEKL staining with 25-D1.16 mAb (Right) is restricted to GFP-positive cells. (C) EL4/SCRAP or HeLa K^b cells transfected with SCRAP DNA were infected with VSV or rVV, respectively, and both K^b–SIINFEKL and GFP levels were determined 4 hpi. The efficiency of presentation was determined by dividing the MFI of K^b–SIINFEKL staining by the MFI of GFP. Values were compared with uninfected cells and are plotted as a percentage. (D) Levels of K^b–SIINFEKL derived from DRiP substrates (in the presence of Shield-1) from SCRAP expressed as transfected DNA or by rVV in HeLa K^b cells were normalized to GFP expression. Cells (both transfected and nontransfected) were briefly washed in mild acid and cultured for 2 h in complete media. Cells were then harvested and were either left uninfected or infected with control rVV at an MOI of 10, followed by an additional 4 h in culture before FACS analysis. Background levels of K^b–SIINFEKL were determined immediately following rVV infection. These data are representative of three independent experiments.

Table 3.

Transfected SCRAP is presented more efficiently than SCRAP expressed by rVV infection

	Transfected SCRAP			rVV-infected SCRAP
Experiment	Kb–SIINFEKL	GFP-degraded	Efficiency (%)	Kb–SIINFEKL	GFP-degraded	Efficiency (%)	Fold increase (transfectant/infected)
1	2,550	2.9 × 105	0.88	857	8.9 × 105	0.10	8.8
2	597	1.1 × 105	0.54	419	7.1 × 105	0.06	9.0
3	879	3.5 × 105	0.25	289	9.4 × 105	0.03	8.3
4	1,543	2.0 × 105	0.79	830	5.9 × 105	0.14	5.6
5	1,680	2.5 × 105	0.69	440	6.0 × 105	0.07	9.9
						Average	8.3 ± 1.6

HeLa K^b cells were transfected with DNA constructs encoding SCRAP or infected with rVV expressing SCRAP and treated with or without Shield-1 for 4 h. The number of K^b–SIINFEKL complexes derived from a defined number of SCRAP molecules, determined by quantitative immunoblotting for GFP, is shown. The difference in efficiencies between transfected and infected cells is statistically significant (Student’s t test, P < 0.05).

Notably, the efficiency of K^b–SIINFEKL generation from transfected SCRAP in HeLa vs. EL4 cells was similar (0.6% vs. 2.4%), particularly when considering that a mismatch between the mouse class I molecule (K^b) and human antigen-processing machinery (TAP, tapasin, other chaperones, and endoplasmic reticulum aminopeptidase) likely lowers peptide loading efficiency. There was a reasonably good match between the efficiency of generating K^b–SIINFEKL from the Shield-1–sensitive pool of VV-encoded SCRAP (1/1,250) and our previous determination of the efficiency of generating K^b–SIINFEKL from rapidly degraded NP in various mouse cell types (as high as 1/1,400 in DC 2.4 cells) or slowly degraded, misfolded NP (1/440) (6).

What is the contribution of virus-induced alterations in cell function to modulating antigen presentation efficiency? We infected EL4/SCRAP cells with VSV or HeLa K^b/transient SCRAP cells with rVV, acid-stripped preexisting K^b–SIINFEKL complexes, and measured the efficiency of K^b–SIINFEKL generation from nascent, Shield-1–sensitive SCRAP synthesized during viral infection (Fig. 4C). Infection with VSV or VV resulted in a partial decrease in both K^b–SIINFEKL and GFP expression, but had only a minor (less than 20%) inhibitory effect on the overall efficiency of antigen presentation.

Taken with the data in Table 3, these findings point to the remarkable conclusion that the antigen-processing machinery has the capacity to distinguish folded proteins based on their synthesis from cell- vs. virus-encoded mRNA. The same peptide from the same folded protein pool in the same cell line is generated approximately eightfold more efficiently when synthesized from cell- vs. VV-encoded RNA.

What about DRiP efficiency? The SCRAP system allows us to separate the presentation of peptides from retired vs. DRiP substrates, which are the only source of peptides when saturating amounts of Shield-1 are added to SCRAP-expressing cells (10). We compared the amount of GFP to K^b–SIINFEKL synthesized in acid-stripped HeLa cells treated with Shield-1 and expressing SCRAP from transfected vs. VV-expressed genes. This revealed that VV infection of HeLa K^b cells per se has little effect on K^b–SIINFEKL processing from DRiPs derived from cell-encoded SCRAP (Fig. 4D, infected), whereas DRiPs derived from VV-encoded SCRAP have approximately twofold reduced efficiency. Thus, viral infection is either associated with fewer SCRAP DRiPs synthesized or diminished efficiency of DRiP conversion to K^b–SIINFEKL complexes.

Discussion

We have made two findings, each startling, that point to major gaps in our understanding of the class I antigen-processing pathway.

First, we show that cellular gene products in the form of the model antigen SCRAP can be processed from a folded state at remarkably high efficiency, at 2% (1 K^b–SIINFEKL complex generated per 50 precursors degraded). This value is ∼40 times higher than our previous determination of K^b–SIINFEKL generation from a rapidly degraded VV-encoded protein (6). All things being equal, a 2% antigen-processing efficiency is extremely difficult to square with cellular protein economy (6, 12, 15). Even in the absence of DRiPs and other rapidly degraded polypeptides (16), normal cellular protein turnover amounts to ∼10⁹ d⁻¹, or 7 × 10⁵ proteins min⁻¹, potentially spawning peptides that bind a given class I allomorph at 3.5 × 10⁶ min⁻¹ [a 500-residue protein possesses 500 potential n-mer peptides, of which 5 (1%) bind with immunogenic affinity (17)]. A 2% efficiency would mean that greater than 7 × 10⁴ peptides are loaded onto class I molecules min⁻¹, which would competitively preclude K^b–SIINFEKL generation, because there are only ∼10² class I molecules exported min⁻¹ per cell. One or more of these numbers, therefore, must either be in error or unrepresentative.

Two factors potentially contribute to the remarkably high efficiency of SCRAP retiree processing. First, SIINFEKL may be a far above average peptide. Many other defined antigenic peptides demonstrate a similar affinity for class I molecules and copy number (18), but these may all be far above average (the Massachusetts Institute of Technology student-body effect). Second, there may be something peculiar about SCRAP that enables it to access the class I pathway at high efficiency. As a chimeric protein derived from several sources originating in different organisms, SCRAP did not experience the honing forces of evolution that assure its integration into the cellular landscape. In some manner, cells may be able to detect SCRAP as a foreign entity and shunt it for high-efficiency antigen presentation. This may also contribute to the similarly high efficiency of antigen processing from Listeria proteins secreted into the cytosol reported in the seminal work of Pamer and colleagues (4, 5), the high efficiency of SIINFEKL presentation when liberated from Listeria- or VV-synthesized Ub fusion proteins (8), and the remarkably high class I occupancy exhibited by a some peptides (19–21).

Our second startling finding is the approximately eightfold difference in efficiency of K^b–SIINFEKL generation from retired SCRAP synthesized from cellular vs. VV mRNA. Ironically, from the perspective of immunosurveillance, viral SCRAP is less efficiently processed. As discussed above, however, this may in part relate to the artificial nature of SCRAP, and not to an intrinsic lower efficiency of viral antigen processing. Indeed, Reits et al. (22) reported that influenza A virus infection rapidly increases the overall supply of TAP-transported peptides, a critical finding for immunosurveillance that still begs an explanation.

However artificial, it is intriguing and potentially important that cells can distinguish cellular from viral SCRAP. We cannot attribute this to VV-induced changes in cellular physiology, because VV infection does not greatly modulate the efficiency of cellular Shield-1–sensitive SCRAP presentation. How can the antigen-processing machinery distinguish between these ostensibly identical antigen sources?

VV SCRAP is likely synthesized in regions of the cytosol organized specifically for viral translation that serve as the precursors for viral factories (23, 24), whereas cellular SCRAP is synthesized on “normal” ribosomes. It is therefore possible that there are subtle (or even not so subtle) differences between the two in their folding/posttranslational modifications. Indeed, because tRNA aminoacyl synthetases are recruited to locally translating ribosomes (24), alterations in their specificity (25) could generate SCRAP with altered primary sequences that enhance antigen processing. Still, there is nothing known about antigen processing to explain why one form of a folded gene product would be a superior source of a given peptide, pointing to a significant lacuna in our understanding.

One explanation is that SCRAP synthesized from cellular mRNA is better-partitioned into a local antigen-processing compartment(s) defined by lack of competition (26) or localized presentation (27). Whereas it is hard to imagine compartmentalizing folded proteins in a manner kinetically dissociated from their synthesis by hours, it is possible that nascent proteins are immediately sequestered in small amounts in regions that favor antigen processing when the protein is eventually degraded by proteasomes. These concepts are admittedly vague, the real point being that we still have a lot to learn about MHC class I antigen processing.

Regardless of the mechanism(s), the present findings point to features of antigen processing that could be handy in discriminating peptides encoded by innocuous vs. danger-associated mRNAs. This would be of obvious utility in immunosurveillance of tumors. Here the immune system often succeeds in finding the needle in the haystack, by recognizing peptides generated at high efficiency from genes whose translation is minimal (2, 28).

Materials and Methods

Cells, Antibodies, and Viruses.

L-K^b, EL4, EL4/SCRAP, and HeLa K^b cell culture has been previously described (10, 29). Recombinant vaccinia virus expressing influenza nucleoprotein fused to the SIINFEKL peptide and GFP (NP-S-GFP) or its rapidly degraded counterpart (Ub-R-NP-S-GFP), SCRAP, or recombinant virus expressing no protein were previously described (10, 3). Recombinant vesicular stomatitis virus expressing NP-S-GFP has been previously described (30). Generation of rVSV expressing Venus-Ub-SIINFEKL (VFP-Ub-S) was as follows. Plasmid pVSV-XN2 was kindly provided by J. Rose (Yale University, New Haven, CT). Venus-Ub-SIINFEKL was polymerase chain-amplified from recombinant vaccinia virus (3) with the primers Venus 5′ XhoIXmaIMluI (5′-CTCGAGCCCGGGACGCGTCCATGGTGAGCAAGGGCGAGGAGC-3′) and Ub-SIINFEKL 3′ NheI (5′-GGCTAGCTTATCATAGCTTTTCGAAGTTGATGATCGAACCACCTCTTAGTCTTAAG-3′) using Platinum Taq High Fidelity DNA polymerase (Life Technologies). PCR product was then cloned into pCR4-topo (Life Technologies) according to the manufacturer’s instructions. Insert DNA containing Venus-Ub-SIINFEKL was excised with XhoI and NheI restriction enzymes (Roche) and ligated with similarly digested pVSV-XN2. Final plasmid pVSV-Venus-Ub-SIINFEKL was verified by restriction digestion patterns and DNA sequencing of the inserted gene. Baby hamster kidney cells (BHK-21; American Type Culture Collection) were maintained in DMEM supplemented with 10% (vol/vol) FBS. Cells were infected with vTF7-3 vaccinia virus expressing T7 polymerase at a multiplicity of infection (MOI) of 10 for 1 h, followed by transfection with plasmids pBS-N, pBS-P, pBS-L (31, 32), and pVSV-VFP-Ub-S using Lipofectamine 2000 (Life Technologies) according to the manufacturer’s instructions. Supernatant media containing recombinant virus were recovered at 48 hpi and used to make viral stocks. Viral titers were determined by plaque assays on BHK cells. The mAb 25D-1.16 was previously described (11). Anti-GFP antibodies were from Roche. Infrared secondary antibodies were from LI-COR.

Cloning and Transfections.

To enhance SCRAP expression in transient transfectants, a SCRAP construct was generated by PCR amplifying the original SCRAP cassette with the primers upstream SacI 5′-TCTAGAGAGCTCCCACCATGGGAGTGCAGGTGGAAACCA-3′ and downstream XhoI 5′-AGATCTCTCGAGTTACTTGTACAGCTCGTCCATGCCCAG-3′ to introduce an upstream SacI and downstream XhoI restriction site. The PCR product was cut with both enzymes and ligated with similarly digested pCAGGS (a gift from Ronald Harty, University of Pennsylvania, Philadelphia, PA). Plasmid DNA was purified (HiSpeed Midi Kit; Qiagen) and used for transfections. HeLa K^b cells (4 × 10⁵) were resuspended in 20 μL of Amaxa Solution SF (Lonza) and mixed with 200 ng DNA. Cells were transfected using the Amaxa 96-well Shuttle System using the program CN-114. Following transfection, cells were cultured overnight and used the following day.

Antigen Presentation Assays and in Vitro Viral Infections.

MHC class I antigen presentation in EL4/SCRAP cells in the presence, absence, or following removal of Shield-1 was examined as previously reported (10). All flow cytometry experiments were conducted using a BD LSR II flow cytomter. To quantify the number of peptide–MHC complexes on the cell surface, acid-washed EL4/SCRAP cells were stained with FITC-coupled 25D-1.16 at different times postwash, and the mean fluorescence intensity (MFI) of the FITC signal was converted to molecular equivalents using FITC-coupled beads (Spherotech) as previously described (6). Cells were also stained in parallel with Alexa 647-coupled 25D-1.16, and the number of peptide–MHC complexes calculated from FITC-labeled cells was used to generate a standard curve for Alexa 647-labeled cells for comparison of Shield-1–treated and –untreated cells. For all other antigen presentation experiments, cells were stained with Alexa 647-coupled 25D-1.16 monoclonal antibody and analyzed by flow cytometry. For rVV and rVSV infections, cells were resuspended at a concentration of 2 × 10⁶ in the appropriate solution (saline solution with 0.1% BSA for rVV, serum-free MEM for rVSV) and virus was added at an MOI of 10. Cells were incubated at 37 °C for 30 min with occasional agitation, washed, and cultured at 10⁶ cells/mL in complete media. When transiently transfected HeLa K^b cells were to be infected with rVV, cells were first acid-washed as described above and recultured for 2 h before infection. In this scenario, the background levels of K^b–SIINFEKL were determined immediately following rVV infection. In transiently transfected cells, K^b–SIINFEKL levels were determined on GFP⁺ cells rather than on the entire population.

Quantitative Western Blot and Antigen Presentation Efficiency.

Quantitative Western blots were performed based on previous experiments (6). Briefly, total cell lysates were prepared from cells at various times after the indicated treatments by boiling cells in SDS sample buffer (Quality Biological) at 10⁷ cells/mL for 20 min, and then an equal volume of 0.05 M DTT (in water) was added to the lysates and boiled for an additional 10 min. Recombinant GFP (Clontech) was added to similarly prepared EL4 cell lysates (ranging in concentration from 1 to 10 nM), and both experimental samples as well as recombinant GFP were resolved by SDS/PAGE, blotted onto nitrocellulose, and analyzed by Western blot using the Odyssey Imaging System (LI-COR). Odyssey software was used to quantitate GFP signal in samples and standards. A standard curve was generated to determine the concentration of GFP in each experimental sample. This concentration was then divided by the number of cell equivalents in the sample (generally 100,000 cells) and multiplied by Avogadro’s number to determine the average number of GFP molecules present in each cell. The difference between Shield-1–treated and –untreated samples is reported. For transiently transfected cells, the cell equivalent was adjusted based on the transfection efficiency, the number of GFP⁺ cells as determined by flow cytometry. The efficiency of antigen presentation was determined by dividing the number of peptide–MHC complexes calculated above by the number of GFP molecules per cell. A two-tailed Student’s t test was used to statistically analyze the datasets using GraphPad Prism software.