Vertebrates express three cytokine-inducible proteasome subunits that are incorporated in the place of their constitutively synthesized counterparts. There is increasing evidence that the set of peptides generated by proteasomes containing these subunits (immunoproteasomes) differs from that produced by standard proteasomes. In this study, we use mice lacking one of the immunoproteasome subunits (LMP2) to show that immunoproteasomes play an important role in establishing the immunodominance hierarchy of CD8+ T cells (TCD8+) responding to seven defined determinants in influenza virus. In LMP2−/− mice, responses to the two most dominant determinants drop precipitously, whereas responses to two subdominant determinants are greatly enhanced. Adoptive transfer experiments with naive normal and transgenic TCD8+ reveal that the reduced immunogenicity of one determinant (PA224–233) can be attributed to decreased generation by antigen presenting cells (APCs), whereas the other determinant (NP366–374) is less immunogenic due to alterations in the TCD8+ repertoire, and not, as reported previously, to the decreased capacity of LMP2−/− APCs to generate the determinant. The enhanced response to one of the subdominant determinants (PB1F262–70) correlates with increased generation by LMP2−/− virus–infected cells. These findings indicate that in addition to their effects on the presentation of foreign antigens, immunoproteasomes influence TCD8+ responses by modifying the repertoire of responding TCD8+.
Proteasomes are multicatalytic enzymes complexes that are responsible for the turnover of most cellular proteins and also for the generation of the bulk of antigenic peptides transported by transporter associated with antigen presentation (TAP) and presented by MHC class I molecules 1. 26S proteasomes consists of catalytic 20S proteasomes and either the 19S or 11S regulatory complexes. The 20S proteasome is a four-ring structure with seven different subunits in each ring, arrayed as α7β7β7α7 2. The regulatory complexes provide the specificity of polypeptide recognition. They also open the gated channel formed by the outer ring of 20S proteasomes to control substrate access to the catalytic chamber 3. Substrates of 26S proteasomes are largely targeted to 19S regulatory subunits by the addition of polyubiquitin chains 4. Three of the β subunits (β1, β2, β5) of 20S proteasomes are known to possess protease activity. Exposing cells to IFN-γ induces the synthesis of alternatives for these subunits (respectively, LMP2, MECL1, LMP7) that together are incorporated into an alternative form of proteasomes known as immunoproteasomes 5.
Differences in the cleavage patterns of standard and immunoproteasomes have been demonstrated using purified 20S proteasomes and synthetic peptide substrates. Although there were several conflicts in the original reports, it now appears that in general, 20S immunoproteasomes are more adept at producing peptides with hydrophobic and positively charged COOH-terminal residues. These are precisely the types of residues preferred by class I molecules (for a review, see reference 6). More recently, these findings have been extended to the generation of defined peptides by APCs 7,8,9,10,11.
The most direct method for studying the influence of immunoproteasomes on immune responses in vivo is to compare wild-type mice to mice with targeted knockout of genes encoding the immunoproteasome subunits. Although mice lacking either LMP2 or LMP7 have been available for several years, characterization of their TCD8+ responses has been performed at a fairly rudimentary level, both in the methods used for enumeration of TCD8+ responses and the number of determinants surveyed. The effects of the targeted disruptions on the immunodominance hierarchy of TCD8+ responding to a viral infection remain to be characterized.
Immunodominance is the term given to the universal phenomenon that only a small percentage of the multitude of peptides present in a given antigen elicit TCD8+ responses. Even among the chosen few, the numbers of responding TCD8+ can vary enormously. The mechanisms underlying immunodominance have recently come under increased scrutiny because of its obvious importance for vaccine design and TCD8+-determinant prediction (for a review, see reference 12). The rekindled interest is also due to technological advances in T cell quantitation, such as the introduction of intracellular cytokine staining (ICS; reference 13) and MHC-peptide tetramers 14. This has enabled direct counting of multiple specific CTL directly ex vivo without the intervening in vitro culture that can skew the true frequencies of responding cells. With the improved technology, it is clear that greater numbers of determinants are recognized by TCD8+ than originally believed. Thus, understanding the rules of determinant hierarchy becomes an even more interesting (and difficult) problem. In this study, we have used LMP2−/− mice to examine the effects of immunoproteasomes on the immunodominance hierarchy in the TCD8+ response to influenza virus (IV) infection.
Materials And Methods
The DC cell line DC2.4 (H-2b; provided by Dr. K. Rock, University of Massachusetts Medical School, Worcester, MA; reference 15) and the thymoma cell line EL-4 were maintained in RPMI 1640 containing 10% fetal calf serum, 5 × 10−5 M 2-mercaptoethanol, antibiotics, and 2 mM glutamine (RP-10). To generate Con A or LPS blasts, 2 × 107 spleen cells were cultured with RP-10 medium in 6-well plates in the presence of 5 μg/ml of Con A or LPS for 48 h. TCD8+ were stimulated and maintained in RP-10 medium supplied with 10 U/ml recombinant human IL-2 (see below). Peritoneal exudate cells were harvested from mice 3 d after intraperitoneal injection with 1 ml thioglycollate. Dendritic cells were prepared as described 16.
mAbs and Other Reagents.
All mAbs used for flow cytometry were purchased from BD PharMingen. All were fluorescein labeled except anti-CD8α which was Cy-Chrome labeled. PE-labeled Db-NP366–374-tetramers were provided by the MHC Tetramer Core Facility of National Institute of Allergy and Infectious Diseases, National Institutes of Health. Anti-CD4 mAb culture supernatants from hybridoma GK1.5 (TIB 205; American Type Culture Collection) and unlabeled anti-B220 Ab RA3-6B2 (BD PharMingen) were used to coat M450 Dynal beads (Dynal) for depleting CD4+ cells and B220+ expressing NK cells when establishing TCD8+ lines.
Mice, Viruses, and CTL Stimulation.
C57BL/J6 (B6) mice and B6.SJL-ptprc (B6.SJL) mice, which carry a CD45.1 marker, were purchased from Taconic. B6.LMP2−/− mice were provided by L. Van Kaer (Vanderbilt University, Nashville, TN) and were maintained under barrier conditions by Taconic. For priming, 8–10-wk-old female mice were infected with 600 hemagglutination units (HAU) of IV A/Puerto Rico/8/34 by intraperitoneal injection. IVs were propagated in the allantoic cavity of 10-d embryonated chicken eggs. Splenic and peritoneal cells were generally prepared 7 d after priming unless otherwise indicated. For generation of TCD8+ lines, animals were generally used >30 d after priming. TCD8+ stimulation was always carried out in RP-10 with 10 U/ml recombinant human IL-2. In brief, 3 × 107 splenocytes were stimulated with 1/20–1/25 the number of IV-infected or peptide-pulsed APCs, irradiated with 200 Gy. Stimulated live T cells were harvested through Ficoll-Hypaque gradient and enriched for CD8+ cells by depleting B220+ and CD4+ cells using mAb-coated M-450 Dynal beads (Dynal).
Intracellular Cytokine and Tetramer Staining.
CTL activities were tested either with ICS using peptides corresponding to the known immunodominant and subdominant determinants in H-2b–restricted response to influenza or assayed with Db-NP366–374-tetramers. ICS was performed as described 17. In brief, splenic or peritoneal cells were incubated with synthetic peptides at 0.5–1 μM for 2 h at 37°C in Iscove's modified DMEM supplemented with 10% FBS. After addition of brefeldin A (BFA) to a concentration of 10 μg/ml, cells were incubated for an additional 4 h at 37°C. Cells were then incubated on ice with Cy-labeled anti-CD8a and PE-labeled anti-CD45.1 for 1 h, washed, and fixed with 1% paraformaldehyde. Cells were then incubated in the presence of 0.2% saponin (Calbiochem) with fluorescein-labeled anti–IFN-γ. Tetramer staining was performed at room temperature for 1 h in the presence of Cy-Chrome–labeled anti-CD8 (BD PharMingen). For ICS or tetramer staining, 100,000–300,000 cells were analyzed on a FACScan™ (Becton Dickinson). Total Ag-specific cell numbers were calculated using Ag-specific percentage of total CD8+ cells multiplied by the total TCD8+ percentage of total spleen or peritoneal cells acquired on the FACS®.
Infection of Cells for Antigen Presentation Kinetic Assay.
Cells were washed with Autopow (Life Technologies) supplemented with 0.1% BSA, 25 mM Hepes, and adjusted to pH 6.6, and resuspended in the same buffer containing 60 HAU IV per 106 cells and incubated for 1 h at 37°C in 96-well plates before adding TCD8+ in 200 μl of RP-10. BFA was then added at different time points to a final concentration of 10 μg/ml. 2 h after the last addition of BFA, cells were harvested and T cell activation was determined by ICS staining as described above.
Peptides and Binding Assays.
All peptides were synthesized, HPLC purified, and analyzed by mass spectrometry by or under the supervision of the Biologic Resource Branch, National Institutes of Allergy and Infectious Diseases, National Institutes of Health. All peptides were >95% purity. Peptides were dissolved in DMSO at 1 mM as stock solutions and stored at −30°C.
Naive TCD8+ Precursor Transfer.
Splenocytes were T cell enriched with anti-CD90 (Thy1.2)-coated microbeads (Miltenyi Biotec). Generally, ∼85% purity was achieved as assessed by flow cytometry. ∼1.5 × 107 purified cells were transferred into a naive host animal intravenously. After transfer, animals were immediately primed intraperitoneally with IV. 7 d later, the responses of both host and donor cell population were followed with CD45.1 Ab specific for B6.SJL cells or CD45.2 Ab specific for normal B6 cells.
Transgenic TCR TCD8+ Transfer.
B6 or LMP2−/− mice were irradiated with 800 rads 2 d before T cell transfer. On the day of transfer, mice (in groups of three) were infected by intraperitoneal injection with IV HK/X31 (X31), A/NT60/68 (NT60), or left uninfected. 5 h later, mice were given 5 × 106 transgenic T cells by intravenous injection. T cells were purified from homogenized spleen and lymph nodes (popliteal, inguinal, brachial, axillary, and superficial cervical) obtained from F5 mice. Purification entailed Ficoll gradient centrifugation followed by positive selection of Thy1.2 (CD90) positive as described above. 16 h after T cell transfer, spleens were removed, homogenized, and centrifuged in a Ficoll gradient to yield viable mononuclear cells. Cells were incubated in 20% normal mouse serum and mAb 2.4G2 (Fc block) supernatant for 20 min before staining with directly conjugated FITC-labeled Abs to CD69, CD25, or CD62L. Cells were analyzed by flow cytometry, gating on CD8+Vβ 11+ cells (the Vβ class of the F5 TCR) to ascertain the cell surface levels of these T cell activation markers.
Results And Discussion
We first compared anti-IV TCD8+ responses of wild-type B6 to that of LMP2−/− mice. After intraperitoneal infection, peritoneal exudate cells and splenocytes were assayed from 4 to 9 d after infection for responses to six of the seven defined peptides from IV gene products (see Table) that are known to be recognized by TCD8+ from IV-infected H-2b mice (18; and unpublished data). Two determinants, NP366–374 and PA224–233, have been reported to dominate local and splenic responses of B6 mice after intranasal infection with influenza. As seen in Fig. 1, the same peptides dominate the local and splenic response after intraperitoneal infection. LMP2−/− mice exhibited dramatic differences from B6 mice. Responses to NP366–374 and PA224–233 were greatly reduced whereas responses to NS2114–122 and PB1F262–70 were enhanced. This effect was particularly prominent with peritoneal TCD8+. Overall, the LMP2−/− response was less vigorous both in the percentage of responding virus-specific TCD8+, and in the overall number of responding cells. Of interest, ∼1/2 the number of total TCD8+ were recovered from infected LMP2−/− mice compared with B6 mice.
This experiment indicates that immunoproteasomes play an important role in determining the immunodominance hierarchy (summarized in Table). To distinguish the contributions to this phenomenon of TCD8+ repertoire versus antigen presentation, we examined the response of LMP2−/− mice after transfer of TCD8+ from wild-type mice. To enable discrimination of host versus donor cells we used B6.SJL mice as donors. TCD8+ from these mice express a marker (CD45.1) absent in B6 mice that can be detected by mAb staining. CD90 (Thy1.2)-enriched splenocytes from B6.SJL mice were transferred into either B6 or LMP2−/− mice that were then infected with IV. Spleen and peritoneal cells were harvested 7 d later and analyzed by ICS using the panel of seven IV peptides or by staining with PE-labeled Db-NP366–374 tetramers.
The left panel of Fig. 2 A shows the typical pattern of tetramer staining of splenocytes in obtained from IV-infected B6 mice; in the mouse examined, 7% of TCD8+ are tetramer positive. The middle panel shows that of the tetramer-positive TCD8+ in this spleen, only 2% are derived from donor cells. The right panel shows the summarized composite results of tetramer-positive cells for host- (white bars) and donor-derived (filled bars) TCD8+ in B6 and LMP2−/− mice. Remarkably, donor cells comprise 50% of the tetramer-positive response in LMP2−/− mice: more than 10-fold higher than in B6 mice. These findings were confirmed by ICS (Fig. 2B and Fig. C), which demonstrated a much greater host TCD8+ response to NP366–374 in B6 mice, and conversely a much greater (∼sixfold) donor NP366–374–specific TCD8+ response in spleens of LMP2−/− mice. Note the 20-fold difference in scale used to display the results of host and donor cells. This difference in abundance between host and donor cells can be attributed to both the number of TCD8+ used for transfer (∼1/10 the amount of TCD8+ present in mice), and also an anticipated loss in cells due to imperfect trafficking to spleen and peritoneum. The marked expansion of transferred naive NP366–374–specific TCD8+ in LMP2−/− mice indicates that deficiencies in the TCD8+ repertoire are a major factor in the limited response of LMP2−/− mice to NP366–374.
Examination of the results for host cell responses to the six other determinants tested in this experiment reveals that the typical immunodominance hierarchies of B6 and LMP2−/− host cells were basically unaltered by the cell transfer (compare Fig. 2 B with Fig. 1). In B6 mice, NP366–374 and PA224–233-specific CTL remain dominant in both peritoneal and spleen cell responses. In LMP2−/− mice, TCD8+ specific for PB1F262–70 and NS2114–122 dominate the response. Notably, the total number of responding host cells in LMP2−/− mice is much less than in B6 mice, particularly in the spleen.
In B6 mice, the response of donor cells basically mirrored the host cell response (Fig. 2 C). There were several interesting differences in the response of donor cells in LMP2−/− mice compared with B6 mice. First, there was the noticeable increase in the response of splenic NP366–374-specific cells observed with tetramers. This was somewhat less apparent in the peritoneal cavity, perhaps due to increased immunodomination from PB1F262–70-specific TCD8+. Second, with one important exception, more donor cells specific for each determinant were recovered from LMP2−/− mice than B6 mice, indicating that antigen presentation is not limiting for these determinants. This effect was most marked with PB1F262–70-specific TCD8+. Third, for the exceptional determinant, PA224–233, less specific TCD8+ were recovered from both peritoneal and splenic populations in LMP2−/− mice than from B6 mice, pointing to a defect in antigen presentation in these mice.
To confirm that presentation of NP366–374 was similar in wild-type and LMP2−/− mice, we adoptively transferred purified TCD8+ from F5 TCR transgenic mice 19. Mice were then infected intraperitoneally with NT60 IV, which expresses NP with the cognate Db-restricted NP366–374 peptide (ASNENMDAM) recognized by F5 TCD8+. As a control, mice were infected with X31, which expresses NP with the non–cross-reactive ASNENMETM peptide. Peritoneal exudate cells were harvested 16 h after infection and the activation state of F5 TCD8+ was assessed by enhanced expression of CD25 and diminished expression of CD62 (Fig. 3). Importantly, F5 cells were activated at similar levels after transfer into LMP2−/− and wild-type mice. The antigen specificity of this activation is shown by the similarity between F5 cells recovered from X31 infected mice and uninfected mice.
We next performed an adoptive transfer experiment using LMP2−/− donor splenocytes and B6.SJL recipients (Fig. 4). The response of host TCD8+ was similar to unmanipulated mice. Despite these robust responses, we failed to detect activation of the transferred cells. This finding cannot be attributed to differences in the handling or purification of the transferred TCD8+, as we observed similar results in three separate experiments. Rather, it is consistent with a general defect in the ability of LMP2−/− mice to mount an IV-specific response that is exacerbated in the presence of TCD8+ from normal mice.
Finally, we examined the capacity of LMP2−/− and wild-type APCs to present endogenous IV antigens to antigen-specific TCD8+ lines. To avoid uncertainties associated with the clonal variation of established APC cell lines, we used cell populations obtained directly from mice. APCs were incubated with BFA at various times after infection and antigen presentation was quantitated by the capacity of the cells to trigger determinant-specific TCD8+ lines as measured by ICS. In additional experiments (not shown), we found that this assay parallels presentation kinetics as measured by 51Cr release, which itself is proportional to the number of peptide-class I complexes recovered from cells 20. As seen in Fig. 5 A, Con A blasts from wild-type and LMP2−/− mice showed a similar kinetics of presenting NP366–374. By contrast, LMP2−/− blasts presented PB1F262–70 much more rapidly than B6 blasts. This provides the second published example of immunoproteasomes interfering with production of a specific determinant 7. Unfortunately, we were not able to compare the presentation of PA224–233 in this assay due to its very poor presentation in vitro even under optimal circumstances (unpublished data).
The identical presentation of NP366–374 by wild-type and knockout cells was surprising, as it was reported that LMP2−/− LPS blasts and peritoneal exudate cells exhibit defective presentation of this determinant 21. We therefore repeated this experiment using LPS blasts, peritoneal exudate cells, and dendritic cells prepared from wild-type and LMP2−/− mice (Fig. 5 B). Once again, we failed to observe a significant difference in the presentation of this determinant.
Taken together, these findings indicate that in LMP2−/− mice the: (a) decreased response to NP366–374 is based principally, if not solely, on defects in the TCD8+ repertoire; (b) decreased response to PA224–233 is due principally to defects in antigen presentation; and (c) enhanced response to PB1F262–70 is probably due at least in part to enhanced generation by standard proteasomes relative to immunoproteasomes.
Van Kaer et al. 21 previously reported that LMP2−/− mice exhibit diminished responses to NP366–374. This was attributed to a defect in antigen presentation based on the decreased abilities of LPS-activated splenocytes or peritoneal exudate cells derived from LMP2−/− mice to activate a NP366–374-specific T cell hybridoma. By contrast, we failed to observe a similar defect in presentation by either LPS- or Con A–activated splenocytes, peritoneal exudate cells, or dendritic cells. We previously demonstrated that the kinetics of presentation as determined using BFA reflects the number of peptide class I complexes recovered from cells 20, and we believe that the data in Fig. 5 demonstrate that there is no significant difference between the capacities of wild-type and LMP2−/− cells to generate NP366–374 from the PR8 NP.
As PA224–233 occupies the α-position in the B6 TCD8+ immunodominance hierarchy, its demotion in LMP2−/− mice can influence the response to all of the other determinants due to decreased immunodomination 17. Given the opportunity to assume the α-position, the plunge of NP366–374 down the immunodominance hierarchy is striking, and demonstrates the profound effect of LMP2 on the anti-NP366–374 repertoire. The loss of immunodomination exerted by these two determinants probably contributes to the ascendence of PB1F262–70- and NS2114–122-specific TCD8+ in LMP2−/− mice. The enhanced presentation of PB1F262–70 in LMP2−/− APCs probably contributes to its rise to the α-position.
Our findings demonstrate that the loss of LMP2 has a significant effect on the TCD8+ repertoire. This is not shocking given the observation that LMP2−/− mice exhibit a specific deficit in the numbers of TCD8+ relative to B6 mice. As LMP2 is expressed in the thymus 22, its absence can affect the repertoire by influencing positive and/or negative selection, as well as whatever effects it may exert in the periphery. As LMP7 and MECL1 can assemble functional proteasomes in the absence LMP2, albeit inefficiently 23,24, the characteristics of LMP2−/− mice provide a minimal estimate of the role of immunoproteasomes on antigen presentation and the TCD8+ repertoire. That deficiencies in antigen processing can influence the TCD8+ repertoire was previously established in studies of TAP−/− mice 25,26. Moreover, the effect of LMP2 on repertoire development parallels similar findings in MHC class II–mediated selection of the TCD4+ repertoire, where targeted disruptions in genes encoding a protease (cathepsin L) that contributes to antigen processing or a molecule (H-2M) that aids peptide loading of class II molecules have been shown to modify the repertoire 27,28,29.
It is interesting that the overall number of TCD8+ responding to IV was decreased in LMP2−/− mice. Potentially, this could be due to our failure to provide the proper determinants to LMP2−/− TCD8+ during restimulation for the ICS assay. This possibility is unlikely, however, as a large difference was observed in head to head comparisons of ex vivo TCD8+ activated for ICS staining by IV-infected splenocytes from autologous mice as APCs (data not shown). Rather, it appears that the absence of LMP2 decreases the ability of TCD8+ to respond to foreign antigens, even to determinants that appear to be made equally by standard proteasomes and immunoproteasomes. Consistent with this finding, TCD8+ from LMP2−/− mice failed to expand when transferred to B6 mice. Even if the lion's share of this effect is due to increased relative immunodomination by host TCD8+ (as it probably is), it still points to a diminished capacity of LMP2−/− TCD8+ to proliferate relative to normal TCD8+.
The poor responsiveness of LMP2−/− TCD8+ to IV demonstrates that LMP2, and by inference immunoproteasomes, enhance the ability of TCD8+ to respond to foreign antigens. Although this is reassuring in terms of understanding the evolution of immunoproteasomes, this comfort comes at the cost of a mechanistic conundrum: how can the absence of immunoproteasomes affect responses to foreign determinants produced equally (or more) efficiently by standard proteasomes? Immunoproteasomes may play a role in TCD8+ cell activation and proliferation. Additionally, given that the selection of germ line TCR genes in evolution occurred in the context of MHC molecules presenting a peptide repertoire heavily influenced by immunoproteasomes, it is not hard to imagine that the repertoire suffers when the selection is limited to peptides produced by standard proteasomes. With the rapid advances in sequencing technology, it should be possible to directly compare the naive TCD8+ repertoire of normal mice and those lacking immunoproteasomes.
Bethany Buschling provided outstanding technical assistance.
W. Chen received support from a C.J. Martin Fellowship (967036) from the Australian National Health and Medical Research Council. C. Norbury received support from a Wellcome Traveling Fellowship. We thank Dr. Luc Van Kaer for his generous gift of LMP2−/− mice and Dr. Kenneth Rock for DC2.4 cells.
W. Chen's present address is Cancer Vaccine Unit, Ludwig Institute for Cancer Research, Austin & Repatriation Medical Centre, Studley Rd., Heidelberg VIC 3084, Australia.