Presentation of Type B Peptide–MHC Complexes from Hen Egg White Lysozyme by TLR Ligands and Type I IFNs Independent of H2-DM Regulation

In APCs, presentation by MHC II molecules of the chemically dominant peptide from the protein hen egg white lysozyme (HEL) generates different conformational isomers of the peptide–MHC II complexes (pMHC). Type B pMHCs are formed in early endosomes from exogenous peptides in the absence of H2-DM, whereas in contrast, type A pMHC complexes are formed from HEL protein in late vesicles after editing by H2-DM. Thus, H2-DM edits off the more unstable pMHC complexes, which are not presented from HEL. In this study, we show that type B pMHC complexes were presented from HEL protein only after stimulation of dendritic cells (DC) with TLR ligands or type I IFN. Type I IFN contributed to most TLR ligand-induced type B pMHC generation, as presentation decreased in DC lacking the receptor for type I IFNs (IFNAR1−/−). In contrast, presentation of type A pMHC from HEL and from peptide was minimally affected by TLR ligands. The relative effectiveness of CD8α+ DC or CD8α− DC in presenting type B pMHC complexes varied depending on the TLR ligand used. The mechanisms of generation of type B pMHC from HEL protein with TLR stimulation did not involve H2-DM or release of peptides. DC from H2-DM–deficient mice in the presence of TLR ligands presented type B pMHC. Such DC showed a slight enhancement of HEL catabolism, but peptide release was not evident. Thus, TLR ligands and type I IFN alter the pathways of presentation by MHC II molecules of DC such that type B pMHCs are generated from protein Ag.

T he chemically dominant peptide from the model protein hen egg white lysozyme (HEL) binds to the MHC II molecule, I-A k , in multiple conformations (1)(2)(3). Both the more stable peptide-MHC II (pMHC) complexes, termed type A, and the less stable conformers, type B, bind to the same peptide, DYGILQINS, and in the same register (1)(2)(3)(4). The type B pMHC conformers result when APC interact with peptides in early endosomes but are removed by H2-DM in late endosomes (4). In contrast, the type A pMHCs are formed in late vesicles where H2-DM edits off the unstable complexes, allowing the persistence of only the type A (4). The type B-specific T cells to the HEL 52-60 are frequent (30-50% of CD4 T cells) and escape tolerance, for example, in transgenic mice expressing HEL by all APCs (5). Because T cells to type B pMHC complexes are found to autologous proteins, it is critical to understand how type B responses could be protective during infection or detrimental in the context of inflammation-associated autoimmunity (5)(6)(7). We previously reported that type B pMHC were generated from HEL protein in vivo when administered along with inflammatory stimuli such as LPS, listeriolysin O, or CFA (8). In this study, we explore the signals that regulate TLR-induced generation of type B pMHC in vitro.
Dendritic cells (DC) play a key role in initiating the adaptive immune pathway through the presentation of Ag on MHC II molecules to activate CD4 T cells. Although resting DC are able to process and present exogenous Ag on MHC II molecules, there are noticeable changes that occur following exposure to inflammatory stimulants (9)(10)(11)(12)(13)(14)(15)(16)(17). TLR signaling induces DC maturation involving the production of critical cytokines, increased migration to draining lymph nodes, and increased expression of MHC I, MHC II, and many costimulatory molecules (18)(19)(20)(21). Although many details of how TLR signaling affects the MHC II pathway are still to be considered, it is clear that TLR engagement leads to a transient increase in endocytosis, a transient increase in MHC II transcription, and stabilization of surface pMHC (10-12, 22, 23).
Stimulation through TLR3, -4, -7, and -9 leads to production of type I IFN by DC (24). TLR2 also initiates type I IFN production (25,26). The contribution of type I IFN signaling to TLR-initiated responses has been explored primarily in vivo and shown to be important for events ranging from DC maturation to Ab production by B cells to bystander activation of T cells and NK cells (27)(28)(29)(30)(31)(32)(33). In the MHC I pathway, type I IFN has been shown to be important for TLR-induced cross priming (13,33,34) and specifically for CpG-induced cross presentation of proteins like soluble OVA (35). However, the contribution of type I IFN to TLR-initiated events in the MHC II pathway has not been explored. In this study, we show that most TLR ligands and type I IFN-induced presentation of type B pMHC by DCs from HEL and that TLR-induced type I IFN contributed to induced presentation of type B pMHC.

DC isolation and Flt3 ligand treatment
DC were isolated from spleens of mice injected with 10 mg Flt3 ligand i.p. for 3 consecutive d. On day 8, spleens were harvested and digested with 0.14 U/ml Liberase Blendzyme 3 or 1.67 U/ml Liberase TL (Roche Applied Science, Indianapolis, IN) to make single-cell suspensions from which DC were isolated by CD11c magnetic beads (Miltenyi Biotec, Auburn, CA). Enriched DC were $95% pure as determined by flow cytometry. For sorting, DC isolated as above were stained for surface markers and resuspended in phenol-free DMEM (Invitrogen, Carlsbad, CA) supplemented with 2% FCS and 2 mM EDTA. Cells were sorted on a BD FACSAria II (BD Biosciences). Conventional DC (cDC) were sorted as CD11c high CD45RA 2 B220 2 or CD11c high Siglec H 2 B220 2 . cDC were further separated by CD8a expression. For analysis of H2-DM and H2-DO, cDC were further selected as CD19 2 .

Assays
For Ag presentation assays, 10 5 DC were incubated with HEL protein or the HEL (48-62) peptide (DGSTDYGILQINSRW) with or without stimulants for 18 h in 100 ml volume in V-bottom 96-well plates. Stimulant concentrations used were: 10 mg/ml zymosan A, 10 mg/ml poly (I:C), 1 mg/ml LPS, 1 mg/ml gardiquimod, 6 mg/ml resiquimod, 1 mM ODN 1826 (CpG-B), 100 U/ml IFN-g, 10 U/ml IFN-a, and 10 U/ml IFN-b. DC were washed three times with serum-free DMEM after incubation before adding 5 3 10 4 T cell hybridomas/well in a 200-ml volume; 24 h later, the release of IL-2 was measured in the culture fluid using CTLL as an indicator cell. Most experiments used two previously characterized T cell hybridomas: 11A10, which only reacts with type B pMHC, and 3A9, which recognizes type A pMHC. Sorted DC were pretreated with 1 mg/well anti-IFNAR1 mAb MAR1-5A3 [a kind gift of Dr. Robert Schreiber (37)], or control antihuman IFN-g receptor-1 GIR.208 Ab for 1 h at 37˚C. Without washing, HEL or peptide was added with or without stimulants to the wells. The remaining assay followed the protocol described above. For examination of surface markers by flow cytometry, DC isolated as described above were incubated with or without stimulants for 18 h in 100 ml volume in 96-well V-bottom plates.
Peptide release assay DC from CB.17 mice were incubated overnight with HEL with or without 1 mM CpG-B. Supernatants were collected and spun twice to remove cells and debris. Supernatants were serially diluted and added to paraformaldehyde-fixed DC from B10.BR mice. DC were fixed in 1% paraformaldehyde for 5 min, incubated with 0.2 M DL-Lysine, and washed extensively in media. T cells were added as described above.

TLR ligands induce type B presentation from HEL
To assess the ability of TLR ligands (Table I) to affect type B pMHC presentation, a culture system was set up involving the culture of DC with HEL in the presence of the stimulants for 18 h, after which the stimuli were removed, and the extent of presentation was assayed by adding the indicator T cell hybridomas. In agreement with the basic definition of type B presentation, unstimulated DC presented type B pMHC only when incubated with the HEL:48-61 peptide, but not HEL (Fig. 1A, 1C). However, all TLR ligands except flagellin ( Fig. 2A) allowed for the presentation of type B pMHC conformers from HEL, referred to as induced type B presentation. Notably, TLR ligands had little effect on presentation from peptide (Fig. 1C, 1D), only affecting type B presentation from peptide at low levels (0.01 mM), and modestly augmented presentation to 3A9, a T cell that recognizes type A pMHC, only at low levels of HEL (0.01 mM) (Fig. 1B). IFN-g did not induce type B presentation from HEL (Fig. 1A). The induced type B presentation by TLR ligands was reproducible, testing three different type B T cell hybridomas recognizing the HEL 48-61 epitope (data not shown.) Induced type B pMHC presentation by MyD88-deficient DC followed the expected pattern, with TLR7 and -9 ligands being Table I. TLR ligands used in this study   TLR2/6  TLR3  TLR4  TLR5  TLR7  TLR9 Agonist(s) Zymosan A (also binds dectin-1) completely eliminated, TLR3 ligands being minimally affected, and TLR4 ligands only partially reduced, reflecting the degree to which each TLR depends on MyD88 as a signaling adapter (Fig. 2) (38).

Role of DC subsets in induced presentation
Whether a subset of DC was responsible for the presentation of type B epitopes from HEL was examined. Plasmacytoid DC comprised ∼10% of the total DC population (identified as CD11c int/low Siglec H + B220 + ), but were not essential for induced presentation as sorted cDC, identified as CD11c hi Siglec H 2 B220 2 presented type B pMHC from HEL with TLR stimulation (data not shown.) To examine the role of the cDC subsets, sorted CD8a + and CD8a 2 DC were incubated with HEL protein with or without CpG as in previous experiments. CD8a + DC strongly presented type B pMHC from HEL upon CpG stimulation (Fig. 3B), but not in the absence of stimuli. CD8a 2 DC, however, had some degree of presentation of type B pMHC in the absence of stimulation at high HEL doses: it increased upon CpG stimulation, but not as strongly  as the CD8a + DC (Fig. 3C). The ability of CD8a + DC or CD8a 2 DC to present type B pMHC from HEL slightly differed, depending on the TLR ligands, except for gardiquimod, a ligand of TLR7/8 (Fig. 3D). Gardiquimod induced a small level of presentation only by CD8a 2 DC, which is not surprising as TLR7 is minimally expressed in CD8a + DC (39). There was no significant difference in presentation of HEL peptide by the DCs (Fig. 3E). Of note is that the degree of presentation of HEL at high doses by the unstimulated CD8a 2 DC varied greatly from no response whatsoever to the small response found in Fig. 3B, reflecting most likely a degree of activation by environmental stimuli. In sum, these data indicate that both splenic DC subsets are capable of presenting type B pMHC from protein HEL in response to TLR stimulation.

Role of type I IFN in induced type B presentation
In addition to TLR ligands, rIFN-a and IFN-b induced type B pMHC presentation from HEL (Fig. 4A), but only affected type A presentation marginally at low levels of protein (Fig. 4B). Type I IFNs had minimal effects on presentation from peptide, as with TLR ligands, only affecting type B pMHC presentation at low peptide doses (0.01 mM) (Fig. 4C, 4D).
qRT-PCR was used to assess expression of IFN-b by sorted cDC. Stimulation of cDC with CpG, poly (I:C), LPS, gardiquimod, and zymosan induced expression of IFN-b as indicated by increased mRNA levels after 2 h, decreasing by 6 h after stimulation (Fig. 4E).
To test the role of TLR-induced type I IFN on presentation, MAR1-5A3 Ab was used to block signaling through the type I IFNa/ b receptor (37). DC were pretreated with control or MAR1-5A3 Ab before adding HEL with or without stimulants. MAR1-5A3 completely blocked IFN-b-induced type B presentation by CD8a + DC (99.7% at half-maximum, Fig. 5A). It reduced CpG-induced presentation by 57% (at half-maximum) by CD8a + DC. Blocking the IFNa/b receptor had little effect on presentation from peptide (data not shown) and followed a similar trend with CD8a 2 DC (Fig. 5B). Of note, MAR1-5A3 treatment reduced the level of type B presentation by unstimulated CD8a 2 DC, indicating that cytokine production may explain why CD8a 2 DC were able to present low levels of type B pMHC from protein in the absence of stimulation.
To more critically address the role of TLR-stimulated type I IFN in induced presentation, DC from IFNAR1-deficient mice were examined. Although rIFN-b and IFN-a had no effect on presentation by IFNAR1-deficient DC (Fig. 5C, 5D, Supplemental   1A), CpG-induced presentation was decreased by ∼60% (at half-maximum) in IFNAR1-deficient DC compared with wildtype DC (Fig. 5C, Supplemental Fig. 1A). The role of TLRinduced type I IFN was explored with the remaining TLR ligands shown to induce type B pMHC presentation from protein HEL using IFNAR1-deficient DC. poly (I:C)-and zymosan Ainduced type B presentation were reduced by 25 and 36%, respectively, in IFNAR1-deficient DC (Fig. 5D, Supplemental Fig.  1A). Gardiquimod-induced presentation was completely inhibited, and LPS-induced presentation was unaffected.
These data show that type I IFN signaling contributed differently to each TLR pathway of induced type B pMHC presentation, augmented presentation through TLR3 and TLR9, was the sole mediator through TLR7, and had no effect on induction through TLR4. Of note, presentation from peptide and type A pMHC presentation from protein was unaltered in IFNAR1-deficient DC (Supplemental Fig. 1B, 1C). Together, these results suggest that TLR and type I IFN signaling most likely initiate the same subcellular events and that TLR-induced type I IFN can amplify the signal, allowing for augmented type B pMHC presentation.

Role of type I IFN on expression of costimulatory molecules
Because costimulatory molecule expression is a component of DC maturation, the contribution of type I IFN to expression of co-stimulatory molecules was examined to see if it mirrored the contribution to induced type B pMHC presentation. DC from B10. BR wild-type or IFNAR1 2/2 mice were analyzed for surface expression of MHC II and costimulatory molecules after 18 h in culture with or without stimulation. Although MHC II levels changed little after stimulation, CD40, CD80, and CD86 were all significantly increased on B10.BR DC after exposure to TLR ligands and IFN-b (Fig. 6, Supplemental Fig. 2). Costimulatory molecule expression after stimulation was lower on IFNAR1 2/2 DC than B10.BR DC for all stimulants tested, indicating that type I IFN signaling augmented upregulation of costimulatory molecule expression by all TLR ligands tested. As expected, increased surface expression of costimulatory molecules by IFN-b was completely dependent on type I IFN signaling, as IFNAR1 2/2 DC were unable to respond (Fig. 6F, 6G). Zymosan A-stimulated costimulatory molecule expression was minimally dependent on type I IFN signaling (Fig. 6A). poly (I:C)-and CpG-stimulated costimulatory molecule expression was more significantly dependent on type I IFN signaling (Fig. 6B, 6E). Gardiquimodstimulated costimulatory molecule expression was almost completely dependent on type I IFN signaling, mirroring the complete requirement for type I IFN signaling for induced type B pMHC presentation (Figs. 5D, 6D). The stimulation of costimulatory molecule expression by LPS was dependent partially on type I IFN signaling (Fig. 6C), which was not the case for induced type B presentation (Fig. 5D, Supplemental Fig. 1A).

Role of H2-DM and HEL catabolism in induced presentation
Because H2-DM prevents type B pMHC from being generated (4), regulation of H2-DM was considered as an explanation for the effects of TLR ligands. The function of H2-DM can be modulated by H2-DO, so it was necessary to examine expression of both molecules. To determine if expression of H2-DM or H2-DO was altered by TLR ligand stimulation, sorted cDC were incubated for 2, 6, or 18 h with or without CpG-B, LPS, or IFN-b stimulation, and H2-DM, H2-DO, and H2-Aa (MHC II) were examined by qRT-PCR. Their expression increased 2 h after stimulation, dropping by 6 and 18 h (Fig. 7A). Protein levels of H2-DM and H2-DO were examined by intracellular staining after 18 h of stimulation. CpG-B did not significantly alter levels of H2-DM protein, but did modestly increase expression of H2-DO (Fig. 7B). LPS has previously been shown to have a similar effect on splenic DC, having no detectable effect on H2-DM protein levels; however, they observed a modest decrease in H2-DO protein levels (40).
To examine whether regulation of H2-DM contributed to induced type B pMHC presentation, cDC from B10.BR wild-type or H2-DM-deficient mice were incubated with HEL with or without stimulants as previously described. Consistent with previous reports (36), presentation of the type A pMHC was decreased by ∼10-fold in the H2-DM-deficient cDC, an indication that H2-DM is important for the assembly of HEL pMHC complexes. Importantly, H2-DM-deficient cDC presented type B pMHC from HEL protein in response to CpG and LPS, albeit at reduced levels compared with wild-type cDC. These data indicate that although regulation of H2-DM may have some contribution to TLR and type I IFN-induced type B presentation from HEL, it is not the key mechanism controlling the event.
Finally, we examined if CpG-B changed the rate of catabolism of [ 125 I]-HEL. DC were incubated for 18 h alone or with CpG-B, then incubated with [ 125 I]-HEL for 2 h at room temperature followed by 1-, 2-, 3-, or 4-h chases in media free of HEL. TCAsoluble and -precipitable fractions were collected from intracellular (Fig. 7G) and supernatant (Fig. 7H) fractions at each time point. CpG-B-activated DC showed slight acceleration of HEL catabolism compared with resting DC. We did not find evidence that CpG-B-treated DC released peptides that would contribute to the type B presentation. Supplemental Fig. 3 describes the experiment.

Discussion
The present report adds not only to the understanding of presentation of type B pMHC conformers of HEL, but also to the general effects of TLR ligand and type I IFN on the MHC II processing and presentation pathway. In the current study, exposure of DC to TLR ligands and type I IFN changed HEL protein handling, allowing strong processing and presentation of type B pMHC complexes. Notably, in the absence of TLR or type I IFN stimulation, as reported before, DC did not present type B pMHC complexes from HEL unless at very high concentrations. Both major DC subsets participated in induced type B pMHC presentation of HEL, although the degree to which each participated varied depending on the inflammatory signal.
Type I IFN signaling was important for type B pMHC presentation initiated by CpG, poly (I:C), gardiquimod, and zymosan A, but not for LPS-initiated events. Type I IFN has been argued to be important for both cross-priming of CD8 T cells and direct priming of CD8 T cells (13,34,41) and for cross-presentation induced by CpG (35). In this study, with MHC II, we found a strong effect not on presentation of the conventional type A epitopes, but rather on the type B pMHCs. Induction of CD40 by LPS, poly (I:C), and CpG on GM-CSF bone marrow-derived DC has been shown to be at least partially dependent on type I IFN signaling (28)(29)(30)(31). Similarly, the increase in CD40, CD80, and CD86 expression after stimulation with zymosan A, poly (I:C), LPS, and CpG was partially dependent on type I IFN signaling. Gardiquimod-induced expression of these costimulatory molecules absolutely required type I IFN signaling. In our experiments, we used T cell hybridomas that do not require costimulation, allowing examination exclusively of changes in levels of pMHC.
Previous studies showed effects of TLR ligands on the generation of class II pMHC (15)(16)(17) as well as on the modulation of costimulatory molecules (28)(29)(30)(31). These findings have varied in the kind of assays the APC used for examination and the amounts and forms of ligands, so a general consensus has not emerged on the key mechanisms of action. An issue thus is to separate processing and generation of pMHC from other components of the presentation pathway. We and others (9, 10) have shown subsequently, comparing results with T cell hybridomas and primary T cells, that both resting and TLR-activated DC were capable of processing and presenting Ag about equally.
Although this report does not identify the way by which the type B pMHC epitopes were presented, some issues are pointing to possible mechanisms of action. The initial data indicate that H2-DM, a logical molecule to examine, may not be key in the effects seen in this study with TLR ligands. Pointedly, H2-DMdeficient DC presented type B pMHC from HEL protein after TLR ligands. Although TLR stimulation regulated expression of H2-DO, which controls H2-DM activity, modulation of H2-DM function was ultimately not the key mechanism of type B presentation, as shown with the H2-DM gene knockout DC. We have ruled out the release of peptides from the treated APC, whereas the effects on catabolism by CpG were modest. The observation that TLR and type I IFN stimulation minimally increased type B pMHC presentation from peptide at low doses could be due to increased total MHC II expression and/or surface pMHC t 1/2 . However, these effects were modest, whereas the effects on type B pMHC presentation from protein were robust; it is unlikely that they are the sole mechanism of induced type B pMHC presentation. We posit that presentation may be explained by changes in the dynamics of vesicular traffic, leading to a flow of peptides from late vesicles into endosomes lacking H2-DM. Such a mechanism of action is now the subject of ongoing examination. This could happen through changes in acidification of, or recruitment of proteases to, an early or recycling compartment. Interestingly,  when bone marrow-derived DC were treated with CpG prior to Ag exposure, MHC II processing and presentation occurred in early and recycling compartments (14), a finding compatible with our hypothesis on how TLR ligands could promote presentation of type B pMHC of HEL.
A longstanding issue has been whether T cells recognizing type B epitopes from self-Ags could be involved in autoimmunity. This study indicates that TLR-and type I IFN-activated DC present type B pMHC from soluble HEL. It will be critical to determine whether type B pMHC from self-proteins are presented by TLR-and/or type I IFN-activated DC, allowing priming of autoreactive T cells in vivo. We have previously shown that naturally arising insulinreactive T cells against type B pMHC can transfer diabetes in the NOD model (7).