Ectopic Expression of HLA-DO in Mouse Dendritic Cells Diminishes MHC Class II Antigen Presentation¹

The recognition of pathogen-derived peptide-MHC class II complexes by CD4 T cells is a fundamental mechanism by which the immune system responds to infection. Newly synthesized MHC class II αβ dimers associate in the endoplasmic reticulum (ER)⁴ with the invariant chain and transport as αβI complexes through the Golgi to late endosomal compartments where the invariant chain is degraded by resident proteases until only small fragments of the invariant chain, CLIP, remain bound in the class II binding groove (1, 2, 3). The release of CLIP and subsequent peptide loading of class II molecules is directly catalyzed by HLA-DM (DM) (H-2M in mice), a class II-like endosomal resident molecule that also functions as a class II-specific chaperone and peptide editor (4). Once generated, class II-peptide complexes are transported to the cell surface for subsequent presentation to and activation of CD4 T cells.

DO (H-2O in mice) is another class II-like molecule that is expressed in B cells and thymic epithelia, but not in other APCs (5). Association of the DO αβ heterodimer with DM in the ER is required for DO transport to endosomal compartments, in which DO/DM complexes accumulate (6). This suggests that DO plays a unique role in the class II processing pathway specifically in B cells, either by regulating or altering DM function. Initial studies showed that DO inhibited DM-mediated peptide loading, resulting in a down-modulation of the class II processing pathway (7, 8). However, recent studies have suggested that the function of DO may be more multifaceted (9, 10). Depending on both the experimental system and the Ag studied, it has been shown that DO can inhibit (7, 11, 12, 13, 14, 15), promote (9, 10, 16), or have no effect (9, 10, 12) on class II-peptide loading. Collectively, these studies suggest that DO modifies the loading of specific peptides on to class II molecules, which results in a change in the class II-peptide repertoire (13). However, the biological relevance of a cell-specific modulator of DM function has proven to be enigmatic and it remains to be determined whether modulation of DM activity is the only, or even the primary, function of DO.

Given the conflicting results concerning the effect of DO on class II Ag presentation, we set up a system to independently evaluate the role of DO in vivo and to better define the function of DO in the class II processing pathway. Dendritic cells (DCs) do not express DO (5), yet express the components required for class II Ag processing and presentation, including class II, H-2M, and the invariant chain (17, 18, 19). Therefore, we ectopically expressed DO in mouse DCs by transgenesis and examined the class II processing pathway in DCs in the presence and absence of DO expression. Our results showed that DO expression resulted in increased cell surface expression of A^b-CLIP and reduced class II-peptide levels, further supporting that DO inhibits DM/H-2M function. Analysis of the effect of DO expression on presentation of specific antigenic epitopes showed that DO significantly inhibited the presentation of some, but not all, exogenously supplied Ags, but had no affect on the presentation of endogenously expressed Ags. Counterintuitively, we found that DO expression in DCs resulted in increased H-2M protein accumulation. Consistent with a role for DO as a modulator of DM levels, we also found that B cells from H-2O-deficient mice had decreased levels of H-2M.

C57BL/6, B10.BR, and AND TCR transgenic (Tg) H-2Ma^−/− (all from The Jackson Laboratory, Bar Harbor, ME), H-2Oa^−/− (from L. Karlsson, Johnson and Johnson Pharmaceutical Research and Development, San Diego, CA), D10 TCR Tg (20), and DO Tg mice were bred and maintained under specific pathogen-free conditions in Memorial Sloan-Kettering Cancer Center (MSKCC) animal facilities. Use of animals was in accordance with MSKCC Institutional Animal Care and Use Committee guidelines. The human HLA-DOA and HLA-DOB genes were placed under control of the human CD11c promoter that is expressed primarily in DCs (21). Tg mice were generated by coinjection of the CD11c-DOA and CD11c-DOB constructs by the MSKCC Transgenic Core Facility. Human DO was used to take advantage of Ab reagents we had previously generated to the DO heterodimer (14) and to the DOβ cytoplasmic tail (7).

Pala and T2 cells were maintained in Iscove’s DMEM supplemented with 5% FCS (HyClone Laboratories, Logan, UT) as described (22). CTLL-2 and T hybridoma cell lines were maintained in RPMI 1640 supplemented with 5% FBS, 10 mM HEPES (pH 7.2), 50 μM 2-ME, and 10 μg/ml gentamicin sulfate as complete medium (CM) at 37°C and 5% CO₂. CTLL-2 cell cultures were also supplemented with 5 U/ml IL-2.

Anti-mouse Abs used for FACS analyses and purchased from BD Pharmingen (San Diego, CA) were as follows: PE-conjugated anti-CD11c (HL3); FITC-conjugated anti-A^kβ (10-3.6), anti-A^kα (11-5.2), anti-E^k (14-4-4S), anti-A^bβ (AF6-120.1), anti-H-2K^b (AF6-88.5), and anti-H-2K^k (36-7-5); biotinylated anti-A^bβ (25-9-17), anti-CD86 (GL1) and anti-A^b (KH74); PerCP-conjugated anti-CD45R (B220, RA3682); and purified anti-rat IgG (R3-34, isotype control) and anti-H-2M (2E5A). The anti-rat IgG isotype control and anti-H-2M mAbs were coupled with Alexa Fluor 488 (Molecular Probes, Eugene, OR) according to the supplied protocol. Biotinylated mAbs were detected with Alexa Fluor 633-conjugated streptavidin (Molecular Probes).

The hybridoma cell lines, 15G4 (anti-A^b-CLIP) and C4H3 (anti-A^k hen egg lysozyme (HEL)_46–61) (23), were provided by S. Rudensky (University of Washington, Seattle, WA) and J. Drake (Albany Medical College, Albany, NY), respectively. Monoclonal Abs Mags.DO5 (14), MaP.DM1 (isotype control) (24), Y-Ae (anti-A^b-Eα_52–68) (25), 15G4, and C4H3 were purified from bioreactor supernatants using standard protein G-Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ) or protein A-Sepharose (Sigma-Aldrich, St. Louis, MO) affinity chromatography and used as purified mAbs or conjugated with Alexa Fluor 647 (Molecular Probes) according to the supplied protocol.

The mouse mAb YoDMA.1 was generated following the immunization of mice with affinity-purified, denatured DM/DO complexes (7) emulsified in Titer Max (CytRx, Norcross, GA). Fused splenocytes from an immune-reactive animal were screened by immunoblotting using DM-positive and DM-negative human B cell lines. One hybridoma secreted a mAb that recognized DMα. This clone (YoDMA.1, IgG1) specifically recognizes denatured DMα and also cross-reacts with H-2Mα.

DCs were derived in vitro from H-2Ma^−/−, nontransgenic (nTg) and DO Tg mice by culturing bone marrow cells for 7 day with CM supplemented with 25 ng/ml GM-CSF as described (26, 27). GM-CSF was produced at MSKCC Monoclonal Antibody Core Facility in bioreactors from P885L cells transfected with mouse GM-CSF cDNA (provided by C. A. Janeway, Jr., deceased, Yale University School of Medicine, New Haven, CT). Where indicated, bmDCs were harvested on day 6 and treated with 5 μg/ml LPS (Escherichia coli type 0111.B4; Sigma-Aldrich) overnight to generate mature bmDCs.

Fluorescence-based quantitative immunoblot analyses were performed, as previously described (14) with the following Abs: rabbit anti-DOβ cytoplasmic tail (R.DOB/c) (14), YoDMA.1, 15G4, KH74 (BD Pharmingen), rabbit anti-A^bα cytoplasmic tail (a gift of S. Rudensky), anti-A^kβ (10.2.16), and polyclonal rabbit sera specific for the N terminus of calnexin (Stressgen Biotechnologies, Victoria, BC, Canada). Primary Abs were detected with alkaline phosphatase-conjugated goat anti-mouse IgG and mouse anti-rabbit IgG secondary Abs (Jackson ImmunoResearch Laboratories, West Grove, PA). Fluorescence (Vistra ECF; Amersham Pharmacia Biotech) was quantitated with a Molecular Imager FX System and QuantityOne software (Bio-Rad, Richmond, CA). Cell numbers or protein concentrations used for each blot are indicated in each figure. Protein concentrations were determined using BCA Protein Assay (Pierce, Rockford, IL).

Cells were fixed and prepared for indirect immunofluorescence as previously described (24). Primary Abs used were rabbit anti-DM serum, which also cross-reacts with H-2M (provided by H. Zweerink, Merck Research Laboratories, Rahway, NJ), rabbit anti-A^b-α cytoplasmic tail and Mags.DO5 (14). Primary Abs were detected with Alexa Fluor 488-conjugated goat anti-rabbit IgG and Alexa Fluor 594-conjugated goat anti-mouse IgG secondary Abs (Molecular Probes). Cells were visualized using an Axiophot 2 fluorescence microscope (Zeiss, Oberkochen, Germany). Digital images were acquired with a charge-coupled device camera (Hamamatsu Photonics, Hamamatsu-City, Japan), deconvoluted using Openlab software (Improvision, Coventry, U.K.), and processed in Adobe Photoshop (Adobe Systems, Mountain View, CA).

Bone marrow-derived DCs from day 6 cultures or splenocytes from wild-type or H-20a^−/− mice were stained with PE-conjugated Abs specific for CD11c or B220 (BD Pharmingen), respectively, and purified by MACS (Miltenyi Biotec, Auburn, CA) using anti-PE-conjugated microbeads according to the supplied protocol. Purified cells were >95% pure as determined by FACS analysis.

Bone marrow-derived DCs were extracted in 20 mM Tris, 130 mM NaCl, pH 8.0 (TBS) containing 1% Triton X-100 and 0.5 mM PMSF at 1 × 10⁶ cells/ml. Following the removal of nuclear material by centrifugation, lysates were precleared for 1 h at 4°C with 10% zysorbin (Zymed Laboratories, San Francisco, CA), 2 μl normal rabbit serum (Jackson ImmunoResearch Laboratories), and 50 μl of 50% (v/v) slurry of the nonspecific mAb OX68 beads. Precleared lysates were split into two equal aliquots and immunoprecipitated sequentially three times with 50 μl of Mags.DO5 beads or negative control MaP.DM1 beads at 4°C for 3 h for the first 2 and 12 h for the final precipitation. Proteins remaining in the supernatants of the final immunoprecipitation were precipitated by the addition of TCA (10% final concentration; Sigma-Aldrich) for 30 min on ice and pelleted by centrifugation. Precipitated proteins were washed twice with acetone and dissolved in nonreducing Laemmli sample buffer, and 2 M Tris (Sigma-Aldrich) was added until sample returned to basic pH. Two equal aliquots of each pellet were separated by SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes and analyzed for remaining H-2Mα and DOβ protein by immunoblotting with Yo DMA.1 and R.DOb/c, respectively. For generation of mAb-conjugated beads, purified OX68, Mags.DO5, and MaP.DM1 mAb were coupled to cyanogen bromide-activated Biogel A15m resin (Bio-Rad) at 1 mg/ml bed volume, as described (28).

Samples were stained for FACS analyses as previously described (29) and analyzed using a BD 3-laser LSR flow cytometer (BD Biosciences). Dead cells were excluded from analysis by the addition of the cell vital dye 4′,6′-diamidino-2-phenylindole (DAPI) (30). For intracellular staining, samples were incubated with Abs specific for surface proteins, fixed, and permeabilized with Cytofix/Cytoperm (BD Pharmingen) and stained according to the supplied protocol. FACS analysis of splenic DCs was performed following digestion of spleens in 400 U/ml collagenase D (Roche Applied Science, Indianapolis, IN) for 30 min at 37°C.

Day 6 bmDCs (1 × 10⁶ cells/ml) were plated in 10 cm tissue culture dishes and 1 mg/ml HEL (Sigma-Aldrich) in PBS or an equivalent volume of PBS was added to each plate. Cells were harvested 24 h later, stained, and analyzed by FACS.

T hybridoma cell lines specific for A^b complexed with peptides from IgM_377–392 (77.1), β₂-microglobulin_48–58 (4.1), and actin_163–177 (15.10) (31, 32) were provided by A. Rudensky (University of Washington, Seattle, WA). The T hybridoma cell line specific for A^b bound to OVA_258–276 (C8) was provided by C. Doyle (Duke University Medical Center, Durham, NC) and the 3A9 T hybridoma cell line (33) specific for A^k complexed with HEL_46–61 was provided by P. Cresswell (Yale University, New Haven, CT). Naive T cells specific for conalbumin (CA) and pigeon cytochrome c (pCC) were purified by negative selection from D10 (20) and AND TCR Tg mice (34), respectively, as described (20).

For exogenous Ag presentation assays, purified A^b (IgM and OVA) or A^k/E^k (HEL, CA, and pCC) bmDCs (1 × 10³/well in 96-well round-bottom plates) in CM were pulsed with increasing concentrations of Ag, as indicated, for 2 h at 37°C. Ag-loaded bmDCs were washed at room temperature in CM to remove unbound Ag and 5 × 10⁴/well T hybridomas or naive T cells were added. IL-2 production by T hybridoma cell lines was determined 16 h later by CTLL-2 bioassay as described (35), and proliferation of naive T cells after 48 h was measured by the addition of 1 μCi/well of [³H]thymidine. Results are expressed as mean cpm (± SEM) of triplicate cultures.

For endogenous Ag presentation assays, 5 × 10⁴/well T hybridomas (actin and β₂-microglobulin) or naive T cells from D10 TCR Tg mice were added to titrated (3-fold dilutions) numbers of purified A^b-expressing bmDCs (0–2500 cells/well) for 16 h and 48 h, respectively. IL-2 production by T hybridomas and proliferation of naive T cells was determined as described above.

To evaluate the biological consequence of DO expression in DCs, we generated Tg mice, in which the human HLA-DOA and HLA-DOB genes were expressed under the control of the DC-specific CD11c promoter (21). Because the mouse and human proteins involved in the class II Ag processing pathway are, for the most part, functionally interchangeable (1, 5), and human DO and mouse H-2O are ∼75% identical and ∼85% similar at the amino acid level, we expressed the human protein to take advantage of Abs specific for the human DO heterodimer. Thus, for the studies presented here, we have made the assumption that DO and H-2O are functionally interchangeable.

Quantitative Western blot analyses of bmDC lysates from two different Tg-positive mouse lines and from a Tg-negative (nTg) littermate control line showed that DO was expressed and accumulated in mouse DCs (Fig. 1,A, lower band). In Fig. 1,A, upper band, the blot resulted from background binding of the DOβ Ab and was used as a loading control. The bmDCs from one Tg line (TgHi) expressed ∼2- to 3-fold more DOβ than did the other Tg line (TgMed) (Fig. 1, B and C). Immunofluorescence analysis of Tg and nTg bmDCs showed that DO colocalized with mouse H-2M, indicating that DO trafficked to endosomal compartments (Fig. 1,D and data not shown). Additionally, DO colocalized with a subset of intracellular class II molecules (Fig. 1 E). In the absence of DM/H-2M, DO/H-2O does not transport out of the ER and is rapidly degraded (6). Indeed, DO failed to transport from the ER when the DO transgene was crossed into the H-2M-deficient background (data not shown). Collectively, these data confirmed that, as expected, human DO functionally interacted with H-2M and localized properly in mouse bmDCs.

Because DO stability and trafficking requires DM/H-2M (6), we next examined whether DO expression altered the steady-state levels of H-2M. The relative levels of H-2Mα in nTg and Tg bmDC lysates were determined by quantitative Western blotting (Fig. 2, A and B). Surprisingly, H2-Mα levels in TgMed and TgHi bmDCs were ∼2-fold and ∼3-fold higher, respectively, compared with DO-negative nTg bmDCs (Fig. 2,C). To rule out the possibility that increased H-2Mα levels in DO-positive bmDCs were an artifact of ectopic DO expression in DCs, we also analyzed H-2Mα levels in purified B cell lysates from H-2O sufficient (H-2O^+/+) and deficient (H-2O^−/−) mice (Fig. 2,D). As seen in the Tg DCs, H-2M levels were higher (∼2-fold) in B cells that expressed H-2O (Fig. 2 E). We have been unable to analyze H-2Mβ levels in Tg bmDCs and B cells, due to lack of Ab reagents. However, it is likely that H-2Mβ levels would also be increased because the expression of one chain of the H-2M heterodimer in the absence of the other leads to its rapid degradation (36).

The observed increased levels of H-2Mα in DO-positive bmDCs suggested that H-2M/DO complexes were accumulating in Tg bmDCs. Therefore, we determined whether all of the H-2M in the Tg DCs were complexed with DO. H-2M/DO complexes were either depleted or mock-depleted from bmDC lysates by immunoprecipitation with a DO-specific mAb or a control mAb, respectively. The amount of H-2Mα remaining in the supernatant was determined by quantitative Western blotting (Fig. 3,A). DO depletion and the amount of H-2Mα remaining after DO depletion relative to mock depletion was determined by quantitative Western blotting (Fig. 3 B). Results showed that both TgMed and TgHi bmDCs had substantial pools of free H-2M (∼90% and ∼40%, respectively). In human primary B cells and B cell lines, ∼50–70% of cellular DM is complexed with DO (5, 15, 37). Thus, the fraction of H-2M molecules that are DO-associated in DO Tg DCs is similar to that in human B cells.

DCs modulate the class II Ag processing pathway during development in vitro. Immature DCs retain class II molecules intracellularly in endosomal compartments, whereas mature DCs redistribute class II to the cell surface (17, 18, 19, 38, 39, 40). To determine whether ectopic DO expression altered DC maturation, Tg and nTg immature (day 4) and mature (day 7 with LPS) bmDCs were analyzed by FACS for CD86 (B7.2) and MHC class I cell surface levels. The surface density of the costimulatory molecule B7.2, which is up-regulated upon DC maturation (18, 38) was unaltered by DO expression in immature and mature bmDCs (Fig. 4,A). Additionally, the MHC class I molecules, I-K^b and I-K^k, were equally expressed on the surface of DO Tg and nTg immature and mature DCs (Fig. 4,A). Thus, Tg DO expression did not alter DC maturation. DO expression levels were also confirmed by intracellular staining and FACS analysis (Fig. 4 B).

In human cell lines and primary B cells, DO expression levels directly correlate with increased cell surface class II-CLIP (7, 8, 13, 14, 15). However, in mice, the surface expression of A^b-CLIP is not altered in either H-2O-deficient B cells (12) or in primary B cells engineered to overexpress H-2O (11). To address this in mouse DCs, we measured the surface density of A^b-CLIP on Tg and nTg bmDCs. Results showed that immature bmDCs from TgHi mice had significantly higher levels of surface A^b-CLIP than did TgMed bmDCs, which had modestly higher levels of A^b-CLIP than nTg immature bmDCs (Fig. 4,B). The same observations were made for mature bmDCs, confirming that A^b-CLIP levels on both immature and mature bmDCs directly correlated with DO expression. Cell surface levels of A^b-CLIP were higher on H-2Mα-deficient bmDCs than on TgHi bmDCs (Fig. 4,B), which suggests that H-2M activity is not completely inhibited in DO Tg bmDCs. This supported our earlier observation of a pool of free H-2M in the DO Tg bmDCs (Fig. 3).

Surface A^b levels on bmDCs were not significantly altered by Tg DO expression (Fig. 4,C, see further discussion). Because we observed that DO Tg DCs had higher A^b-CLIP levels, we expected that DO Tg DCs would have reduced surface levels of A^b-peptide complexes. To test this, we stained Tg and nTg bmDCs from C57BL/6×B10.BR F₁ mice for A^b molecules complexed with the endogenous Eα_52–68 peptide (A^b-Eα), which are specifically recognized by the mAb Y-Ae (25, 41) (Fig. 4,B). In both immature and mature bmDCs, surface levels of A^b-Eα complexes were reduced in bmDCs from TgHi mice, whereas TgMed bmDCs had a slight reduction in A^b-Eα levels, compared with nTg bmDCs. The A^b-specific mAb KH74 exhibits reduced reactivity to A^b-CLIP (42) and therefore preferentially recognizes A^b-peptide complexes. We observed reduced surface levels of A^b-peptide complexes on TgHi bmDCs and, to a lesser extent, on TgMed bmDCs compared with nTg, supporting that DO expression resulted in higher surface levels of A^b-CLIP and lower surface levels of A^b-peptide complexes (Fig. 4 B). These data suggest that high levels of DO are required to inhibit class II-peptide loading of endogenously expressed Ags.

Next, we examined the effect of DO expression on total cell surface levels of class II molecules by staining DO Tg and nTg bmDCs with a panel of mAbs to three different class II alleles (A^b, A^k, and E^k) (Fig. 4,C). As expected, all class II alleles were up-regulated on the surface of mature bmDCs. The overall levels of A^b and E^k were not significantly altered by Tg DO expression, when examined by two A^b-specific mAbs (AF6-120.1 and 25-9-17) and an E^k-specific mAb (14-4-4S). To determine whether total A^b levels were unchanged in DO Tg bmDCs and whether A^b-CLIP and A^b-peptide levels were altered at steady-state in DO TgbmDCs, quantitative Western blotting of lysates from H-2M-deficient, DO Tg and nTg bmDCs was performed (Fig. 5). DO expression did not significantly alter steady-state A^b levels, but increased DO expression levels correlated with increased levels of A^b-CLIP and decreased levels of A^b-peptide (Fig. 5, A and B). These data confirmed the results of our FACS analysis and show that total A^b-peptide and A^b-CLIP, but not overall A^b levels, are altered by ectopic DO expression in bmDCs.

Surprisingly, A^k levels were significantly reduced on immature and mature TgHi bmDCs (Fig. 4,C). Similar results were observed when A^k levels were measured with two other mAbs (10-3.6 and 10.2.16; data not shown). Additionally, quantitative Western blot analysis of bmDC lysates showed that A^k levels were reduced in DO TgHi bmDCs showing that at steady-state, total A^k levels were reduced in the presence of DO (Fig. 5, C and D). These data suggest that DO expression may reduce the half-life of A^k molecules by blocking H-2M stabilization of empty or loosely occupied A^k molecules, as observed for H-2M-deficient H-2K B cells (43).

We also analyzed the cell surface phenotype of DO Tg and nTg splenic DCs that had developed in vivo (Fig. 6, A and B) as well as splenic B cells from DO Tg and nTg mice (Fig. 6,C). As seen with the bmDCs, A^b-CLIP levels were increased and A^b-peptide and cell surface A^k levels were decreased in DO Tg splenic DCs. Although the CD11c promoter should not be active in B cells, intracellular staining for DO clearly showed DO expression in a subset of B220⁺ B cells from the TgMed (∼3%) and TgHi (∼15%) mice (Fig. 6,C). This resulted in a slight elevation of A^b-CLIP staining for total splenic B cells from TgHi mice, but did not significantly alter the levels of A^b-Eα peptide complexes (Fig. 6,C). Gating of DO-positive and DO-negative B cells from the two Tg lines showed that the DO-positive B cell population had elevated surface levels of A^b-CLIP compared with the DO-negative B cell population (Fig. 6 D). Therefore, overexpression of DO in mouse splenic B cells, the cell type in which H-2O is normally expressed, also resulted in inhibition of H-2M function, evident by increased A^b-CLIP levels.

Transgenic DO expression directly correlated with increased cell surface levels of A^b-CLIP, suggesting that H-2M-mediated peptide loading was dramatically reduced in Tg DCs. To directly test this, we used an A^k-HEL_46–61-specific mAb (C4H3) to directly measure the surface density of A^k-HEL_46–61 peptide complexes following the uptake and processing of HEL by Tg and nTg bmDCs in vitro (Fig. 7,A). Because overall A^k levels were reduced on the surface of Tg DCs (Figs. 4–6), A^k-HEL_46–61 levels were normalized to A^k surface levels. Results showed that the proportion of A^k molecules loaded with HEL_46–61 (A^k-HEL_46–61) was significantly lower on TgHi bmDCs than on TgMed and nTg bmDCs (Fig. 7,B). The C4H3 mAb cross-reacts with a subset of A^k molecules occupied with endogenous peptides (23). Recognition by mAb C4H3 of these complexes on mock-treated TgHi, and to a lesser extent, on TgMed bmDCs was significantly reduced relative to mock-treated nTg bmDCs (Fig. 7, A and B). Because mAb C4H3 cross-reacts with A^k complexed with a subset of endogenous peptides, the above experiments do not distinguish an increase in C4H3 reactivity due to the processing and presentation of A^k-HEL_46–61 following HEL internalization from an up-regulation of the endogenous peptide presentation. Therefore, to definitively measure A^k-HEL_46–61 presentation by bmDCs we also measured the functional effect of DO expression on the A^k-restricted T cell response to the HEL_46–61 peptide. Results showed that presentation of A^k-HEL_46–61 by HEL-pulsed bmDCs was significantly reduced in DO-positive bmDCs compared with nTg bmDCs (Fig. 7,C). Reduced T cell recognition directly correlated with DO expression levels, supporting the direct measurement of A^k-HEL_46–61 formation with the C4H3 mAb (Fig. 7 A).

The effect of DO expression on the presentation of other exogenous and endogenous Ags was also examined. For presentation of exogenous Ags, purified bmDCs from A^b or A^k-E^k (Fig. 8,A) nTg and Tg mice were incubated for 2 h with Ag, washed, and presentation was measured by the addition of T hybridoma cells specific for either IgM and OVA (A^b-restricted) or by the addition of purified naive CD4 T cells from TCR Tg mice specific for CA (A^k-restricted) or pCC (E^k-restricted) Ags. Results showed that the effect of DO on exogenous Ag presentation was Ag-dependent (Fig. 8,A). DO, to varying extents, inhibited presentation of HEL (Fig. 7,C), IgM and OVA (Fig. 8,A); but had no effect on the presentation of CA and pCC (Fig. 8,A). Reduced presentation directly correlated with DO expression level. Both nTg and DO Tg DCs efficiently stimulated Ag-specific T hybridomas or naive T cells when pulsed with relevant peptides (Fig. 8 A).

For endogenous Ag presentation assays, T hybridoma cells specific for peptides derived from actin and β₂-microglobulin were added to titrated numbers of purified A^b expressing DO Tg and nTg bmDCs. Presentation of epitopes from these endogenous Ags was not significantly altered by DO expression (Fig. 8 B). We also assayed A^b-restricted endogenous peptide presentation using naive D10 T cells, which recognize A^b bound to endogenous peptides in an alloresponse (20). D10 T cells recognized Tg and nTg bmDCs equally well. As a negative control in both exogenous and endogenous Ag presentation assays, bmDCs from an H-2Ma-deficient mouse (A^b-restricted) that are defective in presentation (44) were used. Collectively, these studies showed that the inhibitory effect of DO on Ag presentation depended upon both the Ag and the route of presentation (exogenous vs endogenous), in agreement with results obtained by others in B cells, B cell transfectants, and B cell receptor (BCR) Tg mice (9, 10, 11, 12).

With the exception of one study (16), initial biochemical and cell culture-based studies collectively showed that DO inhibits DM-mediated peptide loading of class II (7, 8, 12). Subsequent studies that directly analyzed the effect of H-2O expression on Ag presentation to T cells by mouse primary B cells and B cell lines, a mouse spindle cell sarcoma cell line and a human melanoma cell line, yielded conflicting results. Overall, these studies showed that DO can inhibit (7, 11, 12, 13, 14, 15, 45), promote (9, 10, 16), or have no effect (9, 10, 12, 45) on DM-mediated class II-peptide loading. Given the conflicting nature of the data concerning H-2O/DO function, we expressed DO in mouse DCs by transgenesis to further evaluate the role of DO/H-2O in the class II Ag processing pathway. DCs were selected for these studies because they are potent APCs in which the class II pathway has been well characterized (17, 18, 19). DCs from two independent DO transgenic mouse lines that expressed two different levels of DO were evaluated. Results clearly showed that as DO expression level increased, A^b-CLIP levels also increased and class II-peptide levels decreased (Figs. 4–6). In terms of Ag presentation, DO expressing DCs presented some, but not all exogenously supplied Ags less well than nTg DCs (Figs. 7 and 8,A). The presentation of endogenous Ags was unaltered in the presence of DO, as measured with a T cell line and T cell hybridomas (Fig. 8 B). However, when assayed by mAb Y-Ae and C4H3 binding, presentation of both the Eα and C4H3 cross-reactive endogenous epitopes was reduced in the presence of DO. Finally, our results showed that DO/H-2O expression results in the accumulation of H-2M, as both nTg DCs and H-2O-deficient B cells expressed ∼2-fold less H-2M than DO/H-2O expressing cells.

Remarkably, A^b-CLIP levels were increased in the DCs from Tg Med mice although only 10% of the H-2M was complexed with DO. At first these data appear to be inconsistent with published work showing that A^b-CLIP levels are not increased in H-2O over-expressing Tg mice (11) or decreased in H-2O-deficient mice (12). However, we have found that, unlike in human B cells, very little H-2M is complexed with H-2O in wild-type mouse B cells (<10%; our unpublished observations). Therefore, we hypothesize that in our mice the percentage of H-2M in complex with DO exceeds what is typically found in mouse B cells, which results in increased A^b-CLIP levels. Presumably, the previously reported Tg mouse expresses less H-2O than our Tg mice (11). Our results are, however, consistent with data from human cell transfection studies (7, 8, 13) and also from data that correlated DO down-regulation with decreased class II-CLIP levels in germinal centers (14, 15, 37). However, we cannot completely rule out the possibility that DO and H-2O are somewhat functionally divergent. This possible difference in function must be subtle because the proteins are ∼75% identical at the amino acid level. Furthermore, the H-2M/DO complex behaves as expected in many other ways, including proper transport and subcellular localization.

DO expression in DCs resulted in a dramatic increase in surface levels of A^b-CLIP and a corresponding reduction in A^k-HEL_46–61 levels following HEL internalization. We were, therefore, somewhat surprised that the effect of DO expression on exogenous Ag presentation was Ag-specific. This difference was not allele-specific, as DO reduced the presentation of peptides derived from HEL (A^k-restricted), IgM (A^b-restricted), and OVA (A^b-restricted) but not CA (A^k-restricted) and pCC (E^k-restricted) (Figs. 7,C and 8,A). Others have shown that H-2O expression had no effect on endogenous self-Ag presentation (45, 46). We also observed that DO expression did not affect the presentation of the endogenous Ags to T cell hybridomas (Fig. 8,B). In contrast, when we assayed endogenous Ag presentation by measuring surface class II-peptide complex density with mAbs (Y-Ae and C4H3), presentation levels directly correlated with DO expression levels (Figs. 4,B, 6,A, and 7, A and B). Differences in endogenous Ag presentation may be more apparent when using mAbs to examine a larger subset of class II-peptide complexes, compared with T cell hybridomas that are specific for a single class II-peptide complex.

Our data suggest that the free H-2M in Tg DCs (Fig. 3) is sufficient to present some Ags or that the H-2M/DO complex retains some activity in vivo. Alternatively, a combination of both the free H-2M and H-2M/DO complex may contribute to the observed effects on Ag presentation. Differences in Ag structure, stability, interaction with different class II alleles, and retention time in peptide loading compartments may also have contributed to whether or not DO altered the peptide loading of a specific Ag (5, 9, 13). Other studies have reported a marked inhibition of exogenous Ag presentation by H-2O in B cells and B cell transfectants (11, 12). However, DCs are more efficient APCs than B cells. Therefore, there may be sufficient peptide loading, even in the presence of high levels of DO, to induce effective T cell stimulation, especially when the Ag was endogenously expressed (Fig. 8 B).

Cell surface A^k levels are reduced ∼2-fold on H-2M-deficient B cells, suggesting that, in vivo, A^k molecules have a reduced half-life in the absence of the chaperone function of H-2M (43). Our finding of reduced A^k surface levels in bmDCs and resident splenic DCs from the DO Tg mice (Figs. 4,C and 6,B), in addition to decreased steady-state levels of A^k β in DO Tg bmDCs (Fig. 5), strongly suggests that DO association with H-2M blocks its chaperone function. Reduced cell surface A^k levels were observed with three different A^k mAbs (11-5.2, 10.2.16 and 10-3.6; Figs. 4–7 and data not shown), making it unlikely that our results are due to a serological change. Previous studies have shown that DM/DO complexes are associated with HLA-DR in B cells and that DM/DO complexes can function as a class II-specific chaperone in vitro (16), suggesting that the DM/DO complex may function as a class II-specific chaperone in vivo. However, subsequent studies showed that the DM/DO-DR complex also contains the tetraspan molecules, CD82 and CD63 (24). Because CD82 itself has been shown to associate with other tetraspan proteins (CD81, CD53, CD9, and CD37) as well as CD21, CD19, CD4, integrins and MHC class I (47, 48, 49, 50, 51, 52), it is not currently known whether DM/DO complexes interact directly or indirectly with class II in vivo. Our finding of reduced A^k levels in DO-expressing DCs strongly supports that the H-2M/DO complex cannot function as a class II-specific chaperone in vivo and that DM/DO-class II association in cells may be mediated indirectly by other molecules such as CD82 and CD63. Once again, we believe it is unlikely that our results are simply due to the use of DO instead of H-2O for the reasons discussed above.

Perhaps the most surprising result of these studies is the finding that DO/H-2O increases H-2M levels both in DO Tg DCs and also in primary mouse B cells (Fig. 2). These data support what has been observed in human B cells; DM levels are up-regulated in naive and memory B cells, which have increased levels of DO and DM compared with germinal center B cells that express lower levels of both DO and DM (14, 15, 37). How DO/H-2O increases H-2M levels is not clear. H-2M exhibits slow turnover and a long half-life in vivo (data not shown), making kinetic analysis difficult. We have not observed an effect of H-2O on H-2M synthesis in B cells and thus favor the idea that H-2O prevents or decreases the rate of H-2M degradation. Previous studies have reported H-2O expression does not alter H-2M levels (11, 12). These studies examined H-2M levels by either nonquantitative Western blotting or by intracellular staining and FACS analysis in H-2O antisense and over-expressing B cell lines and also in primary B cells from H-2O-deficient and over-expressing Tg mice. We believe the discrepancy between these studies and our own studies can be explained by our use of an extremely sensitive and quantitative fluorescence-based Western blotting technique to measure steady state H-2M levels.

Although it is unclear at this time why DO/H-2O causes the accumulation of H-2M, this finding has several implications. First, there may be some regulated mechanism in vivo by which H-2O is selectively degraded or that promotes the dissociation of H-2O from H-2M. Both of these scenarios would result in an activation of class II-peptide loading. Alternatively, several studies have suggested that H-2M/H-2O complexes specifically inhibit peptide loading only in early endosomes, favoring peptide loading in late endosomes and/or lysosomes (9, 13, 15). The presence of more DO/H-2O may function to prolong the half-life of H-2M (as a H-2M/H-2O complex) in late endosomes and/or lysosomes, which would promote peptide loading in these compartments. We are currently trying to determine whether H-2M accumulates in specific subcellular compartments of H-2O expressing B cells. Finally, although in vitro studies have definitively shown that purified DM/DO complexes are completely inactive in terms of their peptide loading ability (7, 8, 12), studies to date have not ruled out that the DM/DO complex maintains peptide loading activity in the membranes of intact cells. Studies by others showing H-2O can promote presentation of a small number of epitopes supports that the H-2M/H-2O complex may be active in vivo. In this case, accumulation of H-2M by H-2O would promote class II presentation.

The specific expression of DO/H-2O in B cells suggests that it has a unique function in terms of the class II processing pathway. Studies to date concerning the impact of DO/H-2O on Ag presentation specifically by B cells have been largely inconclusive. Although B cells internalize Ag by both fluid phase and by BCR-mediated endocytosis, B cells are relatively inefficient at fluid phase endocytosis. Therefore, the physiologically relevant pathway for Ag uptake by B cells is likely to be via the BCR. The initial report of presentation of Ags internalized via fluid phase uptake showed that Ags were presented less well by H-2O-deficient B cells compared with wild-type B cells (12). These findings were recently substantiated by H-2O over-expression and anti-sense RNA knockdown studies in mouse B cell lines and Tg mice (11). In contrast, however, two additional studies that used B cells from H-2O-deficient mice showed that H-2O had no effect on Ag presentation following fluid phase uptake of Ag. In contrast, these two studies showed that Ags internalized by the BCR were presented less well by H-2O-deficient B cells (13 of 33 epitopes) (9, 10). Collectively, these data support the idea that H-2O promotes Ag presentation via the BCR and dampens fluid phase Ag presentation. Because our studies showed that DO/H-2O promotes H-2M accumulation in mouse B cells (Fig. 2), a possible mechanism by which H-2O could both promote and inhibit Ag presentation would involve H-2M/H-2O accumulation in endocytic compartments of B cells in the inactive form. Dissociation of H-2O from H-2M/H-2O complexes would then result in an immediate activation of the class II-peptide loading via the newly released H-2M, as initially suggested by Karlsson and coworkers (6). Although biochemical studies to date have not shown that H-2M/H-2O dissociates following BCR activation (data not shown and Ref.12), an in vivo evaluation of H-2M/H-2O association will be necessary to more carefully evaluate this idea. Given the conflicting nature of studies to date concerning DO/H-2O function, future studies are warranted to evaluate and define the role of DO/H-2O in the class II Ag-processing pathway in B cells and also to define the impact of DO/H-2O on B cell-dependent immune responses in vivo.

We thank the past and current members of the Denzin and Sant’Angelo laboratories for helpful discussions and Frances Weis-Garcia, Starky Bibb, Jiri Treka, and MSKCC mAb and Mouse Genetics Core Facilities for technical assistance. We also thank Drs. Alexander Rudensky, Lars Karlsson, Peter Cresswell, Charles Janeway, Jr. (deceased), Hans Zweerink, and Carolyn Doyle for mice, cell lines, and reagents.