Distinct Sequences in the Cytoplasmic Domain of Complement Receptor 2 Are Involved in Antigen Internalization and Presentation¹

The B cell receptor (BCR)³ consists of surface Ig (sIg) H and L chain noncovalently associated with the Igαβ heterodimer (1, 2). This receptor is involved in the capture and presentation of Ag to MHC class II-restricted T cells. B cells expressing an Ag-specific BCR are 10³–10⁴ times more efficient at presenting Ag than nonspecific B cells (3, 4, 5). The molecular mechanisms underlying this enhancement have been the subject of considerable investigations during the past 10 years. Data now exist demonstrating that Igα/β and the Ig components of the BCR contain intrinsic sequences directly responsible for improving the efficiency of the BCR. This optimization is mediated by enhanced Ag binding as well as improved internalization and subcellular targeting to MHC class II-containing compartments within the B cell (6, 7, 8, 9, 10).

Other receptors expressed on B cells have also been shown to be involved in Ag internalization (11, 12), subsequent processing, and presentation (13, 14, 15, 16, 17). Unlike the Ag-specific BCR, these receptors recognize Ags either in preformed immune complexes (FcRs, reviewed in Ref. 18) or complexed with proteins derived from the complement system (CRs; reviewed in Ref. 19). In particular, it has been shown in vivo that targeting Ags to the CR1/CR2 (CD35/21)/CD19 complex substantially increases the resulting Ab response (20, 21). Several laboratories have shown in vitro that this is partly due to complement-tagged Ags that cross-link CR2 and the BCR on B cells (22, 23, 24, 25, 26).

As well as having an important role as a BCR modulator, CR2 has also been shown to be capable of Ag presentation in the absence of BCR cross-linking. Studies have demonstrated that anti-CD21 mAbs (13), Ags incubated with nonimmune natural Ab (27), or immune sera (28) were efficiently presented in a CD21-dependent, BCR-independent manner. The underlying molecular mechanisms as well the physiological consequences of CR2-mediated Ag presentation in the absence of BCR cross-linking are unknown. By introducing either wild-type mouse CR2 (WTmCR2) or various mutant mouse CR2 (mCR2) cDNAs along with cDNAs encoding a tetanus toxin C fragment (TTCF)-specific BCR (4, 5) into the CR1/2⁻ mouse B cell line CH27 (29), we have investigated the molecular requirements for mCR2-mediated Ag presentation in both the presence and the absence of BCR cross-linking. Here we demonstrate for the first time that the cytoplasmic domain of mCR2 is required for efficient internalization when cross-linked to the BCR, and we have identified a tyrosine-based motif present in this domain that is required for this. Thus, our study provides clear evidence that mCR2 sequences other than those involved in ligand binding are involved in both the internalization and subsequent presentation to T cells of complex Ags that cross-link mCR2 to the BCR. We also show that additional sequences within the mCR2 cytoplasmic domain are essential for full Ag presentation in the absence of BCR cross-linking. These findings raise the possibility that Ag presentation through mCR2 may occur in vivo in the absence of BCR cross-linking, relying on this domain to target Ag to MHC class II molecule-containing organelles.

Mammalian cells were incubated at 37°C, in an atmosphere of 5% CO₂. Cell culture products unless otherwise stated were purchased from Life Technologies (Gaithersburg, MD). CH27 (CR1/2⁻CD19⁺ H-2E/A^k B lymphoma) and LK35.2 cells (CR1/2⁺CD19⁺ H-2^d/k B lymphoma) were grown in RPMI 1640 containing 10% FCS, 100 μg/ml kanamycin, 2 mM glutamine, 1 mM sodium pyruvate, nonessential amino acids, and 50 μM 2-ME. Splenic B cells were isolated from C57BL/6 mice by negative selection using anti-CD43 microbeads (Miltenyi Biotec, Auburn, CA). The unbound B cells were eluted from a CS column according to the manufacturer’s conditions and used in internalization assays as described below. Insect cells were maintained at 27°C. SF21s originating from the Spodoptera frugiperda IPLBSF-21 cell line (Life Technologies) were grown in TC100 medium with 10% FCS, 2 mM glutamine, 100 U/ml penicillin G sodium, and 100 μg/ml streptomycin sulfate. High Fives (Life Technologies) derived from ovarian cells of Trichoplusia ni were grown in Express Five SFM medium supplemented with 2 mM glutamine, 100 U/ml penicillin G sodium, and 100 μg/ml streptomycin sulfate.

CBA mice were injected s.c. in each hind leg with a total of 100 μg of TTCF emulsified in CFA (Sigma-Aldrich, St. Louis, MO). After 10 days, a single-cell spleen suspension was made and plated out into 24-well plates at 5 × 10⁶ cells/ml in X-Vivo 15 medium (BioWhittaker, Walkersville, MD) supplemented with 100 μg/ml kanamycin, 2 mM glutamine, 50 μM 2-ME, and 1.5 μg/ml TTCF. Cells were then restimulated with Ag initially every 2 wk. In brief, ∼2 × 10⁵ T cells were mixed with 5 × 10⁶ fresh irradiated CBA splenocytes in the presence 1.5 μg/ml TTCF. The third restimulation was as before but with the inclusion of 5% T cell growth factor (a gift from Dr. A. Antoniou, Dundee University, Dundee, U.K.) prepared from Sprague-Dawley rat spleen cells as described (30). After the fourth restimulation, the line was maintained in RPMI supplemented with 100 μg/ml kanamycin, 2 mM glutamine, 50 μM 2-ME, and 5% heat-inactivated FCS and restimulated as before. Using a panel of 88 overlapping 17-mer peptides (a gift from Prof. C. Watts, Dundee University) spanning the entire TTCF established that the dominant specificity of this line was targeted to aa 1028–1043 (ANKWVFITITNDRLSS).

A 3.084-kb cDNA-encoding mCR2, in the mammalian expression vector pAN, was obtained from Drs. H. Molina and V. M. Holers. This plasmid (pANmCR2) was used as WTmCR2 from which the subsequent cytoplasmic domain mutants were derived.

To generate the C-terminal truncated mCR2 mutants, pANmCR2 was amplified with a single 5′-oligonucleotide CD21Seq6, (5′-agggatttcaacctgggaaaa-3′) paired with either CD21KHRSalI (5′-gcggtcgactcatctgtgttttaatatcataca-3′) for the KHR mutant, CD21TMYYTKSalI (5′-gcgtcgactcactttgtataataattgctttc-3′) for the YYTK mutant, CD21TMLETSalI (5′-gcgtcgactcatgtttctaaatgaagagctcc-3′) for the LET mutant, or CD21TMYSISalI (5′-gcgtcgactcaaatagaatatacttctcgtgtttc-3′) for the YSI mutant. Resultant PCR products contained sequence from short consensus repeat (SCR) 15, the transmembrane domain, the required region of the cytoplasmic domain followed by a termination codon and a SalI restriction site. After cloning into pCR4Blunt-TOPO (Invitrogen, San Diego, CA) and confirmation of the truncations by sequencing, a ClaI-SalI fragment was then excised and used to replace the WT sequence in pANmCR2.

A 478-bp ClaI-SalI fragment (containing sequence from SCR15, the transmembrane, and the full length cytoplasmic domain) was excised from pANmCR2 and subcloned into pSL1180 (Pharmacia, Peapack, NJ). The resulting plasmid, pSL1180/mCR2–478, formed a template for PCR site-directed mutagenesis using the QuickChange technology (Stratagene, La Jolla, CA). Oligonucleotides MYFYF5 5′-gtatgatattaaaacacagagaaagcaatttttttacaaagacaagaccc-3′ and MYFYF3 5′-gggtcttgtctttgtaaaaaaattgctttctctgtgttttaatatcatac-3′ for the YY_986/7FF mutant, MYAYA5 5′-gtatgatattaaaacacagagaaagcaatgctgctacaaagacaagaccc-3′ and MYAYA3 5′-gggtcttgtctttgtagcagcattgctttctctgtgttttaatatcatac-3′ for the YY_986/7AA mutant, MYF5 5′-cgagaagtatattctattgatccatttaacccagcaagctg-3′ and MYF3 5′-cagcttgctgggttaaatggatcaatagaatatacttctcg-3′ for the Y₁₁₀F mutant, and MYA5 5′-cgagaagtatattctattgatccagctaacccagcaagctg-3′ and MYA3 5′-cagcttgctgggttagctggatcaatagaatatacttctcg-3′for the Y₁₁₀A mutant were used to incorporate mutations into pSL1180/mCR2-478DNA. After DpnI digestion, amplified products were transformed into TOP10 (Invitrogen)-competent Escherichia coli. After DNA recovery and sequencing to identify mutations, the mutant 478-bp ClaI-SalI fragment was used to replace that in pANmCR2.

The cloning of the H and L chains for the human sIgG from the TTCF-specific 11.3 EBV-transformed B cell clone (4), and its expression has been described (5). To generate a variant BCR that fails to associate with the Igαβ heterodimer, the transmembrane domain of the 11.3 γ₁ heavy chain was replaced with that from human CD4 based on reports by Williams et al. (31) and Wienands et al. (32). pSL1180/276 containing the transmembrane and cytoplasmic domains of the 11.3 γ H chain was mixed with a plasmid containing the extracellular and transmembrane domains of human CD4 joined to a truncated cytoplasmic domain (encoding only the first three amino acids, lysine, valine, lysine) of the 11.3 γ H chain (5) and amplified with the following three oligonucleotides based on a method described (33): pSG5rev 5′-caactagaatgcagtg-3′, M13 (–20) (Stratagene) 5′-tgaccggcagcaaaat-3′, and CD4/γct 5′-gagaagatccacttcaccttgaagaagatgcctagcccaa-3′. The resultant fragment, confirmed by sequencing, contained the extracellular domain of the 11.3 γ-chain joined to the CD4 transmembrane with a truncated 11.3 γ cytoplasmic domain and was used to replace the 276 fragment in the 11.3 γ-chain WT cDNA.

LK35.2 clones expressing both the WT and mutant human sIgG were surface labeled in PBS with 0.5 mg/ml Sulfo-NHS-LC-biotin (Pierce, Rockford, IL) before lysis in 0.5% digitonin (Serva, Heidelberg, Germany). Lysates were precleared at 50,000 × g, and the BCR complex was isolated using TTCF coupled to Sepharose 4B (Pharmacia). Samples were then removed from the matrix using SDS-PAGE loading buffer without reduction and run through primary SDS-PAGE. Gel slices containing immunoprecipitated material were then reduced and reloaded onto secondary SDS-PAGE separation followed by transfer to a nitrocellulose membrane. Membranes were probed with streptavidin peroxidase (Roche, Basel, Switzerland) and developed using an ECL detection kit (Pharmacia) and ECL hyperfilm. The distinctive migration of the mouse Igαβ was used to confirm its association with the WT human sIgG and absence from the mutant BCR.

For both LK35.2 and CH27 B cell transfections, ∼10⁷ cells were washed and resuspended in 0.8 ml of PBS containing 100 μg of linear DNA and electroporated at 1050 μF, 250 V, and 99 Ω. Drug-resistant clones were selected and screened for surface expression using a combination of FACS and radioligand-binding assays. Those clones used in experiments were representative of several and showed similar levels of surface mCR2 or sIgG expression.

B cells were stained with the anti-CR1/2 mAb 7E9 (34) followed by mouse F(ab′)₂ anti-rat IgG (H + L)-FITC (Jackson ImmunoResearch Laboratories, West Grove, PA) as well as with biotinylated Fd mouse anti-human IgG monoclonal (Zymed Laboratories, San Francisco, CA) followed by streptavidin-FITC (Southern Biotechnology Associates, Birmingham, AL) as described (5). On a BD Biosciences FACSCalibur, 10,000 events were collected and analyzed using FlowJo (Tree Star, Ashland, OR).

Biotinylated 7E9 or 7G6 (anti-mCR2) mAbs were ¹²⁵I-labeled with Bolton-Hunter reagent (Amersham, Arlington Heights, IL) as described (5). Splenic B cells, transfected CH27, or LK35.2 cells were incubated with 5 μg/ml ¹²⁵I-labeled mAb in RPMI-HEPES-2% FCS and 10 μg/ml biotinylated goat F(ab′)₂ anti-mouse IgG (H + L) (Jackson ImmunoResearch Laboratories) or biotinylated Fd mouse anti-human IgG (Zymed) for 1 h at 4°C. After unbound label was washed off, an additional 30-min incubation at 4°C with 10 μg/ml streptavidin-FITC (Southern Biotechnology) was used to cross-link surface-bound biotinylated mAb. The cells were then incubated at 37°C for various times before returning to ice. Surface-bound label was stripped using either an acid wash (150 mM HCl, 150 mM NaCl for 5 min) or treatment with Pronase (3 mg/ml; Roche)-DNase I (1 mg/ml; Sigma-Aldrich) for 1 h at 4°C. Each time point was performed in duplicate, and the percentage label internalized at each time point was calculated after the fractions were counted on a Packard MINAXI gamma counter.

The baculovirus transfer vector pBP68-03 (Clontech Laboratories, Palo Alto, CA), encoding three tandem copies of mouse C3d, was a generous gift from M. Steward, V. Cox, and R. Smith. It used extensive third-base redundancy to reduce the tendency for homologous recombination that had been observed between identical wild-type copies of the C3d coding sequence (M. Stewart and R. Smith, unpublished observations). Codon variant forms of murine C3d were generated either by extensive site-directed mutagenesis or by de novo synthesis from overlapping oligonucleotides (Sigma-Genosys, The Woodlands, TX). The pBP68-03 vector therefore contained a mouse (C3d)₃ (wild type-variant 1-variant 2) gene with each module being separated by flexible glycine-serine-rich linkers of ∼15 amino acids. The open reading frame of TTCF was excised from an N-terminal His₁₀ fusion in the E. coli expression vector pET-16b (Novagen, Madison, WI; Ref. 35) (a gift from C. Watts) and subcloned via pcDNA3 (Invitrogen) into a modified form of pBP68-03 using NdeI and XbaI restriction sites. The resulting chimeric open reading frame consists of (C3d)₃ followed by TTCF. Using the Clontech baculovirus expression kit, recombinant viral particles containing either the (C3d)₃ or the (C3d)₃-TTCF open reading frames were generated, and used to infect High Five insect cells. The supernatant from these cells was collected 36 h after infection, supplemented with 100 μM dichloroisocoumarin (Sigma-Aldrich), and applied to a DEAE-Sepharose fast flow anion exchange column (Pharmacia). Fractions eluted with 0.1 M and 0.2 M NaCl, 50 mM Na₂HPO₄ containing (C3d)₃ or (C3d)₃-TTCF were pooled and buffer exchanged several times into PBS using MWCO spin columns (VivaScience, Hannover, Germany). SDS-PAGE analysis confirmed both (C3d)₃ and (C3d)₃-TTCF purity. In addition, gels were transferred to membrane, and the identity of the chimeric Ags was confirmed by probing with anti-TTCF mouse monoclonal 31E11 (a gift from A. Antoniou) (17) or a cross-reactive anti-human C3d polyclonal serum (DAKO, Glostrup, Denmark) and developed using an ECL detection kit (Pharmacia) and ECL hyperfilm. Protein concentrations for use in assays were calculated using a bicinchoninic acid kit (Pierce). TTCF was purified from the lysates of NovaBlue (DE3; Novagen) E. coli transformed with the open reading frame of TTCF as an N-terminal His₁₀ fusion in pET-16b (35) as described (36).

Duplicate wells containing 1 × 10⁵ CH27 cells (or various CH27 transfectants) and 2 × 10⁴ T cells in the presence of graded doses of Ags in a total volume of 200 μl were incubated at 37°C for 24 h in 96-well flat-bottom plates (Costar, Cambridge, MA). T cell stimulation was measured as IL-2 release and detected by an anti-IL-2 capture ELISA. The titration of recombinant IL-2 on each ELISA plate allowed the quantification of IL-2 present in each experimental well.

The role of mCR2 in Ag presentation was addressed by first investigating its ability to internalize ligands from the cell surface. Using splenic B cells, we measured the internalization of ¹²⁵I-labeled, biotinylated anti-mCR1/2 mAbs (7E9 or 7G6; Ref. 34). These Abs both demonstrated negligible FcRγ binding and displayed specific binding after modification. After incubation with subsaturating concentrations of labeled anti-mCR1/2 mAb on ice, and removal of unbound ligand, cells were incubated for various time points at 37°C. After incubation, cells were shifted to ice and pelleted, and the proportion of internalized radiolabel was determined after removing the remaining surface label with either acid or Pronase. Fig. 1,A shows that we failed to detect significant internalization of mCR2 even after 20 min at 37°C. This was also seen when streptavidin-FITC, which cross-links several mCR2 molecules, was added before 37°C incubation (Fig. 1,A). It therefore appears that neither the minimal cross-linking of mCR2 with intact anti-mCR1/2 mAb nor the more extensive cross-linking induced by the addition of streptavidin was sufficient to mediate mCR2 internalization. However, when mCR2 was cross-linked to the BCR by including a biotinylated F(ab)₂ fragment of rat anti-mouse Ig, we were able to detect mCR2 internalization at a rate of ∼2%/min with up to 20% of the total radiolabel internalized by 10 min (Fig. 1 A). It can be also be seen that independent ligation of both mCR2 and BCR does not induce mCR2 internalization. The rate of mCR2 internalization, when cross-linked to the BCR, is very similar to that of the BCR when cross-linked to CR2 measured by Cherukuri et al. (24). Thus, cross-linking to the BCR appears to be essential for the measurement of mCR2 internalization in these assays.

Previous studies have shown that BCR internalization is influenced by its association with the Igαβ heterodimer (6, 7). We therefore investigated whether this dependency could be extended to mCR2 internalization when cross-linked to the BCR. To allow testing of this hypothesis, we conducted internalization assays in the LK35.2 B cell line. Radiolabeled anti-mCR2 mAb reacted with ∼3 × 10⁴ binding sites/cell (not shown) and, as seen with splenically derived B cells, we demonstrated that mCR2 expressed on LK35.2 cells also requires BCR cross-linking for internalization (Fig. 1,B). Using LK35.2 transfectants expressing either a WT (5) or a mutant human BCR (which fails to associate with the mouse Igαβ heterodimer; see Materials and Methods), we cross-linked mCR2 to the BCR and measured the subsequent internalization of mCR2 (Fig. 2,A). Control LK35.2 cells expressing no human BCR, or the two LK35.2 human sIgG transfectants, were incubated with ¹²⁵I-labeled, biotinylated anti-mCR2 mAb in the presence of a biotinylated Fd mouse anti-human IgG mAb (this Ab did not cross-react with the mouse BCR expressed by LK35.2). After 1 h on ice, the cells were washed and treated as before to cross-link mCR2 to the various BCRs. Fig. 2 A illustrates that mCR2 internalization (again at a rate of ∼2%/min) was detected only in cells expressing the WT human BCR. Cells either lacking a human BCR or expressing the mutant human BCR had similar low levels of internalization. We conclude that mCR2 internalization when cross-linked to the BCR is dependent on association of the Igαβ heterodimer.

We also wanted to test whether any sequences in mouse mCR2 were involved in its internalization when cross-linked to the BCR. Previous reports using human CR2 (hCR2) expressed in both fibroblasts and the erythroleukemic K562 cell line showed a requirement for the cytoplasmic domain in the internalization of C3d,g and for EBV infection (11). On examination of the predicted amino acid sequence of the cytoplasmic domain of mCR2, several areas are of interest (highlighted in Fig. 2,C). Tyrosine residues (Y) (at positions 986/7, 1005, and 1110), and a leucine, histidine, leucine (LHL) motif at positions 997/8/9 bear resemblance to motifs present in cytoplasmic domains of other receptors involved in either endocytosis or subcellular targeting (37, 38). To directly address the role of the mCR2 cytoplasmic domain in its internalization, we compared the internalization of WT mCR2 to that of a mutant mCR2 lacking a cytoplasmic domain. cDNAs encoding either WTmCR2 or a mutant mCR2, (KHRmCR2) that lacked a cytoplasmic domain (except for the first three cytoplasmic residues lysine, histidine and arginine) were transfected into the mCR1/2⁻, CD19/sIg-positive mouse B cell line CH27 (29). As seen in splenic mouse B cells and LK35.2 cells, transfected WTmCR2 internalization in CH27 cells is dependent on BCR cross-linking (Fig. 2,B). Because CH27 cells lack both mCR1 and mCR2 and were transfected with the cDNA encoding only mCR2, this finding indicates that the internalization measured in Figs. 1, A and B, and 2,A were not influenced by the presence of mCR1. After confirming comparable levels of mCR2 surface expression on drug-resistant clones (by FACS analysis; not shown), we compared the internalization of WTmCR2 to KHRmCR2 when they were both cross-linked to the BCR. Fig. 2,D shows that the rate of internalization of the cytoplasmic domain deletion mutant, KHRmCR2, was substantially reduced compared with that of WTmCR2 (Fig. 2 D). Thus, mCR2 requires both its cytoplasmic domain and the association of the Igαβ heterodimer with the BCR for full internalization capacity.

To identify residues within the cytoplasmic domain of mCR2 required for internalization when cross-linked to the BCR, we constructed a panel of truncated mCR2 mutants. cDNAs encoding mutant mCR2s lacking the C-terminal tyrosine at position 1110 (YSImCR2), tyrosine 1110, and the tyrosine at position 1005 (LETmCR2) or both tyrosine motifs and the LHL motif at positions 997/8/9 (YYTKmCR2) were made. The predicted amino acid sequence of the WTmCR2 cytoplasmic domain and that of the mutants is shown in Fig. 3,A. Fig. 3,B shows that when expressed in CH27 B cells and cross-linked to the BCR, the YYTKmCR2 mutant (truncated shortly after the two tyrosines at positions 986/7) is capable of full internalization at a rate similar to that of WTmCR2 (∼2%/min). To see whether the two tyrosine residues (YY_986/7) present in YYTKmCR2 were essential for full internalization of mCR2 (when cross-linked to the BCR), we generated two additional mCR2 mutants (Fig. 3,A). These mutants retained the full mCR2 cytoplasmic domain with the exception that the tyrosine residues at positions 986/7 were mutated to either alanine (YY_986/7AAmCR2) or phenylalanine (YY_986/7FFmCR2). After transfection into CH27 B cells, the internalization of these mutant mCR2s (when cross-linked to the endogenous BCR) was measured, comparing them with both WT and KHRmCR2. When cells expressing the YY_986/7AA or the YY_986/7FFmCR2 were analyzed (Fig. 3 B), internalization was reduced to levels seen in the cells expressing the KHRmCR2 mutant. Therefore, the two tyrosine residues YY_986/7 are required for the internalization of mCR2 when it is cross-linked to the BCR.

Ag internalization by B cells is the first stage of a complex process leading to Ag presentation. We therefore investigated whether the cytoplasmic domain of mCR2 was required for efficient mCR2/BCR-mediated Ag presentation to T cells. CH27 cells, as well as CH27s expressing WTmCR2 and KHRmCR2 cDNAs, were transfected with both the H and L chain cDNAs encoding a human anti-TTCF IgG BCR (5). Drug-resistant clones were selected and screened for equivalent levels of both BCR and mCR2 by FACS (not shown). Together with untransfected CH27 cells, clones expressing TTCF-specific human BCR alone, TTCF-specific human BCR, and WTmCR2 or TTCF-specific human BCR and KHRmCR2 were examined for their respective Ag presentation capacities. To cross-link these two receptors and monitor Ag presentation, we constructed a recombinant chimeric Ag consisting of TTCF (35) linked to three copies of the activated C3 component, C3d ((C3d)₃-TTCF) (see Materials and Methods and Fig. 4,A, lane 3). A graded dose of either TTCF or (C3d)₃-TTCF was then added to the various B cell transfectants in the presence of an anti-TTCF TCL (see Materials and Methods). Ag presentation was measured by the detection of IL-2 in the assay supernatants after 24 h. Consistent with our previous studies in other cells (5), expression of an anti-TTCF human BCR either alone or in conjunction with WT or KHRmCR2 in CH27 cells allowed enhanced presentation of TTCF (Fig. 4,B) compared with the untransfected CH27 cells. However, when the same cells were incubated with the chimeric (C3d)₃-TTCF Ag, a different hierarchy in presentation efficiencies was seen (Fig. 4,C). CH27 cells expressing either a TTCF-specific human BCR alone or in conjunction with KHRmCR2 are capable of presenting (C3d)₃-TTCF at concentrations 10³–10⁴ lower than untransfected CH27 cells. (Fluid phase uptake of the chimeric Ag (C3d)₃-TTCF was considerably less than that of TTCF even when TTCF molar equivalents were compared. Experiments with other TTCF conjugates demonstrate this to be related to the significant increase in molecular mass; data not shown). When expressing a TTCF-specific human BCR in conjunction with WTmCR2, however, a further enhancement (∼5-fold) in Ag presentation is seen (Fig. 4 C). This level of enhancement is consistent with that shown by Cherukuri et al. (25) when Ag was targeted to endogenous mCR2 and an introduced BCR. These data, along with those of the Pierce laboratory (25), demonstrate that the substantial improvement in Ag presentation efficiency by BCR engagement can be further improved when Ag cross-links CR2 to the BCR. In addition, our experiments show that this further improvement requires the presence of the mCR2 cytoplasmic domain.

Despite a lack of detectable mCR2 internalization in our biochemical assays (unless cross-linked to the BCR; see Figs. 1 and 2), several reports have previously demonstrated hCR2-dependent Ag presentation in the absence of BCR cross-linking (13, 27, 28). Because Ag presentation to T cells is a more sensitive assay, we wished to examine the fate of Ags targeted to mCR2 in the absence of BCR cross-linking in our model system, particularly with attention to the role of the mCR2 cytoplasmic domain. We therefore repeated the Ag presentation assays using either TTCF, a combination of TTCF and (C3d)₃, or (C3d)₃-TTCF with CH27 transfectants expressing WT, YYTK, or KHRmCR2 this time in the absence of the TTCF-specific human BCR. Fig. 5,A shows that like hCR2, mCR2 does indeed enhance the presentation of TTCF in the absence of BCR cross-linking. CH27 cells expressing the WTmCR2 were capable of presenting (C3d)₃-TTCF at concentrations below that of the uncoupled TTCF when molar equivalents are compared. Thus, despite a lack of measurable mCR2 internalization in the absence of BCR cross-linking in our biochemical assays, we were capable of measuring CR2-mediated, BCR-independent Ag presentation. To examine the role of the cytoplasmic domain in this enhancement, we compared the ability of CH27 cells expressing WT, YYTK, or KHRmCR2 cDNAs to present TTCF or (C3d)₃-TTCF. As can be seen in Fig. 5,B, CH27 cells expressing WT, YYTK, or KHRmCR2 cDNAs present TTCF with similar efficiencies to the anti-TTCF TCL. In addition, the inclusion of recombinant (C3d)₃ in these assays resulted in similar presentation by all cells tested (Fig. 5,C). However, when the chimeric Ag (C3d)₃-TTCF was used, we now saw a requirement of the mCR2 cytoplasmic domain for full Ag presentation capacity (Fig. 5,D). Whereas the untransfected CH27s were very poor at presenting (C3d)₃-TTCF, expression of WTmCR2 increased presentation by ∼3–4 orders of magnitude. Presentation of (C3d)₃-TTCF by the cells expressing either YYTK or KHR mutant mCR2s, however, was ∼2–3 orders of magnitude better than that of untransfected cells but was not as efficient as that seen with cells expressing WTmCR2. Thus, sequences in the cytoplasmic domain of mCR2 are required for full Ag presentation capacity in the absence of BCR cross-linking. Furthermore, although the inclusion of the two membrane proximal tyrosines, YY_986/7, is sufficient to mediate mCR2 internalization when cross-linked to the BCR (Fig. 3 B), they are not sufficient for full Ag presentation.

To identify the sequences in the cytoplasmic domain required for full Ag presentation, we transfected CH27s with two additional truncated mCR2 mutants (see Fig. 3,A). As can be seen in Fig. 6,A, transfectants expressing LET or YSImCR2 still failed to present (C3d)₃-TTCF to the TCL as efficiently as those expressing WTmCR2. Because YSImCR2 lacks only the C-terminal 7 residues of the cytoplasmic domain, including a tyrosine at position 1110, we tested whether this residue was essential for optimal presentation. Two mutants were constructed in which this tyrosine was mutated to either alanine or phenylalanine in the context of an otherwise intact cytoplasmic domain (Fig. 3,A). When CH27s expressing the Y₁₁₁₀A or Y₁₁₁₀F mCR2 mutants were compared with those expressing either WT or KHR mCR2, Ag presentation was seen to be as efficient as that seen in the WTmCR2 (Fig. 6 B). Thus, whereas the last 7 residues of the cytoplasmic tail of mCR2 are essential for full Ag presentation in the absence of BCR cross-linking, the tyrosine residue at position 1110 is not.

Several laboratories have demonstrated enhanced B cell responses (20, 21) as well as improved Ag presentation (22, 23, 24, 25, 26), when Ags cross-link CR2 to the BCR. It has also been shown that when targeted to CR2 without BCR cross-linking, Abs (13) and Ags reacted with complement components (27, 28) are also presented extremely efficiently. In this study, we have examined the role of mCR2 in B cell Ag presentation. Taking a stepwise approach, we began by measuring the internalization kinetics of mCR2 in both ex vivo B cells and various B cell lines. We were surprised to find that, using conventional radiolabeling assays, we could detect mCR2 internalization only when it was cross-linked to the BCR. While we were conducting these experiments, a report from the Pierce laboratory described novel experiments measuring BCR internalization. The internalization rates measured for the BCR were lower when cross-linked to mCR2 (24). This reduced rate is comparable with the rate of mCR2 internalization seen in our experiments with BCR cross-linking. In a later report, the same laboratory failed to measure mCR2 internalization when cross-linked to the BCR (25). However, in this study, mCR2 internalization was measured using ¹²⁵I-labeled anti-mCR2 mAb 7G6 in the presence of C3d-hen egg lysozyme to cross-link mCR2 with a hen egg lysozyme-specific BCR (25). Because this mAb competes directly with C3d for mCR2 binding (39), it is possible that the failure to detect mCR2 internalization on BCR cross-linking in the Pierce report may simply reflect insufficient cross-linking.

Our study further demonstrates that the dependency of mCR2 internalization on the BCR is due to the association of the Igαβ heterodimer. Significantly, we have also shown that sequences within the cytoplasmic domain of mCR2 itself are required. This was evident by a reduction of mCR2 internalization in cells expressing a mutant (KHR) mCR2 lacking the cytoplasmic domain. In particular, we have shown that the two membrane-proximal tyrosine residues, YY_986/7, are sufficient for the internalization seen in WTmCR2. B cells expressing a mutant mCR2 (YYTK), with a truncated cytoplasmic domain containing these tyrosine residues, demonstrate full mCR2 internalization when cross-linked to the BCR. When these two residues were mutated to alanine residues in the context of the cytoplasmic domain, we did not detect internalization when mCR2 was cross-linked to the BCR. This clearly identifies these tyrosine residues as necessary as well as sufficient for internalization. In addition, mutating these residues to two structurally similar bulky phenylalanine residues did not restore internalization. This is in contrast to studies with the low density lipoprotein receptor, for example (40). With this receptor, mutating a tyrosine at position 807 to phenylalanine (but not to alanine) maintained internalization rates. This raises the possibility that the two tyrosine residues (YY_986/7) present in the cytoplasmic domain of mCR2 may act as substrates for phosphorylation events rather than structural determinants. Previous reports have provided limited evidence for ligand-induced phosphorylation of hCR2 (41) in association with a membrane phosphoprotein (p53) (42). A later study showed that this association could be mimicked using a synthetic peptide containing the cytoplasmic residues KHRERNYYTD from hCR2 (43). This sequence is very similar to the residues present in the truncated YYTKmCR2 mutant that maintains full internalization. It is therefore an intriguing possibility that the tyrosine residues YY_986/7 present in the cytoplasmic domain of mCR2 are involved in the interaction with a similar phosphoprotein. This may result in either direct or indirect interaction with the endocytic machinery of the B cell promoting the internalization of mCR2. Recent data have led to the proposal that coligation of mCR2 to the BCR enhances BCR signaling (24, 44). Another possibility, therefore, is that the two tyrosine residues YY_986/7 are either directly or indirectly involved in these signaling events.

To assess whether the cytoplasmic domain of mCR2 could also affect the ability of the mCR2/BCR complex to present Ags, we generated B cell transfectants expressing TTCF-specific BCRs in conjunction with WT and KHRmCR2. A chimeric Ag targeted to the mCR2/BCR complex was presented more efficiently than that targeted to the BCR alone, as has been previously described (25, 26). In addition, we demonstrated that this enhancement of presentation seen upon mCR2 cross-linking to the BCR does indeed require the presence of the mCR2 cytoplasmic domain. Transfectants expressing the specific BCR in conjunction with KHRmCR2 do not present this cross-linking Ag any better than mCR2⁻ cells expressing a TTCF-specific BCR. Again, the downstream events responsible for this remain unclear. It is conceivable that the cytoplasmic domain of mCR2, either directly or indirectly, is involved in subsequent targeting or signaling events leading to the observed enhancement of Ag presentation. As well as peptide-MHC complex recognition, T cell activation also relies on various costimulataory signals from APCs. Previous studies have indeed shown that ligation of the mCR2-CD19-BCR complex leads to increased levels of surface B7-1 (CD80) and B7-2 (CD86) (45). Using various cross-linking reagents on mCR2-BCR-transfected CH27 cells (which constitutively express high levels of both CD80 and CD86) however, we were unable to observe any increased levels on CD80, CD86 or MHC class II expression (data not shown). Thus, mCR2 cytoplasmic domain-dependent Ag presentation seen in CH27 cells is apparently independent of further increases in CD80/86 or MHC class II levels.

We have also investigated the situation in which CR2 acts independently of the BCR to enhance Ag presentation (13, 27, 28). Although we were unable to detect mCR2 internalization in our biochemical assays unless cross-linked to the BCR, we, like previous studies (27, 28), have demonstrated enhanced Ag presentation via CR2 in the absence of BCR cross-linking. We therefore tested the ability of various mCR2 mutants to present Ag when introduced into the CR2⁻ B cell line CH27 in absence of an Ag-specific BCR. As previously seen with hCR2 (27, 28), expression of WTmCR2 improved Ag presentation by several orders of magnitude compared with CR2⁻ cells. Once again a role for the cytoplasmic domain (now in the absence of the BCR) was seen. Cells expressing the KHRmCR2 were not as efficient as those expressing the WTmCR2. It was also apparent that the presence of the tyrosine residues YY_986/7 in the cytoplasmic domain was not sufficient for the restoration of Ag presentation. It was only when the last 7 C-terminal residues (1108–1114) were present that presentation was restored to WT levels. We have also shown that the tyrosine residue at position 1110 is not essential. Thus, two distinct regions of this domain are essential for internalization and Ag presentation. This finding is not without precedent. Previously, we made a similar finding with the cytoplasmic domain of a class-switched human IgG BCR. A residue present in the cytoplasmic domain was required for full Ag presentation despite not being essential for efficient internalization (5). Mice lacking the cytoplasmic domain of sIgG also had impaired secondary B cell responses (46). More recently, it has been shown that this class-switched domain is directly involved in regulating the amplitude of the secondary B cell response (47).

In a recent study, the requirement for intracellular targeting sequences in the BCR was shown to be directly related to the affinity of Ag for the receptor (8). When this was relatively low, mutant BCRs lacking targeting sequences but permitting efficient internalization were not capable of efficient Ag presentation. In contrast, the same mutants were as good as WT BCRs at presenting Ag binding with high affinity. This led the authors to speculate that B cells use a specialized receptor complex such as the BCR-Igαβ to allow efficient presentation of Ags with low affinity for the receptor. In this study, we provide evidence that CR2 also acts as a specialized receptor on B cells for the presentation of Ags complexed with activated components of the complement pathway. Recent crystallographic (48) and kinetic (49) analysis of the interaction of CR2 with its ligand C3d have demonstrated that the interaction does not follow a simple 1:1 model and has a relatively low affinity (K_D 3 μM). Although Ag-binding CR2 will cross-link several molecules, it is still likely that the resultant avidity will be insufficient for presentation by CR2 in the absence of BCR cross-linking unless specific targeting sequences exist. The identification of residues 1108–1114 as being essential for full presentation mediated by mCR2 in the absence of BCR cross-linking is consistent with this idea.

In conclusion, we believe that these studies offer new insights into the mechanisms involved in Ag presentation by CR2. When this receptor recognizes Ag in the presence of an Ag-specific BCR, sequences in its cytoplasmic domain are involved in the internalization of the Ag-mCR2 receptor complex. In the absence of an Ag-specific BCR, our data demonstrate that mCR2 can act as a specialized receptor for the presentation of Ag. Using distinct residues present in the cytoplasmic domain, CR2 allows Ags with a relatively low affinity to be presented by nonspecific B cells. This may be important at the initiation of an immune response where Ag-specific B cells are limiting. Previous reports demonstrating CR2-dependent, BCR-independent Ag presentation (27, 28) have been followed up by the demonstration that a feedback loop exists to prevent uncontrollable expansion and secretion of nonspecific Abs by nonspecific B cells (50). It is therefore possible that the cytoplasmic domain of CR2 may be involved in the initial phase of B cell Ag presentation. As mentioned previously, comparable studies have shown that the cytoplasmic domain of sIgG (containing a tyrosine residue essential for full Ag presentation BCR; Ref. 5) is required for increasing the B cell population size during secondary IgG response (47). This domain was subsequently shown to directly enhance BCR-mediated signaling by a novel mechanism (51). A comparison between the molecular associations with the cytoplasmic domains of these two receptors is currently under investigation.

We thank M. Steward, V. Cox, J. Bright, and R. A. G. Smith (Adprotech, Little Chesterford, U.K.) for providing the (C3d)₃ plasmid; C. Watts for providing the TTCF plasmid and peptides; V. M. Holers and H. Molina for the mCR2 cDNA; and T. Kinoshita for anti-mCR2 Abs. We are also grateful to D. Fearon for advice and to A. Antoniou, K. Brooks, and S. Powis for advice and comments on the manuscript.