|Home | About | Journals | Submit | Contact Us | Français|
The development of approaches for antigen delivery to the appropriate subcellular compartments of APCs and the optimization of antigen persistence are both of central relevance for the induction of protective immunity or tolerance. The expression of the Fc receptor, FcRn, in APCs and its localization to the endosomal system suggest that it might serve as a target for antigen delivery using engineered Fc fragment-epitope fusions. The impact of FcRn binding characteristics of an Fc fragment on in vivo persistence allows this property to also be modulated. We have therefore generated recombinant Fc (mouse IgG1-derived) fusions containing the N-terminal epitope of myelin basic protein that is associated with EAE in H-2u mice. The Fc fragments have distinct binding properties for FcRn that result in differences in intracellular trafficking and in vivo half-lives, allowing the impact of these characteristics on CD4+ T cell responses to be evaluated. To dissect the relative roles of FcRn and the ‘classical’ FcγRs in antigen delivery, analogous aglycosylated Fc-MBP fusions have been generated. We show that engineered Fc fragments with increased affinities for FcRn at pH 6.0–7.4 are more effective in delivering antigen to FcRn-expressing APCs in vitro relative to their lower affinity counterparts. However, higher affinity of the FcRn-Fc interaction at near neutral pH results in decreased in vivo persistence. The trade-off between improved FcRn targeting efficiency and lower half-life becomes apparent during analyses of T cell proliferative responses in mice, particularly when Fc-MBP fusions with both FcRn and FcγR binding activity are used.
The delivery of antigen to elicit protective immunity or tolerance represents an area of considerable interest for both vaccine development and the treatment of autoimmunity (1–4). For CD4+ T cell responses, a primary goal is to achieve efficient delivery of antigen to the site of peptide loading onto MHC Class II molecules, namely the endolysosomal system of APCs (5). However, there is an incomplete understanding as to how the intracellular trafficking pathways of an antigen impact presentation and how this can be modulated. How antigen persistence, which relates to intracellular trafficking, affects both qualitative and quantitative aspects of CD4+ T cell responses is also of fundamental importance for understanding the factors that regulate T cell mediated immunity. Towards addressing these issues, here we use an approach in which we exploit properties of the MHC Class I-related receptor, FcRn, to modulate the uptake/intracellular trafficking and in vivo half-life of antigen as intrinsic properties of the delivery vehicle.
Fc receptors that bind to the Fc region of IgG encompass the classical Fc receptors (FcγRs) and the neonatal Fc receptor, FcRn, that can be distinguished in several important ways. The FcγRs are signaling receptors that can transmit activating or inhibitory signals depending upon whether they associate with the ITAM containing Fcγ chain or have cytosolic ITIM motifs (6, 7). Conversely, the MHC Class I-related receptor, FcRn has no known signaling role and serves as an IgG transporter to maintain antibody levels in vivo (8–14). The expression patterns of FcγRs and FcRn also differ, since FcγRs are primarily expressed by cells of hematopoietic origin (15–17) whereas FcRn is ubiquitously present in cells of diverse origin such as endothelial and epithelial cells (11, 13, 18–20). However, both FcRn and FcγRs are expressed in professional APCs such as dendritic cells (DCs) and macrophages (17, 21–23). Although the role of FcγRs in antigen uptake and presentation is well documented (16, 24–26), there is very limited knowledge concerning a possible function for FcRn.
FcRn transports IgGs within and across cells, and the interaction properties of an IgG with FcRn are key determinants of its in vivo persistence (27–30). The binding of naturally occurring IgGs to FcRn is pH dependent, with relatively strong binding at pH 6.0 that becomes progressively weaker as pH 7.3–7.4 is approached (31–34). The model for FcRn-mediated transport of IgG is as follows: IgGs are taken into cells by fluid phase uptake and enter endosomes where the acidic pH is permissive for binding. IgG molecules that bind to FcRn are recycled or transcytosed, whereas those that do not interact enter lysosomes (35). By contrast with FcRn, in general FcγRs transport bound ligands in the form of immune complexes into degradative compartments that can be involved in antigen presentation within cells (16, 24, 25), although FcγRIIB-mediated antigen recycling has also been observed in DCs (36). The interaction sites for FcRn and FcγRs on IgG are distinct (37–40) and, unlike FcγR-IgG interactions, FcRn binding is not affected by removal of N-linked glycosylation on the CH2 domain (27, 38, 41). This allows the relative contributions of FcγRs and FcRn to functional effects to be evaluated.
The current study is directed towards evaluating a possible role for FcRn in antigen delivery and presentation. As a consequence of the function of FcRn in regulating IgG/Fc persistence, this also enables an analysis of the impact of antigen persistence on cognate CD4+ T cell responses. Central to our studies are a class of engineered IgGs that, relative to their wild type counterparts, bind to FcRn with increased affinities in the pH range 6.0–7.4 (29, 30, 42). Compared with wild type IgGs, we have shown previously that these antibodies accumulate to high levels in FcRn-expressing cells since they can be taken up by receptor-mediated endocytosis and are inefficiently released at the cell surface during exocytic events (30, 42, 43). Such IgGs can lower endogenous IgG levels in vivo by competing for FcRn binding. However, IgGs (or Abdegs, for antibodies that enhance IgG degradation) of this class also have short in vivo half-lives (29, 30). Here we show that in contrast to wild type IgGs that are recycled (35), these mutated variants enter lysosomes following uptake into cells. The accumulation of Abdegs in the endolysosomal pathway, together with the expression of FcRn in APCs (23), has prompted us to compare the CD4+ T cell stimulatory properties of engineered Fc fragments that differ in both intracellular trafficking behavior and in vivo persistence.
For use in our studies, we have generated Fc fusion proteins with distinct FcRn binding properties that are linked to the N-terminal epitope of myelin basic protein (MBP)3, MBP1–9. This peptide represents the immunodominant epitope of MBP in H-2u mice, and is associated with autoreactive CD4+ T cell responses that can culminate in experimental autoimmune encephalomyelitis (EAE) (44). Covalent Fc-epitope fusions have been used in preference to non-covalent antibody-antigen complexes to avoid possible complications of antigen/epitope dissociation. To assess the contribution of the classical FcγRs to the outcome, the activities of engineered Fc fusions that do, and do not, bind to this class of receptors have also been compared. Our studies demonstrate that targeting FcRn with Abdegs can enhance antigen delivery and presentation using in vitro assays for which antigen persistence does not play a role. This enhancement can also be revealed during analyses of T cell proliferative responses in vivo when binding to FcγRs is excluded. However, when FcγR competent fusions are used, the counterbalancing impact of reduced in vivo half-life of Abdegs becomes apparent and longer lived Fc-MBP fusions are more effective.
B10.PL (H-2u) mice were purchased from Jackson Laboratories (Bar Harbor, ME). Mice that transgenically express the 1934.4 T cell receptor [1934.4 tg mice; (45)] or clone 19 T cell receptor [T/R+ tg mice; (46)] were kindly provided by Dr. Hugh McDevitt (Stanford University, CA) and Dr. Juan Lafaille (New York University School of Medicine, NY), respectively. Both the 1934.4 and clone 19 T cell receptors are specific for MBP1–9 complexed with I-Au (45, 46) and have similar affinities and in vitro responsiveness (unpublished data). Transgenic (tg) mice were maintained by backcrossing onto B10.PL mice. All mice were housed and handled in pathogen-free animal facilities in compliance with institutional policies and guidelines and Institutional Animal Care and Use Committee-approved protocols. Male or female mice of 6–9 weeks old were used in experiments.
The MBP1–9:I-Au specific T hybridoma, 46, was generated from mice that transgenically express the β chain of the 172.10 T cell receptor (47) following immunization with 200 µg MBP1–9[4Y] (MBP1–9 with lysine at position 4 substituted by tyrosine) and methods as in (48). The I-Au- expressing B lymphoblastoid line, PL-8, and 1934.4 hybridoma (49) were generously provided by Dr. David Wraith (University of Bristol, Bristol, UK), and the 172.10 hybridoma (47) by Dr. Joan Goverman (University of Washington, Seattle). Stable transfectants of PL-8 cells expressing mouse FcRn tagged with GFP (‘PL-8:FcRn’) were generated by transfection of PL-8 cells with a mouse FcRn-GFP construct (42) followed by selection with G418 (600 µg/ml, Gibco, Grand Island, NY, USA).
A previously described construct encoding the wild type (WT) mouse hinge-Fc region (mouse IgG1 derived) tagged with a C-terminal polyhistidine tag (50) was modified to generate the following derivatives: The WT Fc-hinge (Fc-WT) gene was mutated using splicing by overlap extension (51) to insert the following mutations to generate Fc-mut: Thr252 to Tyr, Thr256 to Glu, His433 to Lys and Asn434 to Phe. Codons encoding Gly-Ser-Gly-Gly and the MBP1–9[4Y] peptide were inserted as overlapping oligonucleotides into a unique BstEII site at the 3’ end of the Fc-WT or Fc-mut genes (5’ to the polyhistidine tag) using standard methods of molecular biology. The Fc-WT gene was subsequently mutated by splicing by overlap extension to insert the H435A mutation, which has been shown previously to ablate binding of human IgG1 to FcRn (52). The resulting three constructs containing Fc-WT, Fc-mut and Fc-H435A linked to MBP1–9[4Y] were then further modified by replacing the prokaryotic pelB leader sequence (50) by the native mouse IgG1 leader peptide using overlapping oligonucleotides and the PCR. BamHI sites were appended at the 5’ and 3’ ends of the Fc-MBP fusion genes, and they were ligated into BamHI restricted pEF6/V5-His vector (Invitrogen, Carlsbad, CA). These constructs were further mutated using splicing by overlap extension to insert the N297A mutation. Fc-WT-MBP(3A6A) was made by inserting oligonucleotides encoding Gly-Ser-Gly-Gly and the MBP1–9[4Y] peptide with Gln3 to Ala, Pro6 to Ala as overlapping oligonucleotides into a unique BstEII site at the 3’ end of the Fc-WT gene. All constructs were sequenced prior to use in transfections of CHO cells. Stable transfectants were selected in CD CHO cell medium (Gibco, Grand Island, NY) containing 8 µg/ml blasticidin (Invitrogen, Carlsbad, CA).
The Fc-MBP fusions were purified from culture supernatants using Ni2+-NTA-agarose columns and previously described methods (50). Purified Fc-MBP fusions were labeled with Alexa 647 using Alexa Fluor 647 succimidyl ester (Molecular Probes, Eugene, OR) and methods recommended by the manufacturer.
Equilibrium dissociation constants of the Fc-MBP fusions for mouse FcRn were determined using surface plasmon resonance and a BIAcore 2000 as described previously (53, 54). Fc-MBP fusions were immobilized by amine coupling chemistry to a density of ~500 RU. Recombinant mouse FcRn was purified from baculovirus-infected High Five cells as in (53), and was used as analyte in PBS plus 0.01% Tween pH 6.0 or 7.4. IgG or Fc has two possible interaction sites, and equilibrium binding data were fitted as in (54). The dissociation constants for the higher affinity interaction sites are presented.
Soluble, recombinant MBP peptide:I-Au (MBP1–9[4Y]:I-Au) complexes were generated and purified using baculovirus-infected High Five cells, and multimeric complexes (‘tetramers’) generated using PE-labeled Extravidin (Sigma-Aldrich, St. Louis, MO) as described in (55).
Single cell suspensions were obtained from homogenized spleens or regional lymph nodes (inguinal and intestinal lymph nodes) as described previously (56). The following fluorescently labeled antibodies were purchased from BD Biosciences (San Jose, CA) and used for flow cytometry: FITC-labeled anti-CD11b (M1/70), PE-labeled anti-CD4 (GK1.5), Vβ8 TCR (F23.1), B220 (RA3–6B2), CD11c (HL3), PerCP-labeled anti-CD4 (RM4–5), Gr-1 (RB6–8C5), B220 (RA3–6B2) and APC-labeled anti-CD4 (RM4–5), CD62L (MEL-14), CD19 (1D3). PE-labeled anti-F4/80 (BM8) was purchased from Invitrogen (Carlsbad, CA). Anti-Vβ8 (F23.1) antibody was purified from culture supernatants of the F23.1 hybridoma using protein G-Sepharose and fluorescently labeled with Alexa Fluor 647 succimidyl ester according to the manufacturer’s instructions. Anti-I-Au antibody was purified from the 10.2.16 hybridoma and labeled with fluorescein-isothiocyanate using standard methods. Cells were treated with fluorescently labeled tetramers or antibodies as described in (56). Flow cytometry analyses were performed using a FACSCalibur (BD Biosciences, San Jose, CA) and data analyzed using FlowJo (Tree Star, Ashland, OR).
PL-8, PL-8:FcRn cells or splenocytes were pulsed with 2–5 µg/ml Alexa 647-labeled Fc-MBP fusions in pre-warmed phenol red free, IgG-depleted cDMEM [DMEM (Lonza, Walkersville, MD) supplemented with 10% IgG-depleted FCS, 100 µM non-essential amino acids, 1 mM sodium pyruvate, 55 µM β-mercaptoethanol, 10 mM HEPES pH 7.2–7.5] at 37°C for 20–30 min, washed or chased following the washes in prewarmed medium at 37°C for 30 min as described in (30, 42). Following these treatments with Alexa 647-labeled Fc-MBP fusions, splenocytes were incubated on ice with fluorescently labeled antibodies to identify cell populations as follows: macrophages (CD11bint+F4/80+), mDCs (CD11b+CD11c+B220−), and B cells (B220+I-Au+). Labeled cells were analyzed by flow cytometry as above. The role of FcγRs in uptake of Fc-MBP fusions was determined by incubating PL-8, PL-8:FcRn cells or splenocytes with 5 µg/ml anti-FcγRIIB/III antibody (2.4G2) or isotype matched rat IgG2b at 4°C for 15 min prior to addition of Fc-MBP fusions.
Single cell suspensions derived from splenocytes were stained with PE-labeled anti-B220 (RA3–6B2) and APC-labeled anti-CD19 (1D3) and B cells sorted by FACS using a MoFlo (Beckman Coulter). Total RNA was isolated from cells using RNA-Bee (Tel-Test, Friendswood, TX) and standard methods. cDNA synthesis was carried out using two ‘forward’ oligonucleotide primers complementary to the 3’ ends of the FcRn (57) and β2-microglobulin genes (FcRnfor: 5’AGA AGT GGC TGG AAA GGC ATT TGC ACC 3’; β2mfor, 5’ CAT GTC TCG ATC CCA GTA GAC GGT CTT 3’). cDNA was used in PCRs with corresponding ‘Back’ primers (FcRnback, anneals to bases 679–705, 5’TAC CCA CCG GAG CTC AAG TTT CGA TTC 3’; β2mback, anneals to bases 1–27, 5’ATG GCT CGC TCG GTG ACC CTG GTC TTT 3’). PCRs were run for 30 cycles under standard conditions and the 415 bp PCR product corresponding to the 3’ end of the FcRn α-chain gene was gel-purified and sequenced.
PL-8:FcRn cells were seeded in 24 well plates containing coverslips (Fisherbrand 1.5; Fisher Scientific, Houston, TX) overnight. PL-8:FcRn cells were incubated with 500 µg/ml Alexa 555-labeled dextran (10,000 MW, Invitrogen, Carlsbad, CA) in phenol red free, IgG-depleted cDMEM for 2 h at 37°C. Cells were then washed and chased in medium for 1 h. Fc-MBP fusions (5 µg/ml in medium) were subsequently added and cells were incubated at 37°C for different times. Cells were washed, fixed with 3.4% paraformaldehyde and mounted as in (35). Cells were imaged using two microscopy setups for Figure 4A and 4B, respectively: a Zeiss (Thornwood, NY) Axiovert 200M inverted fluorescence microscope with a Zeiss 1.4 NA 100× Plan-APOCHROMAT objective and a Zeiss 1.6× Optovar as described in (42); a Zeiss Axiovert S100TV inverted fluorescence microscope with a Zeiss 1.4 NA 100× Plan-APOCHROMAT objective and a Zeiss 1.6× Optovar as described in (35, 43). All data were processed and displayed using the custom-written Microscopy Image Analysis Tool (MIATool) software package (www4.utsouthwestern.edu/wardlab) in MATLAB (Mathworks, Natick, MA). The intensities of acquired data were linearly adjusted. Images were overlaid and annotated. In overlay images, the intensities of the individual color channels were adjusted to similar levels.
Different concentrations of Fc-MBP fusions were added to 96 well plates containing PL-8 or PL-8:FcRn cells (5×104 cells/well) and MBP1–9:I-Au specific T hybridoma cells (5×104 cells/well) or to 96 well plates containing splenocytes (3×105 cells/well) derived from 1934.4 tg mice (45) as in (48, 56). IL-2 levels in culture supernatants and proliferative responses ([3H]thymidine incorporation) were assessed as in (48, 56). Cultures were also set up in the presence of 5 µg/ml anti-FcγRIIB/III antibody (2.4G2) or isotype matched control antibody (rat IgG2b) to assess FcγR-mediated effects.
Splenocytes from 1934.4 tg mice were isolated and labeled with CFSE using previously described methods (56). The activation status of the CD4 T cells was assessed using anti-CD62L antibody and flow cytometry, and the percentage of CD62Llo cells was less than 11% in all experiments. 2×106 CD4 T cells (as mixtures in splenocytes) were transferred into B10.PL mice and subsequently 100 ng of each Fc-MBP fusion was injected intravenously. Three days later, splenocytes and regional LNs were harvested and analyzed by flow cytometry using fluorescently labeled anti-CD4, anti-Vβ8 (F23.1) antibodies and PE-labeled MBP1–9[4Y]:I-Au tetramers.
To examine the longevity of functional antigen:I-Au complexes derived from Fc-MBP fusions in vivo, recipient B10.PL mice were injected intravenously with each Fc-MBP fusion (1 µg/mouse), and the mice were adoptively transferred with 2×106 CD4+ T cells derived from 1934.4 tg mice, or combined from 1934.4 tg and T/R+ tg mice, intravenously at 1 h, 3 and 7 days following Fc-MBP fusion treatment. Injections of Fc-MBP fusions were staggered over a 7 day period, so that all recipient mice were transferred with the same population of antigen specific T cells. Three days following transfer, splenocytes and LNs were harvested and analyzed by flow cytometry as above.
Tests for statistical significance were carried out using the statistical toolbox of Matlab (Mathworks Inc., Natick, MA) and the function ‘multcompare’.
Several different Fc-fusion proteins that vary in the Fc region (mouse IgG1-derived) sequence were expressed for use in these studies (Table I). The wild type Fc (‘Fc-WT’) was mutated to generate the TT-HN mutant [‘Fc-mut’; Thr252 to Tyr, Thr256 to Glu, His433 to Lys and Asn434 to Phe; analogous to a previously described human IgG1 mutant, or Abdeg (42)] and H435A mutant (His435 to Ala). These mutants have different binding properties for mouse FcRn: Fc-mut has markedly increased affinity relative to Fc-WT at both pH 6.0 and 7.4 and H435A shows undetectable binding (Table I). Fc fragments were linked through their C-termini via a Gly-Ser-Gly-Gly sequence to the MBP1–9[4Y] peptide to generate Fc-WT-MBP, Fc-mut-MBP and Fc-H435A–MBP. The peptide variant MBP1–9[4Y] was used in preference to wild type MBP1–9, since the position 4 substitution of lysine by tyrosine results in higher affinity binding to I-Au (58, 59). Glycine was also appended to the N-terminus of the MBP peptide to mimick the N-terminal acetyl group that is necessary for T cell recognition (55, 60).
The Fc-MBP fusions were expressed in both glycosylated and aglycosylated forms in transfected CHO cells. Aglycosylation was achieved by mutation of Asn297 to Ala (N297A), which ablates binding to the classical FcγRs (61, 62) without affecting FcRn interactions (27, 38, 41). Mouse Fc (or IgG1) binds to the low affinity FcγRIIB and FcγRIII, but not to the high affinity FcγRI nor to the recently described FcγRIV (63). In other systems, monomeric IgG-epitope fusions have been shown to have effects through binding to low affinity FcγRs (64, 65). In addition, we cannot exclude the possibility that our recombinant proteins contain low levels of aggregates which are more effective than monomers in mediating signaling through low affinity FcγRs. In the current study we have therefore compared the effects of glycosylated and aglycosylated Fc-MBP fusions.
To analyze the impact of FcRn expression on T cell stimulation in vitro, we have used an I-Au expressing B lymphoblastoid cell line, PL-8 (66), and FcRn transfected PL-8 cells (PL-8:FcRn) as APCs. PL-8 cells were used since, by analogy with other in vitro B cell lines (8, 23), they do not express FcRn (our unpublished observations). Comparison of antigen presentation by PL-8 and PL-8:FcRn cells therefore allows the impact of FcRn expression to be assessed. In vitro stimulation of the MBP1–9:I-Au specific hybridoma, 46, with different Fc-MBP fusions results in distinct patterns of stimulation by the two types of APCs (Figure 1A, B). The T cell receptor of hybridoma 46 is the same as that of the 172.10 hybridoma (47), with the exception of an Ala98 to Glu change in CDR3α (our unpublished data). In PL-8 cells, the stimulatory capacities of glycosylated Fc-WT-MBP, Fc-mut-MBP and Fc-H435A–MBP are similar (Figure 1A). However, this changes when the cells express FcRn, with the stimulatory activity decreasing in the order Fc-mut-MBP > Fc-H435A–MBP > Fc-WT-MBP (Figure 1B). Similar results were observed for the 1934.4 (49) and 172.10 (47) hybridomas that are both specific for MBP1–9:I-Au complexes (data not shown). Importantly, mutation of the T cell contact residues Gln3 and Pro6 in the MBP epitope (67) to alanine [‘MBP(3A6A)’] results in loss of T cell stimulation, demonstrating the specificity of the T cell response (Figure 1C, D). Increasing the affinity of an Fc-MBP fusion for FcRn in the pH range 6.0–7.4 therefore enhances its ability to stimulate antigen specific T cells when the APC expresses this Fc receptor. By contrast, the properties of Fc-WT most likely result in FcRn-mediated recycling (35, 43) of Fc-WT-MBP out of PL-8:FcRn cells. The stimulatory activities of glycosylated or aglycosylated Fc-H435A–MBP fall between those of Fc-mut-MBP and Fc-WT-MBP since these fusions do not accumulate via FcRn-mediated uptake and are not recycled out by FcRn-mediated processes.
In general, the glycosylated proteins were more stimulatory relative to the aglycosylated proteins when either PL-8 or PL-8:FcRn cells were used as APCs (Figure 1A, B). Since glycosylated, but not aglycosylated, proteins bind to FcγRs (61, 62), this suggested that FcγR-interactions might be responsible for the increased activity. Mouse IgG1 can bind to FcγRIIB and FcγRIII but not to the high affinity FcγRI, nor FcγRIV (63). We therefore preincubated APCs with the anti-FcγR antibody, 2.4G2, which binds to FcγRIIB, FcγRIII and FcγRIV (15). Pretreatment with 2.4G2 reduced the levels of stimulation by the glycosylated proteins to those observed for their aglycosylated counterparts (Figure 2), indicating that the differences are due to FcγR binding. Using RT-PCR, we observed that PL-8 cells express both FcγRIIB and FcγRIII (data not shown), indicating that one or both of these receptors are responsible for the enhanced activity of glycosylated Fc-MBP fusions.
We hypothesized that the increased stimulatory activity of Fc-mut-MBP relative to Fc-WT-MBP is due, at least in part, to increased accumulation in APCs. The uptake and retention of fusion proteins containing Fc-mut or Fc-WT by PL-8 and PL-8:FcRn cells in medium at 37°C were therefore compared (Figure 3). Alexa 647-labeled Fc-MBP fusions were added at concentrations of 2 µg/ml (~33 nM) so that fluid phase uptake was low (42), allowing an assessment of enhanced uptake by receptor-mediated processes to be made. In some experiments, cells were also chased for 30 minutes after the pulse to analyze the retention of the recombinant proteins in cells (Figure 3B). Consistent with their high affinities for FcRn at pH 6.0 and 7.4 (Table I), both glycosylated and aglycosylated Fc-mut-MBP fusions are taken up more efficiently by FcRn expressing cells (PL-8:FcRn) relative to PL-8 cells (Figure 3A). The levels of Fc-mut fusions associated with PL-8:FcRn cells do not change over a 30 minute chase period (Figure 3B; data not shown), indicating that these proteins are not recycled following uptake into cells. In addition, the uptake of Fc-mut-MBP fusions in PL-8:FcRn cells was much greater than that of the corresponding Fc-WT-MBP fusions that do not bind detectably to FcRn at pH > 7 (Figure 3A,C). Importantly, these differences were observed in cases where the degrees of labeling (molar ratio of Alexa 647 dye to Fc-MBP fusion) of the Fc-mut-MBP fusions were lower than those for the corresponding Fc-WT-MBP proteins. The uptake of the glycosylated proteins was reduced slightly by preincubating the cells with anti-FcγRIIB/III (2.4G2) antibody, whereas FcγR blockade did not affect the accumulation of aglycosylated fusions (Figure 3C).
Fluorescence microscopy analyses were carried out to determine the subcellular location of internalized Fc-mut-MBP over the course of 24 h (Figure 4). These studies demonstrated that following one hour of incubation, glycosylated and aglycosylated Fc-mut-MBP proteins were extensively colocalized with FcRn within PL-8:FcRn cells (Figure 4A and data not shown). By contrast, the level of Fc-WT-MBP in cells was below the level of detection under the imaging conditions used (Figure 4A). Following several hours of incubation, both glycosylated and aglycosylated Fc-mut-MBP proteins could be detected within lysosomes, and concomitant with a decrease in FcRn colocalization in endosomes, this lysosomal distribution increased up until 24 h post-addition (the last time point of imaging) (Figure 4B). Recent studies have shown that FcRn can enter lysosomes in an invariant chain-dependent way in APCs (68), and/or on the constitutive degradation pathway (Z.G., E.S.W., unpublished observations). Thus, although Fc-mut-MBP fusions are most likely associated with FcRn during entry into lysosomes, FcRn-GFP cannot be detected in these compartments due to the sensitivity of enhanced GFP fluorescence to acidic pH, combined with the susceptibility of GFP to proteolytic degradation (69–71).
The ability of the different Fc-MBP proteins to stimulate antigen specific, naïve T cells derived from mice that transgenically express the 1934.4 T cell receptor [1934.4 tg mice; (45)] was also investigated. Proliferation and IL-2 production in splenocyte cultures derived from these mice were assessed following addition of different concentrations of Fc-MBP fusions (Figure 5A, B). The pattern of stimulation was similar to that observed for PL-8:FcRn cells and T cell hybridomas (Figure 1). Since splenocytes contain distinct APC subsets, we used flow cytometry to analyze the uptake of the Fc-MBP fusions by myeloid DCs (mDCs), macrophages and B cells in freshly isolated splenocytes (Figure 6A). Cells were pulsed with Alexa 647 labeled Fc-MBP fusions for 20 minutes in medium at 37°C, washed and analyzed. These studies demonstrated that Fc-MBP fusions containing Fc-mut accumulate to higher levels in all APC subsets relative to their Fc-WT counterparts (Figure 6A), with the uptake by macrophages being higher than that for DCs and B cells. By contrast, CD4+ T cells which do not express FcRn nor FcγRs (8, 17, 72) show similar, very low levels of (fluid phase) uptake of Fc-mut or Fc-WT fusions that are close to cellular autofluorescence levels (Figure 6A). Although in earlier studies FcRn has been reported to not be expressed in primary B cells and B cell lines (8, 23, 72), RT-PCR analyses indicated the presence of FcRn in sorted primary B cells from B10.PL mice (Figure 6B). In addition, the presence of anti-FcγR antibody (2.4G2) reduced the uptake of glycosylated Fc-MBP fusions, whereas this blockade did not affect the accumulation of aglycosylated Fc-MBP fusions (Figure 6A).
We also compared the uptake of Fc-WT-MBP and Fc-H435A–MBP fusions in different splenic APC subsets (Figure 6C). The levels of cell-associated Fc-WT-MBP fusions were consistently lower than those of Fc-H435A–MBP fusions, and the addition of anti-FcγR antibody reduced the uptake of the glycosylated proteins but did not affect that of the aglycosylated counterparts. Importantly, the degree of labeling of the Fc-H435A–MBP fusions was lower than that of the corresponding Fc-WT-MBP fusions, indicating that the lower levels of cell-associated Fc-WT-MBP fusions could not be accounted for by differences in fluorescent labeling (Figure 6C). The lower accumulation of Fc-WT-MBP fusions in cells is consistent with FcRn-mediated recycling of WT Fc fragments (or IgGs) out of cells (35). By contrast, Fc-H435A–MBP fusions are not salvaged by FcRn into the recycling pathway following uptake (35).
We next analyzed the effects of the Fc-MBP fusions on T cell proliferative responses in vivo. CFSE-labeled MBP1–9:I-Au specific T cells derived from 1934.4 tg mice (45) were transferred into B10.PL recipients, followed by the i.v. injection of relatively low doses (100 ng, ~2 pmole) of glycosylated or aglycosylated proteins. Importantly, delivery of an amount of MBP1–9[4Y] peptide that is equivalent to that present in the injected Fc-MBP fusions (on a molar basis) induced no detectable T cell proliferation (Figure 7A). Further, the antigen specificity of the proliferative response was demonstrated by using an Fc-WT-MBP fusion (‘3A6A’) in which the T cell contact residues in the MBP epitope had been mutated (Gln3 to Ala, Pro6 to Ala (67)) (Figure 7A).
Comparison of the properties of the aglycosylated forms of Fc-WT-MBP, Fc-mut-MBP and Fc-H435A–MBP that do not interact with FcγRs resulted in a similar ranking of stimulatory capacities to that observed in vitro, with aglycosylated Fc-mut-MBP being more effective than fusions containing Fc-WT or Fc-H435A (Figure 7A, B). By contrast, however, analyses of the glycosylated proteins resulted in distinct behavior to that observed in vitro: glycosylated Fc-WT-MBP induced substantially more antigen specific T cell expansion relative to Fc-mut-MBP, with Fc-H435A–MBP inducing an intermediate level of proliferation (Figure 7A, C).
These observations raise the question as to why the relative activities of the glycosylated Fc-MBP fusions in stimulating T cells are different in vivo and in vitro. An obvious distinction between these analyses is that the half-life and distribution of Fc-MBP fusions will have an impact during in vivo studies. The different binding properties of Fc-WT-MBP, Fc-mut-MBP and Fc-H435A–MBP for FcRn are expected to result in variations in longevity in vivo (27–30). We therefore analyzed the persistence of functional antigen derived from each type of protein by determining the proliferative responses of CFSE-labeled, antigen specific T cells that were transferred into B10.PL recipients at different times following delivery of the Fc-MBP fusions. These studies indicated that cognate antigen derived from Fc-WT-MBP persisted for longer than that generated from Fc-mut-MBP or Fc-H435A–MBP, and this was observed for both glycosylated and aglycosylated proteins (Figure 8). In addition, Fc-MBP fusions containing Fc-mut showed a trend towards more rapid clearance than those comprising Fc-H435A, although the differences were not all significant (at a 95% confidence level). This is consistent with our earlier data indicating that loss of pH dependent binding to FcRn results in enhanced clearance (29, 30).
The expression of the MHC Class I-related receptor, FcRn, in professional APCs (23) raises questions concerning its functional relevance. In addition, the location of this receptor in the endosomal pathway (35) suggests that it might be a useful target for enhancing antigen delivery. In the current study we have therefore analyzed the effect of targeting antigen to the FcRn trafficking pathway on cognate T cell responses. This has been achieved by using recombinant Fc-MBP fusions that differ in their binding properties for FcRn. In addition to a wild type, mouse IgG1-derived Fc fragment, we have used a mutated variant (Fc-mut) that binds with substantially increased affinity for FcRn in the pH range 6.0–7.4 and an Fc mutant (H435A) that does not bind detectably to FcRn. These Fc fragments not only have different FcRn targeting abilities, but also have distinct in vivo half-lives, allowing the impact of antigen persistence on T cell responses to be evaluated. In addition, to probe the relative contributions of FcRn and FcγRs to antigen delivery, the Fc fragments have been used in both glycosylated and aglycosylated forms that do and do not, respectively, interact with FcγRs.
In vitro, Fc-mut-MBP fusions are more effective than their counterparts containing Fc-WT or Fc-H435A in stimulating antigen specific T cells when FcRn-expressing APCs are used. This difference is observed with both glycosylated and aglycosylated proteins. These observations can be explained by the high efficiency of uptake of Fc-mut by receptor-mediated processes together with accumulation in the endolysosomal system of FcRn expressing cells [this study; (42)]. By contrast with the in vitro analyses, however, glycosylated Fc-WT-MBP is more effective compared with its counterpart containing Fc-mut in inducing the proliferation of transferred antigen specific T cells in mice. The higher in vivo stimulatory capacity of glycosylated Fc-WT-MBP relative to Fc-mut-MBP can be attributed to several factors: First, delivery of the Fc-WT-MBP fusion in vivo results in greater persistence of antigen (peptide:I-Au) complexes relative to Fc-mut-MBP, which consistent with its FcRn interaction properties is cleared rapidly (29, 30). Second, FcRn is expressed in multiple other cell types, including those of endothelial and epithelial origin (11, 13, 18–20). This is expected to result in dilution of the effective concentration that is available in vivo for uptake of Fc-mut-MBP by APCs such as DCs. The enhancement of in vivo proliferative responses by FcγR interactions with the Fc-MBP fusions is revealed by analyses of their aglycosylated counterparts which show that Fc-WT-MBP and Fc-H435A–MBP, in particular, are markedly reduced in activity when binding to FcγRs is ablated. Consequently, when aglycosylated Fc-MBP fusions are delivered at a relatively low dose of ~2 pmole/mouse, the increase in uptake into (antigen presenting) cells due to high affinity binding to FcRn by Fc-mut-MBP becomes apparent. This results in greater T cell expansion for aglycosylated Fc-mut-MBP relative to analogous Fc-MBP fusions containing Fc-WT or Fc-H435A that both rely on fluid phase uptake for entry into cells.
The shorter in vivo persistence of Fc-mut illustrates the trade-off that is observed when protein engineering is used to generate Fc fragments (or IgGs) that can efficiently target FcRn. Effective uptake and retention in FcRn+ cells requires substantial increases in affinity of an Fc fragment for FcRn binding in the pH range 6.0–7.4, and such Fc fragments/IgGs (or Abdegs) can compete with endogenous IgGs for FcRn binding (30, 42). However, loss of the marked pH dependence of an FcRn-Fc interaction results in reduced in vivo persistence (29, 30, 42). The short in vivo half-lives of such IgGs/Fc fragments most likely result from retention within cells due to reduced exocytic release, combined with their FcRn-directed trafficking during the constitutive degradation of this Fc receptor and/or invariant chain-directed delivery of FcRn to lysosomes [(68), Z.G., E.S.W., unpublished]. Consequently, Fc fragments that are potent, FcRn-mediated antigen delivery reagents have short in vivo half-lives. The dual and opposing effects of increasing FcRn binding vs. in vivo half-life become apparent during analyses in mice when Fc fusions that can bind to FcγRs are used, with the consequence that in vivo persistence is a key determinant of T cell proliferative responses. However, when FcγR binding is ablated, FcRn targeting by an Abdeg becomes dominant and, at the relatively low doses used in this study, the short-lived Fc-mut-MBP is the most effective protein in inducing proliferation in vivo. The role of FcγRs in enhancing immune responses is complex and can occur through both increased antigen uptake and activation of DCs (21, 73, 74). Given that we are using mouse IgG1-derived Fc fragments that in the steady (non-inflammatory) state would be expected to result in a bias towards inhibitory signaling (75), the enhancement of proliferation in response to glycosylated Fc-MBP fusions is expected to be primarily due to increased internalization into cells rather than signaling effects through FcγRs.
The use of I-Au expressing B lymphoblastoid (PL-8) cells that are transfected to express mouse FcRn provides us with a useful system to assess the impact of FcRn expression on antigen presentation. However, DCs are the initiators of immune responses in vivo (76), and there are differences between antigen presentation, processing and trafficking in B cells and DCs (71, 77). In addition, PL-8 cells do not express endogenous FcRn. Importantly, the similar ranking of activities of different Fc-MBP fusions in in vitro antigen presentation assays using splenocytes or PL-8:FcRn cells indicate that our data with PL-8 cells can be correlated with behavior in APCs that have endogenous FcRn expression. However, we cannot exclude the possibility that there might be differences in intracellular trafficking that are not detectable in our assays.
Our data also have relevance to understanding how FcRn might impact antigen presentation pathways when naturally occurring antibodies are bound to antigen. Although we cannot extrapolate the results from the current study to immune complexes, our observations suggest that monomeric IgG-antigen complexes can evade the degradative compartments that play an important role in antigen presentation by being recycled out of FcRn-expressing cells. Specifically, we observe that Fc-WT-MBP fusions accumulate to lower levels in cells and are less effective in stimulating cognate T cells in vitro with FcRn expressing APCs than their counterparts containing H435A that do not bind to FcRn. The behavior of monomeric IgG-antigen complexes contrasts with that described for immune complexes in a recent study (72), in which such complexes were directed into lysosomes following uptake into cells. These observations indicate that FcRn crosslinking enhances lysosomal delivery. In this context, some epitopes can be processed and loaded onto MHC Class II molecules during endosomal recycling (78–82). For such epitopes, FcRn-mediated ‘diversion’ of monomeric IgG-antigen complexes away from degradation and into the recycling pathway might therefore not reduce antigen presentation unless FcRn binding sterically inhibits antigen degradation/loading onto MHC Class II. The impact of FcRn will therefore depend on the processing requirements of the specific epitope and whether it remains stably bound to cognate IgG at the slightly acidic pH of the endosomal recycling system. Although MBP-derived epitopes have been shown to be loaded onto recycling HLA-DR molecules (78), our current data indicating that FcRn-mediated recycling of Fc-WT-MBP fusions can decrease antigen presentation in vitro suggest that this is not the major pathway for the loading of MBP1–9[4Y] onto I-Au. How such recycling impacts in vivo T cell responses is made more complex by the interplay of the influence of FcRn binding properties on intracellular trafficking pathways and in vivo persistence.
An unexpected outcome of the current study is that, although FcRn has been shown to not be expressed by in vitro B cell lines of both human and mouse origin (8, 23), this receptor is present in splenic B cells isolated directly ex vivo. FcRn expression can therefore be extended to this class of professional APCs, and this raises questions concerning its function. In addition to providing an additional depot of hematopoietic cells that might regulate IgG levels in vivo (83), it is possible that this receptor performs functions in this cell type that are related to antigen presentation and/or the intracellular trafficking of IgG. However, given the central role of DCs in initiating immune responses (76), it is likely that these cells are the relevant APCs in our short term proliferation assays in vivo.
Taken together, we demonstrate that targeting FcRn with high affinity, engineered Fc-MBP fusions can elicit T cell responses both in vitro and in vivo. This broadens the previously defined roles of FcRn to encompass a function in the delivery of antigen to endolysosomal compartments in APCs. FcRn might therefore be a useful target for antigen loading, particularly when FcγR-mediated effects are to be avoided to minimize DC activation in the steady state (2, 84). Future studies will be directed towards understanding the factors that lead to immune activation versus tolerance induction using these targeting approaches.
We thank Paula Marcos Mondejar and Tuyetanh Nguyen for assistance with the generation and analysis of recombinant proteins. We are grateful to Hector Perez Montoyo and Carlos Vaccaro for expert assistance, and to Silvia Pastor for providing hybridoma 46. We are also indebted to Dr. Joan Reisch and Angie Mobley for assistance with statistical analyses and FACS, respectively.
1This work was supported in part by grants from the National Multiple Sclerosis Society (RG 2411) and National Institutes of Health (AI/NS ROI 42949).
3Abbreviations used in this paper: MBP, myelin basic protein; EAE, experimental autoimmune encephalomyelitis; Abdegs, antibodies that enhance IgG degradation; CHO, Chinese hamster ovary; MFI, mean fluorescence intensity; mDC, myeloid dendritic cell.