Intercellular Transfer of MHC and Immunological Molecules: Molecular Mechanisms and Biological Significance

Abstract

The intercellular transfer of many molecules, including the major histocompatibility complexes (MHC), both class I and II, costimulatory and adhesion molecules, extracellular matrix organization molecules as well as chemokine, viral and complement receptors, has been observed between cells of the immune system. In this review, we aim to summarize the findings of a large body of work, highlight the molecules transferred and how this is achieved, as well as the cells capable of acquiring molecules from other cells. Although a physiological role for this phenomenon has yet to be established we suggest that the exchange of molecules between cells may influence the immune system with respect to immune amplification as well as regulation and tolerance. We will discuss why this may be the case and highlight the influence intercellular transfer of MHC molecules may have on allorecognition and graft rejection.

MHC Acquisition

Intercellular exchange of major histocompatibility complexes (MHC) molecules has been reported between many cells, including professional and nonprofessional antigen presenting cells (APCs) (Table 1). For example, CD4+ T cells can acquire MHC molecules from other CD4+ T cells as well as APCs (1) and dendritic cells (DCs) quire MHC molecules from other DCs and endothelial cells (EC)(2). This exchange of molecules can either be a unidirectional event as is the case between ECs and DCs or a bidirectional event as is the case between DCs and between DCs and T cells (2,3).

Table 1.

Molecular transfer of MHC class I and class II

Molecule transferred	Donor cell	Recipient cell	Reference
MHC/peptide  complex	APCs	T cells	(4)
(class I and  class II)	APCs	Splenic T cells	(50)
	APCs	Splenic T cells	(50)
	T cells	CD4+ T cells	(5,8,17,44,68,69)
	DCs,APCs	CD8+ T cells	(14,52,70,71)
	DCs,ECs, B cells	DCs (splenic,  follicular, bone  marrow derived)	(2,3,6,9,26,49,72)
	APCs	Thymocytes	(73)
	T cells	B cells	(3)
	APCs	NK cells	(74)
	PBMCs	Polymorphonuclear  neutrophils	(10,11)

Frelinger et al. (1974) published one of the earliest studies detailing MHC acquisition (4). These authors observed the presence of class II MHC molecules on mouse T cells, which are normally devoid of class II MHC, following incubation with MHC class II positive APCs. This initial observation has been confirmed and extended both in vitro and more importantly in vivo, by us and others (1,2,5). Transferred MHC is observed as early as 18 h in vivo (5) and the length of time transferred MHC remains stably expressed on recipient cells appears to be up to two days (6,7).

We have also observed that murine DCs, when cocultured with allogenic DCs or endothelial cells in vitro, can acquire substantial levels of class I MHC:peptide complexes in a temperature- and energy-dependent manner, although transfer is less efficient than that of class II MHC molecules. Importantly we have also reported that acquired class I and class II MHC molecules are functional. Acquired foreign MHC-peptide complexes induced proliferation of allospecific CD4+ and CD8+ T cells in vitro, respectively (2).

Although initial experiments looking at transfer of MHC class I and II were done using whole cells an intact cell membrane is not necessary for this phenomenon to occur. DCs have the ability to acquire preformed class II or class I MHC molecules from tumor cells in which membrane disruption, either through freeze-thaw cycles or osmotic lysis, had occurred (6,7).

Molecular transfer of MHC molecules is not restricted to mouse cells. Indeed, HLA molecules can be acquired by human APCs (8,9) and MHC transfer has also been observed using rat (3) and bovine cells (10,11).

Other Plasma Membrane Molecules Transferred

Major histocompatibility complexes complexes are not the only cell surface molecules transferred between cells. Transfer of many plasma membrane molecules ranging from the T-cell receptor (TCR) to costimulatory molecules, chemokine receptors, adhesion molecules, complement receptors, extracellular matrix organization molecules, tetraspan molecules and viral receptors has been described (Table 2).

Table 2.

Molecular transfer of costimulation, adhesion, extracellular matrix organization molecules as well as viral, chemokine and complement receptors to various recipient cells

Molecule transferred	Donor cell	Recipient cell	Reference
Costimulatory molecules
CD80 (B7.1)	DCs, transfected fibroblasts	CD4+ T cells	(8,75,76)
	APCs	CD8+ T cells	(42,47)
CD86 (B7.2)	APCs	CD4+ T cells	(8)
	APCs, transfected fibroblasts	CD8+ T cells	(47)
OX40L	ECs and monocytes	CD4+ T cells	(46)
Adhesion molecules
CD54 (ICAM-1)	ECs	CD4+ T cells	(77)
	APCs	CD8+ T cells	(42,47)
CD62E	ECs	CD4+ T cells	(77)
CD31	ECs	CD4+ T cells	(77)
Chemokine receptors
CCR5	PBMCs	ECs and monocytes	(78)
Extracellular matrix
organization molecules
CD49d	ECs	CD4+ T cells	(77)
CD61	ECs	CD4+ T cells	(77)
Viral receptors
CD21	EBV infected B cells	NK cells	(15)
Other
TCR/CD3	T cells	B cells	(3,30)
IgM	Daudi B cells	γδT cells	(16)
CD45RA	B cells	T cells	(3)

How Molecules Are Transferred

i) Direct contact

Harashyne et al. (2001, 2003) have reported that DCs can acquire plasma membrane from living cells. Coculturing fluorescently labeled DCs cells with unlabeled cells, for a period of time, revealed membrane exchange between living cells. This intercellular exchange of information was extremely efficient with 43%–66% of unlabeled DCs acquiring fluorescence within 4 h of mixing with labeled cells (12,13). Using live cell fluorescence microscopy, these authors reported that the plasma membrane was physically pulled from one DC to the other during coculturing conditions (12). This phenomenon of ‘nibbling’ is a receptor-mediated process as stripping the plasma membrane of scavenger receptors using a protease treatment resulted in loss of fluorescence uptake (13). Using a similar in vitro coculturing method, using either mouse or human cells, we have also demonstrated that MHC transfer between DCs as well as ECs and DCs requires close cell–cell contact. No transfer was observed when MHC donor or recipient cells were separated using a trans-well system (2,8).

Nibbling might not be the only mechanism by which intercellular transfer occurs. The concept of synaptically captured molecules has been described for a number of cells (14-16). Huang et al. (1999) published that CD8+ T cells can capture fluorescently labeled MHC class I molecules from a variety of APCs. Mixing APC expressing fluorescently labeled MHC class I (L^d) with CD8+ T cells (2C) resulted in clustering of labeled MHC at the+ contact site between the T cells and APC in the presence of an agonist peptide. Within 30 min of coculture small clusters of fluorescent MHC class I molecules were observed on the T cells at sites distal to the contact site. This acquisition was dependent on the strength of interaction between APC and T cell (14). Using live cell imaging to observe intercellular transfer of GFP-tagged MHC:peptide complexes (pMHC) between fibroblasts and T cells, Wetzel et al. (2005) also show that pMHC is acquired by T cells from the immunological synapse. However, these authors suggest that pMHC was acquired as a result of spontaneous T-cell dissociation from APCs, a phenomenon they called transsynpatic transfer (17).

Human NK cells can acquire membrane components from targets cells as early as 5 min after formation of conjugates with labeled target cell (15). In these experiments, fluorescent label was observed in the synapse formed between NK and target cell. Captured membrane patches moved out of the contact site and were observed distal to the contact site (15). γ δ T cells can also acquire molecules from tumor cells after synapse formation (16). Disruption of synaptic transfer occurs in the presence of Src family kinase inhibitors and inhibitors of actin cytoskeleton rearrangement (16,18).

ii) Exosomes/vesicles

Nevertheless direct cell–cell contact is not always essential as acquisition of some plasma membrane molecules can occur even when cells are physically separated. Intercellular transfer, under these conditions, is mediated by the uptake of molecules incorporated into exosomes. Exosomes are small, secreted membrane vesicles (50–100 nm in size) formed by inward budding of the membranes of endosomes or multivesicular bodies (MVB). Exosomes are released into the extracellular space by fusion of the MVB with the plasma membrane and contain both cytoplasmic components as well as membrane receptors (19). Many cells have been shown to secret exosomes including, cytotoxic T cells (CTL), B cells (20-23), DCs (24-29), CD4+ T cells (3,30,31), mast cells (32), megakaryocytes (33), platelets (33,34), reticulocytes (35-37), epithelial (38,39) and tumor cells (40) (Table 3).

Table 3.

Molecules found incorporated into exosomes

Exosome derived from	Molecules present	Reference
BM derived DCs	MHC class I and II	(24,25)
	CD80
	CD86
	CD11 a,b,c
	CD54 (ICAM-1)
	CD9
	CD81
	Milk fat globule
	(MFG-E8/lactadherin)
	Fas ligand
	m TNF
T cells	MHC class I and II	(3,30,31)
	TCR β
	CD3
	CD2
	CD18
	LFA-1
	CXCR4
	Fas ligand
	TRAIL
	CD80 (B7.1)
	CD86 (B7.2)
	CD54 (ICAM-1)
B cells	CD37	(21-23)
	CD53
	CD63
	CD81
	CD82
	CD86

Using fluorescently labeled exosomes derived from bone marrow DCs, Morelli et al. (2004) showed that exosomes are internalized and processed within endosomes of recipient cells and that immature DCs internalized exosomes better than mature DCs. In addition, they also showed that splenic DCs (both CD8− and CD8α+) were also capable of internalizing exosomes both in vitro and in vivo (24). B (25,41) and T cells (42,43) have been shown to acquire exosomes derived from monocytes, immature and mature DCs. In fact, transfer of MHC molecules to T cells using APC derived vesicles has been described (42,44). Vesicles derived from Drosphilia cells containing MHC/peptide and ICAM-1 have been shown to bind to CD8+ T cells (42). In a rat model, by contrast, T-cell-derived vesicles were shown to mediate the transfer of MHC class II/peptide complexes to both B cells and DCs (3).

Possible Consequences of Molecular Transfer Within the Immune System

Although transfer of molecules has been shown in vitro, whether this phenomenon occurs in vivo, as well as its relevance is still a subject of much debate and continuing investigation. Early experiments suggest that MHC transfer may not occur in vivo. When TCR-transgenic OT-1 T cells, specific for an Ovalbumin (OVA) peptide presented by H-2K^b, were injected into mice transgenic for OVA expressed in the pancreas, the transferred T cells divided vigorously in the draining lymph node. This was presumed to result from the capture and processing of OVA by trafficking DCs. However, if the OVA-transgenic mice were made chimeric with H-2^bm1 bone marrow (K^bm1 cannot present the OVA peptide to OT-1 T cells) no OT-1 T cell division was seen. This suggests that the trafficking K^bm1 DCs, did not acquire intact complexes of K^b with OVA peptides from the pancreatic β cells in sufficient quantities to induce OT-1 T-cell proliferation (45). However more recent data suggests that MHC transfer does indeed occur in vivo (2). We have observed MHC transfer in vivo within rIFNγ treated mice. This observation suggests that MHC transfer in vivo maybe more efficient under inflammatory conditions as compared to steady state, this may explain the BM chimera observations above.

Why would cells of the immune system exchange information in this rather haphazard/uncontrolled way? In other words, what is the point of having highly specialized APCs within the immune system when any cell of the body has the potential to become an APC through acquiring pMHC and/or costimulatory molecules from other APCs? Surely such exchange of information between professional and nonprofessional APCs or between APCs and lymphocytes would lead to anarchy rather than order within the immune system, the outcome of which could be the amplification of inappropriate immune responses. The outcome of molecular transfer is dependent on several things. Firstly, the length of time acquired molecules remain on the recipient cell surface and secondly whether the transferred molecules behave like those normally expressed may add a level of regulation to this phenomenon. It appears that acquired MHC molecules are stably expressed on recipient cells. Dolan et al. (2006) observed that transferred MHC class I molecules were stably expressed for up to 2 days (6). By contrast, the expression of acquired costimulatory molecules OX40L and B7 is more transient. Expression of acquired OX40L on CD4+ T cells is lost by 12 h, and transferred CD80 expression is lost 4 h after acquisition (46,47). This was also the case for acquired CD21. CD21 expression on NK cells, that had acquired this molecule from CD21+ EBV-infected B cells, disappeared after 3 h (15). In other words the outcome of molecular transfer may thus be determined by the stability of the transferred molecule.

Many transferred molecules are functional. For instance, transferred pMHC complexes, both class I and II molecules, are capable of inducing CD4+ and CD8+ T-cell responses, respectively (1,2,6,7,48). In addition, acquired coreceptors also appear functional. For example, CD4+ T cells that acquire OX40L are capable of inducing HIV-1 production in OX40-expressing cells transfected with HIV-1. This is dependent on OX40/OX40L interaction. Using B7.2/CD86, B7.1/CD80 double knockout mice and ICAM-1 knock out mice it was shown that ICAM-1 present in exosomes, in combination with B7 molecules, present on recipient APCs, results in efficient T-cell activation (29).

What is the relevance of intercellular transfer of MHC and other molecules in vivo? Acquisition of MHC molecules, as well as chemokine receptors, may also aid trafficking of immune cells to the site of antigen presentation. It has been proposed that transferred MHC class II molecules, attached to the large highly mobile membrane processes of FDC, could provide ‘guidance’ for antigen-specific T helper cells from the outer zone of the follicle into the apical light zone of the lymph node germinal center (49). Another benefit of molecular transfer between cells is that it leads to the modification of the stimulatory capacity of recipient APCs. As many costimulatory and adhesion molecules, such B7 and ICAM-1, are transferred between cells it is possible that such molecules also contribute to T-cell activation. Uptake of exosomes derived from mature DCs confers on B cells the ability to activate näive CD4+ T cells (25,29). In addition B-cell-derived exosomes expressing MHC class II were demonstrated to stimulate CD4+ T cells in vitro ((20).

By contrast, molecular transfer may also play a role in T-cell tolerance and immune regulation. We have shown that MHC class II molecules acquired by activated CD4+ T cells are recognized efficiently by other activated T cells, and that this T: T-cell antigen presentation induces both apoptosis and hyporesponsiveness (1,5). Presentation of acquired MHC molecules containing a self-antigen during T: T-cell interactions may therefore contribute to the maintenance of tolerance (44,50). Acquisition of exosomes containing class I or class II MHC molecules by CD8α+ DCs may also help to maintain peripheral tolerance as CD8α+ DCs in a resting state are the subset of DCs responsible for induction of peripheral tolerance to self- and nonself antigens (51). Recently it has been shown that B cells whom have acquired exosomes derived from anergic T cells deliver a tolerogenic signal to other T cells (3).

It is possible that MHC acquisition is a mechanism for limiting clonal T-cell expansion during the later stages of an immune response. CD8+ T cells become sensitive to peptide specific lysis after acquisition of MHC class I/peptide complexes from APCs (14,52). Like αβ T cells, γ δ T cells also acquire plasma membrane molecules from other cells. It is possible that in doing so they become sensitive to CTL lysis (14). Jurkat T cells, T-cell blasts and human T cells produce microvesicles containing membrane bound and biologically active apoptosis inducers Fas L and TRAIL following activation and before cell death occurs (31). Interaction with Fas would induce apoptosis in these cells. Hence, intercellular transfer of MHC molecules as well as apoptosis inducing molecules could favour the termination of immune responses.

How Would Transfer of MHC Influence the Alloimmune Response Following Transplantation?

In the context of transplantation, DCs can sensitise alloreactive T cells directly or indirectly. In the direct pathway of allorecognition, responder T cells recognize intact foreign major pMHC on the surface of donor cells (53-58). By contrast, in the indirect pathway of allorecognition, recipient T cells recognize peptides derived from foreign MHC after processing and presentation by self-MHC molecules on recipient APCs (56,59-61). Many studies have suggested that the direct and indirect pathways of allorecognition engage in cross-talk (62,63).

Auchincloss and colleagues observed that mice treated in vivo with an anti-CD8 monoclonal antibody rejected MHC class II-deficient, class I mismatched skin grafts. They observed that, in despite of in vivo treatment to remove CD8+ T cells, CD8+ CTLs were present in mice after graft rejection. These CD8+ T cells were specific for donor class I antigen and their generation, was dependent on help from CD4+ T helper cells with specificity to donor peptides presented by recipient class II molecules. These observations suggested that four different cells must work together to produce cytotoxic T cells (62). In other words, helper or suppressor CD4+ T cells with indirect specificity are activated by recipient DCs that reside in secondary lymphoid organs, whereas direct pathway effector CD8+ T cells must recognize determinants expressed on the cells of the donor graft (Figure 1). This model does not comply with the longstanding dogma that CD4+ and CD8+ T cells must be linked through a single APC.

Figure 1. A four-cell, unlinked, model for interactions between direct and indirect pathway T cells in alloimmunity.

(A) This model proposes that helper or suppressor CD4+ T cells with indirect specificity are activated by recipient DCs that reside in secondary lymphoid organs, whereas direct pathway effector CD8+ T cells must recognize determinants expressed on the cells of the donor graft. Intact allogeneic MHC on donor APCs are recognized by recipient CD8+ T cells in the ‘direct pathway’ (B) The semidirect pathway of allorecognition would conform to a three-cell model as one APC stimulates both direct CD8+ T cells responses and indirect CD4+ responses. Recipient DCs could acquire allogeneic MHC class I molecules from donor APCs, endothelial or tissue parenchymal cells, through cell to cell contact and stimulate direct pathway CD8+ T cells as well as stimulating indirect pathway CD4+ T cells following the processing and presentation of peptides derived from allogeneic MHC molecules.

Recently we have described another mechanism for activating alloreactive T cells. The ‘semidirect’ pathway of alloantigen presentation proposes that recipient DCs can acquire intact functional MHC molecules from donor cells (2). Acquisition of intact donor pMHC complexes by recipient APCs which simultaneously process and present allogeneic MHC molecules as peptides, would allow T cells with direct and indirect allospecificity to recognize their ligands on the same APC surface. In other words, trafficking recipient DCs could acquire allogeneic MHC class I from donor tissues and stimulate direct pathway CD8+ T cells as well as stimulating indirect pathway CD4+ T cells following the processing of some allogeneic MHC molecules, and presenting these as peptides bound to recipient MHC class II molecules. The existence of this pathway may help resolve the aforementioned four-cell problem of direct and indirect allorecognition and suggests that one APC can stimulate both direct CD8+ T cells responses and indirect CD4+ responses (Figure 1).

Unfortunately, at present no in vivo transplantation data exist to support this model. However, this does not mean that MHC transfer does not happen in vivo. Indeed, we have observed intercellular transfer of MHC class II molecules between DCs in vivo. Injecting either immature or mature DCs into a recipent animal previously challenged with rINFγ, resulted in acquisition of donor MHC molecules by DCs. Trafficking H-2 I-E−/− (B10A.4R) DCs injected into H-2 I-E+ (B10A.2R) recipent mice acquired intact MHC molecules from allogeneic cells in vivo under inflammatory conditions. These acquired complexes are fully functional, in as much as, H-2 I-E−/− DCs were able to present the H-Y peptide to H-2 I-E restricted T cells when purified, by cell sorting, from lymphoid tissue of the recipient mice. Whether transfer of MHC class II in this model is via direct cell-to-cell interaction or through production of exosomes is unknown at this point. Indeed how MHC is transferred in vivo has yet to be elucidated.

In the context of transplantation, exosomes are an important source of alloantigen. Morelli et al. (2004) have shown that alloantigens present on exosomes, are acquired by bone marrow derived DCs, leading to activation of alloantigen specific CD4+ T cells in vitro (24,64). In addition, they also showed that intravenous injection of alloantigen expressing exosomes into a host animal resulted in the presentation of preformed complexes of alloantigen plus MHC molecule (semidirect pathway) to host T cells. In this model it was the splenic CD8α+ DCs that were responsible for the presentation of alloantigen in vivo (24,64).

Peche et al. (2003 and 2006) showed that the intravenous administration of exosomes derived from immature donor DCs, before and after transplantation, prolonged heart allograft survival in a fully mismatched recipient (65,66). Exosome administration after transplantation prolonged graft survival tolerance in recipients treated with a drug that inhibited DC maturation, suggesting that immature DCs acquiring exosomes containing alloantigen/MHCs can present these molecules to T cells (66).

If the hypothesis that acquisition of pMHC by trafficking DCs is favoured by local inflammation is correct, the semidirect pathway may have a significant influence on alloimmunity in transplant recipients. A phenomenon that we, and others, have shown in human transplant recipients is a progressive decline in the frequency of T cells with direct antidonor allospecificity (67). Previously we have attributed this to encounter, by direct pathway T cells, with donor MHC on graft parenchymal cells that fail to provide costimulation. The semidirect pathway raises an alternative possibility, namely that the decline in antidonor alloreactivity, is due to the exhaustion that can result from chronic T-cell stimulation. As described earlier, T cells can acquire pMHC molecules from other cells leading to tolerance. However, we have also shown that human CD4+ T cells that have acquired allogenic HLA-DR and B7 from APCs can act as APCs to themselves and resting autologous T cells resulting in amplifying the alloresponse (8). Such an observation implies that acquisition of MHC and costimulatory molecules may be an amplifying step in the direct pathway.

Conclusions

In vitro data demonstrating the transfer of membrane-bound molecules from one cell to another have accumulated over the last two decades. This phenomenon clearly includes MHC:peptide complexes. Recent evidence indicates that MHC transfer can occur in vivo and that the acquired molecules stimulate cognate T-cell responses. Furthermore, this event may solve the ‘four-cell problem’ and restore the rules of linkage both for T cell help and suppression. The challenge for the future is to determine how these events influence alloimmunity in the context of transplantation and whether this phenomenon has a biological role in immunity to pathogens.