The role of the cellular prion protein in the immune system

Abstract

Prion protein (PrP) plays a key role in the pathogenesis of prion diseases. However, the normal function of the protein remains unclear. The cellular isoform (PrP^C) is expressed widely in the immune system, in haematopoietic stem cells and mature lymphoid and myeloid compartments in addition to cells of the central nervous system. It is up-regulated in T cell activation and may be expressed at higher levels by specialized classes of lymphocyte. Furthermore, antibody cross-linking of surface PrP modulates T cell activation and leads to rearrangements of lipid raft constituents and increased phosphorylation of signalling proteins. These findings appear to indicate an important but, as yet, ill-defined role in T cell function. Although PrP^–/– mice have been reported to have only minor alterations in immune function, recent work has suggested that PrP is required for self-renewal of haematopoietic stem cells. Here, we consider the evidence for a distinctive role for PrP^C in the immune system and what the effects of anti-prion therapeutics may be on immune function.

Introduction

The prion diseases or transmissible spongiform encephalopathies (TSEs) are invariably fatal neurodegenerative diseases that include Creutzfeldt–Jakob disease (CJD) and kuru in humans, scrapie in sheep and goats, chronic wasting disease (CWD) in deer and elk and bovine spongiform encephalopathy (BSE) in cattle [1,2]. Prion diseases have been subject to much recent attention due to the epidemic of BSE and the subsequent emergence of its human form, variant CJD (vCJD) [3–5]. Although the number of cases of vCJD has thus far been relatively small (∼160 in the United Kingdom), it is unclear how many individuals are harbouring the disease subclinically. This, together with continued outbreaks of BSE [6,7] and the rising prevalence of CWD, means that prion diseases continue to be of major public health significance and their underlying biology subject to intense scrutiny.

The central player in the pathogenesis of prion disease is the prion protein (PrP), a highly conserved 32-kDa glycophosphatidylinositol (GPI)-anchored sialoglycoprotein [8,9] expressed in neurones, glia and a variety of non-neuronal tissues. According to the protein-only or prion hypothesis [10], the key event in TSE aetiology is the conversion of normal, cellular PrP (denoted PrP^C) to an alternate isoform (PrP^Sc) [11], characterized by increased beta-sheet content, resistance to proteases and detergent insolubility [12]. The conversion event involves only conformational change with no detectable difference in primary structure [13]. The resultant pathological agent, termed the ‘prion’, or ‘protein-only infectious particle’ because it lacks a nucleic acid genome, is proposed to consist principally, if not entirely, of PrP^Sc.

Prions can be generated sporadically, as a result of an as-yet uncharacterized stochastic event causing PrP^C to PrP^Sc conversion, or by dominant mutations in the gene encoding PrP (PRNP in humans), producing mutant PrP^C that more readily undergoes spontaneous conversion to PrP^Sc. However, uniquely among neurodegenerative disorders, prion disease can also be caused through infection with exogenous prions; the latter inducing host-encoded PrP^C to undergo conformational change, via seeding or template-directed refolding [14].

The relationship between prions and the immune system is complex. The lack of a clear immune response in prion disease is assumed to be due to tolerance to PrP^Sc. Further, the immune system actually contributes to pathogenesis by amplifying prion ‘load’ in lymphoid compartments thereby facilitating efficient neuroinvasion (reviewed in [15]). This process is dependent at least partly on expression of PrP^C by immune cells [16].

Because prion diseases arise by structural changes in a single protein, PrP is an attractive target for therapeutic intervention. This could be achieved using compounds that specifically bind PrP^C or PrP^Sc or potentially by gene silencing of PRNP. Alternatively, host immune mechanisms could be adapted to block prion conversion or clear abnormal protein. However, because the constitutive function of PrP^C in the lymphoid system, as in the central nervous system (CNS), remains obscure it is not yet clear how therapies targeting PrP^C directly or indirectly will affect immune function. Further, uncovering the role of PrP^C in the immune system may provide novel insights into the peripheral pathogenesis of prion disease and immune function generally. In this review we focus on the putative physiological function of PrP^C in the immune system and how this might be affected by anti-prion therapeutics.

Structure and function of PrP^C

The mature PrP^C species consists of an unstructured N-terminal region of about 100 amino acids and a C-terminal segment, also around 100 amino acids in length. The N-terminus contains the five octapeptide-repeat region, which has a tight binding site for a single Cu²⁺ ion [17]. A second tight copper site is present downstream of the octapeptide-repeat region but before the structured C-domain [17]. The remaining ∼ 100 amino acids at the C-terminus are folded into a series of three α-helices and a small two-strand β-sheet and are stabilized by a disulphide bond linking residues Cys179 and Cys214 [18]. Further post-translational modifications include attachment of the GPI-anchor and addition of oligosaccharide chains at Asn180 and/or Asn197 [19]. The final PrP molecule may be un-, mono- or diglycosylated; with all three glycoforms being present, albeit at variable ratios, in tissues where PrP^C is expressed. After the traversing the Golgi apparatus PrP is trafficked mainly to the cell surface, where its GPI anchor allows it to enter lipid raft domains preferentially. Pulse-chase labelling experiments have shown that the half-life of PrP^C in murine splenocytes is similar to that in neurones; about 1·5–2 h [20].

PrP^C is highly expressed in the CNS, and as this is the major site of prion pathology most interest has focused on defining the role of PrP^C in neurones. Although PrP^–/– mice have a grossly normal neurological phenotype [21–23], even when neuronal PrP^C is knocked out postnatally [24], they do have subtle abnormalities in synaptic transmission [24–26], hippocampal morphology [27], circadian rhythms [28], cognition [29,30] and seizure threshold [31]. Other postulated neuronal roles for PrP^C include copper-binding [17,32], as an anti- [33–35] and conversely, pro-apoptotic protein [36], as a signalling molecule [37,38], and in supporting neuronal morphology and adhesion [39,40].

Although a variety of molecules have been proposed to be PrP^C ligands (reviewed in [41]), it remains unclear whether the principal receptor for PrP^C has yet to be identified or whether PrP^C is a nodal point in a variety of cellular pathways with multiple binding partners and functions.

If the major role of PrP^C in neurones relates to neurotransmission, the protein must have additional functions as PrP^C is expressed in many non-excitable cells, including glia [42,43], and in numerous lymphoid and non-lymphoid organs [44,45]. Functional heterogeneity of PrP^C between tissues could be mediated by different glycosylation patterns. For example, in human lymphocytes, the unglycosylated form is under-represented [46].

Expression of PrP^C during lymphoid and myeloid ontogeny

PrP^C is expressed by haematopoietic stem cells (HSCs) and may define a particular population of murine HSCs with long-term repopulation potential [47]. Human CD34⁺ HSCs express PrP, but this is down-regulated upon CD15⁺ granulocyte differentiation [48]. Similarly, CD43⁺ Gr-1⁺ granulocyte precursors in murine bone marrow express PrP^C, unlike mature neutrophils [49]. In contrast, maturation of monocytes and dendritic cells (DCs) leads to PrP^C up-regulation [50–52].

Studies in mice show a trend towards down-regulation of PrP^C with B and T cell maturation, and mature T lymphocyte expression during quiescence is low [49,53]. However, in humans and sheep PrP^C expression on mature blood and lymphoid cells remains high. Indeed, human umbilical cord blood lymphocytes express lower levels of PrP^C than those from adults [46], with levels increasing further with ageing [54]. Thus, modelling the cellular function of PrP^C in murine lymphoid cells may underestimate the effects of disrupting this in humans.

PrP^C expression in mature immune cells

PrP^C has been detected on human T and B lymphocytes, natural killer (NK) cells, platelets, monocytes, dendritic cells and follicular dendritic cells [46,48,50,51–59, ]. PrP^C expression increases during human NK cell differentiation, with particularly high levels on CD56⁺ CD3⁺ NK T cells [50]. PrP^C expression may be somewhat higher in peripheral blood T cells than in B lymphocytes, while CD8⁺ cells express slightly more PrP^C than CD4⁺ cells [50,54]. PrP^C expression is also higher in CD45RO⁺ memory compared to CD45RA⁺ naive T lymphocytes [46].

Gene expression microarrays have revealed murine Prnp to be up-regulated in certain types of regulatory T cell [60], via a Stat6-dependent mechanism during interleukin (IL)-4 driven Th0 to Th2 differentiation [61] and in CD8⁺ memory T cells [62]. Hence, PrP^C may be more important in certain types of functionally differentiated lymphocyte that operate in particular immune environments.

PrP^C expression during lymphocyte activation

PrP^C is up-regulated within a few hours in T cells following mitogenic activation with concanavalin A (Con A), phytohaemagglutinin (PHA) or anti-CD3 antibodies [46,55,63,64]. Interestingly, up-regulation of PrP^C on human tonsillar and peripheral blood lymphocytes does not occur in response to phorbol myristate acetate (PMA)/ionomycin [57], suggesting that signalling via elements of the T cell receptor (TCR) may be required to induce changes in PrP transcription. In contrast to the relative ease of PrP^C up-regulation in activated T cells, treatment with lipopolysaccharide (LPS) does not increase PrP^C expression on B cells [64].

Intracellular pathways linking T cell activation with increased PRNP mRNA have not been characterized, and regulation of PRNP expression in vivo is poorly understood. The PRNP promoter does not contain a TATA box (a generic promoter motif), but a GC-rich region with SP1 transcription factor binding sites reminiscent of housekeeping genes is present [65–67]. Recently, sequencing and in silico analysis has identified putative binding sites for transcription factors nuclear factor (NF)-IL6, myogenic determination factor (Myo)D, myeloid zinc finger (MZF)-1, myocyte enhancer factor (MEF)-2, octamer (Oct)1, myelin transcription factor (MyT)1 and nuclear factor activated T cell (NFAT) [68,69]. Heat shock elements identified in the PRNP promoter [68,70] have been shown to interact with heat shock transcription factor (HSTF)-1 and heat shock increases Prnp mRNA and protein levels in neurones [70].

Localization and trafficking of PrP^C in T cells

In human T cells PrP^C is present in lipid rafts where it co-localizes and/or co-immunoprecipitates with ganglioside GM3 [71], and to a lesser extent GM1 [72] and with Fyn, Reggie-1, and lymphocyte-specific tyrosine kinase P56lck (Lck), but not sedimented red cells (Src) [73,74]. Following activation with anti-CD3 and anti-CD28, PrP^C co-immunoprecipitates with zeta chain associated protein 70 kDa (Zap70) [73]. In Jurkat lymphocytes, PrP^C has been seen to co-localize with CD3 in caps induced by hypothermia [75]. However, although PrP^C accumulates at sites of T cell-DC contact during major histocompatibility complex (MHC)-peptide stimulation, it does not co-localize with CD3, leucocyte function antigen (LFA)-1, CD43, linker for activation of T cells (LAT) or thyroid hormone (Thy1) [52]. Thus, these observations might reflect a non-specific clustering effect of activation on lipid raft components rather than a specific role for PrP^C within the immune synapse.

In neurones, PrP^C is endocytosed from the cell surface within minutes and enters early endosomes, from whence a large proportion is quickly recycled to the cell surface [76]. This raises the possibility that PrP^C may exert some intracellular function through being constitutively endocytosed (reviewed in [77–79]). Interestingly, internalization of a recombinant PrP–Fc fusion protein by monocytes led to increased tyrosine phosphorylation of Syk and Pyk2 and activation of extracellular regulated kinase (ERK1) and 2 and Akt kinase [80]. Incubation of T cells with anti-PrP monoclonal antibodies (MoAbs) for > 10 min results in internalization of a large proportion of surface PrP into Limp-2 positive endosomes [74]. However, to what extent and via what mechanism constitutive endocytosis of PrP^C occurs in lymphocytes is not clear.

A further mechanism of PrP^C trafficking that may apply in lymphocytes is release into the extracellular milieu or onto other cells. PrP^C molecules have been shown to transfer between cells, including lymphocytes, although this required PMA or Con A activation of the donor or recipient cells and cell–cell contact [81]. PrP^C may be shed constitutively from lymphocytes [20], perhaps in a variety of forms. PrP⁺ microparticles are shed from apoptotic T lymphocytes [82] and epithelia and activated platelets release PrP⁺ exosomes and microvesicles [83–85]. Combined, these phenomena may explain the presence of large quantities of soluble PrP^C in the plasma [86,87]. However, the functional consequences of PrP^C release from lymphocytes remain unknown.

Effects of PrP ligation in lymphocytes

Cross-linking surface PrP^C on lymphocytes with anti-PrP antibodies induces clustering of PrP in caps containing Thy1, Reggie-1, Reggie-2, CD3, F-actin, Fyn, Lck, LAT and GM1, increased release of reactive oxygen species (ROS) and increased phosphorylation of Src family kinases (but not Fyn) and ERK1/2 [72,74,88]. The effects on ROS may be mediated by nicotinamide adenine dinucleotide phosphate (NAPDH) oxidase and those on ERK1/2 by NAPDH oxidase and MEK1/2 [88]. Cross-linking PrP^C in CEM cells modulates ionophore-induced calcium entry and intracellular release [72], while in Jurkat cells PrP^C cross-linking alone induces calcium fluxes, although less marked than those produced by anti-CD3 [74].

These effects seem generally stimulatory, yet anti-PrP antibodies can also block activation of human T cells by Con A [55] or anti-CD3 [46] and murine TCR tg T cells by MHC-peptide [52]. Interestingly, some of these effects may be epitope-specific; for example, anti-PrP MoAbs block anti-CD3-induced proliferation only when certain PrP epitopes are targeted [46].

The mechanism by which these blocking effects occur remains unknown. PrP^C is present in or near the immunological synapse and shows enhanced co-localization with the TCR in lipid rafts upon cross-linking [74]. It also co-localizes with MHC class II in DCs [51]. Thus, the blocking effects of anti-PrP could be mediated via the T cell or antigen-presenting cell (APC). Indeed, in a mixed lymphocyte reaction, anti-PrP inhibited proliferation when PrP expression was restricted to either DCs or T cells [52]. This effect was maintained when Fab fragments were used in place of total IgG to eliminate PrP cross-linking and reduce the degree of steric hindrance [52]. Thus, the effects of anti-PrP on T cell activation may not be be due simply to mechanical interference, but mediated via destabilization of the immune synapse or altered signalling.

The relationship between these observations and the normal function of PrP^C remains unclear. Most of these experiments involved cross-linked surface PrP^C even though there is no evidence that any potential PrP ligand need be polyvalent. Similar effects have been observed upon cross-linking of other GPI-anchored proteins [89], thus they may not inform on the specific function of PrP^C.

Immunological phenotype of PrP^–/– mice

PrP^–/– mice have been reported to have normal MHC class I and II expression, DC maturation and numbers of haematopoietic stem cells, CD4⁺, CD8⁺ and B cells [21,47,52,53], suggesting that they are not grossly immunodeficient. Spontaneous tumours or greater susceptibility to infections than PrP^+/+ mice have not been reported. Further, PrP^–/– mice have proved a useful tool for production of anti-PrP MoAbs of all major IgG classes by immunization with PrP^Sc or recPrP [22,90–92], suggesting that they are able to mount effective B cell responses and provide T cell help where appropriate.

Based on these data it is perhaps unsurprising that PrP^–/– mice kept in pathogen-poor facilities do not develop immunopathology spontaneously; this may become manifest only following focused immunological challenge such as infection or immunization. Published data on in vivo immune responses in PrP^–/– animals are limited. Mice lacking both PrP and its downstream partial homologue Doppel have grossly normal CD8 expansion following lymphocytic choriomeningitis virus (LCMV) infection and specific antibody production after infection with LCMV or vesicular stomatitis virus (VSV) [93].

However, there is a lack of consensus from knockout experiments as to whether PrP^C is required for optimal T cell activation in vitro. In the initial report of the Zurich I PrP knock-out mouse, responses to Con A were normal [21], whereas a recent report described reduced proliferation [94]. Splenocytes from the knock-out line derived in Edinburgh were reportedly hyporesponsive to Con A [63]; conversely, Liu and colleagues reported normal responses to Con A and plate-bound anti-CD3 in T cells from the same strain [49].

In an attempt to resolve these discrepancies, Ballerini and coworkers studied mixed lymphocyte reactions in which PrP was deleted from either the T cells or DCs [52]. Interestingly, PrP^–/– T cells responded normally to PrP^+/+ DCs, whereas PrP^–/– DCs were less efficient than PrP^+/+ DCs at stimulating PrP^+/+ T cells. Thus, discrepancies in the literature on PrP^–/– T cell proliferation may relate to the degree to which this was dependent on APC : T cell conjugation.

Absence of PrP may affect other mature immune functions. In zymosan-induced peritonitis, the leucocyte infiltrate in PrP^–/– mice contained more monocytes and significantly fewer neutrophils than in wild-type animals [95]. Based on work in PrP^–/– mice or cells derived from them, PrP^C has been proposed to down-regulate phagocytosis by macrophages [95] but also to be involved in the phagocytic machinery used by Brucella abortus to invade macrophages [96], although this finding has been challenged [97].

Perhaps the most striking abnormality detected thus far through use of PrP knock-out mice is that PrP^–/– HSCs display impaired self-renewal, albeit after multiple rounds of transplantation into lethally irradiated PrP^+/+ recipients [47]. That PrP may be required for cell growth under harsh conditions is supported by gene expression microarray data, suggesting that it is up-regulated in CD8⁺ cells undergoing homeostatic proliferation upon transfer into lymphopaenic mice [62]. The molecular programme underlying lymphoid repopulation is remarkably similar to that of memory differentiation, in which increased Prnp and surface PrP expression have also been observed [46,62]. Moreover, PrP^+/+ TCR tg T cells transferred into peptide-challenged PrP^–/– mice showed reduced numbers of mitoses [52]. Thus, the role of PrP^C in T cell expansion and differentiation may be mediated either by the cell itself or supporting populations from other lineages. However, the mechanisms underlying these observations remain elusive. If PrP^–/– lymphocytes, or PrP^+/+ cells in a PrP null environment, suffer premature senescence, via what pathway is this mediated? Is there failure of specific signalling proteins known to be activated by PrP cross-linking, or is a separate mechanism involved?

Although studies to date have revealed subtle immunological abnormalities in PrP^–/– mice, none has yet involved postnatal knock-out of lymphoid PrP. In a Cre-lox-dependent, conditional Prnp deletion model developed by Mallucci and coworkers, PrP deletion was restricted to neurones only [24]. In the tetracycline-responsive system used by Tremblay and colleagues, although adult mice remained well for > 380 days following doxycycline-mediated suppression of PrP expression, neuronal PrP^C expression was not entirely abolished and lymphoid PrP^C expression prior to and following doxycycline treatment was not reported [98]. Therefore, a compensatory mechanism operating in lymphocytes, myeloid cells and their precursors in all PrP^–/– mice thus far generated cannot be excluded.

PrP ablation versus ligation

In reviewing the available data on PrP^C function in the immune system, it is worth considering that embryonic deletion and antibody ligation of PrP are not necessarily modelling similar processes. There is increasing evidence that GPI-anchored proteins have signalling functionality in lymphocytes, principally through association with Src family non-receptor tyrosine kinases [89], with which PrP is known to associate. While loss of PrP alone may alter lipid raft composition and obviate any PrP-specific function, it is unlikely to impede the ability of other lipid raft components to signal via these pathways. Anti-PrP treatment is likely to induce additional effects, through disruption of protein–protein interactions in lipid rafts or excessive signalling via PrP, and possibly disrupting or enhancing key pathways shared with other GPI-anchored proteins. This is supported by discrepancies between the effects on T cell physiology of embryonic PrP deletion compared to anti-PrP MoAb administration [52]. Similarly, embryonic deletion of Thy1 produces different effects on immune function to treatment of cells with anti-Thy1 antibodies [99].

Conclusion

A definitive function for PrP^C in the immune system, as in neurones, remains to be established. PrP^C appears to be robustly up-regulated in T cell activation but its expression in T cells is not a prerequisite for proliferation. Rather, PrP^C may be important in mediating pro-survival signals in cells undergoing multiple mitoses in stressful conditions, such as lymphopaenia or rapid memory cell expansion. Detailed analysis of immune responses in PrP^–/– mice should provide further insights into the role of PrP^C but may require generation of conditional knock-out models.

A greater understanding is also required of the effects of targeting PrP in the periphery as part of a therapeutic strategy, as the consequences of ligation may be distinctly different from the phenotype produced by genetic ablation. Anti-prion agents that bind surface PrP^C may induce considerable changes in T cell physiology by disrupting lipid raft constituents, enhancing or blocking PrP signalling or endocytosis, or by labelling bound cells or soluble PrP for elimination. Cross-linking surface PrP^C in T cells induces striking physiological changes reminiscent of T cell activation, but has also been shown to interfere with mitogenic activation. Thus it remains unclear whether PrP targeting will predispose the immune system to over-activation or hyporesponsiveness. Furthermore, preferential deletion or senescence of PrP^high immune cells may disrupt immune homeostasis. If high PrP^C expression is a particular characteristic of HSCs required for long-term renewal, or specialized lymphocytes such as memory and regulatory cells, the effects on immune function may be profound. Interference in the function of key components in the immune system may cause severe immunopathology rendering such therapies unsuitable for long-term use, despite the fact that anti-PrP therapy following peripheral exposure may need to be life-long. To date, immunisation with self-PrP or transgenic expression of anti-PrP antibodies [100] have not been reported to cause immunopathology in mice. However, dermatitis with mononuclear cell invasion and destruction of hair follicles has been described in Lewis rats several months after immunisation with PrP 182–202 [101].

GPI-anchored proteins such as the Campath-1 antigen (CDw52) have either proved tractable therapeutic targets in themselves or modifiers of agents directed against other cell surface constituents [102]. Thus, interest in the immune function of PrP^C extends beyond its role in prion disease. As an activation antigen and lipid raft component it may be a potential target for immunomodulatory therapy in other diseases. Further, PrP up-regulation has been detected in a number of tumour lines [103–105]. Because of the intense interest in PrP as the central mediator of prion diseases, many of the tools for studying the contribution of this protein to immune function are readily available. A serendipitous benefit of the tragedy of BSE and vCJD may be a lasting contribution to immunology and immunotherapeutics.