Report

Cross-Linking Cellular Prion Protein Triggers Neuronal Apoptosis in Vivo

See allHide authors and affiliations

Science  05 Mar 2004:
Vol. 303, Issue 5663, pp. 1514-1516
DOI: 10.1126/science.1094273

Abstract

Neuronal death is a prominent, but poorly understood, pathological hallmark of prion disease. Notably, in the absence of the cellular prion protein (PrPC), the disease-associated isoform, PrPSc, appears not to be intrinsically neurotoxic, suggesting that PrPC itself may participate directly in the prion neurodegenerative cascade. Here, cross-linking PrPC in vivo with specific monoclonal antibodies was found to trigger rapid and extensive apoptosis in hippocampal and cerebellar neurons. These findings suggest that PrPC functions in the control of neuronal survival and provides a model to explore whether cross-linking of PrPC by oligomeric PrPSc can promote neuronal loss during prion infection.

The cellular prion protein (PrPC) is a ubiquitously expressed copper-binding glycophosphatidylinositol (GPI)–linked glycoprotein of unknown function. PrPSc, a misassembled conformer of PrPC, is intimately associated with prion-mediated neurodegenerative disease and is the major molecular component of purified prion preparations (1). Prions are thought to propagate by a high-fidelity process in which PrPSc self-replicates by templating the conformational rearrangement of endogenous PrPC, but the molecular events through which prion infection and the resulting accumulation of PrPSc lead to the neuronal dysfunction, vacuolation, and death that characterize prion pathology remain unclear (2). Intriguingly, disease-associated conformers of PrP alone appear insufficient to induce neurotoxicity in the absence of PrPC. For example, neural tissues overexpressing PrPC grafted into the brains of Prnp0/0 mice develop histopathological changes characteristic of scrapie disease when infected with prions, but the adjoining PrPC-deficient tissue remains healthy despite the presence of high levels of PrPSc (3). In addition, interruption of neuronal PrPC expression during an ongoing prion infection within the central nervous system prevents neuronal loss and reverses early spongiform change (4). These findings jointly argue that PrPC itself may play a critical role in the prion neuropathologic cascade. In support of this hypothesis, PrPC-dependent signal transduction was identified following specific antibody-mediated cross-linking in a differentiated neuronal cell line (5).

To determine whether PrPC-dependent signaling could produce neurotoxic effects in vivo, two microliter volumes of three different purified, endotoxin-free, PrPC-specific monoclonal antibodies (MAbs) were stereotaxically injected into the right hippocampus of C57BL/10 mice. Two of the three MAbs, immunoglobulin G (IgG) D13 (6) and IgG P (7), each recognizing epitopes within the 95 to 105 region of PrP, caused extensive neuronal loss throughout the hippocampal region in 4 of 5 mice and 7 of 10 mice injected, respectively (Fig. 1, A to F). Neuronal damage was characterized by pycnotic nuclei in some or all of the CA1, CA2, and CA3 cell-body layers and in the dentate gyrus. Damage could be detected at 24 hours after antibody injection but not at 12 hours, and at an antibody concentration of 1 mg/ml but not at a concentration of 500 μg/ml or less. No local B or T cell infiltration or astrocytic activation was observed, even at 48 hours after injection of antibody. Equivalent contralateral injections of MAb b12, recognizing the human immunodeficiency virus–1 (HIV-1) envelope glycoprotein gp120, into the same mice produced no neuronal damage in any of the 15 mice studied (time of observation 48 hours postinjection) (Fig. 1, A to F). Similarly, no damage was detected when monovalent Fab fragments of IgGs D13 or P were independently injected into mice in the same manner (12 of 12 mice). To determine whether the effects we observed after antibody injection were PrP-specific, IgG P antibody was preincubated for 60 min with a threefold molar excess of recombinant PrP 90-231 antigen prior to hippocampal injection into three mice. When the brains of these animals were examined, damage was restricted to a very small area of neurons in only one animal, indicating that the capacity of IgG P to induce neuronal injury was markedly diminished by recombinant PrP (time of observation 48 hours postinjection) (fig. S1). Additionally, to ascertain whether cell loss could be triggered by antibody binding to GPI-linked neuronal proteins other than PrPC, hippocampal injections of purified polyclonal IgG antibody recognizing both the GPI-linked and transmembrane forms of the mouse neural cell–adhesion molecule (NCAM) were performed (fig. S2). Binding of these antibodies did not cause neuronal damage in any of the four mice that were treated (time of observation 48 hours postinjection). Finally, injecting each of the D13 and P PrPC-specific IgGs into the brains of Prnp0/0 mice, which do not express PrPC, was equally benign at 48 hours postinjection (seven of seven mice).

Fig. 1.

PrPC-specific antibodies mediate neuron death. (A) Photomicrograph (cresyl violet staining) of a coronal section of mouse hippocampus after injection of IgG b12 (left hemisphere) or IgG P (right hemisphere). (B and C) Magnification of right- and left-hand boxed areas shown in (A). (D) Photomicrograph (cresyl violet) of mouse hippocampus after injection of IgG b12 (left hemisphere) or IgG D13 (right hemisphere). (E and F) Magnification of right- and left-hand boxed areas shown in (D).

Hippocampal injection of a third PrP-reactive antibody, IgG D18, binding efficiently in situ to neuronal cell-surface PrPC from residues 133 to 157 (6) (fig. S2), did not manifest any neuronal injury (seven of seven mice at 48 hours postinjection), arguing against any role for antibody Fc-mediated cytotoxic mechanisms in neuronal cell death.

However, because hippocampal neurons express complement components and their receptors (8), and can be susceptible to complement-mediated injury (9), coronal sections of hippocampus that had sustained neuronal loss following PrPC-specific antibody injection were immunostained to detect the presence of the activated complement component C3a. Using this assay, no evidence of complement activation in areas of antibody-mediated neuronal loss was found (fig. S3). To more definitively exclude the possibility that complement-mediated neurotoxicity could explain our experimental findings, and to further investigate whether cross-linking PrPC antigen was critical to triggering neuronal damage, Fab fragments of IgG P were rendered bivalent by complexation with a mouse IgG1 MAb recognizing human antibody κ light chains. Mouse IgG1:Fab complex stereotaxically injected into hippocampus triggered extensive neuronal loss (three of four mice, time of observation 48 hours postinjection) (Fig. 2), whereas injection of the mouse IgG1 anti-κ reagent alone was entirely nontoxic (four of four mice, time of observation 48 hours postinjection). Because mouse antibodies of the IgG1 subclass are considered unable to activate complement (10), these results indicate both that complement played no role in the neurode-generative process we observed and that cross-linking of cell-surface PrPC molecules was critical in precipitating neuronal death.

Fig. 2.

PrPC cross-linking is required to elicit neurotoxicity. Stereotaxic hippocampal injection of mouse IgG1 anti-human κ:P Fab complex causes extensive neuronal injury. (Inset) A magnified view of damaged hippocampal neurons (asterisk). Equivalent injections of monovalent P Fab alone are nontoxic.

In prion-infected brains, neuronal loss is also seen in regions other than the hippocampus. To evaluate whether PrPC-specific antibody-mediated toxicity was operational in other neuronal populations, simultaneous stereotaxic injections of IgGs P and b12 were administered to the left and right halves of the cerebellar cortex. When sections of the antibody-treated cerebellum were examined, neurons exposed to PrPC-specific antibody again suffered extensive injury characterized by pycnotic nuclei, but neurons exposed to b12 control antibody did not (Fig. 3).

Fig. 3.

Neuronal injury following injection of PrPC-specific antibody into mouse cerebellum. Photomicrograph (cresyl violet staining) of a section of mouse cerebellum after injection of IgG P (left side) or IgG b12 (right side). Magnifications of boxed areas are shown in the lower panels.

To identify the mechanism of neuronal cell death triggered by certain PrPC-specific MAbs, coronal sections of antibody-treated brains were tested for terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) positivity (Fig. 4). IgGs D13 and P caused apoptotic neuronal cell death in the brains of Prnp+/+ mice. No apoptotic cell death was evident in D18- or b12-treated brains, or in the brains of antibody-treated Prnp0/0 mice.

Fig. 4.

PrPC-specific antibody triggers neuronal apoptosis. TUNEL analysis of PrPC-specific antibody-treated hippocampal neurons. A field of 4′,6′-diamidino-2-phenylindole (DAPI)–labeled neuronal cell nuclei (left panels), TUNEL staining of apoptotic cells in the equivalent field (center panels), and a merge of these two photomicrographs (right panels). (A) IgG b12, (B) IgG P, and (C) IgG D13.

Our experimental observations could be explained by a loss of PrP function, for example if antibody binding interrupted a PrPC-dependent neuronal survival signal mediated through an association between PrPC and another molecule. If this were the case, however, then monovalent Fab fragments of the D13 and P MAbs would also be expected to interrupt this type of association and initiate the neurotoxic pathway. Instead it appears that, upon complexation with IgG D13 or P, PrPC was efficiently dimerized, thereby initiating an apoptotic cascade in neurons, possibly through docking with an as yet unidentified secondary molecule. We reason that the D18 antibody, although binding effectively to the neuronal cell surface, was either inefficient at cross-linking neighboring PrPC molecules or, alternatively, sterically obscured a region of PrPC that interacts with its putative signaling partner, thus preventing activation of the apoptotic pathway. In prion-infected brains, neuronal loss may occur when oligomeric forms of cell-surface PrPSc (11, 12) undertake the PrPC cross-linking role performed in our experiments by these MAbs. Thus, PrPC may be co-opted twice in prion disease, once as a substrate for conformational conversion into nascent PrPSc molecules and additionally as a signaling vehicle that promotes neuronal injury and death. This model system should permit the study of both PrPC function and PrP-associated neurotoxicity and, moreover, argues that extreme caution must be exercised if potent PrPC-specific antibody inhibitors of prion propagation are to be considered for use within the central nervous system as agents for the therapy or prevention of prion infection (13, 14).

Supporting Online Material

www.sciencemag.org/cgi/content/full/1094273/DC1

Materials and Methods

Figs. S1 to S3

References and Notes

View Abstract

Navigate This Article