Report

Depleting Neuronal PrP in Prion Infection Prevents Disease and Reverses Spongiosis

See allHide authors and affiliations

Science  31 Oct 2003:
Vol. 302, Issue 5646, pp. 871-874
DOI: 10.1126/science.1090187

Abstract

The mechanisms involved in prion neurotoxicity are unclear, and therapies preventing accumulation of PrPSc, the disease-associated form of prion protein (PrP), do not significantly prolong survival in mice with central nervous system prion infection. We found that depleting endogenous neuronal PrPc in mice with established neuroinvasive prion infection reversed early spongiform change and prevented neuronal loss and progression to clinical disease. This occurred despite the accumulation of extraneuronal PrPSc to levels seen in terminally ill wild-type animals. Thus, the propagation of nonneuronal PrPSc is not pathogenic, but arresting the continued conversion of PrPc to PrPSc within neurons during scrapie infection prevents prion neurotoxicity.

Prion diseases are fatal transmissible neurodegenerative diseases, characterized pathologically by widespread neuronal loss, spongiform change, and the accumulation of PrPSc, the pathological isoform of a host-encoded prion protein (PrPc). However, the cause of neuronal death in prion disease remains unclear. Thus, although PrPSc is associated with both infectivity and pathological changes, levels of PrPSc in the brain may not correlate with symptoms in prion-infected mice (14), and there are human prion diseases with little or no detectable PrPSc (5, 6). Further, PrPSc itself is not toxic to brain tissue that does not express PrPc and cannot replicate prions (1, 2, 7). Also, compounds that reduce PrPSc accumulation in prion-infected cells in culture have only modest effects in vivo on disease incubation periods in prion-infected mice and only when coadministered with the prion inoculum or very soon after infection (8, 9). No treatment has been effective in animals after the onset of clinical signs of disease, and no agent has prevented disease progression in mice during the asymptomatic preclinical phase of central nervous system (CNS) scrapie infection, when intervention would have the greatest therapeutic potential (10, 11). The conversion of PrPc to PrPSc is intrinsic to the pathological process, however, because mice devoid of PrPc (Prnp0/0) are resistant to prion disease and do not propagate infectivity (1, 2, 12). Similarly, PrP-null brain tissue surrounding prion-infected Prnp+/+ neurografts does not develop prion neuropathological change (7). Targeting PrPc itself, as opposed to PrPSc, in established prion infection directly removes the substrate for further conversion and prion replication and might be expected to be a more effective therapeutic strategy than reduction of PrPSc accumulation.

We tested the effects of neuronal PrPc depletion in animals with established neuroinvasive scrapie on the subsequent course of disease and neuropathological changes. We have previously shown that acquired neuronal PrPc depletion in adult mice is in itself without major detrimental effects (13). In another mouse model of acquired PrP knock-out, only the effects of PrPc depletion before inoculation with scrapie prions were studied (14), essentially replicating the inoculation of Prnp0/0 animals; the effects of PrPc depletion after scrapie infection were not examined. We used two lines of transgenic MloxP mice, tg37 and tg46, that express PrPc from MloxP transgenes at ∼3 and 1 times wild-type levels, respectively, and that normally succumb to scrapie ∼12 and 18 weeks, respectively, after intracerebral Rocky Mountain Laboratory (RML) scrapie prion inoculation (Figs. 1 and 2). PrPc expression in these mice (which were generated on a Prnp0/0 background) can be eliminated by the action of the enzyme Cre recombinase (15), which excises the (floxed) PrP-coding sequences in the MloxP transgenes. NFH-Cre mice express Cre under the control of the neurofilament heavy chain (NFH) promoter (16) at ∼10 to 12 weeks of age, which results in Cre-mediated recombination of floxed transgenes in neuronal cells but not in astrocytes or other cell types (fig. S1). Thus, double transgenic NFH-Cre/MloxP mice express PrP in neurons and nonneuronal cells until ∼12 weeks of age, when they undergo Cre-mediated depletion of neuronal PrPc (13).

Fig. 1.

Clinical and neuropathological features of scrapie-infected mice with and without neuronal PrPc depletion. HC, hippocampus; –, absent; +, ++, +++ indicate mild, moderate, and severe, respectively; + symptoms represented disheveled appearance and poor grooming. At all time points, n = 3 to 6 animals, except at 48 and 49 wpi, when n = 2. Arrow indicates onset of Cre-mediated PrPc depletion.

Fig. 2.

Survival of scrapie-infected NFH-Cre/MloxP mice after Cre-mediated PrPc depletion. All tg37 and tg46 mice succumbed to scrapie within 12 and 18 wpi, respectively. No animals with Cre-mediated PrPc depletion at 8 wpi have succumbed to scrapie or show any clinical signs of disease by 52 wpi. Onset of Cre-mediated PrPc depletion in NFH-Cre/MloxP mice, 8 weeks into the course of infection, is indicated.

We inoculated NFH-Cre/MloxP mice with RML scrapie prions at 3 to 4 weeks of age on weaning, allowing prion replication and CNS infection to proceed normally until Cre-mediated neuronal PrPc depletion occurred. MloxP mice without the NFH-Cre transgene were inoculated in parallel. Animals were examined daily and were culled for analysis of neuropathological changes at two-week intervals after inoculation (17). We confirmed MloxP transgene recombination in the brains of double transgenic animals by Southern analysis of whole-brain DNA at around 8 weeks postinoculation (wpi) (fig. S2), at which stage they were 11 to 12 weeks old, consistent with our previous findings (13). Correspondingly, total brain PrP expression, determined by semiquantitative immunoblotting (17), was similar in all mice before 8 wpi, at which point it declined in animals with Cre-mediated recombination (fig. S3).

At this time point, all animals with and without Cre expression showed pathological evidence of neuroinvasive CNS scrapie infection to a similar extent. Specifically, there was PrPSc deposition and reactive astrocytosis in thalamus, hippocampus, and cortex (Fig. 3, E, G, Q, and U). In scrapie-infected NFH-Cre/tg37 and tg37 mice (but not in tg46 mice), there was also early spongiform degeneration, notably in the hippocampus (Figs. 1 and 3, A, C, I, and M). The early appearance of spongiosis in these animals may reflect the higher level of expression of PrPc and short scrapie incubation period of tg37 mice compared with tg46 mice. Similar levels of PrPSc were also found in all mice at 8 wpi with the use of semiquantitative immunoblotting, but were just within the limits of detection with the use of conventional methods, and were confirmed by sodium phosphotungstic acid precipitation of PrPSc from brain homogenates (17, 18) (fig. S4).

Fig. 3.

Prevention of neuronal loss and reversal of early spongiosis in scrapie-infected mice after PrPc depletion. Fixed sections show hippocampal region from scrapie-infected tg37 and NFH-Cre/tg37 mice at various time points postinfection. Sections were stained with haematoxylin and eosin (H&E) and immunostained for detection of astrocytosis and PrPSc deposition. There is severe loss of CA1 to CA3 neurons (arrows) (B and D) with shrinkage of the entire hippocampus (B) in terminally ill tg37 mice but no neuronal loss in asymptomatic prion-infected mice with Cre-mediated PrPc depletion [(J), (K), and (L)] up to 48 wpi. Early spongiosis was seen in eight out of eight animals at 8 wpi (C and M), but was not seen at 12, 26, and 48 wpi in NFH-Cre/tg37 mice [(N), (O), and (P)], despite continued PrPSc accumulation and gliosis [(R) to (T) and (V) to (X)]. Scale bar represents 320 μm, except in panels (C), (D), and (M) to (P), where it represents 80 μm.

The depletion of neuronal PrPc in animals with established CNS scrapie prevented progression to clinical prion disease and resulted in the long-term survival of infected animals. To date, prion-infected NFH-Cre/tg37 and NFH-Cre/tg46 mice remain asymptomatic at >57 wpi (n = 6) and >58 wpi (n = 3), respectively, whereas tg37 and tg46 mice infected at the same time died after 12 weeks (84 ± 5 days, n = 6) and 18 weeks (120 ± 2 days, n = 8), respectively (Figs. 1 and 2). This represents a >fourfold increase in scrapie incubation time for NFH-Cre/tg37 animals (∼threefold for NFH-Cre/tg46 mice), approaching the normal life-span of a mouse.

Neuropathologically, asymptomatic prion-infected animals with PrPc depletion were protected from neuronal loss up to 48 wpi. In scrapie-infected tg37 mice without PrP depletion, hippocampal CA1 to CA3 neurons began to degenerate from 10 wpi, with almost complete CA1 to CA3 cell loss by 12 wpi and accompanying shrinkage of the entire hippocampus (Fig. 3, B and D). Similar changes were seen in terminally ill tg46 mice by 18 wpi, establishing a correlation between development of clinical symptoms in RML scrapie infection and neuronal loss in MloxP mice. In contrast, in scrapie-infected NFH-Cre/MloxP mice, CA1 to CA3 neurons remained healthy, and hippocampal structure was completely preserved (Fig. 3, I to L and M to P), demonstrating the neuroprotection afforded by depleting neuronal PrPc in established CNS scrapie infection.

Further, the hippocampus was also free of spongiosis (intraneuronal vacuoles in cells that have not yet degenerated) in scrapie-infected mice after neuronal PrPc depletion. We found early spongiform change in both tg37 and NFH-Cre/tg37 mice at 8 wpi (Figs. 1 and 3, C and M) but not in age-matched uninfected controls (fig. S5) nor in mice with Cre-mediated neuronal PrPc depletion examined as early as 10 wpi, up to 48 wpi (Figs. 1 and 3, N, O, and P, and fig. S5). Thus it appears that there was reversal of early spongiosis in double transgenic animals after PrPc depletion. Because spongiform pathology at 8 wpi was relatively subtle, coinciding with the onset of Cre-mediated PrP depletion, we further inoculated animals at 1 week of age (19) to allow more time for development of prion pathology before PrPc depletion. Under these conditions, there was unequivocal spongiform change at 8 wpi (Fig. 3, C and M), which similarly reversed when Cre was expressed at the relatively later time point of 10 wpi in these animals, when they were aged ∼11 weeks (fig. S5). Although in this model early spongiosis was not associated with the diagnostic neurological signs of murine scrapie, in humans this stage of prion infection may correspond to the early clinical stages of disease, which may be relatively prolonged and where any potential for reversal of pathology would be most important.

The pattern of neuroprotection observed in scrapie-infected NFH-Cre/MloxP mice correlated well with the known pattern of Cre expression in NFH-Cre mice (13), being strongest in neuron-rich regions, in particular in the entire hippocampus. A degree of spongiform change was seen in these mice in other brain regions at late time points, notably in some cortical regions, deep white matter tracts, and cerebellum, and may reflect lower density neuronal populations and hence less overall NFH-Cre transgene expression in these regions. Importantly, however, these mice remained asymptomatic (Fig. 1).

However, we did observe increasing gliosis and PrPSc deposition in infected animals after PrPc depletion. Most notably, PrPSc accumulation progressed over prolonged periods of observation (Fig. 3, Q to T) and was clearly detected by immunoblotting in NFH-Cre/tg37 mice at 12 wpi, and in NFH-Cre/tg46 mice by 18 wpi (Fig. 4), but was not associated with neuronal loss or clinical symptoms. In NFH-Cre/tg46 mice, levels of PrPSc increased significantly by 49 wpi (Fig. 4A), and, in NFH-Cre/tg37 mice, PrPSc levels at 48 wpi were equivalent to those seen in terminally ill tg37 mice at 12 wpi and in end-stage RML-scrapie–inoculated wild-type mice (Fig. 4B).

Fig. 4.

Immunoblot analysis of total PrP in scrapie-infected mice with and without PrPc depletion. SDS–polyacrylamide gel electrophoresis of 10% whole-brain homogenates pre– (–) and post– (+) proteinase K (PK) digestion from scrapie-infected tg46 and NFH-Cre/tg46 mice (A) and tg37 and NFH-Cre/tg37 mice (B) from each time point. 10 μl was loaded per lane. Total PrP was detected with the use of ICSM35 antibody (25). “+Cre” and “–Cre” denote the presence or absence of Cre-mediated PrPc depletion, respectively. PrPSc continues to accumulate in MloxP mice after PrPc depletion and is first easily detected at 18 and 12 wpi in NFH-Cre/tg46 and NFH-Cre/tg37 brains, respectively. Samples from a terminally ill RML-inoculated wild-type mouse (ES RML) and from mock-inoculated (m.i.) NFH-Cre/MloxP mice were included as controls. kDa, kilodaltons.

The continued accumulation of PrPSc in this model after neuronal PrPc depletion is likely to reflect prion replication predominantly in glial cells where Cre was not expressed (fig. S1) and PrPc not depleted: Both microglia (20) and astrocytes (21) support scrapie replication. Because astrocytes are the largest nonneuronal PrPc-expressing population of cells in the brain, they are probably the major source of continued PrPSc generation here. Prominent astrocytosis is a feature even of uninfected FVB mice (22) (fig. S5) [the predominant genetic background of these transgenic lines (13)], but here we found continued astrocytic proliferation in all animals with a notable increase over time (Fig. 3, U to X), in parallel to and correlating with the pattern of continued PrPSc deposition (Fig. 3, Q to T). We found that PrPSc deposits colocalized with astrocytes in the brains of infected mice with neuronal PrPc depletion with the use of dual immunofluorescent labeling for PrP and for the astrocytic marker glial fibrillary acidic protein (GFAP) (17), which was not seen in scrapie-infected control animals without PrP depletion (fig. S6). The fact that these mice remain asymptomatic indicates that even extensive extraneuronal PrPSc replication does not cause clinical disease or neurodegeneration in this model (23). In contrast, the dramatic neuronal loss and clinical signs seen in MloxP mice, where PrPc was not depleted in neurons, suggest that it is the conversion of PrPc to disease-associated isoforms specifically within neurons that is neurotoxic.

Also, we found that after PrPc depletion the infectivity in brain homogenates from scrapie-infected NFH-Cre/MloxP lagged behind that of animals without the Cre transgene by 1 to 2 log infectious units but eventually reached maximally infectious titers by 48 wpi (table S1 and fig. S7). This initial lag is consistent with loss of neuronal propagation of infectivity, with subsequent predominantly astrocytic propagation of infectivity, as occurs in GFAP-PrP mice (21). Indeed, high levels of prion infectivity are also found in microglia extracted from scrapie-infected mouse brains despite low levels of total PrP (24).

It is possible that the rate of PrPSc accumulation is also an important factor in development of symptomatic prion disease. In scrapie-infected wild-type and hemizygous Prnp0/+ mice, levels of total PrP (PrPc and PrPSc) increase markedly during the course of disease, compared to basal levels of PrP expression (1). We found a similar increase in scrapie-infected tg46 mice between 8 wpi and 18 wpi (–Cre, Fig. 3A) and tg37 mice between 8 wpi and 12 wpi (–Cre, Fig. 3B), where the rise to end-stage levels of PrPSc occurs in ∼2 weeks, at which stage the mice become clinically affected and die. Total PrP levels also increase in prion-infected mice with neuronal PrPc depletion, presumably reflecting continued glial PrPSc replication (Fig. 3, A and B, +Cre), but the increase to end-stage levels of PrPSc in NFH-Cre/tg37 takes ∼36 weeks and produces no symptoms.

In conclusion, we have demonstrated an intervention that prevents the development of symptomatic prion disease in mice with established CNS scrapie infection. Our strategy of arresting neuronal conversion of PrPc to PrPSc by depleting the former prevents progression from preclinical CNS prion infection to clinically manifest disease. PrP-null mice and Prnp0/0 brain tissue surrounding prion-infected Prnp+/+ neurografts are resistant to prion disease (1, 2, 7), because these tissues do not express PrPc and cannot propagate prions. In contrast, we have shown reversal of early neurodegenerative changes of CNS prion infection and long-term protection against neuronal loss despite continued prion replication and PrPSc depostion. Our results also argue against direct neurotoxicity of PrPSc, because the continued nonneuronal replication and accumulation of PrPSc throughout the brains of scrapie-infected mice is not pathogenic. Indeed, this may explain the lack of significant efficacy in vivo of therapeutic agents that reduce PrPSc accumulation in vitro. It appears that the conversion of PrPc to disease-related forms must occur within neurons to be pathogenic, consistent with the possibility that a toxic intermediate is generated within neurons during the conversion process (4). These findings provide a rationale for targeting PrPc as a therapeutic intervention in prion disease, which could prevent the progression to clinical disease in presymptomatic individuals infected with prions or with pathogenic PRNP mutations.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5646/871/DC1

Materials and Methods

Figs. S1 to S7

Table S1

References and Notes

View Abstract

Navigate This Article