No influence of amyloid-{beta}-degrading neprilysin activity on prion pathogenesis

Transmissible spongiform encephalopathies are neurodegenerative diseases that are characterized by the deposition of PrP^Sc, an abnormal, relatively protease-resistant isomer of a normal host-encoded cellular glycoprotein termed PrP^C. Prion infectivity copurifies with PrP^Sc, suggesting that this abnormal isomer is a component of the infectious agent (Aguzzi & Polymenidou, 2004). There are no differences in primary structure between PrP^C and PrP^Sc. Formation of PrP^Sc is thought to result from the conversion of PrP^C to PrP^Sc and its subsequent aggregation. The details of this process are not understood (Aguzzi & Polymenidou, 2004).

Formally, the accumulation of PrP^Sc in brain and extraneural organs is the result of the differential between its de novo generation and its clearance. Surprisingly, clearance of PrP^Sc is exceedingly efficient in vivo. Very large amounts of PrP^Sc injected into brain or peritoneum of Prnp-deficient mice (Büeler et al., 1992), which are incapable of replicating prions, resulted in a very fast decrease to concentrations below detectability (Büeler et al., 1993).

Although the relative resistance of PrP^Sc to protease digestion is the basis of laboratory assays for prion infections, it is obvious that PrP^Sc undergoes proteolytic processing in cell-culture models of prion diseases and in vivo (Büeler et al., 1993; Enari et al., 2001; Glatzel et al., 2003). Proteolytic processing of PrP^C and PrP^Sc differs and cleavage of PrP^Sc may lead to the occurrence of a 20 kDa PrP^Sc fragment referred to as C2, whereas cleavage of PrP^C leads to the formation of an 18 kDa fragment termed C1 (Chen et al., 1995; Jiménez-Huete et al., 1998). Although protein kinase C-dependent cleavage of PrP^C by ADAM10 and TACE has been reported, the protease responsible for PrP^Sc cleavage remains enigmatic (Checler & Vincent, 2002; Vincent et al., 2001). It was suggested that the alleged PrP^Sc-processing protease could belong to the metalloprotease family of enzymes, yet no studies have been undertaken until now to address the exact nature of this activity (Jiménez-Huete et al., 1998).

Neprilysin, a zinc metalloprotease, has been shown to catabolize neurotoxic amyloid-β protein that accumulates in Alzheimer's disease (Iwata et al., 2001; Mohajeri et al., 2004). The localization of neprilysin, a type II integral membrane protein, the relatively broad substrate specificity of this peptidase and the amyloid-β-degrading properties suggest a possible involvement of neprilysin in PrP^C/PrP^Sc catabolism (Turner, 2003). Involvement of neprilysin in degradation of aggregated PrP^Sc is further supported by similarities between PrP^Sc and amyloid-β. Both proteins are deposited in the extracellular space in the form of amyloid and both proteins may act as neurotoxins, either as aggregates or in the form of soluble oligomers (Aguzzi & Haass, 2003).

Here, we investigated whether PrP^C or PrP^Sc represents a substrate for the enzymic activity of neprilysin by exposing mice (i) entirely lacking neprilysin, (ii) expressing reduced amounts of neprilysin or (iii) overexpressing neprilysin, to infectious prions.

Generation and husbandry of transgenic mice.
Neprilysin-deficient mice were described previously (Lu et al., 1995). Genetically modified mice overexpressing neprilysin in the central nervous system (CNS) were generated by following this strategy: a haemagglutinin tag was fused in frame with the C terminus of human neprilysin cDNA (Shirotani et al., 2001). This fragment was then inserted into the XhoI site of the MoPrP expression vector (Borchelt et al., 1996). Transgenic mice were generated by standard pronuclear injection of the DNA construct into (C57BL/6xDBA/2) F1 embryos. The presence of the transgene was confirmed by PCR analysis of genomic DNA isolated from tail biopsies by using primers 5'-CAGAACTGAACCATTTCAAC-3' and 5'-CTATGATGGTGAGGAGCAGGACAAG-3', which were specific for the transgene. Transgenic founders were mated to C57BL/6 mice and one transgenic line, designated Nep-tg, was established from the F1 progeny on a mixed C57BL/6xDBA/2 background. All animal experiments and animal husbandry were performed in compliance with national guidelines.

Measurement of the Nep peptidase activity.
Neprilysin enzyme activity was measured in brain homogenates of Nep-tg (n=4) and wild-type (wt) littermates (n=4) as described previously, with minor modifications (Shirotani et al., 2001). In brief, 34 µg protein homogenized in 150 mM NaCl, 100 mM Tris/HCl, 1 % Triton X-100 (pH 7·8) was incubated for 1 h at 37 °C with 100 µM Z-Ala-Ala-Leu-p-nitroanilide (ZAAL-pNA; Bachem) in 50 mM HEPES buffer (pH 7·2). Thereafter, 0·8 mU leucine aminopeptidase (Sigma) was added to the reaction mixtures, incubated for an additional 20 min at 37 °C (Mohajeri et al., 2004) and OD₄₀₅ was measured. To inhibit neprilysin activity, 40 µM thiorphan (Sigma) was added for 5 min at room temperature before the addition of ZAAL-pNA. The results were confirmed by measuring several amounts of the Nep-tg samples diluted in similarly prepared brain homogenate of a Nep-KO mouse that exhibits no neprilysin enzyme activity (data not shown).

Scrapie infections.
Mice were infected intraperitoneally (i.p.) with 100 µl brain homogenate diluted in PBS and containing 6 or 3 logLD₅₀ intracerebral (i.c.) units of the Rocky Mountain laboratory (RML) scrapie strain (passage 5). For i.c. inoculations, 30 µl inoculum with 3x10⁵ or 3x10² LD₅₀ i.c. units was administered. Mice were monitored every second day and scrapie was diagnosed according to standard clinical criteria. Mice were sacrificed on the day of onset of terminal clinical signs of scrapie. The data for terminal disease are given as means±SD.

Infectivity bioassay with tga20 indicator mice.
Assays were performed on 1 % (w/v) brain homogenates. Tissues were homogenized in 0·32 M sucrose with a microhomogenizer (TreffLab) diluted in PBS/5 % BSA. When the solution appeared homogeneous, it was spun for 5 min at 500 g. Supernatants (30 µl) were inoculated i.c. into groups of five or six tga20 mice (Fischer et al., 1996). Incubation time until development of terminal scrapie sickness was determined and infectivity titres were calculated for 1 g inoculated tissue by using the relationship y=11·450·088x, where y is the number of i.c. LD₅₀ units and x is the incubation time (days) to terminal disease (Prusiner et al., 1982).

Western blot analysis.
Brain homogenates were adjusted to 8 mg protein ml¹ and treated with proteinase K where indicated (50 µg ml¹, 30 min, 37 °C). Total protein (50 µg) was electrophoresed through an SDS-PAGE gel (12 %). Proteins were transferred to nitrocellulose by wet blotting. Membranes were blocked with Tris/HCl-buffered saline/Tween 20 (TBST) containing 5 % Top Block, pH 7·4 (Sigma), incubated with mAb POM 1 to PrP (M. Polymenidou, M. Vey & A. Aguzzi, unpublished data) or with mAb 56C6 to neprilysin (Novocastra) and visualized by enhanced chemiluminescence (ECL; Amersham Biosciences) (Glatzel et al., 2001).

Histology and immunohistochemistry.
Paraffin-embedded sections from brain were stained with haematoxylin and eosin. Immunostaining to glial fibrillary acidic protein (GFAP) was performed with a rabbit antiserum against GFAP (1 : 300 dilution; DAKO) and detected with biotinylated swine anti-rabbit serum (1 : 250 dilution; DAKO) and diaminobenzidine (Sigma).

Histoblots.
Histoblots were performed according to Taraboulos et al. (1992). Briefly, frozen sections were mounted on uncoated glass slides and pressed immediately on a nitrocellulose membrane wetted in lysis buffer. Membranes were air-dried for at least 24 h. For detection, they were rehydrated in TBST and limited proteolysis was performed by using proteinase K concentrations of 50 and 100 µg ml¹ at 37 °C for 4 h. Blots were then denatured in 3 M guanidinium thiocyanate for 10 min and blocked for 1 h in 5 % non-fat milk. Incubation with primary antibody XN to PrP was carried out at a dilution of 1 : 2000 in 1 % non-fat milk at room temperature for 1 h (Glatzel & Aguzzi, 2000). Detection was accomplished with an alkaline phosphatase-conjugated goat anti-mouse antibody (1 : 2000). Visualization was achieved with nitro blue tetrazolium and BCIP according to the protocols of the supplier (Sigma).

Overexpression of neprilysin in the CNS of Nep-tg mice
In order to overexpress neprilysin in the CNS, we prepared a DNA construct in which the coding region of the human neprilysin gene was fused to a haemagglutinin tag, placed under the transcriptional control of the murine prion protein promoter and injected into the pronuclei of C57BL/6xDBA/2 zygotes (Fig. 1a).

(34K): Fig. 1. Generation of Nep-tg mice and analysis of PrPC expression in Nep/ and Nep-tg mice. (a) Schematic drawing of the DNA construct used for generation of transgenic mice. Relevant restriction sites are indicated. The locations of primers used to genotype transgenic mice are given as one-sided arrows (not to scale). (b) Neprilysin expression levels were significantly higher in total brain homogenates of mice overexpressing Nep-tg compared with wt littermates. A similar preparation of kidney total protein homogenate was loaded as a positive control. (c) Nep enzyme activity was on average 34 times higher in brain homogenates prepared from Nep-tg mice compared with age-matched littermates (black line represents mean, P0·029, MannWhitney U test). (d) PrPC expression in brains of Nep/ and Nep-tg mice, as assessed by Western blot analysis. Equal amounts of protein were loaded, as evidenced by comparable intensity of β-actin bands (uppermost band).

Transgenic founders were identified by PCR analysis and were bred to C57BL/6 mice. Transgene expression in the resulting mouse colony (mixed C57BL/6xDBA/2) was analysed in brain homogenates of 12-week-old F1 offspring. All subsequent analyses were performed with heterozygous offspring of one line (designated Nep-tg). Western blot analysis of total brain homogenates showed approximately 50-fold overexpression of neprilysin (Fig. 1b) and a 34-fold higher enzyme activity of neprilysin (Fig. 1c) in Nep-tg mice compared with wt controls.

PrP^C is not a substrate for the enzymic activity of neprilysin
Because of the relatively broad substrate specificity of neprilysin, we investigated whether PrP^C might represent a substrate for degradation by neprilysin (Turner, 2003). If this were correct, one would expect decreased protein levels of PrP^C in mice overexpressing neprilysin. Conversely, the complete lack of neprilysin expression in neprilysin-knockout mice could lead to increased protein levels of PrP^C in these mice.

In order to test these possibilities, we assessed the relative quantities of PrP^C in mice overexpressing and lacking neprilysin by Western blot analysis (Fig. 1d). There were no significant alterations of PrP^C protein levels between these transgenic lines of mice and appropriate wt controls, as evidenced by the ratios of the signal intensities of PrP^C to corresponding β-actin bands (0·21, 0·27 for Nep^/; 0·27, 0·24 for Nep-tg; 0·19, 0·23 for wt; Fig. 1d). The fact that neither depletion nor overexpression of neprilysin seems to influence the turnover of PrP^C showed convincingly that PrP^C did not represent a substrate for the enzymic activity of neprilysin.

Depletion of neprilysin does not alter incubation times to terminal prion disease
Mice deficient for neprilysin (Nep^/) and mice with decreased levels of neprilysin expression (Nep^+/) were challenged with various amounts of infectious prions and the incubation time to terminal prion disease was measured. After high-dose (3x10⁵ LD₅₀) or low-dose (3x10² LD₅₀) i.c. prion injection, there was no significant difference in incubation times compared with wt mice (high dose: 164±4 days, n=6 for wt; 161±4 days, n=7 for Nep^+/; 165±7 days, n=7 for Nep^/ mice; low dose: 192, 197 days, n=2 for wt; 200±17 days, n=5 for Nep^+/; 188±10 days, n=8 for Nep^/ mice; Fig. 2).

(16K): Fig. 2. Incubation times in prion-inoculated Nep+/, Nep/ and wt mice. Survival plots displaying the incubation time (days) until development of terminal scrapie in Nep/ (black boxes), Nep+/ (dark-grey boxes) and wt (light-grey boxes) mice inoculated i.p. (a, b) or i.c. (c, d) with prions. The dose of inoculum is expressed above individual graphs. There is no significant difference in incubation times of mice inoculated either i.c. or i.p.

Upon high-dose (6 logLD₅₀) and low-dose (3 logLD₅₀) i.p. prion challenge, Nep^/, Nep^+/ and wt mice developed scrapie with similar incubation times (high dose: 208±18 days, n=8 for wt; 211±20 days, n=15 for Nep^+/; 202±9 days, n=15 for Nep^/ mice; low dose: 239±10 days, n=4 for wt; 225±5 days, n=6 for Nep^+/; 242±20 days, n=9 for Nep^/ mice; Fig. 2). These data demonstrate clearly that prion pathogenesis and neuroinvasion of prions are unaltered by the relative or absolute lack of neprilysin.

The lack of an effect of neprilysin on prion pathogenesis is further corroborated by biochemical and histological analysis of terminally sick mice. Immunoblot analysis confirmed that there were comparable amounts of PrP^Sc in the brains of terminally sick wt, Nep^+/ and Nep^/ mice (Fig. 3). We investigated the distribution of protease-resistant PrP^Sc in wt, Nep^+/ and Nep^/ mice by histoblot analysis of CNS tissue (Fig. 4). This analysis shows an even, cortically accentuated distribution of PrP^Sc throughout the entire cerebral hemispheres. Accordingly, histological analysis of mice with complete or partial absence of neprilysin showed unaltered patterns of vacuolization and reactive astrogliosis within the CNS tissue when assayed 100 days following i.c. prion challenge (Fig. 5).

(34K): Fig. 3. Western blot analysis of brain tissue from terminally scrapie-sick Nep/, Nep+/ and wt mice. Western blots of CNS tissue electrophoresed before () or after (+) digestion with proteinase K (PK). Similar amounts of PK-resistant prion protein (PrPSc) were detected in the brains of Nep+/, Nep/ and wt mice that had developed scrapie (terminally sick) after the time points indicated [days post-i.c. injection (d.p.i.) of 3x105 LD50 prions].

(85K): Fig. 4. Histoblot analysis of Nep/, Nep+/ and wt mice following i.c. scrapie administration. Histoblots showing immunoreactive PrP in brains before (first column) and after (second and third columns) digestion with increasing concentrations of proteinase K (50 and 100 µg ml1). Prnp0/0 mice (first row) do not show any signal for PrPC or PrPSc, whereas control mice (second row) show proteinase K-sensitive PrPC, but no proteinase K-resistant PrPSc. The accumulation of PrPSc in brains of terminally scrapie-sick Nep+/+, Nep/ and wt mice, taken 157 days post-i.c. prion injection, is comparable.

(90K): Fig. 5. Histological analysis of brain in Nep/, Nep+/ and wt mice 100 days following i.c. scrapie administration. Occasional spongiform changes (ac) and only mild astrogliosis (df) were visible in the hippocampal neuronal ribbon of subclinically prion-diseased Nep+/, Nep/ and wt mice. GFAP, Immunohistochemical stain for the astrocytic marker, glial fibrillary acidic protein; HE, haematoxylin/eosin stain. Bar, 50 µm.

Similar infectivity titres in Nep^/ and wt mice
In order to assess whether the absence of neprilysin leads to alterations in the titre of infectious prions, we determined prion titres of CNS tissue 100 days following i.c. prion injection by transmission of homogenized tissue into highly prion-susceptible tga20 indicator mice (Fischer et al., 1996) and comparison of incubation times to a calibration curve. Infectivity in brains of Nep^/ mice and wt mice was comparable (Table 1). Therefore, lack of neprilysin does not impair the kinetics of prion replication within the CNS.

Table 1. Prion infectivity titres in Nep/ and wt mice 100 days following i.c. prion challenge

Overexpression of neprilysin does not alter CNS prion pathogenesis
If neprilysin plays a role in degrading PrP^Sc, genetically modified mice overexpressing this enzyme would be expected to harbour decreased levels of PrP^Sc in their CNS upon prion inoculation. We investigated this possibility by challenging Nep-tg (n=6) and wt (n=6) mice with infectious prions (3x10⁵ LD₅₀, i.c. injection). All mice were terminated 100 days following prion injection and PrP^Sc levels were determined by Western blot analysis of brain samples. Although there was some heterogeneity in the level of PrP^Sc accumulation within the two groups of mice, there was no significant difference in the level of PrP^Sc accumulation between neprilysin-overexpressing and wt mice (Fig. 6). This provides further evidence that neprilysin is not involved in degradation of PrP^Sc.

(33K): Fig. 6. Western blot analysis of brain tissue from Nep-tg and wt mice 100 days following i.c. scrapie administration. Western blots of CNS tissue electrophoresed natively () or after digestion with proteinase K (PK) (+). Although individual mice showed variable amounts of PK-resistant prion protein (PrPSc), no significant differences were detected in the brains of Nep-tg or wt mice.

Aggregation of proteins in the CNS leads to a number of neurodegenerative diseases (Aguzzi & Haass, 2003). Lately, it has become clear that, in many instances, the aggregation of misfolded proteins, which may be the cause of dementia, is a dynamic process that is determined by a balance of generation and clearance (Iwata et al., 2001). Both Alzheimer's disease and prion diseases are characterized by extracellular accumulation of amyloidogenic proteins (Aguzzi & Haass, 2003). Several authors have pointed to similarities between physiological and pathological cleavage of the amyloid-β precursor protein and the prion protein (Checler & Vincent, 2002). Yet, whilst we understand the processing cascade leading to deposition of amyloid-β in Alzheimer's disease in some detail, relatively little is known about the equivalent processes in prion diseases (Aguzzi & Haass, 2003; Checler & Vincent, 2002). In the case of PrP^C, physiological cleavage by ADAM10 and TACE has been reported and it was speculated that cleavage of PrP^C by an unknown protease might result in the generation of a toxic PrP fragment (Vincent et al., 2001). For PrP^Sc, the situation is different. Cell-culture studies have shown that, once generation of newly formed PrP^Sc is arrested, PrP^Sc is degraded rapidly, leading to the disappearance of PrP^Sc within a few days (Enari et al., 2001). Similar processes seem to be active in vivo, leading to the generation of fragmented PrP^Sc in CNS tissue of patients suffering from prion disease, and it has been suggested that the alleged PrP^Sc-degrading enzyme belongs to the metalloprotease family of proteolytic enzymes (Jiménez-Huete et al., 1998). In a recent study, it was shown that cysteine proteases seem to be involved in degradation of PrP^Sc in lymphoid cells and, to a lesser extent, in neuronal cells (Luhr et al., 2004). In the same study, metalloproteases did not seem to play a substantial role in PrP^Sc degradation, raising the possibility that these processes may be regulated in a cell-specific manner.

As a candidate, we have investigated whether neprilysin is involved in degradation of PrP^C or PrP^Sc. This was achieved by subjecting genetically modified mice expressing either reduced levels of neprilysin or no neprilysin at all and mice overexpressing neprilysin to infectious prion preparations. We did not find any evidence that PrP^C or PrP^Sc represents a target for neprilysin. On the contrary, the evidence that neprilysin plays no role at all in prion/PrP^Sc is overwhelming: all investigated parameters of prion pathogenesis, including the amount and the distribution of PrP^Sc in infected brains at various time points, the amount of infectivity in the CNS and neuropathological changes of mid- and end-stage brains, were identical in mice lacking or overexpressing neprilysin and wt mice.

Proteins belonging to the metalloprotease family, most of which are expressed in CNS tissue, seem to be involved in degradation of extracellular deposited amyloid in vivo. It has been shown that neprilysin is the most potent amyloid-degrading metalloprotease (Iwata et al., 2001). As neprilysin displays a broad substrate specificity and is localized subcellularly in the vicinity of PrP (Iwata et al., 2001; Turner, 2003), it is surprising that it has no impact on prion pathogenesis.

The findings described here do not negate the involvement of other metalloproteases in PrP^Sc catabolism. Indeed, studies focusing on possible substrates of metalloproteases provided evidence that matrix metalloprotease MMP-9 is able to degrade amyloid-β and might possess PrP^Sc-degrading properties (Backstrom et al., 1996). The availability of reliable cell-culture models for prion diseases, in combination with high-throughput tools to assess prion infectivity, should significantly facilitate the investigation of PrP^Sc proteolysis (Enari et al., 2001; Klöhn et al., 2003).

Given that our understanding of PrP^Sc metabolism is far from complete, this field of prion research continues to represent an important target for the future. In view of the fascinating developments in Alzheimer's disease research, where insights into amyloid-β precursor protein processing have led to the identification of the mechanisms of amyloid-β generation and, based on this knowledge, to the development of therapeutic strategies, the investigation of the proteolytic cascade leading to the removal of aggregated PrP^Sc might hold the key to the discovery of novel prophylactic or therapeutic strategies against prion diseases (Aguzzi & Haass, 2003).

This work was supported by grants of the Bundesamt für Bildung und Wissenschaft (Biomed), the Swiss National Foundation and the NCCR on neural plasticity and repair. M. G. is supported by a career development award of the University of Zürich and by a grant from the Stammbach Foundation. We thank the Desirée und Niels Yde Foundation, the EMDO Foundation and the EU(TSELAB) for support. We also thank Mauri Peltola and Marianne Koenig for technical help.

Footnotes

^,†, M. Hasan Mohajeri² †These authors contributed equally to this work.