Upregulation of the Genes Encoding Lysosomal Hydrolases, a Perforin-Like Protein, and Peroxidases in the Brains of Mice Affected with an Experimental Prion Disease

Abstract

In an attempt to identify the molecules involved in the pathogenesis of prion diseases, we performed cDNA subtraction on the brain tissues of mice affected with an experimental prion disease and the unaffected control. The genes identified as being upregulated in the prion-affected brain tissue included those encoding a series of lysosomal hydrolases (lysozyme M and both isoforms of β-N-acetylhexosaminidase), a perforin-like protein (macrophage proliferation-specific gene-1 [MPS-1]), and an oxygen radical scavenger (peroxiredoxin). Dramatic increases in the expression level occurred at between 12 and 16 weeks after intracerebral inoculation of the prion, coinciding with the onset of spongiform degeneration. The proteinase K-resistant prion protein (PrP^Sc) became detectable by immunoblotting well before 12 weeks, suggesting a causal relationship between this and the gene activation. Immunohistochemistry paired with in situ hybridization on sections of the affected brain tissue revealed that expression of the peroxiredoxin gene was detectable only in astrocytes and was noted throughout the affected brain tissue. On the other hand, the genes for the lysosomal hydrolases and MPS-1 were overexpressed exclusively by microglia, which colocalized with the spongiform morphological changes. A crucial role for microglia in the spongiform degeneration by their production of neurotoxic substances, and possibly via the aberrant activation of the lysosomal system, would have to be considered.

Prion diseases, including Creutzfeldt-Jakob disease (CJD), Gerstmann-Sträussler-Scheinker syndrome (GSS), and fatal familial insomnia of humans, scrapie of sheep and goats, and bovine spongiform encephalopathy of cattle, are infectious neurodegenerative disorders. The etiological agent, prion, is postulated to consist mainly of a proteinase K-resistant isoform of prion protein (PrP^Sc) which is generated by posttranslational conversion from the proteinase K-sensitive normal version (PrP^C), a membrane glycoprotein expressed constitutively on the neuronal cell surface and to a lesser extent on various other tissues, including glial and lymphoreticular cells (25). The constitutive conversion results in the tremendous accumulation of PrP^Sc in the prion-infected brain. The accumulated PrP^Sc colocalizes well with pathological lesions, and levels are maximal in the brain at the terminal stage of the disease (8). Homozygous disruption of the Prnp gene encoding PrP^C renders mice resistant to prion, and the animals are no longer capable of generating PrP^Sc (3, 26, 29). These findings have indicated an essential role for the accumulation of PrP^Sc in the pathogenesis of prion diseases.

The major pathological characteristics in the central nervous system in prion diseases are spongiform neurodegeneration and gliosis (14, 27). Several reports have indicated the involvement of an apoptotic process in neuronal cell death (13, 15, 34). For instance, Giese et al. (13) demonstrated apoptosis of neurons in the brain of a mouse with experimental prion disease by using the in situ end-labeling technique. A synthetic peptide corresponding to the hydrophobic region of PrP (PrP106-126) was shown to induce apoptosis in primary cultured neurons, suggesting that the apoptotic event might represent the neurotoxicity of PrP^Sc itself (11). A role for glial cells in the pathogenesis of prion diseases has also been presumed since gliosis precedes the neurodegeneration. Increased levels of several glial-cell-derived cytotoxic cytokines, including tumor necrosis factor alpha, interleukin-1α (IL-1α), and IL-1β, were noted in the scrapie-infected mouse brain (4). Activated microglia were found to colocalize with PrP plaques in the brain tissues of patients with GSS (20) and pathological lesions in an experimental mouse model of scrapie (34). Moreover, Brown et al. (2) recently demonstrated that the neurotoxic effect of PrP106-126 in vitro requires the presence of microglia. These lines of evidence have suggested that microglia are actively involved in the pathogenesis of prion diseases, but precise molecular mechanisms remain unclear.

In the present study, we used a cDNA subtraction technique and successive Northern blotting in an attempt to identify the molecules involved. We demonstrated the upregulation of genes for a series of lysosomal hydrolases and a perforin-like protein in the microglia, and peroxiredoxin (Prx) in the astrocytes, of prion-affected brain tissues. The mode of expression of these genes indicated that aberrant activation of lysosomal enzymes and oxidative stress induced by the accumulated PrP^Sc are possible mechanisms involved in the neuronal cell death of the diseases and that microglia are the cell type most actively involved in this process.

MATERIALS AND METHODS

Preparation of infected mouse brains.

Outbred 4-week-old ddY mice were inoculated intracerebrally with a mouse-adapted CJD agent (Fukuoka-1 strain) (33) equivalent to 10^5.5 50% lethal doses per head. The inoculated mice were sacrificed at several time points postinoculation, and their brain tissues were removed, frozen immediately, and stored until use. Control brain tissues were similarly prepared from mice inoculated with phosphate-buffered saline (PBS). All of the animal experiments were conducted in the biohazard prevention area (P3) of the Laboratory Animal Center of our institution and in accordance with the Guidelines for Animal Experimentation of Nagasaki University.

cDNA subtraction.

Total RNA was isolated from the pooled three brain tissues of ddY mice at the terminal stage of the disease and three control mice by using the acid guanidinium thiocyanate-phenol chloroform method and then subjected to an oligo(dT)-cellulose column (mRNA purification kit; Pharmacia Biotech) for the purification of mRNA. The cDNA subtraction was performed by using a PCR-Select cDNA Subtraction Kit (CLONTECH). “Driver” and “tester” double-stranded cDNAs were synthesized from the mRNAs of the prion-infected and control brains, respectively, in accordance with the manufacturer's recommendations, and digested with RsaI to make blunt ends. The digested driver cDNA was divided into two and ligated with adapters 1 and 2, respectively. The subtraction was carried out by using two different hybridization processes, as follows. Each of the adapter-ligated cDNAs was heat denatured and annealed to excess heat-denatured tester cDNA (first hybridization), and then the two samples from the first hybridization were mixed together again with additional excess heat-denatured tester cDNA (second hybridization). This type of subtraction was called “positive” subtraction. “Negative”-subtracted cDNA was made by the same procedure, except that driver and tester cDNAs were prepared from the control and infected brains, respectively. PCR amplification was performed twice for the subtracted cDNA. All of the primers (PCR primer 1 and nested PCR primers 1 and 2) for the PCR were provided in the kit. After the PCR amplification, the amplified subtracted cDNAs were digested by EagI and cloned into a NotI site of pBluescript II SK(−) (Stratagene).

Colony hybridization.

Ampicillin-resistant colonies of XL1-Blue (Stratagene), which had been transformed by pBluescript II KS(−) carrying cDNA inserts, were allowed to grow on nitrocellulose replicative membranes at 37°C overnight. The membranes were denatured in 1.5 M NaCl–0.5 M NaOH for 7 min, neutralized in 1.5 M NaCl–0.5 M Tris-HCl (pH 7.2)–1 mM EDTA for 3 min twice, and then UV irradiated for DNA fixation. One of the membranes was hybridized with a ³²P-labeled positive-subtracted probe, and the other was hybridized with a ³²P-labeled negative-subtracted probe in a buffer containing 50% formamide–5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–2× Denhardt's solution–0.1% sodium dodecyl sulfate (SDS) and 0.1 mg of heat-denatured salmon sperm DNA per ml at 42°C overnight. After a washing with 2× SSC–0.5% SDS at room temperature for 20 min and with 0.2× SSC–0.5% SDS twice at 65°C for 20 min, the membranes were exposed to an X-ray film (Konica) at −80°C.

Dot blot hybridization.

Subtracted cDNA was amplified by PCR with the T7 and T3 primer pair. The same amount of the amplified cDNA in 0.4 N NaOH–10 mM EDTA was dot blotted onto two different replicative Hybond N⁺ membranes. The conditions for hybridization, washing, and exposure were the same as those described above.

Northern blot hybridization.

Total RNAs or mRNAs were separated on a formaldehyde-denaturing agarose gel and transferred onto a Hybond N⁺ membrane in 20× SSC overnight. After fixation of RNA on the membrane by UV light, hybridization was carried out as described above. Subtracted cDNA clones were labeled with ³²P and used as probes. The other DNA probes were prepared by PCR according to the sequences deposited in the database: α-subunit of β-N-acetylhexosaminidase (HEXα) (GenBank number X64331), Cu-Zn superoxide dismutase (Cu/Zn SOD) (number X06683), glutathione peroxidase (GSHPx) (number 03920), macrophage 23-kDa protein (PAG1) (number D16142), glial fibrillary acid protein (GFAP) (number K01347), glycerol-3-phosphate dehydrogenase (G3PDH) (number M25558), and antioxidant protein 1 (Aop1 or MER5) (EMBL M28723). Intensities of the signals were measured by an image analyzer (BAS 2000; Fuji Film, Tokyo, Japan), and the level of upregulation was estimated by comparing these values between prion-affected and normal control brains after standardization with the signals by the internal control probe, G3PDH.

Western blotting.

Twenty-percent homogenates of the brain tissues obtained from mice sacrificed at various time points after the inoculation with the prion were prepared in a buffer containing 40 mM Tris-HCl (pH 7.5)–10 mM NaCl–6 mM MgCl₂ and digested with DNase I at 37°C for 1 h. The homogenates were then solubilized by the addition of sarcosyl to a final concentration of 1% and digested with proteinase K (0.1 mg/ml) at 37°C for 1 h. The digested proteins were separated by SDS–12.5% polyacrylamide gel electrophoresis under a reduced condition and then transferred electrically onto a nitrocellulose membrane. The membrane was treated with 5% dry milk-containing PBST (PBS plus 0.2% Tween 20) and then washed in PBST. PrP^Sc was detected by using a rabbit antiserum against a synthetic peptide corresponding to the N-terminal region of hamster PrP (31) at a dilution of 1:2,000 in 0.2% bovine serum albumin-containing PBST and ¹²⁵I-labeled protein A (Amersham).

DNA sequencing.

Sequences of subtracted cDNAs were determined by the chain termination reaction method by using Texas Red T3 and T7 primers (Amersham) and the Thermo Sequenase premixed cycle sequencing kit (Amersham) in accordance with the manufacturer's recommendations. A nucleic acid homology search was performed by using the BLAST program (National Institutes of Health, Bethesda, Md.).

In situ hybridization.

Linear template DNA at a suitable site for antisense or sense strand cRNA was prepared from each plasmid. The probes were labeled by using digoxigenin (DIG)-UTP (Boehringer Mannheim Biochemicals, Indianapolis, Ind.) with T7 or T3 polymerase (BRL). The brains were fixed for 16 h in 4% buffered paraformaldehyde (pH 7.4) at 4°C and embedded in paraffin. Coronal 5-μm sections were taken at the level of the rostral hippocampus and geniculate nuclei in order to include the caudal median eminence, the arcuate hypothalamic nucleus, and the CA1-4 regions of the hippocampus. Sagittal sections of cerebellum and brain stem were also investigated. To confirm the reproducibility, brain sections derived from four to five mice similarly affected with the prion were simultaneously examined in every hybridization. The sections were mounted onto slides treated with 2% 3-aminopropyltriethoxysilane, deparaffinized, digested with 8 mg of pepsin per ml for 10 min at 37°C, and soaked for 10 min in 0.25% acetic anhydride–0.1 mM triethanolamine hydrochloride (pH 8.0)–0.9% NaCl. A DIG-labeled probe (500 ng/ml) was added to the hybridization buffer composed of 50% formamide, 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 0.6 M NaCl, 0.5 mg of yeast tRNA per ml, 0.25 mg of salmon sperm DNA per ml, 1% skim milk, 0.25% SDS, and 5× Denhardt's solution. After hybridization at 50°C for 16 h, the slides were washed several times in 4× SSC and immersed in 50% formamide–2× SSC at 50°C for 30 min. The sections were then treated with 20 μg of RNase A per ml at 37°C for 30 min and finally washed in 0.2× SSC at 60°C for 20 min. Hybridization signals were detected by immunological detection with alkaline phosphatase-conjugated anti-DIG Fab fragments (diluted 1:500; Boehringer) by using nitroblue tetrazolium–5-bromo-4-chloro-3-indolylphosphate (BCIP) as the chromogenic substrate.

Immunocytochemistry.

Deparaffinized sections were incubated in 0.3% H₂O₂ solution for 30 min at room temperature to abolish endogenous peroxidase activity. After treatment with normal rabbit serum, the tissue sections were allowed to react overnight at 4°C with anti-GFAP (1:50 [Dako]) and anti-F4/80 antigen (1:20 [Serotec, Ltd., Oxford, United Kingdom]). Subsequently, the tissue sections were incubated in the secondary antibody (biotinylated rabbit anti-mouse, diluted 1:500, or rabbit anti-rat immunoglobulin G [mouse absorbed], diluted 1:500; Dako) at room temperature for 30 min. Avidin-conjugated horseradish peroxidase (1:500; Dako) was applied, and the preparation was incubated for 30 min. The antibody-bound peroxidase was revealed with 0.04% diaminobenzidine (Sigma Chemical Co., St. Louis, Mo.) or 3-amino-9-ethyl carbazole substrate chromogen (Dako). The tissue sections derived from four to five mice similarly affected with the prion were simultaneously examined in every experiment to confirm the reproducibility.

RESULTS

The genes for lysosomal hydrolases, a perforin-like molecule, and Prx are upregulated in the prion-affected brain.

cDNA subtraction was performed in prion-affected mouse brains and control brains. Initially, 704 ampicillin-resistant colonies that had been transformed with the subtracted cDNAs were screened by colony hybridization by using two different positive- and negative-subtracted cDNA probes. As result of this screening, 67 colonies were found to be positive, showing stronger signals with the positive-subtracted cDNA probe than with the negative-subtracted cDNA probe. As a secondary screening, the cDNA inserts of the 67 colonies were amplified by PCR by using a T3 and T7 primer pair. The same amount of each of the amplified cDNAs was blotted onto duplicate membranes and was then analyzed by dot blot hybridization with the same probes. Of 67 clones, 42 reproduced positive results, all of which were subsequently sequenced from both ends. Finally, tertiary screening by Northern blot targeting the same RNA samples used in the subtraction confirmed 15 clones consisting of five distinct genes, which were upregulated in the prion-affected brain tissues. These included the genes for lysozyme M (found in nine clones), the β-subunit of β-N-acetylhexosaminidase (HEXβ, one clone), macrophage proliferation specific gene-1 (MPS-1, four clones), and Prx (one clone). The expression levels of lysozyme M, HEXβ, MPS-1, and Prx were ca. 30, 10, 40, and 5 times more abundant, respectively, in the infected brains than in the control brains (Fig. 1A). The secondary screening and successive sequencing also identified the genes for GFAP (three clones) and apolipoprotein E (one clone). The remaining 23 genes selected in the secondary screening showed no difference in the expression level or else their expression was undetectable by Northern blotting. Subsequently, we performed Northern blotting on the mRNA for another subunit of HEX (HEXα) and noted upregulated gene expression in the prion-affected brains (Fig. 1A). Upregulation of genes for two subunits of HEX suggested increased expression of both of the two HEX isozymes, HEXA and HEXB. The former is a heterodimer of α- and β-subunits, and the latter is a homodimer of the β-subunit (22, 23).

FIG. 1.

Northern blot analysis of the genes upregulated in prion-infected mouse brains. (A) The levels of transcripts for lysozyme M (LM), HEXα, HEXβ, MPS-1, and Prx are compared between the prion-infected (I) and control (C) brain tissues. Basically, 20 μg of total RNA was blotted on each lane, but 1 μg of poly(A)⁺ RNA was used for the MPS-1 probe. Each membrane was hybridized with each ³²P-labeled probe (ca. 2 × 10⁶ cpm in 1 ml of hybridization solution) overnight and, after being washed, was exposed for 24 h. The integrity and quantity of RNA were verified with G3PDH probe on the same membranes (m). (B) The levels of transcripts for several radical oxygen scavengers, including common-type (selenium-dependent) GSHPx, Aop1, mouse macrophage 23-kDa protein (PAG), catalase, and Cu/Zn SOD, are similarly compared.

Expression of the genes encoding molecules functionally related to Prx in the prion-affected brain.

Since Prx functions as a scavenger of reactive oxygen species (ROS) (16), its upregulation in the prion-affected brains was likely to be a consequence of ROS activation. We therefore examined the expression of genes for other radical oxygen scavengers including common-type (selenium-dependent) GSHPx, Aop1 (MER5), PAG, catalase, and Cu/Zn SOD. As shown in Fig. 1B, the expression of GSHPx-1 was also upregulated in the prion-affected brain but to a lesser extent (a <2-fold increase). On the other hand, there was no difference in the expression levels of the genes for Aop1, PAG, catalase, and Cu/Zn SOD between the diseased and unaffected brains.

Upregulation of the genes for lysozyme M, HEX, MPS-1, and Prx correlate with pathologic changes.

In our experimental system, the mice inoculated with the CJD prion (Fukuoka-1 strain) began to reveal vacuolar neurodegeneration in their brain tissues at ca. 15 weeks postinoculation (p.i.) and developed characteristic neurological signs at 20 weeks p.i. The expression kinetics of the genes for lysozyme M, HEXβ, MPS-1, and Prx were examined by Northern blotting by using total or poly(A)⁺ RNA isolated from the brain tissues of mice sacrificed at 0, 2, 4, 6, 8, 10, 12, 16, and 18 weeks p.i. As shown in Fig. 2A, all of the genes showed basically the same expression kinetics. There was no significant increase in the expression level in the inoculated brain tissues by 12 weeks p.i., but a dramatic upregulation occurred at between 12 and 16 weeks, with the expression reaching a maximal level at 18 weeks p.i. The gene expression kinetics correlated well with the development of pathologic changes in the brain tissue. The astrocyte-specific GFAP showed basically the same expression profile. As shown in Fig. 2B, PrP^Sc was detectable by immunoblotting in the brain tissues at 12 weeks p.i., suggesting that the PrP^Sc accumulation occurred before the gene activation.

FIG. 2.

Kinetic analysis of the gene upregulation and PrP^Sc accumulation in the brain of mice inoculated with prion. (A) The levels of transcripts for lysozyme M (LM), HEXβ, MPS-1, Prx, and GFAP in the brains of mice sacrificed at the indicated time points after inoculation of prion were analyzed by Northern blotting. Basically, 20 μg of total RNA were blotted on each lane, but 1 μg of poly(A)⁺ RNA was used for the MPS-1 probe. The integrity and quantity of RNA were verified with a G3PDH probe. (B) Western blot analysis of PrP^Sc accumulation in the infected mouse brains at different time points of infection. For more details see Materials and Methods.

Microglia colocalize with the spongiform degeneration in the brain tissue.

Pathological findings in the brain tissues at the end stage of the disease (18 weeks p.i.) are shown in Fig. 3. The fine vacuolation and microcystic cavitation were concentrated in the white matter of the corpus callosum, fimbria hippocampus, and internal and external capsula (Fig. 3A), cerebellum, and brain stem. Fine vacuolation was also seen in the gray matter (cerebral cortex, hippocampus, and thalamus). Gliosis was confirmed by immunohistochemistry with anti-GFAP and anti-F4/80 antibodies. Astrocytes with GFAP immunoreactivity were observed in both white and gray matter (Fig. 3B), along the blood vessels, and in the pia mater. The distribution pattern of F4/80-immunoreactive microglia was distinct from that of astrocytes, being strongest in the white matter and the thalamus and considerably weaker in the cerebral cortex and the hippocampus (Fig. 3C). Of particular interest is the finding that the processes of F4/80-immunoreactive cells were frequently found surrounding neurons (Fig. 3D) and vacuoles (Fig. 3E). Spongiform changes were barely detectable, and only a few cells revealed the immunoreactivities in the brain of uninfected mice (data not shown).

FIG. 3.

Histological findings in the prion-infected mouse brain at the end stage of disease (18 weeks p.i.). (A) Hematoxylin-eosin staining. Many vacuoles are observed mainly in the white matter (corpus callosum, internal and external capsula, and fimbria hippocampus) (original magnification, ×12.5). (B) Immunohistochemical staining for GFAP with hematoxylin counterstaining (original magnification, ×12.5). GFAP-immunoreactive cells (astrocytes) are present in the cerebral cortex, hippocampus, and thalamus, as well as in the white matter. (C to E) Immunostaining for F4/80 with hematoxylin counterstaining. Immunoreactivity for F4/80 (microglia/macrophages) concentrates to the white matter and the thalamus (C) (original magnification, ×12.5). The processes of microglia are frequently observed around neuron (arrowhead in panel D) and spongiform holes (arrows in panel E) (original magnification, ×100). cc, corpus callosum; fh, fimbria hippocampus; ic, internal capsule; hipp, hippocampus; tha, thalamus.

Determination of cell types expressing the upregulated genes.

The distribution of cells expressing lysozyme M, HEXα and -β, MPS-1, and Prx was analyzed by in situ hybridization on sections of the brain tissue at 18 weeks p.i. (Fig. 4). The hybridization signals for lysozyme M mRNA were intense in the corpus callosum, fimbria hippocampus, internal and external capsula, thalamus, cerebellar medulla, and brain stem, where many vacuoles were present (Fig. 4A). They were also scattered throughout the cerebral cortex and hippocampus. Strong HEXβ mRNA signals were detected in similar sites to lysozyme M mRNAs, but the signals for HEXβ mRNA were more widely distributed throughout the whole brain (Fig. 4B). The distribution of MPS-1 (Fig. 4C) and HEXα (data not shown) mRNAs was almost the same as that of lysozyme M mRNA, although the expression levels were very weak. The cells expressing these genes were often found surrounding vacuoles, as seen with the F4/80-immunoreactive microglia (Fig. 4E to G). Intense Prx mRNA hybridization signals were distributed ubiquitously and were seen in the cerebral cortex and hippocampus as well as the white matter. The Prx mRNA-expressing cells were morphologically larger than those expressing lysozyme M, HEXβ, and MPS-1 mRNAs (Fig. 4D and H). In the uninfected control brain, the cells expressing these genes were barely detectable except for a few cells with HEXβ mRNA (Fig. 4I to L).

FIG. 4.

Distribution of cells expressing the upregulated genes. Upper panel shows the distribution of mRNAs for lysozyme M (A), HEXβ (B), MPS-1 (C), and Prx (D) with Nuclear Fast Red counterstaining (magnification, ×16). Lysozyme M and MPS-1 mRNAs are seen mainly in the corpus callosum (cc), internal capsula (ic), fimbria hippocampus (fh), and thalamus (tha). HEXβ mRNA is more widely distributed. Intense signals of Prx mRNA are present in the hippocampus (hipp), as well as in the white matter. Panels E to H are the high-power views of the squared areas in panels A to D, respectively (magnification, ×50). Panels I to L indicate mRNAs for lysozyme M, HEXβ, MPS-1, and Prx, respectively, barely detectable in the uninfected control brain tissues (magnification, ×16).

To determine the cell type, double staining with in situ hybridization and immunohistochemistry was carried out. The F4/80-immunoreactive cells exclusively revealed signals for lysozyme M, HEXβ, and MPS-1 mRNAs, indicating that these genes were expressed by microglia (Fig. 5E to G). On the other hand, the signal for Prx mRNA was restricted to GFAP-positive astrocytes (Fig. 5D) and was barely detectable in other cell types, including neurons (data not shown).

FIG. 5.

Identification of cell types expressing the upregulated genes. The upper and lower panels show mRNAs (blue) for lysozyme M (A and E), HEXβ (B and F), MPS-1 (C and G), and Prx (D and H) as shown by in situ hybridization paired with the GFAP (A to D; red) and F4/80 (E to H; brown) immunoreactivity, respectively (original magnification, ×200). Lysozyme M, HEXβ, and MPS-1 mRNAs are expressed in F4/80-positive cells but not in GFAP-positive cells. On the other hand, the expression of Prx mRNA is found on GFAP-immunoreactive cells but not on F4/80-immunoreactive cells.

DISCUSSION

The mechanisms of neuronal cell death involved in prion diseases, as well as in other neurodegenerative conditions, including Alzheimer's disease, remain to be elucidated. A role for glial cells, and microglia in particular, in the neurodegeneration has been suggested (2, 4, 20). In the present study, cDNA subtraction and Northern blotting efficiently identified a series of upregulated genes, including those encoding lysozyme M, HEX, and MPS-1, in the microglia of prion-affected brains. The upregulation of these genes correlated well with the development of spongiform degeneration in the brain after inoculation of the prion. In contrast to the ubiquitous distribution of astrocytes throughout the infected brain, the accumulation of microglia was restricted to the site of pathological changes, and they were frequently found surrounding vacuoles or degenerating neurons. Furthermore, the accumulation of PrP^Sc in the brains of inoculated mice well preceded the gene activation in microglia and the development of pathological changes, suggesting that the accumulated PrP^Sc itself caused the microglia activation in the prion-infected brains. Supporting this hypothesis is a recent report showing the colocalization of PrP^Sc deposition and vacuolating changes in the brain tissues of mice intracerebrally inoculated with the Fukuoka-1 prion (10).

The augmented expression of certain lysosomal hydrolases such as lysozyme M and both isoforms of HEX in microglia suggests that the lysosomal system is highly activated in prion diseases. Consistent with this idea is the previously reported finding that another lysosomal hydrolase, cathepsin S, exhibits increased expression in the scrapie-infected mouse brain (7). The altered production and distribution of lysosomal hydrolases have also been reported in other neurodegenerative disorders (5, 19, 21). The physiological function of lysozyme M is unclear, but it is known to degrade peptidoglycan components of the bacterial cell wall, and it is a useful marker for mature and activated macrophages (17). HEXA and HEXB are essential enzymes for catabolism of GM2 gangliosides abundantly expressed on the surface of neurons, and their genetic insufficiency results in the accumulation of GM2 in neuronal lysosomes causing a group of disorders collectively known as gangliosidosis (30). Cathepsin S is a member of the cysteine-lysosomal protease family and thought to be essential for the turnover of intracellular proteins. These enzymes once secreted from microglia may variously target the extracellular matrix and/or gangliosides on neurons. In fact, cathepsin S has been demonstrated to be secreted from macrophage/microglia cell lines and is known to destroy extracellular matrix molecules such as laminin, fibronectin, and chondroitin sulfate proteoglycans (24). Furthermore, degradation of laminin has been shown to cause neuronal cell death in the mouse hippocampus (6). It cannot be ruled out at this point in time that the overexpression of these lysosomal hydrolases in microglia may occur simply as a consequence of the neurodegeneration playing a role in the clearance of debris and/or remodeling of degenerated tissues. However, these enzymes could play an important role in the neurodegeneration of the prion diseases by promoting the degradation of molecules essential for the survival of neurons.

MPS-1 was isolated as the macrophage-specific gene with the highest expression in mature macrophages and in good concurrence with lysozyme M (32). So far, there is very limited information on MPS-1 and its physiological function awaits elucidation. However, its primary structure as deduced from the nucleotide sequence contains a domain with a significant homology to perforin, which was originally identified as a granule protein in cytotoxic T lymphocytes and natural killer cells. Perforin polymerizes in the membrane of target cells to form pores that cause target cell destruction. The homologous region corresponds to the putative α-helical domain of perforin, which plays a crucial role in forming pores in the cytoplasmic membrane of target cells (32). An intriguing possibility is that MPS-1 may have similar cytotoxic potential and directly provoke neuronal cell death. Recently, perforin was demonstrated in GFAP-positive reactive astrocytes and GFAP-negative unclassified round cells, possibly microglia, in the brains of patients with several neurodegenerative disorders, including Alzheimer's disease (12). It would be of value to clarify the physiological function of MPS-1 and perforin in the brain and their role in neurodegeneration.

We have also identified upregulation of the ROS scavenger, Prx, in astroglia distributed throughout the prion-affected brains. This represents indirect but clear evidence of the increased production of ROS in response to prion infection. The neurotoxic peptide PrP106-126 has been shown to stimulate cultured microglia to proliferate and produce ROS in the culture medium (2), suggesting that microglia is the cell type producing ROS in the brain. Of particular interest was our observation that Prx was preferentially upregulated in astrocytes but not in neurons. Consistent with this, the PrP106-126 peptide is toxic to neurons but not astrocytes in culture (2). It is well known that, in culture, neurons are more vulnerable to ROS than are astrocytes (1), and the much lower concentration of glutathione in neurons has been presumed to contribute to this high vulnerability (28). Many studies have demonstrated that oxidative stress is an important exacerbating factor in the neurodegenerative process, and it is likely that the lack of protective responses against ROS in neurons is one of the mechanisms involved.

In short, the present study has shown that the microglial production of lysosomal hydrolases, MPS-1, and possibly ROS was closely correlated with the onset and progression of pathological changes in prion diseases and might suggest an important role for microglia in the pathogenesis. Evaluation of the potential neurotoxic effects of these microglia-derived compounds is urgently needed to allow for the development of a pharmaceutical approach to control the progression of prion diseases.