In the present experiments scrapie infection of transgenic mice expressing anchorless PrP resulted in a slow fatal brain disease. These results demonstrated new mechanisms of prion-induced pathogenesis associated with the presence of PrPres amyloid and the absence of GPI-anchored PrP. This disease lacked gray matter spongiosis and differed in this respect from scrapie infection in non-transgenic mice, where the disease is characterized by extensive gray matter spongiosis and non-amyloid PrPres deposition.
The current results raised the question of how lack of GPI-linked membrane anchoring of PrP might facilitate formation of PrPres amyloid. GPI anchorless PrP has a longer biological half-life 
and is secreted by the cell. Both of these attributes might allow more effective and extensive interactions between soluble PrP molecules. In addition, the minimal amount of carbohydrates and the absence of the GPI group on anchorless PrP might favor amyloidogenic hydrophobic protein-protein interactions, particularly at a time of partial protein unfolding during PrP conversion. These features of anchorless PrP are likely to contribute to its enhanced tendency to form amyloid during conversion to PrPres. Anchorless protease-resistant PrP, cleaved at residue 228, comprises 15% of the PrPres in hamster scrapie brain extracts 
, but it is unclear whether this material contributes to the amyloid PrP seen in this model.
Our results differed from those of two interesting mouse prion disease models where PrPres was also found almost entirely in an amyloid form. In the GSS PrP-8kd model 
and the G3-ME7 model 
, which both used PrP mutant mice, PrP amyloid was seen primarily in the corpus callosum, but did not spread significantly to other brain regions. There was no clinical disease in these models, and transmission experiments suggested very low infectivity titers in the GSS PrP-8kd model. Compared to these two models, the three main distinguishing features of the anchorless PrP model are the ability of the PrPres amyloid to accumulate widely throughout the brain (), the resulting fatal brain disease (), and the high titer of transmissible agent () 
The association of amyloid deposition without gray matter spongiosis in our system is reminiscent of the neuropathology seen in certain human familial prion diseases. For example, GSS patients with PrP mutations Y145Stop and Y163Stop had both CAA and parenchymal perivascular amyloid without gray matter spongiosis  
. Both these mutations result in C-terminally truncated PrP lacking the GPI anchor. Parenchymal amyloid deposition without gray matter spongiosis has also been seen in GSS patients with several other PrP mutations including P102L, P105L, A117V and F198S 
. Recently two human GSS patients with new PrP mutations producing nonsense codons at positions 226 and 227 were described 
. Both patients had widespread PrPres amyloid deposition in the absence of gray matter spongiosis, and one had CAA. These patients expressed a nearly full-length form of PrP lacking 6–7 C-terminal residues and the GPI anchor, which was quite similar to the PrP expressed in our anchorless PrP tg mice.
In many GSS patients, amyloid PrPres purified from brain was truncated resulting in a 7–11 kDa protease-resistant fragment from the central region of PrP (approximately residues 81–150)
. Interestingly, presence of this truncation has been correlated with the lack of gray matter spongiosis 
. In contrast, based on previous immunoblot studies, the proteinase K-resistant PrPres amyloid in our model appeared to contain residues 88–231 
, which was similar to the PrPres found in human and animal prion diseases with extensive gray matter spongiosis. Furthermore, PrPres in tissue sections could be stained with anti-PrP serum R24, specific to residues 23–37 (data not shown) suggesting that there was no significant truncation at the N-terminus beyond the signal peptide. Thus, lack of spongiosis in our model appeared dependent on the absence of GPI-anchoring rather than truncation of the PrPres.
Two possibilities might explain the correlation between lack of GPI- anchored PrP and lack of gray matter spongiosis in our infected transgenic mice: (1) anchorless amyloid PrPres might be less neurotoxic than diffuse PrPres, and/or (2) anchored PrPsen might be required for PrPres-mediated neurotoxic membrane interactions. The former explanation could not be proven or excluded by our results. However, the latter interpretation was supported by data from brain graft experiments. After scrapie infection of tg44+/− mice grafted with C57BL/6 brain expressing normal anchored PrPsen, we observed gray matter spongiosis and non-amyloid PrPres deposition in C57BL/6 grafts, but not in adjacent host tissue expressing only anchorless PrPsen. Tissue expressing only anchorless PrPsen appeared to be unable to respond to the presence of GPI-anchored PrPres produced in the nearby grafts, and no gray matter spongiosis was produced. Therefore, lack of anchored PrPsen might by itself explain the lack of gray matter spongiosis in transgenic mice.
However, even in the absence of anchored PrPsen, the amyloid PrPres was able to induce additional pathogenic processes capable of causing fatal neurological disease. By both light and electron microscopy we observed evidence for three distinct pathogenic processes not seen in typical prion disease in C57BL/10 mice (Box 1):
(1) Brain damage caused by tissue distortion by large amyloid plaques. These plaques were associated with neuronal loss, axonal pathology and gliosis ( ). The more rapid accumulation of PrPres in tg44+/+ mice compared to tg44+/− mice () suggested a faster growth of large space-occupying plaques which might explain in part the clinical neurological signs leading to death of tg44+/+ and tg23+/+ mice ().
(2) A second pathogenic process in scrapie-infected transgenic mice was suggested by ultrastructural studies finding that the early aggregation of PrPres into fibrillar amyloid was located at or within vascular basement membranes (, ). This was associated with vascular damage including occlusion (), amyloid replacement of basement membrane and tunica media, and occasional micro-hemorrhages. This pathology was similar to that observed in CAA seen in Alzheimer's disease and several familial amyloid diseases including two prion diseases 
(3) Evidence of a third pathogenic process in the transgenic mice was suggested by finding of small deposits of immunogold-labeled PrPres at the ultrastructural level in the extracellular spaces between glial and neuritic processes in gray matter (). These PrPres deposits were small, and there was no distortion of the extracellular space or visible aggregation into amyloid fibrils. However, the adjacent processes were often highly dystrophic () or swollen and devoid of organelles, and they appeared to coalesce to form empty spaces larger than the original processes (, ). These abnormal areas, which were also noted in heterozygous tg44+/− and tg23+/− mice, appeared to represent a form of damage related to small, rather than large, PrPres deposits, and they did not require the presence of anchored PrPsen for their formation.
Box 1. Three neuropathogenic processes found in scrapie-infected homozygous anchorless PrP transgenic mice (tg44+/+ and tg23+/+)
- Displacement of brain structure by rapidly expanding amyloid plaques
- Associated with neuronal dropout and adjacent axonal and neurite damage
- Early PrPres amyloid on or within basement membranes of endothelial cells, smooth muscle cells and pericytes
- Possible damage to basement membrane and obstruction of the flow of interstitial brain fluid by PrPres amyloid
- Possible role of basement membrane components in assisting PrPres formation, e.g. glucosaminoglycans, collagen, laminin, etc.
- Vascular occlusion and occasional micro-hemorrhages
- Accumulation of PrPres in the interstitial space between neurite and glial processes
- Neuronal and glial toxicity with formation of swollen neurites and glial processes with fewer organelles
The early localization of PrPres at basement membranes (), suggested that the PrP conversion process might initiate at these sites, and implied that basement membrane molecules might facilitate PrP conversion. For example, basement membrane might filter or trap soluble PrPsen molecules or small PrPres oligomers from the extracellular interstitial fluid of brain increasing their local concentration, thus favoring conversion to larger PrPres amyloid aggregates. Serum amyloid P-component which binds to all amyloids and is a constituent of basement membranes might also contribute to local PrP conversion 
. In addition, collagen, laminin and heparin sulfate-containing proteoglycans are major components of basement membranes, and PrP can bind to both the laminin receptor and heparan sulfate which can associate directly or indirectly with PrP 
. Heparan sulfate and other glycosaminoglycan (GAG) moieties can delay scrapie disease in vivo 
, and some GAG molecules can alter PrP conversion in vitro 
. A scaffolding mechanism might account for this effect. For example, soluble anchorless PrPsen monomers might be held in place by GAG polymers to increase local concentration and facilitate conversion by PrPres, analogous to the tethering of anchored PrPsen on cell membranes 
. In addition, attachment of small mobile PrPres oligomers to GAG polymers might assist conversion at the basement membrane. Subsequently newly formed larger less mobile PrPres could serve as an efficient scaffold for further conversion allowing the process to extend out into the brain parenchyma. Eventually this process might form very large PrPres amyloid plaques with blood vessels at the center as we observed ( and ).
The vascular amyloid pathology seen in our scrapie-infected transgenic mice (, , ) was similar to CAA seen in Alzheimer's disease as well as several familial amyloid diseases 
, including two forms of familial prion disease  
. In Alzheimer's disease, amyloid fibrils within vascular basement membranes are thought to impede interstitial fluid drainage leading to an increase in Aβ concentrations within the extracellular space. Such increased soluble Aβ and oligomeric proto-amyloid fragments are considered a likely contributory factor in the cognitive decline of Alzheimer's disease patients 
. Similar processes might contribute to the clinical disease seen in the anchorless PrP scrapie model. Since all these diseases with CAA show amyloid localization with basement membranes, drugs capable of blocking amyloid-basement membrane interactions might be effective treatments for some of these diseases. In the case of prion diseases, one such compound, pentosan polysulfate, a small GAG oligomer, was effective in blocking PrPres generation in an infected cell line 
and delayed onset of clinical scrapie in vivo   
. Similarly a decoy molecule preventing PrP interaction with the laminin receptor (LRP/LR) reduced PrPres levels and delayed disease in vivo 
. Determining the precise glycans and proteins involved in the protein interactions leading to amyloid deposition in all the CAA diseases might be important in designing new therapeutic approaches.