Differential Targeting of Neurons by Prion Strains Part 2

PrPC Glycoform Heterogeneity as a Function of Brain Region

To explain selective targeting of different neuronal populations by prion strain, we hypothesized that there are cell-specific differences in the affinity of an infecting PrPSc for PrPC, which in turn determines brain region differences in the rate of nascent PrPSc formation and accumulation. From the evidence described above, it seemed reasonable to postulate that neuron-specific variations in the complex CHOs may alter the interaction between an infecting PrPSc and a neuron’s PrPC, or the size of the energy barrier that must be traversed during formation of nascent PrPSc. The goal of our next studies was to comprehensively test whether or not brain regions in Tg mice and Syrian hamsters (SHa) synthesize different sets of SHaPrPC glycoforms, as inferred from 2-D gel electrophoresis patterns (52).

Charge Isomers in Transgenic Mice Expressing Glycosylation Site Mutant PrP

Before beginning studies of SHas, we tested whether or not PrPC charge isomers vary as a function of the number of CHOs attached to PrPs in Tg mice expressing wild-type and glycosylation mutant SHaPrPCs (49). Earlier studies had shown that PrP charge isomers resulted from variable sialylation of Asn-linked CHOs (46,47,53). A large number of charge isomers were found with wild-type SHaPrPC; but, mutation of both Asn-linked glycosylation consensus sites reduced the number of isomers to one major and two minor spots (Fig. 5A, D). An intermediate number of charge isomers were found when one of the two consensus sites was mutated to produce either SHaPrPC(T183A) or SHaPrPC(T199A) (Fig. 5B, C).


Regional Differences in the Number and Location of PrP° Isoelectric Points in SHa Brain

In preliminary 1-D gel electrophoresis studies, the relative concentration of wild-type SHaPrPC was found to vary as a function of brain region in SHa brain. These differences were sufficient to influence the number of isoelectric points visible on Western transfers. Specifically, a progressive loss of SHaPrPC charge isomers occurred on the acidic side of 2-D gels which was proportional to the amount of SHaPrPC loaded. Therefore, the author adjusted the amount of sample loaded on the gel from each brain region, to equalize the amount of SHaPrPC. When equal amounts of SHaPrPC from each brain region were analyzed, differences in the number and locations of SHaPrPC isoelectric points could be identified by visual inspection (Fig. 6). To determine whether or not isoelectric patterns for a brain region were reproducible and unique between experimental runs and among different groups of animals, homogenates from 2-4 groups of hamsters were compared, using two ampholyte ratios (ampholyte ratios were created using Bio-Lyte ampholytes from Bio-Rad, as follows: ratio A, BioLyte 3/10:BioLyte 5/7 = 1:2 and ratio B, BioLyte 3/10:BioLyte 5/8 = 1:4).

 The majority of PrPC charge isomers originate from its two Asn-linked oligosaccharides. Wild-type SHaPrPC and deglycosylation mutant SHaPrPCs expressed in hippocampus (Hp) of Tg mice are compared. (A) Wild-type SHaPrPC; (B) SHaPrPC(T199A); (C) SHaPrPC(T183A); (D) SHaPrPC(T183A, T199A). Western transfers of 2-D gels were immunostained with the 13A5 monoclonal antibody.

Fig. 5. The majority of PrPC charge isomers originate from its two Asn-linked oligosaccharides. Wild-type SHaPrPC and deglycosylation mutant SHaPrPCs expressed in hippocampus (Hp) of Tg mice are compared. (A) Wild-type SHaPrPC; (B) SHaPrPC(T199A); (C) SHaPrPC(T183A); (D) SHaPrPC(T183A, T199A). Western transfers of 2-D gels were immunostained with the 13A5 monoclonal antibody.

The greatest number of comparisons were made for the neocortex, hippocampus and cerebellum because there were sufficient quantities of homogenate from these regions to perform multiple electrophoresis runs.

Differences in the PrPC isoelectric point patterns from six brain regions are readily detectable by visual inspection of Western transfers immunostained with the 3F4 mAb. Equal amounts of PrPC from each region were analyzed. Approximate pH range is indicated at bottom. Hp, hippocampus; Cb, cerebellum; Th, thalamus; Hy, hypothalamus; Cd, head of caudate nucleus; NC, frontoparietal neocortex.

Fig. 6. Differences in the PrPC isoelectric point patterns from six brain regions are readily detectable by visual inspection of Western transfers immunostained with the 3F4 mAb. Equal amounts of PrPC from each region were analyzed. Approximate pH range is indicated at bottom. Hp, hippocampus; Cb, cerebellum; Th, thalamus; Hy, hypothalamus; Cd, head of caudate nucleus; NC, frontoparietal neocortex.

To assemble all data from multiple electrophoresis runs, and from multiple animal groups, into a single graph, the isoelectric points from each run were normalized by choosing one spot that could be recognized, and whose pH was known. An isoelectric point between pH 5.8 and 5.9 was present in each run and animal group, and served this purpose well for the two ampholyte ratios used in this study. The distance to that spot from the alkaline end of each electrophoresis pattern was measured, and the fractional distance traveled by other spots calculated (relative isoelectric mobility) (Fig. 7).

Graphic summary of multiple 2-D electrophoresis runs for the neocortex (NC), hippocampus (Hp), and cerebellum (Cb). Isoelectric points, which appeared in every run from each group of animals, are represented by solid circles, and those that failed to occur one or more runs are represented by open circles. The number of elec-trophoresis runs (left-hand number) and the number of animal groups studied (right hand number) are indicated in parentheses.

Fig. 7. Graphic summary of multiple 2-D electrophoresis runs for the neocortex (NC), hippocampus (Hp), and cerebellum (Cb). Isoelectric points, which appeared in every run from each group of animals, are represented by solid circles, and those that failed to occur one or more runs are represented by open circles. The number of elec-trophoresis runs (left-hand number) and the number of animal groups studied (right hand number) are indicated in parentheses.

Analysis of isoelectric points in this way revealed patterns common to each brain region, as well as differences. The isoelectric point at pH 5.8-5.9 was located between another, more acidic point and two more alkaline points. These four isoelectric points and the spacing between them, were similar in all brain regions, including those not shown in Fig. 7. Differences that distinguished each brain region were on both the acidic and alkaline side of the four points. The more acidic isoelectric points resolved into distinct spots; the more alkaline points between pH 6.3 and 7.3 tended to merge together (Fig. 6 and 7). The latter was particularly characteristic of the neocortex with ampholyte ratio A.

The isoelectric point patterns from multiple runs were consistently different for the neocortex, hippocampus and cerebellum. Most of the isoelectric points were present in each run (constant points); however, a few isoelectric points were found in some runs but not others (non-constant points) (Fig. 7). For ampholyte ratio A, the hippocampus had 15 constant isoelectric points ranging in pH from 4.8 to 7.36 (N = 7) whereas the cerebellum had 13, ranging from pH 5.15 to 7.33 (N = 4). For ampholyte ratio B, the hippocampus had 16 constant isoelectric points from pH 5.18 to 7.25 (N = 10), and the cerebellum, 10 constant points from pH 5.51 to 6.75 (N = 11). 2-D gel runs of neocortical homogenates from three groups of animals were different than both the hip-pocampal and cerebellar patterns, particularly with ampholyte ratio A. One striking difference was the number of constant isoelectric points between pH 6.1 and 7.1, which was estimated to be 12 for the neocortex, 7 for the hippocampus (plus 2 inconstant points), and 8 for the cerebellum. These results indicate that the majority of isoelectric points are reproducible from run to run and between animals, and are, therefore, characteristic of a brain region.

PrP Is Similar to Other GPs

The results of our studies show heterogeneity of PrPC charge isomers as a function of brain region. Moreover, because our earlier study (49), and those of others (46,47,53), found that virtually all of PrPC’s isoelectric points originate from its two Asn-linked CHOs, the charge heterogeneity indicates that there are reproducible brain region differences in the structure of PrPC’s CHOs. Cell-type-specific glycosylation patterns are well known for other glycoproteins, so that they are referred to as a cell’s "glycotype" (54). The results with PrPC are, therefore, consistent with what is known about cell-specific characteristics of other glycoproteins. Finding PrPC glycotype variations as a function of brain region is consistent with our hypothesis that selective targeting of neurons in prion diseases is modulated, at least in part, by PrPC’s CHOs.

Because PrPC from dissected brain regions was analyzed, the spectrum of glycoforms in each probably represents the products of multiple nerve cell types. Whether the PrPC located in a neuroanatomically defined brain region is synthesized by resident neurons, or whether it is carried into the region by axonal transport from neurons outside of the region, or both, is unknown. It is reasonable to assume that glial and endothelial cells in each region contribute little to the set of glycoforms, since they express less than three PrP mRNAs per cell; neurons express from 10 to 50 depending on the nerve cell type (55). Although prion strains can be identified by reproducible differences in the neuroanatomic pattern of PrPSc deposition, it is also true that many prion strains target the same brain regions for conversion of PrPC to PrPSc (40,44,45). If indeed PrPC’s Asn-linked CHOs modulate its interaction with PrPSc, finding only minor differences in the isoelectric point patterns among some brain regions may explain why the same brain regions are targeted by different prion strains.

Evidence that PrPC’s Asn-Linked CHOs Probably Influence its Interaction with an Infecting PrPSc

How might Asn-linked oligosaccharides influence PrPC’s interaction with PrPSc and be the basis of selective neuronal targeting? One possibility is that the flexibility and/or conformation of that portion of the PrPC molecule that interacts with PrPSc, and whose conformation is changed, during conversion, to nascent PrPSc, is influenced by variations in PrPC’s CHOs. The PrP molecule has two domains that play different roles in the conversion of PrPC to PrPSc (Fig. 2). First, there is a stable or ordered core domain that contains the two Asn-linked oligosaccharides; two a-helices, designated helix B and helix C, which are stabilized by a disulfide bridge between cysteine179 and cysteine214; the GPI attached to the C-terminus at residue 231, which anchors PrPC to the plasma membrane; and the protein X binding sites, which are believed to lower the energy barrier for conversion of PrPC to PrPSc when PrPC binds to protein X (24,25). Second, there is a variable or disordered, domain that contains the portion of the molecule that interacts with PrPSc and changes its conformation from primarily unstructured in PrPCto P-sheet in PrPSc (13). Nuclear magnetic resonance and nuclear Overhauser effect spectroscopy of two large synthetic PrP fragments, PrP90-231 (56) and PrP29-231 (57), suggest that the variable domain of PrPC is mostly unstructured, but may contain a relatively short a-helix (helix A, residues 144-156) and two short antiparallel P-strands (residues 129-131 and 161-163). Investigations of the steps required for prion propagation and neurodegeneration in Tg mice expressing chimeric mouse-hamster-mouse or mouse-human-mouse PrP transgenes indicate that residues 90 to 150 in the variable region play a particularly important role in the interaction of PrPC with PrP80 leading to the conversion of the former to the latter (23,24). Residues 90-150 are largely unstructured or weakly helical in PrPC (56,57) but are predicted to be P-sheet in PrPSc (13). In addition, putative helix A may be converted to P-sheet, along with other portions of the variable region, during the conversion to PrPSc. Consistent with these possibilities, Fourier transform infrared and circular dichroism spectroscopy indicate that PrPC contains about 40% a-helix and about 3% P-sheet; PrPSccontains about 30% a-helix and 45% p-sheet (14).

Several lines of evidence indicate that Asn-linked CHOs influence protein conformation (48). For example, Asn-linked glycosylation was found to cause an hemagglutinin peptide to adopt a more compact, folded conformation (58). Glycosylation of epitopes in the rabies virus glycoprotein was found to disrupt a-helical structure, and to induce formation of a P-turn; moreover, the most dramatic effects occurred on addition of a single, simple carbohydrate (59). The latter raises the possibility that some PrPC glycoforms may more readily convert to PrPSc than others. Glycosylation appears to influence disulfide bridges between serum immunoglobulin M peptides, possibly by reducing the mobility of the peptide tailpieces (60). Glycosylation of a highly conserved 15-residue loop region in the nicotinic acetylcholine receptor facilitates disulfide bridge formation, apparently by bringing the termini of the loop into closer proximity (61). In some cases, such as the CD2 receptor of T-cells, the oli-gosaccharide binds to a positively charged cluster of five solvent-exposed lysine residues that destabilize the functionally relevant conformation in the deglycosylated molecule (62,63). Regarding the influence of oligosaccharides on specific targeting of cells, a single carbohydrate residue difference among cell surface antigens determines whether or not E. coli targets the urinary tract for infection (64). The foregoing provide precedent for the idea that the CHOs may influence the conformation and flexibility of residues 90-150 in the PrPC molecule.

Prion Strain Specificity of PrPSc Glycotypes

Given the evidence that each neuron population synthesizes PrPC with a different complement of CHOs, and the fact that each prion strain targets a different set of neurons for conversion of PrPC to PrPSc, one would predict that PrPSc glycoforms are also prion-strain-specific. To date, no laboratory has compared PrPSc isoelectric patterns for different strains of prions. However, different proportions of di-, mono-, and nonglycosylated PrP27-30 (the proteinase K digestion product of PrPSc) have been found to characterize human and animal prion strains. For example, a preponderance of diglycosylated PrP27-30 has been found by Western analysis in the central nervous system of patients with new variant CJD of Great Britain, compared with most sporadic and iatrogenic CJD cases (65,66). Similarly, the proportions of unglycosylated to heavily glycosylated PrP 27-30 following passage of prions in mice, are sufficiently different to differentiate seven scrapie prion strains (67).

In animal models of scrapie, the vast majority of PrPC molecules that are converted to PrPSc are probably doubly glycosylated, because deletion of both CHOs, or deletion of the CHO at Asn181 alone, results in retention of PrPC in the cell body, failure of PrPC transport to the plasma membrane of neuritic processes, and rapid degradation of PrPC (49). Lehmann and Harris (68) found that MoPrPC in Chinese hamster ovary cells mutated to delete the CHO at Asn 180 alone or to delete both CHOs at Asn180, and Asn196 failed to reach the cell surface after synthesis; in contrast, both wild-type MoPrPC synthesized in the presence of tunicamycin or mutated MoPrPC, in which the CHO at Asn196 alone was deleted, were detected on the plasma membrane. These results suggest that the CHO at Asn181 in SHaPrP (or at Asn180, in the case of MoPrP) is particularly important for trafficking of PrPC to the plasma membrane, in general, and to neuritic processes of neurons, in particular. PrPC must reach the cell surface prior to conversion to PrPSc because blocking PrPC export from the endoplasmi reticulum-Golgi complex to the plasma membrane inhibits formation of PrPSc (16) and, because exposure of scrapie-infected cells to phosphatidylinositol-specific phospholipase C which releases PrPC from the cell surface, also inhibits formation of PrPSc (17). Although PrPC that is mono-glycosylated at Asn181 (CHO at 197 deleted) is transported to the plasma membrane, it is unlikely that a significant amount of it accumulates as nascent PrPSc. Thus, although Asn 181 monoglycosylated PrPC has a normal distribution and concentration in the brain that is similar to wild-type PrPC, it requires over 500 d for conversion to PrPSc, following inoculation with Sc237 prions, which is more than 3x as long as the time to death in the animals expressing wild-type PrPC (49).

All of the above observations suggest that the variable, but significant, proportions of mono- and nonglycosylated PrPSc which are characteristic of each prion strain, are probably formed after PrPC is converted to PrPSc. Postconversion modification of PrPSc’s CHOs is likely, because PrPSc’s pep-tide component is highly protease-resistant; its CHOs, like those of other plasma membrane glycoproteins, are relatively sensitive to glycosidases (69,70). CHO degradation may occur in lysosomes, where a proportion of PrPSc becomes stored (16,71), or PrPSc molecules may be partially or completely deglycosylated by recycling through nonlysosomal endocytic compartments (72). What then is the relationship of the neuroanatomic site of formation of PrPSc to the origin of prion-strain-specific proportions of di-, mono-, and nonglycosylated PrPSc? It may be that some brain region-specific PrPSc glycoforms are preferentially trafficked to cellular compartments where CHOs are digested. Furthermore, some PrPSc glycoforms may be more easily degraded than others. Regarding the last possibility, degradation of renin glycoforms by the liver is directly related to the proportion of acidic isoelectric points (73). Alternatively, there may be significant differences in the rate or specificity of CHO digestion among neuron populations.

Familial CJD(T183A)

A Brazilian family with an autosomal dominant form of CJD was found to have a T183A mutation of the PRNP gene (74), which is predicted to prevent glycosylation of the Asn at residue 181 in human PrP, as it did in our transgenic mouse studies. Nine members of the family were affected by the disorder with a mean age of onset of 44.8 ± 3.8 yr and a duration of 4.2 ± 2.4 yr. Intense vacuolation was largely confined to the deeper layers of the neocortex and to the putamen. Immunoperoxidase staining of formalin-fixed, paraffin-embedded tissue sections for abnormal PrP, by the formic acid and hydrolytic auto-claving methods, showed deposition in the cerebellar cortex and putamen, but not in the neocortex. Nerve cell body immunostaining was not described, and is not obvious in their published photomicrographs.

This human pedigree is interesting, because it is different than the author’s findings in Tg mice in which we showed that expression of SHaPrPC(T183A), in the absence of wild-type MoPrPC in Tg(SHaPrP T183A)Prnpo/o mice, does not result in spontaneous neurodegeneration, nor does it make the mouse susceptible to infection with prions. In contrast, the family members expressing mutated HuPrP(T183A) died at a relatively young age with spongiform encephalopathy. The difference between the author’s Tg mouse model and the human cases is that, in the latter, mutated HuPrP was expressed concomitantly with wild-type HuPrPC. We believe that Tg(SHaPrP T183A)Prnpo/o mice were immune from disease because mutated SHaPrP(T183A) accumulated at low levels in Tg mouse brains and because of a rapid rate of degradation following synthesis. Also, SHaPrP(T183A) remained confined to nerve cell bodies, and was not transported to neuronal processes, where it would have a chance to interact with an infecting PrPSc. Concomitant expression of HuPrP(T183A) with wild-type HuPrPC raises the possibility that the former alters the conformation of the latter, converting it to abnormal pathogenic PrP, or that the presence of wild-type HuPrPC prevents degradation of mutated PrP, and allows it to be transported to cellular compartments, where it could cause spongiform degeneration of neurons.

PrPC Is the Main Host Factor Determining the Scrapie Phenotype

By viewing prion diseases from the perspective of the neuropathological changes, it has been learned that host-determined variations in PrPC play as important a role in determining the disease phenotype as the conformation of PrPSc comprising an infecting prion. The role of PrPC in the pathogenesis of prion diseases cannot be overemphasized. Indeed, the critics of the protein-only hypothesis often argue that an infectious agent, composed solely of a single, abnormally folded protein, cannot encode all the information necessary to account for the known variations in the prion disease phenotype. In this, they are probably correct because there are most likely only a limited number of different stable strain determining conformations of PrPSc possible, excluding contributions of PrPSc’s CHOs. However, these critics do not differentiate between coding of strain information in prions or the combination of both prion and host factors that generate the disease phenotype. The results of our studies, as well as others reviewed above, indicate there are more than sufficient animal species-determined variations in the amino acid sequence of PrPC, the level of expression of PrPC allotypes, and the CHO structure of PrPC, in combination with variations in the conformation and amino acid sequence of PrPSc to account for all of the known variations in the prion disease phenotype. Other host factors may also influence the disease phenotype, such as the response of microglia or astrocytes (75,76); nevertheless, all of the evidence indicates that PrPC is the preeminent host factor.

Next post:

Previous post: