Prion protein NMR structure and species barrier for prion diseases

METHODS

The structure of mouse PrP(121–231) used for the present analysis is a refinement of the previously reported NMR structure (9), which is based on an input of 1,586 nuclear Overhauser effect distance constraints. The structure determination will be described elsewhere (unpublished work). Key characteristics of the refined structure include that there are no residual violations of the aforementioned distance constraints in excess of 0.1 Å and that the well defined regions comprise residues 124–166 and 172–226, for which a root mean square distance for the backbone atoms N, C^α, and C′ of the 20 final conformers to their mean of 0.8 ± 0.1 Å was obtained. Figs. 2, 3, and 4 of this paper show an energy-refined mean structure of mouse PrP(121–231), of which the atom coordinates have been deposited in the Brookhaven Protein Data Bank (accession code 1AG2).

Information on mammalian PrP sequences was collected from the protein databases swissprot (13) and the nucleic acid database genembl. The structure comparisons in the following section were based on the sequence alignment of Schätzl et al. (10), which is not a critical issue because the pairwise sequence identity is of the order of 90% for all combinations of mammalian species considered.

RESULTS

A group of 23 PrP sequences (Fig. 1) was identified in a database search for entries with over 60% identity and an overlap of at least 60 residues with the polypeptide fragment 124–226 of the mouse PrP. The following additional selection criteria were used to ensure that the focus was on statistically reliable data: (i) In the presence of the complete sequence, partial sequences of the same species were dropped. (ii) Where different sequences were reported for the same species, only the one with the smallest number of changes relative to the mouse sequence was retained. (iii) All sequences were omitted that differ from the sequence of the mouse protein only by conservative mutations in positions where all other species in the databases show identity with the mouse protein (see Fig. 1 legend). (iv) The sequence of the marsupial possum was not included because its evolutionary distance to all other mammals is outstandingly large (see Table II A, B. in ref. 14).

Overall, as indicated in the bottom line of the figure, the 23 sequences in Fig. 1 contain 22 positions with amino acid replacements. Of these, 18 are grouped into the classes A, B, C, and D on the basis of their locations in the three-dimensional structure of PrP(121–231) (Fig. 2) and the chemical properties of the amino acid residues (Figs. 3 and 4). The three-dimensional fold of PrP(121–231) includes three helices with residues 144–154, 175–193, and 200–219, and an antiparallel β-sheet with residues 128–131 and 161–164 (Figs. 1 and 2). The helices 2 and 3 are linked by the disulfide bond Cys-179–Cys-214 (Fig. 2) and represent a stable scaffold for the global molecular structure (9). The class A sites form a dense cluster of nine residues in the loop between the second β-strand and the second helix (positions 164, 166, 168, 170, and 174), and near the C terminus (positions 215, 219, 220, and 223) (upper right in Fig. 2). Class B includes the five hydrophobic residues 138, 139, 184, 203, and 205, which are located in the interface between the first helix and the remainder of the protein. Class C sites include positions 145 and 155 at the two ends of helix 1, and the position 143 immediately preceding this helix. Class D includes only the exchange Q186E in the bovine protein, which introduces an isolated surface-exposed negative charge that might affect intermolecular interactions. The position 225 (“2”) near the C terminus and three positions with largely conservative single-species amino acid substitution (“1”) are not shown in Fig. 2 and are not further considered.

In class A the amino acid replacements in positions 168, 219, 220, and 223 (Fig. 3a) involve changes in the electrostatic charge of the side chains and may modify long-range forces effective in intermolecular recognition. Fig. 3 b-d shows the electrostatic surface potential for the mouse, human, and sheep proteins, where this molecular region has a total net charge of −1, −3, and 0, respectively. In total, seven of the nine sequence positions in class A contain polar or charged side chains in all species of Fig. 1. The class A residues thus form a surface region of PrP^C that has the potential to function as a selective protein–protein interaction site. It is readily apparent from inspection of the NMR structure of mouse PrP(121–231) that many of the A-type amino acid exchanges will cause different surface hydrogen bonding patterns in the PrP from different species, and some of the amino acid replacements also might lead to different hydrogen bonding with a docked molecule. Overall, although the amino acid substitutions between different species can be expected to preserve the global shape of this putative binding site, they may modify its specificity for both long-range electrostatic interactions and short-range hydrogen bonding with other proteins.

Classes B and C contain residues from helix 1, part of helices 2 and 3, and helix-adjacent loops (Fig. 4a). The group C sites are related to a special feature of helix 1, which includes four largely or partially solvent-exposed aromatic side chains in the mouse sequence (Fig. 4b). With the sole exception of Met-154 all other side chains of this helix have polar character (Fig. 4b), and the positions 146–148, 151, and 152 actually contain charged side chains (Fig. 1). The combination of long-range electrostatic potentials, high hydrogen bonding propensity, and the presence of exposed aromatic residues on the solvent-accessible surface of helix 1 again represents structural features of a surface area that could well function as a specific recognition site for other proteins. Replacement of the C-type amino acid residues 145 and 155, in particular replacement of the unusual surface-exposed Trp-145, can be expected to modify the specificity of intermolecular interactions in this molecular region.

In all proteins of Fig. 1 the type B sites contain exclusively hydrophobic side chains of Val, Leu, Ile, or Met, which are in contact with other hydrophobic residues (Fig. 4a). Because of their limited surface accessibility, and considering the absence of electrostatic charges and hydrogen bonding propensity, the molecular region containing these sites is not reminiscent of a potential binding site that would mediate intermolecular interactions of the folded PrP^C molecule. However, upon limited conformational rearrangement of PrP^C, for example, by unfolding of the helix 1 (see ref. 9), several of these hydrophobic residues would become accessible and thus could support propagation of self-association of PrP. The observation in the spatially neighboring positions 184 and 203 (Fig. 4a) that replacement of Ile-184 by Val correlates in all but two species with the replacement of Val-203 by Ile or Met (Fig. 1), appears to further support the view that class B amino acid exchanges leave the PrP^C surface properties largely unperturbed.

In summary, Fig. 2 shows that large regions of PrP(121–231) are not affected by species variations. These include the β-sheet, the central two turns of the first helix, the C-terminal two turns of the second helix, the loop linking the helices 2 and 3, and the third and fourth turn of the third helix. Part of this conservation can readily be explained by requirements to maintain local three-dimensional PrP^C structure. However, some conservation also may be needed to secure further intramolecular or intermolecular interactions, for example, with the polypeptide segment 23–120 in intact PrP^C, with the cell surface to which the PrP is attached, or with a natural ligand. The high frequency of amino acid exchanges between different species in the region of the three-dimensional structure that contains the class A residues makes this region a likely candidate for a specific binding site, where amino acid replacements would affect intermolecular recognition with PrP^C (Fig. 3). The helix 1 with the C-type amino acid substitutions forms another potential binding site (Fig. 4b) on the opposite molecular surface (Fig. 2), where replacement of solvent-exposed aromatic side chains between different species would be expected to modify the specificity by which this region is recognized by other proteins.

DISCUSSION

Inoculation experiments with chimeric PrP suggested two protein binding sites in the PrP amino acid sequence. These experiments so far have been the only information available for mapping of species-related differences in PrP^C. One of these binding sites would be located in the segment with residues 96–167, which would bind an infecting PrP^Sc particle. The second site would be composed of residues outside of this segment, which could possibly interact with the postulated protein X (4, 12). The variable C-type residues in helix 1 hence would contribute to a protein surface area that might be involved in interactions with PrP^Sc, and the cluster of type A residues would be a potential binding site for protein X. The in vitro PrP^C conversion experiments of Kocisko et al. (8) suggested that the hamster/mouse species barrier is dependent on the three amino acid exchanges M139I, N155Y, and N170S (Fig. 1). It is striking that the segment 96–167 as well as the group of residues 139/155/170 includes at least one residue from each of the classes A, B, and C (Fig. 1). Other experimental data attribute an exclusive role for the species barrier between mouse and hamster to position 139 (16): The substitution of Met in hamster PrP by Ile in mouse PrP appears to be sufficient to protect hamster from infection by mouse PrP^Sc. This would place the critical structural feature for the species barrier in the interface between helix 1 and the protein core (Figs. 2 and 4a). The residue 139 presumably would become operational only after initial conformational transitions of PrP^C induced by complexation with PrP^Sc and possibly protein X (9). The present structural observations can be expected to support future refinement of in vivo and in vitro experiments with chimeric PrP.

The class D exchange Q186 E in bovine PrP introduces a negative electrostatic charge in a surface location that is clearly separated in space from the two presumed binding sites characterized by A- and C-type residues, respectively (Fig. 2). Interestingly, recent results indicate the presence of Glu-186 also in the related species Watussi, Banteng, and Wisent (H. Schätzl, personal communication). The newly determined PrP sequences of the house dog, the Canadian wolf, and the dingo (H. Schätzl, personal communication) further extend the available information on D-type mutation sites, because the otherwise strictly conserved Asn-159 (Fig. 1) is replaced by Asp, and this negative charge is located in close spatial proximity to position 186 (Fig. 2). This leaves the intriguing option that a third variable surface area in PrP^C might emerge by further extension of the sequence database, and that this site might be functional in the species barrier with cattle, including possibly also transmission of disease from bovine spongiform encephalopathy-infected cows to other species.

Finally, although the NMR structure of mouse PrP(121–231) represents only part of PrP^C and the ensemble of all accumulated data indicates that N-terminal parts of the sequence also may influence the species barrier (for example, see refs. 4 and 5), there is evidence that the data presented in this paper are highly relevant with regard to the intact system, and that PrP(121–231) has a key part in PrP^C physiology (8, 12, 16). Structure predictions and experimental observations (9, 17) indicate that PrP(121–231) is probably the only polypeptide segment with a global fold in intact PrP^C, and initial NMR studies with intact PrP^C performed in our laboratory indicate that the segment 121–231 in the intact protein has the same fold as in PrP(121–231) (unpublished data).