Mutational analysis of topological determinants in prion protein (PrP) and measurement of transmembrane and cytosolic PrP during prion infection.

The prion protein (PrP) can adopt multiple membrane topologies, including a fully translocated form (SecPrP), two transmembrane forms (NtmPrP and CtmPrP), and a cytosolic form. It is important to understand the factors that influence production of these species, because two of them, CtmPrP and cytosolic PrP, have been proposed to be key neurotoxic intermediates in certain prion diseases. In this paper, we perform a mutational analysis of PrP synthesized using an in vitro translation system in order to further define sequence elements that influence the formation of CtmPrP. We find that substitution of charged residues in the hydrophobic core of the signal peptide increases synthesis of CtmPrP and also reduces the efficiency of translocation into microsomes. Combining these mutations with substitutions in the transmembrane domain causes the protein to be synthesized exclusively with the CtmPrP topology. Reducing the spacing between the signal peptide and the transmembrane domain also increases CtmPrP. In contrast, topology is not altered by mutations that prevent signal peptide cleavage or by deletion of the C-terminal signal for glycosylphosphatidylinositol anchor addition. Removal of the signal peptide completely blocks translocation. Taken together, our results are consistent with a model in which the signal peptide and transmembrane domain function in distinct ways as determinants of PrP topology. We also present characterization of an antibody that selectively recognizes CtmPrP and cytosolic PrP by virtue of their uncleaved signal peptides. By using this antibody, as well as the distinctive gel mobility of CtmPrP and cytosolic PrP, we show that the amounts of these two forms in cultured cells and rodent brain are not altered by infection with scrapie prions. We conclude that CtmPrP and cytosolic PrP are unlikely to be obligate neurotoxic intermediates in familial or infectiously acquired prion diseases.

Prion diseases are fatal neurological disorders of humans and animals that appear in sporadic, familial, and infectiously acquired forms. These disorders are caused by conversion of a normal neuronal glycoprotein (PrP C ) 1 into a conformationally altered isoform (PrP Sc ) that is infectious in the absence of nucleic acid (1,2). PrP C , which is soluble and protease-sensitive, consists of an ␣-helical, C-terminal domain and an unstructured N-terminal domain. In contrast, PrP Sc is rich in ␤-sheets, aggregated, and protease-resistant. The physiological function of PrP C is uncertain but may be related to transport of copper ions or protection from oxidative stress (3).
PrP C is unusual because it can adopt multiple membrane topologies. Most PrP C molecules are attached to the outer leaflet of the plasma membrane through a C-terminal glycosylphosphatidylinositol (GPI) anchor (this topology is designated Sec PrP) (4,5). However, some PrP C molecules assume a transmembrane orientation when synthesized in vitro or in cells (6 -11). These forms, designated Ntm PrP and Ctm PrP, span the lipid bilayer once via a highly conserved hydrophobic region in the center of the molecule (amino acids 111-134), with either the N or C terminus, respectively, on the extracytoplasmic side of the membrane. It has been shown that these species are generated in small amounts (Ͻ10% of the total PrP) as part of the normal biosynthesis of wild-type PrP in the endoplasmic reticulum (ER). However, mutations within or near the transmembrane domain, including an A117V mutation linked to GSS as well as several "artificial" mutations not seen in human patients, increase the relative proportion of Ctm PrP to as much as 20 -30% of the total (7,12).
Recent studies have begun to define the mechanisms responsible for determining PrP topology during the translation process. We discovered that a non-conservative substitution (L9R) within the hydrophobic core of the signal sequence dramatically increased the proportion of Ctm PrP (13). Combining this mutation with a triple substitution (3AV) within the transmembrane domain resulted in a molecule that was synthesized exclusively as Ctm PrP. These results indicated that the signal sequence as well as the transmembrane domain were major determinants of PrP topology. Work by Hegde and colleagues (9 -11) has demonstrated that these two determinants act in mechanistically distinct ways. The signal sequence serves a dual function, first targeting the nascent polypeptide chain to the translocon channel in the ER membrane via binding to the signal recognition particle, and subsequently gating the translocon to allow passage of the N terminus into the ER lumen. In contrast, the transmembrane domain acts primarily to trigger integration of the polypeptide into the lipid bilayer. The combined action of both domains operating during the transloca-tion process serves to regulate the proportions of the three topological variants of PrP. Regulatory factors associated with the translocon, in addition to sequence determinants within the PrP molecule itself, have also been shown to influence the final topology achieved (14,15).
Work from our laboratory has identified several novel cell biological features of Ctm PrP. First, Ctm PrP contains an uncleaved, N-terminal signal peptide (13). This characteristic makes Ctm PrP unusual among other type II transmembrane proteins, most of which have internal signal-anchor sequences. Second, Ctm PrP has a C-terminal GPI anchor in addition to a transmembrane domain, thus displaying an unusual, dual mode of membrane attachment (13,16). Finally, by using the L9R/3AV mutant, we found that Ctm PrP expressed in cultured cells remains core-glycosylated and is retained completely in the ER (13). This result implies that the protein is recognized as abnormal by the ER quality control machinery that monitors folding of newly synthesized polypeptides.
A great deal of interest in the subject of PrP membrane topology derives from the possibility that topological variants of PrP may play an important pathogenic role in prion diseases. Although PrP Sc is widely agreed to be the infectious form of PrP, there is considerable debate about whether it is the form responsible for neuronal loss in these disorders (17). The amount, anatomical distribution, and time course of accumulation of PrP Sc often correlates with the development of neuropathology and clinical symptoms, but there are notable exceptions to this association. These discrepancies have led to the hypothesis that alternate forms of PrP, distinct from both PrP C and PrP Sc , are the proximate causes of neurodegeneration.
One candidate for such a neurotoxic intermediate is Ctm PrP. Two major pieces of evidence have been used to argue that Ctm PrP plays a key pathogenic role. First, transgenic mice have been generated that synthesize PrP molecules carrying the A117V mutation or one of the other Ctm PrP-favoring mutations (7,12). Animals expressing the mutant proteins above a threshold level synthesize Ctm PrP in their brains and spontaneously develop a scrapie-like neurological illness, but without PrP Sc detectable by Western blotting or infectivity assays. This result implies that certain familial forms of PrP may be due directly to increased levels of Ctm PrP. Second, mice have been constructed in which a wild-type hamster PrP transgene serves as a reporter of Ctm PrP formation (12). When these animals are inoculated with mouse prions, the amounts of Ctm PrP as well as PrP Sc in the brain are found to increase during the course of the infection. This result has been interpreted to indicate that PrP Sc induces formation of Ctm PrP, which is then the proximate cause of neurodegeneration during infectiously acquired prion diseases. In this view then, Ctm PrP is a key intermediate in both genetic and infectious prion diseases.
Another topological variant of PrP that has been proposed as a neurotoxic intermediate is cytosolic PrP. Expression of an artificial form of PrP lacking a signal sequence, which presumably favors accumulation of PrP in the cytoplasm, has been found to be toxic to cultured cells and transgenic mice (18). However, there is debate about whether PrP is found in the cytoplasm under normal circumstances and, if so, what mechanisms are responsible for delivering it there. Based on the observation that cytosolic PrP accumulates in cells that have been treated with proteasome inhibitors, it has been suggested that some molecules are retrotranslocated into the cytoplasm from the ER lumen as part of normal ER quality control mechanisms (19 -21). In contrast, our experiments indicate that cytosolic PrP molecules represents untranslocated chains that have never entered the ER (22). These chains, which are ob-served primarily under conditions of protein overexpression, contain an uncleaved N-terminal signal peptide and lack a GPI anchor.
A key gap in the experimental evidence supporting roles for Ctm PrP and cytosolic PrP in prion-induced neurodegeneration is the lack of data demonstrating that the amounts of these forms increase during the course of a prion infection. In part, this difficulty is due to the absence of direct methods for detecting Ctm PrP and cytosolic PrP in infected cells and tissues. In this paper, we present characterization of an antibody that reacts with both Ctm PrP and cytosolic PrP by virtue of their uncleaved signal peptides and our use of this antibody to assay Ctm PrP and cytosolic PrP in infected samples. In addition, we carry out a mutational analysis of several sequence determinants in PrP to better understand the factors that influence the topology of the protein.

EXPERIMENTAL PROCEDURES
Plasmids-Synthetic oligonucleotides encoding point mutations in the PrP signal sequence (Fig. 1A) were used to amplify a portion of the PrP DNA sequence by PCR. DNA fragments carrying the mutation were digested with HindIII and PshA1 and cloned into a pcDNA3 plasmid (Invitrogen) containing the WT mouse PrP sequence from which the HindIII-PshA1 fragment had been removed. Plasmids encoding PrP molecules with a FLAG epitope (DYKDDDDK) inserted at position 22/23, and with altered numbers of octapeptide repeats (PG0, ⌬51-90; PG1, ⌬51-82; PG2, ⌬67-90; PG14, ϩ9 repeats) have been described previously (5,13,23). Other PrP mutants were constructed by PCR (16). All PrP coding regions carried an epitope tag for monoclonal antibody 3F4, created by changing residues 108 and 111 to methionine. Prior to in vitro transcription, plasmids were linearized with XbaI.
In Vitro Translation and PK Protection-mRNAs encoding WT and mutant PrP molecules were transcribed using the mMessage mMachine kit (Ambion, Austin, TX) and were translated using rabbit reticulocyte lysate (Promega, Madison, WI) containing [ 35 S]methionine as directed by the manufacturer, except that the final lysate concentration was 50%. Translation reactions were supplemented with microsomal membranes from mouse BW5174.3 cells (24) or from canine pancreas (Promega). After translation, 5-l aliquots of lysate were incubated for 60 min at 4°C in a final volume of 50 l with or without 100 g/ml PK (Roche Applied Science) in the presence or absence of 0.5% Triton X-100. PK was inactivated with phenylmethylsulfonyl fluoride for 5 min, and 12-l aliquots were added to gel sample buffer containing phenylmethylsulfonyl fluoride for analysis by SDS-PAGE. In some cases, PrP was immunoprecipitated from translation reactions (as described below) prior to SDS-PAGE. For enzymatic deglycosylation, PrP was eluted from protein A-Sepharose beads with 1% SDS, 50 mM Tris-HCl (pH 7.5) and was then diluted 10-fold with 50 mM Tris-HCl (pH 7.5), 0.5% Triton X-100 containing 0.33 units/ml N-glycosidase F (New England Biolabs). After incubation at 37°C for 1 h, proteins were precipitated with methanol and analyzed by SDS-PAGE. Radioactive bands on gels were quantitated using a PhosphorImager SI (Amersham Biosciences).
Scrapie Infection of N2a Cells-Highly scrapie-susceptible sub-clones of N2a cells were prepared as described (25). Briefly, N2a cells from the ATCC (CCL131) were first sub-cloned by limiting dilution. Each subclone was then tested for scrapie susceptibility by incubation for 3 days with an extract of N2a cells that had been infected previously with the Chandler strain of scrapie (26). Cells were then passaged for 6 weeks and analyzed for PrP 27-30 by cell blotting or by Western blotting after PK digestion. The susceptible sub-clone used for the experiment shown in Fig. 7 was designated N2a.3. It was used in the infected state, as well as in the uninfected state as a matched control.
Transfection, Metabolic Labeling, and Immunoprecipitation-CHO and N2a cells were transiently transfected with PrP-encoding plasmids using LipofectAMINE or LipofectAMINE 2000 (Invitrogen) according to the manufacturer's directions. Twenty-four hours after transfection, cells were labeled for 6 h in methionine-and cysteine-free medium containing 100 -200 Ci/ml of [ 35 S]methionine/cysteine (Promix; Amersham Biosciences). Cultures were then lysed in 0.5% SDS, 50 mM Tris-HCl (pH 7.5), heated at 95°C for 5 min, and diluted with 10 volumes of RIPA buffer (150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl (pH 7.5)). Diluted lysates were incubated with anti-PrP antiserum for Ͼ1 h at 4°C and then with 20 l of protein A-Sepharose beads for 30 min at 4°C. Beads were washed four times in RIPA buffer, and PrP was eluted by heating at 95°C in gel sample buffer.
Scrapie Infection of Mice and Hamsters-Tg(WT)/Prn-p 0/0 mice (E1 line) expressing wild-type PrP carrying a 3F4 epitope have been described previously (27). Tg(L9R/3AV)/Prn-p 0/0 mice (B line) express mouse PrP carrying an L9R/3AV mutation and a 3F4 tag, a construct we have expressed previously in cultured cells (13). Full characterization of these mice will be provided elsewhere. 2 Scrapie inocula included the hamster 263K strain, the mouse RML strain, and the RML strain that had been passaged once in Tg(WT) mice to introduce the 3F4 epitope. To prepare inocula, infected brains were homogenized (10%, w/v) in phosphate-buffered saline using sterile, disposable tissue grinders. After clearing by centrifugation at 900 ϫ g for 5 min, the homogenates were diluted to a final concentration of 1 or 2.5% in PBS, and 25 l was injected intracerebrally into the right parietal lobe of 4 -6week-old recipient mice or hamsters using a 25-gauge needle.
Western Blots of Brain Homogenates-Brain lysates were prepared in 0.5% SDS, 50 mM Tris-HCl (pH 7.5). Samples were heated at 95°C for 5 min and diluted 10-fold with 50 mM Tris-HCl (pH 7.5), 0.5% Triton X-100 containing 0.33 units/ml N-glycosidase F. After incubation at 37°C for 2 h, proteins were precipitated with methanol, separated by SDS-PAGE, and subjected to Western blotting with 3F4 antibody.

Mutations in the Hydrophobic Core of the Signal Peptide Increase the Proportion of Ctm PrP and Reduce Translocation
Efficiency-For these experiments, we translated PrP mRNA in the presence of microsomes from murine thymoma cells, which are efficient at adding the C-terminal GPI anchor (16). Translation reactions were subsequently subjected to PK digestion to reveal protease-protected products corresponding to Sec PrP and Ctm PrP. Ntm PrP was not quantitated in these ex-periments, because negligible amounts of this form are produced in the presence of thymoma microsomes (16). We showed previously that substitution of arginine for leucine at position 9 (L9R) in the hydrophobic core (h-region) of the signal peptide had a dramatic effect on PrP membrane topology, with ϳ50% of the translocated protein assuming the Ctm PrP orientation, compared with ϳ10% for WT PrP (13). We then tested the effects of other amino acid substitutions at this site. The results are shown in Fig. 1 and summarized in Table I (lines 1-12). Substitution of either positively charged residues (Arg and Lys) or a negatively charged residue (Asp) for leucine at position 9 increased the proportion of Ctm PrP, with two non-polar residues (Pro and Gly) having very little effect. All of the substitutions also significantly reduced the efficiency of translocation (the total percentage of PK-protected chains) from ϳ25% for WT PrP to 5-15% for the mutants (data not shown). To examine the effect of substitutions at another residue in the h-region of the signal peptide, we analyzed V13R and V13D. Both of these mutations completely abolished translocation (Fig. 1).
Previous work demonstrated that certain mutations in the transmembrane segment also increased the amount of Ctm PrP. One mutation that has been studied extensively is the triple substitution designated 3AV (substitution of valine for alanine at positions 112, 114, and 117) (7,13). When this mutation was combined with mutations in the signal sequence, an additive increase in the percentage of Ctm PrP was observed ( Fig. 1; Table I, lines 1-12). PrP molecules carrying the 3AV mutation along with substitution of a charged amino acid (Arg, Lys, and Asp) at position 9 were synthesized almost exclusively as Ctm PrP.
Lack of Signal Peptide Cleavage Does Not Cause Production of Ctm PrP-We have shown previously (13) that Ctm PrP has an uncleaved signal peptide. However, it remained unknown whether the lack of signal peptide cleavage was a cause or a consequence of Ctm PrP formation. To address this question, we introduced mutations that prevent cleavage by signal peptidase, and we assayed their effect on synthesis of Ctm PrP. The  (30), are indicated. The upward arrow indicates the site of signal peptide cleavage between residues 22 and 23. The amino acid substitutions that were made at positions 9 and 13 within the signal sequence are written below the wild-type residues at these positions. The 3AV triple mutation at positions 112, 114, and 117 in the transmembrane domain is also indicated. The wild-type sequence shows methionine residues at positions 108 and 111, which constitutes the 3F4 epitope present in all constructs. B, PrP mRNAs encoding the indicated constructs were translated using reticulocyte lysate supplemented with murine thymoma microsomes. Translation reactions were subjected to PK digestion, and then analyzed by SDS-PAGE and autoradiography. The positions of the PK-protected fragments corresponding to Sec PrP and Ctm PrP are indicated to the left of lane 1. Aliquots of each translation reaction were also incubated without PK to visualize total translation products or in the presence of PK plus Triton X-100 to demonstrate that all products were completely digested (not shown). Molecular size markers (to the right of lane 12) are in kilodaltons. been shown to have the greatest influence on cleavage (30); small polar residues are strongly preferred. We therefore substituted large, hydrophobic residues at these positions to block the action of signal peptidase ( Fig. 2A). Fig. 2B (lanes 2, 5, and 8) shows that these mutations had no effect on PrP topology. As in the WT molecule, PrP containing either G20W or C22Y substitutions produced primarily Sec PrP, with ϳ10% being synthesized as Ctm PrP (Table I, lines [13][14][15]. This result was verified by assaying the topology of these mutants in transfected cells (data not shown). The mutant C22Y/L9R/3AV produced 90% Ctm PrP (Fig. 2B, lane 11; Table I, line 16), indicating that lack of signal peptide cleavage does not inhibit formation of Ctm PrP. We confirmed that the signal peptide remained uncleaved in the G20W and C22Y mutants by observing the slightly reduced mobility of the Sec PrP bands on SDS-PAGE (Fig. 2B, compare lane 2 to lanes 5 and 8). We conclude from these results that lack of signal peptide cleavage is a consequence, and not a cause, of Ctm PrP formation.
Deletion of the Signal Peptide Prevents Translocation-Ctm PrP has the topology of a type II transmembrane protein (N terminus in the cytoplasm). Most proteins of this type contain an internal signal-anchor sequence that initiates translocation by binding to the signal recognition particle, and also anchors the polypeptide chain in the lipid bilayer. To test whether the transmembrane segment of PrP could serve a membrane targeting function independent of the N-terminal signal sequence, we assayed PrP constructs in which the N-terminal sequence had been deleted. We observed that removal of the signal peptide completely abolished translocation of WT PrP, as evidenced by the failure to detect any protected fragments after PK digestion of translation reactions (Fig. 2C, lane 14). The same result was observed after introduction of the 3AV mutation, which normally increases Ctm PrP formation and might therefore be expected to enhance the ability of the transmembrane segment to function as a signal-anchor sequence (Fig. 2C, lane 17). Thus, the transmembrane segment cannot function independently to target PrP to the translocon to produce Ctm PrP but requires cooperation with the N-terminal signal sequence.
Residues on the C-terminal Side of the Signal Peptide Cleavage Site Influence PrP Topology-To demonstrate that Ctm PrP has an uncleaved signal, we had constructed previously a PrP molecule with a FLAG epitope inserted at the signal peptide cleavage site (amino acids 22/23) (Fig. 3A) (13). Insertion of this epitope (DYKDDDDK) changes the downstream amino acid context of the cleavage site from basic (KKRPKPGG) to acidic. We found that the presence of the FLAG epitope increases the proportion of Ctm PrP compared with untagged controls. This effect is relatively modest when microsomes from mouse thymoma cells are used (not shown) but more dramatic when  19 -22 and 29 -32). The Ctm PrP and Sec PrP fragments produced after PK digestion were quantitated by Phosphor-Imager analysis of SDS-PAGE gels. The percentage of Ctm PrP was expressed as Ctm Prp/( Ctm PrP ϩ Sec PrP) ϫ 100. Ntm PrP was not included in this calculation because it is present in negligible amounts in translations performed with thymoma microsomes (16). Each value represents the mean of 2-8 replicates.   Table I, lines 19 -22). In this system, the percentage of Ctm PrP is doubled by introduction of the FLAG sequence into either WT or 3AV PrP. We have shown previously that, for all PrP constructs, pancreatic microsomes produce a higher Ctm PrP/ Sec PrP ratio than thymoma microsomes (16), possibly due to differences between these preparations in their content of translation accessory factors (14,15). Insertion of the FLAG epitope does not alter the position of signal peptide cleavage, based on immunoreactivity with the monoclonal antibody M1, which is specific for the FLAG sequence containing a free N terminus (13).
Changing the Spacing between the Signal Peptide and the Transmembrane Domain Affects Membrane Topology-It has been proposed that the two topogenic determinants in PrP (the N-terminal signal peptide and the transmembrane domain) interact with other during the translocation process (9,11). To test this idea, we determined whether altering the spacing between these determinants affected PrP membrane topology. The spacing was altered by deleting or inserting octapeptide repeat units (P(H/Q)GG(G/S/T)WGQ), five copies of which are normally found in the N-terminal region of PrP (Fig. 4A). As shown in Fig. 4B and Table I (lines [23][24][25][26][27][28], decreasing the number of octapeptide repeats to 2, 1, or 0 progressively increases the proportion of Ctm PrP, whereas increasing the number of repeats to 14 has the opposite effect. Similar results were obtained when translations were performed using canine pancreatic microsomes, in which the proportions of Ctm PrP were higher, and it was easier to appreciate the reduction in Ctm PrP produced by insertion of additional repeats (PG14 construct) (data not shown). These results demonstrate that PrP membrane topology is dependent upon the timing with which the two topological determinants are presented to the translocation machinery.
Deletion of the GPI Addition Signal Does Not Affect Membrane Topology-PrP has a hydrophobic segment at its C terminus (residues 231-254) which serves as a signal for attachment of the GPI anchor. This segment, which is normally cleaved off during addition of the anchor structure in the ER, is capable of inserting into the membrane post-translationally when the N-terminal signal sequence and transmembrane domains have been deleted (8). To test the effect of the GPI addition sequence on the membrane topology of PrP in the context of the full-length protein, we analyzed constructs in which this sequence was deleted. As shown in Fig. 5 and Table  I (lines 29 -32), deletion of the GPI addition signal had no effect on the proportion of Ctm PrP, either in the WT protein or in a protein carrying the 3AV mutation. We also observed that replacement of the GPI addition signal with an unrelated sequence (the KDEL ER retention signal) had no effect on membrane topology (data not shown).

An Antibody Directed against the PrP Signal Peptide Specifically Recognizes Ctm PrP and Untranslocated PrP-Because
Ctm PrP, unlike Ntm PrP and Sec PrP, has an uncleaved signal peptide, we reasoned that an antibody raised against the signal peptide would be a useful tool for specifically assaying Ctm PrP in cells and tissues. Because of the hydrophobicity of the signal sequence made synthesis of synthetic peptides encompassing this region problematic, we chose as an immunogen a synthetic peptide that spanned the signal peptide cleavage site (Fig. 6A). This peptide included several positively charged residues on the C-terminal side of the cleavage site that facilitated synthesis of the peptide and improved antigenicity. This peptide was coupled to keyhole limpet hemocyanin via both amino and sulfhydryl groups and was used to raise a polyclonal antiserum (denoted anti-SP) in rabbits.
Initial characterization of the antiserum was performed using PrP synthesized by in vitro translation (Fig. 6B). When WT PrP was immunoprecipitated with an antibody (P45-66) that detects all forms of PrP, two bands were seen (Fig. 6B, lane 1): a 32-kDa species representing core-glycosylated Sec PrP chains, and a 27-kDa species representing unglycosylated, untranslocated chains that have not been processed at either their N or C termini (i.e. they retain both their signal and GPI addition peptides). The anti-SP antibody reacted only with the latter repeats. The stippled region represents the signal peptide, the lightshaded boxes represent the octapeptide repeats, and the cross-hatched region represents the transmembrane domain. B, PrP mRNAs encoding the indicated constructs were translated in reticulocyte lysate supplemented with murine thymoma microsomes. Reactions were then incubated with (ϩ PK lanes) or without (Ϫ PK lanes) proteinase K in the presence (ϩ Det. lanes) or absence (Ϫ Det. lanes) of Triton X-100. Products were immunoprecipitated with 3F4 antibody and deglycosylated with PNGase F prior to analysis by SDS-PAGE and autoradiography. Arrowheads and arrows indicate the positions of Sec PrP and Ctm PrP, respectively. On the autoradiographic exposures shown here, the Ctm PrP band is not visible for PG2, PG5, and PG14. However, small amounts of this form can be detected above background levels by Phos-phorImager analysis (see Table I, lines 25-27). form (Fig. 6B, lane 2), and this reactivity could be abolished by pre-incubation of the antibody with the peptide immunogen (Fig. 6B, lane 3). When L9R/3AV PrP was immunoprecipitated with P45-66, both core-glycosylated as well as unglycosylated, untranslocated forms were also observed, but the glycosylated form migrated at 33 kDa rather than 32 kDa because of the presence of an uncleaved signal peptide characteristic of Ctm PrP (Fig. 6B, lane 4) (13). Anti-SP recognized both the 33and 27-kDa species (Fig. 6B, lane 5), and again reactivity could be blocked by pre-incubation with the peptide immunogen (lane 6). These results demonstrate that the anti-SP serum recognizes PrP translation products with a retained N-terminal signal peptide and does not react with molecules whose signal peptide has been cleaved.
We then tested the ability of the anti-SP serum to recognize PrP synthesized in transiently transfected CHO cells (Fig. 6C). Cells were metabolically labeled with [ 35 S]methionine, and PrP was immunoprecipitated from lysates using either anti-SP antibody or 3F4 antibody (which recognizes all forms of PrP). Cells expressing WT PrP produced two mature, glycosylated species recognized by 3F4 antibody: a doubly glycosylated form at 38 kDa and a singly glycosylated form at 32 kDa (Fig. 6C,  lane 1). In addition, small amounts of unglycosylated PrP at 25-27 kDa were present. In contrast, cells expressing L9R/3AV PrP produced a 33-kDa glycosylated form that we have shown previously (13) represents an endoglycosidase H-sensitive form of Ctm PrP that has an uncleaved signal peptide (Fig. 6C, lane  2). Anti-SP antiserum immunoprecipitated the 33-kDa Ctm PrP species from cells expressing L9R/3AV PrP (Fig. 6C, lane 4), and trace amounts of the same band were detected in cells expressing WT PrP (lane 3). However, the doubly and singly glycosylated forms of WT PrP were not recognized by this antibody (Fig. 6C, lane 3), demonstrating its selectivity for signal peptide-bearing forms synthesized in cells. We noted that anti-SP also recognized an unglycosylated form of PrP (27 kDa), which we have shown represents untranslocated, unprocessed molecules that accumulate in the cytoplasm at high expression levels (22). These molecules are analogous to the untranslocated species observed after in vitro translation (Fig.  6B). Taken together, the results shown in Fig. 6, B and C, demonstrate that anti-SP antibody is capable of selectively recognizing PrP molecules that contain an uncleaved signal peptide, even in the presence of an excess of N-terminally processed forms. These signal peptide-bearing molecules, which are synthesized both in vitro and in cultured cells, are composed of Ctm PrP as well as untranslocated forms that remain on the cytoplasmic side of the ER membrane (13,22).
The Amounts of Ctm PrP and Untranslocated PrP Are Not Altered by Scrapie Infection of Cultured Cells and Brain-Ctm PrP has been proposed to be a neurotoxic intermediate whose levels are increased during infectiously acquired as well as familial prion diseases (7,12). To test this hypothesis, we used the anti-SP antibody to assay the amount of Ctm PrP in scrapie-infected N2a cells. Infected and uninfected N2a cells were labeled to steady state with [ 35 S]methionine, and then PrP was immunoprecipitated with either anti-SP antibody or with an antibody (8H4) that recognizes all forms of PrP. Because the level of endogenous mouse PrP in N2a cells is relatively low, some cultures were transiently transfected to express high levels of WT PrP, with the idea that this might enhance the ability to detect small amounts of Ctm PrP. In addition, some cells were transfected with constructs encoding PrP mutants (A116V, 3AV, and L9R/3AV) that would serve as positive controls for synthesis of Ctm PrP. The transfection efficiency in these experiments, measured using a GFP-encoding plasmid, was sufficiently high (ϳ30%) that we could be confident a substantial number of scrapie-infected cells in the culture expressed transfected PrP. The N2a cells used for these experiments represented a sub-clone (N2a.3) we had isolated that was highly susceptible to scrapie infection (see "Experimental Procedures"), so that the infected and uninfected versions could be directly compared without further cloning (25).  1 and 4), anti-SP antibody (lanes 2 and 5), or anti-SP antibody pre-incubated with the peptide immunogen (lanes 3 and 6). C, transiently transfected CHO cells expressing WT or L9R/3AV PrP were labeled for 6 h with [ 35 S]methionine. PrP was then immunoprecipitated from cell lysates using either 3F4 antibody (lanes 1 and 2) or anti-SP antibody (lanes 3 and 4) and analyzed by SDS-PAGE and autoradiography. The positions of doubly and singly glycosylated forms of mature PrP, unglycosylated PrP, Ctm PrP, and unprocessed PrP are indicated. Fig. 7A shows that there was no difference between infected and uninfected cells in the amounts of signal peptide-containing PrP they produced. Although untransfected cells synthesized low levels of PrP that required long autoradiographic exposures to visualize (data not shown), cells transfected with WT or mutant PrP plasmids expressed considerably higher levels of 8H4-reactive protein (lower panels). In cells transfected with the WT PrP plasmid, anti-SP antibody immunoprecipitated two bands of 27 and 33 kDa (upper panels, lanes 2 and 7). The 33-kDa species, which was sensitive to digestion with endoglycosidase H (not shown), was present in increased amounts in cells expressing A116V, 3AV, and L9R/3AV PrP (upper panels, lanes 3-5 and 8 -10), confirming its identity as Ctm PrP (13). The unglycosylated 27-kDa band corresponds to untranslocated PrP, which we have shown to be present in increased amounts in transiently transfected cells (22). Importantly, the amounts of both the 33-and 27-kDa species did not differ between infected and uninfected cells. We conclude from these results that scrapie infection of N2a cells does not detectably increase the amount of either Ctm PrP or untranslocated PrP. Western blotting confirmed that that the scrapie-infected cells produced protease-resistant PrP Sc (Fig. 7B). We note that the results shown in Fig. 7A are a further demonstration of the specificity of the anti-SP antibody for signal peptide-bearing forms of PrP, because Ctm PrP and untranslocated PrP are selectively immunoprecipitated in the presence of a large excess of processed PrP (compare anti-SP and 8H4 panels).
We also sought to measure the amounts of signal peptidebearing PrP in scrapie-infected brain samples. Initial experiments failed to detect any proteins that reacted specifically with anti-SP antiserum on Western blots of lysates prepared from either infected or control brains, a result that is presumably due to the low levels of these proteins in brain and to the presence of bands that react non-specifically with the antibody on blots (data not shown). As an alternative assay, we used SDS-PAGE to detect the small (ϳ2 kDa) difference in size between PrP molecules with and without a signal peptide. Proteins were enzymatically deglycosylated prior to SDS-PAGE to eliminate size differences due to differential glycosylation. When the gels were run long enough, we found we could reliably detect the difference in migration between PrP molecules containing a cleaved signal peptide (25 kDa) and an uncleaved signal peptide (27 kDa), as demonstrated by visualization of two bands in brain samples from Tg(L9R/3AV) mice (Fig. 8, lane 5). These mice, which spontaneously develop a severe neurodegenerative illness, synthesize Ctm PrP as well as signal peptide-cleaved forms. 3 The identity of the 27-kDa form as a signal peptide-bearing form was confirmed by immunoprecipitating it with anti-SP antibody from [ 35 S]methionine-labeled neurons cultured from Tg(L9R/3AV) mice (not shown). Importantly, we did not detect any of the 27-kDa band in brain samples from uninfected mice or from mice and hamsters infected with several different scrapie inocula (Fig. 8, lanes 1-4). Because samples were deglycosylated prior to SDS-PAGE, this analysis would not distinguish Ctm PrP from untranslocated PrP. We conclude from these data that scrapie infection of mice and hamsters does not increase the amount of either Ctm PrP or untranslocated PrP in the brain to detectable levels. DISCUSSION In this work, we have analyzed several structural features of the PrP molecule that influence its membrane topology. In addition, we have developed an antibody that selectively recognizes Ctm PrP and cytosolic PrP, topological variants of PrP implicated in prion-related neurodegeneration. We have used this antibody, as well as the distinctive gel mobility of Ctm PrP and cytosolic PrP, to measure the amounts of these forms in scrapie-infected cells and brain. Our results have implications for the mechanisms of protein translocation in the ER and for the role of transmembrane and cytosolic PrP in neurodegenerative disease.
Topological relative amount of Ctm PrP synthesized (7,12). Some mutations (e.g. A116V, K109I/H110I, and 3AV) increased the proportion of Ctm PrP, whereas other mutations (e.g. G122P) decreased it. Subsequently, we found that a non-conservative substitution (L9R) within the hydrophobic core of the N-terminal signal sequence dramatically increased the amount of Ctm PrP (13). Combining this mutation with the 3AV mutation resulted in a molecule (L9R/3AV) that was synthesized exclusively as Ctm PrP both in vitro and in cultured cells. We also found that disease-associated mutations that lie outside the transmembrane region did not alter the membrane topology of PrP (16). Taken together, these results indicated that there were two major topogenic determinants in the PrP molecule: the Nterminal signal peptide and the hydrophobic transmembrane domain. Another potential topological determinant, the C-terminal, GPI addition signal, does not appear to play a significant role. Although the GPI signal can mediate post-translational translocation of PrP molecules lacking an N-terminal signal sequence (8), deletion or substitution of this segment has no effect on the membrane topology of the full-length protein (this paper and see Ref. 11).
To explore further the role of the signal peptide in determining PrP topology, we introduced several additional substitutions in the hydrophobic core (h-region) of the signal peptide. We found that introduction of either positively or negatively charged residues at position 9 significantly increased the proportion of Ctm PrP. These substitutions also decreased the overall efficiency of translocation. Introduction of either a positively or negatively charged residue at position 13 completely abolished translocation. These results are consistent with the suggestion (9) that the PrP signal sequence serves two distinct functions: 1) targeting the polypeptide chain to the translocon; and 2) determining the location of the N terminus with respect to the membrane (luminal or cytoplasmic). Introduction of charged residues in the h-region affects both of these functions, reducing the targeting activity of the signal sequence, as well as its ability to promote translocation of the N terminus to the luminal side of the membrane. Previous work has shown that mutations in the n-region of the signal peptide (the polar, N-terminal segment) can also influence the two functions of the signal sequence (9,31).
Several other features of PrP topogenesis are also addressed by our results. First, we found that mutations in the c-region that block signal peptide cleavage do not alter the topology of PrP. Thus, the presence of an uncleaved signal peptide, which we have found is a characteristic of Ctm PrP, does not by itself favor formation of Ctm PrP. Rather, failure of signal peptide cleavage is likely to be a consequence of the fact that, in Ctm PrP, the N terminus of the protein remains in the cytoplasm and therefore does not come into contact with signal peptidase in the ER lumen.
We have also observed that deletion of the signal peptide prevents translocation, resulting in cytosolic molecules that are fully accessible to added protease. This result, which confirms other published reports (8,10), indicates that the hydrophobic, transmembrane segment cannot by itself target the polypeptide chain to the translocon. Thus, in contrast to the case for other type II membrane-spanning proteins (whose N termini are cytoplasmic), the transmembrane domain of PrP is not capable of functioning as a signal-anchor sequence. These considerations are consistent with the proposal (9, 11) that the primary function of the PrP transmembrane domain is to determine whether polypeptide chains that have already been targeted to the translocon by the N-terminal signal sequence will integrate into the lipid bilayer. In the model of Hegde and colleagues (11), the signal sequence and transmembrane domain thus mediate two sequential events within the translocon channel that determine the partitioning of PrP chains among the possible topological variants. Consistent with such a model, decreasing the distance between the signal sequence and transmembrane domain increases synthesis of Ctm PrP (this paper and see Ref. 11).
We find that introduction of a FLAG epitope just C-terminal to the signal peptide cleavage site increases the proportion of Ctm PrP synthesized. This result indicates that, although the signal sequence and transmembrane domain are the primary determinants of PrP topology, other parts of the protein can also have an influence. This observation is reminiscent of experiments in which the translocation of chimeric proteins containing the same signal sequence but different mature domains was measured (31). These experiments indicated that signal sequences and mature domains cooperate in luminal gating of the translocon pore, a step that is correlated with synthesis of Ntm PrP and Sec PrP. In this light, introduction of the FLAG epitope would decrease the efficiency of luminal gating and thus increase the relative amount of Ctm PrP.
An Immunological Assay for Ctm PrP and Cytosolic PrP-Previously, the only method available for direct measurement of Ctm PrP has been the protease-protection assay using microsomal membranes, which relies upon the generation of a protected fragment representing the luminal domain of Ctm PrP (7-9, 11, 16). However, this assay can be cumbersome to perform and is problematic when applied to tissue samples because of the difficulty of preparing purified, ER-derived, microsomal membranes. The assay is also not applicable to scrapie-infected samples because of the intrinsic protease resistance of PrP Sc . Another published method (7), which involves PK digestion of detergent extracts under "mild" conditions, has also been claimed to detect Ctm PrP, but there is no direct evidence that the PK-resistant fragment produced in this assay represents authentic Ctm PrP. Detection of cytosolic PrP has relied upon immunofluorescence staining and the use of proteasome inhibitors (19)(20)(21).
The presence of an uncleaved signal peptide on both Ctm PrP (13) and cytosolic PrP (22) provides a simple and direct method to detect these forms using an antibody directed against the signal sequence. We have prepared an antiserum against a synthetic peptide that spans the signal peptide cleavage site, and we have shown that this antiserum specifically immunoprecipitates Ctm PrP synthesized in vitro and in cells, even in the presence of a large excess of signal peptide-cleaved forms ( Sec PrP and Ntm PrP). The anti-SP antiserum also detects untranslocated forms of PrP that have not been processed at their N or C termini and that we have shown (22)  cytoplasmic side of the ER membrane. Thus, this antiserum is a useful reagent for specifically detecting untranslocated, cytosolic PrP as well as Ctm PrP.
Effect of Scrapie Infection on the Amounts of Ctm PrP and Cytosolic PrP-It has been proposed that Ctm PrP is a key pathogenic intermediate in both familial and infectiously acquired prion diseases (7,12). In this scheme, certain pathogenic mutations in PrP are thought to increase directly the synthesis of Ctm PrP, whereas in cases of infectious origin PrP Sc is hypothesized to indirectly cause accumulation of Ctm PrP. Although it is clear that Ctm PrP-favoring mutations cause a neurodegenerative phenotype when expressed in transgenic mice, there is a paucity of evidence that the amount of Ctm PrP actually increases during the course of a natural prion infection in either humans or animals. The only published experiment that supports this conclusion involved the use of a PrP reporter construct to monitor Ctm PrP levels in scrapie-infected mice (12). However, this assay did not directly quantitate Ctm PrP, and in addition, the maximal increase in Ctm PrP observed was only 3-fold, much smaller than the increase in PrP Sc . Cytosolic PrP has also been proposed as a neurotoxic intermediate, but the only experiments to support this claim derive from expression of artificial forms of PrP that lack an N-terminal signal sequence (18).
We report here that scrapie infection of N2a cells and mouse brain does not alter the amount of signal peptide-bearing PrP, based on reactivity with anti-SP antibody, or on SDS-PAGE to detect the slightly larger molecular weight of this form. These results indicate that scrapie infection does not affect the levels of either Ctm PrP or untranslocated (presumably cytosolic) PrP, both of which have an uncleaved signal peptide. Of course, it is possible that these forms were present below the level of detectability of our assay methods. Arguing against this possibility is the fact that we can easily visualize signal peptide bearing forms of PrP in uninfected and scrapie-infected N2a cells that have been transfected to express WT molecules or molecules carrying Ctm PrP-favoring mutations. We estimate that Ctm PrP represents ϳ0.5-1% of total PrP in cells expressing the WT protein. In addition, we can detect Ctm PrP in brain extracts from Tg(L9R/3AV) mice that spontaneously develop a severe neurodegenerative illness (Fig. 8). 3 If the amount of Ctm PrP had been elevated in the brains of scrapie-infected mice to levels similar to those observed in the brains of Tg(L9R/3AV) mice, it is thus likely we would have been able to detect this change.
A growing body of evidence indicates that, although PrP Sc is the infectious form of PrP, one or more alternate forms of the protein are responsible for the neurodegeneration observed in prion disorders (17). The results presented here, in conjunction with our previous work showing that levels of Ctm PrP and cytosolic PrP are not altered by disease-associated PrP mutations (16,22), suggest that these two forms are unlikely to be obligate neurotoxic intermediates in familial or infectiously acquired prion diseases. It will be important now to identify other neurotoxic forms of PrP (32) and to develop specific and sensitive assays to detect these species during the course of prion diseases.