Address correspondence and reprint requests to Dr Ruth Gabizon, Department of Neurology, Hadassah University Hospital, Jerusalem, Israel, 91120. E-mail: firstname.lastname@example.org
As many GPI anchored proteins, PrPC and its abnormal conformer PrPSc, are inserted into membrane microdomains known as rafts. Upon raft disruption, PrPC becomes soluble, while PrPSc aggregates into insoluble structures. It was recently published that, as opposed to PrPC, PrPSc, as well as its protease resistant core PrP27-30, can bind specifically to plasminogen and other serum components. These findings were suggested to have important physiological implications in transmissible spongiform encephalopathies (TSE) diagnosis and pathogenesis. In this work, we show that the binding of PrPSc or PrP 27–30 to serum proteins occurs only at specific detergent combinations, in which disease associated PrPs are present in aggregated structures. At detergent conditions in which rafts are intact, it is actually PrPC. that binds to blood proteins, albeit not directly, but through neighboring rafts components. Our results therefore indicate that the binding of PrPSc to blood components has no physiological relevance.
PrPSc, an abnormal conformer of PrPC, is the only known component of the prion, an infectious agent which causes transmissible spongiform encephalopathies (TSE; Prusiner et al. 1998). PrPSc and PrPC, which share the same amino acids sequence, are inserted via a glycophosphatidylinositol (GPI) anchor to cholesterol rich membranes microdomains known as rafts (Taraboulos et al. 1995; Naslavsky et al. 1997). Detergent extraction and proteinase K (PK) digestion of membranes from scrapie infected brains resulted in typical rod shape aggregates comprising the protease resistant core of PrPSc (PrP 27–30; McKinley et al. 1991).
In recently published work it was found that PrPSc and PrP27-30, as opposed to PrPC, interacts specifically with plasminogen and other serum components (Fischer et al. 2000; Maissen et al. 2001). In the experiments described by Fischer et al. (2000) normal mouse serum components (NMS) were attached to activated magnetic beads, to which brain homogenates of normal and scrapie infected mice were subsequently loaded. The brain homogenates applied to the resin were prepared by homogenization in the presence of NP-40/DOC (sodium deoxycholate), followed by dilution of the homogenates into 3% Tween20/3%NP-40 in PBS. At these conditions, only the prion specific PrP peptides were bound to the beads (Fischer et al. 2000). It was suggested that these findings may be used to develop a blood test for the diagnosis of TSEs. More important, it was also claimed by Fischer et al. that their data constitutes evidence for the possibility of TSE diseases transmission through blood.
In this work, we show that the binding of PrPSc to serum components is restricted to the specific combination of detergents used in the Fischer et al. manuscript (Fischer et al. 2000), and therefore has no physiological significance. We show here that these conditions of extraction resulted in rafts disruption, and subsequently in the aggregation of PrPSc. When the same experiments were performed in the presence of detergents in which rafts remained intact, it was actually PrPC and not PrPSc that was retained by the resin, suggesting the normal prion protein binds to serum proteins through its interaction with neighboring raft components.
Brain samples from either normal or scrapie infected FVB mice were homogenized on ice with 10 volumes of PBS (phosphate-buffer saline) in the presence of different detergents; 0.5% DOC/0.5% NP-40, 1% TritonX100, 1% NP-40, 1% DOC, 1% NOG or 1% sarkosyl. Following the addition of 100 µm PMSF, the homogenates were centrifuged at 3000 r.p.m. for 30 min, and the supernatants used for the different experiments. PrP27-30 samples were prepared by the addition of 40 µg/mL PK for 30 min at 37°C, prior to the PMSF addition.
Coupling to magnetic beads
The coupling was done as described (Fischer et al. 2000). Briefly, 1 mg of normal mouse serum (NMS) was diluted in 1 mL coupling buffer (0.1 m borate pH 9.5) and added to PBS washed Tosyl-activated paramagnetic Dynabeads M-280 (Dynal, 2 mL). After incubation for 24 h at 37°C, the beads were washed twice with 0.1% BSA/PBS (bovine serum albumin) for 5 min. Blocking buffer (1%BSA/PBS) was added to the beads for 4 h at 37°C.
10 µL of each brain homogenate was diluted into 1 mL PBS containing the appropriate detergent, as described in the text. After spinning the samples twice, the supernatants were added to the magnetic beads (after coupling with NMS) and shaken gently for 1 h at 37°C. Samples were washed twice with PBS in the presence of the appropriate detergent, and subsequently boiled in sample buffer and immunobloted with αPrP mAb 6H4 (Korth et al. 1997).
Mouse serum agarose (MSA; 50 µL; Sigma) were washed first with PBS with the appropriate detergent before coupling. Blocking was done by the addition of BSA 1% with the same buffer for 1 h at 4°C. The MSA-beads were washed twice with the appropriate buffer before the addition of the brain homogenate samples, for 1 h at 4°C. Beads were separated from unbound material by low centrifugation and washed twice in the incubation buffer. Pellets were subjected to immunoblotting with αPrP mAb 6H4.
Flotation of detergent insoluble complexes was performed as described by Naslavsky et al. (Naslavsky et al. 1997). Briefly, 100–150 mg of the appropriate brain samples were extracted with 700μl of ice cold buffer, containing 150 mm NaCl, 25 mm Tris HCl,pH-7.5, 5 mm EDTA and 1% Triton X100 (TX100), or 0.5%NP-40/0.5% DOC/PBS extracted brain homogenates were diluted into ice cold 3% NP-40/3% Tween20. All lysates were loaded into ultracentrifugated tubes (TLS 55, Beckman Industry). An equal volume of 70% Nycodenz in TNE (25 mm Tris-HCl, pH-7.5, 150 mm NaCl, 5 mm EDTA) was added and mixed with the lysate. An 8–35% Nycodenz linear step gradient in TNE was then overlaid above the lysate. The tubes were spun at 55 000 r.p.m. for 4 h at 4°C in a TLS-55 rotor. Fractions of 200μl each were collected from the top to the bottom of the tube and immunobloted with the anti-PrP mAb 6H4.
PrPSc interacts with serum components only at very specific conditions
For the evaluation of all experiments described in this manuscript, it is important to stress that undigested scrapie brain homogenates (panel b in all figures) was shown to comprise both PrPC and PrPSc isoforms (Meyer et al. 1986; McKinley et al. 1991), which are indistinguishable by their molecular weight or immunological properties. The two proteins can only be separated by testing their protease resistance, as depicted in panels c of Figs 1, 2 and 3.
We prepared magnetic beads coupled to NMS at the same conditions described by Fischer et al. 2000. Two different extract preparations (from normal or scrapie mouse brains in the presence and the absence of PK) were applied to these beads. For the first one the brain samples were homogenized in 0.5% NP-40 and 0.5% DOC and a sample from these homogenates was subsequently diluted (10 µL in 1 mL) in PBS containing 3% NP-40 and 3% Tween20 (see methods and Fischer et al. 2000). For the second one the brain samples were homogenized and diluted in 1% Triton X100, to contain the same final protein concentration as the first set of extracts. All extracts (normal and scrapie infected in the presence and absence of PK) were applied to the NMS-beads and washed after the incubation, each in the same detergent used for its coupling. As can be seen in Fig. 1, only PrPSc and PrP27-30 remained attached to the beads in the NP-40/Tween20 buffer. We conclude this from the fact that no PrP from normal brain was bound to the beads in this conditions (panel a), and therefore the full length PrP in panel b1, probably represents PrPSc. In the Triton buffer, it was actually PrPC that was specifically bound to the beads, as can be seen from the fact that the full length PrP in panel b2 was protease sensitive. The band seen in the control lane, reacted also with secondary α mouse antibody, suggesting it represents light chain IgG leaking from the NMS beads.
PrPSc extracted with the NP-40/DOC/Tween20 detergent combination is no longer inserted in membranal rafts
While both PrP isoforms are inserted into cholesterol rich membrane microdomains, denominated rafts, PrPSc was shown to reside in rafts heavier than those containing PrPC (Naslavsky et al. 1997). Following extraction in cold Triton X100, rafts float in a nycodentz gradient, carrying all proteins inserted or attached indirectly to them. The exact gradient fractions in which each protein is present will depend on the nature of the raft it is inserted in. Soluble or aggregated proteins will be found at the bottom of the gradient. PrPC and PrPSc are not inserted into the same rafts (Vey et al. 1996; Naslavsky et al. 1997), therefore disruption of rafts with detergents may subject each of the prion proteins to a different change in gradient location.
To understand the disparate results obtained in Fig. 1, we subjected the brain extracts (normal and scrapie infected with and without PK, each at both detergent combinations) to a floatation assay. Figure 2 shows an immunoblot developed with α PrP mAb 6H4 of the gradient fractions for the different extracts. While extraction of the normal brain homogenates in cold Triton X100 retains most of the PrPC in the higher fractions representing the light rafts, in the brain extracts prepared as described by Fischer et al. 2000, PrPC was present in lower fractions, probably representing heavier rafts or partially disrupted ones. When the scrapie homogenates were subjected to the same experiments, the NP-40/Tween20 detergent conditions produced complete disruption of the rafts carrying PrPSc and PrP 27–30, as can be seen from the fact that protease resistant PrP moved to the lowest gradient fractions (panels b and c). Since PrP 27–30 is known to aggregate in the presence of non-denaturing detergents (Meyer et al. 1986; McKinley et al. 1991), we speculate it is present in the lowest fractions of the floatation gradient in aggregates resembling prion rods. As stated above, in its aggregated state, PrPSc or its protease resistant core, PrP 27–30 cannot be purified by chromatography procedures.
PrPC binds to serum components indirectly
PrPC may interact with serum components directly or through neighboring raft components, i.e. ‘Raft chromatography’ (Keshet et al. 2000). To distinguish between these possibilities, we tested the binding of PrPC, PrPSc and PrP 27–30 to NMS-beads in the presence of several detergents; 1% TritonX100, 1% NP-40 and 1% NOG (n-octyl-beta-d-glucopyranoside), mild detergents in which rafts are insoluble, and 1% DOC, 1% Sarkosyl, 0.5%NP-40/0.5% DOC, 3%NP-40/3% Tween 20, detergent conditions which solubilize rafts (Brown and Rose 1992; Naslavsky et al. 1997; Keshet et al. 2000). In these experiments, we used sepharose bound to NMS instead of magnetic beads to show that the results are due to the ligand (NMS) and not to a specific resin (magnetic beads or sepharose).
Figure 3 depicts the results of such an experiment. As can be seen in the figure, PrPC interacted with the NMS beads only in the presence of detergents that do not disrupt rafts, such as Triton X100 and NOG, suggesting its interaction with serum components occurs through the association of PrPC with neighboring rafts components. In contrast to PrPC, PrPSc and PrP27-30 did not bind to the beads at any of the tested conditions except those used in the Fischer manuscript (Fischer et al. 2000).
The fact that PrPSc interacted with the NMS-beads in the presence of NP-40/Tween20 but not in the presence of sarkosyl or DOC, detergents that also cause the aggregation of the prion isoform (McKinley et al. 1991), is highly significant. We speculate that the prion aggregates formed in the presence of NP-40/Tween20 may be less tight and expose to the outside other components of prion rods, which bind easily to serum components. Another possibility is that the binding of the prion aggregates to the beads is less effective in the presence of negatively charged detergents such as DOC or sarkosyl.
Heparin blocks the binding of PrPSc aggregates to NMS
The prion proteins were shown to have a close relationship with heparin and heparin like molecules. While PrPC was shown to bind to an heparin sepharose resin (Gabizon et al. 1993; Caughey et al. 1994), similar experiments were not performed for PrPSc due to its insoluble nature. Even though, aggregates of PrPSc were shown to contain sulfated sugar polymers (Appel et al. 1999; Snow et al. 1990; Shaked et al. 2001a). In addition, sulfated sugars were shown to inhibit the accumulation of PrPSc in scrapie infected neuroblastoma cells (Caughey et al. 1993; Caughey and Raymond 1993; Gabizon et al. 1993). We therefore tested whether the binding of PrPSc aggregates to serum components in the Fischer experiments were mediated by molecules such as heparin. Heparin is known to interact with several serum proteins, including plasminogen (Ledoux et al. 2000).
To this effect, we incubated the NMS-beads with heparin before their incubation with scrapie brain extracts prepared at the Fischer et al. (2000) conditions. It is important to point out that heparin was added to the beads, which were washed to remove unbound excess of the sugar polymer, before their incubation with the brain extracts. At such conditions, the binding of PrPSc or PrP 27–30 to the NMS beads was mostly inhibited (Fig. 4). It is therefore conceivable that the binding of PrPSc aggregates to serum proteins, under the conditions described by Fischer et al. (2000) occurs through the interaction of such proteins with heparin like compounds present in prion aggregates.
The conversion of PrPC to PrPSc has been attributed to changes in protein conformation, from α helix to β sheet, by a mechanism which is yet to be determined (Pan et al. 1993; Telling et al. 1995; Post et al. 1998; Cohen 1999; Jansen et al. 2001). No reagents have been shown to date to efficiently separate between the two prion isoforms, however, PrPC and PrPSc can be recognized by their physical and biochemical properties (Prusiner et al. 1998). While PrPC is soluble in detergents and sensitive to protease digestion, PrPSc aggregates in the presence of detergents and results following protease digestion in a protease resistant peptide known as PrP 27–30 (McKinley et al. 1991).
The results presented here indicate that the binding of PrPSc to serum proteins represents another method to distinguish in vitro between PrPSc and PrPC. Because of their different membrane location and aggregation properties, both proteins bind to serum components at different conditions, PrPC when rafts are intact, and PrPSc when rafts are disrupted by non-ionic detergents. However, and since PrPSc did not bind to serum components unless it was aggregated at specific conditions, none of these findings bears any significance as to the infectivity of blood from TSE infected animals or humans since in physiological conditions, no detergent induced aggregation of PrPSc has been shown to occur.
Some important conclusions can be drawn from the results presented here. The fact that only PrPC interacted with serum components when rafts were intact, reinforces the floatation results(Naslavsky et al. 1997) suggesting that both proteins are not present at the same membrane microdomains. Also, the fact that heparin could inhibit PrPSc binding to NMS-beads indicates yet again the presence of heparin like molecules in prion rods, as suggested by previous experiments (Snow et al. 1990; Appel et al. 1999; Shaked et al. 1999).
Interestingly, we have recently shown that PrPSc like molecules are present in the urine of TSE infected animals and humans, and that they may originate from blood (Shaked et al. 2001b). Whether these findings imply that blood samples can carry prion infectivity remains to be established.
The findings presented in this manuscript emphasizes the importance of the condition of detergent extractions in determining whether protein–protein interactions are direct or indirect, especially when raft associated membrane proteins are part of such interaction (Keshet et al. 2000).