The chromophore of GFP is made up of the tripeptide sequence Ser-Tyr-Gly that cyclizes in the folded form of the protein 4445. Denaturation and reduction abolish fluorescence but it can be recovered by dilution into a refolding buffer where the rate of fluorescence re-acquisition parallels protein folding 4647. GFP extended at the N-terminus can also be refolded with a similar recovery of fluorescence 4041. To assess this technology as a measure of PrP^c folding, we expressed GFP appended at the N-terminus with the complete mature prion protein (residues 23–231) and a short fragment of PrP^c, residues 76–156, as a control for the effect of size of the amino terminal extension on refolding (Fig 1A). Expression of the PrP^c _23–231-GFP and PrP^c _76–156-GFP fusion protein in E.coli led to the accumulation of non-fluorescent insoluble inclusion bodies that were purified to ~90% (Fig 1B) and then denatured before dilution into refolding buffer. GFP fluorescence (510 nm) rose with time to a maximum refolding level of ~6 fold for PrP^c _23–231-GFP and ~25 fold for PrP^c _76–156-GFP over background within 3 hrs under the conditions of the experiment (Fig 1C). Ranging experiments showed optimal refolding to occur at >pH8 and at 21°C (not shown). We conclude from this data that 1) PrP^c-GFP fusion proteins can refold in vitro to regenerate the GFP chromophore and 2) the level of refolding is related to the PrP^c sequence fused to GFP as alteration of the fragment size altered the rate of fluorescence recovery. The observed fluorescence was directly attributable to refolding of PrP^c-GFP as re-examination of the fusion proteins after the refolding assay showed full length protein in solution with no evidence of breakdown to release free GFP (Fig. 1B). As PrP^c binds both copper 34264849 and RNA 505152 the effect of both of these ligands on the refolding reaction of full length prion protein present in construct PrP^c _23–231-GFP was assessed. However, neither addition of copper (100 nM) nor RNA, prepared as described 51, significantly altered the rate of fluorescence recovery for the full length prion protein, which remained slow when compared to the shorter variant (see Additional file 1).

Previous expression of PrP^c-GFP fusion proteins within eukaryotic cells indicated a marked effect of the extreme N-terminal basic residues 23–28 on prion protein processing 5354 and further studies have suggested an interaction between the extreme amino terminus and the C terminal folded domain 43 extending an earlier antibody binding study 55. In order to assess directly if the N terminal sequence affects folding per se, amino terminal truncations were made in which PrP^c residues 23–156, 29–156, 23–169 and 29–169 (see Additional file 2) were appended to the N terminus of GFP and the fusion proteins purified as an insoluble fraction prior to dilution into the refolding reaction (Fig. 2A). When equimolar amounts of each fusion protein were subjected to the refolding assay, the rates of fluorescence reacquisition were found to vary considerably (Fig. 2B). The presence of residues 23–28 at the extreme N-terminus of PrP^c severely limited refolding in the context of a fragment truncated at residue 156 with overall refolding little better than the complete 23–231 PrP^c sequence despite being a considerably shorter fragment (cf Fig. 1). Deletion of residues 23–28 (construct PrP^c _29–156-GFP) enhanced fluorescence recovery ~4 fold when compared to PrP^c _23–156-GFP (Fig. 2B). However, a fragment starting at residue 23 but with an extended C-terminal truncation point at residue 169 (PrP^c _23–169-GFP), refolded far more efficiently than PrP^c _23–156-GFP (Fig. 2B) and deletion of the amino terminal 6 residues in PrP^c _29–169-GFP failed to improve the level of fluorescence observed. Fluorescence recovery was associated with equivalent quantities of soluble full-length fusion protein as no free GFP was apparent when the refolded samples were analysed by SDS-PAGE after removal from the refolding reaction (Fig. 2A). Thus, recovery of fluorescence by PrP^c-GFP fusion proteins in vitro following denaturation and renaturation measures a direct role for residues 23–28 and 156–169 in folding and mirrors the expression patterns observed for prion protein fragments of the same endpoints in vivo 43.

That GFP fluorescence recovered in vitro reflected properties measured in eukaryotic cells suggested that PrP^c-GFP fusions retained a degree of physiological significance. We sought therefore to use fluorescence for the direct selection of prion mutants with altered folding properties. To do this we used the plasmid encoding PrP^c _23–231-GFP as template for error prone PCR based mutagenesis 56 of the PrP^c sequence followed by substitution of the degenerate amplified material for the wild type sequence in order to generate a library of random PrP^c mutations fused to GFP (see Additional file 3). Nucleotide sequencing of several library members picked at random showed a variety of sequence changes causing premature stop codons as well as single or multiple amino acid changes within the PrP^c coding region (not shown). To select altered folding variants the library was plated at high density, replicated to agar plates containing IPTG and colonies were screened for fluorescence following irradiation with ultraviolet light. The overall number of fluorescent colonies was low and after eradication of false positives three mutants (M17, M22 and M25), which showed particularly strong fluorescence, (Fig. 3) were isolated and characterised further. As a recovery in fluorescence could indicate a change in folding and solubility bacterial cultures of the parental construct and each fluorescent variant were induced, harvested and lysed and the level of PrP^c-GFP fusion protein present in the soluble and insoluble fractions was assessed by western blot using the PrP^c monoclonal antibody 6H4 (epitope 144–152). As noted the parental sequence was wholly insoluble but significant amounts of the fusion protein from variants M22 and M25 and approximately 50% of the protein from mutant M17 were found in the supernatant fraction (Fig. 4). One variant (M22) showed substantial proteolysis leading to loss of full length antibody reactive material in the supernatant fraction, a characteristic of soluble PrP^c expression in E.coli 57. DNA sequencing of each variant revealed that M22 and M25 each had two amino acid changes, E152V+N48S and Y149H+G228E respectively while variant M17 showed only a single amino acid change at H84Q (Figure 5). Thus, changes of as little as one or two amino acids throughout the PrP^c polypeptide chain can cause significant alteration in protein folding. None of the mutations selected by this procedure occurred in the prion hydrophobic sequence (amino acids 111–133) rather, as suggested, change of charge was the predominant feature observed 58.

E.coli Top 10 (Invitrogen) was used throughout for cloning. Plasmids were transformed into E.coli BL21 DE3 (pLysS) (Novagen) for T7 driven protein production.

Mouse Prnpa allele (accession A23544) and enhanced green fluorescence protein (accession AAC53663) were used throughout. cDNA fragments encoding amino acids 23–231 and the N-terminal residues 23–156, 29–156, 76–156, 23–169 and 29–169 were amplified by the polymerase chain reaction (PCR) to be flanked by restriction sites for Bam H1 and first cloned into baculovirus transfer vector pAcVSV_GTMGFP 79 for expression in insect cells 43. Each construct was then used as a template to amplify the sequence encoding the fusion of PrP^c and eGFP flanking the sequence with restriction sites Nde1 and Xho1 at the 5' at the 3' ends respectively. Fragments were digested with the same enzymes and cloned into pET23a (Novagen) through the same sites to produce PrP^c-GFP gene fusions under the control of the T7 promoter.

A degenerate library of prion sequences was created by error prone PCR 56 and cloned en masse into pET23a upstream of, and in frame with, a sequence encoding eGFP. Several library members were picked at random for nucleotide sequencing to ensure errors had been introduced. The library was maintained in E.coli BL21 pLysS in an un-induced state and induced for fluorescence screening by replica plating to agar containing 2 mM IPTG. Colonies were screened visually after a further 5 hours incubation and positives re-streaked to ensure positivity bred true. Once confirmed, uninduced colonies were re-streaked from the master plate and DNA isolated for sequencing.

IBs were prepared by a modified differential solubility regime 80. Following inoculation of a single colony into Luria broth cultures were induced with IPTG (0.2 mM) at an OD600 of 0.5. Cultures were grown for a further 2 hours and bacteria harvested by centrifugation at 4500 rpm for 20 minutes at 4°C. The pellet was resuspended in 10 ml PBS and the IBs released by sonication on ice for 10 minutes, 1% triton X-100 (v/v) was added to complete solublization and the IBs collected by centrifugation at 4500 rpm for 10 minutes. The pellet was washed repeatedly with 1% Triton X-100 until the purity of the IBs was at least 90 % as judged by SDS-PAGE.

IBs were denatured and reduced at 95°C for 5 minute in 4 M Urea and then clarified by ultracentrifugation. Refolding was initiated by a single 7× dilution step into a buffer containing 50 mM Tris.HCl pH8.5, 1 mM KCl, 2 mM MgCl₂. Recovery of fluorescence over time was monitored by periodic fluorescence measurement at 510 nm in a Genios microplate reader (Tecan). Assays were done in duplicate and the average fluorescent units plotted against time. To assess the role of metal ions in refolding buffers were depleted for ions my mixing with chelex-100 (Bio-Rad) as described 25 and filtering prior to constitution of the assay. Ranging studies showed that the addition of copper above 10 micromolar was found to be generally inhibitory (i.e. inhibited the refolding of GFP only).

Total RNA for inclusion in the refolding assay was prepared from SNB cells as described for RNA that stimulates PrP^c-PrP^Sc conversion 51. Briefly, cells were washed with PBS and resuspended in 1 ml of Trizol (Invitrogen). The lysate was extracted with chloroform and the RNA recovered by precipitation with isopropanol. The pellet was washed with 75% ethanol, air dried, resuspended in RNA-ase free water and quantitated by A₂₆₀.

RNA stimulated partial protection of PrP^c was assessed by digestion of the reaction products after refolding with protease K as described 52.

Protein samples to be analyzed were separated on pre-cast 10% Tris-HCl SDS-polyacrylamide gels (Bio-Rad) and transferred onto Immobilon-P transfer membrane (Millipore). Western blotting was performed as described (Burnette, 1981) except that sensitivity was increased through the use of a biotin conjugated secondary antibody followed by extravidin-peroxidase (Sigma). The membrane was finally developed with BM Chemiluminescence (Roche). The primary antibodies used were prion monoclonal antibodies 6H4 (Prionics) and anti-GFP (Clontech).