|Home | About | Journals | Submit | Contact Us | Français|
Ure2p of Candida albicans (Ure2albicans or CaUre2p) can be a prion in Saccharomyces cerevisiae, but the Ure2p of Candida glabrata (Ure2glabrata) cannot, even though the Ure2glabrata N-terminal domain is more similar to that of the S. cerevisiae Ure2p (Ure2cerevisiae) than is Ure2albicans. We show that the N-terminal N/Q-rich prion domain of Ure2 albicans forms amyloid that is infectious, transmitting [URE3alb] to S. cerevisiae cells expressing only the C. albicans Ure2p. Using solid-state NMR of selectively labeled C. albicans Ure2p1–90, we show that this infectious amyloid has an in-register parallel β-sheet structure, like that of the S. cerevisiae Ure2p prion domain and other S. cerevisiae prion amyloids. In contrast, the N/Q-rich N-terminal domain of Ure2glabrata does not readily form amyloid, and that formed on prolonged incubation is not infectious.
A prion is an infectious protein, a protein that can transmit an infection without a required nucleic acid. The non-chromosomal genes [URE3] and [PSI+] were identified as prions of Ure2p and Sup35p of S. cerevisiae based on their unique genetic properties: (i) reversible curability, (ii) prion generation induced by overproduction of the corresponding protein and (iii) phenotype of prion similar to that of mutants in the corresponding gene needed for maintaining the prion (1). [PIN+] was detected as a non-chromosomal factor needed for inducing [PSI+] appearance by overproduction of Sup35p (2), and shown to be a prion of Rnq1p by the above genetic criteria (3). The [SWI+] and [OCT+] prions (4, 5) were uncovered because their respective proteins, Swi1p and Cyc8p, were found, when overproduced, to have properties like the [PIN+] prion. [MOT3+] was found in a screen of proteins with Q/N rich regions (6). Each of these prions is based on self-propagating amyloid formation by a Q/N - rich protein domain (the prion domain) (e.g. (7–9).
Prions of Ure2p from other Saccharomyces species have also been described (10–13), but the Ure2p of Saccharomyces castellii cannot be a prion in S. cerevisiae (13). When expressed in S. cerevisiae, the Ure2p's of Candida albicans, Kluyveromyces lactis and Schizosaccharomyces pombe were reported unable to form a prion, even after overexpression of the respective putative prion domain (11), and the K. lactis Ure2p cannot be a prion in K. lactis (14). Sup35 domains N, M and C, from N- to C-terminus, are the prion domain (necessary for normal mRNA turnover (15)), a charged middle domain and the C domain necessary for translation termination (9). The N-terminal domains of Sup35 of several yeast species, including Candida albicans, can be prion domains when fused to the S. cerevisiae C domain (16–18). This is a significant qualification because prion - forming ability of prion domains of Ure2p, Sup35p and HET-s are consistently inhibited by the presence of the remainder of the molecule (8, 19, 20), perhaps by some stabilizing effect. Hsp104 is a disaggregating chaperone that is necessary for the propagation of each of the amyloid-based yeast prions (2, 4, 5, 21, 22). Its role appears to be that of breaking amyloid filaments to form new prion 'seeds' (23). The Candida albicans Hsp104 homolog is capable of substituting for the S. cerevisiae Hsp104 in propagating [PSI+], suggesting that C. albicans may have an environment compatible with prion propagation (24).
Amyloid is a filamentous protein polymer that is rich in β-sheet, shows special dye-binding properties and is usually more protease resistant than the non-amyloid form of the protein. The amyloids of the prion domains of S. cerevisiae Ure2p, Sup35p and Rnq1p are infectious (25–28), and each has an in-register parallel β-sheet structure (29–31). Measurements of mass per unit length for each are consistent with this structure and inconsistent with a β-helix model (32–34), and the fact that the Ure2p and Sup35p prion domains may each be shuffled in sequence and yet still form prions predicts the same structure (35–37). The in-register parallel β-sheet structure also provides a simple explanation of prion variants, with the different locations of the folds of the β-sheet 'inherited' by new molecules joining the end of the filament (38, 39). In brief, the same hydrogen bonds and hydrophobic interactions between identical side chains of residues aligned in the parallel in-register beta sheets that hold the beta strands in-register, will direct the monomer joining the end of the filament to acquire the same conformation as the other molecules already in the filament. The location of turns (folds of the sheet) and the extent of b-sheet structure will be faithfully propagated, but may differ among prion variants.
We have found that the full length C. albicans Ure2 protein can form a [URE3] prion in S. cerevisiae, but the C. glabrata Ure2p cannot (40). Here we show that the C. albicans Ure2p prion domain forms amyloid more readily than that of C. glabrata, which forms amyloid only over years. Moreover, the C. albicans Ure2p prion domain amyloid is infectious, transmitting [URE3alb] to S. cerevisiae cells expressing C. albicans Ure2p. We present solid - state NMR data suggesting that, like the S. cerevisiae prion domains, the C. albicans Ure2p prion domain forms amyloid with an in-register parallel β-sheet structure.
S. cerevisiae strain BY302 (MATa his3 leu2 trp1 CaURE2 PDAL5:ADE2 PDAL5:CAN1 kar1) and BY304 (MATa his3 leu2 trp1 CgURE2 PDAL5:ADE2 PDAL5:CAN1 kar1) contain the cerevisiae URE2 ORF perfectly replaced by the C. albicans and C. glabrata URE2 ORFs, respectively (40). Strain YHE1161 is BY302 carrying [URE3alb] (40). Rich medium (YPAD) contains 1% yeast extract, 2% peptone, 2% dextrose, 2% agar and 0.4 g/l adenine sulfate. Minimal medium (SD) contains 0.67% Difco Yeast Nitrogen Base without Amino Acids, 2% dextrose, and 2% agar. YES medium, for scoring [URE3], contained 0.5% yeast extract, 3% dextrose, 30 mg/l tryptophan and 2% agar.
C. albicans and C. glabrata URE2 expression vectors were all constructed in pET vector backbones with either the N-terminal domain (or “PD”) or the full length proteins attached to a C-terminal His-6 tag (Table 1).
Strains BY302 and BY304 were used as transfection recipients using the method of Brachmann et al. (27, 41). Clones were selected for transformation using pRS425 (42). Filaments or extracts of the [URE3alb.] strain YHE1161 (40) were sonicated three times for 20 seconds each at output level 2–3 with 20–30% duty cycle (Branson Sonifier 250) with cooling on ice between sonications.
Plasmids in BL21 (DE3) RIPL (Stratagene) were grown overnight with 34ug/mL chloramphenicol, 50ug/mL streptomycin, and 100ug/mL ampicillin and innoculated into LB medium, or synthetic complete medium(43) with 200 mg/l of Ile-1-13C or Val-1-13C or 1g/l of Ala-3-13C for site-specific labeling. Mass spectrometry showed that Ala residues were 60% labeled and no significant leakage of label into other amino acids was detected. The latter medium without amino acids, but containing 2 g/l 15NH4Cl and 5 g/l uniformly 13C-labeled glucose was used for uniform labeling of CaUre2p1–90. At A600 ~0.6–1.0 1mM isopropyl-β-D-thiogalactosidase (IPTG) was added and after four hours of further growth, cells were harvested. CaUre2p1–90 or CgUre2p1–100 were obtained from bacteria lysed by vigorous agitation with 6 M guanidine-HCl, 0.1 M Tris, 150 mM NaCl, pH 8.0–8.5 and incubation at room temperature for 1–2 hours. Full length protein of both C. glabrata and C. albicans was obtained by lysing cells using a French pressure cell in 50 mM NaHPO4, 300 mM NaCl pH8 with Roche protease inhibitors (complete EDTA-free) . Lysates were cleared of cell debris by ultracentrifugation at 30000 rpm for 30–60min (Beckman-Coulter Optima L-90k Ultracentrifuge, Type 45Ti rotor). Cleared lysates were incubated with nickle-nitrilotriacetic acid (NiNTA) agarose (Qiagen) at room temperature with gentle agitation, loaded into a glass column (Biorad), washed with urea buffer (8M urea, 0.1M Tris, 150mMNaCl) for the N-terminal fragments or 50 mM NaHPO4, 300 mM NaCl pH8 for the full length proteins. After washing the column with the same buffer with 20mM imidazole, proteins were eluted with the same buffer with 200 mM imidizole. Protein concentrations were roughly determined by 4–12% NuPAGE Bis-tris gel (Invitrogen) stained with SimplyBlue Safestain (Coomassie G-250 stain, Invitrogen) and by BCA assay (Pierce).
If necessary, further purification used a Q-sepharose column (HighQ, Biorad). Fractions containing protein were diluted 5-fold with zero salt urea buffer (8 M urea, 50mM Tris) and loaded onto the Q-column. The column was washed with same zero salt urea buffer. Protein was eluted with a gradient of 0–200mM NaCl added in the same buffer.
Selected fractions were pooled and concentrated and dialyzed against 50 mM Tris Cl,150 mM NaCl pH 8 to induce amyloid formation. Filaments were formed at room temperature with agitation for two days to one week. A variety of buffers were tested in an effort to induce filament formation by C. glabrata Ure2p1–100. Buffers tested included 5 mM KPO4, 150 mM NaCl, pH 6.8; 50 mM TrisCl, 300 mM NaCl, pH 8; 3 M urea, 0.1 M Tris Cl, 150 mM NaCl, pH 8.3; 20 mM Tris Cl, 200 mM KCl, pH 8; 50 mM Tris Cl, 150 mM NaCl, pH 8. Prolonged incubation of C. glabrata Ure2p1–100 produced only limited amyloid (see Results) and SDS gel analysis of the still soluble material showed no change in protein pattern after 2 years.
Sample preparation, X-ray fiber diffraction procedure and data analysis were carried out as described previously (31).
Ten μl of a suitable dilution of filaments was deposited on a carbon-coated copper grid, washed once with water and stained with 2% uranyl acetate for two minutes. The stain was blotted off, the grid allowed to air dry and images obtained using an FEI Morgagni transmission electron microscope at 80 kV.
Solid-state NMR experiments on selectively labeled CaUre2p1–90 were carried out on a Varian InfinityPlus spectrometer at 9.39 T (100.4 MHz 13C NMR frequency) using magic angle spinning (MAS) in 3.2 mm rotors at room temperature. Although some cooling was used, rotors spun at 20 kHz may warm up to ~40C. One dimensional spectra were acquired at 20 kHz MAS using 1H-13C cross polarization (44) and two-pulse phase-modulated 1H decoupling (45). The dipolar recoupling measurements were carried out at 20 kHz MAS using the PITHIRDS-CT method (46) with pulsed spin-lock detection for improved signal-to-noise ratios (47). C.a.Ure2p1–90 filament samples were washed with water, lyophilized and packed in thick-walled 3.2 mm rotors. 13C T2 values under PITHIRDS-CT measurement conditions were measured by incrementing the total pulse sequence length while keeping the effective recoupling time equal to zero, to assure that no less than 25% of the signal remained at the end of the dipolar recoupling period.
PITHIRDS-CT data for an unlabeled sample of C.a.Ure2p1–90 were measured (Fig. 3A) and used to correct for the 1% natural abundance 13C, assuming linear signal decay: natural abundance signal = Sna = 100 - 0.2 t, where t is the effective dipolar dephasing time in milliseconds. The number of natural abundance 13C nuclei per C.a.Ure2p1–90 molecule, Nna, is 0.011 × (total sites - labeled nuclei), where total sites is 139 carbonyls or 30 methyls. Labeled nuclei, Nlabel, was 2 for Ile-1-13C, 5 for Val-1-13C and 2 for Ala-3-13C. The fraction of natural abundance 13C is Fna = Nna/(Nlabel + Nna). Experimentally determined 13C T2 values were used to correct the magnitude of the natural abundance signal when its T2 differed from that of the specifically labeled sample. Raw PITHIRDS-CT data were corrected according to the following: S(t) = (Sraw - Fna*(100 - 0.2t))/(1-Fna).
Two dimensional NMR experiments with uniformly 15N, 13C-labeled rehydrated amyloid filaments of C.a.Ure2p1–90 in 3.2 mm rotors were carried out at 17 kHz MAS. The mixing period comprised 48 rotor periods (2.82 ms) of finite pulse RFDR (radio-frequency driven recoupling) with 18 μs 13C π pulses. Continuous proton decoupling at 110 kHz was used during the mixing period, and TPPM (two pulse phase-modulated) decoupling (45), also at 110 kHz, during the t1 and t2 periods. 2D NMR data were analyzed using NMRPipe software (48) and displayed relative to tetramethylsilane using an Ala-1-13C standard.
The N-terminal domain of the C. albicans Ure2p is presumed to be the prion domain based on its Q/N richness and the finding that its overproduction induces the de novo appearance of [URE3] (40). We chose to study C.a.Ure2p1–90 because beyond residue 90, homology with the C-terminal domain of the S. cerevisiae Ure2p begins. C.a.Ure2p1–90H6 was purified under denaturing conditions from E. coli (see Methods) and allowed to form filaments in the absence of denaturant. Filaments formed within a few days of gentle agitation at room temperature in 50 mM Tris Cl pH 8.0, 150 mM NaCl (Fig. 1). Fibrils of C.a. Ure2p1–90H6 were straight, apparently unbranched and about 6 nm in diameter, with frequent lateral association (Fig. 1). Some “twisted pairs” were occasionally observed (Fig. 1A). These filaments showed yellow-green birefringence on staining with Congo Red (Fig. 1B), another trait typical of amyloid.
X-ray fiber diffraction of the C. albicans Ure2p1–90 fibers shows an intense sharp band at 4.7 Å and a broader band centered at 9.4 Å (Fig. 1C, D). This pattern is diagnostic of β-sheet structure with the 4.7 Å peak due to reflections between adjacent β-strands and the 9.4 Å peak reflects either the distance between neighboring β-sheets, or possibly the distance between alternate beta strands (2 × 4.7 Å).
In contrast to the rapid formation of amyloid filaments by the C. albicans Ure2 N-terminal domain, that of C. glabrata only slowly and partially did so, in spite of many attempts under a variety of buffer conditions. After about 2 years of incubation at 4C, filaments of C. glabrata Ure2p1–100 were observed, but the bulk of the protein remained soluble. Even seeding with a sonicated portion of these filaments did not promote filament formation after incubation for 2 more months in Tris or potassium phosphate buffers. X-ray fiber diffraction of the fibers formed showed a typical β-sheet pattern and the fibers showed birefringence on staining with Congo Red (not shown). Full length C. glabrata Ure2p did not form detectable amyloid under the conditions used.
Ure2p is a negative regulator of nitrogen catabolism, inhibiting the transcription of genes for transporters and enzymes needed for the utilization of poor nitrogen sources when a good nitrogen source is available (reviewed in (49, 50)). Transcription of DAL5, encoding the alantoate transporter, is particularly strongly repressed by Ure2p action (51). By replacing the ADE2 promoter with that of DAL5, we can use adenine auxotrophy (and red colony color on adenine-limiting media) as a sign of Ure2p activity (27, 52). Thus, [URE3] clones are Ade+ and white because of Ure2p inactivity, while [ure-o] clones are Ade- and red.
Various sonicated filament preparations were introduced (27) into S. cerevisiae strains in which the URE2 gene had been replaced by that of C. albicans (BY302) or of C. glabrata (BY304) (40). Cells that have taken up a DNA plasmid (pRS425) are selected as Leu+ colonies and these are tested on -Ade plates. Those that are Ade+ (possible [URE3] transfectants) are grown on 3 mM guanidine to test for curing to confirm that the Ade+ is due to prion infection (Table 2). Guanidine inhibits Hsp104 (53), a disaggregating chaperone required for propagation of [URE3alb] (40) as it is for other yeast prions. We found that a substantial portion of Leu+ clones had acquired [URE3alb] after infection with either filaments of C.a.Ure2p1–90 or with an extract of the [URE3alb] strain YHE1161 (Table 2). Infection required that the cell's URE2 gene encode the same Ure2p as the infecting material, so S.c.Ure2p1–89 filaments and C.g.Ure2p1–100 filaments were not infectious for cells carrying the C. albicans URE2(Table 2).
This infectivity of C.a.Ure2p1–90 filaments shows that Ure2p1–90 includes the prion domain and is consistent with the ability of overexpression of this region to induce [URE3alb] formation in vivo (40). Restreaking Ade+ [URE3alb] clones on adenine-limiting medium showed that the transfectants harbored an array of prion variants, differing in their stability and in the intensity of the Ade+ phenotype (Fig. 2), similar to results observed for S. cerevisiae Ure2p filaments (27).
The infectivity of the C.a.Ure2p1–90 filaments implies that the structural studies we report here are relevant to the prion phenomenon. One dimensional NMR spectra of dry fibrils of C.a.Ure2p1–90 labeled with Val-1-13C, Ile-1-13C, Ala-3-13C or unlabeled (natural abundance) are shown in Fig. 3. Note that we have shown that the infectivity of amyloid fibrils of Ure2pcerevisiae and Sup35NM are unaffected by drying (54). The Ile-1-13C - labeled amyloid showed a major peak (85%) at 172.6 ppm, whereas the chemical shift expected for random coil structure is 174.7 ppm (55). For Val-1-13C - labeled C.a.Ure2p1–90amyloid the major peak (68%) was at 171.9 ppm while the random coil value is 174.6 ppm. In general, carbonyl carbon resonance frequencies are shifted to lower values for residues in β-sheet structure, and shifted up for a-helical residues (55). These results indicate that most Ile and Val residues are in β-sheet structure. The X-ray fiber diffraction studies (above) indicate that β-sheet structure is more extensive than the few residues that could be examined individually by solid-state NMR.
In an amyloid with parallel in-register β-sheet structure, each residue is aligned in a row of identical residues extending along the long axis of the fibril. Therefore, if a particular amino acid, say isoleucine, is fully labeled specifically in the carbonyl carbon (Ile-1-13C), then the distance from each 13C label to the nearest neighbor 13C will be the 4.7 Å distance between main chains in the β sheet(56, 57) reviewed by (58). In other forms of β-sheet -antiparallel, β-helix or parallel out-of-register - the distance is far greater than 4.7 Å. This distance is measured by a dipolar recoupling experiment, such as the PITHIRDS-CT method used here (46). In this experiment, the rate of signal decay is roughly proportional to the inverse of the cube of the distance from one labeled atom to its nearest labeled neighbor.
An amyloid sample of C.a.Ure2p1–90 labeled with Ile-1-13C showed signal decay reflecting a nearest neighbor distance of about 5 Å indicating a parallel in-register structure (Fig. 4A). About 1.1% of all carbon is 13C, so that the same experiment can be done using unlabeled amyloid. Of course, in this case, the nearest neighbor 13C will nearly always be quite distant, and accordingly, a very slow signal decay was observed (Fig. 4A). Val-1-13C - labeled C.a.Ure2p1–90 amyloid showed an apparent biphasic signal decay curve, suggesting that some Val residues are in an in-register parallel structure and others are not (Fig. 4A). This result is consistent the 1D NMR spectrum of this material which indicates ~2/3 of Val residues are in β-sheet structure and the remainder are not (Fig. 3).
In another experiment, we labeled the methyl groups of the two Ala residues of C.a.Ure2p1–90 and formed amyloid in vitro. For a parallel in-register β-sheet, a nearest-neighbor distance of ~5 Å is expected. Since adjacent side chains in a β-strand point in opposite directions, if the parallel β-sheet is out of register by even a single residue, the nearest neighbor distance for this side-chain label will be > 8 Å. The rapid signal decay seen in a PITHIRDS-CT experiment with Ala-3-13C C.a. Ure2p1–90 (Fig. 4B) is again indicative of an in-register parallel structure.
A two dimensional 13C-13C solid state NMR experiment using uniformly 13C-labeled rehydrated C.a. Ure2p1–90 amyloid showed broad peaks which could only be assigned amino acid type but not to individual residues (Fig. 5). This result indicates heterogeneity of structure, consistent with the heterogeneity of prion variant observed on transfection of yeast cells (Fig. 2).
Our previous work showed that C.a.Ure2p can form a prion in S. cerevisiae, but that C.g.Ure2p cannot (40). Here we find that C.a.Ure2p1–90 readily forms amyloid, but that C.g.Ure2p1–100 only forms a small amount of amyloid over a period of two years, providing a possible explanation for the inability of C.g.Ure2p to be a prion. Of course, it is possible that there exists some other buffer condition under which C.g.Ure2p would readily form amyloid. Whether amyloid forming ability (or lack of it) will be a general explanation of whether or not a protein can be a prion will require examination of a wide range of species.
The C.a.Ure2p1–90 amyloid formed is infectious for S. cerevisiae expressing C.a.Ure2p in place of the S.c.Ure2p, but we were not successful in inducing C.g.Ure2p to form [URE3] by introduction of the small amount of available C.g.Ure2p1–100 amyloid. Of course it remains possible that with larger amounts of C. glabrata amyloids, or different conditions of amyloid formation, infection might be observed. The infection produces a variety of [URE3alb] variants with varying degrees of strong or weak phenotype and stable or unstable inheritance. This array of prion variants presumably reflects an array of amyloid structures, and the wide peaks we observe in 2D solid state NMR experiments (Fig. 5) are likely a reflection of such structural heterogeneity.
Our structural studies indicate that the amyloid fibers formed in vitro by C.a.Ure2p1–90 have an architecture like that of the other amyloid-based yeast prions, namely, an in-register parallel β-sheet. The 6 nm diameter of the filaments of C.a.Ure2p1–90 indicates that the sheet is multiply folded along its long axis, since a fully extended monomer would be approximately 30 nm in length. More detailed structural studies will be needed to ascertain the location of the folds and other details of the structure, but this will require development of a method to prepare structurally homogeneous filaments.
While mammalian prions are uniformly fatal, those of yeast and fungi are not, so that whether prions are diseases or of some adaptive value must be judged by other evidence. The conservation of prion formation among closely related species would not prove that it has a function for the host (any more than the conservation of occasional broken wings among birds suggests a benefit). However, the failure of conservation among close relatives would argue against a functional role for prion formation. Residues 10–39 of the S. cerevisiae Ure2p prion domain is conserved among a range of yeasts (10, 11, 59), and this led Harrison et al. to propose that this region is conserved to enable prion formation (59). Ross et al. demonstrated that sequence is of little importance for prion formation by Ure2p or by Sup35p (35, 36), so the conserved part of the Ure2p prion domain is not likely conserved for prion formation: a need for prion formation would not result in conservation of the sequence. Moreover, C. glabrata is closely related to S. cerevisiae and C.g.Ure2p has the conserved prion domain sequence (Fig. S1), but does not form prions. In contrast, C. albicans is more distantly related to S. cerevisiae and its Ure2p lacks the conserved sequence (Fig. S1), but we find that it readily forms prions. This suggests that the sequence is conserved for some other reason, perhaps for the stabilization of the full length protein by the prion domain as demonstrated for S. cerevisiae (60).
While definitive data is scarce, the pattern seems to be that prion-forming ability of a given protein is scattered, rather than concentrated among close relatives. For example, the full length S. castellii Ure2p cannot form a prion in S. cerevisiae (13), but the Ure2p's of several other Saccharomyces species can form [URE3]. One fourth of wild S. cerevisiae strains tested had a large deletion in their Sup35 prion domains making them unable to form the [PSI+] prion (61). Several N-terminal domains of Sup35 of other species fused to the S. cerevisiae Sup35C have been found to be able to act as prion domains (16–18). However, it is known that for Ure2p, Sup35p and HET-s, the prion domain is substantially stabilized by the remainder of the molecule (8, 19, 20), so that the context of the full length molecule is important.
The absence of the [URE3] and [PSI+] prions in wild strains, the inability of some species closely related to S. cerevisiae to form these prions, the fact that the 'prion domains' have clear-cut non-prion functions and the rapid evolution of the prion domains producing species barriers all indicate that these are yeast diseases (reviewed in (39). The cells themselves signal their displeasure at infection with [URE3] and [PSI+] by undergoing the stress-response, inducing at least Hsp70s and Hsp104 (62, 63). Moreover, the existence of lethal and very pathological variants of [PSI+] and [URE3] further argue against an adaptive role for these prions (64). The study of these yeast prion diseases is making important contributions to our understanding of mammalian prion diseases and amyloid diseases in general.
The authors thank Rob Tycko for help with solid-state NMR.
†This work was supported by the Intramural Program of the National Institute of Diabetes Digestive and Kidney Diseases of the Natioinal Institutes of Health.
Supporting Information Available. An alignment of the prion domains of Ure2p from S. cerevisiae, C. albicans and C. glabrata is shown in Fig. S1. This material is available free of charge via the Internet at http://pubs.acs.org.