Molecular Population Genetics and Evolution of a Prion-like Protein in Saccharomyces cerevisiae

Corresponding author: Mark A. Jensen, Department of Microbiology, University of Washington, Seattle, WA 98195-8070., majensen{at}u.washington.edu (E-mail)

Communicating editor: W. STEPHAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The prion-like behavior of Sup35p, the eRF3 homolog in the yeast Saccharomyces cerevisiae, mediates the activity of the cytoplasmic nonsense suppressor known as [PSI⁺]. Sup35p is divided into three regions of distinct function. The N-terminal and middle (M) regions are required for the induction and propagation of [PSI⁺] but are not necessary for translation termination or cell viability. The C-terminal region encompasses the termination function. The existence of the N-terminal region in SUP35 homologs of other fungi has led some to suggest that this region has an adaptive function separate from translation termination. To examine this hypothesis, we sequenced portions of SUP35 in 21 strains of S. cerevisiae, including 13 clinical isolates. We analyzed nucleotide polymorphism within this species and compared it to sequence divergence from a sister species, S. paradoxus. The N domain of Sup35p is highly conserved in amino acid sequence and is highly biased in codon usage toward preferred codons. Amino acid changes are under weak purifying selection based on a quantitative analysis of polymorphism and divergence. We also conclude that the clinical strains of S. cerevisiae are not recently derived and that outcrossing between strains in S. cerevisiae may be relatively rare in nature.

PRIONS are infectious agents responsible for a wide variety of debilitating diseases in humans and other mammals (HORWICH and WEISSMAN 1997 ; PRUSINER et al. 1998 ). While the dynamics of prion transmission are similar to that of infectious microorganisms, prions appear to consist solely of protein, having no associated nucleic acid to mediate replication. In the mammalian case, the prevailing hypothesis suggests that a prion is an ordinary host protein, PrP^C, that has assumed an alternative stable pathogenic structure, called PrP^Sc. PrP^Sc interacts with PrP^C, inducing PrP^C to change to PrP^Sc conformation. A cascading process of recruitment and conversion converts most of the cell's PrP^C protein to the pathogenic form, producing disease and new infectious material.

The discovery of prion-like phenomena in the yeast Saccharomyces cerevisiae has led to a new model system for prion study (TUITE and LINDQUIST 1996 ; LIEBMAN and DERKATCH 1999 ; WICKNER and CHERNOFF 1999 ; TRUE and LINDQUIST 2000 ). More importantly, it is representative of a novel mechanism for the inheritance of genetic information. A change in phenotype can be passed between generations via a self-perpetuating change in the conformational state of a protein, with no underlying changes in genotype. A series of investigations (TER-AVANESYAN et al. 1994 ; CHERNOFF et al. 1993 , CHERNOFF et al. 1995 ; DERKATCH et al. 1996 ; PATINO et al. 1996 ; GLOVER et al. 1997 ; PAUSHKIN et al. 1997 ) confirmed the model (WICKNER 1994 ) that the cytoplasmic nonsense-suppression factor [PSI⁺] (COX 1965 ) is prion-like, consisting of the protein product of the gene SUP35. Sup35p is the yeast homolog of the translation termination factor eRF3. Sup35p in its prion-like conformation forms intracellular aggregates. This sequesters Sup35p, leading to reduced efficiency of polypeptide release and readthrough of all three nonsense codons. Yeast strains exhibiting this prion-mediated nonsense suppression are called [PSI⁺]. [PSI⁺] is a metastable phenomenon, which spontaneously arises and disappears at a rate less than once in 10⁶ generations (Y. O. CHERNOFF, unpublished data).

The potentially deleterious effects of low-level translation termination disruption suggest that [PSI⁺] could be a disease of yeast, by analogy with the mammalian case. However, several authors (CHERNOFF et al. 1992 ; TUITE and LINDQUIST 1996 ; TRUE and LINDQUIST 2000 ) have suggested that the ability to switch to the [PSI⁺] state may have adaptive value in S. cerevisiae. Their arguments depend in part upon the structure of Sup35p. Sup35p can be divided into three regions, N-terminal, M, and C-terminal (Fig 1 and Fig 3). The N domain is extremely rich in glycine, glutamine, asparagine, and tyrosine, while the M domain harbors an unusual concentration of both positively and negatively charged amino acids. The C region is highly conserved among disparate taxa (KUSHNIROV et al. 1990 ; GAGNY and SILAR 1998 ; WICKNER and CHERNOFF 1999 ) and can function alone in the cell to allow wild-type translation termination efficiency (TER-AVANESYAN et al. 1993 ). When the NM or N regions are deleted, however, strains normally able to propagate [PSI⁺] become unable to do so and are referred to as "[PSI⁺] no more" (PNM; TER-AVANESYAN et al. 1994 ). On the other hand, the formation of the [PSI⁺] prion can be induced by transiently overexpressing just the NM or N region on a plasmid (DERKATCH et al. 1996 ). Thus, the prion-forming function associated with the NM domain appears to be separable from the translation termination function associated with the C domain. The persistence of the NM region in S. cerevisiae in spite of potential deleterious effects on polypeptide termination, the presence in other yeasts and fungi of a flexible glutamine-rich region in the N terminus of their eRF3 homologs (KUSHNIROV et al. 1990 ; GAGNY and SILAR 1998 ), and the fact that the ability to acquire and propagate [PSI⁺] is dependent upon the presence of NM and an appropriate balance of the chaperones hsp104 and hsp70 (CHERNOFF et al. 1995 , CHERNOFF et al. 1999 ; NEWNAM et al. 1999 ; and S. LINDQUIST and J.-J. LIU, unpublished results) make it tempting to speculate that [PSI⁺] itself could serve an adaptive purpose.

View larger version (7K): In this window In a new window Download PPT slide   Figure 1. Structure of SUP35 coding region. Arrows indicate position and orientation of sequencing primers. Hash mark indicates start codon. Coordinates correspond to GenBank accession no. M21129. Primer sequences (5'–3'): 725U, GCCCACTAGCAACAATGTCGG; 741, GTCGGATTCAAACCAAGGCA; 1144U, GCCGCTCCCAAACCAAAGAAG; 1538L, ATTTACCGGCATCAACATGAC; 1550U, CCGGTAAATCTACTATGGGTG; 1143L, ATACCGGAACTGGAGACAAGC; 382Ls, CTGCTTTTGTTGCTTTTGAAAGTC; REV-1551, GGCATCAACATGACCCATGA; 2064L, CCGGATACTGGCATAAATACA.

View larger version (44K): In this window In a new window Download PPT slide   Figure 2. SUP35 polymorphic sites. Strains received from (a) S. Liebman, (b) C. Holm, and (c) J. McCusker (and YJM strains). (d) Site variant is a nonsynonymous or synonymous change. (e) Location of first sampling, if known. (f) SUP35 region.

View larger version (45K): In this window In a new window Download PPT slide   Figure 3. Amino acid comparison of S. cerevisiae and S. paradoxus. (a) n, m, and c denote MET codons delimiting the N, M, and C domains. (b) For polymorphic amino acids, the rare amino acid is given in this line.

As a first step toward understanding the adaptive and other evolutionary properties of SUP35, we examined nucleotide polymorphism and divergence within a DNA fragment spanning the N and M domains among 23 S. cerevisiae alleles and a longer fragment containing N, M, and a portion of C for a set of 10 alleles (Fig 2). Clinical S. cerevisiae strains isolated from different immunocompromised patients worldwide (MCCUSKER et al. 1994 ; CLEMONS et al. 1997 ) provided 13 of the NM alleles. Lab strains of different origins, baking strains, and brewery strains were also included in this analysis. We compared S. cerevisiae sequence to homologous sequence from the sibling species S. paradoxus (MONTROCHER et al. 1998 ) to examine between-species divergence. The sequence of this species can be aligned base-for-base with the entire S. cerevisiae sequence. Other SUP35 divergence studies (e.g., KUSHNIROV et al. 1990 ) have used yeasts and other taxa too distantly related to allow an unambiguous alignment of the prion-associated NM region.

Our analysis provides evidence that the prion-determining domain of Sup35p, if not an adaptation, is at worst only weakly deleterious. We find that the amino acid sequence of all three regions is constrained, presumably by purifying selection against nonsynonymous nucleotide variants. SUP35 is also biased in codon usage toward preferred codons, suggesting that selection can influence the frequency of weakly deleterious variants in this gene. It is possible that this purifying selection has relaxed in the NM region in the S. cerevisiae lineage. Purifying selection in the C region appears to be ongoing. Using recombination estimates based on linkage disequilibrium, we calculate a crude lower bound for the rate of outcrossing between clonal S. cerevisiae lineages.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
We sequenced alleles from 13 clinical isolates, seven laboratory strains, two commercial strains, and one wild vineyard strain of S. cerevisiae and one allele from a wild-caught S. paradoxus isolate. Details of the provenance of the strains are in Fig 2. We also included the GenBank sequence of KUSHNIROV et al. 1988 in our analyses.

Most of the NM sequences were derived from SUP35 genes cloned during the course of ongoing studies of phenotypic variation in yeast strains carrying [PSI⁺] in the Lindquist lab. Sequences that include a portion of the C region were obtained as part of a preliminary study of polymorphism in a subset of these strains; industrial strain sequences and part of the S. paradoxus sequence were provided by the Chernoff lab.

The sample, although not intentionally assembled for a population genetics study, nevertheless represents a random worldwide sample of S. cerevisiae and as such provides a reasonable first approximation of species-wide nucleotide variation at this locus.

Gene regions:
For 23 alleles, we analyzed a 754-bp region spanning the N and M region of SUP35 [corresponding to coordinates 744–1497 of GenBank accession no. M21129 (KUSHNIROV et al. 1988 )]. For 10 of these, we also analyzed a region that extended 534 bp into the C domain (coordinates 1501 to 2034 of accession no. M21129).

DNA sequencing:
Strains were sequenced by one of two methods. For sequencing performed at Chicago, the desired region was amplified from genomic template using PCR and cloned into either pZERO-2.1 (Invitrogen, San Diego) or pmCNMG (Lindquist lab yeast vector, sequence available on request). Two or three clones were sequenced in both directions, with the consensus base among clones accepted. Sequences were run by the Genotyping and Sequencing Center of the Department of Ecology and Evolution at the University of Chicago. For sequencing performed at Georgia Tech, genomic PCR product was used directly in a cycle-sequencing reaction. Sequences were run by the Molecular Genetics Instrumentation Facility of the University System of Georgia, Athens. The differences reflect differences in standard laboratory protocol; no particular difficulty in sequencing any strain was encountered by either group.

For nine of the longer fragments, we obtained sequence by first amplifying genomic DNA using primers SUP35C725U and SUP35C2064L (see Fig 1 legend for primer sequences). PCR products were cloned into a derivative of plasmid pZERO-2.1 (Invitrogen). Clones were sequenced in both directions, using universal M13 primers M13-20 and M13rev and internal primers SUP35C1143L, SUP35C1144U, SUP35C1538L, and SUP35C1550U.

For the strains Red Star, Bass Ale, and SK-1, genomic DNA was amplified with primers SUP35C725U and SUP35C2064L and was gel purified and sequenced directly, using these primers plus internal primers as above.

The NM regions of the remaining S. cerevisiae alleles were cloned into Lindquist lab yeast vector pmCNMG at BamHI/SacII sites from genomic PCR products. Plasmid DNA was sequenced directly in both directions using 5CNMSEQ (5'-GTGCAATATCATATAGAAGTCATCG-3'), 3CNMSEQ (5'-CAACAAGAATTGGGACAACTCCAG-3'), and internal primers SUP35C1144U and SUP35-382Ls. 5CNMSEQ and 3CNMSEQ were designed from vector sequence, with 3' ends lying 51 and 26 bp from the ends of the insert, respectively.

The NM region of S. paradoxus was amplified with primers SUP35-741 and SUP35REV-1551, gel purified, and sequenced directly by using the same primers and internal primers SUP35C1143L and SUP35-741. To amplify the C region of S. paradoxus, we performed a first round amplification on genomic DNA with primers SUP35C725U and SUP35C2064L. This product was then amplified in second round reactions containing SUP35C2064L and either SUP35C725U or SUP35C1144U. Both long and short products were cloned and sequenced using universal primers.

Sequences were managed with Sequencher 3.0 software (Gene Codes, Ann Arbor, MI). S. cerevisiae sequences have been deposited in GenBank under accession nos. AY052599, AY052600, AY052601, AY052602, AY052603, AY052604, AY052605, AY052606, AY052607, AY052608, AY052609, AY052610, AY052611 and AY052613, AY052614, AY052615, AY052616, AY052617, AY052618, AY052619, AY052620, AY052621. S. paradoxus sequence is at accession no. AY052612.

Polymorphism and divergence analysis:
Sequence alignments were performed using Sequencher and MegAlign (DNASTAR, Madison, WI). Nucleotide polymorphism and divergence were analyzed using programs DNAsp 3.0 (ROZAS and ROZAS 1999 ) and SITES 1.1 (WAKELEY and HEY 1997 ). Simulated distributions for Tajima's test were generated using a web-based program of Y.-X. Fu (http://www.hgc.sph.uth.tmc.edu/fu/; FU 1997 ). Recombination was estimated using the method of HUDSON 1987 as implemented in SITES. Estimates of the average selection against replacement substitutions were obtained using the method of SAWYER and HARTL 1992 , as extended by AKASHI 1995 , using a C program (written by M. A. Jensen). SAWYER's (1989) gene conversion test was performed using a C program (written by M. A. Jensen). Source code and other details are available upon request.

Protein constraint and codon bias analysis:
Simulation estimates of the substitution rates K_a, K_s, and their distributions were obtained using a program of COMERON 1999 . We calculated the empirical distribution of the codon adaptation index (CAI; SHARP and LI 1987 ) for all named genes >300 bp in length (obtained October 1999 from the Saccharo-myces Genome Database, http://genome-www.stanford.edu/Saccharomyces/) and compared the CAI for SUP35 to this distribution.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Polymorphism and divergence:
We did not observe statistically significant differences between regions N and M for any of our parameter estimates. We therefore combine the two, referring to the NM region in the following analysis. For the NM region of 754 bp, we found 17 segregating sites in 23 alleles. For 534 bp of the C region, we found 1 singleton segregating site in 10 alleles. Polymorphic sites are displayed in Fig 2, and amino acid sequences between S. cerevisiae and S. paradoxus are compared in Fig 3. Evolutionary parameters are summarized in Table 1. Tajima's D, a measure of the departure of the allelic frequency spectrum from neutrality (TAJIMA 1989 ), is not significant for either region.

The overall divergence in the region between S. cerevisiae and S. paradoxus is 0.10, using KIMURA's (1981) two-parameter model. The two species' sequences can be unambiguously aligned, without gaps, over the regions analyzed. Table 2 contains numbers of polymorphic sites and fixed differences for each of the regions, separated into synonymous and nonsynonymous changes. Independence tests performed on the 2 x 2 comparisons for each region constitute the McDonald-Kreitman (MK) test (MCDONALD and KREITMAN 1991 ) of evolutionary neutrality; P values (two-tailed, by Fisher's exact test) are provided in Table 2. Under neutrality, the number of polymorphisms relative to fixed differences should be the same for both synonymous and nonsynonymous sites; the MK test determines whether the data depart significantly from this hypothesis. The MK test within NM shows no significant departure from neutrality. The C region exhibits a departure significant at the 3% level, with the number of synonymous fixed differences exceeding nonsynonymous differences. The test is significant for the entire sequenced region at the 0.5% level.

Protein constraint:
We measure evolutionary constraint against amino acid change in SUP35, using the rates of nonsynonymous nucleotide substitution, K_a, and synonymous substitution, K_s. The program of COMERON 1999 uses nucleotide divergence between the sister species to calculate observed K_a and K_s for the two protein regions. It then determines confidence intervals for those rates by simulation, generating a distribution by repeatedly applying Poisson mutation processes to the sequence data. In Comeron's method, the means of the synonymous and nonsynonymous processes are the observed rates, corrected for any bias imparted by the multiple-hit correction used to obtain those rates. Mutations to stop codons are not accepted in the simulations. Simulation results for SUP35 and two other genes are given in Table 3.

Strength of selection:
To assess the strength of purifying selection against amino acid changes, we applied the method of (SAWYER and HARTL 1992 ; AKASHI 1995 ) to our polymorphism and divergence data. The Sawyer-Hartl procedure can be interpreted as a quantitative MK test that calculates the average selection on nonsynonymous changes required to give the observed departure from neutral expectation. The quantity estimated is Ns, the product of the effective population size and the average selection coefficient.

Assuming free recombination among sites in a coding region in which replacement changes are under selection with constant selection coefficient s and synonymous changes are neutral, SAWYER and HARTL 1992 use a diffusion approximation with selection to obtain an expression involving Ns for the expected ratio of the number of nonsynonymous polymorphisms to the number of nonsynonymous fixed differences between samples from two divergent but closely related species. This expression requires an estimate of the divergence time between the two species, which is obtained using the ratio of synonymous polymorphisms to synonymous differences. These expressions can be found conveniently in AKASHI 1995 . Estimates of Ns are obtained by equating the observed nonsynonymous polymorphism/fixed difference ratio to the expected ratio and solving numerically for Ns. To obtain confidence intervals, following AKASHI 1995 , we generate a distribution of Ns values by drawing Poisson random variates for each class of site (nonsynonymous polymorphic, nonsynonymous fixed, synonymous polymorphic, and synonymous fixed), from a distribution with mean equal to the observed number in that class, and calculate Ns as above. Source code in C written to perform this analysis is available upon request.

Table 2 contains the median estimates and 95% confidence intervals for Ns, the product of the effective population size and the average selection coefficient. Limits of confidence intervals are the Ns values at the 2.5th and 97.5th percentiles of the simulated distribution. The average amino acid replacement is under weak purifying selection across the two regions, with stronger selection against changes evident in the C-terminal region.

Codon bias:
The level of codon bias complements protein sequence constraint as a measure of the long-term importance of a gene to organism fitness. Amino acid conservation implies functional importance of the particular protein secondary structure. If the choice of codons encoding those constrained amino acids is highly biased, then selection for translational accuracy and efficiency is great enough to significantly constrain the nucleotide sequence of the gene at synonymous sites (see SHARP et al. 1995 for review). This provides further support for long-term functional importance of the gene under study.

We calculated the empirical distribution of the CAI (SHARP and LI 1987 ) for all S. cerevisiae named genes of >300 bp in length, as in MORIYAMA and POWELL 1998 . This is a set of 3106 open reading frames (October 1999), obtained from the Saccharomyces Genome Database (http://genome-www.stanford.edu/Saccharomyces/). The CAI measures the extent to which codon usage in a particular gene is skewed toward use of the most frequently occurring codon for each amino acid in a set of highly expressed genes. It takes values between 0.0 and 1.0; a gene with a CAI of 1.0 always uses the most frequent codon to encode each amino acid. Our calculation gives a mean CAI of 0.213, comparable to 0.22 obtained for 2169 genes by MORIYAMA and POWELL 1998 . CAI values are as follows: SUP35 (entire gene), 0.333 (88.8th percentile); NM alone, 0.262 (83.7th percentile); C alone, 0.372 (90.3rd percentile). On the basis of the distribution of simulated coding sequences whose amino acid composition and GC content were equivalent to that of the yeast genome, COGHLAN and WOLFE 2000 suggest a gene may be considered biased if its CAI value is >0.141, two standard deviations above the mean of their simulated CAI distribution. SUP35 clearly meets this criterion. These values are consistent with the high level of expression of SUP35, which is in the 93rd percentile of genes, as determined by microarray hybridization (WODICKA et al. 1997 ).

Estimates of genetic exchange:
Polymorphism data allow us to make inferences concerning the rates of recombination and gene conversion, scaled to the population's effective size (HUDSON and KAPLAN 1985 ; HUDSON 1987 ; HEY and WAKELEY 1997 ). The data for NM show evidence of recombination, with the four-gamete test of Hudson and Kaplan revealing three apparent recombination events. HUDSON's (1987) estimator of the scaled recombination rate C = 4NR, where R is the recombination fraction and N is the effective population size, is 11.25, or 8.7 x 10^-3 per base. The value of C/_SS is 1.24; this is comparable to that found in Drosophila (median for 24 Drosophila melanogaster loci, 1.5; ANDOLFATTO and PRZEWORSKI 2000 ) and humans (1.3; HEY and WAKELEY 1997 ). This value is much lower than expected given the map length in this region and suggests an excess of linkage disequilibrium among variants (see DISCUSSION). SAWYER's (1989) runs test for gene conversion is not significant (P = 0.45; number of permutations, 10,000).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Amino acid constraint in Sup35p:
We can draw several conclusions of interest. First, the termination function is clearly very highly constrained. There are no nonsynonymous changes between species among the 178 codons of this region that we have analyzed. This is in spite of the fact that the synonymous rates of the two regions are the same; the distributions of K_s for the two regions overlap almost completely (Table 3). We can therefore assume that the intrinsic mutation rates for the two regions are the same. Although we consider only a portion of the termination function, the lack of divergence is still significant. Consider the following comparison with the NM region. Out of 534 sites, 415 are nonsynonymous in the C region (Table 2). Using the lowest rate of nonsynonymous substitution for the NM region, we expect to see at least (415)(0.019) = 7 nonsynonymous substitutions in the portion of the C region we examine, were the two regions evolving at the same rate. This suggests that the two regions are under different selective regimes and complements the functional analysis that distinguishes the two regions.

The NM region clearly has been under constraint, though it is more labile than the C region. We see this in two ways. First, K_a is less than K_s, and the distributions do not overlap (Table 3). If K_s is taken as an estimate of the neutral mutation rate, and selection is assumed to be the agent that causes substitution rates to vary, significantly slower replacement site evolution is evidence of purifying selection against replacement changes or "selective constraint" on the NM amino acid sequence (KIMURA 1983 , p. 44).

Note that this conclusion is conservative. SUP35 is highly codon biased, so that synonymous changes are also constrained; the actual neutral rate is likely to be higher than K_s. In other words, if synonymous changes are under purifying selection, then nonsynonymous changes must be as well, according to these results.

Second, we can compare the evolutionary rates at NM with rates at other loci. Ideally, we would compare the K_a of NM with the K_a values of a large set of genes having similar rates of synonymous substitution. We could then ask whether NM is relatively highly constrained, with respect not to a neutral model but to the average expressed coding sequence. At this time, only two S. paradoxus coding sequences encoding nonduplicated nuclear genes are currently available for such a comparison. The genes REC104 and REC114 are so-called "early exchange" genes, required for meiotic recombination and accurate meiotic division (JIAO et al. 1999 ), and their products are 180 amino acids (aa) and 428 aa long, respectively. REC104 and REC114 are under strong constraint between the species, as their respective K_a and K_s distributions are distinct at the 0.1% confidence level (Table 3). While the K_s 95% confidence interval for NM overlaps that of both genes, K_a for NM is less than K_a for the two other genes, its distribution is distinct from the others at the 5% confidence level, and the K_a/K_s ratio of NM is less than that of either gene. The NM region therefore appears to be at least as strongly constrained in amino acid sequence as these important meiotic genes.

Evolutionary conservation of the prion-determining domain:
Prions are proteins that can switch to different stable conformations that alter their function in a self-perpetuating way. If the protein in the prion state disrupts the function of the protein in wild-type conformation, as in the case of [PSI⁺], the presence of the prion may be deleterious. In such a case, one may expect natural selection to lead to elimination of the capacity of the protein to adopt the alternate conformation, if constraints on the wild-type function of the protein permit this. In this light, the results of this study are somewhat counterintuitive. In spite of its ability to foster prion formation and its apparent dispensability with respect to the termination function, the prion-determining NM region of Sup35p is significantly conserved in amino acid sequence. In addition, SUP35 is biased in codon usage across its length, which is evidence that translational accuracy and efficiency are important for both regions.

The biological function of the NM region is unknown. Deletion of the homologous region of eRF3 in the fungus Podospora anserina reduces reproductive efficiency (GAGNY and SILAR 1998 ), while, in yeast, the NM region is involved in interactions with a variety of other yeast proteins, including a cytoskeletal assembly component (BAILLEUL et al. 1999 ). Whatever function NM may have, loss of this function is evidently more deleterious than deleterious effects caused by a heritable reduction of termination efficiency upon periodic manifestation of [PSI⁺].

How deleterious is the prion state in yeast? If a prion is sufficiently deleterious, there will be selective pressure to eliminate it while preserving other functions of the prion-determining domain, if this is possible. Our study suggests that selection against the prion state need not be very strong to affect the sequence of SUP35. Codon bias results show that synonymous changes away from preferred codons have been eliminated over time. Such changes are under very weak selection (AKASHI 1995 ). Thus, the sequence of SUP35 is sensitive to very weak changes in selective pressures; that is, its effective population size is quite high.

Single replacement changes in NM have been identified that prevent [PSI⁺] propagation (PNM mutations) or inhibit nonsense suppression [anti-suppressor (ASU) mutations; DOEL et al. 1994 ; DEPACE et al. 1998 ]. If the prion state were at least as deleterious as the average nonpreferred synonymous change, such mutations would confer an advantage that might cause at least one of them to reach high frequency. Nevertheless, we do not observe any of these mutations in any of our strains. This may be so because the prion state is only weakly deleterious. However, it is possible that in some of these strains the genetic background cannot support [PSI⁺] or that some of the replacement variants observed in this study are PNM or ASU mutations not detected in DEPACE et al.. 1998 comprehensive screen. It is also possible that these replacements cause deleterious pleiotropic effects on NM's primary function. Since NM can be deleted without affecting viability, and the average replacement change is only weakly deleterious, such a pleiotropic effect is probably weak.

Mode of selection on SUP35:
The significant McDonald-Kreitman test over the entire fragment indicates a deficiency of amino acid replacements relative to polymorphism. That is, weak purifying selection against replacement changes appears to be acting in the region (see, e.g., the discussion in NACHMAN 1998 ). The Sawyer-Hartl analysis indicates that Ns against the average replacement change is very likely negative but is relatively small in magnitude, on the order of unity. The distribution of polymorphism within NM is consistent with neutrality, according to the McDonald-Kreitman and Tajima's tests, while Ns for changes specifically in NM tends toward purifying selection but is not significantly different from zero. It is possible that a relaxation of purifying selection at NM has taken place in S. cerevisiae. Note that this result is not in conflict with our primary result that purifying selection has generally conserved amino acid and codon usage over the period of divergence of the two species. The Sawyer-Hartl analysis is based on polymorphism within the species. This means that it necessarily considers relatively recent evolution of gene frequencies. It also means that sites that are rarely polymorphic, e.g., those sites at which mutation leads to lethality, are not considered, though selection is clearly operating at such sites. That is, the Sawyer-Hartl results are limited to sites that are not subject to strong selection.

Origin of clinical isolates:
The neutral distribution of polymorphism in clinical isolates (see Table 1) suggests that these are not highly "clonal," confirming a study of genomic restriction fragment length polymorphism variation in these strains (CLEMONS et al. 1997 ). That is, these virulent strains do not appear to be recently descended from a strain that spontaneously acquired virulence determinants. One caveat to this conclusion is that strong clonality (i.e., a low number of haplotypes for a given number of segregating sites) might be expected only in regions rather closely linked to virulence genes, and there is no reason to imagine that SUP35 would be linked to such genes. The data suggest, however, that recombination events that would break up long-range associations are infrequent; see discussion below.

Frequency of outcrossing:
We used HUDSON's (1987) method to estimate the scaled recombination rate C = 4NR in the NM region, where R is distance in centimorgans. Since the scaled neutral mutation rate equals 4NU, where U is the mutation rate for the region, the estimated value of C/_SS would closely estimate the biological parameter c/u in a panmictic sexual population, where c is the coefficient of exchange, or R divided by the number of bases, and u is the mutation rate per base pair per generation.

Taking c for the SUP35 region to be 3.1 x 10^-6 [estimated by comparing physical and genetic maps of chromosome IV (CHERRY et al. 1997 )] and u = 2.2 x 10^-10 (the estimated rate of spontaneous mutation per base pair per replication in S. cerevisiae; DRAKE et al. 1998 ), c/u for this region is 1.41 x 10⁴, over 4 orders of magnitude greater than our estimate of C/_SS, 1.24. This discrepancy indicates an excess of linkage disequilibrium among variants compared to a neutrally evolving panmictic population (ANDOLFATTO and PRZEWORSKI 2000 ).

Several processes are known to increase linkage disequilibrium for a given rate of recombination, including population subdivision (see, e.g., STROBECK 1987 ), inbreeding (NORDBORG 2000 ), and various types of selection, such as transient directional selection (HUDSON et al. 1997 ) or epistatic selection (OHTA 1982 ). In organisms like Saccharomyces for which sexual reproduction is facultative and infrequent, clonal reproduction will increase linkage disequilibrium as well.

If we make the simplifying assumption that population subdivision, inbreeding, and asexual reproduction are the principal causes of disequilibrium, we can make a crude estimate of the rate at which genetically divergent partially geographically isolated S. cerevisiae clones interbreed. Supposing that such outcrossing occurs once every m generations (meiosis or mitosis), we must replace c in the above ratio with c/m. Then C/_SS estimates c/mu; that is, the frequency of outcrossing is = = 8.79 x 10^-5/generation. Outcrossing in nature, therefore, may be rare. Ignoring selective forces that increase disequilibrium, however, makes this rough calculation a lower bound.

	ACKNOWLEDGMENTS

We are grateful to S. M. Belenkiy for help in some of the experiments. We thank C. Holm, J. H. McCusker, and S. W. Liebman for generous gifts of yeast strains, and T. Serio for the pmCNMG construct. We also thank P. Andolfatto, J. M. Comeron, M. Kreitman, M. Long, P. Sniegowski, and G. Wyckoff for insightful discussions and comments; B. Charlesworth and R. R. Hudson for their thorough critiques of the manuscript; and K. H. Jensen for support. This work was supported in part by a National Science Foundation Doctoral Dissertation Improvement Grant to M.A.J. and grants R21GM55091 and R01GM58763 from the National Institute of General Medical Sciences to Y.O.C. H.L.T. is a fellow of the Jane Coffin Childs Memorial Fund for Medical Research.

Manuscript received January 20, 2000; Accepted for publication August 2, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKASHI, H., 1995  Inferring weak selection from patterns of polymorphism and divergence at "silent" sites in Drosophila DNA. Genetics 139:1067-1076[Abstract].

ANDOLFATTO, P. and M. PRZEWORSKI, 2000  A genome-wide departure from the standard neutral model in natural populations of Drosophila. Genetics 156:257-268[Abstract/Free Full Text].

BAILLEUL, P. A., G. P. NEWNAM, J. N. STEENBERGEN, and Y. O. CHERNOFF, 1999  Genetic study of interactions between the cytoskeletal assembly protein Sla1 and prion-forming domain of the release factor Sup35 (eRF3) in Saccharomyces cerevisiae. Genetics 153:81-94[Abstract/Free Full Text].

CHERNOFF, Y. O., M. V. PTYUSHKINA, M. G. SAMSONOVA, G. I. SIZONENCKO, and Y. I. PAVLOV et al., 1992  Conservative system for dosage-dependent modulation of translational fidelity in eukaryotes. Biochemie 74:455-461[Medline].

CHERNOFF, Y. O., I. L. DERKACH, and S. G. INGE-VECHTOMOV, 1993  Multicopy SUP35 gene induces de novo appearance of psi-like factors in the yeast Saccharomyces cerevisiae. Curr. Genet. 24:286-270.

CHERNOFF, Y. O., S. L. LINDQUIST, B. ONO, S. G. INGE-VECHTOMOV, and S. W. LIEBMAN, 1995  Role of the chaperone protein hsp104 in propagation of the yeast prion-like factor [PSI⁺]. Science 268:880-884[Abstract/Free Full Text].

CHERNOFF, Y. O., G. P. NEWNAM, J. KUMAR, K. ALLEN, and A. D. ZINK, 1999  Evidence for a "protein mutator" in yeast: role of the Hsp70-related chaperone Ssb in formation, stability and toxicity of the [PSI] prion. Mol. Cell. Biol. 19:8103-8112[Abstract/Free Full Text].

CHERRY, J. M., C. BALL, S. WENG, G. JUVIK, and R. SCHMIDT et al., 1997  Genetic and physical maps of Saccharomyces cerevisiae. Nature 387:67-73[Medline].

CLEMONS, K. V., P.-S. PARK, J. H. MCCUSKER, M. J. MCCULLOUGH, and R. W. DAVIS et al., 1997  Application of DNA typing methods and genetic analysis to epidemiology and taxonomy of Saccharomyces isolates. J. Clin. Microbiol. 35:1822-1828[Abstract].

COGHLAN, A. and K. H. WOLFE, 2000  Relationship of codon bias to mRNA concentration and protein length in Saccharomyces cerevisiae.. Yeast 16:1131-1145[Medline].

COMERON, J. M., 1999  K-Estimator: calculation of the number of nucleotide substitutions per site and the confidence intervals. Bioinformatics 15:763-764[Abstract/Free Full Text].

COX, B. S., 1965  , a cytoplasmic suppressor of super-suppressor in yeast. Heredity 20:505-521.

DEPACE, A. H., A. SANTOSOS, P. HILLNER, and J. S. WEISSMAN, 1998  A critical role for amino-terminal glutamine/asparagine repeats in the formation and propagation of a yeast prion. Cell 93:1241-1252[Medline].

DERKATCH, I. L., Y. O. CHERNOFF, V. V. KUSHNIROV, S. G. INGE-VECHTOMOV, and S. W. LIEBMAN, 1996  Genesis and variability of [PSI] prion factors in Saccharomyces cerevisiae. Genetics 144:1375-1386[Abstract].

DOEL, S. M., S. J. MCCREADY, C. R. NIERRAS, and B. S. COX, 1994  The dominant PNM^- mutation which eliminates the psi factor of Saccharomyces cerevisiae is the result of a missense mutation in the SUP35 gene. Genetics 137:659-670[Abstract].

DRAKE, J. W., B. CHARLESWORTH, D. CHARLESWORTH, and J. CROW, 1998  Rates of spontaneous mutation. Genetics 148:1667-1686[Abstract/Free Full Text].

FU, Y.-X., 1997  Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:915-925[Abstract].

GAGNY, B. and P. SILAR, 1998  Identification of the genes encoding the cytosolic translation release factors from Podospora anserina and analysis of their role during the life cycle. Genetics 149:1763-1775[Abstract/Free Full Text].

GLOVER, J. R., A. S. KOWAL, E. C. SCHIRMER, M. M. PATINO, and J.-J. LIU et al., 1997  Self-seeded fibers formed by Sup35, the protein determinant of [PSI⁺], a heritable prion-like factor of S. cerevisiae. Cell 89:811-819[Medline].

HEY, J. and J. WAKELEY, 1997  A coalescent estimator of the population recombination rate. Genetics 145:833-846[Abstract].

HORWICH, A. L. and J. S. WEISSMAN, 1997  Deadly conformations—protein misfolding in prion disease. Cell 89:499-510[Medline].

HUDSON, R. R., 1987  Estimating the recombination parameter of a finite population model without selection. Genet. Res. 50:245-250[Medline].

HUDSON, R. R. and N. L. KAPLAN, 1985  Statistical properties of the number of recombination events in the history of a sample of DNA sequences. Genetics 111:147-164[Abstract/Free Full Text].

HUDSON, R. R., A. G. SAEZ, and F. J. AYALA, 1997  DNA variation at the Sod locus of Drosophila melanogaster: an unfolding story of natural selection. Proc. Natl. Acad. Sci. USA 94:7725-7728[Abstract/Free Full Text].

JIAO, K., S. A. BULLARD, L. SALEM, and R. E. MALONE, 1999  Coordination of the initiation of recombination and reductional division in meiosis in Saccharomyces cerevisiae. Genetics 152:117-128[Abstract/Free Full Text].

KIMURA, M., 1981  Estimation of evolutionary distances between homologous nucleotide sequences. Proc. Natl. Acad. Sci. USA 78:454-458[Abstract/Free Full Text].

KIMURA, M., 1983 The Neutral Theory of Molecular Evolution. Cambridge University Press, Cambridge, UK.

KUSHNIROV, V. V., M. D. TER-AVANESYAN, M. V. TELCKOV, A. P. SURGUCHOV, and V. N. SMIRNOV et al., 1988  Nucleotide sequence of the SUP2 (SUP35) gene of Saccharomyces cerevisiae. Gene 66:45-54[Medline].

KUSHNIROV, V. V., M. D. TER-AVANESYAN, S. A. DIDICHENKO, V. N. SMIRNOV, and Y. O. CHERNOFF et al., 1990  Divergence and conservation of SUP2 (SUP35) gene of yeasts Pichia pinus and Saccharomyces cerevisiae. Yeast 6:461-472[Medline].

LIEBMAN, S. W. and I. L. DERKATCH, 1999  The yeast [PSI⁺] prion: making sense of nonsense. J. Biol. Chem. 274:1181-1184[Free Full Text].

MCCUSKER, J. H., K. V. CLEMONS, D. A. STEVENS, and R. W. DAVIS, 1994  Genetic characterization of pathogenic Saccharomyces cerevisiae isolates. Genetics 136:1261-1269[Abstract].

MCDONALD, J. H. and M. KREITMAN, 1991  Adaptive protein evolution and the Adh locus in Drosophila. Nature 351:652-654[Medline].

MONTROCHER, R., M.-C. VERNER, J. BRIOLAY, C. GAUTIER, and R. MARMEISSE, 1998  Phylogenetic analysis of the Saccharomyces cerevisiae group based on polymorphisms of rDNA spacer sequences. Int. J. Syst. Bacteriol. 48:295-303[Abstract/Free Full Text].

MORIYAMA, E. N. and J. R. POWELL, 1998  Gene length and codon usage bias in Drosophila melanogaster, Saccharomyces cerevisiae, and Escherichia coli. Nucleic Acids Res. 26:3188-3193[Abstract/Free Full Text].

NACHMAN, M. W., 1998  Deleterious mutations in animal mitochondrial DNA. Genetica 102(103):61-69.

NEWNAM, G. P., R. D. WEGRZYN, S. L. LINDQUIST, and Y. O. CHERNOFF, 1999  Antagonistic interactions between yeast chaperones Hsp104 and Hsp70 in prion curing. Mol. Cell. Biol. 19:1325-1333[Abstract/Free Full Text].

NORDBORG, M., 2000  Linkage disequilibrium, gene trees and selfing: an ancestral recombination graph with partial self-fertilization. Genetics 154:923-929[Abstract/Free Full Text].

OHTA, T., 1982  Linkage disequilibrium due to random genetic drift in finite subdivided populations. Proc. Natl. Acad. Sci. USA 79:1940-1944[Abstract/Free Full Text].

PATINO, M. M., J.-J. LIU, J. R. GLOVER, and S. LINDQUIST, 1996  Support for the prion hypothesis for inheritance of a phenotypic trait in yeast. Science 273:622-626[Abstract].

PAUSHKIN, S. V., V. V. KUSHNIROV, V. N. SMIRNOV, and M. D. TER-AVANESYAN, 1997  In vitro propagation of the prion-like state of yeast Sup35 protein. Science 277:381-383[Abstract/Free Full Text].

PRUSINER, S. B., M. R. SCOTT, S. J. DEARMOND, and F. E. COHEN, 1998  Prion protein biology. Cell 93:337-348[Medline].

ROZAS, M. and R. ROZAS, 1999  DNASP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175[Abstract/Free Full Text].

SAWYER, S., 1989  Statistical tests for detecting gene conversion. Mol. Biol. Evol. 6:526-538[Abstract].

SAWYER, S. A. and D. L. HARTL, 1992  Population genetics of polymorphism and divergence. Genetics 132:1161-1176[Abstract].

SHARP, P. M. and W.-H. LI, 1987  The codon adaptation index: a measure of directional synonymous codon usage bias, and its potential applications. Nucleic Acids Res. 15:1281-1295[Abstract/Free Full Text].

SHARP, P. M., M. AVEROF, A. T. LLOYD, G. MATASSI, and J. PEDEN, 1995  DNA sequence evolution: the sounds of silence. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 349:241-247[Medline].

STROBECK, C., 1987  Average number of nucleotide differences in a sample from a single subpopulation: a test for population subdivision. Genetics 117:149-153[Abstract/Free Full Text].

TAJIMA, F., 1989  Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 129:753-762[Abstract].

TER-AVANESYAN, M. D., V. V. KUSHNIROV, A. R. DAGKESAMANSKAYA, S. A. DIDICHENKO, and Y. O. CHERNOFF et al., 1993  Deletion analysis of the SUP35 gene of the yeast Saccharomyces cerevisiae reveals two non-overlapping functional regions in the encoded protein. Mol. Microbiol. 7:683-692[Medline].

TER-AVANESYAN, M. D., A. R. DAGKESAMANSKAYA, V. V. KUSHNIROV, and V. N. SMIRNOV, 1994  The SUP35 omnipotent suppressor gene is involved in the maintenance of the non-Mendelian determinant [PSI⁺] in the yeast Saccharomyces cerevisiae. Genetics 137:671-676[Abstract].

TRUE, H. L. and S. L. LINDQUIST, 2000  A yeast prion provides a mechanism for genetic variation and phenotypic diversity. Nature 407:477-483[Medline].

TUITE, M. F. and S. LINDQUIST, 1996  Maintenance and inheritance of yeast prions. Trends Genet. 12:467-471[Medline].

WAKELEY, J. and J. HEY, 1997  Estimating ancestral population parameters. Genetics 145:847-855[Abstract].

WATTERSON, G. A., 1975  On the number of segregating sites in genetical models without recombination. Theor. Popul. Biol. 7:256-276[Medline].

WICKNER, R. B., 1994  [URE3] as an altered URE2 protein: evidence for a prion analog in Saccharomyces cerevisiae. Science 264:566-568[Abstract/Free Full Text].

WICKNER, R., and Y. O. CHERNOFF, 1999 Prions of fungi: [URE3], [PSI], and [Het-s] discovered as heritable traits, pp. 229–272 in Prion Biology and Diseases, edited by S. B. PRUSINER. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

WODICKA, L., H. DONG, M. MITTMAN, M. H. HO, and D. J. LOCKHART, 1997  Genome-wide expression monitoring in Saccharomyces cerevisiae. Nat. Biotechnol. 15:1359-1367[Medline].

This article has been cited by other articles:

				I. J. Tsai, D. Bensasson, A. Burt, and V. Koufopanou Population genomics of the wild yeast Saccharomyces paradoxus: Quantifying the life cycle PNAS, March 25, 2008; 105(12): 4957 - 4962. [Abstract] [Full Text] [PDF]

				B. Chen, G. P. Newnam, and Y. O. Chernoff Prion species barrier between the closely related yeast proteins is detected despite coaggregation PNAS, February 20, 2007; 104(8): 2791 - 2796. [Abstract] [Full Text] [PDF]

				N. Talarek, L. Maillet, C. Cullin, and M. Aigle The [URE3] Prion Is Not Conserved Among Saccharomyces Species Genetics, September 1, 2005; 171(1): 23 - 34. [Abstract] [Full Text] [PDF]

				J. C. Fay and J. A. Benavides Hypervariable Noncoding Sequences in Saccharomyces cerevisiae Genetics, August 1, 2005; 170(4): 1575 - 1587. [Abstract] [Full Text] [PDF]

				T. Nakayashiki, C. P. Kurtzman, H. K. Edskes, and R. B. Wickner Yeast prions [URE3] and [PSI+] are diseases PNAS, July 26, 2005; 102(30): 10575 - 10580. [Abstract] [Full Text] [PDF]

				H. Qin, W. B. Wu, J. M. Comeron, M. Kreitman, and W.-H. Li Intragenic Spatial Patterns of Codon Usage Bias in Prokaryotic and Eukaryotic Genomes Genetics, December 1, 2004; 168(4): 2245 - 2260. [Abstract] [Full Text] [PDF]

				J. Baum, A. W. Thomas, and D. J. Conway Evidence for Diversifying Selection on Erythrocyte-Binding Antigens of Plasmodium falciparum and P. vivax Genetics, April 1, 2003; 163(4): 1327 - 1336. [Abstract] [Full Text] [PDF]