The SSD1 gene of Saccharomyces cerevisiae is a polymorphic locus that affects diverse cellular processes including cell integrity, cell cycle progression, and growth at high temperature. We show here that the SSD1-V allele is necessary for cells to achieve extremely long life span. Furthermore, addition of SSD1-V to cells can increase longevity independently of SIR2, although SIR2 is necessary for SSD1-V cells to attain maximal life span. Past studies of yeast aging have been performed in short-lived ssd1-d strain backgrounds. We propose that SSD1-V defines a previously undescribed pathway affecting cellular longevity and suggest that future studies on longevity-promoting genes should be carried out in long-lived SSD1-V strains.
AGING in Saccharomyces cerevisiae can be studied by mutations that extend the replicative life span of mother cells, defined as the number of daughters produced by a given mother cell prior to senescence. One cause of aging in yeast is the accumulation of extrachromosomal ribosomal DNA circles (ERCs), circular DNA molecules derived from homologous recombination within the ribosomal DNA (rDNA; Sinclair and Guarente 1997). ERCs are self-replicating and asymmetrically segregated to the mother-cell nucleus during S-phase, resulting in an exponential increase in ERC copy number with age and, ultimately, in cell death (Sinclair and Guarente 1997).
One important determinant of yeast longevity is the Sir2 protein (Kaeberleinet al. 1999). Sir2p is an NAD-dependent histone deacetylase (Imaiet al. 2000; Landryet al. 2000; Smithet al. 2000) required for transcriptional silencing at telomeres (Gottschlinget al. 1990), silent mating (HM) loci (Ivyet al. 1986; Rine and Herskowitz 1987), and the rDNA (Bryket al. 1997; Smith and Boeke 1997). Mutation of SIR2 results in increased rDNA recombination (Gottlieb and Esposito 1989), increased ERC formation (Kaeberleinet al. 1999), and decreased life span (Kennedyet al. 1995), whereas overexpression extends life span by 30–40% (Kaeberleinet al. 1999). Sir2p is also required for life-span extension by calorie restriction (CR), demonstrating the importance of this protein as a central regulator of longevity (Linet al. 2000). Overexpression of a Sir2p homolog, Sir-2.1, has been shown to extend life span in the nematode Caenorhabditis elegans, suggesting that Sir2 proteins regulate aging in higher eukaryotes as well (Tissenbaum and Guarente 2001).
SSD1 is a polymorphic locus that affects diverse cellular processes. Two allele classes, designated SSD1-V and ssd1-d, have been identified for SSD1. SSD1-V alleles confer viability in the absence of the Sit4 protein phosphatase and code for functional Ssd1 protein. In contrast, strains carrying ssd1-d alleles are inviable in the absence of Sit4p (Suttonet al. 1991), and d-type alleles are likely null for Ssd1p function. Both V- and d-type alleles have been found in natural isolates and in laboratory strains of S. cerevisiae. A recent report (Wheeleret al. 2003) suggests that the SSD1 allele type affects pathogenicity of yeasts, indicating that allelic variation at the SSD1 locus may be important for survival under various environmental conditions.
A potential role for SSD1-V as a regulator of cell life span was suggested by the observation that SSD1-V suppresses many phenotypes associated with mutation of the MPT5/UTH4 gene (Kaeberlein and Guarente 2002). MPT5 is a post-transcriptional regulator (Tadauchiet al. 2001) involved in regulating the pheromone response (Chen and Kurjan 1997), cell-wall stability (Kaeberlein and Guarente 2002), telomere silencing (Cockellet al. 1998), and longevity (Kennedyet al. 1995). Like SIR2, MPT5 is a limiting factor for longevity: overexpression of MPT5 extends life span, whereas the deletion of MPT5 has the opposite effect (Kennedyet al. 1997).
SSD1-V suppresses the temperature-sensitive growth defect caused by mutation of MPT5 as well as the sensitivity to calcofluor white (CFW) and sodium dodecyl sulfate (SDS; Kaeberlein and Guarente 2002). In strains lacking SSD1-V, deletion of MPT5 is synthetically lethal in combination with loss of function in either of the SBF or CCR4 transcriptional complexes (Kaeberlein and Guarente 2002), both of which function downstream of protein kinase C (Pkc1p) to promote cell-wall biosynthesis (Igualet al. 1996; Maddenet al. 1997; Changet al. 1999). These results were interpreted to suggest that Mpt5p, Ssd1p, and Pkc1p define three parallel pathways that function to ensure cell integrity (Kaeberlein and Guarente 2002).
In addition to suppressing the cell integrity defects, SSD1-V suppresses the shortened life span caused by deletion of MPT5 (Kaeberlein and Guarente 2002). These observations raise the possibility that SSD1-V might also promote longevity in wild-type cells. Here we show that addition of a single copy of SSD1-V to ssd1-d wild-type cells extends life span in at least two different strain backgrounds. Furthermore, life-span extension by SSD1-V does not require the Sir2 protein, although the presence of both SSD1-V and SIR2 is necessary for maximal longevity.
MATERIALS AND METHODS
Strains and genetic techniques: The strains used in this study are listed in Table 1. All strains were derived from W303R (described in Millset al. 1999), PSY316 (described in Parket al. 1999), or BKY5 (described in Kennedyet al. 1995). Genetic crosses, sporulation, and tetrad analysis were carried out as described (Sherman and Hicks 1991). The genotype of inviable spore clones was inferred, when possible, on the basis of marker segregation in viable spore clones from the same tetrad. Unless otherwise noted, cells were cultured in YPD or synthetic media prepared using conventional methods (Guthrie and Fink 1991). Yeast transformation was accomplished by the lithium acetate method (Gietzet al. 1992). All other gene deletions were generated by transforming cells with PCR-amplified disruption cassettes as described (Kaeberleinet al. 1999). In each case, the entire open reading frame was removed. All disruptions were verified phenotypically or by PCR. The SSD1-V integrating plasmids p406SSD1 and p405SSD1 were previously described (Kaeberlein and Guarente 2002). Unless otherwise indicated, all SSD1-V strains contain SSD1-V integrated at the marker locus and still carry the ssd1-d allele at the SSD1 locus. Deletion of ssd1-d does not affect any of the phenotypes tested, including life span, growth at 30°, 37°, or 40°, or sensitivity to calcofluor white.
Determination of SSD1 allele: To determine which SSD1 allele was present in PSY316, one copy of the SIT4 gene was deleted in diploid cells. Sporulation of these cells revealed that deletion of SIT4 always resulted in lethality in haploid spore clones (n > 20 sit4Δ spores). This lethality was suppressed by integration of a single copy of SSD1-V at the URA3 locus. Therefore, we conclude that in PSY316 the SSD1 allele type is ssd1-d.
Life span, recombination, and ERC analysis: Life spans were performed as described (Kaeberlein and Guarente 2002). Statistical significance was determined by a Wilcoxon ranksum test. Average life span is different for P < 0.05. Figures 1, 2, 3, 4, 5 represent data derived from a single experiment, unless otherwise stated. ERC levels were determined as described (Defossezet al. 1999; Kaeberleinet al. 1999). ERCs were separated on a 0.6% agarose gel without addition of ethidium bromide at 1 V/cm for 48 hr. rDNA recombination rate was determined as described (Kaeberleinet al. 1999).
Microarray analysis: RNA isolation and microarray analysis were performed essentially as described (Linet al. 2002). For the purposes of comparative statistical analyses, candidate genes with altered transcript levels in SSD1-V cells relative to wild type were defined as such if the average ratio of SSD1-V (Cy5) to ssd1-d (Cy3) was >1.5 or <0.667 (1.5-fold decrease) in three independent experiments. All such genes are listed in supplemental Table 1 at http://www.genetics.org/supplemental/). As a control, two independent microarray analyses were performed on ssd1Δ (Cy5) cells relative to ssd1-d (Cy3) cells. The 124 genes defined as regulated by CR represent the subset of genes found to show significant changes in mRNA expression both in cells lacking HXK2 and in cells grown on 0.5% glucose (Linet al. 2002). Statistical significance of the overlap between regulated genes in different experiments shown in Table 2 was calculated using a hypergeometric distribution. P values were obtained from the online hypergeometric distribution calculator at http://www.alewand.de/stattab/tabdiske.htm. Gene function annotation was obtained from the Saccharomyces Genome Database. Supplemental Table 1, normalized ratio (Cy5/Cy3), and spot intensity data sets for all experiments presented in this article are available at http://web.mit.edu/biology/guarente/arrays/kaeberlein.
SSD1-V extends life span and improves growth at high temperature: Mpt5p is a limiting factor for longevity and functions in a pathway parallel to SSD1-V for cell integrity (Kaeberlein and Guarente 2002). On the basis of this genetic interaction, we hypothesized that SSD1-V might also regulate longevity. We first determined that our wild-type strain PSY316 carries the ssd1-d allele at the SSD1 locus, on the basis of the inviability caused by deletion of SIT4 in this background (see materials and methods). A single copy of SSD1-V integrated at the URA3 locus results in an ∼50% increase in mean life span (Figure 1A). Integration of SSD1-V at the SSD1 locus has a similar effect on life span (not shown). Deletion of the chromosomal ssd1-d allele of PSY316 has no effect on life span and does not affect life-span extension by SSD1-V (Figure 1A). Therefore, ssd1-d is a null allele with respect to life span. All subsequent experiments were carried out in the parental ssd1-d background.
PSY316 is a moderately long-lived yeast strain; however, many yeast aging studies have been carried out in short-lived strain backgrounds having mean life spans of 10–15 generations. To determine whether the life-span extension by SSD1-V was strain specific, we integrated SSD1-V into the short-lived strain BKY5. We verified that BKY5 carries an ssd1-d allele (see materials and methods) as well as a previously identified C-terminal truncated allele of MPT5 (Kennedyet al. 1997). Addition of SSD1-V to BKY5 results in an 85% increase in mean life span (Figure 1B). Thus, SSD1-V promotes long life span in at least two different ssd1-d strain backgrounds.
We had previously observed that cells from strain PSY316 grow normally at 37°, but are incapable of sustained growth at 40°. Cells grown at 40° generally arrest as large-budded cells with a significant fraction undergoing lysis (data not shown), consistent with a loss of cell-wall integrity at the restrictive temperature. Addition of SSD1-V fully suppresses these phenotypes and allows growth of PSY316 at 40° (Figure 1C). Addition of SSD1-V to PSY316 also improves growth in the presence of the cell-wall-perturbing agents CFW and SDS (data not shown), as previously reported for strain W303R (Kaeberlein and Guarente 2002).
SSD1-V extends life span in the absence of SIR2: The Sir2 protein is a central regulator of yeast longevity, necessary for life-span extension in response to environmental signals such as reduced nutrient availability (Linet al. 2000) and osmotic stress (Kaeberleinet al. 2002). To place SSD1-V into a genetic pathway relative to Sir2p, we integrated the SSD1-V allele into a strain lacking SIR2. Deletion of SIR2 shortens wild-type life span by ∼50% (Kaeberleinet al. 1999). Surprisingly, addition of SSD1-V resulted in a significant life-span extension in the absence of Sir2p (Figure 2A), although sir2 SSD1-V cells are shorter lived than SIR2 wild-type cells.
Due to the extremely short life span caused by lack of Sir2p, we also wished to determine the effect of SSD1-V on sir2 fob1 double-mutant cells, which have an almost wild-type life span (Kaeberleinet al. 1999). As predicted by our previously proposed model (Lin et al. 2000, 2002), CR by growth on low glucose fails to extend life span in the absence of SIR2 (Figure 2B). In contrast, sir2 fob1 SSD1-V cells grown on 2% glucose have a life span that is significantly longer than that of sir2 fob1 ssd1-d cells (Figure 2C). However, as was the case for sir2 SSD1-V cells, sir2 fob1 SSD1-V cells do not live as long as SIR2 FOB1 SSD1-V cells. Therefore, SSD1-V acts in a novel, SIR2-independent pathway for longevity, although Sir2p is required for maximum longevity in SSD1-V cells.
The Sir2-dependent life-span extension caused by CR is the result of a metabolic shift from fermentation to respiration (Linet al. 2002). It is possible that a portion of the longevity conferred by SSD1-V is achieved by causing the cell to undergo a similar metabolic shift. To address this possibility, the effect of SSD1-V on life span was examined in a respiration-deficient strain lacking the CYT1 gene encoding cytochrome c1. Mutation of CYT1 prevents life-span extension by CR or by overexpression of the Hap4p transcription factor (Linet al. 2002). In contrast, SSD1-V extends the life span of cells lacking CYT1 to the same extent as that of wild-type cells (Figure 2D), suggesting that the mechanism of life-span extension by SSD1-V is independent of mitochondrial function and respiration.
High osmolarity extends the life span of ssd1-d but not of SSD1-V cells: We have previously demonstrated that the osmotic concentration of media is a determining factor for mother-cell life span. Addition of 1 m sorbitol suppresses the short life span and cell-wall defects of mpt5 ssd1-d cells in the W303 strain background (Kaeberlein and Guarente 2002). Growth on YPD supplemented with 1 m sorbitol (YPDS), 1 m xylitol, or 1 m glucose also extends the life span of MPT5 ssd1-d cells in PSY316 through a SIR2-dependent mechanism (Kaeberleinet al. 2002). We were therefore interested in determining whether high osmolarity would affect the life span of long-lived SSD1-V cells.
As observed in W303R, growth on YPDS suppressed the temperature sensitivity and short life span of mpt5 ssd1-d cells in the PSY316 strain background (not shown). Growth on YPDS also dramatically increased life span in wild-type ssd1-d cells by ∼60% (Figure 2E). In contrast, growth on YPDS resulted in only a modest 10% increase in the life span of SSD1-V cells.
SSD1-V acts independently of ERC formation or accumulation: Sir2p and calorie restriction promote longevity by decreasing the formation and accumulation of ERCs in mother cells (Kaeberleinet al. 1999; Linet al. 2000). To determine whether the long life span of SSD1-V cells is also due to fewer ERCs, we measured the rate of ERC formation and the amount of ERCs present in ssd1-d and SSD1-V cells. ERC formation was estimated by determining the frequency at which an ADE2 marker integrated into the rDNA is lost, as demonstrated by the presence of half-sectored colonies (Kaeberleinet al. 1999). No difference was observed between SSD1-V and ssd1-d cells by this assay (Figure 3A). Upon direct quantitation of ERCs from unsorted cells, we observed that SSD1-V cells often had higher levels of ERCs than ssd1-d cells (Figure 3B), indicating that the life-span extension caused by SSD1-V is unlikely to be the result of decreased ERC formation or accumulation. This is consistent with the observation that SSD1-V does not require Sir2p to extend life span and may indicate that SSD1-V increases the resistance of cells to ERCs (see discussion).
Transcriptional analysis of SSD1-V: We previously observed that calorie restriction by growth in 0.5% glucose, which promotes long life span, causes characteristic changes in gene expression that are reproduced in two genetic models of CR, namely overexpression of HAP4 and deletion of HXK2 (Linet al. 2002). More recently, we demonstrated that growth in the presence of high external osmolarity also extends life span in a SIR2-dependent manner and results in a gene expression profile with significant similarity to calorically restricted cells (Kaeberleinet al. 2002). Although SSD1 has been implicated in many different genetic pathways, little is known regarding the function of the Ssd1 protein. To further understand the effect of SSD1-V on cell physiology and longevity, we used microarray analysis to examine the transcriptional profile of SSD1-V cells relative to wild-type ssd1-d cells.
Messenger RNA was harvested from logarithmically growing ssd1Δ, ssd1-d, or SSD1-V cells. Using RNA derived from three independent experiments, a total of 97 genes were observed to undergo a change in expression >1.5-fold in SSD1-V cells relative to ssd1-d cells (supple mental Table 1 at http://www.genetics.org/supplemental/). Of these 97 genes, only 6 underwent similar transcriptional changes in calorically restricted cells (Table 2). This is only slightly greater than the number of genes expected to overlap between the SSD1-V and CR data sets by chance and is in contrast to the highly significant overlap in transcriptional changes observed between CR and HAP4 overexpression (Linet al. 2002) or between CR and high external osmolarity (Kaeberleinet al. 2002). Intriguingly, of the 6 genes that show similar transcriptional changes in calorically restricted cells and SSD1-V cells, 4 are involved in iron-siderochrome transport: FIT1, FIT2, FIT3, and ARN1 (supplemental Table 1 at http://www.genetics.org/supplemental/).
NCA3 mRNA is increased in SSD1-V cells: One particularly interesting candidate gene that shows altered mRNA levels in SSD1-V cells is NCA3 (supplemental Table 1 at http://www.genetics.org/supplemental/). Nca3p is a member of the SUN family of proteins (Sim1, Uth1, Nca3, and Sun4) and functions to promote maturation of the mitochondrially encoded ATP8-ATP6 cotranscript (Pelissieret al. 1995). Upregulation (3.7-fold) of NCA3 by SSD1-V is striking because Nca3p shares extensive homology (60%) with the aging protein Uth1p. Interestingly, we find that NCA3 mRNA is increased 4-fold in long-lived cells lacking Uth1p (data not shown). Overexpression of NCA3 from the ADH1 promoter results in a slight, but reproducible, increase in life span (Figure 4A), suggesting that Nca3p dosage can affect longevity. However, NCA3 is not required for the majority of the life-span extension seen in SSD1-V cells, as demonstrated by the finding that nca3 SSD1-V cells have a life span comparable to that of NCA3 SSD1-V cells (Figure 4B). Therefore, we conclude that increased transcription of NCA3 accounts for, at most, a minor fraction of the longevity-promoting activity of SSD1-V.
MPT5 and SIR2 affect longevity in a pathway parallel to SSD1-V: We have previously demonstrated that MPT5 and SSD1-V act in parallel pathways to promote cell integrity and that SSD1-V suppresses the short life span caused by deletion of MPT5 in the W303R strain background (Kaeberlein and Guarente 2002). As expected, addition of SSD1-V similarly suppresses the short life span of cells lacking MPT5 in PSY316 (Figure 5A). It is interesting to note, however, that mpt5 SSD1-V cells have a life span intermediate between cells lacking both MPT5 and SSD1-V and cells with functional copies of both genes. In fact, mpt5 SSD1-V cells have a life span not significantly different from that of the wild-type MPT5 ssd1-d strain, suggesting that MPT5 and SSD1-V have additive effects on longevity, as would be expected for genes functioning in parallel pathways.
Like SSD1-V, overexpression of MPT5 increases mother-cell life span (Kennedyet al. 1997). Since SSD1-V is capable of extending the life span of cells lacking SIR2, we wished to determine whether overexpression of MPT5 would have a similar effect. In contrast to SSD1-V, MPT5 overexpression fails to extend the life span of sir2 fob1 cells (Figure 5B), suggesting that MPT5 and SIR2 act in the same pathway to promote longevity. Furthermore, overexpression of SIR2 fails to further extend the life span of cells in which MPT5 is overexpressed (Figure 5B).
It was previously observed that cells with altered dosage of MPT5 display changes in telomeric and rDNA silencing (Kennedy 1996), suggesting a further link between MPT5 and SIR2. Overexpression of MPT5 increases rDNA silencing and decreases telomeric silencing, while deletion has an opposite effect (Table 3). Integration of SSD1-V, in contrast, has no detectable effect on silencing at either locus. Consistent with the inability of MPT5 overexpression to extend life span in the absence of SIR2, the enhanced rDNA silencing observed in cells overexpressing MPT5 is fully suppressed by deletion of SIR2 (Table 3). On the basis of these results, we propose that overexpression of MPT5 increases life span by relocalizing Sir2p from telomeres to the rDNA (see discussion), thus enhancing the ability of Sir2p to inhibit ERC accumulation in aging mother cells.
One cause of aging in yeast is the accumulation of ERCs (Sinclair and Guarente 1997). A central regulator of ERC formation and longevity is the Sir2p histone deacetylase (Kaeberleinet al. 1999). Several genes that regulate yeast life span act by altering Sir2p activity or dosage (Kaeberlein et al. 1999, 2002; Lin et al. 2000, 2002). Here we present evidence that SSD1-V defines a novel Sir2p-independent pathway necessary for cells to achieve extreme longevity.
Two pathways promoting longevity: We initially began studying SSD1-V on the basis of its ability to suppress the temperature sensitivity caused by mutation of the UTH4/MPT5 gene. Like Sir2p, Mpt5p is limiting for life span in wild-type cells (Kennedyet al. 1997). Overexpression of Mpt5p increases life span and rDNA silencing in a Sir2p-dependent manner (Figure 5B, Table 3), suggesting that Mpt5p promotes longevity by increasing Sir2p activity at the rDNA.
In contrast to overexpression of MPT5, addition of a single copy of SSD1-V extends life span in both SIR2 wild-type cells and cells lacking Sir2p (Figure 2, A and C). However, the sir2 fob1 SSD1-V strain has a life span that is shorter than that of the SIR2 FOB1 SSD1-V strain, demonstrating that Sir2p is required for maximum longevity in SSD1-V cells. This is consistent with the observation that MPT5 SSD1-V cells have a longer life span than mpt5 SSD1-V cells (Figure 5A) and suggests a model whereby Mpt5p and Sir2p function in one pathway to increase life span while Ssd1p functions in a parallel pathway (Figure 6).
Mechanism of life-span extension by SSD1-V: How does SSD1-V act to extend life span? The effect of Sir2p on life span is, at least partially, due to its ability to deacetylate rDNA histones and inhibit ERC formation (Kaeberleinet al. 1999). There is no evidence to suggest that SSD1-V affects the rate of ERC formation or accumulation. Addition of SSD1-V had no detectable effect on rDNA recombination (Figure 3A) or on rDNA silencing (Table 3) in PSY316. Moreover, we failed to detect a decrease in ERC levels in SSD1-V cells relative to those in ssd1-d cells (Figure 3B). While it is still possible that SSD1-V affects ERC replication or segregation specifically in aged cells, we feel that this is unlikely to be the case. An alternative possibility is that SSD1-V makes cells more resistant to ERCs, rather than reducing ERC levels. In support of this hypothesis, we often observed that steady-state ERC levels were increased in unsorted SSD1-V cells relative to those in wild-type ssd1-d cells (Figure 3B), although this was not always the case. The mechanism by which ERCs induce senescence is currently unknown. One hypothesis is that ERCs bind to and titrate key cellular replication or transcription factors away from their normal targets. Alternatively, the rapid amplification of rDNA sequence could alter rRNA transcription and/or processing, resulting in ribosome dysregulation. Ssd1p has been shown to bind RNA and is predicted to have RNase activity (Uesonoet al. 1997). Perhaps SSD1-V alters rRNA or ribosome biogenesis in a manner that makes cells more resistant to ERCs.
One attractive hypothesis is that SSD1-V promotes longevity by increasing cell-wall stability and cell integrity. SSD1-V suppresses several temperature-sensitive mutations that weaken the cell wall (Table 4) and has been found to directly affect cell-wall composition (Wheeleret al. 2003). SSD1-V also improves resistance to the cell-wall-perturbing agents CFW and SDS (Kaeberlein and Guarente 2002), increases the maximum temperature at which PSY316 is capable of growth (Figure 1C), and alters the transcription of cell-wall biosynthetic and structural genes. Perhaps the cell wall becomes limiting in very old cells and SSD1-V extends life span by stabilizing it. How might cell-wall stability limit replicative life span? The terminal phenotype of yeast cells in the life-span assay is cell cycle arrest often accompanied by cell lysis (McVey et al. 2001). Enhanced cell-wall stability may prevent cell lysis late in life and allow additional cell divisions to occur.
Genetic diversity and the study of aging: The data presented here identify a genetic polymorphism that has a profound effect on mother-cell life span. Genetic polymorphisms have also been proposed to affect the likelihood of achieving extreme longevity in human populations (e.g., Pucaet al. 2001), as well as in other model systems. Since both ssd1-d and SSD1-V allele types have been isolated from natural yeast populations, the SSD1 locus represents a true polymorphic locus affecting longevity. In the past, researchers studying aging in yeast have tended to avoid using long-lived wild-type backgrounds. We speculate that the majority (if not all) of these shorter-lived yeast strains carry ssd1-d alleles. A comprehensive reevaluation of previously identified mutations affecting life span in a long-lived SSD1-V background would be of value to the field.
SSD1-V confers extreme longevity on yeast mother cells by a pathway independent of Sir2p. Sir2 proteins have been found to extend life span in animals and, like SIR2, SSD1 homologs are present in yeast, worms, flies, and mammals. Might SSD1 family members also promote longevity outside of yeast? The mechanism by which SSD1-V cells achieve up to 85% longer life span is still unknown. Further work should be devoted to testing candidate longevity genes regulated by SSD1-V and to defining the molecular function of Ssd1p in cells.
We thank N. Bishop, B. Kennedy, and T. Kaeberlein for helpful discussion and insight. This work was supported by grants to L.G. from the National Institutes of Health (NIH), The Ellison Medical Foundation, The Seaver Institute, and the Howard and Linda Stern Fund. G.R.F. is supported by grants from the NIH and is an American Cancer Society Professor of Genetics. A.A.A. is supported by an NIH Training Grant in Genomic Sciences, sponsored by the Biotechnology Process Engineering Center.
Communicating editor: B. J. Andrews
- Received August 13, 2003.
- Accepted December 23, 2003.
- Copyright © 2004 by the Genetics Society of America