rDNA copy number variation affects yeast fitness in response to different environments

Abstract

Ribosomal DNA (rDNA) in eukaryotes is maintained in hundreds of copies with rDNA copy number varying greatly among individuals within a species. In the budding yeast Saccharomyces cerevisiae, the rDNA copy number across wild isolates ranges from 90 to 300 copies. Previous studies showed that 35 rDNA copies are sufficient for ribosome biogenesis in this yeast and enable wild-type-like growth in standard laboratory growth conditions. We addressed 2 major questions concerning rDNA copy number variation in this yeast: (1) What are the fitness consequences of rDNA copy number variation outside and within the natural range in standard laboratory growth conditions? (2) Do these fitness effects change in different growth conditions? We used growth competitions to compare the fitness effects of rDNA copy number variation in otherwise isogenic strains whose rDNA copy number ranged from 35 to 200. In standard growth conditions, we found that fitness gradually increases from 35 rDNA copies until reaching a plateau that spans from 98 to 160 rDNA copies, well within the natural range. However, rDNA copy number-dependent fitness differed across environments. The gradual fitness increase with increasing rDNA copy number in standard growth conditions gave way to a markedly lower fitness of strains with copy numbers below the natural range in these 2 stress conditions. These results suggest that selective pressures drive rDNA copy number in this yeast to at least ∼100 copies and that a higher number of copies might buffer against environmental stress. The similarity of the S. cerevisiae rDNA copy number range to the ranges reported in C. elegans, D. melanogaster, and humans points to conserved selective pressures maintaining the range of natural rDNA copy number in these highly diverse species.

Introduction

Ribosomal DNA (rDNA) encodes the ribosomal RNAs that provide the structural and enzymatic features of ribosomes. The rDNA is maintained in tandem arrays at hundreds of copies per cell (Kobayashi et al. 1998; Nelson et al. 2019). The repetitive nature of the rDNA arrays makes them prone to copy number variation within species. There are 90 to 300 copies in the yeast Saccharomyces cerevisiae, 70 to 400 copies in the worm Caenorhabditis elegans, 80 to 600 copies in the fly Drosophila melanogaster, 500 to 2,500 copies in the plant Arabidopsis thaliana, and 100 to 600 copies in humans (Mohan and Ritossa 1970; Thompson et al. 2013; Morton et al. 2020, 2023; Hall et al. 2021, 2022). Given the breadth of variation, both in copy number and the fraction of the genome it represents, rDNA may be an underappreciated source of genetic variation influencing fitness.

rDNA copy number variation is increasingly recognized as an influential factor in cell physiology. A minimum number of rDNA copies is required for the high level of rRNA transcription necessary for ribosome biogenesis with rRNAs making up ∼80% of total cellular RNA in yeast (Warner 1999). For S. cerevisiae, this minimum number of rDNA copies is 35 (French et al. 2003; Kim et al. 2006; Ide et al. 2010; Kwan et al. 2023), a number that is significantly lower than observed in wild yeast isolates and laboratory strains (Morton et al. 2020; Hall et al. 2022). Strains with substantially reduced rDNA copy numbers sufficient for ribosome biogenesis (35–40 copies) show compromised genome replication and increased susceptibility to DNA damage but no growth defects in standard laboratory growth conditions (Ide et al. 2010; Salim et al. 2017; Kwan et al. 2023). As few as 8–10 rDNA copies suffice to maintain yeast viability (Sanchez et al. 2019; Jiang et al. 2024); however, the resulting limitation on ribosome production creates a strong selective pressure for suppressors that ameliorate their slow-growth phenotype (Sanchez et al. 2019). In animals and humans, reductions of rDNA copy number below the naturally occurring range of variation is associated with developmental abnormalities and disease phenotypes (Ritossa and Atwood 1966; Xu et al. 2017; Valori et al. 2020; Morton et al. 2023). However, the intrinsic differences among tissues in multicellular animals make it difficult to determine the extent to which rDNA copy number variation affects cellular fitness.

We therefore wanted to investigate in yeast cells (1) if rDNA copy number variation below and within the natural range affects cellular fitness; (2) if increases in rDNA copy number from the minimal copy number up to levels observed in wild and laboratory strains affect fitness gradually or if there are copy number thresholds; and (3) what additional biological processes are affected by substantial reduction in rDNA copy number. To tackle these questions, we generated a set of isogenic strains in the yeast S. cerevisiae containing between 35 and 200 copies of rDNA, which spans the range from the threshold required for ribosome biogenesis up to the natural variation present in the species. We examined these strains by competitive fitness assays in different growth conditions and used transcriptome analysis of a strain with 35 rDNA copies and a control strain with 180 copies to identify additional, potentially altered pathways.

We found that rDNA copy number variation profoundly affects fitness in haploid yeast. In standard laboratory conditions, strains below ∼100 copies showed reduced fitness that correlated with the severity of copy number reduction. In contrast, strains with copy numbers ranging from 98 to 160 rDNA copies displayed similarly high fitness, establishing a fitness plateau that roughly coincides with the naturally occurring rDNA copy number range reported for this yeast (Morton et al. 2020; Hall et al. 2022). Additional rDNA copies are detrimental, as fitness decreased in strains with more than 160 rDNA copies. A comparison of the transcriptomes of the 35 and 180 rDNA copy number strains showed the expected altered expression of genes involved in mitigating DNA replication stress, in addition to unexpected expression changes in canonical stress response genes. We followed up on the latter findings with phenotyping and competition assays at increased temperature and with growth on a nonfermentable carbon source. We found that the rDNA copy number variants showed different fitness in different growth conditions, and that the gradual fitness changes observed in standard growth conditions had given way to threshold effects. Taken together, our findings suggest that the range of rDNA copy numbers observed in wild yeast strains acts as a buffer under varied environments.

Methods

Yeast media

Yeast strains used in 30 and 37°C yeast competition assays were grown in synthetic complete media buffered with 1% succinic acid (per liter: 1.45 g yeast nitrogen base, 20 g glucose, 10 g succinic acid, 6 g NaOH, 5 g (NH4)₂SO₄, 2.8 g amino acid powder mix with pH adjusted to 5.8). Yeast strains used in glycerol yeast competition assays were grown in synthetic complete media buffered with 1% succinic acid (per liter: 1.45 g yeast nitrogen base, 30 g glycerol, 10 g succinic acid, 6 g NaOH, 5 g (NH4)₂SO₄, 2.8 g amino acid powder mix with pH adjusted to 5.8). Yeast strains used in caloric restriction yeast competition assays were grown in synthetic minimum media buffered with 1% succinic acid (per liter: 1.61 g yeast nitrogen base, 1 g glucose, 11.10 g succinic acid, 6.67 g NaOH, 100 mg (NH4)₂SO₄ with pH adjusted to 5.8) (Table 1).

Table 1.

Yeast strains used.

Strain	Source	Identifier
S288c MATa fob1::cloNAT; BY rDNA (35 copies)	This study	KT003
S288c MATa fob1::cloNAT; BY rDNA (45 copies)	This study	KT088
S288c MATa fob1::cloNAT; BY rDNA (50 copies)	This study	KT095
S288c MATa fob1::cloNAT; BY rDNA (55 copies)	This study	KT004
S288c MATa fob1::cloNAT; BY rDNA (80 copies)	This study	KT005
S288c MATa fob1::cloNAT; BY rDNA (88 copies)	This study	KT006
S288c MATa fob1::cloNAT; BY rDNA (95 copies)	This study	KT007
S288c MATa fob1::cloNAT; BY rDNA (98 copies)	This study	KT102
S288c w/ RM11-1a MATa fob1::cloNAT; BY rDNA (100 copies)	This study	KT121
S288c MATa fob1::cloNAT; BY rDNA (104 copies)	This study	KT097
S288c MATa fob1::cloNAT; BY rDNA (140 copies)	This study	KT132
S288c MATa fob1::cloNAT; BY rDNA (160 copies)	This study	KT125
S288c MATa fob1::cloNAT; BY rDNA (180 copies)	This study	KT001
S288c MATa fob1::cloNAT; BY rDNA (200 copies)	This study	KT128
S288c MATa fob1::cloNAT; HO::GFP-KanMX; BY rDNA (180 copies)	This study	KT013
S288c MATa fob1::cloNAT; HO::GFP-KanMX; BY rDNA (180 copies)	This study	KT014
S288c MATα fob1::cloNAT; BY rDNA (35 copies)	This study	KT015
S288c MATa FOB1; BY rDNA (170 copies)	This study	KT055
S288c MATa FOB1; BY rDNA (59 copies)	This study	S35-1 G30
S288c MATa FOB1; BY rDNA (57 copies)	This study	S35-2 G30
S288c MATa FOB1; BY rDNA (52 copies)	This study	S35-3 G30
S288c MATa FOB1; BY rDNA (170 copies)	This study	S170-1 G30
S288c MATa FOB1; BY rDNA (170 copies)	This study	S170-2 G30
S288c MATa FOB1; BY rDNA (170 copies)	This study	S170-3 G30

Preparation of DNA embedded in agarose

Cells were inoculated into 2 mL synthetic complete media buffered with 1% succinic acid and allowed to grow overnight to stationary phase (1.5–3 × 10⁸ cells/mL). 200 μL of cells were pelleted for 1 min at 15,000 rcf and the supernatant was discarded. Cells were washed with 85 μL 50 mM EDTA, resuspended in 90 μL 1% SeaPlaque GTG agarose in 50 mM EDTA then transferred into plug molds. Plugs were allowed to solidify for 15 min at 4°C then incubated in 1 mL spheroplasting solution (1.0 M sorbitol, 20 mM EDTA pH 8.0, 10 mM Tris-HCl pH 7.4, 14.3 mM β-mercaptoethanol, 0.5 mg/mL Zymolyase-20T [Amsbio]) for 2–4 h at 37°C with gentle shaking. Plugs were washed once with LDS (1% lithium dodecyl sulfate, 100 mM EDTA pH 8.0, 10 mM Tris–HCl pH 8.0) and incubated overnight at 37°C in LDS. Plugs were then washed 3 × 20 min in 0.2X NDS (1X NDS pH 9.5: 0.5 M EDTA, 10 mM Tris base, 1% Sarkosyl) and 5 × 20 min in TE pH 8.0. All processed plugs were stored at 4°C in TE pH 8.0 until use.

Determination of rDNA copy numbers

Intact chromosomes were resolved by contour-clamped homogeneous electric field (CHEF) gel electrophoresis. A small slice (5 mm × 2 mm × 3 mm) of all genomic DNA agarose plugs was embedded in a 0.8% low electroendosmosis agarose gel containing filtered 0.5X TBE. CHEF gels were run in 2.3 L of 0.5X TBE using a Bio-Rad CHEF-DRII electrophoresis cell at 100 V for 68 h (switch time = 300 to 900 s). All gels were stained with ethidium bromide to visualize chromosome XII, which contains the rDNA, and all other chromosomes. Hansenula wingei (H. wingei) chromosomal DNA size marker standards were included in each CHEF gel electrophoresis run for size comparison. All CHEF gels were transferred to Genescreen Hybridization membrane using standard Southern blotting protocols (Tsuchiyama et al. 2013). We then hybridized the sequences of interest using a ³²P-labeled probe. The blots were exposed to X-ray film and to Bio-Rad Molecular Imaging FX phosphor screens for visualization and quantification of signal intensity. rDNA copy number estimates and error rates as previously described (Morton et al. 2020).

Yeast competitions

Cells were streaked out on YEPD plates and allowed to incubate overnight at 30°C. Cells were inoculated into 2 mL synthetic complete media with 2% glucose buffered with 1% succinic acid and allowed to grow overnight to stationary phase (1.5–3 × 10⁸ cells/mL). 100 μL of overnight cultures was transferred to 5 mL synthetic complete media (1:50 dilution) and allowed to grow at 30°C for ∼5 h. After the 5 h incubation period, 500 μL of test strain cultures were mixed with 500 μL of 180 rDNA GFP competitor strain culture and vortexed thoroughly. 500 μL of each test strain and GFP competitor strain culture mix was transferred to 1.5 mL sterile water (1:4 dilution) to record cell number of all strains with the Gilford Stasar Spectrophotometer at a wavelength of 660. 5 μL of each test strain and GFP competitor strain culture mix was transferred to 5 mL synthetic complete media (1:1,000 dilution) and incubated at 30°C overnight. Cell fitness was determined by measuring the change in GFP population of cells over time with flow cytometry of cell culture aliquots from day 0 and day 5. Batches are biological replicates that involved either 3 technical replicates for each pair-wise competition (Figs. 1c and 4a) or 2 (Fig. 4b). All pair-wise competitions within a given batch were conducted at the same time for maximum comparability. Raw flow data are provided in Supplementary File 1, percentage change in Supplementary Table 1, cell numbers and equations in Supplementary Table 2.

Fig. 1.

rDNA copy number affects fitness in standard growth conditions. CHEF gel electrophoresis was performed to verify rDNA copy number in triplicate clones used to inoculate fitness competitions under standard growth conditions. a) Ethidium bromide-stained gels and b) the resulting Southern blots hybridized with a single copy Chr. XII probe (CDC45). c) Graph of fitness values calculated by the rate of population change for each rDNA copy number test strain against the GFP competitor strain (180 rDNA copies). Dots indicate individual fitness calculated for 3 technical replicates for each strain; shared dot and line colors indicate each of 3 biological replicates (batches). Competitions within a given batch were conducted at the same time. For visual clarity, a LOWESS curve for each batch is superimposed based on the batch data points (Cleveland 1979). Inset: distribution of rDNA copy number genotypes in wild S. cerevisiae isolates (Hall et al. 2022). Raw flow data are in Supplementary File 1, percentage change in Supplementary Table 1, cell numbers and equations in Supplementary Table 2, selection coefficients in Fig. 4c, Forward scatter comparisons in Supplementary Fig. 1d, P-values in Supplementary Table 5.

Fig. 4.

rDNA copy number variants show altered fitness in response to different growth conditions. Graph of fitness values calculated by the rate of population change for each rDNA copy number test strain against the GFP competitor strain (180 rDNA copies) at 37°C a) or grown in glycerol b). Dots indicate individual fitness calculated of 3 a) or 2 b) technical replicates for each strain; shared dot and line colors indicate biological replicates (batches). Competitions within a given batch were conducted at the same time, for visual clarity, a LOWESS curve for each batch is superimposed based on the batch data points (Cleveland 1979). Raw flow data are in Supplementary File 1, percentage change in Supplementary Table 1, cell numbers and equations in Supplementary Table 2, selection coefficients in Supplementary Fig. 1, P-values in Supplementary Table 5. c) Coefficients of selection were derived from the percent difference in growth for each genotype in the competition assays across 3 growth conditions (Supplementary Table 1).

Flow cytometry

∼350 μL of day zero yeast competition experiment cultures were suspended in 1 mL 50 mM sodium citrate and sonicated in preparation for flow cytometry. 100 μL of day 1 through day 5 yeast competition experiment overnight cultures were suspended in 1 mL 50 mM sodium citrate and sonicated in preparation for flow cytometry. 10,000 cells were analyzed on a BD Canto II flow cytometer and flow cytometry data was analyzed using FlowJo software. All data are provided in Supplementary File 1, forward scatter for standard growth conditions is shown in Supplementary Fig. 1d.

Passaging experiments and transient introduction and deletion of Fob1

To enable expansion of the rDNA array, we first transiently restored Fob1 expression by crossing the 35 rDNA fob1 strain to the 170 rDNA Fob1 + strain. Next, the resulting diploids were sporulated, and 3 Fob1 + spores with short rDNA arrays were identified. The spore-derived cultures were then passaged for 300 generations with samples collected at regular intervals (Supplementary Table 3). To capture rDNA copy numbers throughout the passaging experiment, Fob1 was deleted through yeast replacement transformation. rDNA copy numbers were determined by CHEF gel electrophoresis and Southern blotting as described (Morton et al. 2020).

RNA extraction for RNA-Seq

Asynchronous and late S phase logarithmic phase cells were collected, and RNA was isolated using an acid phenol: chloroform extraction protocol by resuspending frozen pellets in 200 μL lysis buffer (10 mM Tris, pH 8.0, 10 mM EDTA, 5% SDS), and 200 μL acid phenol and vortexed for 2 min. Cells were resuspended and incubated at 65°C for 1 h and vortexed occasionally. Incubated samples were kept on ice for 10 min before centrifuging samples at 4°C at 15,000 rcf for 10 min. After transferring the aqueous layer to new 1.5 microcentrifuge tubes, we added equal volumes of chloroform, and vortexed vigorously before centrifuging samples at 15,000 rcf for 10 min. To remove ribosomal RNA (rRNA), DNA oligos complimentary to rRNA sequences were incubated with RNA samples and then treated with RNase H to degrade RNA:DNA hybrids. RNA was then purified again using phenol:chloroform and ethanol precipitated. rRNA-depleted RNA was purified with a 2.2X RNAClean SPRI bead treatment (Beckman Coulter). Purified, rRNA-depleted RNA was treated with Turbo DNase (Invitrogen) to remove the rRNA-hybridizing DNA oligos: 0.1 volumes (2 μL) of 10X TURBO DNase Buffer was added to the RNA, followed by 2 μL TURBO DNase. The mixture was incubated at 37°C for 45 min. The DNase-treated RNA was then cleaned with 2.2X RNA SPRI beads, and a total volume of 12 μL was eluted. 9 μL of the eluted volume was used immediately for poly-A mRNA capture.

9 μL of final DNase-treated RNA and 16 μL of nuclease-free ultrapure water were added to 25 μL of RNA Purification Beads, mixed, and incubated at 65°C for 5 min then at 4°C for 30 s and finally at 23°C for 5 min. Samples were placed on a magnetic stand, supernatant was discarded, and 100 μL Bead Washing Buffer was added. After removing all residual supernatant, 25 μL of elution buffer was added to the samples, and they were incubated at 80°C for 2 min. 25 μL Bead Binding Buffer to samples and then washed with 100 μL Bead Washing Buffer. After removing supernatant, 19 μL ice-cold Fragmentation Master Mix (10.5 μL Nuclease-free ultrapure water and 10.5 μL Elute, Prime, Fragment High Mix per sample) was added to each sample and incubated at 94°C for 8 min. 17 μL of mRNA PCR product was added to new PCR tubes on ice in preparation of first strand cDNA synthesis.

First strand cDNA synthesis was performed by adding 8 μL First Strand Synthesis Master (9 μL First Strand Synthesis Act D Mix and 1 μL Reverse Transcriptase per sample) to 17 μL of each mRNA PCR product and incubating at 25°C for 10 min then at 42°C for 15 min and finally at 70°C for 15 min. Second strand cDNA synthesis was performed by adding 25 μL of the first cDNA strand reaction products with 25 μL Second Strand Marking Master Mix and incubating at 16°C for 1 h. After the incubation period, second cDNA strand reaction products were captured with 90 μL AMPure XP beads, supernatant was discarded, and AMPure XP beads were cleaned up twice with 175 μL fresh 80% ethanol on a magnetic stand. After removing all residual ethanol, AMPure XP beads were resuspended in 19.5 μL Resuspension Buffer and 17.5 μL supernatant was transferred into new PCR tubes. The first and second cDNA products were used immediately or stored at −20°C overnight for use the next day.

In preparation for adenylating 3′ ends, 12.5 μL A-Tailing Mix was added to the first and second cDNA products and incubated at 37°C for 30 min then at 70°C for 5 min. Anchors were ligated to samples by adding 2.5 μL Resuspension Buffer, 2.5 μL RNA Index Anchors (Illumina), 2.5 μL Ligation Mix and incubated at 30°C for 10 min followed by adding 5 μL Stop Ligation Buffer. 34 μL AMPure XP beads were added to samples, supernatant was discarded, and AMPure XP beads were cleaned up twice with 175 μL fresh 80% ethanol on a magnetic stand. After removing all residual ethanol, AMPure XP beads were resuspended in 22 μL Resuspension Buffer and 20 μL supernatant was transferred into new PCR tubes. The ligated cDNA products were used immediately or stored at −20°C overnight for use the next day.

Dual-indexed libraries were generated by adding 10 μL index adapters (Illumina) and 20 μL Enhanced PCR Mix to the ligated cDNA products. The mixture was mixed and incubated for 13 cycles at 98°C for 10 s, 60°C for 30 s, and 72°C for 30 s followed by an incubation at 72°C for 5 min. 50 μL AMPure XP beads were added to samples, supernatant was discarded, and AMPure XP beads were cleaned up twice with 175 μL fresh 80% ethanol on a magnetic stand. After removing all residual ethanol, AMPure XP beads were resuspended in 17 μL Resuspension Buffer and 15 μL supernatant was transferred into new PCR tubes. The concentrations and quality of all libraries were examined for quality using TapeStation 2,200 (Agilent Technologies) and Qubit (Thermo Fisher Scientific) and then subsequently diluted to the appropriate starting concentrations for sequencing. Libraries were sequenced using a NextSeq 550 with a 75 Hi kit (Illumina). Read lengths used were Index 1: 8 bp, Index 2: 8 bp, Read 1: 38 bp, Read 2: 38 bp.

Transcriptome profiling

We aligned to the S288C reference genome version R64.1.1 and obtained gene counts using hisat2 version 2.2.1 (Kim et al. 2019). DESeq2 was used to identify differentially expressed genes (DEGs) (Love et al. 2014). Genes with an adjusted P-value < 0.01 were further analyzed: 6 that differed between the 35- and 180-rDNA copy number in asynchronous culture (PRP11, HSP12, STR3, NCE103, YGP1, and CIN5) and 708 that differed between the 35- and 180-rDNA copy number in synchronous culture during late S phase, with 3 genes in common between these 2 sets (PRP11, STR3, and HSP12).

We performed gene ontology enrichment analysis for the following 4 sets of genes, each of which consists of only genes determined to be differentially expressed in late S phase between the 35- and 180- rDNA copy number strains: (1) genes that were more expressed in the 35-rDNA copy number strain than the 180-rDNA copy number strain; (2) genes that were more expressed in the 180-rDNA copy number strain than the 35-rDNA copy number strain; (3) genes that were at least 1.5× more expressed in the 35-rDNA copy number strain than the 180-rDNA copy number strain; (4) genes that were at least 1.5× more expressed in the 180-rDNA copy number strain than the 35-rDNA copy number strain. Set (1) was enriched for ribosomal components, for example 110 of the 420 genes in this set were in the “cytosolic ribosome” category (GO:0022626; enrichment P-adj < 1e−95). Set (2) was enriched for proteolysis-related genes, for example 60 of the 288 genes in this set were in the “proteolysis involved in protein catabolic process” category (GO:0051603; enrichment P-adj < 1e−22). Set (3) was enriched for chromatin-related genes, for example 8 of the 88 genes in this set were in the “structural constituent of chromatin” category (GO:0030527; enrichment P-adj < 1e−10). Set (4) was enriched for stress-related genes, for example 36 of the 83 genes in this set were in the “response to stress” category (GO:0006950; enrichment P-adj < 1e−7). We colored each differentially expressed gene that was a member of any of these 4 enriched GO categories, regardless of whether the gene is in the set enriched for that GO category.

Spot assays

Cells were grown to log-phase, diluted in sterile water in 3-fold dilutions starting with a cell concentration of 4 × 10⁵ cells/mL. 2.5 μL was spotted onto YEPD (1% yeast extract, 2% peptone, and 2% glucose), YEPD + 0.7 M Sodium chloride, YEPD + 3% ethanol, YEPD + 6% ethanol, and YEPG plates. All plates were scanned after 48 h of growth at 30°C. Thermotolerance assays were adapted from Lindquist and Kim (1996).

Petite frequency assays

The procedure adapted from Dimitrov et al. (2009) was used (Dimitrov et al. 2009). Medium-sized colonies from direct-from-freezer-stock streakouts on YEPD plates were inoculated 2 mL culture tubes filled with WF-N media. WF-N is a minimal media containing WF base salts and lacking nitrogen (Wickerham 1946). A portion of 2 mL culture was used to determine cell densities of cultures and diluted to 1 × 10⁵ cells/mL. Cultures were diluted again to a cell concertation of 2.5 × 10³ cells/mL. Diluted cultures were sonicated and 150 μL was plated onto YEPDG (1% yeast extract, 2% peptone, 0.1% glucose, and 3% glycerol) plates and allowed to grow at 30°C for 5 days. After the incubation period, all plates were scored for petite and grande colonies.

Results

rDNA copy number affects yeast fitness in standard growth conditions

To generate a set of isogenic S. cerevisiae strains with variable rDNA copy numbers, we started with a fob1Δ strain that stably maintains 35 copies in a S288c strain background (Kobayashi et al. 1998; Ide et al. 2010; Kwan et al. 2023). After a transient re-introduction of FOB1 to allow for copy number expansion, we obtained multiple clones for strains with copy numbers ranging from 45 to 180. A strain with 200 rDNA copies was isolated by chance during a cross. We verified rDNA copy number for 3 clones of each strain by CHEF gel electrophoresis and Southern blotting (Fig. 1a, b; Supplementary Fig. 1b, c) (Tsuchiyama et al. 2013; Kwan et al. 2016; Morton et al. 2020).

Strains with copy numbers between 35 and 200 showed similar growth rates (Supplementary Fig. 1a), consistent with previous observations (Jiang et al. 2024). To magnify small phenotypic differences between strains (Conti et al. 2022), we performed competition assays to assess the relative fitness of strains with varying rDNA copy numbers. The competitor strain for each experiment was a GFP-tagged strain with 180 rDNA copies that was otherwise isogenic. To determine the competitive fitness of each strain against the GFP 180 rDNA copy number strain, we mixed the 2 starting cultures in triplicate in a 1:1 ratio and grew them to saturation (Gresham and Dunham 2014). Every 24 h, we took samples to measure the ratio of GFP-positive to GFP-negative cells by flow cytometry. We then diluted the competition cultures 1:1,000 into fresh media and grew them again to saturation, allowing for ∼10 generations of growth. We repeated this process for ∼50 generations, and estimated fitness by tracking the ratios of the GFP competitor strain to each test strain over the indicated number of cell divisions. The comparison of the competitor GFP 180 rDNA copy number strain with the 180 rDNA copy number strain yielded an estimate for the fitness cost of GFP expression of 0.08% per generation (Fig. 1c, Supplementary Tables 1 and 2, Supplementary File 1).

We noticed 3 trends in the fitness data: (1) reduced fitness in strains with fewer than 98 rDNA copies; (2) a fitness plateau for strains with 98 to 160 rDNA copies; and (3) subtly reduced fitness in strains with more than 160 rDNA copies. The strain with the fewest rDNA copies (35 copies) showed the greatest fitness defect (Fig. 1c). Strain fitness appeared to increase gradually with steep gains as rDNA copies increased from 35 to 55, more moderate fitness gains from 55 to 95 copies, and steep fitness gains from 95 to 104 copies. There was an extended fitness plateau beginning around 100 copies. The fitness plateau, encompassing strains with 98 to 160 rDNA copies (Fig. 1c), mirrored the distribution of rDNA copy number genotypes in wild S. cerevisiae isolates (Fig. 1c, inset) (Hall et al. 2022). Additional rDNA copies above 160 resulted in moderately reduced fitness. A strain with 180 rDNA copies showed reduced fitness compared with the strains with 98 to 160 rDNA copies, but greater fitness than the strains with fewer than 98 rDNA copies (Fig. 1c). Strains with 200 rDNA copies also showed moderately reduced fitness compared with the strains with 98 to 160 rDNA copies (Fig. 1c). We conclude that rDNA copy number variants with fewer than 98 and more than 160 rDNA copies in the S288c strain background show reduced fitness under standard growth conditions. Strain fitness coincided with the lower boundary of the natural variation range across S. cerevisiae strains.

rDNA expansion upon FOB1 re-introduction into a short rDNA strain slows upon reaching ∼100 rDNA copies

Strains with substantially reduced rDNA arrays increase their rDNA copy numbers back to wild-type levels when FOB1 is re-introduced (Kobayashi et al. 1998). The re-introduction of FOB1 and passaging of cells to an rDNA copy number equilibrium enables another, possibly more nuanced analysis of the fitness effects of rDNA copy number than in the described competition experiments. Instead of competing 2 strains with fixed rDNA copy numbers, the FOB1 re-introduction produces variation in rDNA copy number among the cells in the culture and allows for selection among different rDNA copy number variants. Moreover, assessing fitness of wild-type FOB1 cells avoids any possible effects of the fob1Δ mutation. We hypothesized that the expansion of rDNA arrays upon FOB1 re-introduction would slow when cells reached ∼100 copies in standard growth conditions, consistent with fitness being the primary driver for rDNA expansion. Alternatively, rDNA copy number might expand to the 150–180 rDNA copies reported for this strain background (Ide et al. 2013; Kwan et al. 2016; Morton et al. 2020), supporting a model in which strain-specific rDNA copy number is determined by a counting mechanism (Iida and Kobayashi 2019a, 2019b).

To conduct this experiment, we reintroduced FOB1 by crossing a 35 rDNA fob1Δ strain with a 170 rDNA FOB1 strain, picked three 35 rDNA FOB1 spores, and passaged the resulting cultures (and the 2 parental strains) for 300 generations. Samples were collected at regular intervals (Supplementary Table 3) for subsequent CHEF gel electrophoresis and Southern blotting to determine rDNA copy number (Fig. 2a, b). As evident in Fig. 2, the control strain with a short rDNA array and the fob1 deletion showed a copy number of ∼26 rather than ∼35 at the start of this experiment. After ∼30 generations, the 3 spore-derived cultures showed an average of ∼56 rDNA copies (Fig. 2c, Supplementary Table 3). After 150 generations, they had ∼100 rDNA copies, and ∼107 rDNA copies after 300 generations, suggesting that rDNA copy number expansion slows considerably once ∼100 rDNA copies are reached. As expected, the controls 26 rDNA fob1Δ and 175 rDNA FOB1 changed little in their respective rDNA copy numbers throughout this experiment (Fig. 2c, Supplementary Table 3). We conclude that there is no significant fitness advantage in having more than ∼100 rDNA copies in standard growth conditions, consistent with our observation of a sharp fitness increase associated with this approximate copy number.

Fig. 2.

rDNA array expansion slows upon reaching ∼100 rDNA copies. Examination of chromosome XII size by a) CHEF gel electrophoresis and b) Southern blotting. c) Results of passaging experiment over 300 generations with samples collected at 30, 60, and 90 generations (Supplementary Table 3). After ∼30 generations, the 3 spore-derived cultures showed an average of ∼56 rDNA copies. The strains 26 rDNA fob1Δ and 170 rDNA FOB1 are included as controls. Copy number estimates at the beginning and end of the experiment for each genotype are indicated.

A strain with 35 rDNA copies shows altered gene expression in late S-phase

Strains with substantial reduction of rDNA copy number (20–40 copies) show increased sensitivity to mutagens, defects in genome replication, and cell cycle misregulation (Ide et al. 2010; Kwan et al. 2023). Although these phenotypic effects might suffice to cause the observed fitness defects, we used RNA-seq analysis to discover other biological processes that might be altered between a short (35 copies) and a control (180 copies) rDNA strain. Because low rDNA copy number causes genome replication delays (Kwan et al. 2023), we collected cells in late S-phase (40 min after synchronization) and in asynchronous, logarithmically-growing cells. We identified only 6 genes that were DEGs between the 35 and the 180 rDNA strains in asynchronous cultures (P-adj < 0.01), and 708 genes that were differentially expressed between these strains in late S-phase, with 3 genes overlapping between the 2 sets (Fig. 3, Supplementary Table 4). Of these 3, PRP11, implicated in splicing, and STR3, implicated in peroxisome function, were expressed at lower levels in the 35 rDNA strain. The third gene, HSP12, was expressed at higher levels in the 35 rDNA strain. This gene is induced in response to many stressful conditions, including heat stress, ethanol and salt stress, in addition to DNA replication stress. The small number of genes with expression differences between the 35 and the 180 rDNA strain in asynchronous culture is consistent with previous observations that strains with these rDNA copy numbers do not differ significantly in growth rates, rRNA expression, or ribosome function (Mohan and Ritossa 1970; French et al. 2003; Kim et al. 2006; Ide et al. 2010; Kwan et al. 2023).

Fig. 3.

Transcriptome analysis reveals expression changes specific to late S phase. Volcano plot showing genes with significantly different expression (P-adj < 0.01) between the 35 rDNA copy number strain and the 180 rDNA copy number strain in late S phase. The 3 genes (HSP12, PRP11, STR3) that were also differentially expressed in asynchronous growth conditions are represented by open black circles; HSP12 is also labeled in red because of its function in the stress response. Genes within select GO categories are indicated with colored dots: blue for cytosolic ribosome (GO:0022626), yellow for proteolysis involved in protein catabolic process (GO:0051603), and red for response to stress (GO:0006950). Many canonical heat stress proteins and their co-chaperones were significantly downregulated in the 35 rDNA copy number strain in late S phase (labeled in red). 1.5-fold change in expression is denoted by a dashed gray line.

However, in late S-phase, genes upregulated in the 35 rDNA strain were enriched for GO terms related to ribosomal structure and assembly. For example, 115 of the 420 upregulated genes in the 35 rDNA strain belonged to the “cytosolic ribosome” category. In contrast, the 288 genes downregulated in the 35 rDNA copy strain were enriched for GO categories related to protein degradation (e.g. proteolysis, response to stress, and proteasome complex).

To gain further insight into the significance of these S-phase-specific expression changes, we focused on genes with at least a 1.5-fold change in expression. We found 88 genes upregulated in the 35 rDNA strain (adj-P value < 1e−6), with GO term enrichments for structural constituent of chromatin, protein heterodimerization activity, and nucleosome. The top 10 most upregulated genes were HSP12; 5 genes encoding small nucleolar RNAs involved in rDNA processing; PCNA, a gene encoding a sliding replication clamp; DDR2, a multistress response gene named for its function in the DNA damage response; SOE1, a tRNA gene and suppressor of CDC8, a gene encoding a kinase functioning in the biosynthesis of deoxyribonucleotides; and SIP18, a gene involved in salt stress.

There were 83 downregulated genes in the 35 rDNA strain (adj-P value < 1e−7), with GO term enrichments for protein folding, response to stress, and MCM helicase complex. The latter is consistent with the genome replication defects in the 35 rDNA strain: MCM helicase is essential for DNA replication initiation and elongation and is recruited to the origins of DNA replication as part of the prereplicative complex (Bell and Labib 2016). The top 10 most strongly downregulated genes included SSA1 and SSA2, which encode the 2 Hsp70s; SIS1, which encodes a co-chaperone of Hsp40; STI1, which encodes a co-chaperone of Hsp90; and HSP104, which encodes a disaggregase functioning with Hsp70 and Hsp40 required for the acquisition of thermotolerance (Lindquist and Kim 1996). The other genes induced as part of the canonical heat stress response, the 2 genes encoding Hsp90 (Hsc82 and Hsp82) were the 13 and 15th most strongly downregulated genes, respectively. While some of the DEGs and GO annotations are consistent with prior studies (Ide et al. 2007, 2010; Salim et al. 2017; Kwan et al. 2023), the unexpected S-phase-specific finding of altered chaperone expression warranted further investigation as to whether rDNA copy number variation might play a role in response to stress or altered growth conditions.

rDNA copy number variants show altered fitness at 37°C and in glycerol media

To determine whether rDNA copy number variation might cause phenotypic effects in response to heat stress, we performed competition experiments at 37°C. Results differed markedly from those observed in standard growth conditions. Specifically, the gradual, statistically significant fitness increase for strains with up to 95 copies in standard conditions gave way to a plateau of low fitness for strains with copy numbers ranging from 45 to 95 (Fig. 4a, c, Supplementary Table 5). Although the 35 rDNA copy number strain showed by far the greatest fitness defect among the rDNA copy number variants, compared with its fitness in standard laboratory conditions, it was less affected by heat stress (Fig. 4a, c, Supplementary Table 5). Strains with higher copy numbers, from 80 to 95 copies, showed decreased fitness in response to heat stress compared with their performance in standard growth conditions (Fig. 4a, c, Supplementary Table 5) with the fitness threshold remaining at approximately 100 copies; however, the coefficients of selection remained similar between 30 and 37°C (Fig. 4c). Strains with 180 and 200 rDNA copies showed somewhat greater fitness in response to heat stress than in standard growth conditions (Fig. 4a, c, Supplementary Table 5). To rule out the possibility that heat stress altered rDNA copy numbers during the course of these experiments, we examined CHEF gels and Southern blots of all strains at the start of competitions (day 0) and at the end of competitions (day 5). rDNA copy numbers did not change during the heat stress competitions (Supplementary Fig. 3).

To determine whether shifts in fitness among the rDNA copy number variants were specific to heat stress or common across other nonstandard growth conditions, we performed competition experiments at 30°C in the nonfermentable carbon source glycerol. Results from these competitions differed from those obtained in standard laboratory condition and in response to heat stress (Fig. 4b). Although strong batch effects were present for low and high copy number strains, strains carrying 80 and 100 copies reproducibly showed severe fitness defects in glycerol, suggesting that the previously observed fitness plateau shifted in glycerol to above ∼100 copies. The plateau of high fitness extended to strains with the longer rDNA arrays of 160 to 200 copies. We confirmed that rDNA copy numbers did not change during the competitions (Supplementary Fig. 4).

To probe the extent of the phenotypic consequences of rDNA copy number variation with a less sensitive assay, we conducted spot assays examining responses to heat stress, osmotic stress, and ethanol as well as the ability to acquire thermotolerance. No differences were found among the tested strains (Supplementary Fig. 2). Nevertheless, the results of our competition experiments strongly argue that rDNA copy number variation subtly but significantly contributes to a strain's fitness in response to different environments. Yeast cells face variable environments throughout their culturing in the laboratory (exponential growth, stationary phase, different carbon sources, freeze-thaw, etc.) and outside of it. Thus, the excess of rDNA copies not required for ribosome biogenesis in standard growth conditions may buffer condition-specific requirements for ribosome biogenesis and/or genome replication, the 2 processes previously identified as affected by rDNA copy number variation.

A strain with substantially reduced rDNA copy number generates fewer petites than a control strain

Intrigued by the profound shift in fitness in response to glycerol, a nonfermentable carbon source, we examined mitochondrial function in the 35 rDNA copy number strain compared with the control 180 rDNA copy number strain. Specifically, we assessed mitochondrial function and mitochondrial genome stability by measuring the frequency of petite formation (Dimitrov et al. 2009). Petites are yeast colonies that have lost mitochondrial respiration function, and therefore are much smaller than respiration-competent, grande colonies on specialized glycerol media (Fig. 5a). We performed petite frequency assays by selecting 15 medium-sized colonies from both strain backgrounds (35 rDNA copies, 180 rDNA copies), diluting them and plating ∼350 cells onto glycerol-containing plates. After incubation for 5 days at 30°C, we counted all petite and grande colonies to determine the petite percentage for both strains.

Fig. 5.

A strain with substantially reduced rDNA copy number generates fewer petites than a wild-type strain. a) Representative image of petite and grande yeast cells. b) The 35 rDNA copy number strain generated significantly fewer petite colonies than the 180 rDNA copy number strain in 2 biological replicates (batches). For each batch, the number of plates and the number of colonies that were scored are indicated (Supplementary Table 6). Significance for each biological replicate is indicated (t-tests).

The 35 rDNA copy number strain produced significantly fewer petite colonies (48.2 and 43.1% in the 2 replicates) than the 180 rDNA copy number strain (57.5 and 57.8%, respectively) (Fig. 5b, Supplementary Table 6). Although this result appears inconsistent with the 35 rDNA copy number strain showing lower fitness than a 180 rDNA copy number variant in the glycerol competition experiments, it is likely that rDNA copy number variation affects these 2 traits in different ways.

Discussion

Here, we generated a set of rDNA copy number variants in otherwise isogenic S. cerevisiae strains to study the fitness consequences of rDNA copy number variation below and within the naturally occurring range of variation. In standard laboratory conditions, strains with the highest fitness had 98 to 160 rDNA copies, a range that overlaps with the distribution found in wild yeast isolates (Morton et al. 2020; Hall et al. 2022). Strains with rDNA copy numbers either higher or lower than that range showed reduced fitness, with extreme variants, especially those with the lowest copy numbers, exhibiting the most severe fitness defects. However, rDNA copy number-dependent fitness was not static: the fitness of rDNA copy variants shifted in response to growth at increased temperature or in glycerol. These shifts in fitness were not due to shifts in ploidy; changes in ploidy have been observed under stress conditions previously (Harari et al. 2018). We conclude that rDNA copy number variation plays a causal role in the response of yeast cells to different environments and that this variation should be considered when comparing strain responses across experimental growth conditions.

Upon re-introduction of FOB1, strains with 35 rDNA copies expanded their rDNA arrays to ∼100 rDNA copies but further expansion slowed, consistent with a lack of fitness benefits with further expansion. Of the haploid wild yeast strains previously examined, 75% have rDNA copy numbers within the 98–160 fitness plateau (Morton et al. 2020; Hall et al. 2022). The remaining 25% of strains with higher rDNA copy numbers may experience more diverse environmental conditions than the other strains or carry genetic variants associated with rDNA copy expansion. Indeed, previous studies hint at a connection between rDNA copy number variation and genetic background. Kobayashi et al. report that a short rDNA strain fully expands its array to 150 copies after 150 generations (Kobayashi et al. 1998), far more rDNA copies than the ∼100 we observed after 300 generations. The two studies differ in strain background: the strains in Kobayashi et al. 1998 were derived from W303 (Nogi et al. 1991; Yano and Nomura 1991) while our strains were derived from S288c. Although they are closely related laboratory strains, W303 and S288c differ in ∼7,000 nonsynonymous polymorphisms (Liti et al. 2009; Matheson et al. 2017; Peter et al. 2018) and strains in the W303 background show higher rDNA copy numbers (250–300 copies) than those derived from S288c (140–150 copies) (Michel et al. 2005; Kwan et al. 2016; Lynch et al. 2019; Morton et al. 2020). The W303 strain might harbor genetic variants that facilitate rDNA expansion to higher copy numbers than in S288c strains, such as nonsynonymous SIR2 variants (Jack et al. 2015). Identifying these variants should further understanding of the regulation of rDNA copy number across all S. cerevisiae strains, and possibly in metazoans.

We show that environmental stress alters the relationship between rDNA copy number and fitness. The fitness plateau shifted to include higher rDNA copy numbers, and in glycerol, it reproducibly excluded strains with ∼100 copies that showed high fitness in standard growth conditions. Second, the gradual fitness increase up to 95 copies found in standard laboratory conditions gave way to a worse performance of rDNA copy number variants below the naturally occurring range. Thus, loss of rDNA copies, even by only a few copies, produced strong deleterious effects in adverse conditions, and copy number at the higher end of the naturally occurring copy number range appears to buffer yeast cells against fluctuations in environment.

This interpretation is consistent with our observation that the S288c background slows rDNA array expansion at ∼100 copies and shows high fitness at this copy number, yet the S288c strain typically contains ∼150 rDNA copies (Ide et al. 2013; Kwan et al. 2016; Morton et al. 2023). Since the fitness plateau shifted to higher copy numbers in response to stress, the strain may have been selected under laboratory conditions for 150 rDNA copies rather than ∼100 copies through exposure to a myriad of intermittent stresses: e.g. cold/freezing temperatures, heat shock, reduced nutrient availability (Kwan et al. 2016). We speculate that conducting the expansion experiment in fluctuating environments might have resulted in rDNA arrays of ∼150 copies or higher; however, it is nontrivial to design such an experiment to be informative and artifact-free.

At higher rDNA copy numbers, such as 180 and 200, fitness can decrease in standard growth conditions, which may reflect the more sparse representation of wild yeast isolates with higher rDNA copy numbers. The somewhat lower fitness of the 180 and 200 rDNA copy number strains seems inconsistent with a previous model that implicated the excess of rDNA copies over the number required for ribosome biogenesis in maintaining rDNA and genome stability (Ide et al. 2010). If this were the case, the additional rDNA copies would be expected to increase fitness in standard growth conditions rather than reduce it. A possible mechanism explaining the reduced fitness in strains with increased rDNA copy number is the cost of maintaining the additional copies, especially since an extra 25 rDNA copies (∼227 kb) represent a similar amount of DNA as the smaller S. cerevisiae chromosomes I (230 kb) and III (316 kb) (Cherry et al. 2012). Dedicating resources to maintaining an additional chromosome's worth of repetitive, presumably silenced, DNA could plausibly reduce fitness (Sunshine et al. 2015). Further studies with strains containing a wider range of rDNA copy numbers above the natural range are needed to distinguish among possible fitness effects, although such strains are not easy to construct.

Although the 35 rDNA copy number strain showed lower fitness than a 180 rDNA copy number variant in the glycerol competition experiments, the strain produced significantly fewer petites, suggesting that rDNA copy number variation affects yeast fitness and mitochondrial maintenance in different ways. Ours is not the first study to report an association of rDNA copy number with mitochondria: Gibbons et al. found an inverse association of rDNA copy number and mitochondrial abundance in humans (Gibbons et al. 2014). In yeast, mitochondrial abundance is dynamically regulated through the evolutionarily conserved Mec1/Rad53 intra-S-phase checkpoint pathway, which results in increased de novo synthesis of dNTPs and increased mitochondrial abundance (Taylor et al. 2005). Specifically, deletion of RRM3, a DNA helicase involved in rDNA replication (Ivessa et al. 2000), results in an increase in mitochondrial DNA in yeast (Taylor et al. 2005). As we have shown previously, the 35 rDNA copy number strain has replication defects (Kwan et al. 2023). This replication defects together with a likely increased supply of dNTPs due to the radically shorter size of the rDNA array may plausibly account for the observed result. In addition, the low number of extrachromosomal copies of rDNA in fob1 mutants (Defossez et al. 1999) may contribute to our finding. Previous studies found that petites activate RNA polymerase II transcription of rDNA largely from extrachromosomal copies of rDNA (Conrad-Webb and Butow 1995). This switch from RNA polymerase I to RNA polymerase II might be less likely in the petites of a short rDNA strain.

The phenotypic consequences of rDNA copy number variation have been increasingly studied in multicellular animals and plants (Xu et al. 2017; Hall et al. 2022; Kasselimi et al. 2022; Morton et al. 2023). In the nematode C. elegans, individual animals with rDNA copy numbers below the naturally occurring range display a gradient of developmental defects, ranging from subtle developmental delays to developmental arrest and strikingly variable morphological defects in postembryonic development (Cenik et al. 2019; Morton et al. 2023). These observations suggest that specific stages of development and specific tissues have their own rDNA copy number requirements, perhaps because each stage and tissue represent different cellular environments. That rDNA copy number-dependent fitness shifts in yeast with environmental conditions may be analogous; although yeast cells do not form specific tissues, they experience a wide range of environmental conditions.

The shifts in environmental response of cells with different rDNA copy numbers are particularly intriguing in the context of cancer. Reduction of rDNA copy number has been reported in mTOR-related tumors compared with healthy, noncancerous tissue (Xu et al. 2017; Kasselimi et al. 2022). Because rDNA copy number reduction causes defects in genome replication and cell cycle control in yeast (Ide et al. 2010; Kwan et al. 2023), rDNA copy number reduction in human cells may contribute to the characteristic loss of genome stability that precedes cancer pathogenesis (Hanahan and Weinberg 2011). Moreover, the reduction in rDNA copy number might shift the fitness of precancerous and/or cancer cells in response to the stresses of host environment or treatment, analogous to our observations in yeast.

Data availability

Transcriptome raw sequencing reads can be found at the NCBI short read archive under the BioProject PRJNA1061514. Supplementary File 1 is available on GSA FigShare at https://doi.org/10.25386/genetics.28327211.

Supplemental material available at GENETICS online.

Funding

This work was supported by the following funding sources: NIGMS R01 GM122088 and R35 GM139532 and NHGRI grant RM1 HG010461 to C.Q., and NIGMS R35 GM122497 to B.J.B. Flow cytometry reported in this work was performed at the DLMP Flow Cytometry Core at the University of Washington.