Profiling single chromosome strain growth
To characterize perturbations in growth and/or metabolism, we performed batch fermentations with triplicate bioreactors for the reference strain (S. cerevisiae strain BY4742) or the chromosomal fusion strain SY14, which has a single large chromosome instead of sixteen distinct chromosomes (Fig. 1). Analysis of the CO2 evolution rate of the gas emitted from fermenters suggested that SY14 had an increased lag time prior to exponential growth on glucose (Fig.1a). In addition, the doubling time during growth on glucose was increased by 8% for cultures of SY14 (120 min) compared with BY4742 (111 min) (Fig. 1c). Biomass accumulation, monitored via OD600, was diminished and became more apparent in the later stages of growth, culminating in 28% less biomass after 48 h of growth (Fig. 1b, 1d). Shake flask experiments showed a similar decrease in final biomass after 10 days (Additional File 1: Fig. S1). Despite a longer lag phase, SY14 cultures exhibited similar profiles of carbon source uptake, including complete uptake of glucose and production, followed by consumption of ethanol (Fig. 1e,f,g,h,i). Together, these findings indicate that SY14 exhibited a delay in growth after inoculation, had decreased glucose phase growth, and accumulated less biomass than the reference strain.
The single chromosome strain (SY14) exhibits impaired growth on non-fermentable carbon sources and is sensitive to ethanol
The results presented in Fig. 1 warranted further analysis of the various stages of growth for SY14 and the reference strain. Similar to the fermentation results, microplate growth assays showed that cultures of SY14 exhibited a longer lag phase, increased doubling time during growth on glucose, and lower final biomass yield than BY4742 (Fig. 2a). Notably, the magnitude of these differences was larger for lag phase and final biomass than for glucose doubling time in microplate assays and fermenters. These differences suggested that SY14 cultures might struggle to grow on, and emerge from growth on non-fermentable carbon sources. To test this, we plated strains on glucose (YPD), ethanol (YPE), and glycerol (YPGly) plates (Fig. 2b). The results showed that growth of SY14 was diminished compared to BY4742 on non-fermentable carbon sources, but was similar on glucose. These phenotypes did not appear to be due to oxidative stress that might occur during growth on non-fermentable carbon sources, as addition of 3mM H2O2 did not disproportionately influence SY14 doubling time or lag phase duration (Additional File 1: Fig. S2). Further, total protein levels (Additional File 1: Fig. S3a), as well as ribosomal RNA expression and processing were similar in wildtype and SY14 strains (Additional File 1: Fig. S3b).
The results in Fig. 2b showed that growth for SY14 was particularly diminished in the presence of 6% ethanol. To test for ethanol sensitivity, we cultured SY14 and BY4742 in YPD (glucose) media +/- 5% ethanol (Fig. 2c). The lag phase after inoculation was longer for SY14 with 5% ethanol (Fig. 2d), and the doubling time during growth on glucose increased by 55% for BY4742 and 100% for SY14 (Fig. 2e). These findings suggest that SY14 is sensitive to ethanol, even in the presence of glucose. This sensitivity may influence the observed increase in cell death in the SY14 background (Fig. 2f).
SY14 exhibits decreased expression of diauxic shift related genes in the ethanol phase
Analysis of transcriptomic measurements during growth on glucose discovered relatively few differentially expressed genes in the SY14 background (53 genes) (Additional File 1: Fig. S4a, Additional File 2). The number of differentially expressed genes is intriguing as chromosomal fusion drastically altered genome arrangement and disrupted many interchromosomal interactions, which are important for gene regulation in higher eukaryotes17–19. Furthermore, chromosomal fusion removed the majority of telomeres and centromeres which have previously been shown to influence gene silencing13,20. However, these gene expression results may not encapsulate the deficiencies of the strain during growth on non-fermentable carbon sources or in the presence of ethanol (Fig. 2). To further understand these phenotypes, we performed RNAseq to compare gene expression between the reference (BY4742) and single chromosome (SY14) strains during growth on ethanol following a glucose batch phase and the diauxic shift (Fig. 3a). This analysis resulted in identification of a modest number of differentially expressed genes (109). Interestingly, genes with significantly lower expression in SY14 were enriched for functions related to growth on non-fermentable carbon sources (Fig. 3b). Specifically, SY14 exhibited lower gene expression for enzymes involved in ethanol, carnitine, propionate, and fatty acid metabolism (Fig. 3c,d,e,f), all of which enable S. cerevisiae to generate ATP after glucose depletion. The diminished gene expression observed might predict diminished growth post diauxic shift, which was observed in Fig. 2b.
The diminished activation of the aforementioned genes was not associated with changes in sequence of the ORF, promoters, or terminators, with one exception, CIT3, which encodes a mitochondrial citrate and methylcitrate synthase. The promoter and 5’ coding region of CIT3, a gene known to be involved in propionate metabolism21, was removed during construction of the SY14 strain. Reintroduction of CIT3 via a URA3-marked 2µm plasmid did not alter the growth or ethanol tolerance of SY14 (Additional File 1: Fig. S5), which might be expected as this gene was shown to encode a minor isoform of citrate synthase21. Further analysis of the RNAseq data identified several subtelomeric genes that were not expressed in the single chromosome strain (e.g. HSP33, PAU4, and AAD4), analysis of the SY14 genome sequence showed that these genes were likely removed during chromosomal fusion (Additional File 1: Fig. S6). The majority of these deleted genes lacked a functional description (54%), or were members of the duplicated gene families PAU, COS, and AAD (24%). These genes were not amongst gene sets known to be essential22, associated with slow growth22, or known transcription factors23. Further, as a group, these deleted genes represented a small percentage of the total RNAseq reads in the wildtype strain during glucose (0.10%) or ethanol (0.16%) phase.
Next, we compared gene expression between glucose phase and ethanol phase for each strain to understand the transition between growth on different carbon sources. This analysis showed that some genes that were proximal to the remaining telomeres were upregulated during ethanol phase in the reference strain, but were not upregulated in the single chromosome strain (Additional File 1: Fig. S7), suggesting that SY14 has increased subtelomeric silencing. Further analysis showed that several metabolic genes that were not telomere proximal exhibited lower induction during ethanol phase in SY14 compared to wildtype strains. We refer to these genes as poorly induced, as they are significantly upregulated (log2FC>1 FDR<0.01) in the wildtype strain upon transition from glucose to ethanol phase, but were at least two-fold less induced in the SY14 background compared to the reference strain (Fig. 3g red dots). The 111 genes that were poorly induced in SY14 represented 3.56% of all RNAseq reads in ethanol phase samples, in contrast, these genes accounted for 7.1% of reads amongst reference samples. These data suggest that the significantly decreased ethanol phase expression of metabolic genes like ACS1, YAT1, YAT2, CIT2, PDC6, and ADH2 in Fig. 3c,d,e,f was due to a failure to upregulate these genes after the transition from glucose to non-fermentable carbon source growth in the SY14 background. The similarity of the functional annotations of the poorly induced genes in SY14 may indicate disruption of a global mechanism for regulating non-fermentable carbon source gene expression.
Metabolic modeling of SY14 predicts an ATP bottleneck during ethanol growth
Our transcriptomic and genomic analyses showed that 248 enzyme-encoding genes exhibited altered expression (FDR<0.05 compared to reference) in the SY14 background during growth on ethanol. These genes and/or the subtelomeric genes that were disrupted during chromosomal fusion (Additional File 1: Fig. 6), might explain the growth phenotype of the SY14 strain. To further understand how these changes in expression and deletions might influence glucose and non-fermentable carbon source growth, we constructed enzyme-constrained Genome-scale Metabolic Models (ecGEM) for both reference and SY14 strains24, using the magnitude of gene expression changes to constrain enzyme usage in SY14 relative to the reference. As SY14 and reference cells exhibited remarkably similar metabolic profiles (Fig. 1) and gene expression profiles (Additional File 1: Fig. S4) in glucose-phase growth, the minor growth defect of SY14 pointed to an increased ATP expenditure for non-growth associated maintenance (NGAM), indicating that resources were being diverted to deal with stress (Fig. 4a).
In contrast, modeling growth on ethanol as the carbon source showed that differentially expressed metabolic enzymes drastically limited the ability of SY14 to grow (Fig. 4b) and utilize ethanol (Fig. 4c). Of note, the calculated maximum ATP expenditure on NGAM was comparable between wildtype and SY14 during growth on ethanol (Fig. 4d), indicating that a bottleneck in ATP generation from ethanol underlies the reduction in biomass formation for SY14. Together, this analysis suggested that when using glucose as a carbon source, the growth defect in SY14 cells arose from an increased ATP expenditure to handle stress (Fig. 4e). Conversely, the model predicts that reduced cell growth on ethanol was a result of a disruption in metabolism leading to a reduced capacity to generate energy (Fig. 4f). In silico rescue experiments identified 70 of the 248 perturbed enzyme-encoding genes as candidates that could rescue the ethanol-phase growth defect of SY14 (Additional File 3). Some of these genes (7/70) were deleted during chromosome fusion and were members of multicopy gene families whose individual contributions to metabolism are unclear. The remaining genes (63/70) were downregulated metabolic enzyme-encoding genes whose coding sequences were not perturbed in SY14, like ACS1, PDH1, and YAT1. The 70 rescue candidate genes were enriched amongst GO-slim terms related energy generation, including lipid metabolism, nucleotide metabolism, and carbohydrate metabolism (Fig. 4g), consistent with our model of SY14 showing a growth defect using ethanol as the carbon source in Fig. 4f.
SY14 growth on ethanol is rescued by the sirtuin-family deacetylase inhibitor nicotinamide
Fig. 3 shows diminished induction of several genes involved in metabolic processes that are expected to be upregulated after the diauxic shift. These poorly induced genes were distributed throughout the genome in the single chromosome strain (Fig. 5a), which suggested that the change in expression was not restricted to a single locus. One possibility is that these widespread changes could occur due to a disturbance in chromatin silencing complex activity. During SY14 construction, ~300 kb of subtelomeric DNA was removed to facilitate chromosomal end-to-end fusions (Additional File 1: Fig. S6b). These subtelomeric regions represent ~2.6% of the genome, but in part due to sirtuin family histone deacetylase mediated silencing25, these regions represent only 0.1-0.2% of total transcripts in our reference data (Additional File 1: Fig. S6b,S6c). Further, sirtuins have been shown to downregulate widespread genes when subtelomeric structures are disrupted in S. cerevisiae20. To test whether diminished ethanol growth for SY14 was influenced by sirtuins, growth assays were performed in the presence of nicotinamide (NAM), an inhibitor of sirtuin family deacetylases25. Growth on ethanol for SY14 was rescued by NAM (Fig. 5b), indicating that the enzymes that are necessary to efficiently catabolize ethanol are encoded in the SY14 genome, but that these genes, or their activators, may be repressed by sirtuins. This finding is congruent with the poor induction of genes like CIT2, ACS1, and PDC6 in the ethanol phase for SY14 compared to the reference strain (Fig. 3).
The above findings suggest that sirtuin-mediated silencing impedes growth on ethanol in the SY14 background. Next, the ethanol sensitivity of an independently constructed chromosomal fusion strain that has two large chromosomes (yJL402) was assessed14. Unlike SY14, the yJL402 growth was similar to the reference strain with ethanol as the sole carbon source (Fig. 5b). These data show that S. cerevisiae can thrive on ethanol with major changes to chromosome size, subtelomeric DNA content, and a large number of deleted subtelomeric ORFs. Specifically, yJL402 harbors two very large chromosomes (6Mb each) instead of sixteen, lacks 200kb of subtelomeric DNA, and 70 subtelomeric ORFs compared to the reference strain. In contrast, SY14 harbors a single chromosome (12Mb), lacks 300kb of subtelomeric DNA, and 107 subtelomeric ORFs (Additional File 1: Fig. S6b). Fifty-seven ORFs are absent in both SY14 and yJL402, including multiple members of subtelomeric gene families, such as Aryl Alcohol Dehydrogenases (AAD), Conserved Sequence (COS) genes, seripauperin (PAU) genes, and YRF family helicases (Additional File 4). In each case, SY14 lacks more members of each of these families compared to yJL402. Another possibility is that one or more of the 51 ORF deletions unique to SY14 cause diminished ethanol growth. Of these genes, only THI12, YEL073C, AAD14, THI13, YGL258W-A, SNZ2, YGL262W, and COS6 were expressed at >1 mRNA copy per cell (>25 transcripts per million) during the ethanol phase in our reference dataset (Additional File 4).