Strand-specific RNA sequencing reveals extensive regulated long antisense transcripts that are conserved across yeast species
- Moran Yassour1, 2, 3,
- Jenna Pfiffner†1,
- Joshua Z Levin†1,
- Xian Adiconis1,
- Andreas Gnirke1,
- Chad Nusbaum1,
- Dawn-Anne Thompson1Email author,
- Nir Friedman3, 4Email author and
- Aviv Regev1, 2Email author
© Yassour et al; licensee BioMed Central Ltd. 2010
Received: 15 February 2010
Accepted: 26 August 2010
Published: 26 August 2010
Recent studies in budding yeast have shown that antisense transcription occurs at many loci. However, the functional role of antisense transcripts has been demonstrated only in a few cases and it has been suggested that most antisense transcripts may result from promiscuous bi-directional transcription in a dense genome.
Here, we use strand-specific RNA sequencing to study anti-sense transcription in Saccharomyces cerevisiae. We detect 1,103 putative antisense transcripts expressed in mid-log phase growth, ranging from 39 short transcripts covering only the 3' UTR of sense genes to 145 long transcripts covering the entire sense open reading frame. Many of these antisense transcripts overlap sense genes that are repressed in mid-log phase and are important in stationary phase, stress response, or meiosis. We validate the differential regulation of 67 antisense transcripts and their sense targets in relevant conditions, including nutrient limitation and environmental stresses. Moreover, we show that several antisense transcripts and, in some cases, their differential expression have been conserved across five species of yeast spanning 150 million years of evolution. Divergence in the regulation of antisense transcripts to two respiratory genes coincides with the evolution of respiro-fermentation.
Our work provides support for a global and conserved role for antisense transcription in yeast gene regulation.
Antisense transcription plays an important role in gene regulation from bacteria to humans. While the role of antisense transcripts is increasingly studied in metazoans , less is known about its relevance for gene regulation in the yeast Saccharomyces cerevisiae, a key model for eukaryotic gene regulation. Recent genomic studies using tiling microarrays showed evidence of stable antisense transcription in S. cerevisiae [2, 3] and Schizosaccharomyces pombe [4, 5].
It is unclear how broad the role of antisense transcription is and what key functional processes in yeast it affects. A few functional antisense transcripts have been implicated in the control of several key genes, including the meiosis regulator gene IME4 , the phosphate metabolism gene PHO84 , the galactose metabolism gene GAL10 , and the inositol phosphate biosynthetic gene KCS1 . In contrast, genome-scale analysis in yeast suggested that antisense transcripts largely arise from bi-directional, possibly promiscuous, transcription from nucleosome free regions in promoters or 3′ UTRs of upstream protein coding genes [2, 3]. The ability to massively sequence cDNA libraries (RNA-seq) can facilitate the discovery of novel transcripts [10–12], but most studies have not distinguished the transcribed strand.
Here, we used massively parallel sequencing to sequence a strand-specific cDNA library from RNA isolated from S. cerevisiae cells at mid-log phase. We found 1,103 putative antisense transcripts in those cells, ranging from short ones that cover only the 3′ UTR of sense genes to over a hundred long ones that cover the entire sense ORF. Many of the putative sense targets encode proteins with important roles in stationary phase, stress responses, or meiosis. We validated the differential regulation of 67 antisense transcripts and their sense targets in conditions ranging from nutrient limitation to stress, and show that the exosome component Rrp6 affects their levels, but that the histone deacetylase Hda2 does not. Furthermore, for a few examples we show that antisense transcripts and their differential regulation are conserved over 150 million years across five yeast species. Our results support a potential conserved role for antisense transcription in yeast gene regulation.
Strand-specific RNA-seq of S. cerevisiaecells
To identify antisense transcripts in yeast, we used massively parallel sequencing (Illumina) to sequence a strand-specific cDNA library from S. cerevisiae during mid-log growth in rich media. The approach we used  relies on the incorporation of deoxy-UTP during the second strand synthesis, allowing subsequent selective destruction of this strand (Materials and methods). Our sequencing yielded 9.22 million 76-nucleotide paired-end reads that map to unique positions in the genome.
Of the reads that map to regions with a known annotation for uni-directional transcription (from the Saccharomyces Genome Database (SGD) ), only 0.62% were mapped to the opposite (antisense) strand, demonstrating the strand-specificity of our library  (Materials and methods). We next combined these reads to define consecutive regions of strand-specific transcription (Materials and methods), and found 8,778 units, covering 4,944 of the 5,501 (90%) genes expressed in this condition (top 85% ) at the correct orientation, for at least 80% of the length of each gene (Materials and methods; Additional files 1 and 2).
Identification of 1,103 antisense transcripts that vary in sense coverage from the 3′ UTR to the entire ORF
We found 1,103 putative units that have an antisense orientation relative to annotated transcripts and cover at least 25% of a known transcript on the opposite strand, using published UTR estimates  (Materials and methods; Additional file 1). While antisense reads are only a small minority (0.62%) of the total reads, they aggregate in a relatively small number of loci, with 62% of the antisense reads concentrated in the 1,103 units we defined. The remaining 38% are mostly isolated reads scattered across the genome (Figure S1 in Additional file 3).
The 402 antisense units are supported by several lines of evidence. First, comparing the units to published data from strand-specific tiling arrays , we find that 143 of our 402 units (36%) are at least 80% covered by stable antisense units as previously defined , while 224 units were not detected at all on tiling arrays (Additional file 1; Materials and methods). Finally, 336 of the 402 units are supported by an independent RNA-seq experiment we conducted using an RNA ligation protocol  for strand-specific library preparation (Materials and methods) . The lower number of units detected using the independent library reflects the less continuous nature of the data collected by the alternative protocol .
Antisense units are unlikely to result solely from leaky transcription
We next assessed the previously suggested possibility [2, 17] that antisense transcription is a consequence of leaky transcriptional regulation, through either unterminated transcription, bi-directional transcription initiation from promoters, or transcription from potential nucleosome-free regions (NFRs) in 3′ UTRs. We found that 48 and 27 units might reside within a long 3′ or 5′ UTR, respectively. Of the remaining 333 antisense units, 149 appear to share the (divergent) promoter of a known neighbor transcript, consistent with previous reports [2, 3]. An additional 43 units may be transcribed from potential NFRs in the 3′ UTR of an adjacent transcript . The remaining 141 units (35%) cannot be accounted for by transcription from a known promoter or 3′ UTR (when considering 400-bp margins; Figure S4 in Additional file 3).
We compared the change in expression of antisense units and such neighboring genes between cells grown in rich media containing glucose (yeast peptone dextrose (YPD)) and ethanol (yeast peptone ethanol (YPE)) as the main carbon source . We reasoned that 'leaky transcription' would result in strong positive correlation in expression between the antisense transcript and the neighboring gene. However, we found a very low correlation (R2 = 0.07; Figure S5 in Additional file 3), suggesting only weak co-regulation through leaky transcription, from divergent promoters or 3′ NFRs, if at all. Thus, even among the units that could hypothetically arise from leaky transcription, there is little if any evidence of such events.
We also examined the hypothesis that antisense is transcribed to prevent the neighboring gene from runthrough transcription. Of the 402 units, 72 (18%) end relatively close (< 200 bp) to the 3' ends of known genes (for example, Unit3689 ends close to the NOP10 gene shown in Figure 1a). On average, the 3′ UTRs of these 72 genes are shorter than those of other genes (P < 0.0058, Wilcoxon test; Figure S6 in Additional file 3). This minority of units could thus potentially play a role in curbing runthrough transcription.
Stress, meiosis and nutrient limitation genes are associated with antisense transcripts at mid-log phase
To explore the potential function of the antisense units, we examined the known function and expression pattern of their associated sense transcripts. We found that the set of ORFs with 75% or more antisense coverage is enriched for genes induced after the diauxic shift (P < 6 × 10-14) or in stationary phase (P < 2 × 10-10), during stress (P < 2 × 10-27), and in some meiosis and sporulation experiments (for example, 85 of 805 genes induced at 8 h in a sporulation time course, P < 3 × 10-6), and include multiple central genes in these processes. For example, the genes encoding the key meiosis proteins IME4, NDT80, REC102, GAS2, SPS19, SLZ1, RIM9, and SMK1 are all associated with long antisense transcription. This is consistent with previous studies in S. pombe  showing a preponderance of antisense transcription in genes induced during meiosis. Long antisense is also found in many key respiration and mitochondrial genes, including HAP3, COX8, CYB2, CYC3, COX5B, MMF1, NCA3, CYC1, MBR1, PET10, COX12, and ATP14. Genes from other processes repressed during mid-log phase are also associated with long antisense transcripts. Notably, these include at least five members of the PHO regulon (VTC1, PHO5, PHM8, ICS2, PHO3) and three genes from the GAL regulon (GAL4, GAL10, GAL2). This suggests that antisense regulation may be prevalent across these regulons rather than at single target genes (as found in [6–8]). Furthermore, the expression of 149 of the antisense transcripts is inversely related to that of their sense targets, as measured on tiling arrays  in several conditions (glucose versus ethanol, versus galactose, and in Δrrp6; Figure S7 in Additional file 3). Certain key genes that are highly expressed in mid-log phase are also associated with detectable transcription of long antisense units. These include some of the ribosomal protein genes (for example, RPS26A, RPS20), glycolytic enzymes (for example, CDC19, PGK1), and cell cycle regulators (for example, PCL2, APC11, ASK1). Nevertheless, these observations suggest that antisense transcription may be regulated in a condition-specific manner in S. cerevisiae and may be involved in the repression of stress, stationary phase and meiosis genes in rich growth conditions.
Differential regulation of antisense-sense pairs in nutrient limitation and stress
Finally, to test the generality of these suggestive patterns, we expanded the nCounter assay to measure the expression of 67 sense-antisense pairs in log-phase, early stationary phase, and after 15 minutes under heat shock conditions (Figure 2c; Additional file 6). We found 25 pairs where the sense was induced while the antisense was repressed in either early stationary phase or heat shock (12 in early stationary phase, 21 in heat shock, 8 in both), and 12 pairs where the sense was repressed while the antisense was induced (6 in early stationary phase, 8 in heat shock, 2 in both). Notably, 17 of the 25 pairs with induced sense and repressed antisense in early stationary phase (relative to mid-log) involved sense genes important in respiration, mitochondrial function, alternative carbon source metabolism and starvation response (for example, PET10, MBR1, FMP46, POT1, MOH1, TKL2, ICL1, CTA1). Conversely, four of the six pairs with the opposite pattern involved sense genes with key roles in glycolysis and fermentation (for example, GPM1, PGK1). Many of the pairs with induced sense and repressed antisense following heat shock overlapped with those responsive to early stationary phase (consistent with known metabolic changes under stress ). Furthermore, they also included four genes known to be important under environmental stresses (the regulators CRF1 and MRK1, and the effectors HSP31 and GRE2). Thus, antisense regulation may play a regulatory role at coordinating the major metabolic changes in the diauxic shift and early stationary phase, and some of the changes in the environmental stress response [21–24].
The exosome component Rrp6 affects antisense levels, but the histone deacetylase Hda2 does not
To explore the mechanistic regulation of antisense transcription, we measured the expression of the 67 pairs of sense and antisense units using the nCounter assay in strains deleted for the exosome component RRP6 (Δrrp6), the histone deacetylase HDA2 (Δhda2), or both (Δrrp6Δhda2). Previous studies [2, 7] have suggested that Δrrp6 increases the levels of antisense transcription in the PHO84 locus, and that Hda2 is required for mediating the effect of antisense transcription on the sense transcripts in this locus. If these findings apply more broadly, we expect higher levels of antisense transcripts in Δrrp6, and a change in the relative levels of sense to antisense in either the Δhda2 or Δrrp6Δhda2 strains.
Evolutionary conservation of six antisense transcripts and their regulation in five species of yeast
We found that the tested antisense units are largely conserved in the sensu stricto species, and less so at increasing evolutionary distances. All six units were detected in at least one species besides S. cerevisiae. Five of the six units are present in sensu stricto Saccharomyces, and four are still observed in S. castellii and K. lactis. The absence in K. lactis of an antisense transcript to the PET10 gene, important for respiratory growth, is consistent with its respiratory lifestyle, and suggests that antisense transcription in this gene may have appeared after the whole genome duplication. We cannot rule out the possibility, however, that other antisense units are present in the K. lactis genome, or that the missing antisense units are expressed under different conditions.
The anti-correlation between sense and antisense units observed in S. cerevisiae is conserved in most post-WGD species, but not in the pre-WGD K. lactis. The differential expression of five sense-antisense pairs (PET10, MRK1, MBR1, CRF1, CTA1) is conserved in at least two out of three other sensu stricto species. The more distant S. castellii shows less conservation of transcriptional regulation, most prominently in the PET10 gene. In contrast, although we could detect four of the antisense units in K. lactis, their differential expression was not conserved. This is consistent with the lack of repression of the corresponding sense gene in mid-log K. lactis cultures. The absence of antisense (for two genes) and the observed correlated (rather than anti-correlated) regulation (for three others) in K. lactis may reflect either the increased phylogenetic distance or may be more directly related to the shift to a respiro-fermentative lifestyle. In the latter case, either antisense transcription or its regulatory pattern in those genes may have evolved concomitantly with the emergence of fermentative growth, and the repression of respiratory genes, such as PET10 and MBR1. Further experiments are needed to elucidate this relationship.
In this study, we used strand-specific mRNA sequencing to explore the extent of antisense transcription in yeast, and found 1,103 putative antisense transcripts expressed in mid-log phase in S. cerevisiae, ranging from 39 short ones covering only the 3′ UTR of sense genes to 145 long ones covering the entire sense ORF. We focus on 402 long antisense units (each spanning over 75% of a coding unit). In this category, our sequencing based methodology allowed us to identify 224 new antisense transcripts that, in previous studies based on tiling microarrays , were either undetected or annotated as long UTRs of neighboring genes.
What could be the role of such prevalent antisense transcription? To date, functional studies have identified a regulatory role for a few antisense transcripts [6–8], whereas genome-wide analyses have suggested that antisense transcripts may represent promiscuous leaky transcription from NFRs at the promoter of a neighboring gene or the 3′ UTR of the sense gene [2, 3, 28]. The diversity of lengths in our 1,103 antisense units - ranging from long antisense units covering entire ORFs to shorter ones mostly at the 3′ UTR - suggests that there may be more than a single underlying mechanism for their formation and function.
Our results do not support promiscuous or aberrant transcription as the primary cause of the observed antisense transcripts. We find antisense transcription at only 18% of the genes. Moreover, many of the units are long and show robust sequence coverage, in contrast to what we might expect in a noisy process. Finally, antisense genes are only very weakly correlated to their neighbors, inconsistent with leaky transcription from divergent promoters or 3′ NFRs.
Characterizing the functional effect of each unit requires delicate assays to disable the antisense unit, without harming the sense gene, which have been successfully performed only in a few examples [6–8]. We therefore instead examined whether the changes in expression of sense and antisense are consistent with a regulatory function. We chose to focus on the long antisense units because they exhibit strong signal in our data, are less well-studied, are less likely to reflect noise, and can be verified more rigorously.
We found that the sense transcripts corresponding to longer antisense units are significantly enriched for key processes in S. cerevisiae, including stress response, the differential regulation of growth and stationary phase, and possibly meiosis and sporulation. The high level of antisense expression is consistent with the repression of these processes in fast growing yeast, and suggests a potential global function. Indeed, when we examined the relative change in expression in sense and antisense units across multiple conditions using three technologies (tiling arrays , strand-specific qPCR, and nCounter measurements), we found a strong and consistent anti-correlation between sense genes and the corresponding antisense units. While these results are consistent with regulatory function of antisense units (for example, reduction of antisense transcription leads to increased sense transcription), we cannot rule out the possibility that anti-correlation can occur without active regulation of the antisense transcript. For example, it is possible that when a sense gene is repressed, there is a relieved hindrance of antisense-transcription. Notably, we found support for the role of Rrp6 in the regulation of antisense levels, resulting in an increase in antisense levels in the Δrrp6 mutant, and a concomitant, albeit very mild, decrease in sense levels. We could not demonstrate a general effect of Hda2 on the levels of sense or antisense transcripts (either alone or together with Rrp6), and - in all mutants - the differential expression of sense and antisense remained highly correlated to the wild-type regulation. This suggests that it may be challenging to generalize the mechanisms shown for specific transcripts (PHO84) to all antisense transcripts.
Independent support for a potential function is the conservation of expression and regulation of six antisense units tested across five species that have diverged more than 150 million years ago, suggesting purifying selection. Notably, previous studies in mammals have shown that certain non-coding RNAs (that are not antisense) can be conserved at the sequence level [17, 29], but the applicability of such analyses to antisense transcripts that cover ORFs is limited, and hence experimental data are needed to show conservation. We find that both the presence and the regulation of antisense transcripts are most diverged in the distant, pre-WGD species K. lactis. This may reflect either the increased phylogenetic distance per se, or an evolved role in regulating respiration genes in post-WGD species. Another possibility for the lack of conservation in expression or absence of antisense in S. castellii and K. lactis may be the presence of RNA interference in these species . Further experiments will be needed to elucidate these possibilities and characterize the full functional scope of antisense transcription in yeasts.
Our results expand and strengthen the existing body of evidence that antisense transcription is a substantial phenomenon in yeast, and not solely a noisy by product of imprecise transcription regulation. While the mechanism and function of antisense transcription is still elusive, our results indicate that antisense transcription is often conserved and plays a regulatory role in the yeast transcriptional response.
Materials and methods
All tables, figures, raw sequenced reads, and a link to a browser with the mapped reads appear on our supplementary website .
Strains and growth conditions
Strains and growth conditions
Gift from Leonid Kruglyak's lab
MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0
Gift from Andrew Murray's lab
Same as above with rrp6Δ::KANMX6
Same as above with hda2Δ::URA3
Gift from Oliver Rando's lab
Same as above with rrp6Δ::KANMX6, hda2Δ::NatMX4
NCYC Stock Center
CLIB Stock Center
CLIB Stock Center
To generate strain RGV 69(rrp6Δ::KANMX6, hda2Δ::NatMX4), strain RGV 71(rrp6Δ::KANMX6) was transformed with a PCR product constructed by using the pAG25 containing the NatMX4 cassette using the following primers: GTAAAAGTATTTGGCTTCATTAGTGTGTGAAAAATAAAGAAAATAGATACAATACTATCGACGGTCGACGGATCCCCGGGTT and AAGAAAGTATATAAAATCTCTCTATATTATACAGGCTACTTCTTTTAGGAAACGTCACATCGATGAATTCGAGCTCGTT . Correct integration of this construct was confirmed with the following: (5′ left) left TGGCGTATATGGTTCATTGC; (5′ right) GTATGGGCTAAATGTACGGG; (3′ left) left TGGCGTATATGGTTCATTGC; (3′ right) GGTTGGAGAGGCAAATTGAG.
Overnight cultures of S. cerevisiae were grown in 650 ml of media at 22°C to between 3 × 107 and 1 × 108 cell/ml, OD600 = 1.0. The overnight culture was split into two 300 ml cultures and cells from each were collected by removing the media via vacuum filtration (Millipore - Billerica, MA, USA). The cell-containing filters were re-suspended in pre-warmed media to either control (22°C) or heat-shock temperatures (37°C). Density measurements were taken approximately 1 minute after cells were re-suspended to ensure that concentrations did not change during the transfer from overnight media. We harvested 12 ml of culture at 15 minutes and quenched by adding to 30 ml liquid methanol at -40°C, which was later removed by centrifugation at -9°C, and stored these overnight at -80°C. Cell density measurements were repeatedly taken every 5 to 15 minutes for the first 2 hours after treatment. Harvested cells were later washed in RNase-free water and archived in RNAlater (Ambion - Austin, TX, USA) for future preparations. Cells were also harvested from cultures just before treatment for use as controls.
Overnight cultures of S. cerevisiae (BB32) were grown in 600 ml of media at 30°C until reaching a final concentration of 3 × 107 and 1 × 108 cell/ml. The culture was split into two parallel cultures of 250 ml and sodium chloride was added to one culture for a final concentration of 0.3 M NaCl. Cells were harvested by vacuum filtration at 15 minutes after the addition of sodium chloride and from cultures immediately before the addition of sodium chloride for use as controls (t = 0 minutes). Filters were placed in liquid nitrogen and stored at -80°C and were later archived in RNAlater for future use.
Overnight cultures for each species were grown to saturation in 3 ml rich medium. From the 3 ml overnight cultures, 300 ml of rich media was inoculated at the OD600 corresponding to 1 × 106 cell/ml: S. cerevisiae 0.016, S. paradoxus 0.016, S. mikatae 0.023, S. bayanus 0.016, S. castellii 0.020, and K. lactis 0.024. The density measurements were taken approximately 1 minute after cells were re-suspended to ensure that concentrations did not change during the transfer from overnight media. Cells were harvested and quenched at a final concentration of 60% methanol at the mid-log and early stationary phase time points. Mid-log was taken at the following OD600 values: S. cerevisiae, 0.35; S. paradoxus, 0.40; S. mikatae, 0.40; S. bayanus, 0.30; S. castellii, 0.35; and K. lactis, 0.30. The early stationary phase time points were taken 2 hours after the glucose levels reached zero. Glucose levels were monitored hourly using the YSI 2700 Select Bioanalyzer (YSI Life Sciences - Yellow Springs, OH, USA). OD600 values for early stationary phase time points were: S. cerevisiae, 4.6; S. paradoxus, 3.9; S. mikatae, 4.3; S. bayanus, 2.8; S. castellii, 3.2; and K. lactis, 5.0. Harvested cells were later washed in RNase-free water, archived in RNAlater (Ambion) for future preparations, and frozen at -80°C.
Stationary phase was done for S. cerevisiae (BB32) only. This experiment was set up identically to the diauxic shift, but samples were taken at mid-log, and 5-day time points. The 5-day samples were taken at the same time of day as the mid-log samples.
Strand-specific cDNA library
The library was created by modifying the previously described dUTP second strand method . All reagents were from Invitrogen (Carlsbad, CA, USA) except as noted. We fragmented 200 ng of S. cerevisiae polyA+ RNA by heating at 98°C for 40 minutes in 0.2 mM sodium citrate, pH 6.4 (Ambion). Fragmented RNA was concentrated to 5 μl, mixed with 3 μg random hexamers, incubated at 70°C for 10 minutes, and placed on ice. First-strand cDNA was synthesized with this RNA primer mix by adding 4 μl of 5× first-strand buffer, 2 μl of 100 mM DTT, 1 μl of 10 mM dNTPs, 4 μg of actinomycin D (USB), 200 U SuperScript III, and 20 U SUPERase-In (Ambion) and incubating at room temperature for 10 minutes followed by 1 hour at 55°C. First-strand cDNA was cleaned up by extraction twice with phenol:chloroform:isoamyl alcohol (25:24:1), followed by ethanol precipitation with 0.1 volumes 5 M ammonia acetate to remove dNTPs and re-suspension in 104 μl H2O. Second-strand cDNA was synthesized by adding 4 μl 5× first-strand buffer, 2 μl 100 mM DTT, 4 μl 10 mM dNTPs with dTTP replaced by dUTP (Sigma - Aldrich, St Louis, MO, USA), 30 μl 5× second strand buffer, 40 U Escherichia coli DNA polymerase, 10 U E. coli DNA ligase, 2 U E. coli RNase H and incubating at 16°C for 2 hours. A paired-end library for Illumina sequencing was prepared according to the instructions provided with the following modifications. First, five times less adapter mix was ligated to the cDNAs. Second, 1 U USER (New England Biolabs - Ipswich, MA, USA) was incubated with 180- to 480-bp size-selected, adapter-ligated cDNA at 37°C for 15 minutes followed by 5 minutes at 95°C before PCR. Third, PCR was performed with Phusion High-Fidelity DNA Polymerase with GC buffer (New England Biolabs) and 2 M betaine (Sigma). Fourth, PCR primers were removed using 1.8× volume of AMPure PCR Purification kit (Beckman Coulter Genomics - Danvers, MA, USA).
Strand-specific library based on the RNA ligation method
The RNA ligation library was created using a previously described method  starting from 1.2 μg of polyA+ RNA with the following modifications. RNA was fragmented by incubation at 70°C for 8 minutes in 1× fragmentation buffer (Ambion) and 65- to 80-nucleotide RNA fragments were isolated from a gel. RNA was reverse transcribed with SuperScript III (Invitrogen) at 55°C and cDNA was amplified with Herculase (Agilent - Santa Clara, CA, USA) in the presence of 5% DMSO for 16 cycles of PCR followed by a clean up with 1.8× volumes of AMPure beads (Beckman Coulter Genomics - Danvers, MA, USA) rather than gel purification.
Both cDNA libraries were sequenced with an Illumina Genome Analyzer II (San Diego, CA, USA). The dUTP library was sequenced using 1 lane of 76-nucleotide paired reads, and the RNA ligation library was sequenced using 2 lanes of 51-nucleotide reads. All RNA-seq data are available in the Gene Expression Omnibus [GEO:GSE21739].
We used the Arachne mapper  to map the reads to the genome. We next identified consecutive regions of transcription by segmenting the centers of the paired-end segments with coverage >1 and maximum signal gaps of size 20 nucleotides.
Assessment of the strand specificity of the library
To evaluate the strand specificity of our library, we used the known annotation from SGD , and published estimates of UTR lengths , or when absent an estimation of 100 bp. According to these annotations we found that only 53,803 reads (0.62%) mapped to the opposite strand of known transcripts.
Identification of sense and antisense transcriptional units
We assigned a putative unit to a known gene if it is in the same orientation as the unit and it overlaps the known transcript boundaries, including published estimates of UTR length , or when absent an estimation of 100 bp was used. When comparing our transcription units to known annotations in the SGD , we examined the top 85% of expressed genes, as previously described .
Manual annotation of 402 antisense units
We have manually annotated the boundaries of antisense units covering 75% or more of an opposite ORF, resulting in 402 antisense units covering 75% or more of 412 ORFs.
Comparing the antisense units to published data from strand-specific tiling arrays
We compared our units to the published catalog of  using the following criteria. For each of our units, we searched for units in the catalog of  that are on the same strand and overlap it. We chose the unit with the highest overlap, and required a minimal threshold of 50% overlap.
Functional analysis of sense units
We constructed a gene set from the 377 sense genes, for which at least 75% of the ORF is covered by an antisense unit, and tested it for functional enrichment using a collection of functional categories as previously described . We also tested the genes for enriched induction or repression in a compendium of 1,400 annotated arrays, as previously described .
Identification of candidate regions in other species
We searched for orthologs of the sense gene in other species, using our published orthogroup catalog , and used the relative coordinates of the antisense transcripts in S. cerevisiae relative to the sense gene to predict their locations in other species. In cases where there were no clear candidates for orthologs, or the synteny block was broken , we did not define a candidate.
Strand-specific RT-PCR followed an adaptation of a published protocol . Total RNA was isolated from strain Bb32(3) at late log time point for two biological replicates. RNA was Turbo DNase treated (Ambion) following the manufacturer's stringent protocol followed by phenol chloroform extraction. For each assay, a gene-specific, strand-specific reverse transcription (RT) was performed. The four reactions for each sample were: +RT L-primer (sense), +RT R-Primer (antisense), +RT no primer, -RT both primers. First strand cDNA synthesis started RNA denaturation and the hybridization of the 2 pmol of gene specific primer. Total RNA with primer (10 ng) was heated to 70°C for 10 minutes and incubated on ice for at least 1 minute. A primer targeting ACT1 mRNA was always included as an internal control for strand specificity. This was followed by adding a Master mix containing 200 U SuperScript III (Invitrogen), 40 U RNaseOut (Invitrogen) and 10 mM dNTP mix for at 55°C for 15 minutes. The enzyme was heat-inactivated at 70°C for 15 minutes. RNA complementary to the cDNA was removed by E. coli RNase H (10 U; Ambion) and remaining RNAs were digested with 20 U of RNase Cocktail (Ambion) by incubating at 37°C for 20 minutes. PCR was performed for the sense and antisense transcripts independently. We added 5 μl of RT to each reaction as template with two gene-specific primers each at 250 nM final concentration (the same primers that were used for the sense and antisense RT; Additional file 7), 300 μM dNTP and 1 U of Ampli Taq Gold (Applied Biosystems - Carlsbad, CA, USA), in a 50 μl reaction. RNA contaminated with genomic DNA was used as a positive control. The touch down amplification program used was as follows: incubation of 95°C for 5 minutes followed by 10 cycles of 95°C for 30 s, 60°C for 30 s -1 degree per cycle, 70°C for 45 s, then followed by 17 to 20 cycles of 95°C for 30 s, 50°C for 30 s, 70°C 45 s, 72°C for 10 minutes (a step required for future Topo TA cloning (Invitrogen)).
Strand-specific RT-PCR across species
Strand-specific RT-PCR across species used an adaptation of a published protocol . Total RNA was isolated from each species at both the mid-log and early stationary phase time points. Genomic DNA contamination was removed with Turbo DNase (Ambion) using the stringent protocol, and phenol:chloroform to extract the RNA and to inactivate the DNase. For each of the species two biological replicates of the mid-log and early stationary phase time points were tested. Four reactions were performed for each sample: +RT L-primer (sense), +RT R-primer (antisense), +RT no primer, -RT. The sense, antisense, and -RT reactions were done with 2 pmol of primer (Additional file 7; only the primers with A1 in the title were used for the initial RT-PCR, and all primers used were designed for the target species). RT was done with first strand synthesis only in 20-μl reactions, using 4 units of Omniscript reverse transcriptase (Qiagen - Valencia, CA, USA) and 500 ng of total RNA. Each reaction was carried out at 50°C for 20 minutes, and heat inactivated at 70°C for 15 minutes. PCR was conducted as for the S. cerevisiae RT-PCR described above.
Strand-specific qRT-PCR across species
The same RT protocol was followed for the qRT-PCR across species as for the RT PCR above. For each sense-antisense pair validated, two sets of primers were tested, and primers for two internal control genes (ACT1 and PDA1) were included in each reaction. Control primers ('right primer', Additional file 7) were added at a concentration of 2 pmol to each of the RT reactions. qPCR was done using the Roche Light Cycler 480 in 12-μl reactions in a 384 well plate (Roche - Indianapolis, IN, USA). qPCR was done independently for sense, antisense, and control genes. RT samples were diluted 1:40 in water then 1:2 in Light Cycler 480 SYBR Green I Master with gene specific primer pair (each primer at 200 nM final concentration). The program protocol used was as follows: activation, 95°C for 5 minutes; cycling, 95°C for 15 s and 60°C for 45 s; melt, 95°C continuous.
Analysis of strand-specific qRT-PCR data
The ratios reported in Additional file 5 and Figure 2a are log2 ratios of early stationary phase and mid-log qRT-PCR reads (after normalization by the control gene PDA1), averaged over the two sets of primers and the two biological repeats.
The following experiments were done in biological duplicates: heat shock - 0 and 15 minutes; salt stress - 0 and 15 minutes; diauxic shift - log and early stationary phase; and stationary phase - log and 5 days. Details on the nCounter system are presented in full in . In a nutshell, the nCounter system uses pre-defined probes labeled with molecular barcodes ('code sets') and single molecule imaging to detect and directly count millions of unique transcripts (from up to hundreds of genes) in a single reaction. The assay is performed in cell lysates, involves no enzymatic steps prior to detection, and is highly accurate. Code sets were constructed to detect putative antisense units and sense genes and additional controls (Additional file 8). We lysed 7 × 107 (or 2 × 107, depending on the code set) cells according to the RNeasy (Qiagen) yeast mechanical lysis protocol. The protocol was stopped after spinning the lysate to remove debris, and 3 μl of the lysate was hybridized for 16 hours followed by processing in the nCounter Prep Station and quantification by the nCounter Digital Analyzer. We normalized the nCounter data in two steps as previously described . In the first step, we controlled for small variations in the efficiency of the automated sample processing. To this end, we followed the manufacturer's instructions, and normalized measurements from all samples analyzed on a given run to the levels of a chosen sample (in all cases we used the first sample in the set). This was done using the positive spiked-in controls provided by the nCounter instrument. In the second step, we used the control genes for which we designed probes to normalize for sample variation.
open reading frame
quantitative reverse transcriptase PCR
Saccharomyces Genome Database
whole genome duplication
yeast peptone dextrose
yeast peptone ethanol.
MY was supported by the Canadian Friends of the Hebrew University. The work was supported by the Howard Hughes Medical Institute, the Human Frontiers Science Program, a Career Award at the Scientific Interface from the Burroughs Welcome Fund, an NIH PIONEER award, the Broad Institute, and a Sloan Fellowship (AR), the National Human Genome Research Institute (NHGRI) (CN), and by a US-Israel Bi-national Science Foundation award (NF and MY).
- Faghihi MA, Wahlestedt C: Regulatory roles of natural antisense transcripts. Nat Rev Mol Cell Biol. 2009, 10: 637-643. 10.1038/nrm2738.PubMedPubMed CentralView ArticleGoogle Scholar
- Xu Z, Wei W, Gagneur J, Perocchi F, Clauder-Münster S, Camblong J, Guffanti E, Stutz F, Huber W, Steinmetz LM: Bidirectional promoters generate pervasive transcription in yeast. Nature. 2009, 457: 1033-1037. 10.1038/nature07728.PubMedPubMed CentralView ArticleGoogle Scholar
- Neil H, Malabat C, d'Aubenton-Carafa Y, Xu Z, Steinmetz LM, Jacquier A: Widespread bidirectional promoters are the major source of cryptic transcripts in yeast. Nature. 2009, 457: 1038-1042. 10.1038/nature07747.PubMedView ArticleGoogle Scholar
- Wilhelm BT, Marguerat S, Watt S, Schubert F, Wood V, Goodhead I, Penkett CJ, Rogers J, Bahler J: Dynamic repertoire of a eukaryotic transcriptome surveyed at single-nucleotide resolution. Nature. 2008, 453: 1239-1243. 10.1038/nature07002.PubMedView ArticleGoogle Scholar
- Dutrow N, Nix DA, Holt D, Milash B, Dalley B, Westbroek E, Parnell TJ, Cairns BR: Dynamic transcriptome of Schizosaccharomyces pombe shown by RNA-DNA hybrid mapping. Nat Genet. 2008, 40: 977-986. 10.1038/ng.196.PubMedPubMed CentralView ArticleGoogle Scholar
- Hongay , Grisafi , Galitski , Fink : Antisense transcription controls cell fate in Saccharomyces cerevisiae. Cell. 2006, 127: 735-745. 10.1016/j.cell.2006.09.038.PubMedView ArticleGoogle Scholar
- Camblong J, Iglesias , Fickentscher , Dieppois , Stutz : Antisense RNA stabilization induces transcriptional gene silencing via histone deacetylation in S. cerevisiae. Cell. 2007, 131: 706-717. 10.1016/j.cell.2007.09.014.PubMedView ArticleGoogle Scholar
- Houseley , Rubbi , Grunstein , Tollervey , Vogelauer : A ncRNA modulates histone modification and mRNA induction in the yeast GAL gene cluster. Mol Cell. 2008, 32: 685-695. 10.1016/j.molcel.2008.09.027.PubMedView ArticleGoogle Scholar
- Nishizawa , Komai , Katou , Shirahige , Ito , Toh-E : Nutrient-regulated antisense and intragenic RNAs modulate a signal transduction pathway in yeast. PLoS Biol. 2008, 6: 2817-2830. 10.1371/journal.pbio.0060326.PubMedView ArticleGoogle Scholar
- Nagalakshmi U, Wang Z, Waern K, Shou C, Raha D, Gerstein M, Snyder M: The transcriptional landscape of the yeast genome defined by RNA sequencing. Science. 2008, 320: 1344-1349. 10.1126/science.1158441.PubMedPubMed CentralView ArticleGoogle Scholar
- Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B: Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods. 2008, 5: 621-628. 10.1038/nmeth.1226.PubMedView ArticleGoogle Scholar
- Yassour M, Kaplan T, Fraser HB, Levin JZ, Pfiffner J, Adiconis X, Schroth G, Luo S, Khrebtukova I, Gnirke A, Nusbaum C, Thompson DA, Friedman N, Regev A: Ab initio construction of a eukaryotic transcriptome by massively parallel mRNA sequencing. Proc Natl Acad Sci USA. 2009, 106: 3264-3269. 10.1073/pnas.0812841106.PubMedPubMed CentralView ArticleGoogle Scholar
- Parkhomchuk D, Borodina T, Amstislavskiy V, Banaru M, Hallen L, Krobitsch S, Lehrach H, Soldatov A: Transcriptome analysis by strand-specific sequencing of complementary DNA. Nucleic Acids Res. 2009, 37: e123-10.1093/nar/gkp596.PubMedPubMed CentralView ArticleGoogle Scholar
- Cherry JM, Adler C, Ball C, Chervitz SA, Dwight SS, Hester ET, Jia Y, Juvik G, Roe T, Schroeder M, Weng S, Botstein D: SGD: Saccharomyces Genome Database. Nucleic Acids Res. 1998, 26: 73-79. 10.1093/nar/26.1.73.PubMedPubMed CentralView ArticleGoogle Scholar
- Levin JZ, Yassour M, Adiconis X, Nusbaum C, Thompson DA, Friedman N, Gnirke A, Regev A: Comprehensive comparative analysis of strand-specific RNA sequencing methods. Nat Methods. 2010Google Scholar
- Lister R, O'Malley RC, Tonti-Filippini J, Gregory BD, Berry CC, Millar AH, Ecker JR: Highly integrated single-base resolution maps of the epigenome in Arabidopsis. Cell. 2008, 133: 523-536. 10.1016/j.cell.2008.03.029.PubMedPubMed CentralView ArticleGoogle Scholar
- Berretta J, Morillon A: Pervasive transcription constitutes a new level of eukaryotic genome regulation. EMBO Rep. 2009, 10: 973-982. 10.1038/embor.2009.181.PubMedPubMed CentralView ArticleGoogle Scholar
- Tsankov A, Thompson DA, Socha A, Regev A, Rando OJ: The role of nucleosome positioning in the evolution of gene regulation. PLoS Biol. 2010, 8: e1000414-10.1371/journal.pbio.1000414.PubMedPubMed CentralView ArticleGoogle Scholar
- Amit I, Garber M, Chevrier N, Leite AP, Donner Y, Eisenhaure T, Guttman M, Grenier JK, Li W, Zuk O, Schubert LA, Birditt B, Shay T, Goren A, Zhang X, Smith Z, Deering R, McDonald RC, Cabili M, Bernstein BE, Rinn JL, Meissner A, Root DE, Hacohen N, Regev A: Unbiased reconstruction of a mammalian transcriptional network mediating pathogen responses. Science. 2009, 326: 257-263. 10.1126/science.1179050.PubMedPubMed CentralView ArticleGoogle Scholar
- Geiss GK, Bumgarner RE, Birditt B, Dahl T, Dowidar N, Dunaway DL, Fell HP, Ferree S, George RD, Grogan T, James JJ, Maysuria M, Mitton JD, Oliveri P, Osborn JL, Peng T, Ratcliffe AL, Webster PJ, Davidson EH, Hood L, Dimitrov K: Direct multiplexed measurement of gene expression with color-coded probe pairs. Nat Biotechnol. 2008, 26: 317-325. 10.1038/nbt1385.PubMedView ArticleGoogle Scholar
- Brown MP, Grundy WN, Lin D, Cristianini N, Sugnet CW, Furey TS, Ares M, Haussler D: Knowledge-based analysis of microarray gene expression data by using support vector machines. Proc Natl Acad Sci USA. 2000, 97: 262-267. 10.1073/pnas.97.1.262.PubMedPubMed CentralView ArticleGoogle Scholar
- Reisdorf P, Boy-Marcotte E, Bolotin-Fukuhara M: The MBR1 gene from Saccharomyces cerevisiae is activated by and required for growth under sub-optimal conditions. Mol Gen Genet. 1997, 255: 400-409. 10.1007/s004380050512.PubMedView ArticleGoogle Scholar
- Martin DE, Soulard A, Hall MN: TOR regulates ribosomal protein gene expression via PKA and the Forkhead transcription factor FHL1. Cell. 2004, 119: 969-979. 10.1016/j.cell.2004.11.047.PubMedView ArticleGoogle Scholar
- Agarwal S, Sharma S, Agrawal V, Roy N: Caloric restriction augments ROS defense in S. cerevisiae, by a Sir2p independent mechanism. Free Radic Res. 2005, 39: 55-62. 10.1080/10715760400022343.PubMedView ArticleGoogle Scholar
- Chechik G, Oh E, Rando O, Weissman J, Regev A, Koller D: Activity motifs reveal principles of timing in transcriptional control of the yeast metabolic network. Nat Biotechnol. 2008, 26: 1251-1259. 10.1038/nbt.1499.PubMedPubMed CentralView ArticleGoogle Scholar
- Byrne KP, Wolfe KH: Visualizing syntenic relationships among the hemiascomycetes with the Yeast Gene Order Browser. Nucleic Acids Res. 2006, 34: D452-455. 10.1093/nar/gkj041.PubMedPubMed CentralView ArticleGoogle Scholar
- Wapinski I, Pfeffer A, Friedman N, Regev A: Natural history and evolutionary principles of gene duplication in fungi. Nature. 2007, 449: 54-61. 10.1038/nature06107.PubMedView ArticleGoogle Scholar
- He Y, Vogelstein B, Velculescu VE, Papadopoulos N, Kinzler KW: The antisense transcriptomes of human cells. Science. 2008, 322: 1855-1857. 10.1126/science.1163853.PubMedPubMed CentralView ArticleGoogle Scholar
- Guttman M, Amit I, Garber M, French C, Lin MF, Feldser D, Huarte M, Zuk O, Carey BW, Cassady JP, Cabili MN, Jaenisch R, Mikkelsen TS, Jacks T, Hacohen N, Bernstein BE, Kellis M, Regev A, Rinn JL, Lander ES: Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature. 2009, 458: 223-227. 10.1038/nature07672.PubMedPubMed CentralView ArticleGoogle Scholar
- Drinnenberg IA, Weinberg DE, Xie KT, Mower JP, Wolfe KH, Fink GR, Bartel DP: RNAi in budding yeast. Science. 2009, 326: 544-550. 10.1126/science.1176945.PubMedPubMed CentralView ArticleGoogle Scholar
- Supplementary Website. [http://compbio.cs.huji.ac.il/YeastAntisense]
- Goldstein AL, McCusker JH: Three new dominant drug resistance cassettes for gene disruption in Saccharomyces cerevisiae. Yeast. 1999, 15: 1541-1553. 10.1002/(SICI)1097-0061(199910)15:14<1541::AID-YEA476>3.0.CO;2-K.PubMedView ArticleGoogle Scholar
- Batzoglou S, Jaffe DB, Stanley K, Butler J, Gnerre S, Mauceli E, Berger B, Mesirov JP, Lander ES: ARACHNE: a whole-genome shotgun assembler. Genome Res. 2002, 12: 177-189. 10.1101/gr.208902.PubMedPubMed CentralView ArticleGoogle Scholar
- Perocchi F, Xu Z, Clauder-Munster S, Steinmetz LM: Antisense artifacts in transcriptome microarray experiments are resolved by actinomycin D. Nucleic Acids Res. 2007, 35: e128-10.1093/nar/gkm683.PubMedPubMed CentralView ArticleGoogle Scholar
- Haddad F, Qin AX, Giger JM, Guo H, Baldwin KM: Potential pitfalls in the accuracy of analysis of natural sense-antisense RNA pairs by reverse transcription-PCR. BMC Biotechnol. 2007, 7: 21-10.1186/1472-6750-7-21.PubMedPubMed CentralView ArticleGoogle Scholar
- Integrative Genomics Viewer. [http://www.broadinstitute.org/igv]
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.