Open Access

Functional characterisation of long intergenic non-coding RNAs through genetic interaction profiling in Saccharomyces cerevisiae

  • Dimitris Kyriakou1,
  • Emmanouil Stavrou1,
  • Panayiota Demosthenous1,
  • Georgia Angelidou1,
  • Bryan-Joseph San Luis2,
  • Charles Boone2,
  • Vasilis J. Promponas1 and
  • Antonis Kirmizis1Email authorView ORCID ID profile
BMC Biology201614:106

https://doi.org/10.1186/s12915-016-0325-7

Received: 23 September 2016

Accepted: 9 November 2016

Published: 7 December 2016

Abstract

Background

Transcriptome studies have revealed that many eukaryotic genomes are pervasively transcribed producing numerous long non-coding RNAs (lncRNAs). However, only a few lncRNAs have been ascribed a cellular role thus far, with most regulating the expression of adjacent genes. Even less lncRNAs have been annotated as essential hence implying that the majority may be functionally redundant. Therefore, the function of lncRNAs could be illuminated through systematic analysis of their synthetic genetic interactions (GIs).

Results

Here, we employ synthetic genetic array (SGA) in Saccharomyces cerevisiae to identify GIs between long intergenic non-coding RNAs (lincRNAs) and protein-coding genes. We first validate this approach by demonstrating that the telomerase RNA TLC1 displays a GI network that corresponds to its well-described function in telomere length maintenance. We subsequently performed SGA screens on a set of uncharacterised lincRNAs and uncover their connection to diverse cellular processes. One of these lincRNAs, SUT457, exhibits a GI profile associating it to telomere organisation and we consistently demonstrate that SUT457 is required for telomeric overhang homeostasis through an Exo1-dependent pathway. Furthermore, the GI profile of SUT457 is distinct from that of its neighbouring genes suggesting a function independent to its genomic location. Accordingly, we show that ectopic expression of this lincRNA suppresses telomeric overhang accumulation in sut457Δ cells assigning a trans-acting role for SUT457 in telomere biology.

Conclusions

Overall, our work proposes that systematic application of this genetic approach could determine the functional significance of individual lncRNAs in yeast and other complex organisms.

Keywords

Long intergenic non-coding RNAs Stable unannotated transcripts Synthetic genetic array Genetic interactions Telomere

Background

Genome-wide transcriptional studies have revealed that eukaryotic genomes are pervasively transcribed, producing thousands long (>200 nucleotides) non-coding RNAs (lncRNAs) [1]. Nevertheless, functional characterisation of lncRNAs is lagging considerably behind their discovery rate [2]. One reason for this discrepancy is the fact that only a small proportion of these molecules are essential for life as shown by knockout studies of mammalian lncRNAs [3]. Therefore, the cellular effects of a mutated lncRNA might be masked by other factors with compensatory functions, indicating that the role of lncRNAs should be interrogated under synthetic knockout conditions. Another challenge for determining lncRNA functions stems from the fact that genetic manipulations of certain lncRNAs often affect the expression of other overlapping genes [2]. This issue is, of course, less true for long intergenic non-coding RNAs (termed lincRNAs), which do not overlap any other genomic features.

In Saccharomyces cerevisiae, approximately 85% of the genome is transcribed, generating a large number of lncRNAs [4, 5] that are reminiscent of those found in higher eukaryotes [6]. Among these non-coding transcripts there are classes of lncRNAs sensitive to RNA decay machinery, such as cryptic unstable transcripts, Xrn1-sensitive unstable transcripts and telomeric repeat-containing RNAs [4, 713], as well as lncRNAs expressed only under specific conditions such as meiotic unannotated transcripts [14], cytoplasmically degraded-cryptic unstable transcripts [15], stress-inducible lncRNAs [16] and telomerase mutant lncRNAs [17]. In addition, there are stable lncRNAs in wild-type yeast appropriately defined as stable unannotated transcripts (SUTs), which represent approximately 12% of the yeast transcriptome [11]. SUTs evade degradation in the nucleus and are processed in the cytoplasm in a similar manner to mRNAs [6]; thus, it was suggested that these transcripts might be functionally important [2, 18].

Only a handful of SUTs have been experimentally investigated and most of these have been assigned roles in regulating gene expression. In fact, it is the transcriptional process of some of these characterised lncRNAs that has a functional output and not the transcripts themselves [2]. For example, transcription of the lncRNA designated as IRT1 (IME1 Regulatory Transcript 1, also known as SUT643) represses the expression of its adjacent IME1 protein-coding gene by establishing repressive chromatin modifications at the IME1 promoter [19]. Additionally, transcription of the GAL10 lncRNA (aka SUT013) recruits similar chromatin modifying activities to alter nucleosome occupancy and to repress the expression of GAL1 and GAL10 [20, 21]. Likewise, the transcription of two other lncRNAs, ICR1 and PWR1, modulates chromatin structure and transcription factor binding at the FLO11 promoter [22]. Other steady-state yeast lncRNAs control gene expression through transcriptional interference and not through recruitment and modulation of chromatin modifications. For instance, transcription of the IME4-antisense lncRNA blocks sense strand transcription of the IME4 gene [23] and transcription of the SRG1 lncRNA prevents transcription initiation at the SER3 promoter [24]. A common feature of almost all characterised yeast lncRNAs, including the SUTs described above, is that they function in cis by regulating their cognate genes. Therefore, it remains unclear whether a significant number of yeast lncRNAs could also function in trans, distant from the locus from which they are transcribed, in order to regulate gene expression and other DNA-based processes.

The RNA component of the telomerase complex, known as telomerase component 1 (TLC1) [25], constitutes the single most characterised yeast lincRNA that functions in trans. TLC1 has a functional orthologue in human cells, known as TERC [26], but these two RNAs vary in size and nucleotide sequence. TLC1 is transcribed on chromosome II and then interacts physically with proteins Est1, Est2 and Est3 to form the telomerase complex, whose role is to solve the end replication problem by synthesising telomeric DNA repeats and preventing telomere shortening during each cell division [27, 28]. TLC1 is recruited to telomeres through a chain of physical interactions coordinated by the heterodimeric Ku complex (yKu70/yKu80) [2931]. Once at the telomeres, TLC1 functions as the template for telomere DNA synthesis by Est2, the reverse transcriptase component of the telomerase complex [27]. In the absence of functional telomerase, telomeres are significantly shortened and cells stop dividing and senesce [32]. Within a population of senescing cells some survivors arise by fixing their telomere length through a DNA recombination-based mechanism [33]. This alternative telomere maintenance mechanism is mediated by two distinct Rad52-dependent DNA recombination pathways; a commonly induced Type I pathway, which requires the Rad51, Rad54, Rad55 and Rad57 proteins, or a Type II pathway, which involves the trimeric MRX complex (consisting of Mre11, Rad50 and Xrs2) and Rad59 [33, 34]. Additional proteins, including the Ino80 chromatin remodelling complex, function as regulators of the telomere recombination mechanisms [34] in order to preserve telomere length and function.

Telomere homeostasis is also dependent on the presence of 3’-end single stranded DNA (ssDNA) at chromosome ends, known as telomeric overhangs. In most organisms, the telomeric overhangs consist of G-rich repeats and extend over the C-rich strands at both ends of a chromosome [27]. Telomeric overhangs are necessary during telomere replication because they provide a substrate for the RNA moiety of the telomerase complex. Formation of telomeric overhangs is closely linked to DNA replication, occurs independently of telomerase action and is regulated by evolutionary conserved mechanisms [35]. In S. cerevisiae, various exonucleolytic activities are involved in the formation of telomeric ssDNA, including the MRX complex and the 5’–3’ double-strand-specific exonuclease Exo1 [36]. The nuclease activity of these enzymes towards chromosome ends is inhibited by telomeric capping factors, such as Rap1, Rif1 and Rif2, that block MRX access, and the CST (Cdc13-Stn1-Ten1) and Ku complexes, which protect telomeres from Exo1-mediated degradation [37, 38]. Defects in telomere capping that can impact on the activity of exonucleases towards telomere ends lead to various problems, including premature senescence, cell-cycle arrest and accumulation of telomeric ssDNA [37, 3941]. Hence, complete understanding of the factors and mechanisms that control telomere-end processing is important for determining how telomeres maintain their structure and function [27].

In this study, we sought to identify genetic interactions (GIs) between yeast lincRNAs and protein-coding genes in order to evaluate their functional relationships using synthetic genetic array (SGA) technology. As proof of principle, we initially demonstrated that the GI network of the lincRNA TLC1 is consistent with its function in telomere maintenance. Therefore, we next applied the same systematic genetic approach to six other uncharacterised intergenic SUTs. Interestingly, the GI profile of SUT457 connected this lincRNA to telomere organisation and follow-up experiments established SUT457 as a novel factor of telomere overhang homeostasis. This study proposes that systematic analysis of GIs could unveil the function of lincRNAs in S. cerevisiae and other complex organisms.

Results

The GI network of TLC1 corresponds to its cellular function

Although genome-wide transcriptional studies have revealed an enormous amount of non-coding RNAs that can be synthesised from the S. cerevisiae genome, a relatively small number of these lncRNAs have been functionally characterised [2, 42]. We reasoned that we could obtain insights about the function of yeast lincRNAs by comprehensively mapping their GIs through an approach that has been previously applied for protein-coding genes [43]. Therefore, we employed the SGA methodology to construct double mutants in which the deletion of an intergenic SUT (sutΔ) is systematically combined with individual deletions of non-essential genes in budding yeast. GIs are scored in double mutants that show significant deviation in fitness compared to the growth of the corresponding single gene deletion strains generated during the control SGA screen (Fig. 1a and Methods). Specifically, a negative GI (NGI) refers to a more severe fitness defect in the double mutant compared to the corresponding single gene deletion mutants, while a positive GI (PGI) corresponds to growth with a less severe fitness defect in the double mutant in comparison to the single deletion mutants. The precise stepwise procedure used for the SGA screens and the deviations from a conventional approach [44] are shown in Fig. 1a (see also Methods).
Fig. 1

TLC1 validates the SGA approach for interrogating the function of lincRNAs. a Schematic illustration of the SGA screening procedure used for studying lincRNAs. A query strain (MATα) carrying deletion of a lincRNA (white box) is mated to an array of 4309 single knockout strains (MATa), each lacking a non-essential protein-coding gene (white X mark). Following several selection steps, double mutant strains are isolated and their growth fitness is compared to the corresponding single deletion strains derived from a parallel control SGA screen. b Genetic interactions (GIs) of TLC1 associated with the three indicated Gene Ontology (GO) biological processes. Gene nodes are coloured according to their assigned GO annotation. Negative (bold letters) and positive (regular font) GIs that belong to the three GO terms are shown. Previously established TLC1 cofactors are highlighted in light grey boxes. c Venn diagram showing the significant overlap (P = 2 × 10–29) between genes classified as having a negative GI with TLC1 and EST1 in this study. The P value was generated using the hypergeometric test (phyper)

To demonstrate proof of concept for the aforementioned rationale, we initially applied the SGA procedure to the yeast lincRNA TLC1, which has a well-defined cellular role in telomere maintenance [27, 28]. The SGA screen was carried out in duplicate by combining the tlc1Δ strain against an ordered array of approximately 4300 viable protein-coding gene deletion strains. A total of 116 NGIs and 261 PGIs of TLC1 were identified within the two SGA screens (Additional file 1: Table S1 and Additional file 2: Table S2). We identified NGIs with Ku complex components (YKU70/80) whose synthetic lethal interactions with TLC1 were previously reported [30]. Moreover, TLC1 showed NGIs with genes (RAD51, RAD52, RAD54, RAD57) whose proteins are integral components of the Rad52-mediated telomere recombination pathway [33] and the Ino80 chromatin remodelling complex that is also implicated in telomere lengthening via homologous recombination [34] (Fig. 1b). Since PGIs are typically detected among components of the same protein complex [45, 46], we also found that TLC1 has a positive GI with its co-factors EST1 and EST2, which are subunits of the telomerase complex. To further support the relevance of the identified GIs with the function of TLC1, we then performed gene ontology (GO) analysis of our SGA data, as an objective metric for deriving functional utility from GI datasets [47]. We chose to perform GO enrichment analysis using only NGIs, since the SGA technique demonstrates higher precision rate on detecting true NGIs compared to true PGIs and, in addition, NGIs frequently occur between genes with overlapping functions [43, 46, 48]. In agreement with the individual GIs mentioned above, GO analysis of all TLC1 NGIs significantly enriched the biological process terms, telomere organisation (P = 0.00081) and DNA recombination (P = 0.0049) (Figs. 1b and 2a). Furthermore, TLC1 NGIs significantly enriched the term chromatin remodelling (P = 0.0027), a cellular activity that has been associated with telomere maintenance [34, 4951] (Figs. 1b and 2a). These results show that the SGA-derived GIs of TLC1 are consistent with the cellular function of this yeast lincRNA.
Fig. 2

Genetic interactions link intergenic SUTs to diverse biological processes. a Heatmap of gene ontology (GO) terms enriched by genes whose deletions result in negative genetic interactions. The synthetic genetic array (SGA) procedure was performed for TLC1, EST1 and six intergenic SUTs. The negative genetic interaction (NGIs) from the different SGA screens were analysed as described in Methods to identify enriched GO terms with P ≤ 0.05 (generated using Fisher exact test). b and c Heatmap of GO terms enriched by genes classified as having a NGI with TLC1 (b), SUT457 (c) or their corresponding flanking genes. The NGIs of the genes flanking the tested lincRNAs were extracted from the Drygin database (http://drygin.ccbr.utoronto.ca/) and Szappanos et al. [53]

It was previously demonstrated that genes with highly similar GI profiles have strong functional relationship to the extent that they can be physical partners within the same complex [43]. Hence, we hypothesised that TLC1 should have a GI profile that is highly correlated to the profile of other components of the telomerase complex. To test this hypothesis and support the validity of our TLC1 SGA screens, we performed an independent SGA screen for the protein-coding gene EST1, which encodes a subunit of the telomerase complex. We identified 34 NGIs for EST1, of which 22 interactions were common with TLC1 NGIs (P = 2.004 × 10–29), indicating that their GI profiles are highly related (Fig. 1c). The shared GIs between EST1 and TLC1 enrich for the GO terms telomere organisation and DNA recombination that are related to their cellular function (Fig. 1c). To further examine the validity of our SGA screens, we also compared our identified TLC1 GIs to EST1 GIs reported in previous studies. Notably, about 40% of TLC1 NGIs identified in this study were also identified as EST1 negative GIs by two independent SGA studies [43, 52], and once again, the common NGIs between TLC1 and EST1 enrich the GO terms telomere organisation and DNA recombination (Additional file 3: Figure S1a, b). Overall, these findings demonstrate that SGA-derived GIs can illuminate the biological role of TLC1, and support the utility of this approach in identifying the function of other uncharacterised yeast lincRNAs.

SGA analysis implicates intergenic SUTs in diverse cellular processes

The S. cerevisiae genome encodes approximately 850 SUTs [11], of which 95 are intergenic (Additional file 4: Table S3) since they do not overlap any other known genomic features (Additional file 5: Table S4). We applied the SGA procedure to six intergenic SUTs that differ in size, distance from adjacent genes and level of expression [4, 11] (Additional files 4 and 6: Table S3 and S5). The SGA screen for each deleted lincRNA was paired with a control SGA screen (Fig. 1a) to minimise the ‘batch effect’ [46]. We identified in total 606 negative GIs and 1079 positive GIs for all six tested SUTs (Additional files 1 and 2: Tables S1 and S2). We subsequently analysed the NGIs of each SUT (Additional file 1: Table S1) for the enrichment of GO biological process terms as described for TLC1 above. Notably, all tested lincRNAs enriched at least one GO term with the exception of SUT123 which did not highlight any terms despite displaying 74 NGIs (Fig. 2a and Additional file 1: Table S1). It is possible that this lincRNA is involved in multiple processes that are diluted among the identified GIs. The other five SUTs exhibited NGIs that result in distinct GO enrichment profiles (Fig. 2a). In particular, the SGA screen using the sut457Δ strain enriched for the two GO terms, telomere organisation (P = 0.022) and not-yet-annotated (Fig. 2a). The latter term consists of functionally uncharacterised genes and therefore does not provide any insights into the function of this lincRNA. Furthermore, SUT042 also showed a specific profile since its screen enriched GO terms (membrane fusion P = 0.039; organelle fusion P = 0.020; vesicle organisation P = 0.020) associated with vesicle function (Fig. 2a). On the other hand, the other three lincRNAs, SUT014, SUT451 and SUT469, exhibited GO profiles that are more diverse in function. SUT451 and SUT469 SGA screens each enriched seven different biological processes, while SUT014 showed the most varied profile with enrichment of ten different GO terms (Fig. 2a). The heterogeneous GO term profiles suggest that the latter three SUTs are pleiotropic or that the enriched cellular processes are interconnected [43]. Altogether, these results indicate that, under physiological conditions, intergenic SUTs are implicated in a broad spectrum of biological processes that are important for normal cell growth.

SUT457 and SUT042 exhibit distinct GI profiles in comparison to their adjacent genes

Previously characterised SUTs and other yeast non-coding RNAs have been linked to transcriptional regulation of their neighbouring genes [2]. Therefore, we reasoned that SGA analysis of lincRNAs, which control their cognate genes, will lead to the discovery of GIs which would be similar to those identified for their neighbouring genes. To compare the GI networks of the examined intergenic SUTs with those of their neighbouring genes, we exploited the GI data of protein-coding genes available on the DRYGIN dataset (http://drygin.ccbr.utoronto.ca/; accessed 29 Feb 2016) and those generated by Szappanos et al. [53], which are compatible with the SGA approach used in this study. We first compared the GO enrichment profiles of TLC1 with its upstream and downstream gene, PDX3 and CSG2, respectively, since TLC1 is not linked to the transcriptional control of its neighbours. As predicted, the GIs of TLC1 enrich GO terms that are distinct to those enriched by the GIs of its flanking genes (Fig. 2b). Similarly to TLC1, two other lincRNAs, SUT457 and SUT042, exhibit GO enrichment profiles that are completely different from those of their upstream and downstream neighbouring genes (Fig. 2c and Additional file 7: Figure S2). In contrast, SUT014, SUT451 and SUT469, whose GIs enrich diverse biological processes, display GO profiles that partially overlap those of their adjacent genes (Additional file 7: Figure S2). These findings suggest that SUT014, SUT451 and SUT469 might have functions that are linked to the expression of their neighbouring genes, while SUT457 and SUT042 have roles that are independent to the function of their adjacent genes.

Deletion of SUT457 accelerates senescence in telomerase-deficient cells

SUT457 showed a very specific GO enrichment profile that was distinct from its flanking genes (Fig. 2) and, therefore, we decided to further characterise its cellular role. Initially, we verified by northern blot analysis the loss of SUT457 transcript (345 bp) in the sut457Δ strain which was processed in the SGA screens (Additional file 8: Figure S3). Then, we examined whether construction of the sut457Δ strain genetically perturbs the expression of its neighbouring genes (Fig. 3a). Quantitative RT-PCR analysis showed that in sut457Δ cells the expression of SUT457-adjacent genes SYP1, SNR65 and RPS14A remains unaffected compared to a wild-type control strain (Fig. 3b). This result demonstrates that construction of SUT457 deletion does not lead to the previously reported neighbouring gene effect [54, 55] and, hence, it confirms that the GI network of sut457Δ is directly associated with SUT457 itself and not with its adjacent genes.
Fig. 3

Deletion of SUT457 accelerates senescence in telomerase-deficient cells. a Schematic representation of the SUT457 locus on Chromosome III. Arrows indicate direction of transcription. b Expression levels of SYP1, SUT457, SNR65 and RPS14A determined by qRT-PCR using total RNA extracted from isogenic wild-type and sut457Δ strains. The expression levels of each gene were normalised to the expression of ACT1. Error bars represent standard error of the mean, resulting from three independent replicates. *** P < 0.001; n/s, not significant (generated by t-test). c Genetic interactions of SUT457 associated with the gene ontology term telomere organisation. Gene names in bold represent negative genetic interactions while gene names in regular font correspond to positive genetic interactions. d Telomere length analysis of genomic DNA isolated from the indicated wild-type and mutant strains at passage 1. The extracted DNA was fragmented with XhoI and subjected to denatured southern blotting using a biotinylated probe against telomeric repeats. e Senescence assays performed using liquid cultures of the indicated isogenic wild-type and mutant strains. The strains were generated through tetrad dissection of the heterozygous diploid double mutant SUT457/sut457ΔTLC1/tlc1Δ. This plot represents one of three independent tetrads examined (see also Additional file 9: Figure S4). f Telomere length analysis performed as in (d) using genomic DNA isolated from the indicated wild-type and mutant strains during passages 1 and 8

SUT457 genetically interacts with 12 out of 42 telomere organisation genes (P = 0.0145) found in the SGA deletion collection (Fig. 3c). These genes are associated with roles in supporting telomerase function (HEK2, PBP2, YKU70, TGS1, HSC82) [5659], modulating subtelomeric chromatin changes (PBP2 and HEK2) [56], regulating telomere-end processing (YKU70, GBP2, MRE11, CGI121, TGS1 and HSC82) [37, 58, 6063], and controlling telomere maintenance via recombination (RAD51, RAD52, RAD57, RAD59, MRE11, CGI121) [33, 34]. The genetic link between SUT457 and telomere organisation initially prompted us to determine if loss of this lincRNA leads to changes in telomere length. To do this, we used denaturing southern analysis and found that deletion of SUT457 alone does not affect telomere length compared to an isogenic wild-type strain (Fig. 3d, compare lanes 1 and 2, 28 and 29). Then, we examined whether sut457Δ, in combination with deletion of any of its 12 genetic interactors (Fig. 3c), affects telomere length. As expected, significantly shorter telomeres were detected in the yku70Δ and mre11Δ single mutants compared to wild-type cells, but telomere length was not further affected in the yku70Δsut457Δ or mre11Δsut457Δ double mutants (Fig. 3d, compare lanes 3 and 4, 5 and 6). Consistently, sut457Δ, in combination with deletion of the remaining telomere-related genetic interactors (Fig. 3c), does not have a significant change in telomere length compared to its isogenic single deletion mutants (Fig. 3d, lanes 7–14 and 16–27).

Based on the fact that some of the telomere-associated GIs of SUT457 (e.g. RAD51, RAD52, RAD57, MRE11) are known to affect entry into senescence in telomerase-deficient cells [34, 52], we next sought to determine if deletion of SUT457 influences the rate of senescence in TLC1 null cells. Therefore, we performed liquid senescence assays using the four isogenic strains, namely wild-type, sut457Δ, tlc1Δ and the double mutant sut457Δtlc1Δ. Consistent with the fact that single deletion of SUT457 had wild-type telomeric length (Fig. 3d), we observed that sut457Δ and wild-type strains had a very similar growth profile in senescence assays (Fig. 3e and Additional file 9: Figure S4). However, we found that the double mutant sut457Δtlc1Δ had an accelerated entry into senescence compared to the tlc1Δ single mutant (Fig. 3e and Additional file 9: Figure S4). Interestingly, this faster induction of senescence correlates with the appearance of shorter telomeres in the sut457Δtlc1Δ double deletion strain compared to the tlc1Δ single mutant (Fig. 3f, left panel). Moreover, deletion of SUT457 did not affect the generation of Type II survivors in strains lacking telomerase (Fig. 3f, right panel). Altogether, these findings link the function of SUT457 to telomere organisation and specifically show that SUT457 participates within a telomerase-independent pathway.

Loss of SUT457 leads to accumulation of telomeric single-stranded DNA

It has been previously reported that faster senescence in telomerase-negative cells can be attributed to abnormal accumulation of telomeric ssDNA [40, 41]. Thus, we hypothesised that deletion of SUT457 may lead to the accumulation of ssDNA at telomeres. This hypothesis was further supported by the SGA findings above, which show that SUT457 genetically interacts with the genes YKU70, MRE11, TGS1, CGI121, HSC82 and GBP2 (Fig. 3c) that encode for factors implicated in telomere-end protection and in the formation of telomeric ssDNA overhang [35]. To examine this hypothesis, we extracted genomic DNA from isogenic wild-type and sut457Δ cells at an early passage and performed native southern blotting using a C-rich oligonucleotide probe. In addition, we extracted genomic DNA from a later passage to mimic the subculturing of cells resulting from the serial pinning steps performed during the SGA procedure. Absence of SUT457 led to an increase in the signal of telomeric ssDNA compared to wild-type cells in passage 1 (P1), which became even more intense after subculturing yeast cells for five passages (P5) (Fig. 4a). In order to verify that the detected ssDNA signal is due to an increase in 3’ terminal overhang and not due to accumulation of internal DNA replication intermediates, we treated the native DNA with bacterial Exonuclease I that can only degrade terminal ssDNA in a 3’ to 5’ direction. The single stranded DNA that accumulated in sut457Δ mutant was sensitive to bacterial Exonuclease I digestion (Additional file 10: Figure S5), indicating that the detected ssDNA indeed corresponds to telomeric 3’ overhang. We also validated the increase in telomeric overhang by comparing the levels of ssDNA among 11 different sut457Δ mutant clones and their corresponding wild-type strains. As above, we observed statistically significant accumulation of telomeric ssDNA in sut457Δ strains at an early passage (P1) which became even more substantial at a later passage (P5) (Fig. 4b).
Fig. 4

Loss of SUT457 leads to Exo1-dependent accumulation of telomeric ssDNA. a Analysis of telomeric ssDNA overhangs in isogenic wild-type and sut457Δ strains. Yeast colonies were restreaked on agar plates and genomic DNA was extracted at passages 1 (lanes 1 and 2) and 5 (lanes 3 and 4), fragmented with XhoI and subjected to native southern blotting (left panel) using a biotinylated probe against telomeric repeats. The southern blot was then treated with 0.4 N NaOH and the denatured DNA was re-probed to monitor equal loading (right panel). b Accumulation of telomeric ssDNA in eleven sut457Δ strains compared to their respective wild-type strains during passages 1 and 5 (P1 & P5). The ssDNA signal for each wild-type and mutant strain detected in the native southern blot was normalised to the corresponding signal in the denatured southern blot. Error bars represent standard error of the mean. * P < 0.05; ** P < 0.01; n/s, not significant (generated by nonparmetric one-tailed Mann–Whitney t-test). c Cell growth was monitored for the indicated isogenic strains at passages 1 and 5. d Analysis of telomeric ssDNA levels at passage 5 of wild-type, sut457Δ, exo1Δ and exo1Δsut457Δ isogenic strains generated through dissection of the heterozygous diploid double mutant EXO1/exo1ΔSUT457/sut457Δ. Southern blotting was performed as in (a) above. e Telomeric ssDNA levels analysed at passage 5 in the indicated isogenic wild-type and mutant strains generated from three independent tetrads (n = 3). The quantification of the ssDNA signal for each wild-type and mutant strain was performed as in (b) above. Error bars represent standard error of the mean. ** P < 0.01; n/s, not significant. The statistical analysis was performed using one-way ANOVA (Dunnett’s test)

Accumulation of telomeric ssDNA has been proposed to be a signal for cell cycle arrest [39]. However, we did not observe a growth reduction in the sut457Δ mutant strain compared to wild-type cells (Figs. 3e and 4c). It was previously demonstrated that growth arrest associated with telomeric overhang accumulation is masked by telomeric ssDNA binding proteins like Gbp2 [60]. Interestingly, GBP2 was identified as a SUT457 NGI (Fig. 3c) and, therefore, we next examined how lack of Gbp2 can affect the growth of sut457Δ cells. We observed a strong growth arrest of gbp2Δsut457Δ cells at passage 5 (Fig. 4c), which coincides with the robust accumulation of telomeric ssDNA (Figs. 4a, b). Altogether, these data suggest that SUT457 controls the levels of telomeric ssDNA overhang in order to maintain proper cellular growth.

Exo1 nuclease is required for the accumulation of telomeric ssDNA in sut457Δ cells

The 5’–3’ exonuclease Exo1 plays a key role in telomere G-rich overhang formation in yeast by mediating C-rich strand degradation [64]. The action of Exo1 is blocked by the Ku complex since lack of Yku70 leads to unprotected telomere ends and results in Exo1-dependent accumulation of telomeric ssDNA [37, 38, 65]. Because SUT457 genetically interacts with YKU70 (Fig. 3c), we hypothesised that Exo1 activity may be responsible for the accumulation of telomeric overhang in sut457Δ cells. To test this hypothesis, we wanted to generate an exo1Δsut457Δ double mutant strain by crossing the exo1Δ and sut457Δ single mutants, but we realised that EXO1 was not properly deleted in the exo1Δ strain found in the SGA knockout library. Therefore, we constructed a new exo1Δ deletion strain, mated it to the sut457Δ mutant and generated isogenic haploid strains that were analysed for telomeric ssDNA levels. Consistent with the above findings, we detected a strong signal in sut457Δ corresponding to the accumulation of telomeric ssDNA after subculturing yeast for 5 passages (Fig. 4d). Notably, the telomeric overhang signal in the exo1Δsut457Δ double mutant cells is reduced back to wild-type levels (Fig. 4d, compare lanes 1, 3 and 4). This result was verified by quantifying the levels of telomeric ssDNA in isogenic mutant strains generated from three independent tetrads (Fig. 4e). Furthermore, deletion of EXO1 in gbp2Δsut457Δ cells rescued their growth arrest at passage 5 suggesting that exo1Δ suppresses the effect of sut457Δ (Additional file 11: Figure S6). Altogether, these findings show that SUT457 functions within an Exo1-dependent pathway to affect telomeric overhang accumulation.

The lincRNA SUT457 acts in trans to regulate the levels of telomeric ssDNA

The GI profile of SUT457 was distinct from that of its adjacent genes (Fig. 2c) and this is also consistent with the fact that loss of SUT457 does not affect the expression of its neighbours (Fig. 3b). Therefore, we hypothesised that, unlike most characterised yeast lncRNAs, which control the expression of their cognate genes, SUT457 may function in trans distant from its locus of synthesis. To address this hypothesis, we examined whether ectopic expression of SUT457 could rescue the phenotype of telomeric overhang accumulation observed in sut457Δ cells (Fig. 4a). For this purpose, we deleted SUT457 from its endogenous locus on chromosome III and re-introduced it on chromosome V under the control of the URA3 promoter (Fig. 5a, sut457Δ [ectopSUT457]). We also constructed another strain in which SUT457 was ectopically expressed at the URA3 locus in wild-type cells to serve as an additional control (Fig. 5a, wild-type [ectopSUT457]). Native southern blot analysis indicated that deletion of sut457Δ from its endogenous locus results in robust increase in telomeric ssDNA at passage 5 as previously shown (Figs. 4 and 5b, compare lanes 1 and 3), but concomitant ectopic expression of SUT457 reduces the telomeric overhang signal to corresponding wild-type levels (Fig. 5b). This result was verified by quantifying the levels of telomeric ssDNA in five independent constructed strains (Fig. 5c). Collectively, these findings demonstrate that SUT457 can control the levels of telomeric ssDNA overhang regardless of its genomic location, indicating that this lincRNA functions in trans.
Fig. 5

SUT457 regulates the levels of telomeric ssDNA by acting in trans. a Schematic illustration indicating the presence of endogenous (endogSUT457, light grey box) and ectopic SUT457 (ectopSUT457, dark grey box) in wild-type and sut457Δ strains. The level of SUT457 expression for each constructed strain is indicated on the right: + (wild-type levels of expression), +++ (higher expression than wild-type), and – (no expression). Arrows indicate the direction of SUT457 transcription at each locus. b Analysis of telomeric ssDNA levels in wild-type (lane 1), wild-type [ectopSUT457] (lane 2), sut457Δ (lane 3), and sut457Δ [ectopSUT457] (lane 4) strains at passage 5. Genomic DNA was extracted, fragmented with XhoI and subjected to native southern blotting using a biotinylated probe against telomeric repeats (left panel). The southern blot was then treated with 0.4 N NaOH and the denatured DNA was re-probed to monitor equal loading (right panel). c Telomeric ssDNA levels were analysed in five independent clones of the indicated wild-type and mutant strains at passage 5. The ssDNA signal for each strain detected in the native southern blot was normalised to the corresponding signal in the denatured southern blot. Error bars represent standard error of the mean. * P < 0.05; n/s, not significant. Statistical analysis was performed using one-way ANOVA (Dunnett’s test)

Discussion

Despite the plethora of long non-coding RNAs identified by genome-wide approaches, rigorous genetic studies that could aid in the functional characterisation of individual lncRNAs are missing [2]. Here, we took advantage of the genetic tractability of yeast and employed a high-throughput SGA methodology to systematically catalogue the GIs of six intergenic SUTs. Initially, we verified the utility of the approach by demonstrating that GIs identified for the telomerase RNA TLC1 enrich GO terms corresponding to its known cellular function. Subsequent SGA screens and analysis of GI profiles implicated the six tested intergenic SUTs in diverse biological processes. One of these lincRNAs, SUT457, exhibited GIs that enriched a definitive GO annotation linking it to telomere biology. Further functional characterisation unveiled the role of SUT457 in preventing the accumulation of telomeric ssDNA. Moreover, phenotypic rescue experiments showed that SUT457 acts in trans to maintain physiological levels of telomeric ssDNA. Our work reveals that systematic mapping of the GIs of lncRNAs could associate these molecules with specific biological processes and pinpoint their individual functions.

In yeast, the function of identified lncRNAs has been mainly linked to the regulation of gene expression. Most lncRNAs characterised so far regulate their cognate genes either through transcriptional interference [23, 24, 66] or by modulating their local chromatin structure [1922]. However, emerging evidence suggest that lncRNAs could have roles beyond gene regulation [42]. The GO enrichment profiles of the tested intergenic SUTs have linked them to nuclear functions such as DNA repair, meiosis and telomere organisation but also to cytoplasmic processes including membrane fusions and vesicle formation, for example, in the case of SUT042 (Fig. 2a). The latter observation is consistent with the fact that a large number of SUTs are transported to the cytoplasm and have been proposed to represent functional transcripts within this cell compartment [18]. Moreover, human lncRNAs are co-expressed with cytoplasmic membrane proteins [67] and are present within extracellular vesicles [68]. Therefore, future in-depth characterisation of lincRNAs, like SUT042, could elucidate the role of these molecules in cytoplasmic processes such as vesicle biogenesis and trafficking.

Only rare examples of trans-acting yeast lncRNAs have been reported to date. Specifically, the PHO84 antisense lncRNAs and a cryptic unstable transcript associated with Ty1 retrotransposons control transcriptional silencing of their target genes in trans [69, 70]. We show here that GI profiling could also provide insights about the mode of action of lincRNAs in addition to inferring their biological function. This is achieved by comparing the GI profiles of lincRNAs with those of their adjacent protein-coding genes. For example, TLC1 has a GI network that completely differs from that of its upstream and downstream genes (Fig. 2b) which is in line with the fact that TLC1 has a function unrelated to its neighbours and acts in trans at chromosome ends [71]. Similarly, SUT457 and SUT042 have distinct GO profiles compared to their neighbouring genes (Fig. 2c and Additional file 7: Figure S2), suggesting that these SUTs may also work in trans. Accordingly, SUT457 deletion does not affect the expression of its neighbouring genes (Fig. 3b) and regulates a telomeric process occurring distal from its transcriptional site (Fig. 5). However, there is still a possibility that the SUT457 locus is proximal to a gene involved in telomere-end processing within the three-dimensional architecture of the genome. Hence, in such a scenario, SUT457 would have a role in telomere-end processing by controlling its spatially proximal gene. Nevertheless, the phenotypic rescue obtained by expressing SUT457 from a different chromosome (Fig. 5) makes this scenario rather unlikely. Three of the other intergenic SUTs tested (SUT014, SUT451 and SUT469) exhibit GIs with GO enrichment profiles that overlap those of their flanking genes (Additional file 7: Figure S2), indicating that these lincRNAs may be involved in the regulation of their nearby genes in cis. Of course, we cannot eliminate, at this point, the likelihood that construction of these three SUT deletions leads to a neighbouring gene effect [54, 55], which might be responsible for the observed overlap among enriched GO profiles (Additional file 7: Figure S2). We note that two of the six tested intergenic SUTs (SUT042 and SUT457) exhibit GI profiles consistent with a trans-acting role and, therefore, in contrast to current evidence in the literature, we anticipate that a considerable number of budding yeast lincRNAs would function at sites distal from their locus of synthesis.

Lack of SUT457 exhibited accelerated senescence in telomerase-negative cells (Fig. 3e and Additional file 9: Figure S4) associated with enhanced telomere shortening (Fig. 3f) and an accumulation in telomeric ssDNA overhang (Fig. 4a). Interestingly, analogous phenotypes have been reported for mutations of factors involved in telomere-end protection [38, 40, 41, 52]. Consistent with the above observations, SUT457 shows a positive GI with the telomere-capping factor YKU70 (Fig. 3c and Additional file 12: Figure S7), implying that these two molecules affect the same molecular process [46]. Notably, Yku70 binds to and protects telomeres from Exo1 nucleolytic processing, thus preventing the accumulation of telomeric ssDNA overhang [30, 37, 64]. Our data also show that the Exo1 nuclease is required for the increase in telomeric ssDNA detected in the absence of SUT457 (Figs. 4d and e). Furthermore, in agreement with the fact that EXO1-dependent accumulation of telomeric ssDNA induces cell-cycle arrest [37, 72], we show that loss of the telomeric ssDNA masking protein Gbp2 in sut457Δ cells results in growth inhibition after subculturing (Fig. 4c) and, importantly, this growth arrest is rescued by deletion of EXO1 (Additional file 11: Figure S6). Based on the above evidence, we speculate that SUT457 functions in a pathway that protects telomere-ends from nucleolytic processing in order to regulate the levels of telomeric ssDNA overhangs. Interestingly, the telomere-associated lncRNA telomeric repeat-containing RNA has also been implicated in this pathway but instead facilitates the nuclease activity of Exo1 at chromosome ends [73].

The primary nucleotide sequence of SUT457 (Additional file 6: Table S5) does not display significant sequence similarity with any other regions within the S. cerevisiae genome and lacks conservation even amongst closely related yeast species ([11] and data not shown). Nevertheless, considering its unveiled role in the fundamental and conserved cellular process of telomere overhang homeostasis [35], we believe that functional analogues of SUT457 may exist in other eukaryotes. Likewise, TLC1 is a lncRNA whose primary sequence is not evolutionarily conserved but which has functional analogues in other organisms [26].

Conclusions

Transcriptome studies have enabled the discovery of a large number of lncRNAs in various species. However, it remains unclear how many of these lncRNAs serve a biological function or are a mere result of transcriptional noise [74, 75]. We show here that systematic analysis of pairwise GIs can provide insights about the function of individual lncRNAs. The functional maps obtained by GIs could complement the information derived by other high-throughput approaches [76, 77] to help unravel the biological significance of lncRNAs. Although the present study focuses on stable lincRNAs, future work could also analyse GIs for yeast antisense lncRNAs since methodologies are being developed that abrogate the transcription of a specific non-coding RNA without disrupting the expression of its associated sense mRNA [78]. Furthermore, global mapping of GIs for lncRNAs can be conceivably extended to complex organisms because recent advances in RNAi and CRISPR/Cas technologies allow for efficient pairwise gene disruptions in human cells [7982]. Such analyses may unveil interspecies conservation of GI maps [83] that could shed light on the evolution and functional conservation of lncRNAs [84, 85].

Methods

Yeast strains and plasmids

Yeast strains used in this study are described in Additional file 13: Table S6. SGA query strains were constructed by substituting the TLC1, EST1 or a candidate SUT locus with the NatMX4 cassette, which confers cloNAT (Nourseothricin) antibiotic resistance in Y7092 background strain (MATalpha can1Δ::STE2pr-Sp_his5 lyp1Δ his3Δ1 leu2Δ0 ura3Δ0 met15Δ0). The PCR fragment used for the above deletions was generated by primers listed in Additional file 14: Table S7. The SGA library consists of 4309 BY4741 (MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0) single knockout strains with each one carrying deletion of a non-essential gene. The genes were replaced with the antibiotic marker KanMX4, which confers resistance to G418 (Geneticin). The exo1Δ strain was constructed from scratch by transforming BY4741 cells with a PCR fragment that replaced the EXO1 Open Reading Frame with the KanMX4 cassette. The Y8835 strain in which the HIS3 gene is substituted by the NatMX4 cassette is used as the query strain in control SGA screens [44].

All isogenic strains used for follow-up experiments were generated by crossing the sut457Δ single mutant with single knockout strains of interest. Diploid cells were sporulated on 2% potassium acetate for 7 days at room temperature. Tetrads were isolated and subsequently manipulated using a dissection microscope (SporePlay, Singer Instruments) to generate isogenic wild-type, single and double mutant strains. All genotypes were verified by PCR analysis. For the construction of strains ectopically expressing SUT457, isogenic wild-type and sut457Δ cells were subcultured to passage 5 and then transformed with a PCR fragment containing the SUT457 sequence. The PCR fragment was integrated on chromosome V by homologous recombination in front of the URA3 promoter. The ectopic expression of SUT457 was verified by quantitative real-time PCR (qRT-PCR).

Selection of intergenic SUTs

To select intergenic SUTs, we developed an in-house perl script (filter-SUTs-2.pl) that identified, among all SUTs listed in the supplementary material of Xu et al. [11], the ones that did not overlap any of the genomic features (file SGD_features.tab, date stamp 20070811, listed in Additional file 5: Table S4) annotated in the SGD database (http://www.yeastgenome.org; accessed: Jan 2013). For any given SUT, genomic elements in both strands were taken into account. The list of selected intergenic SUTs is shown as Additional file 4: Table S3.

Synthetic genetic array (SGA)

The knockout query strain harbouring the deletion of interest (Y7092), in this case of a lincRNA or EST1 gene, was crossed against the SGA single deletion library described above. The diploid cells were sporulated, germinated and passaged as previously described [44], using a BM3-BC colony processing robot (S&P Robotis Inc.). Haploid mutants containing the deletion of interest (lincRNA or EST1) and/or the corresponding protein-coding gene deletion were isolated as described previously [44], apart from the following modifications in the protocol in order to improve population purity: (1) strains were pinned two times instead of one on media selecting for can1Δ, lyp1Δ and STE2pr-Sp_his5, and (2) strains were pinned two times instead of one on media selecting for double deletions (Fig. 1a). The same procedure was followed in parallel SGA screens using the control query strain (Y8835).

Quantification of yeast colony size on the final selection plates of the SGA screen was accomplished using spImager (S&P robotics Inc., Toronto, Canada). For each plate, we carried out a normalisation procedure (using a custom built perl script) based on the average colony growth detected on the specific plate, in order to correct for uneven plate growth. We excluded from further analysis: (1) colonies on the periphery of the plate (YOR202W), (2) genes for which the single deletion strain harbouring the relative library gene deletion grew less than 60% of the average growth, and (3) linkage group loci (composed of 30 genes upstream and downstream from the query gene of interest) to avoid artefacts related to potential linkage disequilibrium [44]. Finally, we report, as potential GIs, those genes for which the double deletion strain has at least a ±30% change in growth fitness compared to the single deletion strain. This threshold was defined based on growth changes observed for known TLC1 GIs.

Gene ontology (GO) analysis

The genes corresponding to the NGIs identified in each SGA screen were grouped into their annotated GO biological processes according to the Saccharomyces Genome Database (SGD) GO Slim Mapper (http://www.yeastgenome.org/cgi-bin/GO/goSlimMapper.pl). A comprehensive list of all 102 GO biological processes used during this analysis is provided in Additional file 15: Table S8. The Fisher exact test (significance level α = 0.05) was used to identify statistically significantly enriched GO terms. The same GO analysis was performed on NGI datasets generated in this work and NGIs obtained from previous studies.

Venn diagrams

Venn diagrams for GI datasets were generated using the on-line tool Venny 2.0 (http://bioinfogp.cnb.csic.es/tools/venny/). Significance of the overlap between gene sets was evaluated using the hypergeometric test, as implemented in the R function ‘phyper’ (http://www.R-project.org/), setting the significance level to α = 0.05.

RNA isolation and gene expression analysis

Total RNA from logarithmically grown (OD600 of 0.8) yeast cells was isolated using the hot phenol extraction method [86] and treated with the TURBO DNA-free DNase kit (Ambion, AM1907). For cDNA preparation, 0.5 μg RNA (DNase treated) was treated with PrimeScript Reverse Transcriptase (Takara, 2680A). The quantity of cDNA was determined by qRT-PCR analysis performed on a Bio-Rad CFX96 Real-Time PCR system using SYBR Green (Kapa SYBR Fast Master Mix # KK4602) and the primers listed in Additional file 14: Table S7.

Northern blot analysis

Isolated total RNA was first treated with DNAse and then mixed with 10 μL deionised formamide, 3.5 μL 37% formaldehyde and 2 μL loading buffer (0.1 M MOPS pH 7, 40 mM sodium acetate, 5 mM EDTA pH8.0) to a final volume of 20 μL. The sample was then heated at 70 °C for 10 min and electrophoresed on agarose-formaldehyde gel (1% agarose, 200 mM MOPS pH 7, 10 mM EDTA, 50 mM NaOAC, 6.7% formaldehyde) at 100 volts for 30 minutes. The gel was stained with ethidium bromide (0.75 μg/mL) to ensure RNA integrity and then equilibrated with 10× SSC buffer for 30 min (150 mM NaCl and 15 mM sodium citrate, pH 7) prior to transferring onto Hybond-N + membrane (GE Healthcare, RPN303B) using overnight capillary transfer with 20× SSC. The RNA was UV cross-linked (700 Joules/cm2) onto the membrane and hybridised with 50 ng/mL biotinylated probes (Additional file 14: Table S7) in 25 mL Church buffer (0.5 M NaPO4 pH 7.2, 1 mM EDTA pH 8, 7% SDS, 1% BSA) at 50 °C overnight. The membrane was treated with Chemiluminescent Nucleic Acid Detection Module (ThermoFisher Scientific, 89880) and then exposed using the UVP Bioimaging system (Syngene).

Native and denatured southern blotting

Genomic DNA from non-synchronised saturated cell cultures was digested overnight with XhoI (Takara, 1094A) and then separated on 1% agarose gel (15 cm in length) for 18 hours at 25 volts. E. coli exonuclease I digestion (New England biolabs, M0293S) was performed prior to XhoI digestion of the genomic DNA. Analysis of the telomeric single-stranded overhangs was performed under native conditions by treating the agarose gel with 10× SSC buffer (0.17 M trisodium citrate, 1.5 M NaCl) for 30 minutes at room temperature. The DNA bands were then transferred onto Hybond-N + membrane (GE Healthcare, RPN303B) using a Trans-Blot Semi-Dry electrophoretic transfer cell (Bio-Rad Laboratories) as previously described [87]. The DNA was UV cross-linked (700 Joules/cm2) on the membrane and hybridised with 50 ng/mL biotinylated C1-3A probe (Additional file 14: Table S7) in 25 mL Church buffer (0.5 M NaPO4 pH 7.2, 1 mM EDTA pH 8, 7% SDS, 1% BSA) at 50 °C overnight. Probe-bound DNA fragments corresponding to telomeric single-stranded overhangs were treated with the Chemiluminescent Nucleic Acid Detection Module Kit (ThermoFisher Scientific, 89880) and detected using the UVP Bioimaging system (Syngene). Following detection of the single-stranded telomeric overhang, the blot was incubated under denaturing conditions (0.4 M NaOH, 0.1% SDS) at 45 °C for 30 minutes. The blot was then probed and processed as above to detect denatured telomeric fragments. The intensity of the signals corresponding to native and denatured Y’ telomeric bands were quantified by histogram analysis (Adobe Photoshop CC 2015).

Single-colony re-streaking assay

Isogenic haploid strains derived from tetrad dissection are streaked on solid YPAD (Yeast extract, Peptone, Adenine hemisulfate, Dextrose) medium and incubated at 30 °C for 2 days until the appearance of single colonies (~25 cell divisions) corresponding to passage 1. Then, individual colonies from each wild-type and mutant strain were sequentially re-streaked on new solid YPAD plates until the appearance of colonies corresponding to passage 5 [34].

Senescence assay

A single colony from each indicated isogenic haploid strain was used to inoculate YPAD liquid medium and the culture was grown overnight at 30 °C until saturation. The following day, each culture was diluted to OD600 of 0.01 in fresh YPAD medium, incubated for 24 hours at 30 °C and the OD600 was then determined corresponding to the value of passage 1. The culture was then re-diluted to OD600 of 0.01 and the same procedure was repeated for passages 2–10 [34].

Abbreviations

GIs: 

genetic interactions

GO: 

gene ontology

lincRNAs: 

long intergenic non-coding RNAs

NGIs: 

negative genetic interactions

PGIs: 

positive genetic interactions

SGA: 

synthetic genetic array

ssDNA: 

single stranded DNA

SUTs: 

stable unannotated transcripts

Declarations

Acknowledgments

We thank Martin Kupiec, Maria Pia Longhese, Michael Costanzo and Anastasia Baryshnikova for fruitful discussions and advice. We also thank all members of the AK lab for discussions and assistance on experimental procedures and Sasan Ragibizadeh for technical support.

Funding

This work was supported by a grant to AK from the European Research Council (ERC-2010-Stg, N.260797, ChromatinModWeb).

Availability of data and material

All data generated or analysed during this study are included in this published article and its additional files. The datasets supporting the results of Figs. 4e and 5c are available in Additional file 16.

Authors’ contributions

DK, CB and AK conceived and designed experiments. DK, ES, PD and B-JSL performed experiments; DK, GA, VJP and AK analysed data; VJP critically read the manuscript; DK and AK wrote the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Consent for publication

Not applicable.

Ethics approval and consent to participate

Not applicable.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Authors’ Affiliations

(1)
Department of Biological Sciences, University of Cyprus
(2)
The Donnelly Centre, University of Toronto

References

  1. Jensen T, Jacquier A, Libri D. Dealing with pervasive transcription. Mol Cell. 2013;52:473–84.View ArticlePubMedGoogle Scholar
  2. Quinn JJ, Chang HY. Unique features of long non-coding RNA biogenesis and function. Nat Rev Genet. 2016;17:47–62.View ArticlePubMedGoogle Scholar
  3. Sauvageau M, Goff LA, Lodato S, Bonev B, Groff AF, Gerhardinger C, et al. Multiple knockout mouse models reveal lincRNAs are required for life and brain development. Elife. 2013;2013:1–24.Google Scholar
  4. David L, Huber W, Granovskaia M, Toedling J, Palm CJ, Bofkin L, et al. A high-resolution map of transcription in the yeast genome. Proc Natl Acad Sci U S A. 2006;103:5320–5.View ArticlePubMedPubMed CentralGoogle Scholar
  5. Nagalakshmi U. The transcriptional landscape of the yeast genome defined by RNA sequencing. Science. 2008;320:1344–9.View ArticlePubMedPubMed CentralGoogle Scholar
  6. Marquardt S, Hazelbaker DZ, Buratowski S. Distinct RNA degradation pathways and 3’ extensions of yeast non-coding RNA species. Transcription. 2011;2:145–54.View ArticlePubMedPubMed CentralGoogle Scholar
  7. Wyers F, Rougemaille M, Badis G, Rousselle JC, Dufour ME, Boulay J, et al. Cryptic Pol II transcripts are degraded by a nuclear quality control pathway involving a new poly(A) polymerase. Cell. 2005;121:725–37.View ArticlePubMedGoogle Scholar
  8. Davis CA, Ares M. Accumulation of unstable promoter-associated transcripts upon loss of the nuclear exosome subunit Rrp6p in Saccharomyces cerevisiae. Proc Natl Acad Sci U S A. 2006;103:3262–7.View ArticlePubMedPubMed CentralGoogle Scholar
  9. Houalla R, Devaux F, Fatica A, Kufel J, Barrass D, Torchet C, et al. Microarray detection of novel nuclear RNA substrates for the exosome. Yeast. 2006;23:439–54.View ArticlePubMedGoogle Scholar
  10. Luke B, Panza A, Redon S, Iglesias N, Li Z, Lingner J. The Rat1p 5’ to 3’ exonuclease degrades telomeric repeat-containing RNA and promotes telomere elongation in Saccharomyces cerevisiae. Mol Cell. 2008;32:465–77.View ArticlePubMedGoogle Scholar
  11. Xu Z, Wei W, Gagneur J, Perocchi F, Clauder-Münster S, Camblong J, et al. Bidirectional promoters generate pervasive transcription in yeast. Nature. 2009;457:1033–7.View ArticlePubMedPubMed CentralGoogle Scholar
  12. Neil H, Malabat C, d’Aubenton-Carafa Y, Xu Z, Steinmetz LM, Jacquier A, et al. Widespread bidirectional promoters are the major source of cryptic transcripts in yeast. Nature. 2009;457:1038–42.View ArticlePubMedGoogle Scholar
  13. van Dijk EL, Chen CL, D’Aubenton-Carafa Y, Gourvennec S, Kwapisz M, Roche V, et al. XUTs are a class of Xrn1-sensitive antisense regulatory non-coding RNA in yeast. Nature. 2011;475:114–7. Nature Publishing Group; Available from: /Users/yurikoharigaya/Documents/ReadCube Media/Nature 2011 van Dijk EL.pdf\nhttp://dx.doi.org/10.1038/nature10118.View ArticlePubMedGoogle Scholar
  14. Lardenois A, Liu Y, Walther T, Chalmel F, Evrard B, Granovskaia M, et al. Execution of the meiotic noncoding RNA expression program and the onset of gametogenesis in yeast require the conserved exosome subunit Rrp6. Proc Natl Acad Sci U S A. 2011;108:1058–63.View ArticlePubMedGoogle Scholar
  15. Toesca I, Nery CR, Fernandez CF, Sayani S, Chanfreau GF. Cryptic transcription mediates repression of subtelomeric metal homeostasis genes. PLoS Genet. 2011;7(6):e1002163.View ArticlePubMedPubMed CentralGoogle Scholar
  16. Nadal-Ribelles M, Sole C, Xu Z, Steinmetz LM, DeNadal E, Posas F. Control of Cdc28 CDK1 by a stress-induced lncRNA. Mol Cell. 2014;53:549–61.View ArticlePubMedPubMed CentralGoogle Scholar
  17. Niederer RO, Papadopoulos N, Zappulla DC. Identification of novel noncoding transcripts in telomerase-negative yeast using RNA-seq. Sci Rep. 2016;6:19376.View ArticlePubMedPubMed CentralGoogle Scholar
  18. Tuck AC, Tollervey D. A transcriptome-wide atlas of RNP composition reveals diverse classes of mRNAs and lncRNAs. Cell. 2013;154:996–1009.View ArticlePubMedPubMed CentralGoogle Scholar
  19. van Werven FJ, Neuert G, Hendrick N, Lardenois A, Buratowski S, Van Oudenaarden A, et al. Transcription of two long noncoding RNAs mediates mating-type control of gametogenesis in budding yeast. Cell. 2012;150:1170–81.View ArticlePubMedPubMed CentralGoogle Scholar
  20. Houseley J, Rubbi L, Grunstein M, Tollervey D, Vogelauer M. A ncRNA modulates histone modification and mRNA induction in the yeast GAL gene cluster. Mol Cell. 2008;32:685–95.View ArticlePubMedGoogle Scholar
  21. Pinskaya M, Gourvennec S, Morillon A. H3 lysine 4 di- and tri-methylation deposited by cryptic transcription attenuates promoter activation. EMBO J. 2009;28:1697–707.View ArticlePubMedPubMed CentralGoogle Scholar
  22. Bumgarner SL, Dowell RD, Grisafi P, Gifford DK, Fink GR. Toggle involving cis-interfering noncoding RNAs controls variegated gene expression in yeast. TL - 106. Proc Natl Acad Sci U S A. 2009;106(43):18321–6.View ArticlePubMedPubMed CentralGoogle Scholar
  23. Hongay CF, Grisafi PL, Galitski T, Fink GR. Antisense transcription controls cell fate in Saccharomyces cerevisiae. Cell. 2006;127:735–45.View ArticlePubMedGoogle Scholar
  24. Martens JA, Laprade L, Winston F. Intergenic transcription is required to repress the Saccharomyces cerevisiae SER3 gene. Nature. 2004;429:571–4.View ArticlePubMedGoogle Scholar
  25. Singer MS, Gottschling DE. TLC1: template RNA component of Saccharomyces cerevisiae telomerase. Science. 1994;266:404–9.View ArticlePubMedGoogle Scholar
  26. Feng J, Funk WD, Wang SS, Weinrich SL, Avilion AA, Chiu CP, et al. The RNA component of human telomerase. Science. 1995;269:1236–41.View ArticlePubMedGoogle Scholar
  27. Wellinger RJ, Zakian VA. Everything you ever wanted to know about Saccharomyces cerevisiae telomeres: beginning to end. Genetics. 2012;191:1073–105.View ArticlePubMedPubMed CentralGoogle Scholar
  28. Kupiec M. Biology of telomeres: lessons from budding yeast. FEMS Microbiol Rev. 2014;38:144–71.View ArticlePubMedGoogle Scholar
  29. Stellwagen AE, Haimberger ZW, Veatch JR, Gottschling DE. Ku interacts with telomerase RNA to promote telomere addition at native and broken chromosome ends. Genes Dev. 2003;17:2384–95.View ArticlePubMedPubMed CentralGoogle Scholar
  30. Lopez CR, Ribes-Zamora A, Indiviglio SM, Williams CL, Haricharan S, Bertuch AA. Ku must load directly onto the chromosome end in order to mediate its telomeric functions. PLoS Genet. 2011;7(8):e1002233.View ArticlePubMedPubMed CentralGoogle Scholar
  31. Pfingsten JS, Goodrich KJ, Taabazuing C, Ouenzar F, Chartrand P, Cech TR. Mutually exclusive binding of telomerase RNA and DNA by Ku alters telomerase recruitment model. Cell. 2012;148:922–32.View ArticlePubMedPubMed CentralGoogle Scholar
  32. Lundblad V, Szostak JW. A Mutant with a defect in telomere elongation leads to senescence in yeast. Cell. 1989;57:633–43.View ArticlePubMedGoogle Scholar
  33. Chen Q, Ijpma A, Greider CW. Two survivor pathways that allow growth in the absence of telomerase are generated by distinct telomere recombination events. Mol Cell Biol. 2001;21:1819–27.View ArticlePubMedPubMed CentralGoogle Scholar
  34. Hu Y, Tang HB, Liu NN, Tong XJ, Dang W, Duan YM, et al. Telomerase-null survivor screening identifies novel telomere recombination regulators. PLoS Genet. 2013;9(1):e1003208.View ArticlePubMedPubMed CentralGoogle Scholar
  35. Bonetti D, Martina M, Falcettoni M, Longhese MP. Telomere-end processing: Mechanisms and regulation. Chromosoma. 2014;123:57–66.View ArticleGoogle Scholar
  36. Bonetti D, Martina M, Clerici M, Lucchini G, Longhese MP. Multiple pathways regulate 3′ overhang generation at S. cerevisiae telomeres. Mol Cell. 2009;35:70–81.View ArticlePubMedGoogle Scholar
  37. Maringele L, Lydall D. EXO1-dependent single-stranded DNA at telomeres activates subsets of DNA damage and spindle checkpoint pathways in budding yeast yku70Δ mutants. Genes Dev. 2002;16:1919–33.View ArticlePubMedPubMed CentralGoogle Scholar
  38. Bonetti D, Clerici M, Anbalagan S, Martina M, Lucchini G, Longhese MP. Shelterin-like proteins and Yku inhibit nucleolytic processing of Saccharomyces cerevisiae telomeres. PLoS Genet. 2010;6:1.View ArticleGoogle Scholar
  39. Garvik B, Carson M, Hartwell L. Single-stranded DNA arising at telomeres in cdc13 mutants may constitute a specific signal for the RAD9 checkpoint. Mol Cell Biol. 1995;15:6128–38.View ArticlePubMedPubMed CentralGoogle Scholar
  40. Grandin N, Charbonneau M. Mrc1, a non-essential DNA replication protein, is required for telomere end protection following loss of capping by Cdc13, Yku or telomerase. Mol Genet Genomics. 2007;277:685–99.View ArticlePubMedGoogle Scholar
  41. Luke-Glaser S, Luke B. The Mph1 helicase can promote telomere uncapping and premature senescence in budding yeast. PLoS One. 2012;7(7):e42028.View ArticlePubMedPubMed CentralGoogle Scholar
  42. Yamashita A, Shichino Y, Yamamoto M. The long non-coding RNA world in yeasts. Biochim Biophys Acta. 1859;2015:1–8.Google Scholar
  43. Costanzo M, Baryshnikova A, Bellay J, Kim Y, Spear ED, Sevier CS, et al. The genetic landscape of a cell. Science. 2010;327:425–31.View ArticlePubMedGoogle Scholar
  44. Tong AHY, Evangelista M, Parsons AB, Xu H, Bader GD, Pagé N, et al. Systematic genetic analysis with ordered arrays of yeast deletion mutants. Science. 2001;294:2364–8.View ArticlePubMedGoogle Scholar
  45. Dixon SJ, Costanzo M, Baryshnikova A, Andrews B, Boone C. Systematic mapping of genetic interaction networks. Annu Rev Genet. 2009;43:601–25.View ArticlePubMedGoogle Scholar
  46. Baryshnikova A, Costanzo M, Kim Y, Ding H, Koh J, Toufighi K, et al. Quantitative analysis of fitness and genetic interactions in yeast on a genome scale. Nat Methods. 2010;7:1017–24.View ArticlePubMedPubMed CentralGoogle Scholar
  47. Myers CL, Barrett DR, Hibbs MA, Huttenhower C, Troyanskaya OG. Finding function: evaluation methods for functional genomic data. BMC Genomics. 2006;7:187.View ArticlePubMedPubMed CentralGoogle Scholar
  48. He X, Qian W, Wang Z, Li Y, Zhang J. Prevalent positive epistasis in Escherichia coli and Saccharomyces cerevisiae metabolic networks. Nat Genet. 2010;42:272–6.View ArticlePubMedPubMed CentralGoogle Scholar
  49. Askree SH, Yehuda T, Smolikov S, Gurevich R, Hawk J, Coker C, et al. A genome-wide screen for Saccharomyces cerevisiae deletion mutants that affect telomere length. Proc Natl Acad Sci U S A. 2004;101:8658–63.View ArticlePubMedPubMed CentralGoogle Scholar
  50. Yu EY, Steinberg-Neifach O, Dandjinou AT, Kang F, Morrison AJ, Shen X, et al. Regulation of telomere structure and functions by subunits of the INO80 chromatin remodeling complex. Mol Cell Biol. 2007;27:5639–49.View ArticlePubMedPubMed CentralGoogle Scholar
  51. Hass EP, Zappulla DC. The Ku subunit of telomerase binds Sir4 to recruit telomerase to lengthen telomeres in S. cerevisiae. Elife. 2015;4:1–19.View ArticleGoogle Scholar
  52. Chang H-Y, Lawless C, Addinall SG, Oexle S, Taschuk M, Wipat A, et al. Genome-wide analysis to identify pathways affecting telomere-initiated senescence in budding yeast. G3 (Bethesda). 2011;1(3):197–208.View ArticleGoogle Scholar
  53. Szappanos B, Kovács K, Szamecz B, Honti F, Costanzo M, Baryshnikova A, et al. An integrated approach to characterize genetic interaction networks in yeast metabolism. Nat Genet. 2011;43:656–62.View ArticlePubMedPubMed CentralGoogle Scholar
  54. Ben-Shitrit T, Yosef N, Shemesh K, Sharan R, Ruppin E, Kupiec M. Systematic identification of gene annotation errors in the widely used yeast mutation collections. Nat Methods. 2012;9:373–8.View ArticlePubMedGoogle Scholar
  55. Baryshnikova ABA. Neighboring-gene effect: a genetic uncertainty principle. Nat Methods. 2012;9:341–3.View ArticlePubMedGoogle Scholar
  56. Denisenko O, Bomsztyk K. Yeast hnRNP K-like genes are involved in regulation of the telomeric position effect and telomere length. Mol Cell Biol. 2002;22:286–97.View ArticlePubMedPubMed CentralGoogle Scholar
  57. Bertuch AA, Lundblad V. The Ku heterodimer performs separable activities at double-strand breaks and chromosome termini. Mol Cell Biol. 2003;23:8202–15.View ArticlePubMedPubMed CentralGoogle Scholar
  58. Franke J, Gehlen J, Ehrenhofer-Murray AE. Hypermethylation of yeast telomerase RNA by the snRNA and snoRNA methyltransferase Tgs1. J Cell Sci. 2008;121:3553–60.View ArticlePubMedGoogle Scholar
  59. Toogun OA, DeZwaan DC, Freeman BC. The Hsp90 molecular chaperone modulates multiple telomerase activities. Mol Cell Biol. 2008;28:457–67.View ArticlePubMedGoogle Scholar
  60. Pang TL, Wang CY, Hsu CL, Chen MY, Lin JJ. Exposure of single-stranded telomeric DNA causes G2/M cell cycle arrest in Saccharomyces cerevisiae. J Biol Chem. 2003;278:9318–21.View ArticlePubMedGoogle Scholar
  61. Larrivée M, LeBel C, Wellinger RJ. The generation of proper constitutive G-tails on yeast telomeres is dependent on the MRX complex. Genes Dev. 2004;18:1391–6.View ArticlePubMedPubMed CentralGoogle Scholar
  62. Downey M, Houlsworth R, Maringele L, Rollie A, Brehme M, Galicia S, et al. A genome-wide screen identifies the evolutionarily conserved KEOPS complex as a telomere regulator. Cell. 2006;124:1155–68.View ArticlePubMedGoogle Scholar
  63. DeZwaan DC, Toogun OA, Echtenkamp FJ, Freeman BC. The Hsp82 molecular chaperone promotes a switch between unextendable and extendable telomere states. Nat Struct Mol Biol. 2009;16:711–6.View ArticlePubMedPubMed CentralGoogle Scholar
  64. Ngo HP, Lydall D. Survival and growth of yeast without telomere capping by Cdc13 in the absence of Sgs1, Exo1, and Rad9. PLoS Genet. 2010;6(8):e1001072.View ArticlePubMedPubMed CentralGoogle Scholar
  65. Vodenicharov MD, Laterreur N, Wellinger RJ. Telomere capping in non-dividing yeast cells requires Yku and Rap1. EMBO J. 2010;29:3007–19.View ArticlePubMedPubMed CentralGoogle Scholar
  66. Ard R, Tong P, Allshire RC. Long non-coding RNA-mediated transcriptional interference of a permease gene confers drug tolerance in fission yeast. Nat Commun. 2014;5:5576.View ArticlePubMedPubMed CentralGoogle Scholar
  67. Fang K, Han BW, Chen ZH, Lin KY, Zeng CW, Li XJ, et al. A distinct set of long non-coding RNAs in childhood MLL-rearranged acute lymphoblastic leukemia: biology and epigenetic target. Hum Mol Genet. 2014;23:3278–88.View ArticlePubMedGoogle Scholar
  68. Mohankumar S, Patel T. Extracellular vesicle long noncoding RNA as potential biomarkers of liver cancer. Brief Funct Genomics. 2015;15(3):249–56.View ArticlePubMedPubMed CentralGoogle Scholar
  69. Berretta J, Pinskaya M, Morillon A. A cryptic unstable transcript mediates transcriptional trans-silencing of the Ty1 retrotransposon in S. cerevisiae. Genes Dev. 2008;22:615–26.View ArticlePubMedPubMed CentralGoogle Scholar
  70. Camblong J, Beyrouthy N, Guffanti E, Schlaepfer G, Steinmetz LM, Stutz F. Trans-acting antisense RNAs mediate transcriptional gene cosuppression in S. cerevisiae. Genes Dev. 2009;23:1534–45.View ArticlePubMedPubMed CentralGoogle Scholar
  71. Lue NF. Sequence-specific and conformation-dependent binding of yeast telomerase RNA to single-stranded telomeric DNA. Nucleic Acids Res. 1999;27:2560–7.View ArticlePubMedPubMed CentralGoogle Scholar
  72. Zubko MK, Guillard S, Lydall D. Exo1 and Rad24 differentially regulate generation of ssDNA at telomeres of Saccharomyces cerevisiae cdc13-1 mutants. Genetics. 2004;168:103–15.View ArticlePubMedPubMed CentralGoogle Scholar
  73. Pfeiffer V, Lingner J. TERRA promotes telomere shortening through exonuclease 1-mediated resection of chromosome ends. PLoS Genet. 2012;8(6):e1002747.View ArticlePubMedPubMed CentralGoogle Scholar
  74. Ponjavic J, Ponting CP, Lunter G. Functionality or transcriptional noise ? Evidence for selection within long noncoding RNAs. Genome Res. 2007;17(5):556–65.View ArticlePubMedPubMed CentralGoogle Scholar
  75. Marques AC, Ponting CP. Intergenic lncRNAs and the evolution of gene expression. Curr Opin Genet Dev. 2014;27:48–53.View ArticlePubMedGoogle Scholar
  76. Alcid EA, Tsukiyama T. Systematic approaches to identify functional lncRNAs. Curr Opin Genet Dev. 2016;37:46–50.View ArticlePubMedGoogle Scholar
  77. Kashi K, Henderson L, Bonetti A, Carninci P. Discovery and functional analysis of lncRNAs: methodologies to investigate an uncharacterized transcriptome. Biochim Biophys Acta Gene Regul Mech. 1859;2016:3–15.Google Scholar
  78. Huber F, Bunina D, Gupta I, Theer P, Steinmetz LM, Knop M. Protein abundance control by non-coding antisense protein abundance control by non-coding antisense transcription. Cell Rep. 2016;15:1–12.View ArticleGoogle Scholar
  79. Cong L, Ran FA, Cox D, Lin S, Barretto R, Hsu PD, et al. Multiplex genome engineering using CRISPR/VCas systems. Science. 2013;339:819–23.View ArticlePubMedPubMed CentralGoogle Scholar
  80. Roguev A, Talbot D, Negri GL, Shales M, Cagney G, Bandyopadhyay S, et al. Quantitative genetic-interaction mapping in mammalian cells. Nat Methods. 2013;10:432–7.View ArticlePubMedPubMed CentralGoogle Scholar
  81. Laufer C, Fischer B, Huber W, Boutros M. Measuring genetic interactions in human cells by RNAi and imaging. Nat Protoc. 2014;9:2341–53.View ArticlePubMedGoogle Scholar
  82. Blomen VA, Májek P, Jae LT, Bigenzahn JW, Nieuwenhuis J, Staring J, et al. Gene essentiality and synthetic lethality in haploid human cells. Science. 2015;350:1092–6.View ArticlePubMedGoogle Scholar
  83. Ryan CJ, Roguev A, Patrick K, Xu J, Jahari H, Tong Z, et al. Hierarchical modularity and the evolution of genetic interactomes across species. Mol Cell. 2012;46:691–704.View ArticlePubMedPubMed CentralGoogle Scholar
  84. Johnsson P, Lipovich L, Grandér D, Morris KV. Evolutionary conservation of long non-coding RNAs; sequence, structure, function. Biochim Biophys Acta. 1840;2014:1063–71.Google Scholar
  85. Hezroni H, Koppstein D, Schwartz MG, Avrutin A, Bartel DP, Ulitsky I. Principles of long noncoding RNA evolution derived from direct comparison of transcriptomes in 17 species. Cell Rep. 2015;11:1110–22.View ArticlePubMedPubMed CentralGoogle Scholar
  86. Schmitt ME, Brown TA, Trumpower BL. A rapid and simple method for preparation of RNA from Saccharomyces cerevisiae. Nucleic Acids Res. 1990;18:3091–2.View ArticlePubMedPubMed CentralGoogle Scholar
  87. Ortega LM, Hengartner CJ, Vega LR. Nonradioactive method to detect native single-stranded G-tails on yeast telomeres using a modified southern blot protocol. Biotechniques. 2011;50:407–10.View ArticlePubMedGoogle Scholar

Copyright

© Kirmizis et al. 2016