Making sense of the natural antisense transcript puzzle

Natural antisense transcripts are very common in plant genomes, with thousands of genes associated with antisense transcription.Despite many sense–antisense RNA pairs having the potential to form long double-stranded RNAs (i.e., substrates for dicer proteins), there is weak evidence that gene silencing plays a prominent role.Natural antisense transcripts are rather likely to be involved in a broad range of regulatory processes, including translation enhancement, with indications that several other mechanisms have yet to be discovered.We present a roadmap for the functional dissection of antisense transcript action in gene regulation and suggest testable hypotheses relying on an experiment-based decision tree. In plants, thousands of genes are associated with antisense transcription, which often produces noncoding RNAs. Although widespread, sense–antisense pairs have been implicated in a limited variety of functions in plants and are often thought to form extensive dsRNA stretches triggering gene silencing. In this opinion, we show that evidence does not support gene silencing as a major role for antisense transcription. In fact, it is more likely that antisense transcripts play diverse functions in gene regulation. We propose a general framework for the initial functional dissection of antisense transcripts, suggesting testable hypotheses relying on an experiment-based decision tree. By moving beyond the gene silencing paradigm, we argue that a broad and diverse role for natural antisense transcription will emerge. In plants, thousands of genes are associated with antisense transcription, which often produces noncoding RNAs. Although widespread, sense–antisense pairs have been implicated in a limited variety of functions in plants and are often thought to form extensive dsRNA stretches triggering gene silencing. In this opinion, we show that evidence does not support gene silencing as a major role for antisense transcription. In fact, it is more likely that antisense transcripts play diverse functions in gene regulation. We propose a general framework for the initial functional dissection of antisense transcripts, suggesting testable hypotheses relying on an experiment-based decision tree. By moving beyond the gene silencing paradigm, we argue that a broad and diverse role for natural antisense transcription will emerge. Natural antisense transcripts (NATs) (see Glossary) are widespread in eukaryotic genomes. The bulk of our current knowledge on the role of NATs in gene regulation is for cis-NATs (i.e., NATs and sense transcripts are transcribed from the same genomic locus), which are the focus of this work. Early work based on strand-specific RNA sequencing (RNA-seq) using short-read RNA sequencing of polyA+ mRNA, revealed that approximately 30% of loci of arabidopsis (Arabidopsis thaliana) are associated with cis-NATs [1.Wang H. et al.Genome-wide identification of long noncoding natural antisense transcripts and their responses to light in Arabidopsis.Genome Res. 2014; 24: 3Crossref Scopus (216) Google Scholar]. More recent analysis of arabidopsis transcriptomics based on strand-specific RNA seq of either polyA+ mRNAs [2.Deforges J. et al.Control of cognate sense mRNA translation by cis-natural antisense RNAs.Plant Physiol. 2019; 180: 305-322Crossref PubMed Scopus (21) Google Scholar] or RNAs transcriptionally engaged with RNA Polymerase II (RNAPII) [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar], reported 4300 to 5400 distinct cis-NATs, indicating that approximately 15–20% of arabidopsis loci possess a cis-NAT. Work in rice (Oryza sativum) using single-molecule long-read RNA sequencing reported that nearly 60% of loci are associated with a cis-NAT [4.Chen M. et al.Full-length transcript-based proteogenomics of rice improves its genome and proteome annotation.Plant Physiol. 2020; 182: 1510-1526Crossref PubMed Scopus (10) Google Scholar]. Similar high levels of genes associated with cis-NATs are also found in mice and humans [5.Faghihi M.A. Wahlestedt C. Regulatory roles of natural antisense transcripts.Nat. Rev. Mol. Cell Biol. 2009; 10: 637-643Crossref PubMed Scopus (515) Google Scholar]. Our current understanding is that most (>91%) cis-NATs are long noncoding RNAs (lncRNAs), and their expression appears to be specific to certain growth conditions and tissues [1.Wang H. et al.Genome-wide identification of long noncoding natural antisense transcripts and their responses to light in Arabidopsis.Genome Res. 2014; 24: 3Crossref Scopus (216) Google Scholar,2.Deforges J. et al.Control of cognate sense mRNA translation by cis-natural antisense RNAs.Plant Physiol. 2019; 180: 305-322Crossref PubMed Scopus (21) Google Scholar]. Consequently, the sampling of the cis-NAT landscape is variable in different studies and likely underestimated, given that most reports are limited to one or a few growth conditions and tissue types. Choice of methodologies for RNA isolation and analysis also impact estimation of the cis-NAT pool, since it influences the depth of analysis of the transcriptome and the type of RNA analyzed, including capped and polyA+ RNAs or RNAPII-associated RNAs, which can include unstable cis-NAT that are rapidly degraded by the nuclear exosome pathway [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar,6.Mayer A. et al.Native elongating transcript sequencing reveals human transcriptional activity at nucleotide resolution.Cell. 2015; 161: 541-554Abstract Full Text Full Text PDF PubMed Scopus (191) Google Scholar] (Box 1). Despite these variables, the current consensus is that several thousand genes in plants are associated with a cis-NAT. However, few of these sense–antisense transcript pairs have been functionally characterized in detail in plants. Considering that sense–antisense overlapping nucleotides are thought to form double-stranded RNAs (dsRNAs) that are substrates for dicer proteins, gene silencing would appear to be an important mode of action of cis-NATs. In this opinion piece, we argue that this notion is at odds with current knowledge and that it is likely that only a small fraction of sense–antisense interactions results in silencing. We observed that historical reasons, rather than empirical data, have perhaps (mis)guided plant biologists to adopt a simplistic view of the functions of cis-NATs (Box 2). This is evidenced by the contrastingly varied functions that this RNA class displays in animals [5.Faghihi M.A. Wahlestedt C. Regulatory roles of natural antisense transcripts.Nat. Rev. Mol. Cell Biol. 2009; 10: 637-643Crossref PubMed Scopus (515) Google Scholar].Box 1Detection of cis-NATsFor detection of cis-natural antisense transcripts (NATs), any technology relying on sequencing of double-strand DNA generated from reverse-transcribed RNA needs to be able to unambiguously differentiate between the sense and antisense strands. Numerous techniques and protocols produce strand-specific RNA-seq data, thus retaining the strand information. However, RT-PCR can generate various artefacts through, for example, reverse transcriptase template switching or mispriming, which can lead to erroneous assignment of antisense transcripts [65.Perocchi F. et al.Antisense artifacts in transcriptome microarray experiments are resolved by actinomycin D.Nucleic Acids Res. 2007; 35e128Crossref PubMed Scopus (152) Google Scholar, 66.Houseley J. Tollervey D. Apparent non-canonical trans-splicing is generated by reverse transcriptase in vitro.PLoS One. 2010; 5e12271Crossref PubMed Scopus (98) Google Scholar, 67.Mourão K. et al.Detection and mitigation of spurious antisense expression with RoSA.F1000Research. 2019; 8: 819Crossref Google Scholar]. As cis-NATs are often weakly expressed relative to their cognate sense mRNAs, care must be exercised when the ratio of sense mRNA to cis-NAT expression level is very large (>1000) [2.Deforges J. et al.Control of cognate sense mRNA translation by cis-natural antisense RNAs.Plant Physiol. 2019; 180: 305-322Crossref PubMed Scopus (21) Google Scholar]. RT-based artefacts can be avoided with long-read direct RNA sequencing, albeit read depth is lower as compared with strand-specific RNA-seq [68.Parker M.T. et al.Nanopore direct RNA sequencing maps the complexity of Arabidopsis mRNA processing and m6A modification.eLife. 2020; 9e49658Crossref PubMed Scopus (66) Google Scholar].In recent years, various methods have been developed to identify and sequence RNAs associated with RNAPII. Methods such as global run-on sequencing (GRO-seq) and neosynthesized 5-ethynyl uridine RNA sequencing (NEU-seq) identify nascent RNAs via the labeling of RNAs with modified nucleotides [69.Szabo E.X. et al.Metabolic labeling of RNAs uncovers hidden features and dynamics of the Arabidopsis transcriptome.Plant Cell. 2020; 32: 871-887Crossref PubMed Scopus (0) Google Scholar, 70.Hetzel J. et al.Nascent RNA sequencing reveals distinct features in plant transcription.Proc. Natl. Acad. Sci. 2016; 113: 12316-12321Crossref PubMed Scopus (68) Google Scholar, 71.Zhu J. et al.RNA polymerase II activity revealed by GRO-seq and pNET-seq in Arabidopsis.Nat. Plants. 2018; 4: 1112-1123Crossref PubMed Scopus (47) Google Scholar]. Native elongating transcript sequencing (NET-seq) and a similar method adapted for plants (plaNET-seq) rely on the immunoprecipitation of RNAPII–RNA complexes [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar]. In addition to providing a high-resolution view of RNAPII transcription dynamic along a gene, these methods allow the detection of unstable RNAs that are subject to rapid degradation through the nuclear exosome pathway, including transcripts that are not polyadenylated at their 3′ end. Analysis of RNA transcripts using these methods in wild type and nuclear exosome mutants, such as hen2, enabled the identification of numerous novel unstable and co-transcriptionally degraded cis-NATs, as well as revealing that previously well-defined capped and polyadenylated cis-NAT, such as COOLAIR, are nevertheless targets of the exosome pathway [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar,69.Szabo E.X. et al.Metabolic labeling of RNAs uncovers hidden features and dynamics of the Arabidopsis transcriptome.Plant Cell. 2020; 32: 871-887Crossref PubMed Scopus (0) Google Scholar]. Analysis of RNAPII–RNA complexes also showed that numerous cis-NATs overlapping at their 3′ end (tail-to-tail configuration) arise via RNAPII continuing a few hundred nucleotides past the polyA site, a phenomenon influenced by temperature and the histone demethylase FLD [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar,72.Inagaki S. et al.Chromatin-based mechanisms to coordinate convergent overlapping transcription.Nat. Plants. 2021; 7: 295-302Crossref PubMed Scopus (1) Google Scholar].Box 2The long-lasting excitement of the 1990s around antisense RNAsThe natural occurrence of antisense RNAs was first demonstrated 40 years ago in prokaryotes, in which an antisense RNA was found to regulate maturation of the ColE1 primer for plasmid DNA replication [73.Itoh T. Tomizawa J. Formation of an RNA primer for initiation of replication of ColE1 DNA by ribonuclease H.Proc. Natl. Acad. Sci. 1980; 77: 2450-2454Crossref PubMed Google Scholar] (see timeline in Figure I). It was later found in mammals [74.Williams T. Fried M. A mouse locus at which transcription from both DNA strands produces mRNAs complementary at their 3′ ends.Nature. 1986; 322: 275-279Crossref PubMed Google Scholar], and in 1992, cis-natural antisense transcripts (NATs) were first described in plants [75.Schmitz G. Theres K. Structural and functional analysis of the Bz2 locus of Zea mays: characterization of overlapping transcripts.Mol. Gen. Genet. 1992; 233: 269-277Crossref PubMed Scopus (16) Google Scholar]. However, for over a decade, very few cis-NATs have been reported in plants [76.Terryn N. Rouzé P. The sense of naturally transcribed antisense RNAs in plants.Trends Plant Sci. 2000; 5: 394-396Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar], possibly because of technical challenges associated with very low cDNA sequencing throughput and poor strand information. For instance, in RNA probing assays (e.g., in situ hybridization and Northern blot), the sense probe was often used as a negative control, and signals (i.e., associated with the antisense transcript) were disregarded as artifacts, evidencing a lack of interest or momentum in the field. Moreover, the discovery that artificial expression of antisense RNAs leads to downregulation of specific gene(s) in mammals [77.Izant J.G. Weintraub H. Inhibition of thymidine kinase gene expression by anti-sense RNA: a molecular approach to genetic analysis.Cell. 1984; 36: 1007-1015Abstract Full Text PDF PubMed Scopus (350) Google Scholar] and plants [78.Ecker J.R. Davis R.W. Inhibition of gene expression in plant cells by expression of antisense RNA.Proc. Natl. Acad. Sci. 1986; 83: 5372-5376Crossref PubMed Google Scholar,79.van der Krol A.R. et al.An antisense chalcone synthase gene in transgenic plants inhibits flower pigmentation.Nature. 1988; 333: 866-869Crossref Scopus (323) Google Scholar] resulted in antisense RNA becoming a widespread tool for functional analysis.In petunia, the pigments that give flowers color are derived from the flavonoid biosynthesis pathway, in which chalcone synthase is an essential enzyme. Constitutive expression of an antisense sequence for the chalcone synthase gene downregulated endogenous gene expression and the enzyme itself, producing various pigmentation patterns that correlated with the downregulation strength [79.van der Krol A.R. et al.An antisense chalcone synthase gene in transgenic plants inhibits flower pigmentation.Nature. 1988; 333: 866-869Crossref Scopus (323) Google Scholar]. This pivotal work opened the floodgates for crop improvement using antisense-based transgenic strategies [80.Mol J.N.M. et al.Regulation of plant gene expression by antisense RNA.FEBS Lett. 1990; 268: 427-430Crossref PubMed Scopus (92) Google Scholar].It is interesting to note the inversion between research and development in the history of antisense RNA (i.e., the technology was developed much ahead of any meaningful basic knowledge of the process). In fact, in addition to poor knowledge of natural antisense RNAs, the mechanistic understanding of how artificial antisense RNAs inhibit gene expression was over a decade away from their development as a technology. In 1998, it was discovered that dsRNA formation by sense–antisense RNA pairs was a substrate for RNase degradation, triggering gene silencing [81.Waterhouse P.M. et al.Virus resistance and gene silencing in plants can be induced by simultaneous expression of sense and antisense RNA.Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 13959-13964Crossref PubMed Scopus (856) Google Scholar]. Thus, the decade-long excitement around antisense RNAs was based on gene silencing. Furthermore, antisense RNA was very inefficient as a tool for post-transcriptional gene silencing (PTGS), as only a small proportion of transgenic plants would produce silenced individuals, and its use was rapidly superseded by more efficient approaches, such as expression of intron-spliced RNA with hairpin structure [82.Smith N.A. et al.Total silencing by intron-spliced hairpin RNAs.Nature. 2000; 407: 319-320Crossref PubMed Scopus (658) Google Scholar]. However, the assumption that plant cis-NATs function similarly to their artificial counterparts in gene silencing persisted [83.Yamada K. Empirical analysis of transcriptional activity in the Arabidopsis Genome.Science. 2003; 302: 842-846Crossref PubMed Scopus (720) Google Scholar], even after evidence for roles in other processes in animals [84.Werner A. Natural antisense transcripts.RNA Biol. 2005; 2: 53-62Crossref PubMed Google Scholar] and evidence that cis-NATs’ primary role(s) might not involve transcript cleavage in humans [85.Faghihi M.A. Wahlestedt C. RNA interference is not involved in natural antisense mediated regulation of gene expression in mammals.Genome Biol. 2006; 7: R38-9Crossref PubMed Scopus (52) Google Scholar]. For detection of cis-natural antisense transcripts (NATs), any technology relying on sequencing of double-strand DNA generated from reverse-transcribed RNA needs to be able to unambiguously differentiate between the sense and antisense strands. Numerous techniques and protocols produce strand-specific RNA-seq data, thus retaining the strand information. However, RT-PCR can generate various artefacts through, for example, reverse transcriptase template switching or mispriming, which can lead to erroneous assignment of antisense transcripts [65.Perocchi F. et al.Antisense artifacts in transcriptome microarray experiments are resolved by actinomycin D.Nucleic Acids Res. 2007; 35e128Crossref PubMed Scopus (152) Google Scholar, 66.Houseley J. Tollervey D. Apparent non-canonical trans-splicing is generated by reverse transcriptase in vitro.PLoS One. 2010; 5e12271Crossref PubMed Scopus (98) Google Scholar, 67.Mourão K. et al.Detection and mitigation of spurious antisense expression with RoSA.F1000Research. 2019; 8: 819Crossref Google Scholar]. As cis-NATs are often weakly expressed relative to their cognate sense mRNAs, care must be exercised when the ratio of sense mRNA to cis-NAT expression level is very large (>1000) [2.Deforges J. et al.Control of cognate sense mRNA translation by cis-natural antisense RNAs.Plant Physiol. 2019; 180: 305-322Crossref PubMed Scopus (21) Google Scholar]. RT-based artefacts can be avoided with long-read direct RNA sequencing, albeit read depth is lower as compared with strand-specific RNA-seq [68.Parker M.T. et al.Nanopore direct RNA sequencing maps the complexity of Arabidopsis mRNA processing and m6A modification.eLife. 2020; 9e49658Crossref PubMed Scopus (66) Google Scholar]. In recent years, various methods have been developed to identify and sequence RNAs associated with RNAPII. Methods such as global run-on sequencing (GRO-seq) and neosynthesized 5-ethynyl uridine RNA sequencing (NEU-seq) identify nascent RNAs via the labeling of RNAs with modified nucleotides [69.Szabo E.X. et al.Metabolic labeling of RNAs uncovers hidden features and dynamics of the Arabidopsis transcriptome.Plant Cell. 2020; 32: 871-887Crossref PubMed Scopus (0) Google Scholar, 70.Hetzel J. et al.Nascent RNA sequencing reveals distinct features in plant transcription.Proc. Natl. Acad. Sci. 2016; 113: 12316-12321Crossref PubMed Scopus (68) Google Scholar, 71.Zhu J. et al.RNA polymerase II activity revealed by GRO-seq and pNET-seq in Arabidopsis.Nat. Plants. 2018; 4: 1112-1123Crossref PubMed Scopus (47) Google Scholar]. Native elongating transcript sequencing (NET-seq) and a similar method adapted for plants (plaNET-seq) rely on the immunoprecipitation of RNAPII–RNA complexes [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar]. In addition to providing a high-resolution view of RNAPII transcription dynamic along a gene, these methods allow the detection of unstable RNAs that are subject to rapid degradation through the nuclear exosome pathway, including transcripts that are not polyadenylated at their 3′ end. Analysis of RNA transcripts using these methods in wild type and nuclear exosome mutants, such as hen2, enabled the identification of numerous novel unstable and co-transcriptionally degraded cis-NATs, as well as revealing that previously well-defined capped and polyadenylated cis-NAT, such as COOLAIR, are nevertheless targets of the exosome pathway [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar,69.Szabo E.X. et al.Metabolic labeling of RNAs uncovers hidden features and dynamics of the Arabidopsis transcriptome.Plant Cell. 2020; 32: 871-887Crossref PubMed Scopus (0) Google Scholar]. Analysis of RNAPII–RNA complexes also showed that numerous cis-NATs overlapping at their 3′ end (tail-to-tail configuration) arise via RNAPII continuing a few hundred nucleotides past the polyA site, a phenomenon influenced by temperature and the histone demethylase FLD [3.Kindgren P. et al.Native elongation transcript sequencing reveals temperature dependent dynamics of nascent RNAPII transcription in Arabidopsis.Nucleic Acids Res. 2020; 48: 2332-2347Crossref PubMed Scopus (25) Google Scholar,72.Inagaki S. et al.Chromatin-based mechanisms to coordinate convergent overlapping transcription.Nat. Plants. 2021; 7: 295-302Crossref PubMed Scopus (1) Google Scholar]. The natural occurrence of antisense RNAs was first demonstrated 40 years ago in prokaryotes, in which an antisense RNA was found to regulate maturation of the ColE1 primer for plasmid DNA replication [73.Itoh T. Tomizawa J. Formation of an RNA primer for initiation of replication of ColE1 DNA by ribonuclease H.Proc. Natl. Acad. Sci. 1980; 77: 2450-2454Crossref PubMed Google Scholar] (see timeline in Figure I). It was later found in mammals [74.Williams T. Fried M. A mouse locus at which transcription from both DNA strands produces mRNAs complementary at their 3′ ends.Nature. 1986; 322: 275-279Crossref PubMed Google Scholar], and in 1992, cis-natural antisense transcripts (NATs) were first described in plants [75.Schmitz G. Theres K. Structural and functional analysis of the Bz2 locus of Zea mays: characterization of overlapping transcripts.Mol. Gen. Genet. 1992; 233: 269-277Crossref PubMed Scopus (16) Google Scholar]. However, for over a decade, very few cis-NATs have been reported in plants [76.Terryn N. Rouzé P. The sense of naturally transcribed antisense RNAs in plants.Trends Plant Sci. 2000; 5: 394-396Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar], possibly because of technical challenges associated with very low cDNA sequencing throughput and poor strand information. For instance, in RNA probing assays (e.g., in situ hybridization and Northern blot), the sense probe was often used as a negative control, and signals (i.e., associated with the antisense transcript) were disregarded as artifacts, evidencing a lack of interest or momentum in the field. Moreover, the discovery that artificial expression of antisense RNAs leads to downregulation of specific gene(s) in mammals [77.Izant J.G. Weintraub H. Inhibition of thymidine kinase gene expression by anti-sense RNA: a molecular approach to genetic analysis.Cell. 1984; 36: 1007-1015Abstract Full Text PDF PubMed Scopus (350) Google Scholar] and plants [78.Ecker J.R. Davis R.W. Inhibition of gene expression in plant cells by expression of antisense RNA.Proc. Natl. Acad. Sci. 1986; 83: 5372-5376Crossref PubMed Google Scholar,79.van der Krol A.R. et al.An antisense chalcone synthase gene in transgenic plants inhibits flower pigmentation.Nature. 1988; 333: 866-869Crossref Scopus (323) Google Scholar] resulted in antisense RNA becoming a widespread tool for functional analysis. In petunia, the pigments that give flowers color are derived from the flavonoid biosynthesis pathway, in which chalcone synthase is an essential enzyme. Constitutive expression of an antisense sequence for the chalcone synthase gene downregulated endogenous gene expression and the enzyme itself, producing various pigmentation patterns that correlated with the downregulation strength [79.van der Krol A.R. et al.An antisense chalcone synthase gene in transgenic plants inhibits flower pigmentation.Nature. 1988; 333: 866-869Crossref Scopus (323) Google Scholar]. This pivotal work opened the floodgates for crop improvement using antisense-based transgenic strategies [80.Mol J.N.M. et al.Regulation of plant gene expression by antisense RNA.FEBS Lett. 1990; 268: 427-430Crossref PubMed Scopus (92) Google Scholar]. It is interesting to note the inversion between research and development in the history of antisense RNA (i.e., the technology was developed much ahead of any meaningful basic knowledge of the process). In fact, in addition to poor knowledge of natural antisense RNAs, the mechanistic understanding of how artificial antisense RNAs inhibit gene expression was over a decade away from their development as a technology. In 1998, it was discovered that dsRNA formation by sense–antisense RNA pairs was a substrate for RNase degradation, triggering gene silencing [81.Waterhouse P.M. et al.Virus resistance and gene silencing in plants can be induced by simultaneous expression of sense and antisense RNA.Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 13959-13964Crossref PubMed Scopus (856) Google Scholar]. Thus, the decade-long excitement around antisense RNAs was based on gene silencing. Furthermore, antisense RNA was very inefficient as a tool for post-transcriptional gene silencing (PTGS), as only a small proportion of transgenic plants would produce silenced individuals, and its use was rapidly superseded by more efficient approaches, such as expression of intron-spliced RNA with hairpin structure [82.Smith N.A. et al.Total silencing by intron-spliced hairpin RNAs.Nature. 2000; 407: 319-320Crossref PubMed Scopus (658) Google Scholar]. However, the assumption that plant cis-NATs function similarly to their artificial counterparts in gene silencing persisted [83.Yamada K. Empirical analysis of transcriptional activity in the Arabidopsis Genome.Science. 2003; 302: 842-846Crossref PubMed Scopus (720) Google Scholar], even after evidence for roles in other processes in animals [84.Werner A. Natural antisense transcripts.RNA Biol. 2005; 2: 53-62Crossref PubMed Google Scholar] and evidence that cis-NATs’ primary role(s) might not involve transcript cleavage in humans [85.Faghihi M.A. Wahlestedt C. RNA interference is not involved in natural antisense mediated regulation of gene expression in mammals.Genome Biol. 2006; 7: R38-9Crossref PubMed Scopus (52) Google Scholar]. The realization that cis-NATs are widespread in plants and the fact that dsRNA is the substrate for RNA interference (RNAi) generation triggered an interest in understanding the role of cis-NATs in gene silencing. However, growing evidence suggests a broader range of roles for cis-NATs, in which the formation of natural antisense short interfering RNA (nat-siRNA) might play a minor role. cis-NAT–mRNA pairs are not strong sources of short interfering RNA (siRNA) production, and in fact, compared with nonoverlapping mRNAs, cis-NAT–mRNA pairs often generally show lower, not higher, siRNA density [7.Henz S.R. et al.Distinct expression patterns of natural antisense transcripts in Arabidopsis.Plant Physiol. 2007; 144: 1247-1255Crossref PubMed Scopus (74) Google Scholar]. In arabidopsis and rice, only ~100 cis-NAT–mRNA pairs produce appreciable amounts of siRNAs from the overlapping region [8.Zhang X. et al.Genome-wide analysis of plant nat-siRNAs reveals insights into their distribution, biogenesis, and function.Genome Biol. 2012; 13: R20Crossref PubMed Scopus (84) Google Scholar]. Analysis of siRNA showed less than 5% of arabidopsis cis-NATs can potentially generate detectable 24-nt nat-siRNAs or 21-nt phased siRNAs [8.Zhang X. et al.Genome-wide analysis of plant nat-siRNAs reveals insights into their distribution, biogenesis, and function.Genome Biol. 2012; 13: R20Crossref PubMed Scopus (84) Google Scholar, 9.Wang X.-J. et al.Genome-wide prediction and identification of cis-natural antisense transcripts in Arabidopsis thaliana.Genome Biol. 2005; 6: R30Crossref PubMed Google Scholar, 10.Bazin J. et al.Global analysis of ribosome-associated noncoding RNAs unveils new modes of translational regulation.Proc. Natl. Acad. Sci. 2017; 114: E10018-E10027Crossref PubMed Scopus (71) Google Scholar, 11.Li S. et al.Integrated detection of natural antisense transcripts using strand-specific RNA sequencing data.Genome Res. 2013; 23: 1730-1739Crossref PubMed Scopus (41) Google Scholar]. Counterintuitively, the accumulation of cis-NATs is not globally affected in siRNA biogenesis mutants [7.Henz S.R. et al.Distinct expression patterns of natural antisense transcripts in Arabidopsis.Plant Physiol. 2007; 144: 1247-1255Crossref PubMed Scopus (74) Google Scholar]. Furthermore, the biogenesis pathways for the few nat-siRNA production shown to impact cognate sense mRNA are strikingly heterogeneous, with one being dependent on RNA polymerase IVa [12.Swiezewski S. et al.Small RNA-mediated chromatin silencing directed to the 3’ region of the Arabidopsis gene encoding the developmental regulator, FLC.Proc. Natl. Acad. Sci. 2007; 104: 3633-3638Crossref PubMed Scopus (106) Google Scholar] and most implicating either DCL1 or DCL2, but not DCL3 [8.Zhang X. et al.Genome-wide analysis of plant nat-siRNAs reveals insights into their distribution, biogenesis, and function.Genome Biol. 2012; 13: R20Crossref PubMed Scopus (84) Google Scholar,13.Borsani O. et al.Endogenous siRNAs derived