High throughput transcriptome studies in the last decade have led to the discovery of 1000s of novel RNA molecules that do not appear to code for functional peptides. These transcripts, named the long non-coding RNAs (lncRNAs), can be found in both eukaryotes and prokaryotes. However, they are especially abundant in more complex eukaryotes including animals and plants (Rinn and Chang, 2012; Morris and Mattick, 2014). Interestingly, their abundance seems to correlate with the level of complexity of the organism, for example while protein-coding sequences constitute < 2% of the human genome, a significantly larger fraction of our genomes is coding for lncRNAs (ENCODE Project Consortium, 2012; Clark et al., 2013). lncRNAs have diverse modes of biogenesis and mechanism of action and constitute a heterogeneous group of transcripts. Many of them are transcribed by RNA polymerase II and are spliced and polyadenylated, although on average they contain fewer introns compared to protein-coding genes (Carninci et al., 2005; Derrien et al., 2012; Niazi and Valadkhan, 2012). However, several studied lncRNAs are RNA polymerase I and III transcripts and a significant fraction of lncRNAs are not polyadenylated (Carninci et al., 2005; Derrien et al., 2012; Djebali et al., 2012). Similarly, they show a wide heterogeneity in size and can range from a couple of hundred to tens of thousands of nucleotides in length. An arbitrary lower length limit of 200 nucleotides which has been proposed for this class of RNAs should not be applied too strictly, as it mainly serves as a general guideline for distinguishing lncRNAs from the small non-coding RNA such as small nuclear and nucleolar RNAs, microRNAs, etc. (Clark and Mattick, 2011; Rinn and Chang, 2012; Mattick and Rinn, 2015).

Despite intense research in recent years, the sheer number and diversity of lncRNAs have limited our understanding of the scope of function of lncRNAs in higher eukaryotes. Evidence emerging from existing research indicate that the expression of many lncRNAs is strongly dependent on the cell type and cellular state and is tightly controlled by various cellular signals (Rinn and Chang, 2012; Amaral et al., 2013). Thus, it is likely that many lncRNAs remain to be described and annotated as more cellular states are probed with high throughput transcriptomic studies. Even among the annotated lncRNAs, the vast majority remain unstudied. Finally, many alternatively processed unstudied isoforms of protein-coding RNAs do not have a significant protein-coding capacity and thus, can potentially affect the cellular function as non-coding RNAs (Carninci et al., 2005; Djebali et al., 2012; ENCODE Project Consortium, 2012). Therefore, it is likely that future research will elucidate a much larger share for lncRNAs in cellular function than currently assumed.

Nonetheless, data from the small number of lncRNAs currently studied indicate their involvement in diverse aspects of cellular function (Wapinski and Chang, 2011; Moran et al., 2012; Rinn and Chang, 2012; Amaral et al., 2013; Ulitsky and Bartel, 2013; Yang et al., 2014). Emerging data suggest that a major functional mechanism of lncRNAs involves regulation of nuclear events, including transcriptional regulation and control of the epigenetic state of chromatin (Rinn and Chang, 2012; Amaral et al., 2013; Rinn, 2014). Accordingly, localization of a lncRNA to the nuclear compartment and association with chromatin modifying complexes or transcription factors serve as potential clues into the function of a lncRNA.

Another potential indicator of the function of a lncRNA is the genomic locus from which it originates (Figure 1). Genes coding for lncRNAs frequently overlap protein-coding genes or other lncRNA genes in the sense or antisense orientation with the lncRNA initiating from a different promoter. In many cases, such promoters are located within or near the 3′ UTR of protein-coding genes (Carninci et al., 2005; Derrien et al., 2012; Djebali et al., 2012). Further, many lncRNAs result from selective stabilization of a certain region of a protein-coding gene, such as the 3′ UTR (Mercer et al., 2011). Such overlapping genes may affect the biogenesis and/or function of the other genes in the locus via several potential mechanisms, ranging from epigenetic regulation of the activity of the locus, to transcriptional interference, to masking or competing with functionally critical elements of the other transcripts arising from the overlapped locus through base-pairing (Valadkhan and Nilsen, 2010). Over 10% of human genes originate from the so-called bidirectional promoters, and in many cases at least one of the two promoter-sharing genes is a lncRNA (Figure 1) (Adachi and Lieber, 2002; Wakano et al., 2012; Uesaka et al., 2014). Data from studied examples has shown that in such loci, one member of the pair can regulate the expression of the other transcript which originates from the same promoter (Wei et al., 2011; Uesaka et al., 2014). Another subclass of lncRNAs, those originating from promoters in enhancer loci, are thought to be crucial for the function of the enhancers from which they are transcribed (Figure 1) (Lam et al., 2014). Finally, many lncRNAs are located in vicinity of other genes without overlapping them or sharing architectural elements with them (Figure 1). Such intergenic vicinal lncRNAs can potentially affect the expression of their nearby genes via transcriptional interference or epigenetic regulation (Valadkhan and Nilsen, 2010; Rinn and Chang, 2012; Mattick and Rinn, 2015). It should, however, be mentioned that many functional lncRNAs are located far away from their target genes and thus, locus proximity or even overlap is not necessarily a requirement for a regulatory relationship. On the other hand, some lncRNAs may function in cis to regulate the expression of genes located far in the genome but in the same nuclear territory (Hacisuleyman et al., 2014).

As mentioned above, current evidence point to the involvement of lncRNAs in virtually every aspect of cellular function. Several pioneering studies in recent years have provided evidence for the critical role of lncRNAs in regulation of diverse aspects of the immune response, including both innate and adaptive immunity (Fitzgerald and Caffrey, 2014; Heward and Lindsay, 2014; Imamura and Akimitsu, 2014; Stachurska et al., 2014; Marques-Rocha et al., 2015; Satpathy and Chang, 2015; Sigdel et al., 2015; Yu A.D. et al., 2015; Yu Q. et al., 2015). While the main focus of this review is on the host-derived lncRNAs involved in the interferon arm of the antiviral response, recent research has revealed that a number of virally coded lncRNAs also affect the IFN response (summarized in Valadkhan and Gunawardane, 2015). The following sections include a discussion of the host-derived lncRNAs induced by and/or affecting the interferon response, followed by a more focused overview of the role of lncRNAs in hepatitis C infection as a well-studied example of a viral infection. Comprehensive reviews on the role of lncRNAs in development and function of the immune response are included elsewhere (Valadkhan and Gunawardane, 2015; Carpenter, 2016; Atianand et al., 2017).

A crucial and nearly ubiquitous component of the innate immune response against viruses and many other microbial pathogens is mediated through the interferon (IFN) signaling cascade. IFNs are traditionally divided into three major groups (I, II, and III). While binding distinct receptors, type I (IFN-α, -β, -κ, -𝜀, and -ω) and type II (IFN-γ) IFNs show extensive overlap in their downstream signaling cascades and regulated genes (Hertzog et al., 2011; Pollard et al., 2013; Rusinova et al., 2013; Bolen et al., 2014). Similarly, type III IFNs (IFN-λ1, IFN-λ2, IFN-λ3, also called IL-29, IFN-λ4, IL-28A, and IL-28B, respectively), which are predominantly expressed in plasmacytoid dendritic cells in addition to a number of other cell types, bind a distinct receptor but show downstream overlap with type I IFNs (Meyer, 2009; Pollard et al., 2013; Bolen et al., 2014). The IFN response is initiated through the induction of expression of IFNs via activation of a class of cellular factors that act as the initial sensors of pathogen-associated molecular patterns. Among them, RNA sensors such as retinoic acid inducible gene I (RIG-I), melanoma differentiation-associated gene-5 (MDA5) and membrane-bound Toll-like receptors (TLR 3, 7, or 8) are particularly relevant to antiviral response against RNA viruses, although cells also contain several DNA sensors (Figure 2) (Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014). Activation of the sensor molecules, in turn, leads to signal transduction cascades and ultimately induction of the expression of genes with specific inhibitory functions against micro-organisms.

In the case of the type I IFN response, recognition of pathogen-associated patterns by RIG-I and MDA5 leads to activation of the IFN-β promoter stimulator 1 (IPS-1), while TLR3 and TLR7/TLR8 induce signaling via TRIF and MyD88, respectively (Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014). The activation of these pathways, in turn, will result in phosphorylation and dimerization of transcription factors including the interferon regulatory factor 3 (IRF-3), and IRF-7 followed by their nuclear translocation (Figure 2). Next, together with NF-κB and ATF-2/c-jun, they nucleate the formation of active transcriptional complexes through interactions with several transcription factors and regulatory proteins such as the transcriptional cofactor (CREB)-binding protein (CBP)/p300 (Kawai and Akira, 2006; Hiscott, 2007; Randall and Goodbourn, 2008; Takeuchi and Akira, 2009; Onomoto et al., 2010; Jensen and Thomsen, 2012). This, in turn, results in induction of the expression of antiviral cytokines including type I IFNs (IFN-α and IFN-β). While IFN-β is produced in most cells, IFN-α is predominantly produced in hematopoietic cells such as monocyte/macrophages and dendritic cells, especially the plasmacytoid dendritic cells. IFN-γ is induced in response to different extracellular signals, including interleukins IL2, IL12, and IL18, preferentially, in NK and T cells (Green et al., 2017). These cytokines activate JAK/STAT, NF-κB, JNK, ERK and p38 MAPK pathways, depending on the cell line, and induce transcription of IFN-γ in response to NFAT, NF-κB, STAT and AP1 transcription factors.

IFN-α and -β bind to the IFN-α/β receptor (IFNAR) in an autocrine and paracrine manner, triggering the induction of the Janus kinase/signal transducers and activators of transcription (JAK/STAT) signaling cascade (Figure 2) (Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014). The outcome is recruitment, followed by phosphorylation, dimerization and finally nuclear translocation of STAT1 and STAT2. Once in the nucleus, the complex of STAT1 and STAT2 is bound to IRF9/p48, forming the IFN-stimulated gene factor 3 (ISGF3) complex which induces the transcription of hundreds of IFN-stimulated genes (ISGs) (Figure 2) (Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014). In a highly analogous manner, IFN-γ, the sole member of the type II IFN family, binds the IFN-γ receptor and activates signaling through the JAK/STAT pathway resulting in STAT1 phosphorylation followed by nuclear translocation (Hu and Ivashkiv, 2009). Once activated, STAT1 binds the gamma-activated sequence (GAS) close to its target genes as a homodimer, triggering the transcriptional induction of STAT1-regulated genes. As mentioned above, the majority of IFN-γ-regulated genes are also induced by type I IFNs and thus, type I and type II IFN response show significant overlap in terms of the downstream effects (Hu and Ivashkiv, 2009; Pollard et al., 2013).

The final outcome of the IFN signaling is the induction of ISGs, many of which exert antimicrobial activity via regulation of different steps of cellular gene expression in addition to direct antimicrobial effects. For example, several ISGs including IFIT-1, IFIT-2, IFITM3, ISG15, ISG20, RNase L, PKR, viperin and BST2/Tetherin are known to have antiviral activity (Randall and Goodbourn, 2008). Thus, activation of the IFN response results in the induction of a cell-intrinsic antimicrobial state in both the infected and the neighboring cells which limits the spread of infectious agents. Further, these potent cytokines also affect both the innate and adaptive immune response through promoting antigen presentation, natural killer cell function, production of antibody in B cells and T cell effector function (Ivashkiv and Donlin, 2014). As mentioned above, most cells are able to launch the type I IFN response, however, cell type and context have a strong effect on the magnitude of the induced response and the specific subsets of effector genes activated (Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014). On the other hand, the IFN response itself is subject to negative feedback regulation by a number of ISGs and other cellular signaling pathways (Yoshimura et al., 2007; Hertzog and Williams, 2013; Ivashkiv and Donlin, 2014; Schneider et al., 2014; Porritt and Hertzog, 2015).

While the IFN cascade is mainly activated in response to the presence of exogenous viral RNAs, a group of long non-coding cellular RNAs, namely the retrotransposon-derived transcripts, also activate the IFN response (Yu Q. et al., 2015). It has been shown that activation of the Long Interspersed Element-1 (LINE-1) retroelements results in induction of the expression of IFN-β and ISGs (Yu Q. et al., 2015). Prior treatment of cells with IFN leads to suppression of LINE-1 replication, and mutations that inactivate different steps of the IFN signaling pathway cause an increase in LINE-1 replication (Yu Q. et al., 2015). Although the mechanism behind the induction of the IFN response by LINE-1 replication is not known, it is likely that the replication state of LINE elements and other retrotransposons are detected by either the same or similar sensors to those that detect the presence of exogenous viral RNA in the cell. Existing data suggest that the replication of another class of retrotransposons, the Small Interspersed Elements (SINEs) similarly results in activation of the IFN response (Leonova et al., 2013). Interestingly, it has been suggested that LINE-1 elements may contribute to the pathogenesis of autoimmune disorders such as Systemic Lupus Erythematosus (SLE) (Crow, 2010; Nakkuntod et al., 2011), which is known to involve activation of the IFN response (Bronson et al., 2012; Elkon and Wiedeman, 2012). Additional studies of the connection between autoimmunity and the replication state of cellular repeat elements will likely provide novel insight into the mechanistic basis of development of autoimmune diseases.

A number of high throughput transcriptomic analyses in both human and mouse have revealed the presence of strong changes in the expression of lncRNAs during the IFN response (Peng et al., 2010; Carnero et al., 2014; Josset et al., 2014; Kambara et al., 2014, 2015; Barriocanal et al., 2015). In the mouse studies, which were performed in both mouse lung tissue and cultured mouse embryonic fibroblasts, a significant fraction of the differentially expressed genes belonged to lncRNAs, which were either up or down-regulated after interferon stimulation and infection with respiratory viruses (Peng et al., 2010; Josset et al., 2014). Interestingly, promoter analysis and expression correlation studies raised the possibility that a significant fraction of the induced lncRNAs may be novel ISGs that were directly induced through the IFN signaling cascade (Peng et al., 2010; Josset et al., 2014).

The results obtained in the murine models were closely analogous to those obtain using human cells. In one such study, primary human hepatocytes from five donors of different ages and genders were subjected to high throughput transcriptome analysis before and at three time points of 3, 9, and 24 h after treatment with IFN-α (Kambara et al., 2014). Similar to what had been observed in the mouse, IFN stimulation resulted in both induction and repression of expression of a large number of lncRNAs. Many of the induced lncRNAs were likely direct targets of the JAK-STAT signaling pathway and thus, novel ISGs. While the majority of the induced lncRNAs remained upregulated for the three time points analyzed in this study, a fraction of them showed a shorter duration of induction and were limited to the earlier time points of the study (Kambara et al., 2014).

A third set of high throughput studies in HuH7 human hepatocytes focused on changes in gene expression observed at later time points (after 72 h) of treatment with a high dose of IFN-α2 using both microarray analysis (Carnero et al., 2014) and RNA-seq (Barriocanal et al., 2015). Interestingly, similar to the results of the study by Kambara et al. (2014), the lncRNAs that showed differential expression were almost equally divided between upregulated and downregulated ones in both microarray and RNA-seq-based datasets, while almost all differentially expressed protein coding genes were upregulated. In addition, both sets of studies revealed a significant number of differentially expressed lncRNAs which were either vicinal to or overlapped protein-coding genes that function in the immune response (Carnero et al., 2014; Kambara et al., 2014, 2015; Barriocanal et al., 2015). As mentioned above, in some cases the function of a lncRNA depends on its genomic locus and thus, these studies potentially point to the presence of an IFN-activated regulatory network of lncRNAs which may function in fine-tuning the immune response.

Hepatitis C (HCV) is a deadly virus that affects ∼2% of the world’s population (60–170 million people) (Manns et al., 2017). About 500,000 people die every year from HCV-related diseases (Webster et al., 2015; Bukh, 2016). After the initial infection, most patients develop an asymptomatic chronic infection that causes liver damage and therefore, may progress to liver steatosis, fibrosis, cirrhosis and hepatocellular carcinoma (HCC). In the course of 30 years of HCV infection, ∼20% of patients will develop cirrhosis. In fact, HCV infection is nowadays one of the major inducers of liver cirrhosis. As HCV infection induces the expression of several oncogenic factors and liver cirrhosis is a niche for HCC, ∼3–7% of cirrhotic patients will develop liver tumors every year (Llovet and Villanueva, 2016).

In spite of the efforts of many groups, an effective vaccine to prevent HCV infection has not yet been developed. Recently, several inhibitors that target different viral proteins have been approved for the treatment of HCV infection (D’Ambrosio et al., 2017). The efficacy of these treatments grazes a complete sustained viral response, allowing the claim that HCV infection can be cured (Bourlière et al., 2017). A fraction of the small number of patients that resist the treatment and do not show viral clearance have high levels of liver cirrhosis. Some of the patients with HCV infection and HCC show a transient response with a later relapse of the virus, probably because HCC architecture and/or composition impairs drug penetration and serves as a reservoir for infected cells (Prenner et al., 2017). Some authors have observed an increased risk of HCC recurrence in patients that have cleared HCV infection (Llovet and Villanueva, 2016; Reig et al., 2016). In spite of this, there are good reasons for optimism. After sustained viral responses, liver fibrosis can regress, the risk of cirrhosis-related complications, including HCC, is reduced and the overall survival of the patients increases (van der Meer and Berenguer, 2016; Nahon et al., 2017; van der Meer et al., 2017). However, there is still a long way to achieve the goal of HCV cure. Patients cured of chronic HCV infection are at risk for reinfection if exposed. Work toward the development of an effective vaccine to prevent HCV infection must be intensified to cure the disease (Bukh, 2016). Research about HCV infection should not stop and a rigorous epidemiological follow up should be carried out given the high prevalence of HCV infection and the ability of HCV to generate escape mutants resistant to the treatments (Ramirez et al., 2016).

In vitro studies on the role of lncRNAs in the IFN response in the absence of pathogens point to the presence of a strong perturbation in the expression of the non-coding transcriptome following IFN stimulation. From the very small number of studies on the role of lncRNAs in the antiviral response, it is evident that changes in the expression of this class of cellular effectors do play a critical role in negative regulation of the downstream steps of the IFN response. Together with proteins that act as negative regulators, this class of lncRNAs create a complex, decentralized regulatory network with overlapping and likely partially redundant functions which allows for extreme fine tuning of the magnitude and duration of the IFN response. However, in the presence of pathogens such as HCV, such negative feedback loops can favor viral replication and in some cases are actively hijacked by the virus to help in viral survival. There are two possible scenarios that can explain this from an evolutionary perspective. One possibility is that the co-evolution of the IFN response and pathogens has tipped in favor of viral survival due to the faster rate of viral evolution, with the viruses gaining the ability to exploit the immune response. Alternatively, it is plausible that at least in the case of some chronic infections, such a reduction in the magnitude of the immune response may be beneficial for the organism through limiting the damage incurred on the infected tissue. Future studies, by providing a more complete picture of the interplay of lncRNAs and viral infections, will shed additional light on the highly complex interplay of the antiviral response and pathogens.

SV and PF wrote and reviewed the manuscript. PF built the table and SV the figures.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.