New Insights into the Role of RNase L in Innate Immunity

Abstract

The interferon (IFN)-inducible 2′-5′-oligoadenylate synthetase (OAS)/RNase L pathway blocks infections by some types of viruses through cleavage of viral and cellular single-stranded RNA. Viruses induce type I IFNs that initiate signaling to the OAS genes. OAS proteins are pathogen recognition receptors for the viral pathogen-associated molecular pattern, double-stranded RNA. Double-stranded RNA activates OAS to produce p_x5′A(2′p5′A)_n; x = 1–3; n > 2 (2-5A) from ATP. Upon binding 2-5A, RNase L is converted from an inactive monomer to a potently active dimeric endoribonuclease for single-stranded RNA. RNase L contains, from N- to C-terminus, a series of 9 ankyrin repeats, a linker, several protein kinase-like motifs, and a ribonuclease domain homologous to Ire1 (involved in the unfolded protein response). In the past few years, it has become increasingly apparent that RNase L and OAS contribute to innate immunity in many ways. For example, small RNA cleavage products produced by RNase L during viral infections can signal to the retinoic acid-inducible-I like receptors to amplify and perpetuate signaling to the IFN-β gene. In addition, RNase L is now implicated in protecting the central nervous system against viral-induced demyelination. A role in tumor suppression was inferred by mapping of the RNase L gene to the hereditary prostate cancer 1 (HPC1) gene, which in turn led to discovery of the xenotropic murine leukemia-related virus. A broader role in innate immunity is suggested by involvement of RNase L in cytokine induction and endosomal pathways that suppress bacterial infections. These newly described findings about RNase L could eventually provide the basis for developing broad-spectrum antimicrobial drugs.

Introduction

The 2′-5′-oligoadenylate synthetase (OAS)/RNase L system was one of the first antiviral pathways to be discovered during investigations in the mid-1970s on how interferon (IFN) inhibits viral infections (Hovanessian and others 1977; Clemens and Williams 1978; Kerr and Brown 1978; Ratner and others 1978; Slattery and others 1979). IFNs induce transcription of a family of OAS genes in cells of higher vertebrates (see article by H. Kristiansen and others 2010, this issue). In addition, RNase L is IFN inducible in mouse cells, but only minimally if at all in human cells (Jacobsen and others 1983; Rusch and others 2000). OAS proteins are pathogen recognition receptors (PRR) for the viral pathogen-associated molecular pattern, double-stranded RNA (dsRNA). Once stimulated by dsRNA, OAS uses ATP to synthesize 2-5A molecules of the formula [p_x5′A(2′p5′A)_n; x = 1–3; n > 2], yielding mainly the 5′-triphosphorylated triadenylate (Fig. 1). At subnanomolar levels, 2-5A activates RNase L to cleave single-stranded RNA (ssRNA) in U-rich sequences, typically after UU or UA dinucleotides leaving a 5′-OH and 3′-monophosphate (Floyd-Smith and others 1981; Wreschner and others 1981b). Murine OAS1b, also known as the flavivirus resistance gene (Flv^r), does not synthesize 2-5A but has an alternative antiviral mechanism of action (Kakuta and others 2002; Mashimo and others 2002; Perelygin and others 2002). In addition, the classical OAS/RNase L pathway has an antiviral effect in cells from both Flv^r (resistant) and Flv^s (susceptible) mice (Scherbik and others 2006). RNase L is negatively regulated by RNase L inhibitor/ATP-binding cassette, sub-family E member 1 [RLI (ABCE)], which inhibits rRNA cleavage by RNase L (Bisbal and others 1995; Malathi and others 2004) and by the 2′-phosphodiesterase that degrades 2-5A (Kubota and others 2004). Beyond its antiviral function, RNase L has recently been implicated in innate immunity against bacterial infections (Li and others 2008). We begin with a review of the structure and function of RNase L. From there we highlight and discuss new findings on the role of RNase L in innate immunity and tumor suppression published in the last 3 years (for a review of prior literature see Silverman 2007).

FIG. 1.

Role of RNase L in innate immunity against viruses and bacteria. The diagram illustrates how RNase L is activated in response to viral double-stranded RNA activating OAS to produce 2-5A from ATP. The RNase L domains shown are “A,” ARD; “K,” PK-like; and “N,” RNASE. 2-5A binding to the ARD causes inactive RNase L monomers to form activated dimers that degrade viral and cellular single-stranded RNA. RNA cleavage products activate RIG-I to amplify production of IFN-β. Bacterial infections signal through RNase L to inhibit Cathepsin E synthesis enhancing levels of lysosome-associated membrane proteins, LAMP 1 and 2, required for the terminal step in phagosome maturation and destruction of the microbial contents. NOD2 interacts with OAS2, whereas NOD2 overexpression leads to enhanced RNase L activity in the presence of double-stranded RNA. NOD2 responds to bacterial MDP triggering NF-κB activation contributing to type I IFN, IL-1β and TNF-α synthesis. ARD, ankyrin repeat domain; IFN, interferon; MDP, muramyl dipeptide; NOD2, nucleotide-binding and oligomerization domain-2; OAS, 2′-5′-oligoadenylate synthetase; PK, protein kinase; RIG-I, retinoic acid-inducible gene-I.

Biochemistry of RNase L

RNase L is a highly regulated, latent endoribonuclease (hence the “L” in RNase L) first cloned in 1993 when it was referred to as the 2-5A-dependent RNase (Zhou and others 1993). It is widely expressed in most, if not all, mammalian tissues (Zhou and others 2005), lying in wait for virus to infect and produce dsRNA, which triggers the pathway through OAS activation. RNase L is composed of 3 major domains: an N-terminal regulatory ankyrin repeat domain (ARD), a protein kinase (PK)-like domain, and an C-terminal ribonuclease domain (RNASE) (Fig. 2). The PK and RNASE domains [collectively referred to as the kinase-extenstion-nuclease (KEN) domain] have homology with Ire1, both a kinase and an endoribonuclease, that functions in the unfolded protein response (UPR) from yeast to humans (Sidrauski and Walter 1997; Lee and others 2008; Korennykh and others 2009). Analogous to the ∼120-amino-acid long linker domain that tethers the KEN domain to the transmembrane region in Ire1, there exist a much shorter, ∼30 amino acid, linker between ARD and PK homology domain of RNase L. Within the RNASE domain there is also a PUG (or PUB) domain similar to that in peptide N-glycanase, which removes glycans from misfolded glycoproteins, also present in some ubiquitin-protease system proteins (Allen and others 2006) and a possible protein–protein interaction domain (Suzuki and others 2001) (Fig. 2). There are other proteins that have both ARD and PK domains, such as integrin-linked kinase (Hannigan and others 1996), and death associate PK, DAP kinase (Bialik and others 2004). As mentioned, Ire1 proteins have both PK and RNASE domains (Sidrauski and Walter 1997), but RNase L is the only protein yet identified that has all 3 domains (ARD, PK, and RNASE). RNase L sequences are currently known for a wide range of mammalian species and 1 avian species (chicken, Gallus gallus), allowing for comparisons (Table 1). In general, the ARD is more highly conserved than the KEN domain. For instance, the Gallus RNase L has 47% and 32% similarity to the ARD and KEN domain of human RNase L, respectively. Sequence analysis of RNase L using GlobPlot 2 (Linding and others 2003) from different species predicted intrinsic disordered segments, ∼20 amino acids, at both the N- and C-termini (Fig. 1). These disordered segments are dispensable for human RNase L function (ie, 2-5A binding and ribonuclease activity) (Dong and Silverman 1997).

FIG. 2.

Domain structure of RNase L. The main structural and functional domains of RNase L are shown, including ankyrin repeat domain, ARD; protein kinase-like domain, PK; ribonuclease domain, RNASE; kinase-extension-nuclease domain, KEN, and peptide N-glycanase/UBA or UBX-containing protein domain, PUG or PUB. The region responsible for binding 2-5A is indicated.

Table 1.

Homology Between Different Mammalian and Avian Species of RNase L Protein

	Identity in domains of RNase L from different species compared with human RNase L (%)
Species	Full length (aa 1–741)a	ARD (aa 1–335)a	KEN (aa 336–741)
Homo sapiens	100	100	100
Pan troglodytes	97	95	98
Pongo abelii	95	94	96
Macaca mulatta	92	91	93
Equus caballus	77	78	77
Canis lupus familiaris	71	72	70
Bos taurus	71	73	70
Sus scrofa	70	69	71
Loxodonta africana	67	68	66
Cavia porcellus	65	64	66
Mus musculus	64	63	65
Rattus norvegicus	62	64	62
Gallus gallus	39	47	32

^aNumbers of amino acid (aa) are from human RNase L Acc No. Q05823 (www.ncbi.nlm.nih.gov/protein/Q05823.2). The protein sequences were obtained from the NCBI database (www.ncbi.nlm.nih.gov/sites/entrez?db=cdd), and align using Clustlaw 2.0 (www.ebi.ac.uk/Tools/clustalw2/index.html). The percent identities were rounded to the nearest whole number.

ARD, ankyrin repeat domain; KEN, kinase-extenstion-nuclease.

Ankyrin repeats are one of the most common amino acid motifs, typically functioning in mediating protein–protein interactions, and are conserved in all kingdoms of life (reviewed in Barrick and others 2008). They are 33 amino acid units of paired antiparallel α-helices connected by a β-hairpin turn (Gorina and Pavletich 1996). RNase L has 8 complete ankyrin repeats and 1 partial repeat appearing as a disordered segment in the crystal structure of the N-terminal (amino acid 1–333) of human RNase L (Tanaka and others 2004). Ankyrin repeats 2 to 4 in RNase L constitute the 2-5A binding domain, a unique function among ARDs. In the absence of 2-5A, the ARD represses the RNASE domain. 2-5A binding to the ARD is believed to alter the conformation of RNase L, thereby exposing protein–protein interaction domains and releasing the RNASE domain from internal inhibitory sequences (Dong and others 2001) (Fig. 1). However, the precise activation mechanism will not be known until structures of full-length RNase L (as a free protein and in a complex with 2-5A) are solved. Removing ankyrin repeat 7 or 8, or Lys to Asn mutations at positions 240 and 274 in ARDs 7 and 8, respectively (Zhou and others 1993; Dong and Silverman 1997), result in loss of 2-5A binding and enzyme activities, suggesting that these changes affect the structural integrity of the 2-5A binding domain. Amino acid residues W60, N65, Q68, and K89 in repeat 2 and F126, E131, and R155 in repeat 4 directly interact with 2-5A either via H-bonding or by stacking interactions (Tanaka and others 2004). All the above residues are conserved among all known protein sequences of RNase L, except asparagine-65, which is substituted with a serine in Gallus RNase L. Mutagenesis studies in ARD 2 and 4, coupled with biochemical analysis, indicated that each of K89A, F126A, E131A, and R155A results in loss of 2-5A binding and enzyme activity, whereas N65A, Q68A, and Y135A had reduced 2-5A binding activity (Nakanishi and others 2005). Although RNase L has the only ARD known to bind to an oligonucleotide, the ARDs in transient receptor potential vanilloid (TRPV) channel proteins TRPV1, TRPV3, and TRPV4 have a multiligand binding site for ATP and calmodulin (Lishko and others 2007; Phelps and others 2010). Asparagine-233 in ankyrin repeat number 6 of human RNase L is hydroxylated by factor inhibiting hypoxia-inducible factor (HIF-1) alpha, factor inhibiting hypoxia-inducible factor (FIH), suggesting that RNase L stability and/or activity could be regulated in response to oxygen-related signals (Cockman and others 2009).

The KEN domain in RNase L functions in dimerization and catalysis (Dong and Silverman 1995, 1997; Carroll and others 1997). The RNASE domain becomes constitutively active upon removal of the ARD (although at 6-fold reduced activity compared with activated full-length protein) (Dong and Silverman 1997). Residues 365 to 586 in the KEN domain contain PK homology motifs (Fig. 1). RNase L is a pseudokinase due to lack of several amino acids required for an active kinase (including lack of critical residues in activation loop, substrate binding site, and active site). However, RNase L has highly conserved residues responsible for ATP binding and stimulation of enzyme activity (Wreschner and others 1982; Dong and others 1994; Lee and others 2008; Korennykh and others 2009). In the human kinome, the PK domain of RNase L is most closely related to that of protein kinase R (PKR) (Manning and others 2002). In addition to being related in their respective PK domains, PKR and RNase L are both antiviral enzymes whose levels and/or activities are regulated by IFN and viral dsRNA. Could this mean that the PK domains of PKR and RNase L evolved from a common ancestral precursor gene (Silverman and Williams 1999)?

Although RNase L shares some domain architecture with Ire1, there are both similarities and differences between these proteins. The N-terminal regions are unrelated, but both receive the activation signal (Korennykh and others 2009). In Ire1, activation is triggered by unfolded proteins through its endoplasmic reticulum (ER)-luminal domain that titrate off BiP, an ER chaperone, whereas RNase L activation occurs in the cytoplasm in response to 2-5A produced during viral infections. Both proteins are regulated at the level of dimerization (or oligomerization). Whereas a high-order assembly has been shown for Ire1, RNase L has thus far been shown to only form dimers (Dong and Silverman 1995; Carroll and others 1997; Korennykh and others 2009). In a highly unusual RNA splicing pathway, Ire1 cleaves pre-mRNAs for transcription factors HAC1 (yeast) or XBP1 (mammals) at unique sites (Korennykh and others 2009). HAC1 and XBP1 regulate UPR gene expression. In addition, Ire1 causes degradation of host mRNAs (mostly for membrane proteins), in a process named “regulated Ire1-dependent decay” (RIDD) (Hollien and others 2009). RIDD requires the nuclease activity of Ire1, but does not appear to be dependent on its kinase activity (Hollien and others 2009); in this respect, Ire1 is similar to RNase L. RNase L lacks PK activity and does not require phosphorylation for activation of its ribonuclease activity (Dong and Silverman 1999). Ire1 could have antiviral functions, beyond its role in the UPR, that overlap with RNase L. For example, RIDD was suggested to play a role in attenuating viral protein synthesis during hepatitis C virus (HCV) infections (Tardif and others 2004; Hollien and others 2009). Could RNase L be involved in RIDD during viral infections or could Ire1 play a role in the RNase L antiviral effect?

Role in IFN Induction

RNase L blocks different types of viruses (mostly RNA viruses) by different mechanisms depending the specific RNA substrates and the extent of ribonuclease activity (discussed in Silverman 2007). For instance, sustained RNase L activity is thought to eliminate virus-infected cells through apoptosis (Castelli and others 1997; Zhou and others 1997). Even cleavage of some cellular RNAs, such as rRNA in intact ribosomes, likely contributes to the antiviral activity of RNase L (Wreschner and others 1981a; Silverman and others 1983). Some of the RNA cleavage products resulting from RNase L activity, named “suppressor of virus RNA,” either viral or cellular in origin, can contribute to production of type I IFNs (Malathi and others 2007; Malathi and others 2010).

Viral RNAs are typically recognized by different PRRs (reviewed in Kumar and others 2009). For example, toll-like receptor 3 is a sensor for dsRNA in the endosomal compartment, whereas OAS is a PRR for viral dsRNA in the cytoplasm. Retinoic acid-inducible gene-I (RIG-I, also known as DDX58) and melanoma differentiation-associated gene-5 (MDA5, also known as IFIH1) are activated by 5′-triphosphorylated, double-stranded, or uridine and adenosine-rich viral RNAs in the cytoplasm (Saito and others 2008). Upon activation, the PRRs trigger signaling cascades that stimulate transcription of type I IFN genes (Kumar and others 2009). A role for RNase L in IFN induction was apparent from studies on mouse embryonic fibroblasts deficient in RNase L that showed reduced IFN-β production upon treatment with 2-5A, synthetic dsRNA [poly(rI):poly(rC)] and Sendai virus infection (Malathi and others 2007). The ribonuclease function of RNase L was essential for IFN induction with contributions from both RIG-I and MDA5. Total cellular RNA digested with RNase L or just the small cleaved RNA fragments, <200 nt, induced higher levels of IFN induction than uncleaved RNA. RNase L produces small RNA cleavage products with 3′-monophosphate (3′-p) and 5′-hydroxyl (5′-OH) at the termini. The 3′-p of the cleaved RNAs function in the recognition by RIG-I or MDA5, as calf alkaline phosphatase (CIP) treatment compromised their ability to induce IFN. Cellular observations were validated in mice where injection of 2-5A caused IFN-β induction in wild-type mice but not in mice deficient in RNase L. Moreover, mice lacking RNase L had several-fold reduced levels of IFN induction after infections with EMCV and Sendai virus. Therefore, effects of the OAS/RNase L pathway extend beyond initially infected cells to support a prolonged antiviral state in the organism. In essence, RNase L converted self-RNA into small RNA products that appeared to the host cell as nonself RNA. RNase L is also a critical component in IFN induction by a DNA virus, HSV-2 (Rasmussen and others 2009).

RNase L, Cancer, and Xenotropic Murine Leukemia-Related Virus

The cell growth inhibitory and proapoptotic effects of RNase L, and its chromosomal location at 1q25 mapping to some types of cancers, led to early speculation that RNase L could be a tumor suppressor (Hassel and others 1993; Lengyel 1993; Squire and others 1994; Castelli and others 1997; Zhou and others 1997). The idea gained support when the hereditary prostate cancer 1 (HPC1) gene was mapped to the RNase L gene (RNASEL) (Carpten and others 2002). HPC is apparent when there are 3 or more first degree relatives with the disease. Several germline mutations in RNASEL were observed in HPC, including M1I, E265X, 471ΔAAAG, and R462Q (reviewed in Silverman 2003). The missense variant, R462Q, mapping to the PK-like domain, reduces enzymatic activity by a factor of ∼3 (Xiang and others 2003) and doubled the risk of prostate cancer when homozygous (QQ) (Casey and others 2002). A homozyogous variant allele (rs12757998) downstream of the gene was associated with a significant increased risk of prostate cancer, including risk of higher grade tumors, in association with increased plasma biomarkers of inflammation, interleukin-6 (IL-6), C-reactive protein, and TNFR2 (Meyer and others 2010). It is unknown if this variant affects RNase L expression or if it is in linkage disequilibrium with a functional SNP that was not monitored. In the same study, the R462Q variant was not associated with advanced stage disease overall in the PSA era (post-1992); however, the QQ genotype was significantly associated with increased risk in the pre-PSA era, possibly reflecting diagnosis at a later stage in the disease (Meyer and others 2010).

Although some reports confirmed a role for RNASEL mutations in HPC or sporadic prostate cancer, others did not (see Silverman 2003; Meyer and others 2010 for discussion of recent literature). These studies suggest that environmental influences, such as microbial infections, might contribute to varying findings on RNase L/HPC1 in prostate cancer. Therefore, a clinical study was conducted to determine if a low activity variant of RNase L (R462Q) could be correlated to the presence of a viral infection in prostate cancer (Urisman and others 2006). One hundred fifty men with localized prostate cancer were genotyped for the RNase L variant, whereas RNA isolated from tumor-bearing prostate tissues was used to probe a pan-viral DNA microarray known as ViroChip (Wang and others 2002). A novel human retrovirus, the xenotropic murine leukemia virus-related virus (XMRV), was identified, mostly in men who were homozygous for the reduced activity variant of RNase L (Urisman and others 2006) and confirmed in (Arnold 2010) (reviewed in Silverman and others 2010). Another study on XMRV did not find an association with R462Q, although other mutations in the OAS/RNase L were not investigated (Schlaberg and others 2009). More recently, XMRV has been found in association with chronic fatigue syndrome in 1 study (Lombardi and others 2009), but not in others (Erlwein and others 2010; Groom and others 2010; van Kuppeveld and others 2010). Previously, RNase L was shown to be degraded in peripheral blood mononuclear cells (PBMC) of chronic fatigue syndrome (CFS) patients, providing a possible link between RNase L and XMRV infections in CFS (Suhadolnik and others 1997; Demettre and others 2002). RNase L was shown to be necessary for a complete IFN antiviral response against XMRV, thus supporting an association of RNase L mutation with XMRV in prostate cancer (Dong and others 2007). In addition, a 5′-untranslated region (UTR) mutation, rs3738579, in RNASEL associated with cancers of head and neck cancer, uterine cervix, and breast (Madsen and others 2008). The homozygous R462Q allele is also associated with disease aggressiveness and metastasis in familial pancreatic cancer (Bartsch and others 2005) and with age of onset of hereditary nonpolyposis colon cancer (Kruger and others 2005).

Protection Against Virus-Mediated Demyelination in the Central Nervous System

A novel protective role for RNase L in mitigating viral-induced demyelination in the central nervous system has been demonstrated showing that even when RNase L does not inhibit global viral replication it can still protect from virus-mediated disease. A sub-lethal, demyelinating mouse hepatitis virus (the neutropic coronavirus strain JHM) was lethal to a majority of RNase L-deficient mice by 12 days postinfection (Ireland and others 2009). Absence of RNase L in mice enhanced the morbidity rate without affecting overall viral replication in the brain. Further, RNase L deficiency did not impair type I IFN production or interferon stimulated gene (ISG) expression after viral infection, and neither did it alter the inflammatory response in central nervous system (CNS). Instead, there was early onset of severe demyelination with axonal damage in the brain stem and spinal cord of infected animals that lacked RNase L. Foci of infected microglia, sustained brain stem infection, and enhanced apoptosis were observed in the mice lacking RNase L. Therefore, RNase L prevents spread of virus to the microglia, leading to protection of the CNS from virus-induced demyelination. The authors suggest that by contributing to viral tropism in the CNS, RNase L affects the balance between neuroprotective and neutoxic effects of microglia.

A Region of Poliovirus RNA Is an RNase L Inhibitor

Poliovirus RNA is resistant to cleavage by RNase L due to a conserved RNA structure in the group C enteroviruses present in the 3C^Pro open reading frame (known as the “RNase L competitive inhibitor RNA” or RNase L ciRNA) (Han and others 2007; Townsend and others 2008a, 2008b). CiRNA does not affect 2-5A binding, but instead competitively inhibits the RNASE domain (Townsend and others 2008a). The RNase L inhibitory function of the ciRNA requires a putative loop E motif and an H-H kissing loop (Townsend and others 2008b).

Bacterial Targets

Although RNase L is usually thought of in terms of its antiviral functions, it also is protective in mice against infections by Bacillus anthracis and Escherichia coli (Li and others 2008) (Fig. 2). After infections with either type of bacteria, mice lacking RNase L had increased bacterial loads and higher mortality rates than identically infected WT mice. These findings were correlated to reduced levels of proinflammatory cytokines, IL-1β and TNF-α, and IFN-β after bacterial infection in the RNase L-deficient mice. Also, IRF3 dimerization was reduced 2-fold in E. coli–infected RNase L^−/− macrophages. Microarray analysis and subsequent experiments revealed a role of RNase L in regulation of Cathepsin E (Cat E), an endolysosomal aspartyl proteinase. Cat E mRNA was stabilized in macrophages lacking RNase L leading to increased levels in the protein. Elevated Cat E levels correlated with reduction of lysosome-associated membrane proteins, LAMP1 and LAMP2, required for the terminal step in phagosome maturation in which late endosomes fuse with lysosomes to eliminate phagocytosed microbial cargo. Decreased expression of LAMP1/2 in macrophages from RNaseL^−/− mice was correlated to accumulated phagocytic vacuoles and ineffective clearance of bacteria. In addition, overexpression of Cat E mimicked the absence of RNase L by impairing the induction of IL-1β after LPS treatment. The findings identify an essential role for RNase L in antibacterial immunity in which RNase L is required for the optimal induction of proinflammatory cytokines while also regulating Cat E, and associated endolysosomal functions, required for the elimination of phagocytosed bacteria. However, exactly how bacteria or LPS are signaling to RNase L, or whether, if fact, the ribonuclease activity of RNase L is required for its anti-bacterial role has not been reported. Curiously, while the IFN-inducible OAS/RNase L pathway is antibacterial, the type I IFNs themselves have been described in some studies to have probacterial effects (Rayamajhi and others 2009; Shahangian and others 2009).

A completely different line of investigation has indirectly linked the OAS/RNase L pathway to an innate immunity against bacterial infections. The nucleotide-binding and oligomerization domain-2 (NOD2) is a Nod-like receptor member that is activated by bacteria-derived muramyl dipeptide to trigger innate immunity by activating NF-kB and MAP kinases (reviewed in Ting and others) (Fig. 1). OAS2 p69 interacts with NOD2, leading to enhanced RNase L activity in poly(rI):poly(rC)-treated human acute monocytic leukemia cell line THP-1 (Dugan and others 2009). In addition, NOD2 recognizes viral ssRNA (from respiratory syncytial virus) and uses NOD2 to activate IRF3 (Sabbah and others 2009).

Cross-Regulation of HuR and RNase L Impacts mRNA Turnover

Cellular levels of RNase L are controlled in part by regulated turnover of its mRNA through sequences in its 3′-UTR and proteins that interact with these elements (Li and others 2007). The RNase L 3′-UTR contains 8 AU-rich elements (ARE), deletion of which identified both positive and negative regulatory effects. The RNase L 3′-UTR acted in cis to destabilize a heterologous mRNA for β-globin; therefore, the overall effect of the RNase L 3′-UTR was to decrease the half-life of the RNA. However, AREs 7 and 8 exerted a positive, or stabilizing effect on the RNase L mRNA. In addition, expression of the ARE-binding protein, HuR, enhanced RNase L mRNA and protein levels through 3′-UTR sequences between and including AREs 7 and 8. HuR binds RNase L mRNA during myoblast differentiation as determined in RNP immunoprecipitations. Cellular stress induced by heat shock or UVC radiation caused increases in RNase L levels that were dependent on the 3′-UTR.

In an apparent feedback loop, HuR expression increases levels of RNase L, whereas expression of RNase L decreases levels of HuR (Al-Ahmadi and others 2009). Cell growth rates and HuR levels were both elevated in RNase L-null mouse embryonic fibroblasts. The increase in HuR protein levels was correlated to enhanced stability of its mRNA in the absence of RNase L. The RNase L inhibitory effect on HuR mRNA levels was mapped to the HuR 3′-UTR, which contains U-rich/ARE-like sequences. In summary, whereas HuR stabilizes the RNase L mRNA through AREs 7 and 8, RNase L destabilizes the HuR mRNA through AREs in its 3′-UTR. Results suggest posttranscriptional cross-regulation between HuR and RNase L that determines cellular levels, and consequently cellular effects, of both proteins. RNase L effects on HuR expression was found to be most prominent in confluent cells or cells arrested in G1 phase of cell cycle, during which time both RNase L and HuR were in the cytoplasm.

RNase L and Senescence

Because of its antiproliferative and tumor suppression functions, involvement of RNase L in cell senescence and aging were explored (Andersen and others 2007). Ectopic expression of RNase L in mouse Balb-c 3T3 cells resulted in senescent morphological changes, decreased DNA synthesis, increased β-galactosidase activity, and replicative senescence. On the other hand, senescence was retarded in RNase L-null mouse fibroblasts. Further, 2-5A activation of RNase L induced senescence in primary WI38 human diploid fibroblasts. The RNase L effect was correlated with down-modulation of some cellular mRNAs, including several ribosomal protein mRNAs (Andersen and others 2009). RNase L-null mice had lifespan that was significantly prolonged (by about 20 weeks) compared with wild-type mice (Andersen and others 2007). Could polymorphisms or mutations in the human RNASEL gene affect the human lifespan?

Conclusions

Regulated turnover and processing of ssRNA by RNase L is essential for a complete IFN response against some viral infections. Studies on the structure and function of RNase L, and the related nuclease Ire1, are providing a window into how these enzymes are regulated. RNase L participates in the IFN antiviral effect directly or indirectly depending on the type of viral or cellular RNA that is cleaved. Antiviral innate immunity is triggered, in part, by processing of ssRNA into small RNA activators of retinoic acid-inducible gene-I like receptors (RLR) signaling that amplifies production of type I IFNs. RNase L is a suspected suppressor of HPC, and possibly of some other types of cancer, by inhibiting oncogenic viruses, but also through its pro-apoptotic and cell growth inhibitory properties. Determining how bacterial infections regulate RNase L activity resulting in suppression of infections will be of considerable interest. RNase L is ubiquitously expressed in a latent form, activated by 2-5A, and effective against many types of viral infections (Fig. 1). Therefore, RNase L could prove a unique target for a broad-spectrum antiviral, with possible activity against bacteria as well. What is clear after >30 years of research is that RNase L is now firmly established as an important player in overall innate immunity that controls viral pathogens.

Acknowledgments

The authors wish to gratefully acknowledge financial support for their research studies from the NIH (NCI), grant CA044059 (to R.H.S.) and to David Schumick, B.S., C.M.I., Center for Medical Art and Photography, Cleveland Clinic for outstanding artwork.

Author Disclosure Statement

R.H.S.: Abbott Laboratories, Inc. (patents and consulting) and Alios BioPharma, Inc. (patents and consulting).