RNA degradation is a vital post-transcriptional process which ensures that transcripts are maintained at the correct level within the cell. DIS3L2 and XRN1 are conserved exoribonucleases that are critical for the degradation of cytoplasmic RNAs. Although the molecular mechanisms of RNA degradation by DIS3L2 and XRN1 have been well studied, less is known about their specific roles in the development of multicellular organisms or human disease. This review focusses on the roles of DIS3L2 and XRN1 in the pathogenesis of human disease, particularly in relation to phenotypes seen in model organisms. The known diseases associated with loss of activity of DIS3L2 and XRN1 are discussed, together with possible mechanisms and cellular pathways leading to these disease conditions.
Ribonucleases are key enzymes involved in the control of messenger RNA (mRNA) stability, which is crucially important in maintaining RNA transcripts at the correct levels within cells. Studies have shown that 40–50% of changes in gene expression occur at the level of RNA stability , which means that this process is highly important in regulating gene expression. In multicellular organisms, it is increasingly evident that degradation of specific mRNAs is critical for the regulation of many cellular processes, including early development, infection and inflammation, apoptosis and ageing [2–7]. Although progress has been made in identifying the targets regulated by these ribonucleases [8–11], the mechanisms whereby they affect cellular processes leading to disease are largely unknown. This review focusses on two highly conserved cytoplasmic exoribonucleases DIS3L2 and XRN1, and how dysregulation of these enzymes is implicated in human diseases. The association of the related exoribonuclease DIS3 with human disease has recently been reviewed elsewhere .
An mRNA transcript is usually protected from degradation by the 5′ methylguanosine cap and the 3′ poly-A tail. mRNAs have been shown to be held in a circular conformation by interactions of protein factors such as eIF4E (the cap-binding protein), eIF4G and the poly-A-binding protein. Degradation usually begins with decircularization of the RNA and the removal of the poly-A tail. Following this process, degradation can proceed in two ways: in the 3′–5′ direction or in the 5′–3′ direction. In some circumstances [such as during non-sense-mediated decay or microRNA (miRNA)-induced degradation], internal cleavage occurs prior to degradation of the mRNA in both the 3′–5′ and 5′–3′ directions [2,4,5] (Figure 1).
Image depicting deadenylation-dependent RNA degradation pathways.
During 3′–5′ degradation, deadenylases first digest away the poly-A tail leaving an exposed 3′-end. This facilitates 3′–5′ degradation by a family of ribonucleases [9,12], including DIS3 and DIS3L2. DIS3L2 is a 3′–5′ exoribonuclease that is conserved from bacteria to humans. It acts independently of the 3′–5′ exoribonuclease–exosome complex and degrades non-coding RNAs as well as mRNAs [13,14]. In humans and the yeast Schizosaccharomyces pombe, this is often achieved by the addition of a polyuridine tract to the 3′-end of the transcript that provides an ideal landing pad for DIS3L2 to initiate/reinitiate 3′–5′ decay [14–16]. As well as mRNAs, miRNAs, such as pre-let-7, have been shown to be poly-uridylated and then degraded by DIS3L2 in human cells and mouse embryonic stem cells [17,18].
In contrast, there is only one known exoribonuclease involved in cytoplasmic 5′–3′ degradation; XRN1. After deadenylation by the Ccr4/Not deadenylase complex , the Lsm1–7 complex is recruited to the 3′-end of the RNA, together with XRN1 and decapping proteins such as Dcp1 and Dcp2 . This complex activates the decapping enzymes that cleave off the 5′-methylguanosine cap allowing the threading of the exposed 5′-end to the active site of XRN1 where it is digested. Although RNA degradation by XRN1 is likely to be co-translational , XRN1 is also found in cytoplasmic foci, termed processing bodies (P-bodies) [22,23], which constitute a concentration of mRNA decapping machinery, messenger ribonucleoproteins and ribonucleases . Here, mRNAs can also be stored for re-entry into translation at a later time. Being the only cytoplasmic exoribonuclease of its kind, this places XRN1 at a pivotal position in the RNA degradation pathway, with any modulation to this enzyme having the potential to have significant effects on cellular function.
Recent work using human tissue culture cells and model organisms shows that DIS3L2 and XRN1 affect important cellular processes such as proliferation and apoptosis. Experiments using human HeLa cells have demonstrated that knockdown of DIS3L2 results in increased proliferation  and reduced apoptosis . Knockdown of DIS3L2 in HeLa cells was also shown to result in mitotic errors and affect the expression of key cell cycle proteins, including Cyclin D and Aurora B kinase , suggesting that it has a role in regulating the cell cycle. In multicellular organisms, XRN1 is involved in many developmental processes [26,27], but more recently has been shown to be a key player in controlling apoptosis [8,9]. We have recently demonstrated that null mutations in the Drosophila XRN1 homologue pacman result in wing imaginal discs that are 45% the size of wild-type controls as well as resulting in lethality during pupation. Mutant wing discs show an increase in the number of cells undergoing apoptosis, especially in the wing pouch area. The pacman mutant imaginal discs also undergo compensatory proliferation, but this is not sufficient to compensate for the large amount of apoptosis occurring in these discs. The pro-apoptotic RNAs hid and reaper increase at the post-transcriptional level in pacman mutants, suggesting that they are targets of Pacman . Additionally, RNA-seq experiments have shown that a potential target of Pacman is dilp8, a Drosophila insulin-like peptide with similarity to the insulin-like peptides insulin-like growth factor (IGF)-1 and IGF-2 in humans .
Human diseases associated with DIS3L2
The clinical importance of DIS3L2 is demonstrated by its association with Perlman syndrome, which is a congenital overgrowth syndrome inherited in an autosomal recessive manner . Affected children display foetal gigantism, abnormal enlargement of organs (e.g. kidneys), macrocephaly, facial abnormalities, neurodevelopmental delay and high neonatal mortality. Histological examinations reveal nephroblastomatosis, which is an important precursor for Wilms' tumour. Germline mutations in these children include deletions of exon 6, 9 and/or 19, which are consistent with a loss of function of DIS3L2 [25,29].
Mutations in DIS3L2 have also been associated with sporadic occurrence of Wilms' tumour, a kidney cancer also referred to as a nephroblastoma [25,28,29]. Wilms' tumour results from the failure of groups of kidney cells to mature and differentiate, instead of undergoing continuous proliferation. DIS3L2 is also likely to be important in sporadic Wilms' tumour as 30% of these tumours (6/20) show partial or complete DIS3L2 deletion . This tumour is not only associated with aberrations in DIS3L2, but is also linked to mutations in the DROSHA gene, leading to aberrant expression of oncogenic miRNAs , such as miR-562 , which promotes deregulation of EYA1. EYA1 is involved in the retinal determination of gene network, and has been associated with driving various malignancies (such as breast cancer) through the cell cycle, leading to extensive proliferation, while also inhibiting apoptosis . This is interesting because it suggests that Wilms' tumour is partially driven by extensive loss of the RNA degradation machinery itself, highlighting the importance of this pathway.
Other aberrations in the DIS3L2 gene have been associated with a Marfan-like syndrome with skeletal overgrowth . Three patients carrying a chromosomal translocation at the chromosomal region 2q37.1 were shown to have a breakpoint within the DIS3L2 gene at intron 5 (two patients) and intron 6 (one patient) [33–35]. The symptoms of these patients included skeletal overgrowth and malformations, leading to long, slender fingers, curvature of the hands and feet (arachnodactyly), lack of apoptosis between digits (mild syndactyly) and abnormal curvature of the spine (severe scoliosis). However, for these patients, it cannot be ruled out that these phenotypes are also affected by misexpression of the neighbouring gene NPPC (encoding C-type natriuretic peptide) or deletion of the gene encoding hsa-miR-562, which is located within intron 9 of DIS3L2 .
Human diseases associated with XRN1
Wilms' tumour is not the only tumour which has been associated with defects in RNA degradation. Reduced XRN1 mRNA expression has been demonstrated in many osteosarcoma cell lines as well as osteosarcoma patient samples . Osteosarcoma is the third most common cancer affecting adolescents with a second smaller peak occurring between 60 and 80 years, particularly in people affected by Paget's disease. Overall, osteosarcoma has a moderate incidence rate, with 10–26 per million new cases worldwide each year . Osteosarcoma often develops at either end of the long bones of the arms or legs, although tumours can also occur in the skull and pelvis. Tumours typically arise from osteoid-producing neoplastic cells adjacent to the long bone growth plate. The main cause of death in these patients is the spread of cancer cells to the lungs; these secondary tumours are present in 10–20% of cases at diagnosis .
In osteosarcoma cells, XRN1 is often down-regulated, suggesting that lower levels of XRN1 promote proliferation. However, null mutations and hypomorphic mutations of this conserved exoribonuclease in Drosophila result in apoptosis . This apparent contradiction can be explained by hypothesizing that apoptosis is blocked in osteosarcoma cells, meaning that the down-regulation of XRN1 causes an increase in the proliferation of the tumour cells without the associated apoptosis. By this means, it may be that reduction of XRN1 confers a growth advantage on the tumour cells. Alternatively, down-regulation of XRN1 may lead to increased expression of specific miRNAs and/or protein-coding RNAs, which then promote the cancerous phenotype. Possible candidates are IGF-1, which is known to induce proliferation of osteoblasts [38–40] and/or pro-inflammatory RNAs, such as FOS and MYC, which are widely known to be involved in cancer progression.
XRN1 and the host response to viruses
XRN1 also appears to play a pivotal role during the host response to viral infection and has been implicated in the pathogenesis of many viral families. Recent studies have shown that the activity of XRN1 is specifically inhibited by certain Flaviviruses, which include Dengue fever virus, West Nile virus , Yellow fever virus , Japanese Encephalitis virus  and, perhaps more notably at present, the Zika virus. Flaviviruses are single-stranded (+) sense arthropod-borne viruses that are usually transmitted to humans through mosquitos, but can also be transmitted through ticks . Encephalitis is symptomatic of each of these infections, alongside haemorrhagic fevers. The activity of XRN1 is also repressed in a similar manner by Hepatitis C virus (HCV) that primarily infects the liver, and can lay asymptomatic for many years. During this time, the virus can cause liver scarring, potentially leading to cirrhosis, liver cancer and liver failure . HCV, along with Hepatitis B, has also been tenuously linked with pancreatic cancer and leukaemia/lymphoma [45–47]. The targeting of XRN1 by these viruses indicates that XRN1 is normally vitally important in protecting the cell from these viral pathogens.
A series of recent publications have elucidated the molecular mechanisms whereby these viruses inhibit XRN1 as well as shedding light on the disease mechanisms. In the case of Dengue and West Nile virus (Kunjin strain), viral transcripts are able to inhibit the 5′–3′ RNA degradation pathway by stalling XRN1 within the 3′-untranslated region (UTR) of the viral RNA (Figure 2A). The formation of RNA pseudoknots (‘slipknots’) , as well as other RNA structures within the 3′-UTR, stalls the progression of XRN1 by blocking entry of the RNA into the active site . XRN1 degrades the viral RNAs from the 5′-end of the transcript, until it reaches these structures, leaving small flaviviral non-coding RNAs, which accumulate in the cytoplasm. These small, structured RNAs bind to and sequester XRN1, thus reducing 5′–3′ degradation of other cellular RNAs  (Figure 2A). For HCV, similar highly structured regions within the 5′-UTRs of their transcripts stall and repress the activity of XRN1 (Figure 2B). Using tissue culture cells infected with HCV, it was shown that the disruption of the 5′–3′ mRNA decay pathway results in the stabilization of short-lived RNAs such as transcription factors involved in oncogenesis (e.g. MYC, FOS and JUN) and angiogenesis (e.g. VEGFA, HIF1A and CXCL2) . Since HCV infection is associated with the development of hepatocellular carcinoma, these factors might contribute to the development of the diseased state . This study also showed an increase in the levels of capped and presumably functional RNAs, suggesting that repression of XRN1 leads to shutdown of the entire degradation pathway, possibly by sequestration of other decay components . In the case of Flaviviruses, the available evidence suggests that viral infection disproportionately affects short-lived transcripts encoding cytokines and factors involved in innate immunity leading to the inflammatory symptoms observed .
Inhibition of XRN1 in the 5′- and 3′-UTRs of viral transcripts.
The importance of XRN1 in limiting the pathogenesis of HCV is illustrated by the fact that this virus has evolved two ways to inhibit the action of XRN1. Recent publications show that 5′–3′ degradation of HCV transcripts is stalled by the expression of miR-122 in the host cell, which binds via its miRNA seed sequence to the 5′-UTR of the viral transcript (Figure 2C). The association of miR-122/RISC, including Ago2 , at two sites in the 5′-UTR, protects it from degradation by XRN1 . miR-122 is a major determinant of viral RNA replication, as it inhibits XRN1 from degrading the viral RNA, allowing accumulation, increased replication and increased translation of viral RNA. Studies have shown that miR-122 only needs to be bound at one of the two sites for it to have this inhibitory effect, and binding to both sites maintains a co-operative effect, rather than increasing the effectiveness . In addition to the miR-122-binding site, the 5′-UTR HCV viral RNA includes four stem loops of which three (loops 2–4) appear to function as a highly structured internal ribosome entry site, allowing translation initiation of the viral RNA . Translation of the HCV core protein has been shown to activate the proto-oncogene, MYC, allowing the progression of hepatocellular carcinomas . While loops 1 and 2 are involved in viral replication, they also inhibit the activity of XRN1 by presenting structures that stall the progress of XRN1 as described above.
Another way in which viruses can inhibit the 5′–3′ RNA degradation pathway is by disruption of P-bodies, affecting the localization of XRN1. Rotavirus infection is understood to disrupt P-bodies, the sites of concentration of the RNA degradation machinery. Rotavirus is an RNA virus that causes acute diarrhoea across many developing countries, mainly affecting young children, with its spread being attributable to poor sanitation and hygiene . This virus is a double-stranded, non-enveloped virus that utilizes rotavirus protein NSP1 to degrade the deadenylase, PAN3, in the P-bodies, resulting in transcript stabilization, an effect also seen in Poliovirus . Rotavirus also causes the relocalization of XRN1 from the P-bodies to the nucleus, thereby inhibiting the degradation of viral RNA. However, the molecular details of this are unknown. Viral proteins specifically prevent degradation of six structural transcripts and six non-structural transcripts, which are either translated into protein or serve as templates for replication of viral progeny .
When DIS3L2 and XRN1 were first identified, it could not have been envisaged that investigation of these apparent ‘housekeeping’ exoribonucleases would lead to such interesting and fundamental insights into the molecular mechanisms of RNA decay. As described above, recent work on DIS3L2 and XRN1 is now revealing their important roles in cellular function, viability and human disease. Finding a way to combat disease by engineering the RNA degradation machinery to alter patterns of RNA regulation observed during pathogenesis offers a new hope to many people in the future. This is especially true in the cancer field. For example, if DIS3L2 or XRN1 is proved to facilitate overgrowth syndromes or the progression of osteosarcoma, there is potential to develop novel therapies for these diseases. The data implicating XRN1 as a key enzyme in the control of Flavivirus and Hepatitis C viral infection also present an opportunity for the development of broad-spectrum antiviral treatments. This is particularly important in the wake of new viral epidemics such as the outbreak of the Zika virus in South America. It is clear that there is still much work to be done to fully understand the ways in which DIS3L2 and XRN1 regulate cellular pathways resulting in disease progression. A good starting point for this would be to pinpoint the exact RNA transcripts which are being targeted by specific exoribonucleases in these pathways to elucidate how changes in RNA levels could contribute to pathogenesis. From starting at the level of basic science, this could have profound implications in the medical world in the future.
HCV, hepatitis C virus; IGF, insulin-like growth factor; miRNA, microRNA; mRNA, messenger RNA; UTR, untranslated region; PABP, poly-A-binding protein; CAF1, Chromatin Assembly Factor 1.
This work was funded by the University of Brighton Studentship [grant number WC003-30].
The authors thank Sophie Robinson and Peter Bush for their input and advice.
The Authors declare that there are no competing interests associated with the manuscript.