Angiogenin: a unique ribonuclease with multiple cellular functions

Parkinson’s disease (PD) is characterized by the progressive degeneration and loss of dopaminergic neurons in the substantia nigra pars compacta and catecholaminergic neurons in the locus coeruleus. Loss of neurotrophic or neuroprotective support may therefore increase the likelihood of developing PD. Angiogenin is a 14.1 kDa protein that belongs to the superfamily of vertebrate secreted ribonucleases. Other members of this family include RNase 1 (pancreatic ribonuclease), RNase 2 (eosinophil-derived neurotoxin), RNase 3 (eosinophil cationic protein), RNase 4, RNase 6 (k6), RNase 7 and RNase 8 [1]. In contrast to most members of this family, angiogenin exhibits relatively low ribonuclease activity [2]. Angiogenin was initially discovered as a tumour-derived angiogenic factor in human colon adenocarcinoma cells [3]. Subsequent studies focused on its role in angiogenesis, cancer, ischaemia and infection (reviewed in [4]). As the name suggests, angiogenin stimulates endothelial cell proliferation and has been shown to be required for the angiogenic activity of vascular endothelial growth factor (VEGF) and fibroblast growth factor-2 (FGF-2) [5]. While VEGF and FGF-2 signal through tyrosine kinase receptors to activate protein synthesis (via stimulation of mTOR and S6 kinase pathways), angiogenin increases ribosomal RNA (rRNA) transcription in the nucleus [6]. It thereby acts synergistically with VEGF and FGF-2 to increase protein synthesis in endothelial cells and is required for their proliferation [5]. Angiogenin-induced RNA transcription also requires the ribonuclease activity of angiogenin and occurs via an epigenetic activation of the ANG promoter [6, 7]. Angiogenin has also been proposed to indirectly stimulate the PI3-kinase and Akt kinase pathways in endothelial cells and bladder cancer cells [8,9,10].

Additional cytosolic activities of angiogenin have recently emerged. Under conditions of cellular stress such as oxidative stress, disruption of proteostasis or withdrawal of trophic factors, angiogenin accumulates in cytosolic stress granules [11, 12]. Stress granules are cytoplasmic foci containing untranslated mRNA and RNA-binding proteins that are formed in response to cellular stress and function to arrest protein translation [13,14,15]. Angiogenin is also a secreted factor that can be taken up by endocytosis into the cytoplasm of target cells [16,17,18] where it may function in paracrine [16].

As angiogenin is secreted from cells, several studies investigated potential angiogenin receptors and cellular uptake mechanisms. Angiogenin has been shown to be a ligand for surface receptors of the Plexin family. Plexin-B2 was identified by Yu et al. as receptor for angiogenin in endothelial, cancer, neuronal, and normal hematopoietic and leukaemic stem and progenitor cells [19]. Plexin-B2 is expressed in the postnatal and adult nervous system particularly in subventricular zone (SZV)-derived neural stem cells [20].

Angiogenin may, however, also been taken up through additional mechanisms into other cell types. Studies from our group demonstrated that neuronally-secreted or exogenous angiogenin protein is effectively taken up into astroglia [16] and alters their secretome [21]. Indeed, astroglia are now considered an important contributor to neurodegenerative disorders including PD and amyotrophic lateral sclerosis (ALS) [22, 23]. Uptake and subsequent RNA cleavage in astrocytes have been shown to require clathrin-mediated endocytosis (CME) and dependent on heparan sulfate proteoglycans [16]. It is possible that other angiogenin-binding proteins responsible for angiogenin uptake and signalling will be identified going forward.

Regulation of angiogenin activity

The activity of angiogenin is critically regulated by an endogenous inhibitor, ribonuclease/angiogenin inhibitor 1 (RNH1) [24]. Like angiogenin, RNH1 is expressed in endothelial cells and many other cell types, including neurons and glial cells [25,26,27]. RNH1 inhibits the activity of angiogenin by binding to its catalytic triad residues Lys-40, His-13 and His-114 [28]. Under stress and anabolic conditions, RNH1 accumulates in the nucleus where it binds angiogenin and inhibits rRNA transcription to save energy [29]. Angiogenin activity is therefore a process that is determined by the relative abundance and co-localisation of both proteins in cells.

Similar to other pro-angiogenic factors such as VEGF, transcription of the ANG gene is regulated by the transcription factor hypoxia-inducible factor-1α (HIF-1α). Through this mechanisms, increased ANG expression stimulates angiogenesis in tissues that have insufficient oxygen supply [30]. The hypoxia responsive element within the ANG gene has been mapped to the consensus HIF-1α binding site 5′-RCGTG-3′ [30, 31]. HIF-1α has also been shown to be required for ANG expression in neural cells in response to hypoxia [31]. ANG expression is also positively regulated by the transcription factor hepatic nuclear factor-1α (HNF-1α) [32]. This transcription factor is involved in glucose and lipid metabolism in liver and pancreatic beta-cells.

ANG variants in ALS and PD

Angiogenin became of interest to neuroscience with the initial discovery that ANG gene variants were associated with familial and apparently sporadic cases of ALS in Scottish, Irish, English, Swedish and Northern American families [33]. As part of this study, our laboratory identified that angiogenin is expressed and enriched in motor neurons. Subsequent studies have identified ANG variants in Italian, French, German, Dutch, Belgian, Hungarian, Chinese and Indian ALS patients (Table 1). Several of the ANG variants identified were predicted and subsequently validated to affect angiogenin’s ribonuclease activity due to their proximity to the catalytic site of the protein (Table 2). Other angiogenin variants have been shown to inhibit the shuttling of angiogenin between nucleus and cytoplasm [34], or to reduce the stability of the protein [35]. Subsequent studies showed that angiogenin exerts neuroprotective activities in vitro in models of excitotoxic, hypoxic and trophic factor-withdrawal-induced injury to motor neurons and other neural cells, including dopaminergic SH-SY5Y neuroblastoma cells [31, 36, 37]. Many of the ANG variants identified were shown to have reduced neuroprotective activity compared with wild-type ANG when overexpressed at similar levels in neurons [31, 36]. However, it is currently unknown which cell types in the nervous system (including endothelial cells) are susceptible to ANG mutations.

Table 1 ANG  variants associated with ALS
Full size table
Table 2 Biochemical characterisation of ribonucleolytic activity of human ALS ANG variants
Full size table

Of note, subsequent studies also identified ANG variants in familial forms of PD [38, 39]. The two studies identified several non-synonymous ANG variants in Northern American, German, Dutch and Italian PD patients (Table 3). The frequency of PD ANG variants were highly similar in both studies (0.45%/0.47%) compared with controls (0.04%/0%). Furthermore, Van Es et al. reported similar frequency of ALS patients and PD patients carrying ANG variants (0.46%/0.45% compared with 0.04% in controls) [38]. Many of the reported PD ANG variants were predicted to impair angiogenin protein function [38]. In a more recent study, several of these variants were validated to have reduced levels of ribonuclease activity in comparison with wild-type angiogenin [40].

Table 3 ANG variants associated with PD
Full size table

Interestingly, a recent study demonstrated ANG mutations in familial cases of Alzheimer’s disease (AD) [41]. Collectively, these findings point to a general role of angiogenin as a protective factor for central nervous system neurons. Of note, ang1 knock-out mice do not appear to develop neurodegeneration and ALS-, PD-, or AD-like symptoms or neuropathology in their life span [42], highlighting that aging and/or additional disease processes are required to trigger neurodegeneration. ANG can therefore be added to list of genes that regulate stress responses in neurons and are mutated and contribute to neurodegeneration in various neurological disorders (including PD), as seen in the case of TARDBP mutations, FUS mutations, C9ORF72 repeat expansions, and DJ-1 mutations [43,44,45,46].

Angiogenin delivery is neuroprotective in models of ALS and PD

The ANG mutations that have been reported in ALS and PD patients suggested a direct involvement of angiogenin in pathways leading to motoneuron degeneration or degeneration of dopaminergic neurons. We demonstrated that angiogenin protects cultured primary mouse motor neurons against ALS-associated, stress-induced cell death including excitotoxic injury by promoting and sustaining cell survival signalling through PI3-kinase/Akt kinases [36]. In further preclinical work, we demonstrated that daily, systemic (intraperitoneal) delivery of recombinant human angiogenin protein significantly increased life span and improved motor function in SOD1G93A mice, an established mouse model of ALS [36]. Cell survival signalling in motoneurons was preserved in angiogenin-treated mice. Importantly, the effect of angiogenin, when delivered post-symptom onset (from day 90 onward) on life span and disease progression, was comparable to the effect of a pre-symptom angiogenin treatment (from day 50 onward). These results suggested that angiogenin protein delivery may be beneficial in treating patients with newly diagnosed ALS.

To further validate these findings, our group performed a SOD1G93A mouse model study according to the preclinical guidelines for ALS animal studies set by the 2010 European ALS/MND group [47]. In this study, we demonstrated that systemic delivery of human angiogenin three times per week post-symptom onset (from day 90 onward) delayed motor dysfunction, significantly enhanced survival and protected against motoneuron loss and vascular network regression in the lumbar spinal cord [48].

Angiogenin may also be neuroprotective in PD. Steidinger et al. demonstrated significantly decreased levels of endogenous angiogenin in an alpha-synuclein transgenic mouse model of PD and showed that recombinant human angiogenin protected against dopaminergic neuronal cell death and inhibited caspase-3 activation in neurotoxin-induced in vitro models of PD [49]. A subsequent study by the same group found that virally-mediated overexpression of human angiogenin in the substantia nigra did not protect against dopaminergic cell loss in a neurotoxin-based mouse model of PD [50]. These findings suggest that further in vivo studies are required to explore potential neuroprotective functions in animal models of PD, in particular in genetic models.

tRNA-derived fragments (tRFs) and ‘stress-induced tRNA fragments’ (tiRNAs) in PD

During stress conditions, angiogenin accumulates in the cytosol where it cleaves non-coding RNAs, including transfer RNAs (tRNAs). Cleavage of tRNAs by angiogenin generates fragments termed ‘stress-induced tRNA fragments’ or ‘tiRNAs’ [12, 51, 52]. Cleavage of tRNAs by angiogenin occurs in their anticodon loop, a process that is highly regulated by tRNA modifications [53,54,55,56,57], so that only specific subsets of tRNA fragments are generated [58]. tiRNAs have been shown to inhibit ribosome assembly [12] and to inhibit cap-dependent protein translation via interaction with the initiation factor eIF4F [12]. Both processes may facilitate the recovery of cells during stress conditions, so that resource-consuming or error-sensitive cell functions are stalled during periods of stress. tiRNA generation has also been linked to the process of stem cell maintenance and inhibits the proliferation of hematopoietic stem/progenitor cells [42]. Due to their multiple mechanisms of action, lack of tiRNA production could be involved in the pathogenic effects of ANG variants in human disease.

tRFs in general are now considered a new class on small non-coding RNAs with multiple cellular functions, of which tiRNAs are only a subclass. tRFs have been detected in various biological systems suggesting that tRNA cleavage by angiogenin and other ribonucleases is an evolutionary conserved process. Other new functions of tRFs beyond the regulation of protein translation have also been reported. ‘SHOT-RNAs’, which are analogous to angiogenin-produced tiRNAs, were identified in hormone responsive cancer cells where they stimulate their proliferation [59]. Two studies in 2016 showed that tRFs fragments are enriched in sperm cells and delivered to the zygote at fertilization where they modified gene expression by binding to the elements in the zygote’s genome [60, 61]. Multiple ribonucleases other than angiogenin are able to generate such fingerprints, including the ribonucleases Z and Dicer [62,63,64,65,66]. The identification of tRF ‘fingerprints’ in different biological systems and disease conditions is still at its infancy, but interesting observations are now beginning to emerge. A study from our laboratory showed that specific tRFs are associated with epilepsy [67]. Small RNA sequencing analysis of plasma samples collected during video EEG monitoring of patients with focal epilepsy identified significant differences in three specific tRFs fragments compared with healthy controls. Interestingly, these tRFs are different from angiogenin-generated fragments as cleavage did not occur in the anticodon loop. These fragments were elevated in the pre-seizure period, but lower in post-seizure samples, and may represent a novel class of biomarkers indicating seizure risk in epilepsy patients.

A recent study performed in PD patients identified disease-specific tRFs in brain and biofluids [68]. Reanalyses of RNAseq data from three previous studies identified multiple differentially abundant tRFs between PD patients and healthy controls in prefrontal cortex, cerebrospinal fluid and serum. Of note, a subset of the identified tRFs successfully distinguished PD patients from controls with high sensitivity and specificity in each sample collection. Further research is required as to whether these fragments are generated by angiogenin or other ribonucleases. Collectively, these findings suggested that tRF signatures are promising candidates as non-invasive PD biomarkers.

Summary

There is a significant body of evidence suggesting that angiogenin is a stress-induced survival factor for central nervous system neurons. While it has been shown that angiogenin is able to protect against dopaminergic neuron loss in vitro, further research is required to explore its role in animal models of PD. The arrival of new animal models of PD will likely accelerate this translation, as seen in ALS models. Due to its pleiotropic mechanism of action, angiogenin may indeed be an interesting candidate for the treatment of neurodegenerative disorders. It stimulates angiogenesis in endothelial cells and promotes neuronal survival through Akt signalling and possibly through the formation of tiRNAs, thereby facilitating the recovery of stressed neurons. Moreover, tRFs generated by angiogenin and other ribonucleases may deliver novel diagnostic or prognostic tools for the management of neurodegenerative disorders. Studies are now required to explore the biological functions of these fragments in vitro and in vivo.