Angiogenin mutations in Hungarian patients with amyotrophic lateral sclerosis: Clinical, genetic, computational, and functional analyses

Abstract Introduction Mutations in the angiogenin (ANG) gene are known to be associated with both familial and sporadic amyotrophic lateral sclerosis (ALS). The majority of disease‐causing mutations of ANG result in loss of either ribonucleolytic activity, nuclear translocation activity or both. Methods We sequenced ANG gene from a total of 136 sporadic ALS patients and 112 healthy controls of Hungarian origin. To elucidate the role of the R33W mutation in the disease mechanism, computational, and functional analyses were performed. Results Mutation screening revealed a mutation located in the signal peptide (M‐24I) and two mutations that affect the mature protein (R33W, V103I). The R33W mutation, which has not been previously detected in ALS patients, affects the key amino acid of the nuclear translocation signal of the ANG protein. Molecular dynamics simulations suggested that the R33W mutation results in partial loss of ribonucleolytic activity and reduced nuclear translocation activity. The ribonucleolytic assay and nuclear translocation assay of the R33W ANG protein confirmed the molecular dynamics results. Conclusions In the Hungarian ALS population, the observed frequency of ANG mutations was 2.9%, which is higher than previously reported for sporadic cohorts. The evidence from computational and functional analyses support the deleterious effect of the novel R33W variant detected in this study.

indistinguishable. In the past decades, more than 30 genes involved in the etiology of the disease have been identified (Amyotrophic Lateral Sclerosis Online Genetics Database, ALSoD; Abel, Powell, Andersen, & Al-Chalabi, 2012). The prevalence of mutations of the two most frequently mutated genes, SOD1 and C9orf72 repeat expansion was previously reported for the Hungarian population (Tripolszki et al., 2017).
First reported in 2004, substantial evidence has accumulated for angiogenin (ANG) involvement in ALS (Greenway et al., 2004;Sheng & Xu, 2016). To date, approximately 30 different mutations are reported for ANG in the ALSoD (Abel et al., 2012). The ANG gene, located on chromosome 14q11.2, encodes ANG, a 123-residue, 14.1-kDa protein. The ANG protein belongs to the pancreatic ribonuclease superfamily , and it plays an important role in rRNA biogenesis and cellular proliferation (Kishimoto, Liu, Tsuji, Olson, & Hu, 2005) in addition to its crucial role in inhibiting protein translation by cleaving transfer RNA (tRNA; Ivanov, Emara, Villen, Gygi, & Anderson, 2011). ANG is synthesized with a signal peptide of 24 amino acids that is cleaved to form the mature protein Shapiro, Riordan, & Vallee, 1986). The mature protein contains three functional sites: a receptor binding region 60 NKNGNPHREN 68 , the catalytic triad His13, Lys40, and His114, which is responsible for ribonuclease activity, and a nuclear localization signal sequence 29 IMRRRGL 35 , which is responsible for translocation into the nucleus (Figure 1; Hallahan, Shapiro, Strydom, & Vallee, 1992;Leonidas et al., 1999;Moroianu & Riordan, 1994). It is known that ANG mediates neurovascularization and promotes neurite outgrowth during early embryonic development and that it protects against hypoxia-induced motor neuron death (Kishimoto et al., 2005;Moroianu & Riordan, 1994;Sebastià et al., 2009). Mutations in the ANG gene cause loss of either ribonucleolytic activity or nuclear translocation activity or both (Wu et al., 2007).
The aim of this study was to evaluate the contribution of mutations in ANG, a major ALS gene, to the pathogenesis of the disease in Hungarian patients. The patients positive for SOD1 or C9orf72 repeat expansion identified in our previous study (Tripolszki et al., 2017) were not excluded, as there is evidence that ALS is oligogenic (van Blitterswijk et al., 2012). To explore the underlying mechanisms F I G U R E 1 The R33W mutation identified in the angiogenin (ANG) gene in a Hungarian sporadic amyotrophic lateral sclerosis patient. (a) Electropherograms of wild-type and mutant ANG sequences. (b) X-ray structure of human ANG in cartoon representation showing its functional sites and R33W mutation. Cartoon representation of the structure of human ANG (Protein Data Bank: 1B1I) showing its functional sites: catalytic triad residues His13, Lys40, and His114 are shown as stick models; nuclear localization signal is shown in cyan; and receptor binding sites are shown in green. The R33W mutation is represented as stick model in yellow. (c) Alignment of ANG protein sequences from various species shows the conservation of the R33 residue that result in loss of function of the detected R33W mutant protein, computational simulations, ribonucleolytic assays, and nuclear translocation assays were performed.

| Patients
The investigated sporadic patients (n = 136) were recruited from the Department of Neurology, University of Szeged, Szeged, Hungary. All patients fulfilled the revised El Escorial criteria for ALS (Ludolph et al., 2015). Among the patients, 54 were males and 82 were females (M/F ratio was 1:1.51), and the mean age of disease onset was 60.72 years (range 29-86 years). All patients were of Hungarian ancestry. The study was approved by the Internal Ethical Review Board of the University of Szeged. Written informed consent was obtained from all patients, and the study was conducted according to the Principles of the Declaration of Helsinki.

| Genetic analysis
Genomic DNA was isolated from frozen blood using DNeasy Blood  (Liu, Wu, Li, & Boerwinkle, 2016). Amino acid numbering of ANG is according to the ALS Online Database (Abel et al., 2012), which has been used in the previous published reports on ANG mutations.

| System setup and all-atom molecular dynamics simulations
The crystal structure of the human ANG protein (entry: 1B1I) was retrieved from the RCSB Protein Data Bank and used as the starting structure (Leonidas et al., 1999). The R33W variant identified in the Hungarian patient was modeled in silico by mutating Arg33 residue to Trp33, while keeping the secondary structure intact. The cofactor, CIT and crystallographic waters were removed from the structure and hydrogen atoms were added subsequently using the Xleap tool of AMBER 14 (Case et al., 2014). Modeled structures for the wild-type and R33W proteins were then solvated in an octahedral box of TIP3P water with ~10 Å between the protein surface and the box boundary (Jorgensen, Chandrasekhar, Madura, Impey, & Klein, 1983). Each system was then electrostatically neutralized by the addition of Cl − counter ions and simulations were carried out using the SANDER module of the AMBER 14 software package with ff14SB force field (Maier et al., 2015).

| Solvent-accessible surface area
The nuclear translocation activity of the R33W variant was examined by computing the SASA of nuclear localization signal residues 31 RRR 33 . A VMD plug-in, VolArea, was used to calculate the SASA of wild-type and R33W using a probe radius of 1.4 Å (Humphrey et al., 1996;Ribeiro et al., 2013).

| Graphics and figure preparation
All figures for representing ANG structures were generated using PyMOL (http://www.pymol.org) and VMD. Hydrogen bond interactions were visualized using Cytoscape (Shannon et al., 2003).
After 2 hr of incubation at 37°C, 70 μl of 3.4% ice-cold perchloric acid (Sigma) was added to terminate the reaction. After vortexing, the mixtures were incubated on ice for 15 min, and centrifuged at 15,000 g for 10 min at 4°C. Supernatant absorbance was measured at 260 nm with a MaestroNano spectrophotometer. Data were collected from three independent experiments.

| Nuclear translocation assay
Human umbilical vein endothelial cells (HUVEC; Cell Applications Inc., San Diego, CA) were grown in Endothelial Cell Growth Medium (Cell Applications) in a humidified atmosphere with 5% CO 2 at 37°C. Cells were seeded at 5 × 10 5 cells/ml onto an eight-well cell culture slide (Biologix, Shandong, China) (Schneider, Rasband, & Eliceiri, 2012). The ratio of nuclear/whole staining intensity was calculated to determine nuclear localization intensity of ANG, two-tailed T-test was used to determine significance with a significance level set at p ≤ 0.05.

| RE SULTS
Mutation analysis of the ANG gene revealed three coding-region mutations in four ALS patients (4/136 patients; 2.94%). One of the detected mutations affects the signal peptide (M-24I) and the two are located in the mature ANG protein (R33W, V103I). These mutations were not detected in 112 age-matched healthy controls.
ANG sequence analysis confirmed the presence of the rs11701 polymorphism in our ALS and control populations. We did not observe significant differences in allelic distributions between sALS patients, (minor allele frequency, MAF(G) = 0.118) and controls (MAF(G) = 0.171).

| M-24I mutation
The known M-24I mutation (c.3G>T) was detected in two patients.
One of the patients, who was described in our previous study (Tripolszki et al., 2017), also carried a mutation in the SOD1 gene Clinical report: The other patient with the ANG M-24I mutation was a 79-year-old woman. Her symptoms started 6 months before the examination with progressing dysphagia and dysarthria. On examination she was anarthric with increased gag and soft palate reflex and exhibited difficulty moving the tongue, which showed slight muscle atrophy. Deep tendon reflexes were increased in the extremities; however, she did not complain of weakness, even though signs of lower motor neuron damage were detected with electromyography (EMG). The revised ALS Functional Rating Scale (ALSFRS-R) was 33 (the normal score is 48). The Mini-Mental State Examination (MMSE) score was in the normal range: 27 (the maximum score is 30). The patient could communicate in writing. The initial symptoms indicated pseudobulbar palsy.

| V103I mutation
A recurrent missense variant (c.379G>A; V103I) was identified in the mature ANG protein.
Clinical report: The patient with the V103I mutation was a 63- year-old man presenting with severe weakness of the right arm He was on continuous Coumadin treatment. The result of the MMSE was 28 and was comparable with the global impression that the patient had no intellectual deficit. The initial disease sign was lower motor neuron damage of the right arm.

| R33W mutation
A missense variant (c.169C>T; R33W) located in the nuclear localization signal was identified in one patient (Figure 1). This variant has not been previously reported in an ALS patient, but occurs with a low allele frequency in the gnomAD database (MAF(T) = 0.00001624).
Six of the seven computational variant-effect prediction tools used predicted R33W to be pathogenic.

| Molecular dynamics simulations
MD simulation of the R33W variant protein was carried out to model structural and dynamics changes in the functional sites compared to the wild-type protein. The backbone root mean square deviation (RMSD) of the R33W variant obtained from MD simulations showed that, although the fluctuations were higher than the wild-type during the first 15 ns, it subsequently stabilized and converged as observed in the wild-type ANG ( Figure S1). This suggested that the Arg33 to Trp33 mutation may not affect the structural stability of the mutant. Catalytic residue His114 changed its conformation from native −80º during the simulations, indicating its role in probable loss of ribonucleolytic activity, whereas wild-type ANG maintained a stable His114 conformation 100%) (Figure 3). The result of this enzymatic activity assay was consistent with the MD simulation results.

| Local folding of nuclear localization signal abolishes nuclear translocation activity of the R33W variant
The nuclear translocation activity of the R33W variant was investigated from MD simulations. Our 100 ns simulation of R33W variant showed that the key nuclear localization signal residues 31RRW33 undergo local folding and becomes less accessible to solvent as compared to wild-type ( Figure 4A). This resulted in a reduced SASA in the R33W variant compared to the wild-type, which retained an open conformation and loosely packed 31RRR33 residues ( Figure 4B). Since the 31RRR33 residues are known to play a pivotal role in the nuclear translocation of ANG, we predicted that a decrease in SASA due to the local folding and close packing of 31RRR33 residues originated due to the replacement of hydrophilic

| D ISCUSS I ON
In this study, we identified three mutations in the coding region of the ANG gene, one of which has not been reported previously in ALS patients. The frequency of ANG mutations in our Hungarian sALS cohort (4/136; 2.9%) is higher than previously reported. According F I G U R E 3 Ribonucleolytic assay of wild-type (WT) and mutant (MUT) R33W angiogenin (ANG) proteins. Enzyme activity was measured using yeast tRNA as substrate. Increasing concentrations (0.05-0.4 mg/ml) of WT and R33W (MUT) proteins were incubated with yeast tRNA (2 mg/ml) at 37°C for 2 hr. Undigested tRNA was precipitated with perchloric acid. The absorbance of the supernatants was measured at 260 nm. Data were collected from three independent measurements. Relative enzyme activity of the MUT ANG was calculated as compared with the WT protein (100%). The activity differences at concentrations 0.1, 0.2, 0.3, and 0.4 mg/ml were calculated

| M-24I mutation
The M-24I mutation was first described by Conforti et al. (2008) and was detected in two Hungarian patients in this study. The biological function of the signal peptide is not entirely understood; hence it is difficult to predict the effect of the M-24I mutation. Nonetheless, it affects the start codon (ATG) of the gene, which may influence the correct translation of the protein.
One of the patients with the ANG M-24I also carried a mutation in the SOD1 gene (V14M) and was described in our previous study (Tripolszki et al., 2017). The co-occurrence of different variants in ALS-associated genes was also detected in other ALS cohort screenings (Kenna et al., 2013;Krüger et al., 2016).

| V103I mutation
The V103I mutation affects an amino acid that is conserved in mammals and it was first detected in an ALS cohort of Chinese  (Zou et al., 2011). According to previously reported molecular simulation results of the V103I variant structure, a hydrogen bond interaction network connecting from I103 to H114 was proposed (Padhi et al., 2012). Enzyme activity assays showed that the V103I mutant retains 54.1% of enzymatic activity toward tRNA (Bradshaw et al., 2017). Observing SASA values of the V103I mutant, it is predicted that the nuclear translocation activity is also impaired (Padhi et al., 2012).

| R33W mutation
The R33W missense variant, located in the nuclear translocation signal of the mature protein, has not been identified in ALS patients before. To understand the role of the mutation in disease manifestation, a computational and functional analysis of the R33W mutant was performed. Our MD simulations showed that the R33W mutation results in partial loss of ribonucleolytic activity and impaired nuclear translocation activity. The ribonucleolytic assay and nuclear translocation assay of the R33W ANG protein confirmed the MD results. According to the ribonucleolytic activity assay, the R33W mutant protein retained 57.4% of catalytic activity of the wild-type protein. Previous studies described that the reported disease-causing ANG mutations-except variant R121C-show a reduced level of catalytic activity in a range of 0%-95% compared with the wild-type ANG (Bradshaw et al., 2017;Padhi et al., 2014;Wu et al., 2007).
Since the R33W variant affects the key amino acid of the nuclear localization signal of ANG, this function of the mutant protein is also compromised. Our nuclear translocation assay showed that the alteration of Arg33 to Trp33 in ANG significantly reduced its entry into the nucleus. Amino acids 29 IMRRRGL 35 are part of the nuclear localization signal and are involved in nuclear translocation of ANG.
The arginine residues 31 RRR 33 are critical in governing the nuclear translocation of ANG: R33 is essential and residues 31 RR 32 modulate the process. Moroianu and Riordan (1994) described that the R33A mutant ANG protein is not translocated to the nucleus and lacks angiogenic activity (Moroianu & Riordan, 1994). Previous studies showed that other ALS-associated ANG mutations (S28N, L35P, P112L, K17I, K17E, Q12L) also result in partial or complete loss of nuclear translocation capability due to the changes in the folding of the 31 RRR 33 residues and subsequent reduction in SASA (Bradshaw et al., 2017;Padhi et al., 2014;Thiyagarajan, Ferguson, Subramanian, & Acharya, 2012;Wu et al., 2007).
Mature ANG contains three distinct functional sites: a receptor binding region, a catalytic triad responsible, which is for ribonuclease activity, and the nuclear localization signal. All three functional regions are needed for its angiogenic activity (Moroianu & Riordan, 1994). Alterations in the structure of ANG and dynamics result in F I G U R E 5 Nuclear translocation of wild-type (WT) and mutant angiogenin (ANG) in human umbilical vein endothelial cells (HUVEC) cells. HUVEC cells were incubated with 1 μg/ml of WT ANG or mutant ANG for 30 min and subsequently immunostained for ANG. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Images show the representative result of four experiments. (a) WT ANG accumulates both in the cytoplasm and nucleus of the HUVEC cells, whereas mutant (R33W) ANG accumulated predominantly in the cytoplasm of the cells. Individual channels as well as merged images are shown. (b) Nuclear translocation was quantified by measuring mean signal intensity levels of ANG staining in nucleus and cytoplasm of cells for at least three fields of views. Quantification of nuclear translocation of WT and mutant (R33W) ANG showed a significant decrease in nuclear staining in the R33W-treated samples (two-tailed Student's t-test, p < 0.0001). (c) Mean cellular signal intensity levels of ANG was determined for at least three fields of views, which showed a significant difference between the intensity of staining of the WT and mutant protein (two-tailed Student's t-test, p < 0.0001), suggesting a difference in the cellular internalization of the protein

ACK N OWLED G M ENTS
We thank Zsuzsanna Horváth-Gárgyán for her skilled technical assistance. This work was funded by the Hungarian Brain Research Program (Grant No. 2017-1.2.1-NKP-2017.

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest regarding this work.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request.