The catalytic activity and secretion of zebrafish RNases are essential for their in vivo function in motor neurons and vasculature

Angiogenin (hANG), a member of the Ribonuclease A superfamily has angiogenic, neurotrophic and neuroprotective activities. Mutations in hANG have been found in patients with Amyotrophic lateral sclerosis (ALS). The zebrafish (Danio rerio) rnasel-1, 2 and 3 are orthologues of hANG and of these only Rnasel-1 and Rnasel-2 have been shown to be angiogenic. Herein we show that NCI-65828, a potent and specific small molecule inhibitor of hANG inhibits Rnasel-1 to a similar extent. Treatment of early zebrafish embryos with NCI-65828, or with terrein, a fungal metabolite which prevents the secretion of hANG, resulted in spinal neuron aberrations as well defects in trunk vasculature. Our detailed expression analysis and inhibitor studies suggest that Rnasel-1 plays important roles in neuronal migration and pathfinding as well as in angiogenesis in zebrafish. Our studies suggest the usefulness of the zebrafish as a model to dissect the molecular consequences of the ANG ALS variants.

in the RNase activity affect neuronal survival. Neuronal cell lines expressing the ANG-ALS variants also lacked the ability to form stress granules 24 . Zebrafish (Danio rerio) RNase-like proteins (Rnasel1, 2 and 3) are hANG like RNases identified in zebrafish [25][26][27][28] . The Rnasels are secreted RNases and have a signal sequence, the "CKXXNTF" signature motif and the catalytic triad, as well as six conserved cysteine residues similar to hANG (RNase 5) 29 . Previously, we have shown (based on a detailed structure-function study) that Rnasel-1a cleaves tRNA with a specific activity similar to hANG. This is consistent with the finding that the active site in Rnasel-1a is blocked by its C-terminus as in hANG 27 . This is in contrast to Rnasel-3e in which the active site is open and which has 17-20 fold more RNase activity towards tRNA 27 .
In this study, we have used the compound NCI-65828, a small molecule that inhibits the enzymatic activity and terrein, a fungal metabolite known to prevent the secretion of hANG by prostate cancer cell lines, to explore if both the enzymatic activity and the secretion of hANG are essential for its in vivo neuronal and angiogenic functions. For our model systems, we used neuronal cell lines stably expressing HA epitope tagged mouse Ang1 (mAng1) and Tg(fli1a:EGFP) zebrafish which express EGFP in the vascular system 30 . The nervous system of the zebrafish is well characterised, and its relatively simple neuromuscular organization makes it an ideal model to study neurodegenerative disorders 31,32 . Prior to carrying out a detailed functional study, we first investigated whether NCI-65828 inhibits zebrafish RNases.
Here we report that NCI-65828 inhibits Rnasels and that human neuronal cells exposed to terrein accumulate mAng1. We also show that in vivo inhibition of the RNase activity of Rnasels and their secretion leads to defective development of spinal motor axons and intersegmental vessels. Our results show that both the catalytic activity and the secretion of ANG-like Rnasels play important roles during development of the zebrafish nervous system and vasculature.

NCI-65828 is a potent inhibitor of the ribonucleolytic activity of Rnasels.
Prior to studying the effects of NCI-65828 (8-amino-5-(4′-hydroxybiphenyl-4-azo) napathlene-2-sulphate) (Fig. 1D), a selective and potent cell permeable inhibitor of the catalytic activity of hANG, on zebrafish motor neurons and vasculature we sought to establish if indeed NCI-65828 is also able to inhibit the enzymatic activity of zebrafish Rnasels.
We initially established the time courses for cleavage of 6-FAM-mAmArCmAmA-Dabcyl (a fluorogenic substrate) by Rnasels -1a, -3e and hANG which enabled clear measurements of v, F 0 , and F max (Fig. 1A-C). These curves provided highly reproducible estimates of k cat /K m (Fig. 1F). Using this substrate (which has its cleavable bond between a cytidine and an adenine residue, and is thus tailored for hANG), the specific activities of Rnasel-1a and hANG were very similar. This is also the case for the cleavage of tRNA by these two enzymes 27 . This differs slightly from the result obtained with 6-FAM-dArUdAdA-6-TAMRA (cleavable between uridine and adenine), which is cleaved ~3-fold more readily by Rnasel-1a 25 , although this is consistent with structural features in the B 1 subsite of Rnasel-1a that may promote uridine binding 27 . The specific activity of Rnasel-3e with the present substrate is 4-and 6-fold greater than that of Rnasel-1a or hANG, respectively. This is fairly similar to the profile obtained with 6-FAM-dArUdAdA-6-TAMRA 25 but is somewhat less marked than that obtained with tRNA, which is cleaved 17-20-fold more readily by Rnasel-3e 27 .
Having established the time course of RNase activity, we studied the effect of 80 µM NCI-65828 (Fig. 1D), previously reported to inhibit hANG, on the ribonucleolytic activity of Rnasels 9 . 6-FAM fluorescence was quenched significantly by 80 µM NCI-65828. Assay sensitivity could be restored satisfactorily by increasing the fluorimeter's emission slit width, and all data were validated by verifying that the cleavage-induced increase in fluorescence (i.e. the value of F max /F 0 ) fell in the normal range (25)(26)(27)(28)(29)(30). In the presence of 80 µM NCI-65828, the ribonucleolytic activities of Rnasel-1a and -3e were only 3-4% of normal, indicating substantial inhibition (Fig. 1E).
Spatio-temporal expression pattern of zebrafish rnasel1-3. We carried out a systematic analysis of rnasel expression at various embryonic stages as well as in adult organs as described in materials and methods (see online methods) which would enable us to ascertain which of the rnasels was likely to be the cause of the in vivo defects when inhibited by NCI-65828. Expression of all three rnasels was low at 2.5 hours post fertilization (hpf)the earliest embryonic stage analysed, though rnasel-1 was higher than -2 and -3. Expression of rnasel-1 and -3 dramatically increased after 20 hpf, with a peak at 32 hpf after which levels of both rnasel-1 and -3 dropped to initial levels. A small increase in rnasel-2 levels occurred at 32 hpf and peaked at 72 hpf. This peak of expression at 72 hpf was also seen for rnasel-3. The fall in expression levels seen between 32 and 40 hpf for rnasel-1 expression was maintained at consistent levels until 55 hpf after which a small increase was seen, mirroring that of rnasel-2 and -3 ( Fig. 2A).
The expression levels of rnasels relative to each other observed at 72 hpf are similar to the overall pattern seen in adult fish where increased expression of rnasel-2 and -3 is seen over rnasel-1 (Fig. 2B). rnasel-2 and -3 both appear to be strongly expressed in the ovaries with very little expression of rnasel-1. Conversely little expression of any rnasel is seen in the testes. rnasel-1 appears to be strongly expressed in the eye, gut, heart and spinal chord in the adult fish and seems to be the dominant transcript in the eyes, gut and heart. The highest overall expression of all three isoforms appears to be in the spinal cord, while the highest expression of the individual isoforms is seen in the heart for rnasel-1, ovaries for rnasel-2 and liver for rnasel-3. Low expression with comparable levels between transcripts is seen in organs such as the gills, spleen and swim bladder (Fig. 2B). Terrein treatment of SH-SY5Y mAng1HA cells leads to accumulation of intra-cellular Ang.
Prior to investigating the effects of terrein on neuronal development and vascularization in zebrafish, we first investigated if indeed treatment with terrein affected the secretion of Ang using a SH-SY5Y cell line constitutively expressing a HA epitope tagged mouse mAng1 which we will refer to as SH-SY5Y mAng1HA. We analysed the intra-cellular accumulation of the HA tagged mAng1 in these cells by Western blotting. Intracellular levels of mAng1HA increased over the course of the 24 or 48 h treatment with 30 μM terrein relative to GAPDH loading (Fig. 3A). The adverse effects, if any, due to accumulation of Ang in the SH-SY5Y mAng1HA cells treated with terrein was investigated. There were no obvious morphological changes or increased incidence of cleaved caspase 3 positive cells in both untreated cells and cells cultured in terrein for 24 h (Fig. 3B). However, with a longer terrein treatment of 72 h, we observed an increase in cleaved caspase 3 positive cells as compared to the control. The SH-SY5Y mAng1HA cell line had significantly more cleaved caspase 3 positive cells than untreated SH-SY5Y mAng1HA cells at both 20 µM and 30 µM of terrein compared by ANOVA (P < 0.05). The most significant effect was observed at the highest concentration (30 µM) (Fig. 3B).
The intracellular distribution of HA tagged mAng1 in the terrein treated SH-SY5Y mAng1HA cells was different when compared with the untreated cells. There was a high incidence of large aggregates in the cell bodies, which increased in number at higher terrein concentrations. Co-staining with organelle markers showed that a significant amount of mAng1HA no longer co-localised with TGN46. No co-localisation seen with either PDI or LAMP1 (Fig. 3C). It has previously been reported that in P19 embryonal carcinoma cells exposed to high levels of ANG, the apoptosis inducing factor (AIF) does not translocate to the nucleus 33 . In control SH-SY5Y mAng1HA cells where secretion of ANG is not inhibited, AIF is seen evenly distributed throughout the cell body. However, in terrein treated SH-SY5Y mAng1HA cells with an accumulation of mAng1HA, AIF aggregates are localised adjacent to the nucleus (Fig. 3D).
Blocking the RNase activity and secretion of Rnasels in zebrafish embryos leads to motor neuron and vascular defects. Transgenic Tg(fli1a:EGFP) 30 embryos were used to assess both vascular and motor neuron development following treatment with NCI-65828 or terrein. We treated zebrafish embryos with NCI-65828 or terrein at two critical stages in the formation of motor neurons -at 10 hpf when primary motor neurons (1°MNs) are generated in the nascent brain and spinal cord, and at 18 hpf when they extend axons 34,35 (Fig. 4A,B).
The dorsal aorta also puts forth sprouts at around 20 hpf (after 1° MN axon extension) by angiogenesis. These sprouts lengthen longitudinally to connect up and form the dorsal longitudinal anastomotic vessels [36][37][38] .

Effects on motor neurons on treatment with NCI-65828 or terrein prior to their specification.
Tg(fli1a:EGFP) zebrafish embryos were treated either with NCI-65828 or terrein at 10 hpf and allowed to develop to 27 and 36 hpf stages (Fig. 4A,B). We focused on a subset of primary motor neurons. Cell bodies of the caudal primary neurons (CaP) and the variable primary neurons (VaP) located in the middle of each of the spinal cord hemisegment express the LIM homeodomain protein islet2 39 . Control and inhibitor treated embryos were stained with mouse αIslet antibody, labelling both Rohon-Beard sensory neurons in the dorsal spinal cord and motor neurons prior to their migration to the ventral horn. We counted the number of islet + nuclei in the trunk (Fig. 4C) and found a significant reduction in the number of islet + nuclei in embryos treated with both NCI-65828 and 30 µM terrein (Fig. 4D,E) at 36hpf. We also observed misplaced and missing islet + cells.   Inhibitor-treated and control embryos were stained with mouse αZnp1 to visualise CaP motor axons and rabbit αGFP to visualise the vascular system. In embryos treated with inhibitors from 10 hpf, the spinal cord can be seen clearly by Znp1 staining (Fig. 5A). No differences were seen between the longitudinal tracts of the spinal cord in treated and untreated embryos. In both controls and embryos treated with inhibitors, the primary motor neuron axon length correlated with A-P position where anterior axons are more developed (Fig. 5A, dotted lines). Axons in NCI-65828 and terrein treated embryos projected from the spinal cord at the same positions within each somite as untreated and the proximal part of the axon projected at a similar angle dorso-ventrally in all cases. Similarly in all cases each axon had an increasing posterior curve as development continued to 36 hpf, however the distal tip of axons treated with NCI-65828 or terrein were seen to loop back towards the anterior frequently while untreated axon continued extend towards the posterior. At 27 hpf, axons in untreated embryos branched at the distal tip while treated embryos frequently branched in more medial regions of the axons in both anterior and posterior directions. These aberrant medial branches were not noticeable at 36 hpf, where increased branching is seen at the distal tip (Fig. 5A).
Axonal length measured in the trunk of control 27 hpf embryos ranged from 101.4 µm ± 26.6 SD to 85.6 µm ± 23.5 SD along the anterior-posterior axis ( Supplementary Fig. S1). In comparison, axonal length at 27hpf embryos treated with NCI-65828 at 10 hpf was significantly decreased ranging in length from a maximum of 68.3 µm ± 14.7 SD at the anterior to 45.8 µm ± 5.1 SD towards the posterior. Similar effects were in embryos incubated in terrein (either 20 µM or 30 µM) from 10hpf (75.4 µm ± 14.2 SD to 43.9 µm ± 18.1 SD and 75.1 ± 14.0 SD to 49.8 µm ± 9.4 SD respectively). By 36hpf, the axonal length of NCI-65828 treated embryos was similar to untreated embryos suggesting that at earlier stages axonal extension was retarded by NCI-65828 treatment but accelerated between 27-36 hpf. However, axons in terrein treated embryos remained shorter when compared with untreated and NCI-65828 treated embryos. Ectopic branching was more frequent in embryos treated with NCI-65828 and 30 µM terrein in posterior axons ( Supplementary Fig. S1).

Effects on motor neurons on treatment with NCI-65828 and terrein after their specification.
Zebrafish embryos treated with NCI-65828, 20 µM or 30 µM terrein at 18.5 hpf were allowed to develop to 27 and 72 hpf. Embryos exhibited no gross morphological abnormalities when observed by brightfield microscopy at 27 hpf. Embryos under all conditions at this time point moved spontaneously, had beating hearts and visible dorsal, ventral, yolk sac and head vasculature.
Embryos incubated in inhibitors from 18 hpf, had a normal spinal cord similar to embryos treated at 10hpf and to control embryos. The CaP axon out-growth position or its angle relative to the spinal cord at 27 hpf was also similar to the control (Fig. 5B). The axons were shorter though not as severely affected as in embryos exposed to inhibitors from 10 hpf. Treatment with NCI-65828 or terrein caused branching in incorrect medial positions in both anterior and posterior segments. Increased branching was particularly noticeable in 20 µM terrein treated embryos which may be due to the presence of relatively longer axons when compared to those seen in NCI-65828 or 30 µM terrein treated embryos.
Immunostaining for Znp1 revealed a retardation of axon outgrowth, with lengths of trunk CaP axons decreasing in inhibitor treated embryos when compared with untreated embryos particularly in more posterior segments. The length of CaP axons in untreated embryos was consistent between each segment while axons in NCI-65828 treated embryos were significantly affected. This effect was also seen in embryos treated with 20 and 30 µM terrein. The effect on axonal growth was less severe at 20 µM terrein but more pronounced at 30 µM terrein and the retardation of axon outgrowth was comparable to that in embryos incubated in NCI-65828 ( Supplementary  Fig. S1). Few branches were observed on axons of untreated control embryos at 27 hpf. Treatment with NCI-65828 resulted in a loss of branches from the more posterior neurons ( Supplementary Fig. S1). Unexpectedly, treatment with 20 or 30 µM terrein resulted in increased branching across all axons.
The positioning of caudal, rostral and medial primary motor neuron ( Fig. 5B; CaP, RoP and MiP axons, white, red and blue dotted lines) 40 were indistinguishable between treated and untreated embryos by 72 hpf. CaP, RoP and MiP motor neuronal axons in treated and control embryos show the same angle of projection, position of the curve towards the posterior at the midline (in the case of the CaP and RoP), and the curve of the MiP ventrally. Later born secondary motor neuron axons however, were found to be significantly longer in treated embryos under all three conditions with increased branching, which was more pronounced in terrein treated embryos ( Supplementary Fig. S2).
Znp1 staining also revealed mean CaP axon lengths to be consistent across trunk region analysed, both in the untreated controls and in embryos treated with inhibitors at 18.5 hpf and observed at 72 hpf. However, the RoP axons were longer after NCI-65828 treatment when compared to controls (249.7 µm ± 46.4 SD) or 20 and 30 µM terrein ( Supplementary Fig. S2). No difference in branching of RoP axons was seen with 20 or 30 µM terrein treatment when compared to controls. However, incubation in NCI-65828 caused an increase in branching, particularly in axons towards the posterior of the trunk. Secondary motor neuron (2°MN) axon length as well as branching increased after treatment either with NCI-65828 or terrein (Supplementary Fig. S2).
Effects on vasculature on treatment with NCI-65828 or terrein prior to MN specification. The dorsal aorta in both inhibitor treated and untreated embryos appeared to have no obvious defects. Inter-somitic vessels (ISV) were seen sprouting in all cases but were more developed in untreated embryos (Fig. 6A). Many ISVs were found to be sprouting laterally at the ventral side in untreated embryos, prior to formation of the dorsolateral anastomotic vessel (DLAV). This was not observed in any of the inhibitor treated embryos where ISV sprouts barely appear to develop past the midline. By 36 hpf, the DLAV was fully developed in untreated embryos whereas in treated embryos, although the ISVs were positioned correctly and reached the dorsal side, the DLAV did not appear as thick or complete. Growth of ISVs was severely retarded by NCI-65828 as well as both concentrations of terrein at 27 h. Embryos incubated in NCI-65828, 20 µM and 30 µM terrein for 27 h had substantially shorter ISVs (Fig. 6B) but by 36 h the vessels in NCI-65828 had grown to lengths comparable to those of the control. In contrast, ISVs in embryos incubated in 30 µM terrein did not attain the lengths seen in the untreated embryos and vessels stopped around the midline and did not connect with the dorsal vasculature both at 27 and 36 hpf. Besides growth and elongation of ISVs, we also analysed the effects on branching of the blood vessels. At 27 hpf, branching in the ISVs was rare in control embryos but appears slightly reduced in NCI-65828 treated embryos but surprisingly increased in 20 µM terrein alone (Fig. 6C).
Effects on vasculature on treatment with NCI-65828 or terrein after MN specification. When compared to embryos treated with inhibitors from 10 hpf, the ISVs in embryos from 18hpf appeared to be more developed (Fig. 6D). They were longer and thicker, extended past the midline and sprouted laterally towards the dorsal side. Inter-somitic vessels were shorter in embryos incubated in NCI-65828 or 30 µM terrein from 18.5 hpf to 27 hpf when compared to control. These effects were most striking in NCI-65828 treated embryos. However, in embryos treated (from 18 hpf to 72 hpf) with inhibitors, the ISVs were only marginally shorter suggesting that the shorter ISVs seen earlier are due to growth retardation in inhibitor treated embryos which eventually catch up. ISV truncations were rare in all 72 hpf embryos (Fig. 6E). By 72 hpf, the DLAV appeared well developed in both treated and untreated embryos. At 72 hpf, small vessels could be seen branching from the midline of the embryos. In control embryos these appeared typically perpendicular to the ISVs, however, many were found at incorrect angles after treatment with 30 µM terrein ( Fig. 6D; dotted lines, E,F). At 72 hpf, embryos incubated in the inhibitors from 18.5 hpf developed a dorsal curvature. While untreated fish remained straight and of consistent length, NCI-65828 treated embryos developed mild dorsal curvature. Incubation of embryos in 20 or 30 µM terrein caused a much more noticeable dorsal curvature which was severe in the higher concentration of 30 µM terrein. At the higher terrein concentration, a few embryos presented with shortened tails. other phenotypes in inhibitor-treated embryos. Swimming behaviour was abnormal in treated embryos and also more frequently observed than in controls. Embryos appeared to move using only pectoral fins, no sinusoidal movements using lateral muscles were seen (Supplementary Video S1). The fish appeared to be paralysed, as they were unable to straighten out during any movement and were unable to right themselves. These effects were seen embryos exposed to both inhibitors but were most severe in 30 µM terrein. Hearts beat normally in all cases and blood was observed to be circulating through dorsal and ventral vessels.

Muscle development is not affected in embryos treated with NCI-65828 or terrein.
To determine whether the movement defects we see in the inhibitor treated fish are due effects on the muscles, we examined the patterning of the myotomes. Morphologically there appeared to be no defects in the myotome which was further confirmed by staining with the muscle marker mAb F59 41,42 (Supplementary Fig. S3).

Discussion
We have used the well characterised programme of motor neuron development in the zebrafish embryo to demonstrate that the RNase activity of the ANG homolog Rnasels is required for normal developmental progression. By blocking both the RNase activity of the Rnasels using the specific inhibitor NCI-65828 and the secretion of the RNasels by terrein, we show growth retardation and impaired pathfinding in the primary motor neurons of the spine, as well as increased branching in secondary motor neurons. We have previously identified a role for ANG and its mouse homolog in neurite outgrowth in vitro 11,14,24 and here we show this for the first time in vivo. Zebrafish have also been used to study the role of other genes implicated in ALS, such as SOD1 and C9orf72 43,44 . The defects we observed shed light on the effects of ALS-associated human ANG variants with active site mutations resulting in grossly impaired or loss of RNase activity (e.g. K40I, K17I) or those with signal peptide mutations predicted to result in impaired secretion (P-4S).
The effects of NCI-65828 and terrein on axon development are similar to the effects of NCI-65828 seen previously in mouse motor neurons differentiated from stem cells in culture 11,14 . Similar effects are seen in perturbation studies of other genes such as survival motor neuron (SMN) and unplugged mutants [45][46][47] which shows motor neuron truncation and arborisation to a greater extent than that induced by NCI-65828 treatment. Unplugged functions as part of the signalling system, guiding motor neuron axons along their stereotypical pathways, and it is possible that Rnasels perform a similar role between motor neurons and their surroundings. This is suggested by the findings of Wei et al. 16 , that hANG interacts with molecules such as β-actin and α-actinin. They also showed that hANG is present at the leading edge of migrating cells and cultured mammalian cells failed to migrate properly in the absence of hANG. We have also observed that hANG associates with GAP-43 in the early stages of neuronal differentiation from embryonal carcinoma cells 11 .
The results presented here also suggest that Rnasels have physiological roles in angiogenesis in vivo like human ANG. NCI-65828 has been shown to inhibit angiogenesis in HUVEC tube forming assays 9 . Thus ANG like Rnasels function in vivo like VEGF in both vascular and neuronal development 48 . Inhibition of Rnasels by NCI-65828 and terrein shown here cause vascular aberrations primarily in the trunk vasculature of embryos which may or may not become more pronounced in adulthood. Rnasels regulate rRNA production and kinase phosphorylation, suggesting a role -even if indirect -in signalling processes 28,49 . Evidence for a direct role comes from characterisation by Chamoux et al. 50 , of ANG receptors on capillaries in the bovine brain while signalling pathways have been shown by Gho and Chae 51 followed by Weidlocha 52 .
Although we cannot exclude the possibility that NCI-65828 acts on other molecules, three lines of evidence, taken together, point to ANG as the most likely target. Kao et al. 9 found that (i) An analogue of NCI-65828 which was significantly lower effect on the enzymatic activity of ANG did not have a similar effect on neovascularization. Minor changes in ligand structure markedly reduced potency, suggesting that the inhibition of ANG activity was through an active-site rather than nonspecific binding which was supported by observations from a computationally generated model of the ANG.65828 complex (ii) Tumours from mice treated with NCI-65828 had fewer blood vessels in the interior of the tumour as compared control groups. This has also been reported for two antagonists that were demonstrably ANG specific (mAb and antisense) 53,54 (iii) Data reported in the NCI web site (http://dtp.nci.nih.gov/docs/cancer/searches/cancer_open_compounds.html) show that NCI-65828, at concentrations up to 100 μM, did not inhibit the growth in culture of PC-3, HT-29, or any of 57 other human tumour cell lines tested. These data suggest that the effects of NCI-65828 are most likely through its Ang inhibitory activity.
The inhibition of Rnasel activity by NCI-65828 and the inhibition of secretion by terrein occurs ubiquitously throughout the developing embryo, hence it is difficult to conclude whether the defects observed in the developing nervous system are primary or secondary to those seen in the developing vascular system. Both phenotypes are unlikely to be due to general developmental deficits since we did not observe segmental defects in adaxial muscle patterning, defects in anterior/posterior vascular tissues such as the heart, dorsal aorta or cardinal vein, or the positioning of the spinal cord motor neuron nuclei.
Our previous findings from a detailed three-dimensional crystal structure analysis clearly suggest that Rnasel-1 is most similar to hANG (human angiogenin, with a root mean square deviation of 1.6 Å over equivalent 112 C α atoms), including the unique feature such as the active site in obstructed by the C-terminal segment of the molecule (as observed in angiogenin) 27 . The inhibition of Rnasel1 by NCI-65828 is comparable to that observed for hANG. These findings combined with the observations on the spatiotemporal expression of the Rnasels suggest that the effects on the motor neurons and the vascular system are likely to be mediated by Rnasel-1 as it is seen to be expressed at sufficiently high levels in the eye, heart, brain and spine, as with its mouse and human orthologs 11 . Our detailed study on Rnasel expression and inhibitor studies suggest that hANG-like Rnasels play important roles in axonal pathfinding as well as in angiogenesis in zebrafish. Using an in vivo model we demonstrate that the RNase activity and secretion of angiogenins are essential for their function in the development of the nervous system providing insights into the effects seen in ALS patients with mutations in the signal sequence as well as the active site.

Methods
RNase assays and inhibition analysis of zebrafish RNases. All RNases (except RNase A-from Sigma-Aldrich) were prepared in [Met −1 ]-form as reported 55 and authenticated using mass spectrometry. Compound NCI-65828 (Fig. 1D) was the kind gift of Dr Robert Shapiro; NCI-65828 (Tyger Scientific as reported) 9 . The fluorogenic substrate 6-FAM-mAmArCmAmA-Dabcyl was synthesized by Integrated DNA Technologies (Coralville, IA).
Concentrations of protein and 6-FAM-mAmArCmAmA-Dabcyl solutions were determined spectrophotometrically. For the proteins, extinction coefficients calculated by the method of 56

Kinetic analyses.
A modification of earlier fluorimetric methods 9,57 was employed. Briefly, assays were conducted at 37 °C in a Perkin-Elmer LS-50B fluorimeter using stoppered semi-micro quartz cuvettes (Starna Scientific, Essex, UK). The fluorescence increase accompanying substrate cleavage was monitored using λ ex = 495 nm (slit width fixed at 5 nm) and λ em = 520 nm (slit width varied depending on quenching). Assay mixtures (1 ml) contained 0.02 M HEPES·NaOH, 0.1 M NaCl (pH 7.0), 0.002% (w/v) Tween-20 and 100 nM 6-FAM-mAmArCmAmA-Dabcyl. To each assay, either 20 µl DMSO or 20 µl inhibitor solution (solvent = DMSO) were added. Fluorescence was recorded during a 10-min equilibration period, after which the reaction was initiated by addition of 10 µl of enzyme stock to give a final concentration of either 195 nM Rnasel-1a, 77 nM Rnasel-3e or 370 nM hANG. The progress curve was recorded for 20-25 min, after which time 5 µl of 20 µM RNase A was added to take the reaction to completion. Recording continued until the reading stabilized (typically after 10 min).
Initial trials revealed that inclusion of 0.002% (w/v) Tween-20 as a surface-blocking agent increased initial rates by 10-20% and improved the linearity of the initial phase of the progress curve when compared with equivalent assays that included 10 µg/ml BSA 9 instead. Initial rate (v), initial fluorescence (F 0 ) and maximal fluorescence (F max ) were obtained using Grafit 5 (Erithacus Software, Surrey, UK). For v and F 0 estimation, a straight line was fitted to the data from the first 10 min of the progress curve, corresponding to ≤ 2% substrate cleavage. When progress curves showed a transient, data from the 10 min immediately following the transient were used. It was expected that  , which was then blocked in 5%Marvel 0.1% Tween20 (Sigma) in PBS for 1 h RT then incubated with mouse anti-HA (Covance) 1:5000 overnight at 4 °C. The membrane was then washed 4 × 5 min with PBST and incubated with HRP conjugated anti-mouse (Sigma) 1:5000 for 2 h at RT. After another 4 × 5 min wash with PBST the membrane was incubated with ECL reagents and exposed to Amersham Hyperfilm MP. GAPDH was used as loading control, HRP conjugated anti-GAPDH (Abcam) was used as above at 1:5000. Immunocytochemistry. Fixed cells were rehydrated to PBS and washed twice for ten minutes before blocking for 1 h at room temperature in PBS with 0.1% gelatin, 0.5% FBS and 0.1% Triton-X100, and incubated with primary antibodies overnight in the same buffer. After washing four times with PBST (PBS with 0.1% Triton-X100) for ten minutes each cells were incubation with secondary antibodies for two hours again in blocking buffer. After a further four PBST washes, samples were mounted with Mowiol. Z-stacked images were acquired using a Leica DM5500B microscope, DFC 360FX camera and LAS software and deconvoluted. See Table 1 for antibody information.

Treatment of SH-SY5Y and SH
Quantification of cleaved Caspase 3. Counts of total DAPI positive and cleaved caspase 3 positive cells were made in five randomly selected fields on each coverslip from each condition from two independent experiments. Each field contained a mean of 600 cells. Data presented is the mean of the percentage of total cells positive for cleaved caspase 3.
Zebrafish breeding. ABWT Zebrafish 62 were kept in tanks (Aquatic Habitats system, S66U-2 model) at 28 °C containing water of pH 7.5-8.0 and were fed brine shrimp. Breeding was conducted using two pairs of four-month-old fish per breeding tank in a light-dark cycled 28 °C incubation chamber. Light-dark cycle consisted of 10 hours dark and 14 hours light. Fish were acclimated overnight and permitted to breed at first light. When breeding was successful eggs were collected at 1 hpf and kept in E3 saline solution containing methylene blue during development. Mastermix comprising iQ SYBR Green supermix (BioRad), water and cDNA were made for quantitative PCR. Appropriate primers were added and divided into three replicates for each gene on PCR plates (Thermo). Final reactions were 20 µl in volume with 0.1 µM primers and 10 ng cDNA (see Table 2 for primer sequences). Reactions were carried out in a BioRad iQ5 cycler. Dynamic well factors were collected for 2 min 30 sec, then forty cycles at 60 °C and 95 °C for 20 s each followed by a melt curve. Expression levels were determined relative to 18 s rRNA 63 from baseline subtracted curves and corrected using primer efficiencies determined previously from serial dilutions of PCR product. Each was performed on cDNA from two individual pooled adult organ or embryonic preparations.
Treatment of zebrafish embryos with NCI-65828 and terrein. Embryos were collected in E3 at 28 °C then sorted and synchronised at shield stage (6hpf). One hour prior to drug addition, embryos were dechorionated using Pronase (Sigma) and washed extensively in E3 without methylene blue. Methylene blue was omitted from all subsequent E3 during drug treatment due to a reaction with NCI-65828 resulting in precipitation. Up to Wholemount immunohistochemistry. Ten to twenty embryos were rehydrated to PBS and washed four times for 30 minutes with 1 ml incubation buffer (IB; 1% bovine serum albumen (Sigma), 0.5% Triton X100 (Sigma) and 1% DMSO in PBS). After a final 30 minute wash in IB with 1% goat serum (Sigma) embryos were incubated overnight at 4 °C in the same buffer with rocking using primary antibodies at the dilutions indicated in Table 2. Following four washes with 1 ml IB embryos were again incubated overnight at 4 °C with rocking with secondary antibodies diluted in IB with 1% goat serum. After one final wash with IB, three further washes in PBS were performed for 30 min each.
Mounting and imaging of fluorescently-labelled embryos. Embryos were de-yolked using pins and briefly washed with PBS to remove debris then cleared through 25%, 50% to 70% glycerol (VWR) in PBS before mounting and imaging. Heads were cut from the trunk just posterior to the otic vesicles before mounting in glycerol under a coverslip (Thermo, 18 × 18 mm, #1.5) supported by vaseline. Trunks were mounted right-side down while heads were mounted ventral-side down. Embryos were viewed using a Leica DM5500B microscope fitted with a Leica CTR5500 lightbox and DFC360FX camera. Images were acquired using Leica Application Software (LAS) version 2.1.2, Z-stacks of images were deconvoluted in the same software.
Quantification of zebrafish trunk motor neurons and intersomitic vessels. Measurements were taken from seven trunk motor neurons starting from immediately posterior to where the yolk sac joins the yolk proper and identified by Znp1 staining. Neurite lengths and branching were measured by tracing through 3D volumes in Z-stacks using the Fiji distribution of imageJ (fiji.sc/Fiji) and the plug-in Simple Neurite Tracer (fiji.sc/ Simple_Neurite_Tracer). The same process was used to quantify inter-somitic vessel (ISV) length and branching, identified by Fli:GFP expression. Truncations were scored manually where ISV failed to reach the dorsal longitudinal anastomotic vessel and observations were made of sprouting at the midline. Motor neuron nuclei identified by Islet staining were counted in the same region and analysed as three equal-sized bins corresponding to the anterior, middle and posterior area. To ensure any defects seen were not due to underlying body-plan defects, observations were made of myosin heavy chain staining (F59) in the same region, none were found. statistics. All statistical tests were performed in SPSS 22 (IBM). Neurite and vascular quantification was performed on at least ten embryos from each time-point from two experiments were used. Measurements from neurons in equivalent positions posterior to the yolk-yolk sac boundary were pooled between fish in the same group and compared between treatments. Measurements were made from both left and right-side neurons and vasculature and although the same defects were identified consistently different lengths were obtained for equivalent left and right side neurons in fish within the same group (control and treated). These may have been an artefact due to imaging through the embryo for right-side neurons. As these were present in the control also, only the left side data has been presented here. Normality was tested using the D' Agostino-Pearson normality test prior to ANOVA testing with Tukey's post-hoc.