Structural mechanism of angiogenin activation by the ribosome

Abstract

Angiogenin, an RNase-A-family protein, promotes angiogenesis and has been implicated in cancer, neurodegenerative diseases and epigenetic inheritance^1-10. After activation during cellular stress, angiogenin cleaves tRNAs at the anticodon loop, resulting in translation repression^11-15. However, the catalytic activity of isolated angiogenin is very low, and the mechanisms of the enzyme activation and tRNA specificity have remained a puzzle^3,16-23. Here we identify these mechanisms using biochemical assays and cryogenic electron microscopy (cryo-EM). Our study reveals that the cytosolic ribosome is the activator of angiogenin. A cryo-EM structure features angiogenin bound in the A site of the 80S ribosome. The C-terminal tail of angiogenin is rearranged by interactions with the ribosome to activate the RNase catalytic centre, making the enzyme several orders of magnitude more efficient in tRNA cleavage. Additional 80S-angiogenin structures capture how tRNA substrate is directed by the ribosome into angiogenin's active site, demonstrating that the ribosome acts as the specificity factor. Our findings therefore suggest that angiogenin is activated by ribosomes with a vacant A site, the abundance of which increases during cellular stress^24-27. These results may facilitate the development of therapeutics to treat cancer and neurodegenerative diseases.

Extended Data Fig. 1 |. Cryo-EM characterization of angiogenin interactions with the ribosome.

(a) Example of hand-picked ribosome particles (red circles) on one of the higher contrast micrographs (n = 600) containing rabbit reticulocyte lysates treated with angiogenin. Representative micrograph had a defocus of −1.4 μm. Notice low contrast of ribosomes among concentrated lysates. (b) Examples of automatic particle picking of ribosome in cisTEM (red circles) on a micrograph containing in vitro assembled 80S–angiogenin(H13A) ribosome complexes. Micrograph was taken on same microscope and has similar defocus and CTF fitting resolution criteria to micrograph in (a). (c) Maximum-likelihood classification in Frealign of a dataset of in vitro assembled 80S ribosome complexes with angiogenin reveals angiogenin bound to the A site of ribosomes in the non-rotated (classical) state. (d) Masked, Fourier shell correlation curve as a function of inverse resolution for the map derived from 45,850 particles shown in Extended Data Fig. 1c (FSC_part from Frealign v9) (80S–angiogenin, black) or for the map from rabbit reticulocyte lysates derived from 1,960 particles shown in Fig. 1d,e (RRL–angiogenin, purple). (e-j) Examples of interactions of angiogenin with the A-site. Model of 80S–angiogenin is shown with the 2.8 Å cryo-EM map, sharpened by a B-factor of −50 Ų and shown as mesh at σ levels as noted in individual legends. (e) Interaction of angiogenin with H69 of 28S rRNA. Mesh is shown at 2.75 σ. (f) Interaction of angiogenin with 18S rRNA. Mesh is shown at 2.75 σ. (g) Binding of angiogenin near the universally conserved decoding centre residues of H69 (28S rRNA) and h44 (18S rRNA). Mesh for mRNA is shown at 3 σ; for angiogenin and h44, at 4 σ; and for H69 and G626 at 5 σ. mRNA is green for well modelled residues, grey for disordered residues. (h) Interaction of angiogenin with the A-site codon (mRNA is green for well modelled residues, grey for disordered residues). Mesh is shown at 3.5 σ both mRNA and angiogenin. (i) Interaction of angiogenin with P-site tRNA. Mesh is shown at 3.75 σ. (j) Interactions of angiogenin with the first nucleotide of the A-site codon adjacent to the P-tRNA anticodon and codon interaction. Mesh is shown at 4 σ.

Extended Data Fig. 2 |. Angiogenin cleaves various tRNAs but not isolated stem loops.

(a) Rigid-body docking of in vitro assembled 80S–angiogenin–Ala-tRNA^Ala–eEF1A–GDPCP model into the RRL–angiogenin cryo-EM map (grey mesh at 6 σ for most of complex, but due to sub-stoichiometry, blue mesh at 4 σ for ternary complex) shows good agreement of in vitro model with the complex visualized directly in rabbit reticulocyte lysates. The density shows that the C-terminal residues of angiogenin form an extended beta strand in lysates and that the ribosome and ternary complex position the anticodon of tRNA near the active site of angiogenin in lysates. (b) Activity of 10 nM angiogenin alone or together with the 80S ribosome complex (including model mRNA encoding phenylalanine) was checked against total rabbit tRNA purified from rabbit reticulocyte lysates. After 100 min, cleavage reactions were separated by denaturing 15% Urea-PAGE gels stained with SYBR Gold (top) then later detected via Northern blotting (bottom) for the 5′ end of three different rabbit tRNA as noted (see Methods). tRNA fragments of all three tRNAs tested increase in the presence of the 80S–angiogenin complex and are prevented by RNasin (n = 2 independent experiments). (c) Activity of angiogenin alone or together with the 80S ribosome complex (including tRNA^fMet and model mRNA encoding phenylalanine in A site) was checked using in vitro translated yeast tRNA^Ala, an extended 33-mer hairpin including the yeast tRNA^Ala anticodon stem loop, or a 17-mer including only the yeast tRNA^Ala anticodon stem loop (see Extended Data Fig. 2d). At the specified time (in minutes), cleavage reactions were separated by denaturing 15% Urea-PAGE gels stained with SYBR Gold. Full-length tRNA^Ala is depleted over time while fragments arise in the presence of the 80S–angiogenin complex, but the 33-mer and 17-mer are not depleted suggesting tRNA is cleaved more effectively than a simple hairpin. Weak tRNA-fragment-sized bands appearing in 33-mer and 17-mer lanes likely arise from tRNA^fMet, which is present in the 80S complex for positioning in the P-site. (n = 2 independent experiments). (d) Diagrams depicting the secondary structure of in vitro translated yeast tRNA^Ala, the extended 33-mer hairpin including the yeast tRNA^Ala anticodon stem loop, or the 17-mer hairpin used in Extended Data Fig. 2c. Larger, bold residues are the anticodon of tRNA^Ala. (e) Quantification of full length tRNA and hairpins from Extended Data Fig. 2c. The amount of full-length tRNA decreased by 50–66% from 1 to 100 min while the amount of hairpins stays constant. (bar shows mean of n = 2 independent experiments).

Extended Data Fig. 3 |. Purification and characterization of WT, H13A and K54E angiogenin.

(a) SDS PAGE showing inclusion-body expressed, refolded and purified WT, H13A and K54E angiogenin (“refolded” Angiogenin see Methods, Preparation of Angiogenin) (n = 1). (b) 200 nM WT refolded angiogenin co-pellets with purified ribosomes more efficiently than angiogenin(K54E). Bovine pancreatic RNase A does not co-pellet with ribosome under these conditions (n = 1). (c) RRL translation of nanoluciferase-encoding mRNA is inhibited by 90 nM refolded WT but not H13A or K54E angiogenin. These data, excluding RNasin controls, also appear in Fig. 2k. (mean +/− s.d., n = 2 for mutants or 6 for controls, independent experiments). (d) Refolded WT angiogenin and angiogenin(K54E) but not angiogenin(H13A) exhibit weak RNase activity against total yeast tRNA in the absence of ribosomes, at high enzyme concentration and high tRNA concentration^, (50 and 300 times higher, respectively, than in the ribosome-activated cleavage assay), indicating that K54E mutation does not affect the basal catalytic activity (Methods) (mean +/− s.e., n = 2 independent experiments). (e-f) SDS PAGE showing recombinant, soluble WT angiogenin (e) and angiogenin(H13A) (f) (“soluble” angiogenin, Methods, Preparation of Angiogenin) in comparison with 200 ng of R&D angiogenin. * is a contaminant in soluble angiogenin(H13A) prep (n = 1). (g) Western blotting of soluble angiogenin(H13A) and R&D angiogenin shows similar amounts are loaded (n = 1). (h) Comparison of tRNA^Ala cleavage by R&D vs. soluble WT and H13A angiogenin (10 nM). A ternary complex containing Ala-tRNA^Ala, eEF1A and GDPCP was treated with angiogenin in the presence or absence of the 80S ribosome complex with UUC-containing mRNA and tRNA^fMet. After 0 or 10 min, cleavage reactions were separated by denaturing 15% Urea-PAGE gels and stained with SYBR Gold. R&D and soluble WT angiogenin exhibit the same activity, while angiogenin(H13A) is inactive. (mean, n = 2 independent experiments). (i) Translation of nanoluciferase-encoding mRNA is inhibited by soluble WT angiogenin (similarly to R&D angiogenin shown in Fig. 1a), but not by the soluble angiogenin(H13A) (RLU: relative light units). Example trace is shown. (j) Apparent rate of translation, measured as RLU/s (see Extended Data Fig. 3i), in the presence of soluble WT angiogenin or angiogenin(H13A), (mean +/− s.d., n = 3 independent experiments except buffer control which was duplicated for n = 6).

Extended Data Fig. 4 |. Effects of ribosome complex composition on angiogenin activation.

(a) Production of tRNA^Ala fragments by angiogenin is stimulated by an elongation-like 80S ribosome complex more efficiently than by 40S or 60S subunits. A ternary complex containing Ala-tRNA^Ala, eEF1A and GDPCP was supplemented to 80S ribosome complexes assembled using 40S, 60S, leaderless mRNA placing phenylalanine UUC codon in A site, tRNA^fMet complementary to the P-site, and/or angiogenin. After 10 min, cleavage reactions were separated by denaturing 15% Urea-PAGE gels stained with SYBR Gold (top) then later detected via Northern blotting for the 5′ end of yeast tRNA^Ala (bottom), n = 2 independent experiments. (b) Same reactions as in (a) run separately then probed for the 3′ end of yeast tRNA^Ala, n = 2 independent experiments. (c) Close-up of lane 14 from (a) showing SYBR Gold and 5′ Northern side-by-side. Two 5′ tRNA fragments are apparent with the more intense one running above the 30 nt marker. (d) Close-up of lane 14 from (b) showing SYBR Gold and 3′ Northern side-by-side. 3′ tRNA fragment runs close to the 40 nt marker. (e) Secondary structure diagram of tRNA^Ala with the anticodon shown in orange font. Triangles indicate potential cleavage locations in the anticodon stem based on preference of angiogenin to cleaved at pyrimidine-adenine dinucleotides. (f) Cleavage of tRNA^Ala by angiogenin is stimulated by multiple 80S ribosome complexes including ones lacking mRNA, lacking P-site tRNA, and ones encoding amino acids phenylalanine (F, UUC) or lysine (K, AAA), or a stop codon (X, UAA) as well as by truncated mRNA (1, single uridine in A site). A ternary complex Ala-tRNA^Ala, eEF1A and GDPCP was supplemented to ribosome complexes assembled using 40S, 60S, mRNAs as specified, tRNA^fMet complementary to the P-site, and/or angiogenin, and after the indicated time (in minutes) cleavage reactions were stopped then separated by denaturing 15% Urea-PAGE gels stained with SYBR Gold (top) then later detected via Northern blotting (bottom) for the 5′ end of yeast tRNA^Ala. (mean, n = 2 independent experiments).

Extended Data Fig. 5 |. Cryo-EM of the 80S–angiogenin and 80S–angiogenin(H13A) complexes with Ala-tRNA^Ala–eEF1A–GDPCP ternary complex.

(a) Maximum-likelihood classification in Frealign of a dataset of 80S ribosome with angiogenin and ternary complex of Ala-tRNA^Ala–eEF1A–GDPCP reveals classes with angiogenin bound to the ribosomal A site as tRNA is presented to the angiogenin active site. (b) Masked, Fourier shell correlation curves (FSC) as a function of inverse resolution for the cryo-EM map described in Extended Data Fig. 5a. (c) Maximum-likelihood classification in Frealign of a dataset collected from 80S ribosomes with recombinantly expressed angiogenin(H13A) and ternary complex of Ala-tRNA^Ala–eEF1A–GDPCP. (d) Masked, Fourier shell correlation curves (FSC) as a function of inverse resolution for the cryo-EM maps described in Extended Data Fig. 5c. (e) Half reactions for assembling the cryo-EM complexes in Extended Data Fig. 5a (Half reaction 1: 80S ribosome complexes with angiogenin; Half reaction 2: ternary complex of Ala-tRNA^Ala–eEF1A–GDPCP as described in Methods) were mixed in test tubes held on wet ice. After the indicated time, the reactions were quenched with 8 M Urea loading buffer. For the 0-s time point, Half reaction 1 and Half reaction 2 were added directly to 8 M Urea loading buffer. The reactions were separated on a 15% Urea-PAGE gel to visualize full length and newly cleaved tRNA. tRNA is cleaved nearly maximally by 25 s on ice under these conditions, which mimic the conditions used to prepare cryo-EM grids (n = 1). (f) Half reactions for assembling the cryo-EM complexes in Extended Data Fig. 5c (Half reaction 1: 80S ribosome complexes with recombinant angiogenin(H13A); Half reaction 2: ternary complex of Ala-tRNA^Ala–eEF1A–GDPCP as described in Methods) were mixed in test tubes at 30 °C. After the indicated time, the reactions were quenched with 8 M Urea loading buffer and separated showing lack of tRNA cleavage under these conditions, which mimic the conditions used to prepare cryo-EM grids (300 s) (n = 1).

Extended Data Fig. 6 |. Cryo-EM density maps of the 80S–angiogenin(WT) and angiogenin(H13A) complexes formed with Ala-tRNA^Ala–eEF1A–GDPCP.

(a) 3.7-Å cryo-EM map of 80S–angiogenin(WT)–Ala-tRNA^Ala–eEF1A–GDPCP has density consistent with GDPCP (cyan) bound to eEF1A (blue). GDPCP-bound bacterial EF-Tu (cyan) (PDB: 5UYM) was aligned via the homologous domain 1 to domain 1 of eEF1A. Mesh is shown at 2.75 σ. (b). The 2.8-Å cryo-EM map of 80S–angiogenin (WT), sharpened by a B-factor of −50 Ų, has well-resolved density for the active site residue H13. Mesh is shown at 4 σ. (c) The 2.8-Å cryo-EM map of 80S–angiogenin(H13A) (no sharpening) has well-resolved density for the active site of angiogenin supporting mutation of histidine 13 to alanine (H13A) (mesh). The H13A map was rigid-body fitted with our 80S–angiogenin model. Mesh is shown at 4 σ. (d-f) Comparison of density for tRNA^Ala in the 3.7-Å cryo-EM map of 80S–angiogenin(WT)–Ala-tRNA^Ala–eEF1A–GDPCP (d), 3.1-Å cryo-EM map of 80S–angiogenin(WT)–tRNA^Ala (e) or 3.9-Å cryo-EM map of 80S–angiogenin(H13A)–tRNA^Ala (f) show that the anticodon stem–loop (ASL) of tRNA is strongest in the H13A map with the catalytically inactive angiogenin. The cryo-EM maps (grey mesh) are shown without sharpening at 4 σ. (g-i) Comparison of cryo-EM density filtered to 5 Å in the cryo-EM maps of 80S–angiogenin(WT)–Ala-tRNA^Ala–eEF1A–GDPCP (g), 80S–angiogenin(WT)–tRNA^Ala (h) or 80S–angiogenin(H13A)–tRNA^Ala (i) further supports that the ASL of tRNA is stronger in the H13A map with catalytically inactive angiogenin. (j-l) The cryo-EM maps of 80S–angiogenin(WT)–Ala-tRNA^Ala–eEF1A–GDPCP (j), 80S–angiogenin(WT)–tRNA^Ala (k) or 80S–angiogenin(H13A)–tRNA^Ala (l) show an L-shaped density consistent with tRNA interacting with the P-stalk via the tRNA elbow. The cryo-EM maps were each low-pass filtered to 5 Å and shown at 3 σ. (m) Cryo-EM map of 80S–angiogenin–Ala-tRNA^Ala–eEF1A–GDPCP shows density for eEF1A (blue) bound to the shoulder of the 40S (gold). The cryo-EM map was low-pass filtered to 5 Å and is shown at 2.7 σ. (n) The cryo-EM map of 80S–angiogenin–Ala-tRNA^Ala–eEF1A–GDPCP has an unidentified low-resolution density (question mark) next to the anticodon arm of Ala-tRNA^Ala (green) and the P-stalk (uL11, light blue is labelled for reference), which may correspond to a dynamic element of the P-stalk. The cryo-EM map was low-pass filtered to 5 Å and is shown at 2.5 σ.

Fig. 1 |. Angiogenin inhibits translation by binding to 80S ribosomes.

a, Angiogenin (Ang) inhibits translation in RRL in a concentration-dependent manner. Inhibition is relieved by RNasin (RNase inhibitor). The apparent translation rate is shown (relative light units (RLU) s⁻¹). Data are mean ± s.e.m. n = 2 or 3 independent experiments, as shown. b, Incubation of RRL with angiogenin for 15 min leads to the accumulation of tRNA fragments; fragment accumulation is inhibited by RNasin. n = 2 independent experiments. nt, nucleotides. c, Angiogenin co-pellets with the ribosome fraction from RRL, except in the presence of RNasin. n = 2 independent experiments. d, Cryo-EM analysis of RRL with 1 μM angiogenin reveals that over 30% of 80S ribosomes have a novel density in the A site. e, Rigid-body fit of human angiogenin (PDB: 5EOP) into the 3.4 Å cryo-EM map of angiogenin-bound ribosomes in RRL. f, A 3.0 Å cryo-EM map of in vitro assembled 80S–angiogenin complex. The 60S subunit is coloured cyan; 40S subunit is coloured yellow; P-site tRNA is coloured orange; mRNA is coloured green; angiogenin is coloured magenta. g, Angiogenin in the A-site of the ribosome with the secondary structure labelled. h, Cryo-EM density showing well-resolved interactions of angiogenin with P-tRNA and 18S rRNA; the three catalytic residues are shown in red. C-terminal residues of angiogenin (118 to 123) and h30 have been omitted to show P-tRNA interactions. The mesh shows a 2.8 Å cryo-EM map with B-factor sharpening of −50 Ų. i, Sequence alignment of human angiogenin with human RNase I; the secondary structure is indicated for angiogenin in the ribosome complex, including β-strand 5 joining β-strand 6 (5/6) and β-strand 7. The underlined red letters indicate catalytic residues^,. The blue letters indicate angiogenin residues adjacent to 28S rRNA. The gold residues are adjacent to 18S rRNA. The green residues are adjacent to mRNA. The orange residues are adjacent to P-site tRNA. Residues with mutations associated with ALS and Parkinson’s disease are indicated by circles^–. H.s., Homo sapiens.

Fig. 2 |. Angiogenin’s RNase is activated after ribosome binding.

a, The catalytic residues (red spheres) of angiogenin face the tRNA-delivery path, comprising the tRNA-binding P-stalk and SRL of the 60S subunit. b, Interaction between angiogenin and the ribosome is incompatible with angiogenin’s interaction with the RNase/Ang inhibitor 1 (grey) due to steric clash. Structural alignment compares ribosome-bound angiogenin with angiogenin bound to human RNase/Ang inhibitor 1 (PDB: 1A4Y). c, Catalytically active RNase A contains a C-terminal β-strand (left; PDB: 5RSA), whereas isolated angiogenin has an inhibited α-helical C terminus (middle; PDB:5EOP). Ribosome-bound angiogenin features a β-strand C terminus (right; this study) like RNase A. d, The C-terminal β-strand of angiogenin is stabilized by ribosomal h30 of 18S rRNA, uS19 and the β5/6 strand. e, Schematic of the measurement 80S-stimulated tRNA cleavage by angiogenin, showing the order of addition of components from left to right. f, Urea–PAGE gel showing that the production of Trna fragments by 10 nM angiogenin (R&D) is stimulated by the 80S ribosome. g, The ratio of signals of tRNA fragments to full-length tRNAs. Data are mean ± s.d. n = 3 independent experiments. h, The production of tRNA fragments by ribosome-bound angiogenin is further stimulated by eEF1A ternary complex with GTP (T) or GDPCP (C) and Ala-tRNA^Ala. i, The ratio of signals from tRNA fragments to full-length tRNA. Data are mean ± s.d. n = 3 independent experiments. j, WT and H13A, but not K54E, angiogenin co-pellets with ribosomes from RRL supplemented with 1 μM of each angiogenin. n = 2 independent experiments. Western blot analysis of the supernatant and ribosome pellet is shown. OD, optical density units at 260 nm. k, The concentration dependence of translation inhibition in RRL by WT, H13A or K54E angiogenin. IC₅₀ values are shown. n = 2 independent experiments. l, Comparison of ribosome-stimulated tRNA cleavage (in a ternary complex of eEF1A–GDPCP–Ala-tRNA^Ala) by R&D angiogenin and refolded WT, H13A or K54E angiogenin. NA, not applicable. Data are the mean of n = 2 independent experiments.

Fig. 3 |. Cryo-EM structures of 80S–angiogenin complexes formed with eEF1A–GDPCP and tRNA.

a, A 3.7 Å cryo-EM map of 80S–angiogenin bound to eEF1A–Ala-tRNA^Ala, coloured as in Fig. 1 with the addition of green for Ala-tRNA^Ala and dark blue for eEF1A. b, A 3.1 Å cryo-EM map of 80S–angiogenin bound to tRNA^Ala. c, The anticodon stem of Ala-tRNA^Ala (green) in the ternary complex is next to the angiogenin active site (red spheres), but the anticodon loop is not resolved due to cleavage and/or mobility (dashed green line). d, The anticodon stem of tRNA decoding mRNA (grey; from PDB: 5LZS) clashes with angiogenin (clashing residues are black), highlighting that angiogenin binding is incompatible with mRNA decoding. e, Different positions for tRNA^Ala among five different cryo-EM maps suggest tRNA mobility near angiogenin. The 60S P-stalk (shades of cyan) and tRNA (shades of green) exhibit different positions between classes indicative of ~8 Å movement. f, A cryo-EM map of 80S–angiogenin(H13A) bound with tRNA^Ala.

Fig. 4 |. Schematic of angiogenin activation and specificity toward tRNA induced by the 80S ribosome.

a, Under normal conditions, angiogenin is inhibited by RNase/Ang inhibitor 1, and active translation keeps the 80S ribosomal A sites mostly occupied. b, Cellular stress, intercellular signalling or diseases such as cancer can lead to degradation of RNase/Ang inhibitor 1 and/or an increase of free, inactive angiogenin, but the RNase activity of angiogenin is still limited by its inactive C terminus. Cell stress or disease can also cause the accumulation of 80S ribosomes with vacant A-sites, allowing angiogenin to bind and become activated by the refolding of C-terminal residues. Free or eEF1A-bound tRNAs, which are presented by the ribosome to angiogenin’s active site, are efficiently cleaved at the anticodon loop. This cleavage inhibits translation. Nicked tRNAs may affect other downstream processes or be unwound into tRNA halves or fragments to carry out their many roles in translation inhibition, apoptosis, angiogenesis and neuroprotection. Nicked tRNAs may also be returned to the tRNA pool, for example, by the RNA ligase RtcB.