Structure of the ALS Mutation Target Annexin A11 Reveals A

bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Structure of the ALS Mutation Target Annexin A11

2 Reveals a Stabilising N-Terminal Segment

3 Peder August Gudmundsen Lillebostad 1*, Arne Raasakka 1*, Silje Johannessen Hjellbrekke 1, 4 Sudarshan Patil 1, Trude Røstbø 1, Hanne Hollås 1, Siri Aastedatter Sakya 1, Peter D. Szigetvari 1, 5 Anni Vedeler 1,# and Petri Kursula 1,2,# 6 1 Department of Biomedicine, University of Bergen, Jonas Lies vei 91, 5009 Bergen, Norway 7 2 Faculty of Biochemistry and Molecular Medicine & Biocenter Oulu, University of Oulu, Aapistie 7, 90220 8 Oulu, Finland 9 * and # These authors contributed equally 10 Correspondence: email: [email protected], tel: +47 55586438

11 Abstract: The functions of the annexin family of proteins involve binding to Ca2+, lipid 12 membranes, other proteins, and RNA, and the annexins share a common folded core structure at 13 the C terminus. Annexin A11 (AnxA11) has a long N-terminal region, which is predicted to be 14 disordered, binds RNA, and forms membraneless organelles involved in neuronal transport. 15 Mutations in AnxA11 have been linked to amyotrophic lateral sclerosis (ALS). We studied the 16 structure and stability of AnxA11 and identified a short stabilising segment in the N-terminal end 17 of the folded core, which links domains I and IV. Crystal structure of the AnxA11 core highlights 18 main-chain hydrogen bonding interactions formed through this bridging segment, which are 19 likely conserved in most annexins. The structure was also used to study the currently known ALS 20 mutations in AnxA11. Three of these mutations correspond to buried Arg residues highly 21 conserved in the annexin family, indicating central roles in annexin folding. The structural data 22 provide starting points for detailed structure-function studies of both full-length AnxA11 and the 23 disease variants being identified in ALS.

24 Keywords: Annexin A11; crystal structure; solution structure; calcium binding; membrane 25 binding; protein stability; folding; amyotrophic lateral sclerosis 26

27 1. Introduction 28 The annexin (Anx) family of Ca2+-binding proteins is widely distributed in eukaryotes, with 29 twelve Anxs found in vertebrates [1–3]. All Anxs share multiple structural elements, most notably 30 an evolutionarily conserved C-terminal Anx core. This core consists of four similar domains, each 31 comprised of ~70 amino acids arranged into five α helices, termed A-E [1,4]. The only exception is 32 AnxA6, which contains eight domains, most likely due to a fusion of duplicated anxA5 and anxA10 33 genes [5]. The five helices within each domain are connected via short loops, forming a right- 34 handed superhelix. Helices A and B as well as helices D and E are oriented pairwise in an 35 antiparallel fashion, while helix C is positioned as a bridge orthogonally to the others [4]. The loops 36 connecting helix A to helix B and helix D to helix E contain the Ca2+ binding sites [1,6]. 37 The Anxs are structurally distinct from the EF-hand and C2-domain Ca2+-binding proteins [2]. 38 The Anxs reversibly bind negatively charged lipids in a Ca2+-dependent manner, which is a 39 defining feature of this protein family [1]. The binding is mediated through the convex side of the 40 core structure, coupled to a conformational change induced by Ca2+ binding [1,2]. Through lipid 41 binding, Anxs can affect important proteolipid membrane properties, such as curvature and phase 42 separation [7,8] . 43 While the folded C-terminal core structure of the vertebrate Anxs is conserved, the structure of 44 the N terminus varies greatly between different Anxs and determines their individual properties 45 [1]. This region typically contains 10 - 30 residues. However, AnxA7 and AnxA11 harbour 46 unusually long N termini. AnxA11 contains the longest N-terminal domain of all Anxs, with ~200 47 residues [9]. The N termini of AnxA7 and AnxA11 resemble each other, both being hydrophobic,

48 enriched in Gly, Tyr, and Pro [10–12]. The Ca2+-dependent interaction of AnxA11 with 49 phosphatidylserine changes the conformation of its N terminus [13]. Furthermore, the AnxA11 N 50 terminus interacts with the apoptosis-linked gene-2 protein (ALG-2) [14] and S100A6 (calcyclin) 51 [15]. 52 Rabbit AnxA11 appears to exist as two isoforms; AnxA11A and AnxA11B [16]. Only the 53 former binds S100A6 in the presence of Ca2+ via the region Gln49-Thr62. This region is absent in 54 AnxA11B [16–18], suggesting isoform-specific functions. S100A6 also binds to AnxA1 and AnxA2 in 55 addition to AnxA11 [19,20], indicating functional overlap between these three proteins. 56 In certain cell types, AnxA11 is predominantly localised to the cytoplasm, while in others, it is 57 mainly present in the nucleus [15,21]. It is likely that the A isoform is enriched in the nucleus, 58 whereas the B isoform is specifically localised to the cytoplasm [16]. The N terminus was suggested 59 to be responsible for the nuclear localisation [18,22]. In the nucleus, AnxA11 appears to be involved 60 in events related to the cell cycle. It participates in cytokinesis, playing an essential role in the 61 formation of the midbody. Cells lacking AnxA11 cannot complete cytokinesis and undergo 62 apoptosis [23]. 63 Expression levels of AnxA11 have been linked to various cancers [24–29]. Recent studies have 64 revealed that mutations in AnxA11 play an important role in the development of the 65 neurodegenerative disease amyotrophic lateral sclerosis (ALS) [8,30–36]. The molecular basis of 66 these mutations is unclear, as an experimentally determined 3D structure of AnxA11 has been 67 lacking. 68 Since the long N terminus of AnxA11 renders recombinant production of the protein 69 challenging, N-terminally truncated recombinant forms were generated, and the crystal structure of 70 one form, Δ188AnxA11, was determined. We show that a 4-residue sequence motif in the N 71 terminus of the core structure, conserved in most Anxs, functions as a Velcro-like bridging segment 72 and is important for the thermal stability of the Anx core structure. The known ALS mutations are 73 analysed based on the crystal structure, providing a view into conserved features of the Anx core 74 fold.

75 2. Materials and Methods

76 2.1. Cloning of full-length and N-terminally truncated forms of AnxA11 77 In order to obtain full-length rat AnxA11, RT-PCR was performed using total RNA isolated 78 from PC12 cells by the Trizol method [37]. The cDNA of AnxA11 was obtained using the iScript kit 79 (Bio-Rad) according to the manufacturer’s instructions and the forward primer 5′- 80 ATCCGGCCATGGGTATGAGCTATCCAGGCTATCCAC-3´ (NcoI restriction site is underlined) 81 and the reverse primer 5′- ATCCGGGGTACCTCAGTCGTTGCCACCACAGATC-3´ (Acc65I 82 restriction site is underlined). The anxA11 cDNA was digested with NcoI and Acc65I, purified after 83 separation in a 1% agarose gel, and ligated into the same restriction sites of the pETM10 vector (a 84 kind gift from Gunter Stier, Heidelberg). Two truncated versions of AnxA11 were generated by 85 PCR using the pETM10 plasmid harbouring rat full-length anxA11 cDNA as a template and using 86 the same reverse primer as for full-length AnxA11. The forward primers were 5′- 87 ATCCGGCCATGGGTAGAGGCACCATCA-3´ and 5′-ATCCGGCCATGGGTACCGATGCTTCTG - 88 3´(the NcoI restriction site is underlined), resulting in Δ188AnxA11 (starting with 189-RGTI) and 89 Δ192AnxA11 (starting with 193-TDAS), respectively. Both constructs code for the tag sequence 90 MLHHHHHHPMG at the N terminus. The cDNAs were ligated into the pETM10 vector after 91 digestion of PCR fragments with Acc65I and NcoI. Constructs were verified by DNA sequencing.

92 2.2. Expression and purification of full-length and N-terminally truncated forms of AnxA11 93 E. coli BL21(DE3) were transformed with AnxA11 expression plasmids. The bacteria were 94 grown in 5 ml LB medium with 50 µg/ml kanamycin until an OD600 of 0.6, after which 1 mM 95 isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to induce protein expression at +25 °C for 96 4 h or at +15 °C overnight (ON), to determine optimal expression conditions. At the end of protein bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

97 expression, 1 ml of the culture was withdrawn and centrifuged at 2000 g for 10 min at +4 °C. The 98 supernatant was discarded, and the pellet was resuspended in 500 µl lysis buffer (50 mM Na2HPO4, 99 500 mM NaCl, 5% (v/v) glycerol, 10 mM imidazole) containing 0.5 µg/ml DNase I (Sigma), 0.25 100 µg/ml RNase A (Sigma), and EDTA-free protease inhibitor cocktail (cOmplete, Roche). The samples 101 were sonicated on ice, whereafter 100-µl aliquots (total lysate) were withdrawn. Additional 100-µl 102 aliquots of the lysates were transferred into new tubes and centrifuged at 21000 g for 15 min at +4 103 °C. The supernatant was withdrawn. The pellet was resuspended in 100 µl lysis buffer. After 104 addition of denaturation buffer, 20 µl of the protein samples were separated on 4-20% SDS-PAGE 105 (Mini-PROTEAN TGX Precast Protein Gels, Bio-Rad) at 20 mA and 250 V as the limiting voltage. 106 For large-scale purification, the above procedure was repeated using 300 ml LB medium. The 107 full-length (FL) and N-terminally truncated AnxA11 were all induced at +25 °C for 4 h, while FL- 108 AnxA11 was also expressed ON at +15 °C. The cells were collected by centrifugation at 4000 g for 15 109 min. The pellets were frozen at -80 °C in lysis buffer before the addition of 0.5 µg/ml DNase I, 0.25 110 µg/ml RNase A, and EDTA-free protease inhibitor cocktail (cOmplete, Roche). Subsequently, cells 111 were broken by sonication, and the lysates were centrifuged at 16000 g for 30 min at +4 °C. All 112 purification steps were performed at +4 °C. 113 We took advantage of an N-terminal His-tag to purify AnxA11. In addition, a His-tag protects 114 the protein from degradation at the N terminus [13]. Purification on Co2+ instead of Ni2+ resin 115 resulted in higher purity. The lysate supernatant was loaded onto a Co2+-NTA agarose column and 116 incubated for 30 min. Subsequently, the proteins on the Co2+ resin were washed with equilibration 117 buffer (50 mM Na2HPO4, 0.3 M NaCl, 10 mM imidazole; pH 8) and high-salt buffer (50 mM 118 Na2HPO4, 1 M NaCl, 10 mM imidazole; pH 8), both buffers containing EDTA-free protease 119 inhibitors (cOmplete, Roche). Elution buffer (50 mM Na2HPO4, 0.3 M NaCl, 250 mM imidazole; pH 120 8) containing EDTA-free protease inhibitors (cOmplete, Roche) was used to elute the proteins. After 121 adding EGTA to 2 mM, the eluates were quickly transferred onto PD-10 columns for a buffer 122 exchange to 20 mM Tris, pH 8. All recombinant forms of AnxA11 were subjected to size exclusion 123 chromatography using a Superdex 75 or 200 Increase 10/300 GL column (GE Healthcare). Protein 124 concentration was determined by absorbance at 280 nm (using molecular masses of 55513, 36828, 125 and 36401 Da as well as sequence-based extinction coefficients of 42750, 13410, and 13410 M-1 cm-1 126 for full-length AnxA11, Δ188AnxA11, and Δ192AnxA11, respectively). The protein size and purity 127 were confirmed by SDS-PAGE and Coomassie Brilliant Blue staining.

128 2.3. Circular dichroism spectroscopy 129 A Jasco J-810 spectropolarimeter with a Peltier temperature control unit was used for far-UV 130 circular dichroism (CD) spectrocopy. Melting curves were recorded from +25 to +75 °C at 222 nm, at 131 a heating rate of 40 °C/h, to determine the thermal transition temperature (Tm). Tm was estimated in 132 GraphPad Prism using four-parameter logistic regression. 133 Synchrotron radiation CD (SRCD) data were acquired from 0.5 mg/ml Δ188AnxA11 and 134 Δ192AnxA11 in 20 mM Tris-HCl, pH 8.0 on the AU-CD synchrotron beamline at ASTRID2 (ISA, 135 Aarhus, Denmark). Samples containing 1 mM CaCl2 were included in addition to CaCl2-free 136 samples to detect Ca2+-induced changes in the secondary structure content. Closed 100-µm circular 137 cells were used (Suprasil, Hellma Analytics). Spectra were recorded from 170 to 280 nm at +10 °C. 138 Baselines were subtracted, and CD units were converted to Δε (M-1 cm-1) in CDtoolX [38]. The 139 spectra were truncated based on the HT voltage and the CD spectrum noise levels.

140 2.4. Small-angle X-ray scattering 141 SAXS data were collected from 0.41 – 2.96 mg/ml Δ188AnxA11 and Δ192AnxA11 in 20 mM 142 Tris-HCl on the EMBL/DESY P12 beamline, PETRA III (Hamburg, Germany) [39]. Fixed protein 143 concentration series at 0.7 – 1.0 mg/ml were measured with 0 – 1 mM CaCl2 to follow potential 144 oligomerisation. Monomeric bovine serum albumin (66.5 kDa) was used as a molecular weight 145 standard. Data were processed and analysed using the ATSAS package [40], and GNOM was used 146 to calculate distance distribution functions [41]. Theoretical scattering of the crystal structure was bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

147 calculated and fitted against the SAXS data using CRYSOL [42]. Ab initio modelling was performed 148 using GASBOR [43]. SUPCOMB [44] was used for crystal structure and ab initio model 149 superposition. SAXS data collection, processing, and fitting parameters are presented in Table 1.

150 Table 1. Small-angle X-ray scattering parameters and analysis. Data collection parameters Instrument P12, PETRAIII, DESY Wavelength (nm) 0.124 Angular range (nm-1) 0.022 - 7.33 Exposure time (s) 0.045 Exposure temperature (°C) 10 Protein AnxA11Δ188 AnxA11Δ192 Concentration range (mg/ml) 0.77 - 2.96 0.41 - 1.18 Structural parameters

I0 (relative) [from Guinier] 0.02759 0.0274

Rg (nm) [from Guinier] 2.44 2.43

I0 (relative) [from P(r)] 0.02692 0.02748

Rg (nm) [from P(r)] 2.500 2.503

Dmax (nm) [from GNOM] 10.00 9.97

VPorod (nm3) [from GNOM] 58.03 59.53 Molecular mass determination

Molecular mass Mr (kDa) [from I0 using Guinier]* 34.3 34.1

Molecular mass Mr (kDa) [from I0 using P(r)]1 33.5 34.2

Molecular mass Mr (kDa) [from VPorod] 34.1 35.0

Molecular mass Mr (kDa) [from absolute scale] 34.8 34.8

Theoretical Mr from sequence (kDa) 36.8 36.4 Software Primary data reduction & processing PRIMUS Ab initio modelling GASBOR χ2 0.770 0.629 Fitting of theoretical scattering curve to data CRYSOL χ2 0.888 0.749

151 1 Monomeric BSA standard was used as reference (Mr = 66.5 kDa; I0 = 0.0534)

152 2.5. Protein crystallisation, data collection, and structure determination 153 Sitting-drop vapour diffusion crystallisation experiments were set up in 96-well format. X-ray 154 diffraction data were collected on the EMBL P13 beamline, PETRA III (Hamburg, Germany) [45] 155 from a crystal grown in 100 mM Bis-tris (pH 5.5), 3 M NaCl at +20 °C; the drop contained 150 nl of 156 12.7 mg/ml Δ188AnxA11 and 150 nl of reservoir solution. The crystal was cryoprotected by adding 157 a solution containing 25% glycerol and 75% reservoir solution directly onto the crystallization drop. 158 Upon mounting, the crystal was flash-frozen using liquid N2, and kept at 100 K during X-ray 159 diffraction data collection. 160 Diffraction data were processed with XDS [46], and the structure was solved by molecular 161 replacement in Phaser [47] with PDB entry 1ANN (bovine AnxA4) as search model [48]. Refinement 162 was carried out using phenix.refine [49] and model building with coot [50]. The structure quality 163 was assessed with MolProbity [51]. The data processing and refinement statistics are presented in 164 Table 2. 165 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

166

167 Table 2. Crystallographic data collection and refinement. Construct ∆188AnxA11 Wavelength (Å) 0.9763 Space group P1 a = 39.1 Å, b = 86.7 Å, c = 87.3 Å Unit cell parameters = 114.0 °, β = 102.0 °, γ = 97.2 ° Resolution range (Å) 50 – 2.30 (2.36 – 2.30)1 Completeness (%) 95.4 (95.0) ⟨I/σI⟩ 5.2 (1.1)

Rsym (%) 17.5 (164.0)

Rmeas (%) 20.9 (196.6)

CC1/2 (%) 99.2 (47.6) redundancy 3.3 (3.2)

Rcryst/Rfree (%) 22.8 (26.6) Rmsd bond length (Å) 0.002 Rmsd bond angle (°) 0.4 MolProbity score (percentile) 1.58 (98th) Ramachandran favoured/disallowed 97.2/0.3 (%) 168 1 Numbers in parentheses refer to the highest-resolution shell.

169 For structure analysis, the programs PyMOL [52] and APBS [53] were used. The AnxA11 crystal 170 structure was deposited at the PDB with the entry code 6TU2.

171 3. Results and Discussion

172 3.1. Solubility assays, large scale expression and purification of the N-terminally truncated forms of AnxA11 173 Induction was performed at +25 °C for 4 h and at +15 °C ON to determine optimal expression 174 conditions. There was little or no difference in yield between the two modes of expression. FL- 175 AnxA11 was consistently more challenging to express than its truncated forms, yielded remarkably 176 low concentrations and was notoriously prone to aggregation, as reported previously by others 177 [13]. The solubility of the truncated forms of AnxA11 in bacterial lysates was assessed after 178 centrifugation at 21000 g by comparing samples from the total bacterial lysate with the 179 corresponding supernatant and pellet by SDS-PAGE (Figure 1). In contrast to FL-AnxA11, both 180 Δ188AnxA11 and Δ192AnxA11 were mainly soluble (Figure 1A). FL-AnxA11 as well as the N- 181 terminally truncated Δ188AnxA11 and Δ192AnxA11 were purified by affinity chromatography on 182 Co2+-resin, which greatly improved the purity of the recombinant both proteins compared to Ni2+- 183 based chromatography. The N-terminal truncations resulted in a marked enhancement in 184 expression, giving a five-to-tenfold increase in yield compared to FL-AnxA11 (Figure 1B). This 185 suggests that the N terminus is responsible for the aggregation of the full-length protein. This may 186 be related to the observed phase separation phenomena linked to the N-terminal segment [8], or 187 misfolding of the protein in bacterial cells.

188 3.2 Thermal stability of the N-terminally truncated Δ188AnxA11 and Δ192AnxA11 189 To address thermal stability, ellipticity at 222 nm was monitored as a function of temperature 190 for Δ188AnxA11 and Δ192AnxA11. The thermal transition temperatures were estimated from the 191 inflection points of the melting curves (Figure 1C). The Tm of FL-AnxA11 was earlier reported to be 192 +45 °C [13]. The Tm of Δ188AnxA11 showed a Tm of +50.76 ± 0.02°C, while that of Δ192AnxA11 was bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

193 substantially lower (+41.87± 0.02°C). This difference highlights the importance of the four amino 194 acid residues 189-RGTI-192 in thermally stabilising the core structure of AnxA11. Cooperative 195 unfolding of the two truncated forms of AnxA11 remained largely unchanged but decreased 196 slightly in Δ192AnxA11. This increased thermal stability of Δ188AnxA11 compared to Δ192AnxA11 197 may explain why we obtained crystals of Δ188AnxA11, but not of the other N-terminally truncated 198 form. The dramatic difference in Tm is necessarily accounted for by at least one of the four 199 additional residues present in Δ188AnxA11 being involved in the stabilising of the core structure.

200 201 Figure 1. The solubility and purification of N-terminally truncated forms of AnxA11 expressed at 202 15 °C overnight and their thermal stability. (A) Uninduced (U), total (T), supernatant (S), and pellet 203 (P) fractions of bacterial lysates containing Δ188AnxA11 or Δ192AnxA11 as indicated. (B) 10 µl of 204 FL-AnxA11 and the N-terminally truncated forms Δ188AnxA11 and Δ192AnxA11, purified using 205 Co2+-resin. Molecular mass standards are indicated to the right. Proteins were separated in 4-20% 206 SDS gradient gels and Coomassie-stained. (C) CD-monitored thermal disruption of α-helicity of 20 207 µM Δ188AnxA11 (black) and Δ192AnxA11 (red).

208 3.3. The crystal structure of AnxA11 209 Using Δ188AnxA11, we solved the crystal structure of the rat AnxA11 core domain. The 210 asymmetric unit contains 3 molecules of AnxA11, and the structure presents the canonical annexin 211 core fold. The Ca2+ binding sites are partially occupied, whereby 2 monomers have 3 Ca 2+ ions 212 bound, and one monomer has 4 (Figure 2). As is typical for the Anxs, all Ca2+ binding sites are on 213 one face of the molecule, promoting calcium-dependent lipid membrane binding (Figure 2).

214 215 Figure 2. Crystal structure of the AnxA11 core domain. The 4 subdomains are numbered and 216 coloured from blue to red. The N and C termini are indicated, and bound Ca2+ ions are shown in 217 magenta. Shown is one monomer of the 3 in the asymmetric unit.

218 The substantial difference in thermal stability induced by the 4 N-terminal residues of the 219 Δ188AnxA11construct implies an important structural element at this position. This segment ~10 220 residues before the AnxA11 domain I helix A are central in the folded structure, making 221 interactions with the termini of several helices of the Anx fold, and effectively enabling a tight 222 contact between the N terminus of domain I and the C terminus of domain IV (Figure 3A,B). bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

223 224 Figure 3. Rat annexins and conservation of the RGTI motif. (A) The N-terminal Velcro-like 225 bridging segment is shown in surface mode, bound to domain IV. (B) Stereo view of the hydrogen 226 bonding interactions of the motif with three helices in domain IV. (C) Sequence alignment of all rat 227 Anxs. The N-terminal motif is underlined in blue, and the mutations (red) and variants (green) 228 linked to ALS are shown with asterisks. A6 and A6c correspond to the N- and C-terminal halves of 229 AnxA6.

230 Comparisons with other Anxs show that the segment 189-192 is a conserved motif in the Anx 231 family. The only Anx that does not contain the motif at the sequence level is AnxA13; conservation bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

232 is weak for AnxA1 and AnxA8. The 4-residue segment forms conserved interactions with the ends 233 of helices in domain IV (Figure 3B). These interactions are mainly comprised of main-chain 234 hydrogen bonds, which explains the mode of conservation and the importance of the Gly residue in 235 the motif. Interestingly, it has been proposed that AnxA11 is the common ancestor of all 236 mammalian Anxs [54], except for AnxA7 and AnxA13 – the two phylogenetically oldest Anxs [55]. 237 Thus, the AnxA11 core domain has the highest similarity to the other mammalian Anxs, acting as a 238 sort of prototype [56]. 239 Interestingly, the Tyr23 and Ser25 phosphorylation sites of AnxA2 lie within the same 4- 240 residue region (Figure 3C). Phosphorylation at Tyr23 and Ser25 results in a closed or open 241 conformation, respectively, of AnxA2 [57]. These findings can be understood based on the 242 conserved structure of the observed bridging segment. 243 AnxA5 has been the most studied Anx on membrane surfaces, on which it forms a crystal-like 244 lattice of trimers [58–61]. The crystal structure of AnxA11 presents trimers very similar to those 245 formed by AnxA5, and hence, could be expected to interact with membranes in a similar manner. In 246 the crystal, the trimers are arranged in layers, with similar contacts as observed for AnxA4 and 247 AnxA5 between the trimers. Figure 4 shows the trimer arrangement of AnxA11; the arrangement in 248 the crystal resembles that of AnxA5 on membranes [58–61]. Like other Anx assemblies, the Ca2+ 249 binding sites of the AnxA11 trimer are all on the same face, which would be in direct contact with 250 the membrane (Figure 4A). The membrane-facing surface in the dimer gains a positive charge 251 potential upon Ca2+ binding, while the opposite side is relatively featureless (Figure 4B). This 252 observation further highlights the common function of the Anx core structure in binding membrane 253 surfaces, while the N-terminal domain is responsible for most functional differences between the 254 Anxs [1]. One might assume that AnxA11 would arrange on a membrane much like it does in the 255 crystal; the N-terminal segments would then point away from the membrane. This is in line with 256 the observations that AnxA11 induces membrane curvature [7], as well as the suggestion that the N 257 terminus of AnxA11 binds to RNA, while the core structure binds to lipids in lysosomes for long- 258 distance co-transport in neurons [8]. The RNA binding site in AnxA2 has been shown to reside 259 within domain IV [62]. The binding dynamics were recently investigated; AnxA2 binds strongly to 260 an RNA 8-mer (5′-GGGGAUUG-3′) with a Kd of 30 nM [63]. The RNA binding site(s) in AnxA11 261 have not been determined.

262 263 Figure 4. Trimeric assembly of AnxA11 in the crystal, shown as a surface with electrostatic 264 potential. (A) View from the Ca2+-binding face. Ca2+ ions are shown as green spheres. The negative 265 sites are the locations for Ca 2+ binding, and their neutralization will make the surface very positive, 266 enabling lipid membrane binding. (B) The opposite face of AnxA11 is relatively featureless. bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

267

268

269 3.4. Structure-based analysis of ALS disease mutations (ALS) 270 The crystal structure of rat AnxA11 allows to analyse the structural basis of AnxA11 mutations 271 linked to amyotrophic lateral sclerosis (ALS) [34] in humans. The sequence identity between full- 272 length rat and human AnxA11 is 93.3%, being 96.5% for the core structure. The identification of 273 AnxA11 as an ALS gene is rather recent, and mutations are being continuously identified. Thus, it is 274 safe to assume that more ALS mutations will be identified in the future. We studied the mutations 275 thus far linked to ALS with respect to potential structural implications. Many of the mutations, 276 including the best-characterised D40G [34], lie in the N-terminal extension, which is not present in 277 the crystallised construct. The mutations in this region could affect protein-protein interactions, 278 protein localisation, phase separation, or binding to mRNA [8]. More interesting for the current 279 structural study, however, are the mutations identified within the Anx core domain. 280 Five mutations have thus far been described in ALS patients within the AnxA11 core: S229R, 281 R235Q, R302C, R346C, and G491R; in addition, G189E is located two residues before the start of the 282 crystallised rat AnxA11 construct [34,36]. In the 3D structure, the AnxA11 mutations can be 283 observed clustered at two locations (Figure 5A). Interestingly, three of these mutations involve a 284 buried Arg residue, and a close look at the alignment reveals that these Arg residues are fully 285 conserved in all rat Anxs. The remaining two mutations replace an exposed, non-conserved, small 286 residue with an Arg. Details of the environment of each of the three conserved Arg residues are 287 shown in Figure 5B-D. In brief, two of them coordinate carbonyl groups at C termini of helices, and 288 one is involved in a conserved extended buried network of salt bridges between domains II and IV. 289 These locations suggest that the ALS mutations involving these conserved Arg residues will have 290 effects on AnxA11 folding and/or stability. 291 ∆∆G predictions for the five mutations on the SDM server [64] indicate decreased stability for 292 all mutations except S229R. Hence, the affected residues may be important for correct folding 293 and/or stability of AnxA11. Recent functional characterisation of the AnxA11 ALS mutation R235Q 294 shows that the mutant protein is prone to aggregation and sequesters FL-AnxA11. This in turn 295 abolishes the binding to calcyclin [34]. In addition, two rare missense variants A293V and I307M 296 were recently discovered in the Chinese Han population [33]. These residues are located in domain 297 II (Figure 5). bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

298 299 Figure 5. Location of ALS mutations and variants in the AnxA11 structure. (A) The five ALS 300 mutations (see text) are highlighted in red and the two rare variants in green. The two 3D clusters of 301 mutations/variants are shown with blue dashed circles. One cluster lies in domain II, possibly 302 affecting folding and/or interactions with domain III. The other cluster is at the interface between 303 domains I and IV, i.e. the N and C terminus of the AnxA11 core structure. (B) Arg233 304 (corresponding to the human R235Q mutation) makes 4 contacts to carbonyl groups at the C- 305 terminal end of domain I helices B and E. With its buried positive charge, Arg233 interacts with the 306 negative helix dipole from both helices. (C) Arg344 (corresponding to R346C) similarly caps helix B 307 of domain III. (D) An extended salt bridge network exists between domains II (top) and IV (bottom) 308 (stereo view). Arg300 (corresponds to R302C) is central in this network. The charged residues with 309 numbers indicated are fully conserved in all rat Anxs (Arg, blue; acidic, red).

310 The N-terminal tail of AnxA11 has a unique amino acid composition, reminiscent of 311 disordered proteins able to undergo liquid-liquid phase separation (LLPS) [65]. It has the hallmarks 312 of an extended segment poised for π-π interactions, which are central in LLPS [66]. Indeed, recently 313 the AnxA11 N-terminal region was shown to promote LLPS; its N terminus binds to RNA granules, 314 which do not contain membranes, while the core structure binds to lipids in membranes [8]. 315 Furthermore, ALS-associated mutations in both the N terminus (D40G) and C terminus (R346C) of 316 AnxA11 were shown to promote phase transition from liquid-liquid droplets to more stable gel-like 317 states within AnxA11 droplets and in addition impair the reversal of the states. As a consequence, 318 these mutations disrupt the function of AnxA11 to tether RNA granules with lysosomes [8].

319 3.5. The solution behaviour of the AnxA11 core 320 The thermal stability difference between Δ188AnxA11and Δ192AnxA11 prompted us to study 321 their conformation in solution using SAXS and SRCD. In SAXS, both variants were monomeric with 322 identical distance distributions and globular folds that superposed well with the crystal structure 323 (Figure 6A-E), indicating that the absence of the RGTI stretch does not lead to large-scale 324 conformational changes. Despite the trimeric packing in the crystal, AnxA11 is monomeric in 325 solution under the employed conditions. bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

326 Like most Anxs, AnxA11 has been implicated in Ca2+ binding [1,13]. Structural studies of many 327 Anxs have revealed Ca2+ binding to the convex side of the core structure, essentially through 328 coordination by acidic residues. To test for effect of Ca 2+ on the folding of the core structure of 329 AnxA11, we titrated Δ188AnxA11 and Δ192AnxA11 with CaCl2 in SAXS and followed the radius of 330 gyration (Rg; Figure 6F). The Rg of both truncated proteins remained the same in up to 1 mM CaCl 2. 331 Additionally, we carried out SRCD spectroscopy of Δ188AnxA11 and Δ192AnxA11 in the absence 332 and presence of 1 mM CaCl2 (Figure 6G). The spectral quality was excellent, with the data being 333 truncated at 178 nm. The presence of Ca2+ did not alter the spectra, indicating that Ca2+ does not 334 influence the secondary structure content of the core structure. This result confirms that 335 Δ192AnxA11 is folded despite its dramatic decrease in thermal stability, and that the degree of 336 secondary structure content between the two variants is nearly identical. Ca 2+ induced a large 337 thermal stabilisation of FL-AnxA11, indicated by a rise in Tm of ~14 °C [13]. The presence of up to 75 338 mM Ca2+ only induced small increases in α-helical content [13]. Thus, Ca 2+ binding results in only 339 local conformational changes, which primes AnxA11 for lipid membrane binding.

340 341 Figure 6. SAXS and SRCD analysis of Δ188AnxA11 and Δ192AnxA11. (A) SAXS data for 342 Δ188AnxA11 and Δ192AnxA11 have been offset for clarity. GNOM, GASBOR, and CRYSOL fits are 343 plotted over the measured data to denote the accuracy of distance distribution analysis, ab initio 344 modeling, and theoretical scattering calculated from the crystal structure. (B) Guinier analysis. (C) 345 Distance distribution diagram from GNOM. (D) Kratky plot of the SAXS data. (E) GASBOR ab initio 346 models based on the SAXS data. The crystal structure (black) has been fitted inside. χ 2 values of the 347 GASBOR and CRYSOL fits are indicated. (F) Rg as a function of CaCl2 concentration. The error bars 348 represent the fitting error from Guinier analysis. (G) SRCD spectra of Δ188AnxA11 and 349 Δ192AnxA11 in the presence and absence of 1 mM CaCl2.

350 4. Conclusions bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

351 As the diversity of the vertebrate Anx family is mainly realised through the uniqueness of the 352 N-terminal tail [1,16,67,68], many unique functional features are lost by truncation. However, 353 crucial functional properties of interest, such as mRNA binding, Ca 2+-dependent binding to 354 phospholipids, as well as Ca2+ binding itself, may be retained. Thus, truncated Anxs may serve as a 355 foundation for functional studies, dodging practical issues often encountered due to their long N 356 termini. 357 The crystal structure of AnxA11, although similar to known Anx core domain structures, 358 provides important starting points to further understand the structure-function relationships in 359 AnxA11 and other Anx family members. The identification of a strongly stabilizing Velcro-like 360 bridging segment at the N terminus of the Anx core, as well as the fact that many known ALS 361 mutations correspond to highly conserved buried Arg residues in the Anx fold, provide new 362 insights into Anx folding. AnxA11, like many other ALS targets, is part of a membraneless 363 organelle, a ribonucleoprotein complex, involved in neuronal transport of mRNA granules [8,69– 364 71]. It is evident that this process can be affected by mutations in either the N-terminal region or the 365 folded core domain of AnxA11, which in turn can affect binding to both RNA, lipid membranes, 366 and Ca2+. The fact that the bridging segment is a target for phosphorylation in AnxA2 shows that in 367 addition to simply stabilizing the core structure, this segment may be a conformational regulatory 368 element common to most Anxs. 369 Author Contributions: Conceptualization, A.V. and P.K.; methodology, P.A.G.L., A.R., and A.V.; validation, 370 A.R. and P.K.; formal analysis, A.R. and P.K.; investigation, P.A.G.L., A.R., S.J.H., , S.P., T.R., H.H., S.A.S., 371 P.D.S., A.V., and P.K.; resources, P.A.G.L., S.J.H., S.P., T.R., H.H., S.A.S., P.D.S., and A.V.; data curation, A.R. 372 and P.K.; writing—original draft preparation, P.A.G.L., A.R., and A.V.; writing—review and editing, A.R., 373 A.V., and P.K.; visualization, A.R. and P.K.; supervision, A.V. and P.K.; project administration, A.V. and P.K.; 374 funding acquisition, A.V. and P.K. All authors have read and agreed to the published version of the 375 manuscript. 376 Funding: This research was funded by the University of Bergen. Synchrotron visits were supported by the EU 377 Framework Programme for Research and Innovation HORIZON 2020 iNEXT and CALIPSOplus programmes, 378 grant numbers 653706 and 730872. 379 Acknowledgments: We thank Marianne Goris for valuable assistance in the laboratory. Crystallographic and 380 SAXS data were collected on EMBL Hamburg beamlines P13 and P12, respectively, at the PETRA III storage 381 ring (DESY, Hamburg, Germany). We also thank the SRCD beamline support at the ASTRID2 synchrotron 382 facility (Aarhus, Denmark). We thank the BiSS facility at the University of Bergen for providing vital 383 instrumentation for this study. 384 Conflicts of Interest: The authors declare no conflict of interest.

385 References 386 1. Gerke V & Moss SE (2002) Annexins: from structure to function. Physiol Rev 82, 331–371. 387 DOI:10.1152/physrev.00030.2001 388 2. Gerke V, Creutz CE & Moss SE (2005) Annexins: linking Ca2+ signalling to membrane dynamics. Nat Rev 389 Mol Cell Biol 6, 449–461. DOI:10.1038/nrm1661 390 3. Moss SE & Morgan RO (2004) The annexins. Genome Biol 5, 219. DOI:10.1186/gb-2004-5-4-219 391 4. Liemann S & Huber R (1997) Three-dimensional structure of annexins. Cell Mol Life Sci 53, 516–521. 392 DOI:10.1007/s000180050065 393 5. Morgan RO, Jenkins NA, Gilbert DJ, Copeland NG, Balsara BR, Testa JR & Fernandez MP (1999) Novel 394 human and mouse annexin A10 are linked to the genome duplications during early chordate evolution. 395 Genomics 60, 40–49. DOI:10.1006/geno.1999.5895 396 6. Rosengarth A & Luecke H (2003) A calcium-driven conformational switch of the N-terminal and core 397 domains of annexin A1. J Mol Biol 326, 1317–1325. DOI:10.1016/s0022-2836(03)00027-5 398 7. Boye TL, Jeppesen JC, Maeda K, Pezeshkian W, Solovyeva V, Nylandsted J & Simonsen AC (2018) 399 Annexins induce curvature on free-edge membranes displaying distinct morphologies. Sci Rep 8, 10309. 400 DOI:10.1038/s41598-018-28481-z bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

401 8. Liao YC, Fernandopulle MS, Wang G, Choi H, Hao L, Drerup CM, Patel R, Qamar S, Nixon-Abell J, Shen 402 Y, Meadows W, Vendruscolo M, Knowles TPJ, Nelson M, Czekalska MA, Musteikyte G, Gachechiladze 403 MA, Stephens CA, Pasolli HA, Forrest LR, St George-Hyslop P, Lippincott-Schwartz J & Ward ME (2019) 404 RNA Granules Hitchhike on Lysosomes for Long-Distance Transport, Using Annexin A11 as a Molecular 405 Tether. Cell 179, 147–164.e20. DOI:10.1016/j.cell.2019.08.050 406 9. Tokumitsu H, Mizutani A, Muramatsu M, Yokota T, Arai K & Hidaka H (1992) Molecular cloning of 407 rabbit CAP-50, a calcyclin-associated annexin protein. Biochem Biophys Res Commun 186, 1227–1235. 408 DOI:10.1016/s0006-291x(05)81537-2 409 10. Burns AL, Magendzo K, Shirvan A, Srivastava M, Rojas E, Alijani MR & Pollard HB (1989) Calcium 410 channel activity of purified human synexin and structure of the human synexin gene. Proc Natl Acad Sci U 411 S A 86, 3798–3802. DOI:10.1073/pnas.86.10.3798 412 11. Creutz CE, Snyder SL, Husted LD, Beggerly LK & Fox JW (1988) Pattern of repeating aromatic residues in 413 synexin. Similarity to the cytoplasmic domain of synaptophysin. Biochem Biophys Res Commun 152, 1298– 414 1303. DOI:10.1016/s0006-291x(88)80426-1 415 12. Creutz CE (2009) Novel protein ligands of the annexin A7 N-terminal region suggest pro-beta helices 416 engage one another with high specificity. Gen Physiol Biophys 28, F7–F13. 417 13. Lecona E, Turnay J, Olmo N, Guzman-Aranguez A, Morgan RO, Fernandez MP & Lizarbe MA (2003) 418 Structural and functional characterization of recombinant mouse annexin A11: influence of calcium 419 binding. Biochem J 373, 437–449. DOI:10.1042/bj20021721 420 14. Satoh H, Shibata H, Nakano Y, Kitaura Y & Maki M (2002) ALG-2 interacts with the amino-terminal 421 domain of annexin XI in a Ca(2+)-dependent manner. Biochem Biophys Res Commun 291, 1166–1172. 422 DOI:10.1006/bbrc.2002.6600 423 15. Mizutani A, Tokumitsu H, Kobayashi R & Hidaka H (1993) Phosphorylation of annexin XI (CAP-50) in 424 SR-3Y1 cells. J Biol Chem 268, 15517–15522. 425 16. Sudo T & Hidaka H (1998) Regulation of calcyclin (S100A6) binding by alternative splicing in the N- 426 terminal regulatory domain of annexin XI isoforms. J Biol Chem 273, 6351–6357. 427 DOI:10.1074/jbc.273.11.6351 428 17. Tokumitsu H, Mizutani A & Hidaka H (1993) Calcyclin-binding site located on the NH2-terminal domain 429 of rabbit CAP-50 (annexin XI): functional expression of CAP-50 in Escherichia coli. Arch Biochem Biophys 430 303, 302–306. DOI:10.1006/abbi.1993.1287 431 18. Tomas A & Moss SE (2003) Calcium- and cell cycle-dependent association of annexin 11 with the nuclear 432 envelope. J Biol Chem 278, 20210–20216. DOI:10.1074/jbc.M212669200 433 19. Naka M, Qing ZX, Sasaki T, Kise H, Tawara I, Hamaguchi S & Tanaka T (1994) Purification and 434 characterization of a novel calcium-binding protein, S100C, from porcine heart. Biochim Biophys Acta 1223, 435 348–353. DOI:10.1016/0167-4889(94)90094-9 436 20. Powell MA & Glenney JR (1987) Regulation of calpactin I phospholipid binding by calpactin I light-chain 437 binding and phosphorylation by p60v-src. Biochem J 247, 321–328. DOI:10.1042/bj2470321 438 21. Mizutani A, Usuda N, Tokumitsu H, Minami H, Yasui K, Kobayashi R & Hidaka H (1992) CAP-50, a 439 newly identified annexin, localizes in nuclei of cultured fibroblast 3Y1 cells. J Biol Chem 267, 13498–13504. 440 22. Mizutani A, Watanabe N, Kitao T, Tokumitsu H & Hidaka H (1995) The long amino-terminal tail domain 441 of annexin XI is necessary for its nuclear localization. Arch Biochem Biophys 318, 157–165. 442 DOI:10.1006/abbi.1995.1216 443 23. Tomas A, Futter C & Moss SE (2004) Annexin 11 is required for midbody formation and completion of 444 the terminal phase of cytokinesis. J Cell Biol 165, 813–822. DOI:10.1083/jcb.200311054 445 24. Duncan R, Carpenter B, Main LC, Telfer C & Murray GI (2008) Characterisation and protein expression 446 profiling of annexins in colorectal cancer. Br J Cancer 98, 426–433. DOI:10.1038/sj.bjc.6604128 447 25. Fernandez-Madrid F, Tang N, Alansari H, Granda JL, Tait L, Amirikia KC, Moroianu M, Wang X & 448 Karvonen RL (2004) Autoantibodies to Annexin XI-A and Other Autoantigens in the Diagnosis of Breast 449 Cancer. Cancer Res 64, 5089–5096. DOI:10.1158/0008-5472.Can-03-0932 450 26. Hua K, Li Y, Zhao Q, Fan L, Tan B & Gu J (2018) Downregulation of Annexin A11 (ANXA11) Inhibits Cell 451 Proliferation, Invasion, and Migration via the AKT/GSK-3beta Pathway in Gastric Cancer. Med Sci Monit 452 24, 149–160. DOI:10.12659/msm.905372 453 27. Song J, Shih Ie M, Chan DW & Zhang Z (2009) Suppression of annexin A11 in ovarian cancer: 454 implications in chemoresistance. Neoplasia 11, 605–14, 1 p following 614. DOI:10.1593/neo.09286 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

455 28. Wang J, Guo C, Liu S, Qi H, Yin Y, Liang R, Sun MZ & Greenaway FT (2014) Annexin A11 in disease. Clin 456 Chim Acta 431, 164–168. DOI:10.1016/j.cca.2014.01.031 457 29. Xin W, Rhodes DR, Ingold C, Chinnaiyan AM & Rubin MA (2003) Dysregulation of the annexin family 458 protein family is associated with prostate cancer progression. Am J Pathol 162, 255–261. 459 DOI:10.1016/s0002-9440(10)63816-3 460 30. Brenner D & Weishaupt JH (2019) Update on amyotrophic lateral sclerosis genetics. Curr Opin Neurol 32, 461 735–739. DOI:10.1097/WCO.0000000000000737 462 31. Fernandopulle M, Wang G, Nixon-Abell J, Qamar S, Balaji V, Morihara R & St George-Hyslop PH (2019) 463 Inherited and Sporadic Amyotrophic Lateral Sclerosis and Fronto-Temporal Lobar Degenerations arising 464 from Pathological Condensates of Phase Separating Proteins. Hum Mol Genet 28, R187–R196. 465 DOI:10.1093/hmg/ddz162 466 32. Fratta P, Birsa N, Tosolini AP & Schiavo G (2020) Travelling Together: A Unifying Pathomechanism 467 for ALS. Trends Neurosci 43, 1–2. DOI:10.1016/j.tins.2019.10.006 468 33. Liu X, Wu C, He J, Zhang N & Fan D (2019) Two rare variants of the ANXA11 gene identified in Chinese 469 patients with amyotrophic lateral sclerosis. Neurobiol Aging 74, 235.e9–235.e12. 470 DOI:10.1016/j.neurobiolaging.2018.09.020 471 34. Smith BN, Topp SD, Fallini C, Shibata H, Chen HJ, Troakes C, King A, Ticozzi N, Kenna KP, Soragia- 472 Gkazi A, Miller JW, Sato A, Dias DM, Jeon M, Vance C, Wong CH, de Majo M, Kattuah W, Mitchell JC, 473 Scotter EL, Parkin NW, Sapp PC, Nolan M, Nestor PJ, Simpson M, Weale M, Lek M, Baas F, Vianney de 474 Jong JM, Ten Asbroek A, Redondo AG, Esteban-Perez J, Tiloca C, Verde F, Duga S, Leigh N, Pall H, 475 Morrison KE, Al-Chalabi A, Shaw PJ, Kirby J, Turner MR, Talbot K, Hardiman O, Glass JD, De Belleroche 476 J, Maki M, Moss SE, Miller C, Gellera C, Ratti A, Al-Sarraj S, Brown RH, Jr., Silani V, Landers JE & Shaw 477 CE (2017) Mutations in the vesicular trafficking protein annexin A11 are associated with amyotrophic 478 lateral sclerosis. Sci Transl Med 9, DOI:10.1126/scitranslmed.aad9157 479 35. Tsai PC, Liao YC, Jih KY, Soong BW, Lin KP & Lee YC (2018) Genetic analysis of ANXA11 variants in a 480 Han Chinese cohort with amyotrophic lateral sclerosis in Taiwan. Neurobiol Aging 72, 188.e1–188.e2. 481 DOI:10.1016/j.neurobiolaging.2018.07.002 482 36. Zhang K, Liu Q, Liu K, Shen D, Tai H, Shu S, Ding Q, Fu H, Liu S, Wang Z, Li X, Liu M, Zhang X & Cui L 483 (2018) ANXA11 mutations prevail in Chinese ALS patients with and without cognitive dementia. Neurol 484 Genet 4, e237. DOI:10.1212/nxg.0000000000000237 485 37. Chomczynski P & Sacchi N (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate- 486 phenol-chloroform extraction. Anal Biochem 162, 156–159. DOI:10.1006/abio.1987.9999 487 38. Miles AJ & Wallace BA (2018) CDtoolX, a downloadable software package for processing and analyses of 488 circular dichroism spectroscopic data. Protein Sci 27, 1717–1722. DOI:10.1002/pro.3474 489 39. Blanchet CE, Spilotros A, Schwemmer F, Graewert MA, Kikhney A, Jeffries CM, Franke D, Mark D, 490 Zengerle R, Cipriani F, Fiedler S, Roessle M & Svergun DI (2015) Versatile sample environments and 491 automation for biological solution X-ray scattering experiments at the P12 beamline (PETRA III, DESY). J 492 Appl Crystallogr 48, 431–443. DOI:10.1107/S160057671500254X 493 40. Franke D, Petoukhov MV, Konarev PV, Panjkovich A, Tuukkanen A, Mertens HDT, Kikhney AG, 494 Hajizadeh NR, Franklin JM, Jeffries CM & Svergun DI (2017) ATSAS 2.8: a comprehensive data analysis 495 suite for small-angle scattering from macromolecular solutions. J Appl Crystallogr 50, 1212–1225. 496 DOI:10.1107/s1600576717007786 497 41. Svergun DI (1992) Determination of the regularization parameter in indirect-transform methods using 498 perceptual criteria. Journal of Applied Crystallography 25, 495–503. DOI:10.1107/s0021889892001663 499 42. Svergun, D., Barberato, C. & Koch, M. H. J. (1995) CRYSOL - a Program to Evaluate X-ray Solution 500 Scattering of Biological Macromolecules from Atomic Coordinates. Journal of Applied Crystallography 28, 501 768–773. DOI:doi:10.1107/S0021889895007047 502 43. Svergun DI, Petoukhov MV & Koch MH (2001) Determination of domain structure of proteins from X-ray 503 solution scattering. Biophys J 80, 2946–2953. DOI:10.1016/s0006-3495(01)76260-1 504 44. Kozin MB & Svergun DI (2001) Automated matching of high-and low-resolution structural models. 505 Journal of Applied Crystallography 34, 33–41. 506 45. Cianci M, Bourenkov G, Pompidor G, Karpics I, Kallio J, Bento I, Roessle M, Cipriani F, Fiedler S & 507 Schneider TR (2017) P13, the EMBL macromolecular crystallography beamline at the low-emittance bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

508 PETRA III ring for high- and low-energy phasing with variable beam focusing. J Synchrotron Radiat 24, 509 323–332. DOI:10.1107/S1600577516016465 510 46. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66, 125–132. DOI:10.1107/s0907444909047337 511 47. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC & Read RJ (2007) Phaser 512 crystallographic software. J Appl Crystallogr 40, 658–674. DOI:10.1107/s0021889807021206 513 48. Sutton RB & Sprang SR (1996) Three dimensional structure of annexin IV. In Annexins: Molecular Structure 514 to Cellular Function (Seaton BA, ed), Landes Biosciences, Georgetown, TX, USA. 515 49. Afonine PV, Grosse-Kunstleve RW, Echols N, Headd JJ, Moriarty NW, Mustyakimov M, Terwilliger TC, 516 Urzhumtsev A, Zwart PH & Adams PD (2012) Towards automated crystallographic structure refinement 517 with phenix.refine. Acta Crystallogr D Biol Crystallogr 68, 352–367. DOI:10.1107/s0907444912001308 518 50. Emsley P, Lohkamp B, Scott WG & Cowtan K (2010) Features and development of Coot. Acta Crystallogr 519 D Biol Crystallogr 66, 486–501. DOI:10.1107/s0907444910007493 520 51. Davis IW, Murray LW, Richardson JS & Richardson DC (2004) MOLPROBITY: structure validation and 521 all-atom contact analysis for nucleic acids and their complexes. Nucleic Acids Res 32, W615–9. 522 DOI:10.1093/nar/gkh398 523 52. DeLano WL (2002) Pymol: An open-source molecular graphics tool. CCP4 Newsletter on protein 524 crystallography 40, 82–92. 525 53. Unni S, Huang Y, Hanson RM, Tobias M, Krishnan S, Li WW, Nielsen JE & Baker NA (2011) Web servers 526 and services for electrostatics calculations with APBS and PDB2PQR. J Comput Chem 32, 1488–1491. 527 DOI:10.1002/jcc.21720 528 54. Bances P, Fernandez MR, Rodriguez-Garcia MI, Morgan RO & Fernandez MP (2000) Annexin A11 529 (ANXA11) gene structure as the progenitor of paralogous annexins and source of orthologous cDNA 530 isoforms. Genomics 69, 95–103. DOI:10.1006/geno.2000.6309 531 55. Iglesias JM, Morgan RO, Jenkins NA, Copeland NG, Gilbert DJ & Fernandez MP (2002) Comparative 532 genetics and evolution of annexin A13 as the founder gene of vertebrate annexins. Mol Biol Evol 19, 608– 533 618. DOI:10.1093/oxfordjournals.molbev.a004120 534 56. Fernandez MP, Jenkins NA, Gilbert DJ, Copeland NG & Morgan RO (1996) Sequence and Chromosomal 535 Localization of Mouse Annexin XI. Genomics 37, 366–374. DOI:https://doi.org/10.1006/geno.1996.0571 536 57. Ecsédi P, Kiss B, Gógl G, Radnai L, Buday L, Koprivanacz K, Liliom K, Leveles I, Vértessy B, Jeszenői N, 537 Hetényi C, Schlosser G, Katona G & Nyitray L (2017) Regulation of the Equilibrium between Closed and 538 Open Conformations of Annexin A2 by N-Terminal Phosphorylation and S100A4-Binding. Structure 25, 539 1195–1207.e5. DOI:10.1016/j.str.2017.06.001 540 58. Chen Z, Mao Y, Yang J, Zhang T, Zhao L, Yu K, Zheng M, Jiang H & Yang H (2014) Characterizing the 541 binding of annexin V to a lipid bilayer using molecular dynamics simulations. Proteins 82, 312–322. 542 DOI:10.1002/prot.24389 543 59. Hong S, Na S, Kim OH, Jeong S, Oh BC & Ha NC (2020) High-resolution structures of annexin A5 in a 544 two-dimensional array. J Struct Biol 209, 107401. DOI:10.1016/j.jsb.2019.10.003 545 60. Lin YC, Chipot C & Scheuring S (2020) Annexin-V stabilizes membrane defects by inducing lipid phase 546 transition. Nat Commun 11, 230. DOI:10.1038/s41467-019-14045-w 547 61. Mo Y, Campos B, Mealy TR, Commodore L, Head JF, Dedman JR & Seaton BA (2003) Interfacial basic 548 cluster in annexin V couples phospholipid binding and trimer formation on membrane surfaces. J Biol 549 Chem 278, 2437–2443. DOI:10.1074/jbc.M210286200 550 62. Aukrust I, Rosenberg LA, Ankerud MM, Bertelsen V, Hollas H, Saraste J, Grindheim AK & Vedeler A 551 (2017) Post-translational modifications of Annexin A2 are linked to its association with perinuclear 552 nonpolysomal mRNP complexes. FEBS Open Bio 7, 160–173. DOI:10.1002/2211-5463.12173 553 63. Solbak SMØ, Abdurakhmanov E, Vedeler A & Danielson UH (2017) Characterization of interactions 554 between hepatitis C virus NS5B polymerase, annexin A2 and RNA - effects on NS5B catalysis and 555 allosteric inhibition. Virol J 14, 236. DOI:10.1186/s12985-017-0904-4 556 64. Pandurangan AP, Ochoa-Montaño B, Ascher DB & Blundell TL (2017) SDM: a server for predicting 557 effects of mutations on protein stability. Nucleic Acids Res 45, W229–W235. DOI:10.1093/nar/gkx439 558 65. Gomes E & Shorter J (2019) The molecular language of membraneless organelles. J Biol Chem 294, 7115– 559 7127. DOI:10.1074/jbc.TM118.001192 560 66. Vernon RM, Chong PA, Tsang B, Kim TH, Bah A, Farber P, Lin H & Forman-Kay JD (2018) Pi-Pi contacts 561 are an overlooked protein feature relevant to phase separation. Elife 7, DOI:10.7554/eLife.31486 bioRxiv preprint doi: https://doi.org/10.1101/2020.03.27.011445; this version posted March 27, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

562 67. Misaki Y, Van Venrooij WJ & Pruijn GJ (1995) Prevalence and characteristics of anti-56K/annexin XI 563 autoantibodies in systemic autoimmune diseases. J Rheumatol 22, 97–102. 564 68. Sjolin C, Movitz C, Lundqvist H & Dahlgren C (1997) Translocation of annexin XI to neutrophil 565 subcellular organelles. Biochim Biophys Acta 1326, 149–156. DOI:10.1016/s0005-2736(97)00018-7 566 69. Li HR, Chiang WC, Chou PC, Wang WJ & Huang JR (2018) TAR DNA-binding protein 43 (TDP-43) 567 liquid-liquid phase separation is mediated by just a few aromatic residues. J Biol Chem 293, 6090–6098. 568 DOI:10.1074/jbc.AC117.001037 569 70. Nguyen HP, Van Broeckhoven C & van der Zee J (2018) ALS Genes in the Genomic Era and their 570 Implications for FTD. Trends Genet 34, 404–423. DOI:10.1016/j.tig.2018.03.001 571 71. Wei Q, Chen X, Chen Y, Ou R, Cao B, Hou Y, Zhang L & Shang HF (2019) Unique characteristics of the 572 genetics epidemiology of amyotrophic lateral sclerosis in China. Sci China Life Sci 62, 517–525. 573 DOI:10.1007/s11427-018-9453-x 574 575