ALS Variants of Annexin A11’s Proline-Rich Domain Impair Its S100A6-Mediated Fibril Dissolution

Mutations in the proline-rich domain (PRD) of annexin A11 are linked to amyotrophic lateral sclerosis (ALS), a fatal neurodegenerative disease, and generate abundant neuronal A11 inclusions by an unknown mechanism. Here, we demonstrate that recombinant A11-PRD and its ALS-associated variants form liquidlike condensates that transform into β-sheet–rich amyloid fibrils. Surprisingly, these fibrils dissolved in the presence of S100A6, an A11 binding partner overexpressed in ALS. The ALS variants of A11-PRD showed longer fibrillization half-times and slower dissolution, even though their binding affinities for S100A6 were not significantly affected. These findings indicate a slower fibril-to-monomer exchange for these ALS variants, resulting in a decreased level of S100A6-mediated fibril dissolution. These ALS-A11 variants are thus more likely to remain aggregated despite their slower fibrillization.

A nnexins, calcium-dependent phospholipid-binding proteins, harbor multiple copies of a conserved core called an annexin repeat and a variable head domain. 1 Among the 12 human annexins, annexin A11 comprises the largest head domain (196 residues; Figure 1A). This proline-rich domain (PRD, ∼30% prolines; Figure 1B) orchestrates A11's phase separation, which is critical for its role in RNA granule transport. 2 A11-PRD binds to S100A6 (calcyclin), a dimeric calcium-binding protein, and these interactions regulate the cell cycle. 3−5 Four missense point mutations in A11-PRD, namely, G38R, D40G, G175R, and G189E, are linked to amyotrophic lateral sclerosis (ALS), a fatal neurodegenerative disease. 6 Despite its importance, very little is known about A11-PRD. Here, we investigated its aggregation properties and interactions with S100A6. Our results elucidate how ALSassociated variants may promote neuronal A11 inclusions observed in ALS patients.
To obtain residue-specific details about the above-described weak-affinity site, we performed nuclear magnetic resonance (NMR) backbone assignments of S100A6 and A11 2−52 . Unlike A11 2−52 , amyloidogenic A11-PRD constructs are refractory to these measurements due to their fibrillization, and therefore, the corresponding assessment of the high-affinity site is not plausible. NMR chemical shift analysis 11 revealed that A11 2−52 was disordered in solution (Table S3). NMR titration experiments found significant attenuation in the 13 C'-15 N cross-peaks of residues 41 NVATYAGQ 48 of A11 2−52 upon the addition of S100A6 ( Figures 2C and S5), establishing that residues 41−48 of A11-PRD represent the weak-affinity binding site for S100A6. Significant attenuation in multiple 1 H− 15 N cross-peaks of the S100A6 dimer was observed in the presence of A11 2−52 ( Figures 2C and S6). Figure 2D maps these perturbations onto the S100A6 structure, 12 which showed that the affected residues lie in and around the dimerization interface. The Annexin A2−S100A4 complex (other members of the Annexin-and S100-families) is shown in Figure 2D, 13 where the A2 head domain binds to both subunits of the S100A4 dimer. Given the similarities between this complex and our results, we conclude that A11-PRD likely binds to the S100A6 dimer asymmetrically.
The impact of S100A6 on condensates of A11 2−196 was monitored by fluorescence microscopy and turbidity assays ( Figure 2E). S100A6 readily colocalized with A11 2−196 condensates, and its increasing concentrations led to their progressive decrease, establishing S100A6-mediated dissolution of these condensates. ALS variants showed similar results, although A11 2−196 D40G showed a smaller decrease in turbidity at lower S100A6 concentrations. To determine the impact of Figure 2. Impact of S100A6 on A11-PRD aggregation. (A) Sedimentation profiles of S100A6 (green) and S100A6 + A11 2−196 (same numbering scheme as Figure 1C) mixture, black. (B) SPR analysis of A11 2−196 − S100A6 interactions; n = 3, data (circles), fit (solid line). The reduction in cross-peak heights of (C, upper) 200 μM 13 C/ 15 N-labeled A11 2−52 (blue) with 25 μM S100A6 dimer and (C, lower) 100 μM 15 N-labeled S100A6 dimer with 50 μM A11 2−52 (green). X-ray structures of (D, left) S100A6; 12 affected residues in panel (C) are shown as green spheres, and (D, right) S100A4−Annexin A2 complex; 13 residues 33−339 of A2 are not shown for clarity. (E) Dissolution of ATTO488-labeled condensates of constructs 1−5 with ATTO647-labeled S100A6 monitored using turbidity assays and fluorescence microscopy. (F) S100A6's impact on fibrils of constructs 1−5. Each construct (50 μM, n = 3) was incubated for ∼14 h, upon which 100 μM S100A6 dimer was added; TEM images of samples at 50 h are shown (right). Additional three replicates of each construct received buffer at ∼14 h; only A11 2−196 after the addition of buffer is shown (purple). Additional controls included each construct (50 μM, n = 3) with 100 μM S100A6 dimer at 0 h; only A11 2−196 is shown (olive). (G) Model describing the findings of this study. Table 1. Quantitative Analysis of A11-PRD−S100A6 Affinities were measured using SPR (n ≥ 2), unless noted otherwise, and refer to an S100A6 dimer:A11-PRD monomer stoichiometry uncovered by sedimentation. b Measured using ITC (n = 3, Figure  S4). c No binding was detected via sedimentation, SPR, and NMR. S100A6 on A11-PRD fibrils, 2 mol equiv of S100A6 dimer were added to A11 2−196 fibrils at a stationary phase. A significant loss of ThT signal was observed with t 1/2 : ∼3.5 h ( Figure 2F). TEM images showed a complete lack of fibrils, establishing that the A11-PRD fibrillization can be reversed by S100A6. Although a similar decrease in ThT signals was noted for ALS variants, the corresponding t 1/2 were longer (4−13 h). TEM images of A11 2−196 G189E showed no aggregates, while amorphous (nonfibrillar) aggregates were seen in the case of A11 2−196 G38R . For the A11 2−196 D40G variant, ThT signals did not return to baseline, indicating incomplete fibril dissolution, which was confirmed by TEM. Residual fibrillar aggregates were also observed for A11 2−196 G175R , which underwent the slowest dissolution (t 1/2 ∼ 13 h). Figure 2G depicts a model that is consistent with these results. A11-PRD and ALS variants have similar binding affinities and, thus, likely similar exchange kinetics for S100A6. The latter can sequester monomeric A11-PRD in a soluble form, inhibiting its fibrillization. Because of longer half-times but similar growth phases, A11-PRD and its ALS variants may have equivalent forward rate constants for monomer-to-fibril transitions. However, the corresponding fibril-to-monomer rate constant is likely much faster in A11-PRD than its ALS variants, increasing the amount of available monomer, which can be siphoned off by S100A6, leading to fibril dissolution. In contrast, ALS variants would produce fewer free monomers from fibrils over time, resulting in slower S100A6-mediated fibril dissolution and possibly leading to abundant neuronal A11 inclusions. Given that S100-family proteins, including S100A6, can also disassemble nonmuscle myosin-2 filaments, 14 depolymerization is likely one of their many functions, providing a plausible explanation as to why S100A6 is overexpressed in ALS. Finally, we note that PRDs are ubiquitous in eukaryotes and often form dynamic signaling networks due to their disordered conformations and favorable binding properties. 15 The results presented in this study demonstrate that some PRDs can also phase separate and form labile amyloid fibrils that dissolve upon interaction with partner proteins. Thus, while the structural basis of how a PRD forms β-sheet−rich fibrils is not known and is a topic of ongoing research in our laboratory, we argue that A11-PRD may represent a new subclass of PRDs that are prone to reversible aggregation. ■ METHODS Protein Expression and Purification. A11-PRD and S100A6 constructs were custom synthesized (Azenta Life Sciences) and expressed at 16°C. Tobacco etch virus (TEV) protease construct, a generous gift from David Waugh (NIH), was expressed at 37°C. For A11-PRD and S100A6, cells were grown at 37°C in 1 L Luria− Bertani or minimal M9 15−17 medium. The latter was used for NMR isotopic labeling. TEV purification was described previously. 17 A11-PRD constructs, except A11 89−196 , were purified as follows: Cells, resuspended in a lysis buffer comprising 50 mM Tris, pH 8, and 6 M guanidine hydrochloride (GdmCl), were lysed by heat shock (80°C for 5 min) and cleared by centrifugation. The supernatant was filtered (Stericup, Sigma-Aldrich) before being loaded onto a HisTrap column (Cytiva) pre-equilibrated with lysis buffer. Bound protein was washed with a buffer comprising 50 mM Tris, pH 8.0, and 250 mM NaCl and eluted in the same buffer containing 500 mM imidazole. The eluted protein was loaded onto a XK 16/20 column (Cytiva) prepacked with Strep-Tactin Sepharose resin (Cytiva) preequilibrated with 50 mM Tris, pH 8.0, 250 mM NaCl, 1 mM DTT, and 1 mM EDTA, and eluted in the same buffer containing 2.5 mM ddesthiobiotin. The protein was then mixed with TEV to hydrolyze the N-terminal B1 domain of protein G (GB1) 18 fusion tag. The reaction was carried out at room temperature (∼20 h) and produced a poorly soluble hydrolyzed product. The latter was solubilized by using 6 M GdmCl and passed through a HisTrap column. Relevant flow-through fractions were purified using reverse-phase HPLC (Jupiter 10 μm C18 300 Å column) with a 25−37% acetonitrile gradient comprising 0.1% trifluoroacetic acid (TFA). The eluted protein was lyophilized and stored at −80°C. For A11 89−196 , similar lysis and HisTrap purification steps were performed. The eluted protein was then purified using a HiLoad 26/600 Superdex 75 prep-grade column (Cytiva) preequilibrated with 50 mM Tris, pH 8.0, 250 mM NaCl, and 1 mM EDTA. Relevant fractions were incubated with TEV to cleave off the GB1 tag. The hydrolyzed product was passed through a HisTrap column and further purified by reversed-phase HPLC. Eluted fractions were lyophilized and stored at −80°C.
Phase Separation and Turbidity Measurements. Lyophilized A11-PRD constructs were reconstituted in DMSO, followed by immediate dilution in a buffer comprising 25 mM HEPES, pH 7.0, and 5 mM CaCl 2 (final DMSO concentration ∼ 2% v/v). For A11 2−196 + S100A6 mixtures, A11 2−196 condensates were made, followed by the addition of different concentrations of the S100A6 dimer. These mixtures were analyzed after an ∼5 min incubation period. Turbidity was recorded at optical densities (OD) of 330 and 600 nm using an Agilent Cary 50 Bio UV−vis spectrophotometer. The values reported in Figure 2E are relative to those of the corresponding A11-PRD samples without S100A6.
Fluorophore Labeling. A11-PRD constructs were mixed with a 4 mol equivalent of ATTO-488 NHS ester in 25 mM HEPES, pH 8, and 20% v/v DMSO. The reaction was performed at room temperature (30 min) before being quenched by dilution into a buffer comprising 50 mM Tris, pH 8, and 6 M GdmCl. Constructs were further purified using reversed-phase HPLC, mixed with corresponding unconjugated proteins (concentration of fluorophorelabeled protein = 5 mol %), lyophilized, and stored at −80°C. S100A6 was mixed with 4 mol equiv of ATTO-647N maleimide in 25 mM HEPES, pH 7, and 5 mM CaCl 2 . The reaction was performed at room temperature (30 min) and quenched by β-mercaptoethanol (BME). Excess dye and BME were removed by dialysis. The fluorophore-labeled S100A6 was mixed with unconjugated protein (concentration of fluorophore-labeled protein = 5 mol %) and stored at −80°C.
Microscopy Imaging and FRAP Assays. For microscopy experiments, slides were passivated using PEG-silane. 19,20 Differential interference contrast (DIC) imaging was performed on a Nikon Ti2 widefield microscope equipped with a DS-Qi2 CMOS camera and 100×/1.49NA oil DIC N2 Objective as described previously. 21 Condensates of A11-PRD constructs were excited by a 488 nm laser controlled by a Lumencor SpectraX instrument for imaging of ATTO-488. Additionally, a 640 nm laser was used to image the ATTO-647Nlabeled S100A6. For the image shown in Figure 1I, droplets of 50 μM A11 2−196 with 20 μM ThT were incubated at 37°C for 14 h, and image was taken using a 488 nm laser. FRAP measurements of ATTO-488-labeled A11-PRD constructs were performed on a Nikon point scanning confocal C2 with 2 GaAsP PMTs using a Plan Apo λ 100×/1.45 NA oil objective using our previously described procedure. 21 CR Assay, PK Digestion, and TEM. CR assay, PK digestion, and TEM measurements were carried out using our previously described protocols. 16,22 X-ray Diffraction. A11 2−196 (50 μM) was incubated for 3 days (37°C ) and subsequent fibrils were pelleted using Optima XE Ultracentrifuge (Beckman Coulter). A small amount of the sample was dried and loaded onto a Cryoloop. The sample was mounted on a Bruker Microstar 592 diffractometer equipped with an APEX II CCD detector and Cu Kα radiation (λ = 1.54178 Å). X-ray data was collected using a 360°scan with an exposure time of 300 s.
Fibril Formation and Dissolution Kinetics. Measurements were performed at 37°C under nonagitated conditions using a microplate reader (Infinite M Plex; Tecan) and sealed 96-well flat bottom plates containing 100 μL of sample per well. 50 μM samples of individual A11-PRD constructs were prepared as described above. To ensure reproducible results, the concentrations of stock solutions in DMSO and those of the final samples were checked by UV absorbance at 280 nm. Fibrillization was monitored using ThT (20 μM) fluorescence collected every 5 min (excitation and emission wavelengths: 415 and 480 nm, respectively). Because the aggregation was dependent on the time required to plate different samples, we used two batches to determine rates. Batch  . Additionally, negative controls of three replicates received only buffer (sans S100A6). S100A6-dissolution of fibrils was assessed using the drop in ThT signal. Samples of negative controls did not show any noticeable decrease in ThT signal.
Multicycle kinetic experiments were performed with S100A6 dimer as the binding analyte. Concentrations ranged from 700 pM to 5 μM (in monomeric subunits), with association times of 600−720 s and dissociation times of 60−180 s at a flow rate of 20 μL/min. The immobilization surface was regenerated with 6 M GdmCl for 30 s at a flow rate of 30 μL/min. Each experiment was performed in triplicate. However, due to significant aggregation observed with A11 2−196 G175R and A11 2−196 G189E , only two replicates of these two constructs were analyzed. The following equation was used to determine K D .
where R eq is the steady state binding level and C is the analyte concentration. R max (analyte binding capacity of the surface, in response units), offset (residual response at zero concentration, in response units), and K D (in M) are fitted parameters.
The initial experiments were performed with S100A6 concentrations ranging from 10 pM to 10 μM (in monomeric subunits). But we were unable to characterize the low-affinity interaction between A11 2−52 and S100A6, and no binding was detected for A11 89−196 .
Sedimentation Velocity Analytical Ultracentrifugation. Sedimentation experiments were carried out at 50,000 rpm and 20°C on a Beckman Coulter ProteomeLab XL-I analytical ultra-centrifuge and an An-50-Ti rotor following standard protocols. 23 Stock solutions of individual constructs were diluted to ∼20 and ∼50 μM. Mixtures were prepared by mixing the components at appropriate volumes. Samples were spun at 12,000g for 2 min at room temperature to remove any precipitated proteins and then loaded into 12 mm two-channel centerpiece cells. Absorbance sedimentation data were collected at 280 nm and analyzed using our published protocols. 15,17,24−26 NMR. NMR samples were prepared in a buffer comprising 25 mM HEPES, pH 7.0, 5 mM CaCl 2 , 1 mM TCEP, and 7% (v/v) D 2 O. Experiments were carried out at 30°C on Bruker 600 and 800 MHz spectrometers equipped with z-gradient triple resonance cryoprobes. Spectra were processed using NMRPipe 27 and analyzed using CCPN. 28 Backbone resonance assignments of A11 2−52 and S100A6 were carried out using TROSY-based triple resonance experiments and (HACA)N(CA)CON experiment. 29 The latter was used for A11 2−52 due to its high-proline content. NMR titration experiments (2D 1 H N -15 N TROSY-HSQC) were performed using the 0.1 mM 15 N-labeled S100A6 dimer and 0.05 mM unlabeled A11 2−52 . Similar experiments were carried out using the 0.05 mM 15 N-labeled S100A6 dimer and 0.3 mM unlabeled A11 89−196 and established the lack of S100A6−A11 89−196 interactions. Similar titration measurements were carried out using 0.2 mM 15 N/ 13 C-labeled A11 2−52 and 0.025 mM unlabeled S100A6 dimer. Due to high proline content of A11 2−52 , measurements were also performed using 2D 13 C− 15 N CON correlation experiments. 29 ITC. Measurements were performed using a low-volume Affinity ITC calorimeter (TA Instruments) at 25°C in 25 mM HEPES, pH 7.0, and 5 mM CaCl 2 . 3.6−4.2 μL aliquots of 300 μM A11 2−52 were injected (20 injections) into a cell containing 50 μM S100A6 dimer (n = 3). Results were analyzed by using NanoAnalyze software (TA Instruments).

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschemneuro.3c00169. A11-PRD constructs used in current study and AUC, SPR, ITC, and NMR analyses of their interactions with S100A6; parameters obtained from AUC and NMR and technical details of SPR measurements; caption for Video S1 (PDF) Fusion of A11-PRD droplets (MP4) ■ AUTHOR INFORMATION