ATP induces folding of ALS-causing C71G-hPFN1 and nascent hSOD1

The study addresses how ATP, present at millimolar concentrations in cells, may directly modulate protein folding equilibria independently of its energetic roles. Protein folding is challenged by cellular crowding, marginal stability, and disease-causing mutations that promote misfolding and aggregation. ALS pathogenesis involves misfolding/aggregation of proteins such as human SOD1 (hSOD1) and mutant human profilin 1 (hPFN1). Mature hSOD1 is a highly stable, metal-bound, disulfide-bridged β-barrel dimer, whereas nascent hSOD1 is unfolded and requires Zn^2+ to initiate folding. The ALS-linked C71G mutation in hPFN1 yields a protein that coexists between folded and unfolded states and is aggregation-prone. The central questions are whether ATP can directly shift folding equilibria of these disease-relevant proteins, what chemical features of ATP mediate such effects, and how this compares to known osmolytes (e.g., TMAO) and ions (Zn^2+).

Prior work indicates that ATP at millimolar concentrations can act as a biological hydrotrope, dissolving protein LLPS and aggregates, and modulating phase behavior of intrinsically disordered proteins, and inhibiting amyloid fibrillation. ATP has also been proposed to act as a hydration mediator antagonizing crowding-induced destabilization without specific binding. Nascent hSOD1 is unfolded in vitro and in cells folds via a Zn^2+-initiated maturation pathway, followed by CCS-mediated disulfide formation and copper loading. C71G-hPFN1 exists in folded/unfolded equilibrium and is highly aggregation-prone. Traditional osmolyte trimethylamine N-oxide (TMAO) can stabilize proteins at high molar concentrations but may not be effective for aggregation-prone disease proteins at practical ranges. Inorganic polyphosphates have been reported to function as primordial chaperones, suggesting a potential role of phosphate chemistry in protein homeostasis.

Proteins: WT-hPFN1, C71G-hPFN1, and wild-type hSOD1 were expressed in E. coli; hSOD1 was isolated from inclusion bodies, reduced, and purified by RP-HPLC. Isotope labeling (15N/13C) used M9 media with (15NH4)2SO4 and 13C-glucose. Conditions were optimized to minimize aggregation: C71G-hPFN1 prepared in 1 mM phosphate buffer pH 6.0 with 2 mM DTT; nascent hSOD1 in 1 mM acetate buffer pH 4.5. NMR: Experiments were collected at 25 °C on an 800 MHz Bruker spectrometer with cryoprobe. Sequential assignments employed HNCACB, CBCANH, HSQC-TOCSY, HSQC-NOESY. Backbone dynamics: 15N T1, T1ρ, and {1H}-15N steady-state NOE were measured and analyzed with Model-Free formalism (DYNAMICS), deriving S^2 order parameters and overall rotational correlation times (τc). Conformational exchange between folded (F) and unfolded (U) states of C71G-hPFN1 was quantified using 3D HSQC-NOESY to resolve overlaps and measure exchange cross-peak intensities; exchange parameters (populations and kex) were derived from kinetic equations with R1 values. Translational diffusion coefficients were measured by pulsed field gradient NMR DOSY in D2O; 16 gradient strengths, δ=2 ms, Δ=150 ms, data fitted to extract Ds. Titrations: 50 µM 15N-labeled proteins were titrated with ATP, ADP, AMP, adenosine, ATP analogs (ATPP, AMP-PNP, AMP-PCP), inorganic phosphates (PPP, PP, Pi), Zn2+ (zinc acetate), NaCl, and TMAO across specified molar ratios (e.g., ATP/PPP 1:0.5 to 1:30; for C71G-hPFN1 ATP/ADP/AMP up to 1:400; PPP up to 1:4 before precipitation; PP/AMP-PNP/AMP-PCP up to 1:10 before precipitation; phosphate up to 1:100; NaCl up to 1:200; TMAO up to 1:2000 for C71G-hPFN1 and 1:1000 for hSOD1). Folding/aggregation responses were monitored by 1D up-field methyl signals and 2D 1H-15N HSQC spectra. Zn2+-induced folding of hSOD1 was characterized by emergence of well-dispersed HSQC peaks and chemical shift-based secondary structure propensities (ΔCα−ΔCβ), NOE patterns, and hNOE values. Comparative HSQC overlays and triple-resonance assignments assessed differences between ATP- and Zn2+-induced hSOD1 folded states. Thermal stability: Differential scanning fluorimetry (DSF) with SYPRO Orange measured melting behavior; for WT- and C71G-hPFN1 with varying ATP concentrations; attempts with PPP and hSOD1 yielded noisy, non-cooperative curves due to aggregation/coexistence. Data processing used NMRPipe and NMRView; ionic strengths were calculated for salt comparisons.

• C71G-hPFN1 equilibrium and dynamics: The protein exists in two states with average populations of 55.2% (folded) and 44.8% (unfolded), exchanging at kex ≈ 11.7 Hz (~85.5 ms). WT-hPFN1 exhibits high rigidity (mean hNOE ~0.82; S^2 mean 0.89), while the folded state of C71G-hPFN1 is more flexible (mean hNOE ~0.70; S^2 mean 0.73). τc: WT 7.5 ns; C71G folded 7.8 ns. DOSY: WT D = 1.12 ± 0.03 × 10^−10 m^2/s; C71G D = 1.03 ± 0.02 × 10^−10 m^2/s, indicating slightly reduced compactness in C71G. - ATP effects on C71G-hPFN1: ATP lacks a specific binding pocket on WT or C71G-hPFN1 (minimal HSQC shifts except His120/Gly121). For C71G-hPFN1, ATP shifts the equilibrium to the folded state; complete conversion observed at 1:2 (C71G:ATP), with ratio dependence on protein concentration. - Chemical determinants: Triphosphate (PPP) alone induces folding equivalently to ATP (complete conversion at 1:2), but triggers aggregation at higher ratios (≥1:4). Pyrophosphate (PP) requires higher ratios (≥1:8) and induces aggregation. Phosphate (Pi) and NaCl do not induce folding up to tested ranges. ADP induces conversion at 1:8; AMP fails even at 20 mM; adenosine fails up to 5 mM. Among ATP analogs: ATPP induces folding like ATP at 1:2 but strongly promotes aggregation at ≥1:4; AMP-PCP and AMP-PNP induce at ~1:8 (similar to ADP) and promote aggregation at sub-mM to low mM. The atoms linking β–γ phosphates modulate inducing/aggregation propensities. - Nascent hSOD1: Zn2+ induces a folded population coexisting with U; saturation around 1:20 (hSOD1:Zn2+). ΔCα−ΔCβ and NOE patterns show formation of β-barrel; hNOE average ~0.61 for folded population; no exchange cross-peaks detected, indicating a larger barrier than C71G-hPFN1. - ATP and PPP induce nascent hSOD1 folding: ATP induces a coexisting folded population at 1:8 (saturation near 1:8–1:20), but cannot fully convert U to F. PPP similarly induces folding (onset ~1:4; comparable spectra at 1:8) but causes precipitation ≥1:10. ADP and PP do not induce folding of hSOD1 up to tested ratios before aggregation; adenosine is inactive. - Comparative ATP vs Zn2+ hSOD1 states: HSQC overlays reveal many coincident peaks but notable differences, especially in loop regions comprising the Zn-binding pocket. Triple-resonance assignments of the ATP-induced state show missing assignments over loops (indicative of μs–ms exchange). hNOE values similar overall (ATP-induced mean ~0.59 vs Zn-induced ~0.61), with negative hNOE at residues near the Zn pocket in ATP-induced state, implying enhanced dynamics without Zn coordination. Adding ATP to Zn-induced samples (and vice versa) yields spectra resembling the Zn-induced state, highlighting loop stabilization by Zn2+. - TMAO: Fails to induce folding of C71G-hPFN1 even at 1:2000 and of hSOD1 even at 1:1000; high concentrations cause precipitation. - Thermodynamic stability (DSF): WT-hPFN1 Tm ~56 °C; ATP (up to 20 mM) does not change Tm. C71G-hPFN1 shows no cooperative unfolding without ATP; with ATP, cooperative unfolding appears with Tm ~32 °C at 20 µM ATP (1:2), ~38 °C at 1 mM (1:100), and ~40 °C at 20 mM. PPP did not yield cooperative curves, consistent with aggregation or insufficient tight packing. hSOD1 with Zn2+ and/or ATP did not show cooperative DSF transitions under tested conditions. - Mechanistic proposal: ATP and triphosphate enhance intrinsic folding capacity by modulating protein hydration—displacing water hydrogen-bonded to backbone to favor intramolecular H-bonds, promoting folding. Free triphosphate’s strong electrostatic screening promotes aggregation of partially folded/disordered states with exposed hydrophobic patches; the adenine moiety in ATP may dynamically shield hydrophobic patches, reducing aggregation and enhancing stability.

The findings demonstrate that ATP can directly shift folding equilibria of aggregation-prone ALS-relevant proteins: fully converting C71G-hPFN1 to a folded population at low stoichiometry and promoting a folded population in nascent hSOD1. The triphosphate moiety confers folding induction capacity, while covalent linkage to adenosine attenuates aggregation-triggering effects, providing a rationale for ATP’s unique efficacy compared to inorganic triphosphate. Comparisons with Zn2+ show distinct mechanisms: Zn2+ provides specific structural information to nucleate and stabilize the hSOD1 fold (particularly loops forming the metal site), whereas ATP/PPP act generally by enhancing the sequence-encoded folding propensity, likely via hydration effects. These results address the question of ATP’s non-energetic role in proteostasis, suggesting that cellular millimolar ATP contributes to preventing misfolding/aggregation by promoting folding and stability. They also contextualize the high misfolding propensity of C71G-hPFN1 (due to increased flexibility and lower stability) and the kinetic barriers in hSOD1 maturation (linked to cysteines and metal binding). The work supports the concept that polyphosphates function as primordial chaperones, and suggests that age-related declines in cellular ATP could underlie increased risk or onset of ALS and other neurodegenerative disorders by weakening this energy-independent proteostatic function.

This study reveals a previously unrecognized, energy-independent role of ATP in protein homeostasis: directly inducing folding of disease-relevant proteins at low stoichiometric ratios, with the triphosphate group being essential and the adenine moiety mitigating aggregation and enhancing stability. ATP fully shifts C71G-hPFN1 into a folded population (1:2) and induces a folded population in nascent hSOD1 (1:8), in contrast to TMAO’s ineffectiveness at practical concentrations. Zn2+ and ATP induce folded hSOD1 states via fundamentally different mechanisms. The results rationalize the chaperone-like actions of polyphosphates and suggest that ATP contributes broadly to maintaining proteostasis in modern cells. Future work should probe ATP’s energy-independent effects in vivo under physiological ionic strength and crowding, and guide the design of ATP-like molecules that promote folding, suppress aggregation, and stabilize proteins for therapeutic applications in aggregation-associated diseases.

Experiments were conducted in vitro under low-salt buffers optimized to minimize aggregation; translation to cellular environments remains to be tested. For hSOD1, complete conversion to a solely folded state was not achieved with ATP or Zn2+ (wild-type sequence), reflecting persistent kinetic barriers; exchange kinetics for hSOD1 could not be quantified due to large energy barriers. PPP and several analogs induced rapid aggregation at higher ratios, limiting characterization. DSF measurements with PPP and hSOD1 were noisy and non-cooperative, and thermal stability assessments could not be obtained for all states. Triple-resonance assignments of ATP-induced hSOD1 were incomplete, especially in dynamic loop regions, constraining detailed structural analysis.