Protein thermal sensing regulates physiological amyloid aggregation

Marijan, Dane; Momchilova, Evgenia A.; Burns, Daniel; Chandhok, Sahil; Zapf, Richard; Wille, Holger; Potoyan, Davit A.; Audas, Timothy E.

doi:10.1038/s41467-024-45536-0

Download PDF

Article
Open access
Published: 09 February 2024

Protein thermal sensing regulates physiological amyloid aggregation

Nature Communications volume 15, Article number: 1222 (2024) Cite this article

3469 Accesses
4 Altmetric
Metrics details

Subjects

Abstract

To survive, cells must respond to changing environmental conditions. One way that eukaryotic cells react to harsh stimuli is by forming physiological, RNA-seeded subnuclear condensates, termed amyloid bodies (A-bodies). The molecular constituents of A-bodies induced by different stressors vary significantly, suggesting this pathway can tailor the cellular response by selectively aggregating a subset of proteins under a given condition. Here, we identify critical structural elements that regulate heat shock-specific amyloid aggregation. Our data demonstrates that manipulating structural pockets in constituent proteins can either induce or restrict their A-body targeting at elevated temperatures. We propose a model where selective aggregation within A-bodies is mediated by the thermal stability of a protein, with temperature-sensitive structural regions acting as an intrinsic form of post-translational regulation. This system would provide cells with a rapid and stress-specific response mechanism, to tightly control physiological amyloid aggregation or other cellular stress response pathways.

Emergence of fractal geometries in the evolution of a metabolic enzyme

Article Open access 10 April 2024

Optogenetic control of mRNA condensation reveals an intimate link between condensate material properties and functions

Article Open access 15 April 2024

Local environment in biomolecular condensates modulates enzymatic activity across length scales

Article Open access 18 April 2024

Introduction

One of the fundamental features of life is the ability to detect and respond to changing cellular conditions. This feature is critically important, as deviations from a normal growth environment occur frequently, and modulation of metabolic processes is necessary to maintain homeostasis. Collectively, these adaptations are known as stress response pathways, with cell survival depending on sensing the correct stress and eliciting an appropriate response. Stimuli such as heat shock¹, DNA damage², hypoxia³, oxidative stress⁴, and unfolded protein accumulation in the ER⁵ each require unique metabolic changes to facilitate cell recovery. However, significant crosstalk between signaling cascades often occurs, as disparate stimuli can activate common elements of certain stress response pathways. For example, heat shock proteins are upregulated by elevated temperatures, low oxygen availability⁶, heavy metal contamination⁷, and oxidative stress⁸. Thus, gaining insight into the molecular mechanisms that a cell uses to activate specific and common stress-response factors is crucial to understanding these important signaling networks.

Over the last decade, the emergence of biomolecular condensates has altered the landscape of cellular stress response research. Across the eukaryotic domain, the formation of membrane-less organelles have been tightly linked to changing environmental conditions^9,10. In yeast, energy depletion and low pH cause reversible Sup35 condensate formation¹¹, while Sec bodies of Drosophila melanogaster form during amino acid starvation¹². Mammalian nuclei are a hub for stress-induced condensate biogenesis¹³, with foci such as paraspeckles, nuclear stress bodies, and nucleolar aggresomes forming in response to an array of cellular insults^{14,15,16,17,18}. Though much remains to be learned about the functionality of these condensates, common biological effects of their formation include the modulation of biochemical reactions and the buffering of cellular protein concentration¹⁹. Biophysically, many of these condensates possess liquid-like properties, as proteins sequestered within these structures retain much of their dynamic profiles²⁰. The transition of these condensates from a liquid to solid state has been linked to numerous disorders, as striking evidence suggests that aberrant aggregation of disease-associated proteins within these structures may underlie the etiology of prominent human pathologies^10,20. Unlike their liquid-like counterparts, stress-inducible amyloid bodies (A-bodies) were found to naturally possess solid-like properties, as a heterogeneous family of proteins adopt an amyloid-like structure within this subnuclear domain^21,22. The amyloid state is characterized by unbranching fibrils of polypeptides stacked perpendicularly to form a β-strand-rich, insoluble, and immobile amyloid spine²³. Although this protein form has been historically viewed as toxic²⁴, there is growing evidence that eukaryotic cells utilize functional amyloids in numerous physiological roles, including the storage of peptide hormones²⁵, melanin synthesis²⁶, sequestration of organelles and RNA in oocytes²⁷, programmed necrosis signaling²⁸, and memory persistence^29,30. Biogenesis of A-bodies occurs in response to elevated temperature or low environmental pH, while a subsequent return to normal growth conditions results in their disassembly^21,31. These stimuli mediate the expression of low complexity non-coding RNA derived from the ribosomal intergenic spacer region, which nucleates the formation of dense, fibrillar, protease-resistant, and Congo red-positive condensates^21,32,33. As these reversible amyloids have been found in species from across the eukaryotic domain³¹, the evolutionary conservation highlights the significance of this biomolecular condensate and suggests that these foci represent an ancient stress response pathway.

We recently showed that the A-body proteome was dependent on the stressor that induced this physiological amyloid aggregation event²². While some proteins acted as universal targets, other constituents are sequestered only in response to specific stimuli. Based on the low complexity nature and non-specific charge-based interactions ascribed to the seeding noncoding RNA, we hypothesized that a protein-level mechanism must exist to discriminate the constituents upon exposure to A-body-inducing stressors. Here, we use two related protein pairs as model systems to identify critical structural elements that mediate heat shock-specific amyloid aggregation. DDX39A and DDX39B are members of the DEAD box family of RNA helicases, which have been implicated in genomic integrity³⁴, RNA splicing³⁵, and mRNA export³⁶, while the heterogeneous nuclear ribonucleoprotein A0 (hnRNPA0) and A1 (hnRNPA1) have been linked to multiple aspects of RNA biogenesis³⁷. Thus, stress-specific recruitment of these proteins could significantly alter genome stability and RNA metabolism, providing a site of cellular regulation that tailors this stress-response pathway to different environmental perturbations. Our work demonstrates that individual proteins can possess thermo-sensitive structures that act as direct temperature sensors to mediate A-body recruitment and aggregation. Interestingly, the intrinsically disordered domains of these proteins did not contribute to this stress-specific aggregation mechanism, as the highly-ordered structures possessed the thermo-sensing functionality. Collectively, these data represent a post-translational regulatory mechanism where physiological amyloid aggregation can be rapidly and specifically controlled by the thermo-sensitivity of the tertiary structure of individual cellular proteins.

Results

Differential A-body targeting of related RNA helicases during heat stress

The mechanism regulating stress-specific amyloid aggregation within the A-body has not been established, though screening the acidotic²¹ and heat shock A-body proteomes²² revealed a pair of closely related RNA helicases that possess different targeting specificities. DDX39A and DDX39B share 90% sequence identity (Supplementary Fig. 1a) and a similar dumbbell-shape when comparing the predicted (DDX39A) and crystal (DDX39B) structures³⁸ (Fig. 1a). However, proteomics data suggests that DDX39A aggregates within both acidotic- and heat shock-induced A-bodies, while DDX39B is only present in the acidosis-induced structures^21,22. To validate these intriguing mass spectrometry results, we expressed GFP-tagged DDX39A and DDX39B in MCF-7 cells. As predicted, under acidotic conditions both proteins were targeted to Thioflavin S-positive structures that contained the A-body marker molecule CDC73 (Fig. 1b). For these RNA helicases, the putative adoption of an amyloid-like conformation within A-bodies was demonstrated by the hallmark shift towards the insoluble (Supplementary Fig. 1b) and immobile (Supplementary Fig. 1c, d) biophysical state and the inherent capacity of these proteins to generate fibrils, when expressed in a bacterial inclusion body assay (Fig. 1c: a model setting of amyloid aggregation^21,39,40). Heat shock exposure also induced the expected results, as DDX39A formed aggregates in Thioflavin S- and CDC73- positive foci (Fig. 1b and Supplementary Fig. 1b, c, e), while DDX39B remained nuclear, soluble, and mobile (Fig. 1b and Supplementary Fig. 1b, d, e) at elevated temperatures. The stress-specific targeting/aggregation properties of these proteins were also observed in A549 and HEK293 cell lines (Supplementary Fig. 2a, b) and extended to HIS-tagged variants of both proteins (Supplementary Fig. 2c, d), demonstrating that the cell line and GFP-tag did not mediate this divergence in stress-specific A-body aggregation.

**Fig. 1: Closely related proteins, DDX39A and DDX39B, are differentially targeted to A-bodies.**

To uncover the mechanism regulating heat shock-specific aggregation, we took advantage of the high structural and sequence similarity of DDX39A and DDX39B (Fig. 1a and Supplementary Fig. 1a). Here, a series of substitution constructs were created where regions of DDX39A were replaced by the corresponding sequences in DDX39B (Fig. 1d-Left Panel). Using an established method for quantifying A-body targeting efficiency (Supplementary Fig. 2e)^31,41, a central segment emerged as a critical differentiator of heat shock-induced protein sequestration (Fig. 1d, e), insolubilization (Supplementary Fig. 2f, g), and immobilization (Fig. 1f). DDX39A constructs containing amino acids 100-250 of DDX39B remained nuclear, soluble and mobile under heat shock conditions, while the presence of amino acids 100-250 from DDX39A targeted any DDX39B construct into the A-bodies, immobilizing and insolubilizing the chimeric protein (Fig. 1d−f and Supplementary Fig. 2f, g). We next wanted to use purified proteins to assess whether DDX39A and DDX39B responded directly to changing temperatures. Unfortunately, inclusion body formation (Fig. 1c) prevented the isolation of bacterially-expressed protein, and multiple attempts to generate soluble mammalian-expressed protein were unsuccessful. Thus, we shifted our efforts to examine the effects of temperature on DDX39 within harvested lysates, which were pre-cleared of aggregates and cellular debris by centrifugation. Soluble supernatants were incubated at 4 °C or 43 °C (1 h), then separated by non-denaturing- and denaturing-PAGE. Despite the uniform presence of protein in the lysates (Fig. 1g, SDS-PAGE: Total), monomeric DDX39B could be seen at both 4 °C and 43 °C, while DDX39A monomers were absent from the 43 °C sample (Fig. 1g, Native-PAGE). To detect putative temperature-induced aggregates, we centrifuged aliquots of the incubated lysates, and ran the decanted pellets on SDS-PAGE. Here, we can see that incubation at 43 °C clearly shifted DDX39A, but not DDX39B, into an aggregated state (Fig. 1g, SDS-PAGE: Pellet), demonstrating that the divergent responses of these proteins to elevated temperatures persist within an in vitro setting. Swapping the central domains (100-250) of these proteins abrogated the thermo-sensitive aggregation of DDX39A, and imparted temperature-mediated insolubilization on DDX39B (Fig. 1g). Together, these results suggested that the central region allows DDX39 proteins to directly sense changing temperatures and mediates stress-specificity amyloid aggregation.

Protein structure regulates a universal A-body targeting motif

Two mechanistic models arise from the data presented above. Either a heat shock-specific A-body targeting motif is present in the central region of DDX39A (absent in DDX39B), or DDX39B possesses a thermo-stable tertiary structure that masks a generic aggregation motif. To test these alternate hypotheses, we created a series of truncation mutations to map the minimal A-body targeting motif. We focused on the N-terminal lobe of the DDX39 proteins, as this domain contains amino acids 100−250 and possesses the same stress-specific A-body targeting phenotypes as the full-length proteins (Supplementary Fig. 3a). Using quantitative microscopy to gauge the efficiency of A-body recruitment (Supplementary Fig. 2e), our results demonstrate that amino acids 75−250 of both proteins were capable of sequestering GFP under heat shock conditions (Fig. 2a). This suggests that DDX39A and DDX39B contain a generic A-body localization domain, which has the capacity to mediate sequestration and aggregation of both proteins at elevated temperatures. Finer mapping showed that there are potentially three subnuclear targeting motifs present within the central region of both proteins (amino acid positions 100−150, 150−200, and 200−250) (Fig. 2a, b and Supplementary Fig. 3b, c). As AmylPred2⁴² predicts that each of these regions contain aggregation prone clusters (Supplementary Fig. 3b, c – Bottom Panel), we synthesized DDX39 peptides and ran a Thioflavin T (ThT) fibrillation assay to determine whether DDX39 fragments could adopt an amyloid conformation. In this assay, peptides from the central region of DDX39A (160−199) and DDX39B (161−200) generated ThT-positive amyloid fibrils at a significantly higher rate than AmylPred2-negative DDX39A (1−39) and DDX39B (1-40) regions (Fig. 2c and Supplementary Fig. 3d). Thus, these data point towards DDX39 proteins possessing a generic targeting motif that facilitates the amyloid-like aggregation of these proteins within A-bodies.

**Fig. 2: DDX39A and DDX39B contain generic A-body targeting motifs.**

This panel of N-terminal truncation mutations also highlights the essential role of amino acids 50−74 in DDX39B stress-specific aggregation. Both DDX39B (50−250) and DDX39B (75−250) contain the putative A-body targeting motif(s) described above, yet the addition of these 25 amino acid residues prevented heat shock-mediated A-body recruitment (Fig. 2a, Supplementary Fig. 3a). Based on our data, it seemed unlikely that this sequence contains a discrete motif that actively inhibits A-body targeting, as it; (1) falls outside of the stress-specificity region identified in Fig. 1, (2) contains no amino acid substitutions unique to DDX39B (Supplementary Fig. 1a), and (3) fails to disrupt A-body targeting in DDX39B (1-150)/DDX39B (1-200) (Fig. 2a) or DDX39B (100-250) upon direct fusion (DDX39B [50-74 + 100-250], Supplementary Fig. 3e). A more probable explanation for the effects of amino acids 50-74 on A-body stress-specificity, is that this region is important for the proper folding and tertiary structure of this protein. IUPred3 analysis and AlphaFold predictions (AF-Q13838-F1) support this notion, as the first ~50 residues of DDX39B encode an intrinsically disordered domain, while the next 25 amino acids are part of the N-terminal ordered structure (Fig. 2d). Thus, we predict that thermal stability of the tertiary structure could be a critical determinant in heat shock-specific A-body recruitment.

To further characterize the role of the central region of DDX39B in A-body targeting and protein aggregation we generated a panel of single amino acid substitutions that replaced each of the unique residues in DDX39A with their DDX39B equivalents (Fig. 3a). Of the 17 sites that encoded different amino acids within the 100-250 region, multiple substitutions significantly repressed A-body recruitment (Fig. 3b). The F184L and C223V substitution were the most prominent, as they could independently reduce A-body localization (Fig. 3b, c) and increase DDX39A mobility (Supplementary Fig. 4a) to DDX39B levels. Under acidotic conditions neither F184L nor C223V had any effect on full-length DDX39A sequestration (Fig. 3c and Supplementary Fig. 4b), and their incorporation into the minimal DDX39A (99-249) constructs also failed to repress A-body recruitment under both heat shock and acidotic conditions (Supplementary Fig. 4c). This suggests that DDX39B amino acid substitutions are not disrupting the generic acidosis/heat shock A-body targeting motif(s). Next, we sought to perform the reciprocal experiments and impart the heat shock A-body targeting properties of DDX39A on DDX39B. We noted that both F184L and C223V substitutions independently inhibit DDX39A recruitment (Fig. 3b, c), suggesting that these residues exert a dominant-negative effect on this form of heat-induced protein aggregation. Therefore, we expected the presence of either residue in DDX39B would prevent its aggregation. As predicted, the single reciprocal mutations L185F and V224C (DDX39B contains one extra residue at amino acid 18 [Supplementary Fig. 1a], hence the different residue position) failed to induce A-body targeting in DDX39B, as each construct possessed an inhibitory residue at the other established site (Fig. 3d). However, the double substitution DDX39B (L185F + V224C) also did not aggregate within A-bodies at elevated temperatures (Fig. 3d–f and Supplementary Fig. 4a, d), suggesting that other critical inhibitory regions are present within DDX39B. To explore this further, we sequentially added reciprocal amino acid substitutions that were shown to partially impair DDX39A subnuclear targeting (Fig. 3b). Each substitution additively increased heat shock-induced recruitment of DDX39B, culminating in the DDX39B (septuplet: L185F + V224C + S115T + A146S + L107I + H189N + I168V) construct being sequestered, insolubilized, and immobilized within A-bodies as efficiently as DDX39A (Fig. 3d–f and Supplementary Fig. 4a, d). Reverting the L185F or V224C residues back to the wild-type amino acids sequences in the DDX39B (septuplet) construct completely abolished heat shock-induced targeting, insolubilization, and immobilization (Fig. 3d–f and Supplementary Fig. 4a, d). Together, these data generate a putative model where generic A-body targeting motif(s) are masked under heat shock conditions by multiple thermo-stable structural features within the protein.

**Fig. 3: Central amino acid residues can modulate DDX39A and DDX39B heat shock A-body targeting.**

Distinct structural pockets regulate heat shock-mediated A-body recruitment

Prompted by these findings, we examined the location of each residue within the predicted structure of DDX39A. This revealed that all seven amino acid substitutions are spatially clustered into three pockets in the N-terminal domain (Fig. 4a). Two of these pockets are hydrophobic in nature, surrounding residues C223 and F184/S145 (Fig. 4b), while the third pocket is hydrophilic and contains the residues T114, I106, N188, and V167 (Fig. 4c). Based on the additive effect of the latter four amino acids on DDX39B sequestration (Fig. 3d–f and Supplementary Fig. 4a, d), we predicted that the combined inhibition of substituting all four residues in DDX39A might parallel the results seen by altering the F184L and C223V pockets (Fig. 3b). Individually, each of these residues impair DDX39A sequestration (Fig. 3b), however, the DDX39A (quadruple; T114S + I106L + N188H + V167I) construct repressed heat shock-induced A-body targeting to wild-type DDX39B levels (Fig. 4d), validating that these four residues cooperatively coordinate a third site of structural regulation in the DDX39 protein. To probe the properties of the hydrophobic pockets further, we replaced the phenylalanine and cysteine residues at positions 184 and 223 of DDX39A, respectively, with the other 19 amino acids and assessed A-body targeting. At position 184 only medium-sized hydrophobic residues (leucine, isoleucine and methionine) were capable of inhibiting A-body targeting (Fig. 4e), while valine was the only amino acid that significantly repressed heat shock-mediated sequestration at position 223 (Fig. 4f). Interestingly, the substitution of large, polar, and/or charged residues at these sites almost uniformly increased A-body targeting beyond the wild-type protein levels (Fig. 4e, f). These findings mirrored the effects observed with large N-terminal truncation mutations, where putatively unstructured fragments displayed greater A-body targeting efficiency than the properly folded full length (1-427 and 1-428) DDX39A and DDX39B (Fig. 2a). We suspected the biophysical properties of the third pocket to be distinct, as it contains hydrophilic residues. To examine this, we substituted the third pocket residue that shows the greatest inhibitory effect on DDX39A A-body targeting (Fig. 3b: T114), with all 19 other amino acids. The results showed that there was greater variability in the residues that could repress A-body targeting at this location (Fig. 4g), and that charged/polar amino acid substitutions did not dramatically enhance A-body recruitment, unlike the effects seen in the F184L and C223V hydrophobic pockets (Fig. 4e, f). Therefore, these data indicate that structural features regulate A-body targeting, with the three pockets in DDX39B imparting thermo-stability on the protein. Interestingly, the residues comprising the three pockets neatly aligned with each of the three minimal targeting motifs (Supplementary Fig. 5) previously identified (Fig. 2b).

**Fig. 4: Identification of structural pockets that regulate DDX39 targeting to A-bodies.**

To rigorously challenge this model, our aim was to restore heat shock-induced A-body aggregation to the DDX39A (F184L), DDX39A (C223V), and DDX39A (quadruple) mutant constructs by altering the pockets that surround these critical residues. The selected DDX39A constructs only encode one putative thermo-stable pocket each (i.e., with a DDX39B-like sequence/structure), allowing for precise control of their thermo-stability via strategic introduction of alanine mutations that reduce critical residues to a single methyl group within the three predicted pockets (Fig. 5a: pink, cyan, and yellow). As controls we also designed one alanine mutation for each pocket, V159A, K240A, and N110A, which are adjacent to the rescue substitutions, but whose R groups are oriented away from the respective pocket cores (Fig. 5a: green). Once heat shocked, A-body targeting of DDX39A (F184L), DDX39A (C223V), and DDX39A (quadruple) variants were significantly increased by each of the pocket rescue mutations, while mutants carrying the positional controls remained diffusely distributed (Fig. 5b). Additionally, while rescue variants displayed decreased mobility and solubility when detained in heat shock A-bodies, their analogous positional control mutants remained mobile and soluble (Supplementary Fig. 6a–c). Collectively, these data validate the role of thermo-sensitive structural pockets in heat shock-specific sequestration, allowing the DDX39 tertiary structures to act as intrinsic regulators of their own A-body targeting and aggregation.

**Fig. 5: Heat shock A-body targeting is regulated by distinct structural pockets.**

The molecular dynamics of DDX39A are sensitive to elevated temperature

To further investigate the response of DDX39A and DDX39B to heat, we subjected these proteins to a ~ 1.6 microsecond molecular dynamics simulation at 37 °C and 43 °C. The results validate a model where exposure of A-body targeting motifs likely results from local conformational changes driven by thermally-enhanced dynamics. During the simulation the average root mean square deviation (RMSD) of the ordered regions of the N-terminal lobes of DDX39A and DDX39B (amino acids 46-250) were similar at 37 °C (Fig. 6a, b-top panel). However, at 43 °C the same region of DDX39A exhibited a markedly larger RMSD difference, relative to the DDX39B protein, (Fig. 6a, b-bottom panel), highlighting the thermo-stability of the A-body excluded DDX39B and thermo-sensitivity of the A-body targeted DDX39A at elevated temperatures.

**Fig. 6: Elevated temperature enhances the dynamic properties of the thermo-sensitive A-body constituent DDX39A.**

To investigate where these enhanced dynamics are localized, we calculated the root mean square fluctuation (RMSF) of the Cα atoms in each simulation and compared DDX39A and DDX39B at 43 °C. The results clearly demonstrate a large perturbation in dynamics occurring on the N-terminal lobe of DDX39A, centered at serine 183 (Fig. 6c), which constitutes a loop and helix covering the F184 hydrophobic pocket (Fig. 6d). How these dynamics translate to exposure of the A-body targeting motif(s), and how they are altered in the context of the other structural domains is a subject of future investigation. Nevertheless, these results show that the N-terminal domain of DDX39A experiences enhanced dynamics at increased temperature, while DDX39B does not, supporting the overarching hypothesis.

Structural regulation of heat shock A-body recruitment is not exclusive to DDX39 proteins

In order to validate this regulatory mechanism beyond the DDX39 helicases, we searched the MS datasets^21,22 to find related proteins outside the DDX family that may have a heat shock A-body targeting and aggregation pattern akin to DDX39A and DDX39B. This revealed two structurally similar proteins from a heterogeneous nuclear ribonucleoprotein subfamily, hnRNPA0 and hnRNPA1. Both proteins possess two N-terminal globular domains and a large C-terminal IDR (Fig. 7a), with the MS data indicating that only hnRNPA0 is a putative A-body resident during heat shock exposure. By using specific antibodies to visualize the two endogenous proteins, we found that hnRNPA0 and hnRNPA1 acted as expected. In heat shock treated cells, hnRNPA0 was targeted to A-bodies (Fig. 7b) and insolubilized (Supplementary Fig. 7a), while hnRNPA1 remained diffuse (Fig. 7b) and soluble (Supplementary Fig. 7a). Our in vitro assay also showed that hnRNPA0 underwent temperature-sensitive aggregation in extracted cell lysates, while hnRNPA1 remained primarily monomeric and soluble at high temperatures (Fig. 7c). Thus, based on the A-body targeting discrepancy and overall structural similarity we attempted to manipulate the aggregation properties of hnRNPA0 and hnRNPA1 in a fashion analogous to the DDX39A/B experiments above. We confirmed that the GFP-tagged constructs maintain the same heat shock targeting/aggregation properties as their endogenous counterparts (Fig. 7b and Supplementary Fig. 7b) and created chimeric hnRNPA constructs where we substituted the analogous globular domains and/or IDRs (Fig. 7d, e). During heat shock, only constructs carrying the first globular domain of hnRNPA0 (amino acids 1-92) displayed A-body targeting, while those carrying amino acids 1–99 of hnRNPA1 remained diffuse (Fig. 7e). This strongly suggests that regulation of A-body targeting and protein aggregation was mediated by this globular domain of hnRNPA0/A1, and like DDX39, was independent of the intrinsically disordered regions, which were shown not to affect localization (Fig. 7e and Supplementary Fig. 7c) or mobility during heat shock exposure (Supplementary Fig. 7d). Observation of this regulatory domain suggests that it may represent a simplified version of the DDX39 structure (Figs. 4b and 7f), as this globular region of hnRNPA0/A1 contains a single hydrophobic pocket (i.e., a β-sheet and two helices surrounding a hydrophobic core). To test whether disruption of this hydrophobic core can induce heat shock-mediated A-body recruitment of hnRNPA1, we substituted a central phenylalanine (Fig. 7f) for a destabilizing, charged aspartate residue, strategically choosing a mutation site analogous to the L185 position that was crucial for inducing DDX39B targeting (Fig. 3). As an indirect comparison we also generated L185D or V224D substitutions in DDX39B, before observing protein localization in heat shock treated cells (Supplementary Fig. 7e). Introducing the single destabilizing mutations into either of the DDX39B pockets resulted in A-body recruitment comparable to that of the DDX39B (Septuplet) (Supplementary Figs. 7e, 3e). Remarkably, the analogous F34D mutation in hnRNPA1 also resulted in a gain of heat shock A-body targeting/immobility (Fig. 7e and Supplementary Fig. 7d, e). Finally, to make sure this was not too drastic a destabilization effect, due to a charge residue being introduced into the hydrophobic core, we generated an additional hnRNPA1 mutant with a F34A substitution. This construct was also targeted to A-bodies (Fig. 7e, g), where it was insolubilized (Fig. 7h) and immobilized (Fig. 7i) under heat shock conditions. Together, these results suggest that A-body recruitment and aggregation is not based on random protein misfolding, but that specific sensor motifs in unrelated proteins have developed a distinct regulatory role controlling physiological amyloid aggregation.

**Fig. 7: Destabilization of hnRNPA1 regulatory pocket causes heat-induced A-body targeting.**

Discussion

In this study we explored the mechanisms regulating heat shock-specific protein recruitment and aggregation within A-bodies, using sets of related proteins as a model system. Despite overwhelming similarities in their sequences, predicted structures, and aggregation propensity, these proteins are differentially recruited to functional amyloids upon exposure to elevated temperatures. Here, we exploited these similarities to identify discrete structural pockets that control the accessibility of generic A-body targeting sequences. By precisely altering these structural pockets, we were able to modify amyloid aggregation behavior under heat shock conditions, without affecting the trafficking of these proteins under other A-body-inducing stimuli. Thus, we propose a model where the tertiary structure of protein dictates A-body targeting, and acts as an intrinsic sensor to control amyloidogenesis. This effect is likely imparted by a thermo-sensitive conformational shift that occurs in DDX39A and hnRNPA0 (but not the more thermally-stable DDX39B and hnRNPA1), to unmask generic A-body targeting and aggregation domains (Fig. 8).

**Fig. 8: Model depicting the thermo-sensing mechanism of A-body constituents.**

While there has been scientific progress in the phase transition community, the events mediating the formation of physiological amyloid-like assemblies are not well understood. A-body biogenesis is believed to occur through a multi-step process, as the structure initially possesses liquid-like properties before solidifying as the stress persists³³. Liquid phase separation often relies on flexible, transient, and multivalent interaction networks between low complexity domains and RNA^10,43. Similarly, functional amyloid assembly also requires the presence of prion-like or disordered regions in their constituent proteins^{27,29,44,45,46,47}. An interesting finding in our study demonstrated that although DDX39 and hnRNPA proteins contain intrinsically disordered regions, these low-complexity sequences were not required for imparting stress-specificity or aggregation properties in the A-body pathway. Instead, generic A-body targeting motifs were found in the ordered structures of the DDX39 proteins, buried in regulatory pockets that were sensitive to temperature changes. Despite the presence of these localization signals within ordered regions of the native structure, we suspect that the appropriate environmental condition could locally denature the regulatory pockets and generate transient low complexity or prion-like regions that facilitate the phase separation events preceding amyloid fibrillation. This notion is supported by molecular dynamics simulations that indicate the N-terminal lobe of DDX39A (but not DDX39B) is sensitive to increased temperatures. This would suggest that A-bodies not only recruit misfolded proteins⁴⁸ and unstable peptide fragments⁴⁹, but that precise regulatory mechanisms emerged during the structural evolution of A-body constituents to act as intrinsic stress-sensors.

How proteins acquire their native fold is a longstanding and fundamental biological question⁵⁰. However, despite remarkable advances in prediction tools^51,52 and theoretical models⁵³ the question remains largely unresolved^54,55. It is fascinating to observe that proteins are generally only marginally stable, with a relatively small energy difference between the folded and unfolded state⁵⁶. DDX39A/DDX39B and hnRNPA0/hnRNPA1 are related protein pairs, which share substantial protein sequence and structural similarities. However, our data demonstrates that their aggregation potential under heat shock conditions is markedly different. We propose that a critical element in the evolutionary divergence of these protein pairs was to tweak the marginal stability of DDX39A and hnRNPA0, and through the addition of thermally-destabilizing mutations tip the protein towards the aggregation state when temperatures increase. Other amino acid substitutions, like those shown not to affect A-body targeting of DDX39 (Fig. 3b), can potentially be associated with the paralog-specific functions. While DDX39B and DDX39A share some redundancy in mRNA processing and export^57,58,59, the molecular complexes they form allow them to distinguish between different RNA transcripts^36,60. Similarly, hnRNPA family proteins are involved throughout mRNA maturation, with their altered expression linked to tumorigenesis and neurodegenerative diseases such as ALS^61,62,63. By combining functional and structural differences in these RNA interacting proteins, evolution may have allowed the cell to use marginal stability/selective A-body sequestration to tailor gene expression under heat shock conditions via modulation of RNA maturation.

In general, we believe that this thermo-sensing mechanism can be applied to many additional proteins. Mass spectrometry data indicate there are >250 stress-specific A-body constituents²², suggesting there could be hundreds of additional thermo-stable and thermo-sensitive proteins within this aggregation pathway alone. It is tempting to speculate that a mechanism of stress-sensitive protein stability could also be applied to other A-body-inducing stimuli. It is well established that, like heat, changes in the cellular pH can have a destabilizing effect on proteins^64,65. Thus, it is conceivable that generic A-body targeting motif could also be masked by structural pockets that are pH-sensitive. Local unfolding under acidotic conditions could be triggered by amino acid R-groups changing from polar to charged (e.g., histidine: pKa_{(side chain)}=6.0), locally denaturing the protein and allowing A-body recruitment and protein aggregation. Evidence from several functional amyloids gives credence to this idea, with Staphylococcal biofilm-associated protein containing a pH switch domain that enables formation of amyloid aggregates in acidic environments⁶⁶ and the melanosome protein Pmel17 and Neuropeptide Hormone GnRH requiring low pH conditions to fibrillate^67,68. By extension, proteins containing natively-exposed targeting motifs may represent an additional class of universal A-body targets, with their recruitment and aggregation relying purely on upregulation of the seeding RNA^32,33. With future studies, we aim to more fully characterize these mechanisms of stress-specific protein aggregation. We envision that this form of intrinsic structural control developed over the course of evolution to be a potent post-translational regulatory mechanism, which doesn’t rely on cumbersome and energetically costly covalent modifications. This mechanism likely also extends to other stress response networks, allowing cells to harness the power of protein stability/dynamics for the quick and reversible inactivate cellular pathways in an environmental context dependent manner.

Methods

Cell culture, transfection, treatments

MCF-7, A549 and HEK-293 cells (ATCC) were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS and 1% penicillin-streptomycin. Monolayer cultures were grown at 37 °C, 5% CO₂ and passaged every 2−3 days. Transfection using PEI (Millipore-Sigma) was performed 1 day following cell seeding, allowing cells to reach ~60% confluency. Unless specified otherwise, heat shock and acidosis treatments were performed 1 day after transfection for 2 h by placing cells in a 43 °C incubator, or by exchanging growth media with acidic DMEM (pH 6) and placing cells in a humidified Hypoxia chamber at 37 °C, 1%O₂, and 5% CO₂ (H35 HypOxystation, Don Whitley Scientific).

Microscopy imaging, immunofluorescence, staining, and quantification

Cells were seeded in 35 mm plates with 20 x 20 mm, 1 mm thick glass coverslips. Following treatment, cells were immediately fixed with cold methanol. Immunostained samples were blocked (10% horse serum) and incubated with α-His antibody (ThermoFisher Scientific, MA1-21315, 1:1000), α-hnRNPA1 (ThermoFisher Scientific, PA5-19431, 1:200) or α-hnRNPA0 (ThermoFisher Scientific, PA5-57722, 1:200) followed by α-mouse Alexa 488 (ThermoFisher Scientific, A11001, 1:400) or α-rabbit Alexa 488 (ThermoFisher Scientific, A11008, 1:400). Thioflavin S (Millipore-Sigma) staining was performed on formaldehyde-fixed, TritonX-100 permeabilized cells²¹. Coverslips mounted in Fluoromount G (ThermoFisher Scientific) were imaged on a Zeiss LSM880 confocal laser scanning microscope with Airyscan (Carl Zeiss Microscopy) using Zen 2.3 software (black edition, Carl Zeiss Microscopy). Image processing was performed with Zen 3.1 (blue edition, Carl Zeiss Microscopy), and no non-linear adjustments were made.

Fixed samples were imaged, and A-body presence was quantified as the average pixel intensity of 1−2 A-body regions relative to the average pixel intensity of the nuclear background for each cell, following image background normalization (Supplementary Fig. 2e). Intensity measurements were generate using ImageJ 1.52a (National Institutes of Health). Values were calculated by averaging 10 images of representative cells in at least three biological replicates. The data is presented as the mean of biological replicates +/- standard error of the mean. Statistical significance was measured via students two tailed t-test, indicated as significant when p ≤ 0.05.

Fluorescence recovery after photobleaching

Live cells were grown on glass-bottom 35 mm culture dishes (MatTek Corporation) and visualized by confocal microscopy (Zeiss LSM880 with Airyscan, Carl Zeiss Microscopy). Bleached areas were subjected to a 100% argon laser pulse at 488 nm and imaged for the indicated times post-bleaching. For A-body-detained proteins, a partial region of the A-body was photobleached, while diffuse nucleoplasmic or cytoplasmic proteins had an equal size portion of those regions photobleached instead. Intensity measurements were made using ImageJ 1.52a (National Institutes of Health) and normalized^21,22,31. All fluorescence recovery after photo- bleaching experiment data represent an average of 10 cells in three independent replicates. The data was normalized⁶⁹ and presented as the mean of three biological replicates +/- standard error of the mean.

Insoluble fractionation, western blotting, and quantification

Treated cells were harvested in NP40 lysis buffer (1% NP40 + 50 mM Tris-HCl + 150 mM NaCl) and fractionated²¹. Briefly, samples underwent two rounds of sonication (60 s, 1 s pulse on/off at 10% power) and centrifugation, prior to re-suspending the insoluble pellet in fresh NP40 lysis buffer. Western blotting used antibodies to detect GFP (Santa Cruz Biotechnology, sc-9996, 1:1000), GAPDH (Santa Cruz Biotechnology, sc-47724, 1:4000), Histone H3 (Cell Signaling Technology, 9715, 1:10,000), hnRNPA1 (ThermoFisher Scientific, PA5-19431, 1:1,000), hnRNPA0 (ThermoFisher Scientific, PA5-57722, 1:1000), and HIS-tag (ThermoFisher Scientific, MA1-21315, 1:5000). α-mouse HRP (ThermoFisher Scientific, A16011, 1:10,000) or α-rabbit HRP (ThermoFisher Scientific, A16023, 1:10,000) were applied prior to detection on the Amerhsam Imager 600 (GE Healthcare Life Sciences). Relative intensities were calculated as the average pixel intensity of heat shocked insoluble fraction bands compared to the average pixel intensity of the cell lysate bands, normalized against background. Intensity measurements were made using ImageJ 1.52a (National Institutes of Health). For each sample at least three independent blots were measured, and data is presented as the mean value of biological replicates +/- standard error of the mean. Statistical significance was measured via Student’s two tailed t-test.

In vitro aggregation

MCF-7 cells transfected with plasmids encoding DDX39A-GFP, DDX39A(B100-250)-GFP, DDX39B, DDX39B(A100-250)-GFP, hnRNPA0-GFP, or hnRNPA1-GFP were lysed in a non-denaturing buffer (1xPBS + 1 mM PMSF) and sonicated for 60 s (1 s on, 1 s off) to break cell membranes. Insoluble aggregates and other cellular debris were removed from the lysates by centrifugation (5700 x g for 10 min at 4 °C) and total soluble fraction was transferred to a new microcentrifuge tube. Approximately 100 μg of lysate was incubated at 4 °C and 43 °C for 1 h. Non-denaturing loading buffer (0.01% Brilliant Blue + 10% glycerol + 0.3 M Tris-HCl) was added to native-PAGE samples, while SDS-PAGE samples were boiled for 12 min in denaturing loading buffer (0.02% bromophenol blue +2% SDS + 10% DTT + 10% glycerol + 0.3 M Tris-HCl) prior to loading into 10% TGX-Free FastCast Acrylamide gels (BioRad, 1610173). Native-PAGE electrophoresis was done with a Tris-Glycine (no SDS) buffering system. Aliquots of incubated lysates were also pelleted (12,900 x g for 15 min at 4 °C) and following decanting of the supernatant the pellets were re-suspended in non-denaturing loading buffer, prior to boiling (12 min) and SDS-PAGE. Following electrophoresis, native-PAGE gel was agitated for 15 min in a 0.05% SDS solution, then for 15 min in ddH₂O. All samples were transferred to PVDF membrane and immunoblotted with the GFP antibody (Santa Cruz Biotechnology, sc-9996: 1/600).

Bacterial inclusion body formation and Thioflavin T fibrillation

BL21 cultures expressing GFP, DDX39A-GFP, DDX39B-GFP, or β-amyloid (1-42)-GFP were grown at 32 °C for 20 h. Microscopy was performed on fixed (4% formaldehyde, 1 h) and permeabilized (0.5% Triton X-100, 30 min) bacterial cells, stained with amyloidophilic dye (Congo red) and DNA dye (DAPI).

Thioflavin T fibrillation assays were performed using DDX39A (1-39), DDX39B (1-40), DDX39A (160-199), and DDX39B (161-200) peptides synthesized by GenScript, which were re-suspended to a final concentration of 200 µM in ddH₂O⁴¹. Experiments were carried out in 96-well black-walled plates sealed with optical film. Each well contained a glass bead, Thioflavin T (10 µM), DDX39 peptide (10 µM), and the final volume was adjusted to 100 µL with 1xPBS (pH = 10.5). In a microplate reader (Infinite M200 Pro, Tecan), plates were incubated at 37 °C and shook for one minute between readings (60 s of linear shaking with an amplitude of 1 mm and frequency of 886.9 RPM). Thioflavin T was excited (450 nm) and emission (490 nm) fluorescence was measured every 2 min for 16 h. Data was generated from 3 independent replicates, and presented as the mean value of the independent replicates +/- standard error of the mean. Statistical significance was measured via Student’s two tailed t test.

Plasmids and mutagenesis

Wild-type DDX39A, DDX39B, CDC73, hnRNPA0, and hnRNPA1 cDNA were generated by RT- PCR from MCF-7 RNA and cloned into pEGFP-C1, pmCherry, pET-30, pcDNA3.1-His, pcDNA 3.1/CT-GFP-TOPO (Invitrogen) plasmids. Each of the DDX39A/DDX39B and hnRNPA0/hnRNPA1 domain swapping and point mutations were generated in the pEGFP-C1 or pcDNA 3.1/CT-GFP-TOPO backbones, respectively, using a modified form of site-directed mutagenesis⁷⁰. Construct names refer to the DDX39A or DDX39B, and hnRNPA0 or hnRNPA1 protein backbones with the corresponding amino acid(s) or region(s) substitutions in brackets. Note, DDX39B contains a valine at position 18 that is absent from DDX39A, leading to a discrepancy when naming equivalent amino acid positions C-terminal of that residue.

Structural models and protein disorder/aggregation propensity prediction

Structure models were generated, analyzed, and presented using PyMOL Molecular Graphics System (Version 2, Schrödinger, LLC.). Protein structures used in this work were, DDX39B (AF-Q13838-F1 and PDB: 1XTI)³⁸, DDX39A (generated on a 1XTI template via the Fold and Function Assignment System)⁷¹, the hnRNPA1 RNA-recognition motif (PDB: 1L3K)⁷², and full-length hnRNPA1/hnRNPA0 (predicted by AlphaFold)⁵¹. Disordered protein regions were determined using the IUPred3 interface⁷³. Consensus aggregation propensity was calculated using AmylPred 2⁴².

Molecular Dynamics simulation and analysis

DDX39A and DDX39B were first solvated with TIP3P water using a 1.5 nm padding between the edges of the box. NaCl was added to neutralize the system and bring the ionic strength of the solution to 0.15 M. The total number of particles in the resulting systems was ~95,000 atoms (Supplementary Table 1). Simulations are performed with molecular dynamics library OpenMM 7.6 using the Amber99SBILDN forcefield for protein and TIP3P force field for water with rigid bond constraints and neutralized with counterions. Electrostatics was evaluated with the Particle Mesh Ewald Algorithm using a nonbonded cutoff of 1 nm. A 1.5x hydrogen mass repartitioning was employed, allowing a 4 fs timestep with middle scheme Langevin integrator with a collision rate of 1/ps. The pressure was controlled every 25 steps with a Monte Carlo barostat. The energy was minimized with the steepest descent algorithm, and the system was equilibrated for 50 ps. Each protein was subject to a 1.6-microsecond long simulation at temperatures 37 °C and 43 °C with frames recorded at 40 ps intervals. After processing the solvent out of the resulting trajectories, we separated each system into N- and C-terminal domains and aligned them independently with the starting structure. Using the MDAnalysis python library we calculated the RMSF and RMSD of the C alpha atoms. RMSD was calculated using every 100th frame^74,75. Violin plots were generated using Prism 10 (GraphPad).

Statistics and reproducibility

All graphs represent the mean values of at least three biological replicates, with individual data points (dots) overlaid on the bar graph. All sample sizes used in the experiments were in line with those reported in the literature for similar experiments. Error bars represent the standard error of the mean (s.e.m.), and p values were calculated using the two-tailed unpaired Student’s t test, with the significance level of p ≤ 0.05. Representative microscopy and western blot images were captured from experiments that were independently repeated at least three times with similar results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The paper and supplementary information contains all the data needed to evaluate the conclusions of the work. Raw data and p values for all Figures and Supplementary Figs. are present in the Source Data files. Initial coordinate, simulation input files, and coordinate files of the final output of the molecular dynamics simulations are available online (https://github.com/PotoyanGroup). DDX39B (AF-Q13838-F1 and PDB: 1XTI, [https://doi.org/10.2210/pdb1XTI/pdb]), hnRNPA0 (AF-Q13151-F1), hnRNPA1 (AF-P09651-F1 and PDB: 1L3K, [https://doi.org/10.2210/pdb1L3K/pdb]) protein structures were used in this work. Any additional information is available upon request to the corresponding author (T.E.A: taudas@sfu.ca). Source data are provided with this paper.

References

Richter, K., Haslbeck, M. & Buchner, J. The heat shock response: life on the verge of death. Mol. Cell 40, 253–266 (2010).
Article CAS PubMed Google Scholar
Ciccia, A. & Elledge, S. J. The DNA damage response: making it safe to play with knives. Mol. Cell 40, 179–204 (2010).
Article CAS PubMed PubMed Central Google Scholar
Lee, P., Chandel, N. S. & Simon, M. C. Cellular adaptation to hypoxia through hypoxia inducible factors and beyond. Nat. Rev. Mol. Cell Biol. 21, 268–283 (2020).
Article CAS PubMed PubMed Central Google Scholar
Sies, H., Berndt, C. & Jones, D. P. Oxidative stress. Annu. Rev. Biochem. 86, 715–748 (2017).
Article CAS PubMed Google Scholar
Walter, P. & Ron, D. The unfolded protein response: from stress pathway to homeostatic regulation. Science 334, 1081–1086 (2011).
Article ADS CAS PubMed Google Scholar
Baird, N. A., Turnbull, D. W. & Johnson, E. A. Induction of the heat shock pathway during hypoxia requires regulation of heat shock factor by hypoxia-inducible factor-1. J. Biol. Chem. 281, 38675–38681 (2006).
Article CAS PubMed Google Scholar
Wagner, M., Hermanns, I., Bittinger, F. & Kirkpatrick, C. J. Induction of stress proteins in human endothelial cells by heavy metal ions and heat shock. Am. J. Physiol. - Lung Cell. Mol. Physiol. 277, L1026–L1033 (1999).
Article CAS Google Scholar
Kalmar, B. & Greensmith, L. Induction of heat shock proteins for protection against oxidative stress. Adv. Drug Deliv. Rev. 61, 310–318 (2009).
Article CAS PubMed Google Scholar
Franzmann, T. M. & Alberti, S. Protein phase separation as a stress survival strategy. Cold Spring Harb Perspect Biol. 11, a034058 (2019).
Alberti, S. & Hyman, A. A. Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat. Rev. Mol. Cell Biol. 22, 196–213 (2021).
Article CAS PubMed Google Scholar
Franzmann, T. M. et al. Phase separation of a yeast prion protein promotes cellular fitness. Science 359, eaao5654 (2018).
Article PubMed Google Scholar
Zacharogianni, M., Gomez, A. A., Veenendaal, T., Smout, J. & Rabouille, C. A stress assembly that confers cell viability by preserving ERES components during amino-acid starvation. Elife 3, 1–25 (2014).
Article Google Scholar
Lacroix, E. & Audas, T. E. Keeping up with the condensates: the retention, gain, and loss of nuclear membrane-less organelles. Front. Mol. Biosci. 9, 1–17 (2022).
Article ADS Google Scholar
Biamonti, G. & Vourc’h, C. Nuclear stress bodies. Cold Spring Harb. Perspect. Biol. 2, 1–13 (2010).
Article Google Scholar
Latonen, L., Moore, H. M., Bai, B., Jäämaa, S. & Laiho, M. Proteasome inhibitors induce nucleolar aggregation of proteasome target proteins and polyadenylated RNA by altering ubiquitin availability. Oncogene 30, 790–805 (2011).
Article CAS PubMed Google Scholar
Latonen, L. Phase-to-Phase with nucleoli – stress responses, protein aggregation and novel roles of RNA. Front. Cell. Neurosci. 13, 1–10 (2019).
Article Google Scholar
Adriaens, C. et al. P53 induces formation of NEAT1 lncRNA-containing paraspeckles that modulate replication stress response and chemosensitivity. Nat. Med. 22, 861–868 (2016).
Article CAS PubMed Google Scholar
McCluggage, F. & Fox, A. H. Paraspeckle nuclear condensates: global sensors of cell stress? BioEssays 43, 1–18 (2021).
Article Google Scholar
Lyon, A. S., Peeples, W. B. & Rosen, M. K. A framework for understanding the functions of biomolecular condensates across scales. Nat. Rev. Mol. Cell Biol. 22, 215–235 (2021).
Article CAS PubMed Google Scholar
Shin, Y. & Brangwynne, C. P. Liquid phase condensation in cell physiology and disease. Science 357, eaaf4382 (2017).
Article PubMed Google Scholar
Audas, T. E. et al. Adaptation to stressors by systemic protein amyloidogenesis. Dev. Cell 39, 155–168 (2016).
Article CAS PubMed PubMed Central Google Scholar
Marijan, D. et al. Stress‐specific aggregation of proteins in the amyloid bodies. FEBS Lett. 1873-3468.13597 https://doi.org/10.1002/1873-3468.13597 (2019).
Iadanza, M. G., Jackson, M. P., Hewitt, E. W., Ranson, N. A. & Radford, S. E. A new era for understanding amyloid structures and disease. Nat. Rev. Mol. Cell Biol. 19, 755–773 (2018).
Article CAS PubMed Google Scholar
Wang, M., Audas, T. E. & Lee, S. Disentangling a bad reputation: changing perceptions of amyloids. Trends Cell Biol. 27, 465–467 (2017).
Article CAS PubMed PubMed Central Google Scholar
Maji, S. K. et al. Functional amyloids as natural storage of peptide hormones in pituitary secretory granules. Science 325, 328–332 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Fowler, D. M. et al. Functional amyloid formation within mammalian tissue. PLoS Biol. 4, 0100–0107 (2006).
CAS Google Scholar
Boke, E. et al. Amyloid-like self-assembly of a cellular compartment. Cell 166, 637–650 (2016).
Article CAS PubMed PubMed Central Google Scholar
Li, J. et al. The RIP1/RIP3 necrosome forms a functional amyloid signaling complex required for programmed necrosis. Cell 150, 339–350 (2012).
Article CAS PubMed PubMed Central Google Scholar
Majumdar, A. et al. Critical role of amyloid-like oligomers of Drosophila Orb2 in the persistence of memory. Cell 148, 515–529 (2012).
Article CAS PubMed Google Scholar
Si, K., Choi, Y. B., White-Grindley, E., Majumdar, A. & Kandel, E. R. Aplysia CPEB can form prion-like multimers in sensory neurons that contribute to long-term facilitation. Cell 140, 421–435 (2010).
Article CAS PubMed Google Scholar
Lacroix, E. et al. Evolutionary conservation of systemic and reversible amyloid aggregation. J. Cell Sci. 134, jcs258907 (2021).
Article CAS PubMed Google Scholar
Audas, T. E., Jacob, M. D. & Lee, S. Immobilization of proteins in the nucleolus by ribosomal intergenic spacer noncoding RNA. Mol. Cell 45, 147–157 (2012).
Article CAS PubMed Google Scholar
Wang, M. et al. Stress-induced low complexity RNA activates physiological amyloidogenesis. Cell Rep. 24, 1713–1721.e4 (2018).
Article CAS PubMed PubMed Central Google Scholar
Yoo, H. H. & Chung, I. K. Requirement of DDX39 DEAD box RNA helicase for genome integrity and telomere protection. Aging Cell 10, 557–571 (2011).
Article CAS PubMed Google Scholar
Shen, H. et al. Distinct activities of the DExD/H-box splicing factor hUAP56 facilitate stepwise assembly of the spliceosome. Genes Dev. 22, 1796–1803 (2008).
Article CAS PubMed PubMed Central Google Scholar
Yamazaki, T. et al. The closely related RNA helicases, UAP56 and URH49, preferentially form distinct mRNA export machineries and coordinately regulate mitotic progression. Mol. Biol. Cell 21, 2953–2965 (2010).
Article CAS PubMed PubMed Central Google Scholar
Jean-Philippe, J., Paz, S. & Caputi, M. hnRNP A1: the swiss army knife of gene expression. Int. J. Mol. Sci. 14, 18999–19024 (2013).
Article PubMed PubMed Central Google Scholar
Shi, H., Cordin, O., Minder, C. M., Linder, P. & Xu, R. M. Crystal structure of the human ATP-dependent splicing and export factor UAP56. Proc. Natl. Acad. Sci. USA 101, 17628–17633 (2004).
Article ADS CAS PubMed PubMed Central Google Scholar
de Groot, N. S., Sabate, R. & Ventura, S. Amyloids in bacterial inclusion bodies. Trends Biochem. Sci. 34, 408–416 (2009).
Article PubMed Google Scholar
Wang, L., Maji, S. K., Sawaya, M. R., Eisenberg, D. & Riek, R. Bacterial inclusion bodies contain amyloid-like structure. PLoS Biol. 6, 1791–1801 (2008).
Article CAS Google Scholar
Chandhok, S. et al. Stress-mediated aggregation of disease-associated proteins in amyloid bodies. Sci. Rep. 13, 14471 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Tsolis, A. C., Papandreou, N. C., Iconomidou, V. A. & Hamodrakas, S. J. A consensus method for the prediction of ‘aggregation-prone’ peptides in globular proteins. PLoS One 8, 1–6 (2013).
Article Google Scholar
Rhine, K., Vidaurre, V. & Myong, S. RNA droplets. Annu. Rev. Biophys. 49, 247–265 (2020).
Article CAS PubMed PubMed Central Google Scholar
Berchowitz, L. E. et al. Regulated formation of an amyloid-like translational repressor governs gametogenesis. Cell 163, 406–418 (2015).
Article CAS PubMed PubMed Central Google Scholar
Hoashi, T. et al. The repeat domain of the melanosomal matrix protein PMEL17/GP100 is required for the formation of organellar fibers. J. Biol. Chem. 281, 21198–21208 (2006).
Article CAS PubMed Google Scholar
Raveendra, B. L. et al. Characterization of prion-like conformational changes of the neuronal isoform of Aplysia CPEB. Nat. Struct. Mol. Biol. 20, 495–501 (2013).
Article CAS PubMed PubMed Central Google Scholar
Mompeán, M. et al. The structure of the necrosome RIPK1-RIPK3 Core, a human hetero-amyloid signaling complex. Cell 173, 1244–1253.e10 (2018).
Article PubMed PubMed Central Google Scholar
Frottin, F. et al. The nucleolus functions as a phase-separated protein quality control compartment. Science 365, 342–347 (2019).
Article ADS CAS PubMed Google Scholar
Mediani, L. et al. Defective ribosomal products challenge nuclear function by impairing nuclear condensate dynamics and immobilizing ubiquitin. 1–19 https://doi.org/10.15252/embj.2018101341 (2019).
Dill, K. A. & Maccallum, J. L. The Protein-Folding Problem, 50 Years On. Science 338, 1042–1046 (2012).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).
Nassar, R., Dignon, G. L., Razban, R. M. & Dill, K. A. The protein folding problem: the role. Theory J. Mol. Biol. 433, 167126 (2021).
Article CAS PubMed Google Scholar
Outeiral, C., Nissley, D. A. & Deane, C. M. Current structure predictors are not learning the physics of protein folding. Bioinformatics 38, 1881–1887 (2022).
Article CAS PubMed PubMed Central Google Scholar
Moore, P. B., Hendrickson, W. A., Henderson, R. & Brunger, A. T. The protein-folding problem: not yet solved. Science 375, 507–507 (2022).
Taverna, D. M. & Goldstein, R. A. Why are proteins marginally stable? Proteins Struct. Funct. Genet. 46, 105–109 (2002).
CAS Google Scholar
Stäßer, K. et al. TREX is a conserved complex coupling transcription with messenger RNA export. Nature 417, 304–308 (2002).
Article ADS Google Scholar
Pryor, A. et al. Growth-regulated expression and G0-specific turover of the mRNA that encodes URH49, a mammalian DExH/D box protein that is highly related to the mRNA export protein UAP56. Nucleic Acids Res. 32, 1857–1865 (2004).
Article CAS PubMed PubMed Central Google Scholar
Kapadia, F., Pryor, A., Chang, T. H. & Johnson, L. F. Nuclear localization of poly(A)+ mRNA following siRNA reduction of expression of the mammalian RNA helicases UAP56 and URH49. Gene. 384, 37–44 (2006).
Article CAS PubMed Google Scholar
Huang, C., Liang, D., Tatomer, D. C. & Wilusz, J. E. A length-dependent evolutionarily conserved pathway controls nuclear export of circular RNAs. Genes Dev. 32, 639–644 (2018).
Article CAS PubMed PubMed Central Google Scholar
Lu, Y. et al. Heterogeneous nuclear ribonucleoprotein A/B: an emerging group of cancer biomarkers and therapeutic targets. Cell Death Discov. 8, 1–15 (2022).
Article Google Scholar
Geuens, T., Bouhy, D. & Timmerman, V. The hnRNP family: insights into their role in health and disease. Hum. Genet. 135, 851–867 (2016).
Article CAS PubMed PubMed Central Google Scholar
Thibault, P. A. et al. Hnrnp a/b proteins: an encyclopedic assessment of their roles in homeostasis and disease. Biology 10, 712 (2021).
Article CAS PubMed PubMed Central Google Scholar
Anfinsen, C. B. Principles that govern the folding of protein chains. Science 181, 223–230 (1973).
Article ADS CAS PubMed Google Scholar
Matthew, J. B. et al. pH-dependent processes in protein. Crit. Rev. Biochem. 18, 91–197 (1985).
Article CAS Google Scholar
Ma, J. et al. Structural mechanism for modulation of functional amyloid and biofilm formation by Staphylococcal Bap protein switch. EMBO J. 40, 1–16 (2021).
Article ADS Google Scholar
Pfefferkorn, C. M., McGlinchey, R. P. & Lee, J. C. Effects of pH on aggregation kinetics of the repeat domain of a functional amyloid, Pmel17. Proc. Natl. Acad. Sci. USA 107, 21447–21452 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Dharmadana, D., Reynolds, N. P., Conn, C. E. & Valéry, C. PH-Dependent self-assembly of human neuropeptide hormone gnrh into functional amyloid nanofibrils and hexagonal phases. ACS Appl. Biol. Mater. 2, 3601–3606 (2019).
Article CAS Google Scholar
Muzzopappa, F. et al. Detecting and quantifying liquid–liquid phase separation in living cells by model-free calibrated half-bleaching. Nat. Commun. 13, 7787 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Zeng, F. et al. Efficient strategy for introducing large and multiple changes in plasmid DNA. Sci. Rep. 8, 1–12 (2018).
Article ADS Google Scholar
Xu, D., Jaroszewski, L., Li, Z. & Godzik, A. FFAS-3D: Improving fold recognition by including optimized structural features and template re-ranking. Bioinformatics 30, 660–667 (2014).
Article CAS PubMed Google Scholar
Vitali, J. et al. Correlated alternative side chain conformations in the RNA-recognition motif of heterogeneous nuclear ribonucleoprotein A1. Nucleic Acids Res. 30, 1531–1538 (2002).
Article CAS PubMed PubMed Central Google Scholar
Erdős, G., Pajkos, M. & Dosztányi, Z. IUPred3: prediction of protein disorder enhanced with unambiguous experimental annotation and visualization of evolutionary conservation. Nucleic Acids Res. 49, W297–W303 (2021).
Article PubMed PubMed Central Google Scholar
Gowers, R. et al. MDAnalysis: a Python package for the rapid analysis of molecular dynamics simulations. Proc. 15th Python Sci. Conf. 98–105 https://doi.org/10.25080/majora-629e541a-00e (2016).
Michaud-Agrawal, N., Denning, E. J., Woolf, T. B. & Beckstein, O. MDAnalysis: a toolkit for the analysis of molecular dynamics simulations. J. Comput. Chem. 32, 2319–2327 (2011).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This work was supported by the Canadian Institute of Health Research (T.E.A., PJT-162364) and Natural Sciences and Engineering Research (T.E.A., RGPIN/04998-2017). D.A.P. acknowledges funding from the National Institute of General Medical Sciences (NIGMS) of the National Institutes of Health for financial support (R35 GM138243). T.E.A. acknowledges the kind support of the Canada Research Chairs program for a Tier II Canada Research Chair: Cellular Stress (CRC-2021-00117).

Author information

These authors contributed equally: Evgenia A. Momchilova, Daniel Burns, Sahil Chandhok.

Authors and Affiliations

Department of Molecular Biology and Biochemistry, Simon Fraser University, 8888 University Drive, Burnaby, BC, V5A 1S6, Canada
Dane Marijan, Evgenia A. Momchilova, Sahil Chandhok, Richard Zapf & Timothy E. Audas
Centre for Cell Biology, Development, and Disease, Simon Fraser University, 8888 University Drive, Burnaby, BC, V5A 1S6, Canada
Dane Marijan, Evgenia A. Momchilova, Sahil Chandhok, Richard Zapf & Timothy E. Audas
Roy J. Carver Department of Biochemistry, Biophysics and Molecular Biology, Iowa State University, Ames, IA, 50011, USA
Daniel Burns & Davit A. Potoyan
Department of Biochemistry, University of Alberta, Edmonton, Alberta, T6G 2H7, Canada
Holger Wille
Centre for Prions and Protein Folding Diseases, University of Alberta, Edmonton, Alberta, T6G 2M8, Canada
Holger Wille
Neuroscience and Mental Health Institute, University of Alberta, Edmonton, Alberta, T6G 2E1, Canada
Holger Wille
Department of Chemistry, Iowa State University, Ames, IA, 50011, USA
Davit A. Potoyan

Authors

Dane Marijan
View author publications
You can also search for this author in PubMed Google Scholar
Evgenia A. Momchilova
View author publications
You can also search for this author in PubMed Google Scholar
Daniel Burns
View author publications
You can also search for this author in PubMed Google Scholar
Sahil Chandhok
View author publications
You can also search for this author in PubMed Google Scholar
Richard Zapf
View author publications
You can also search for this author in PubMed Google Scholar
Holger Wille
View author publications
You can also search for this author in PubMed Google Scholar
Davit A. Potoyan
View author publications
You can also search for this author in PubMed Google Scholar
Timothy E. Audas
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

DM performed experiments, with assistance from EAM, SC and RZ. DM, HW, DAP and TEA designed the experiments. DB and DAP performed molecular dynamics simulations and analyses. DM and TEA conceptualized the study and wrote the manuscript. All authors contributed to review and editing. TEA supervised the project.

Corresponding author

Correspondence to Timothy E. Audas.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Samir Maji and the other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Marijan, D., Momchilova, E.A., Burns, D. et al. Protein thermal sensing regulates physiological amyloid aggregation. Nat Commun 15, 1222 (2024). https://doi.org/10.1038/s41467-024-45536-0

Download citation

Received: 24 May 2023
Accepted: 25 January 2024
Published: 09 February 2024
DOI: https://doi.org/10.1038/s41467-024-45536-0

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.