A conserved and tunable mechanism for the temperature-controlled condensation of the translation factor Ded1p

Heat shock promotes the assembly of translation factors into condensates to facilitate the production of stress-protective proteins. How translation factors detect heat and assemble into condensates is not well understood. Here, we investigate heat-induced condensate assembly by the translation factor Ded1p from five different fungi, including Ded1p from Saccharomyces cerevisiae. Using targeted mutagenesis and in vitro reconstitution biochemistry, we find that heat-induced Ded1p assembly is driven by a conformational rearrangement of the folded helicase domain. This rearrangement determines the assembly temperature and the assembly of Ded1p into nanometer-sized particles, while the flanking intrinsically disordered regions engage in intermolecular interactions to promote assembly into micron-sized condensates. Using protein engineering, we identify six amino acid substitutions that determine most of the thermostability of a thermophilic Ded1p ortholog, thereby providing a molecular understanding underlying the adaptation of the Ded1p assembly temperature to the specific growth temperature of the species. We conclude that heat-induced assembly of Ded1p into translation factor condensates is regulated by a complex interplay of the structured domain and intrinsically disordered regions which is subject to evolutionary tuning.


31
Temperature affects all aspects of life by setting the pace for cellular processes and 32 biochemical reactions. Proteins are among the most abundant macromolecules in the cell, 33 and temperature-induced changes to the native structure of proteins pose a serious threat to 34 cell survival. With increasing temperature, proteins are prone to unfolding and aggregation, 35 and key proteins lose activity, ultimately leading to heat-induced cell death (Jarzab et al., 2020;36 Leuenberger et al., 2017). 37 more resilient to heat-induced unfolding than the proteomes of mesophilic organisms (Jarzab 48 et al., 2020; Leuenberger et al., 2017) and, accordingly, their HSR is typically activated at 49 higher temperatures (Fields, 2001;Somero, 1995). What determines the thermostability of 50 proteins has been studied with various model proteins. These studies revealed that 51 temperature adaptation of proteins generally does not require large structural changes but can 52 be mediated by few amino acid substitutions that, for instance, introduce (de)stabilizing 53 interactions and/or alter molecular packing (Somero, 1995;Taylor and Vaisman, 2010). 54 While most proteins exhibit unfolding transitions above the viable temperature of an organism, 55 some proteins are only marginally stable and unfold within the physiological temperature range 56 of an organism (Wallace et al., 2015). Examples are several translation factors, which 57 assemble into higher-order structures and change from the soluble to the insoluble fraction in 58 budding yeast exposed to sublethal temperatures (Cherkasov et  The RNA helicase Ded1p is an example of a protein that reversibly assembles into translation 65 factor condensates in the yeast cytoplasm upon exposure to heat. This essential DEAD-box 66 helicase is one of the first proteins to condense upon exposure to heat stress (Iserman et al.,67 2020; Wallace et al., 2015) and this process suppresses the translation of mRNAs that encode 68 housekeeping proteins (Iserman et al., 2020;Sen et al., 2021). Concomitantly, the 69 sequestration of Ded1p has been proposed to promote the preferential production of stress-70 protective proteins, thereby complementing the transcriptional arm of the HSR with 71 translational regulation. In vitro reconstitution experiments revealed that Ded1p autonomously 72 assembles into condensates upon heating by the process of phase separation (Iserman et al., 73 2020). The underlying molecular mechanisms underlying this adaptive response have 74 remained largely undefined. 75 Here, we investigated the molecular mechanism of heat-induced Ded1p condensation and 76 asked if and how the Ded1p assembly temperature is adapted to the temperature niche of 77 organisms. To this end, we identified Ded1p orthologs from different fungi and quantitatively 78 characterized the temperature behavior. We demonstrate that the ability of Ded1p to form 79 condensates upon heating is conserved among the fungal orthologs and that the onset 80 temperature for assembly is adapted to the growth temperature of the respective species. 81 Adaptation of the Ded1p assembly temperature requires evolutionary tuning of the structured 82 helicase domain, and the assembly temperature is determined by the structural stability of the 83 helicase domain. Mutational and structural analyses revealed that the IDRs interact with the 84 helicase domain and that the N-terminal IDR stabilizes the native helicase domain fold of 85 Ded1p. We further show that the IDRs engage in intermolecular protein interactions upon 86 heating and support assembly into micron-sized condensates. We propose that heat-induced 87 assembly of translation factors involves a complex interplay between structured domains and 88 IDRs, which determines both the thermal responsiveness of the protein and its ability to control 89 the translational HSR. We suggest that translation factor assembly into condensates is 90 adapted to the specific thermal niche of an organism and plays a vital role in the execution of 91 the HSR. 92

Identification of Ded1p orthologs from species living in different thermal niches 94
Ded1p is a DEAD-box helicase that uses ATP binding and hydrolysis to unwind complex 95 secondary structures in RNA and to promote translation initiation in yeast (Iost et al., 1999). 96 Upon heat shock, Ded1p assembles into translation factor condensates (Iserman et al., 2020; 97 Wallace et al., 2015). Its sequestration into condensates has been linked to a decrease in the 98 translation of housekeeping and the preferential synthesis of stress protective proteins 99 (Iserman et al., 2020;Sen et al., 2021). While being critical for the cellular response to heat 100 stress, the mechanism by which Ded1p detects changes in environmental temperature and 101 assembles into stress-protective condensates remained unknown. 102 To provide a mechanistic understanding of heat-induced condensation of Ded1p, we set out 103 to identify Ded1p orthologs from different fungi and characterize their behavior upon 104 temperature increases. The availability of a large collection of fungal genomes combined with 105 data on the preferred growth conditions of fungi provided an opportunity to identify Ded1p 106 orthologs and determine how temperature shapes the evolution of Ded1p sequences. We 107 identified a set of 630 orthologous Ded1p sequences from genome analysis and found that 108 Growth assays at different temperatures revealed that Sk did not grow at T ≥ 37°C and Sc did 121 not grow at T ≥ 44°C after 24 h ( Figure 1B). Op grew at all tested temperatures ( Figure 1B). 122 These differences in growth temperature are consistent with published data and suggest that, 123 relative to the mesophile Sc, the growth temperature of Sk is cold-adapted (Salvado et al., 124 2011) and the growth of Op is warm-adapted. Regarding the species from the 125 Sordariomycetes class, Cg has previously been shown to grow well at 34°C but not at ≥45°C, 126 and Tt grew even at 55°C, which contributed to their classification as meso-and thermophile 127 respectively (Morgenstern et al., 2012). In summary, we identified the sequences of Ded1p 128 orthologs from fungi from different classes and these fungi exhibit different temperature growth 129 profiles. Based on the difference in growth temperature, we speculate that the Ded1p 130 orthologs from these species function at different temperature ranges and a quantitative 131 comparison of the temperature response of these orthologs can provide a molecular 132 understanding for heat-induced condensate assembly. 133 The species' growth temperature correlates with the heat-induced Ded1p assembly 134 Previous data using budding yeast suggested that heat induces Ded1p condensation to 135 regulate the translation of housekeeping mRNAs (Iserman et al., 2020). To test if Ded1p 136 orthologs exhibit different assembly temperatures compared to Ded1p from S. cerevisiae, we 137 expressed and purified Sc Ded1p and orthologs as C-terminal GFP-tagged fusion proteins 138 from insect cells and characterized them in vitro. Consistent with published data (Iserman et 139 al., 2020), Sc Ded1p assembled into reversible, spherical and amorphous condensates in the 140 presence or absence of RNA, respectively (Figure 2A). Despite morphological differences, 141 assemblies of Sc Ded1p with and without RNA were detected by microscopy at ~38°C ( Figure  142  Next, we determined the assembly temperatures for Sc Ded1p and the orthologs from Sk, Op,146 Cg and Tt, complementing the dataset from Iserman et al. (2020) with additional Ded1p  147 orthologs. First, we monitored the hydrodynamic radius (rH) as a function of temperature using 148 dynamic light scattering. At 20°C, the rH of the most abundant protein species was similar 149 among Ded1p orthologs (Sc = 6.0 nm ± 0.7 nm, Sk = 6.2 nm ± 0.3 nm, Op = 5.9 nm ± 0.5 nm, 150 Cg = 5.7 nm ± 0.1 nm and Tt = 5.9 nm ± 0.4 nm). Upon temperature increase, the mean rH 151 remained relatively constant until a protein-specific temperature ( Figure 2B) at which the rH 152 increased, indicating assembly into larger particles. Inspecting these samples by fluorescence 153 microscopy revealed that all proteins assembled into morphologically similar condensates 154 (Supplemental Figure 2A-B). Compared to Ded1p from mesophilic Sc, the Ded1p ortholog 155 from the cold-adapted Sk assembled at a lower temperature (DT = -4.3°C) and the ortholog 156 from the warm-adapted Op at a higher temperature (DT = 4.5°C) ( Figure 2B, Table 1). 157 Similarly, the Tonset of the Ded1p ortholog from the mesophilic Sordariomycetes species Cg 158 was 11.3°C below that of the closely related thermophilic ortholog from Tt ( Figure 2B). 159 Comparison of the assembly temperatures and yeast upper growth temperatures ( Heat-induced assembly of Sc Ded1p coincides with tertiary structure changes (Iserman et al., 169 2020). To test if the assembly of Ded1p orthologs is also accompanied by changes in tertiary 170 structure, we recorded fluorescence intensities at 350 and 330 nm as a function of 171 temperature. To distinguish structure changes from temperature-induced fluorescence 172 quenching, we analyzed the fluorescence emission ratio (F350/F330) (Supplemental Figure  173 3A). The fluorescence temperature profiles were characterized by a sigmoid shape with an 174 initial plateau with a drift (native baseline) preceding a steep increase (transition) followed by 175 a final plateau with a drift (unfolded baseline) (Supplemental Figure 3D). The sigmoid change 176 in fluorescence emission ratio is characteristic of cooperative (un)folding transition and 177 demonstrates that all Ded1p orthologs adopted a stable fold at temperatures below the 178 ortholog specific transition temperature. To determine the transition onset (Tonset) and the 179 transition midpoints (Tm), we fitted the data to a two-state transition model. For clarity, the 180 transition profiles are represented as normalized transition data (see methods for details). 181 The Tonset for the structural change was smallest for Ded1p ortholog from the cold-adapted Sk 182 (39.5°C ±0.

Surface-exposed residues in the helicase domain contribute to thermo-adaptation 222
Our data demonstrate that the temperature-induced assembly of Sc Ded1p is determined by 223 the global stability of the helicase domain. Accordingly, the Ded1p orthologs exhibit distinct 224 and different structural stabilities that correlate with their assembly temperature ( Figure 3A).  (Robinson, 2022;Varadi et al., 2022), highlighting residues that were predicted to be either "thermoadaptive" (6-mut, 8-mut, 11-mut and 14-mut) or "non-thermo-adaptive" (Control) and mutated. C. Change in normalized F350/F330 as a function of temperature for 8 µM of the GFP-labelled helicase domains of C. globosum (C.g.) and T. terrestris (Tt) and variants (6-mut, 8-mut, 11-mut, 14-mut and Control). Mean (points), SD (light ribbon), n = 10. D. Change in cumulant radius (rH) as a function of temperature for 8 µM of the GFP-labelled helicase domains of C. globosum (C.g.) and T. terrestris (Tt) and variants (6-mut, 8-mut, 11-mut and Mock). Mean (points), sd (light ribbon), n = 4-5 To test this idea, we set out to increase the structural stability of the helicase domain of Cg  We anticipated that if an amino acid residue is exclusively found in either thermo-or 247 mesophilic Sordariomycetes orthologs, the substitution is more likely to be thermo-adaptive 248 than if that residue occurs in both thermo-and mesophilic orthologs. Following this rationale, 249 we designed four "thermo-adaptive" variants in which we sequentially reduced the number of 250 substitutions between Tt and Cg Ded1p. In these variants, we replaced either fourteen (14-251 mut), eleven (11-mut), eight (8-mut) or six (6-mut) amino acid residues of Cg Ded1p with the 252 corresponding residues from Tt Ded1p ( Figure 5B and Supplemental Figure 4A Figure 4C). This suggests that the key stability-determining residues are in 275 the set of six "thermo-adaptive" amino acid substitutions (6-mut). 276 Next, we analyzed the assembly temperatures for the respective variants ( Figure 5D). In 277 agreement with our fluorescence data, residues which we considered to be non-278 thermoadaptive increased the assembly onset temperature by 2.9°C (Cg 34.4°C ±0.9°C; 279 Control 37.3°C ±0.6°C). In contrast, the thermo-adaptive residues increased the assembly 280 transition temperature by ~8°C (6-mut (42.2°C ±0.8°C), 8-mut (42.6°C ±1.3°C), 11-mut 281 (41.8°C ±0.4°C) and 14-mut (43.2°C ±0.4°C)) ( Figure 5D). In summary, we identified a set of 282 six residues which we consider "thermo-adaptive". These residues appear to be sufficient to 283 increase the stability of the helicase domain as well as the assembly temperature of Ded1p. 284 This suggests that an evolutionary route to adapting the assembly temperature of Ded1p is by 285 changing the temperature stability of the helicase domain. 286

IDR-helicase domain chimeras reveal interactions between the structured domain and 287 the N-terminal IDR 288
Recently, IDRs have been discussed as drivers, as well as modulators of condensate 289 assembly. Our data indicate that the heat-induced assembly of Ded1p is primarily determined 290 by the structural stability of the helicase domain ( Figure 5D). However, our analysis also 291 revealed that Sc Ded1p deletion variants lacking the N-terminal IDRs are structurally 292 destabilized compared to wildtype Sc Ded1p (Figure 3B-C).

The IDRs mediate micron-sized condensate assembly of Ded1p 370
Our NMR data show that the N-terminal and C-terminal IDR are folded back onto the helicase 371 domain, and the interaction between the N-terminal IDR and the helicase domain weakens 372 with increasing temperatures (Figure 7B). We speculate that upon exposure to heat, the IDRs 373 become available to establish intermolecular interactions that drive condensation. Consistent 374 with this reasoning, DLS data reveal that full-length Ded1p assembles into micron-sized 375 condensates upon heating, whereas variants in which the N-terminal, C-terminal or both IDRs 376 were removed assemble into particles that are at least one order of magnitude smaller (8A-377 B). These data suggest that the IDRs provide protein-protein interactions that provide 378 additional valences for growth into condensates. 379 To identify residues that could be important for establishing protein-protein interactions for 381 condensation, we returned to our phylogenetic analysis. While tryptophan residues are 382 typically excluded from IDRs, we found that the tryptophan residues (or aromatic residues in 383 general) in the C-terminal IDR of Sc Ded1p are highly conserved (Supplemental Figure 7). 384 To test whether the C-terminal tryptophan residues influence the heat-induced assembly of 385 Sc Ded1p, we substituted five tryptophan residues in the C-terminal IDR with alanine residues, 386 giving rise to Ded1p-5WA. The Tonset for the structural change of Ded1p-5WA was comparable 387 to that of wildtype Ded1p (Figure 9A). However, the assembly temperature of Ded1p-5WA 388 was increased compared to wildtype Ded1p (Figure 9B), suggesting that the tryptophan 389 residues within the C-terminal IDR contribute to protein-protein interactions for Ded1p 390 condensate assembly. In summary, these data indicate that heat-induced assembly of Ded1p 391 into condensates is regulated by a complex interplay of the IDRs and the structured domain, 392 which is subject to evolutionary tuning. 393  Upon exposure to heat, the structured helicase domain undergoes a change in tertiary structure, and this is sufficient to induce the assembly of nanometer-sized particles. Because of the heat-induced structural change, the IDRs are now available to promote intermolecular assemblies that promote assembly into larger, micron-sized particles. To adapt the assembly temperature of Ded1p to different temperatures, the structural stability of the helicase domain must co-evolve with the IDRs.

397
In this paper, we describe a molecular mechanism for detecting and responding to 398 temperature changes by Ded1p, an essential translation factor that assembles into 399 condensates regulating the HSR (Iserman et al., 2020). Our data suggest that heat triggers a 400 conformational change in the folded helicase domain of Ded1p, which promotes assembly into 401 nanometer-sized particles. This is accompanied by changes in the conformational ensemble 402 of the flanking IDRs, which facilitates assembly into micron-sized condensates. Ded1p 403 assembly into condensates is conserved among various fungi and the assembly temperature 404 is adapted to the species' growth temperature, suggesting that Ded1p's function in the HSR 405  is conserved and adjusted to a species' temperature niche. Given that many other translation 406 factors co-assemble into condensates upon exposure to heat and are composed of both folded 407 and disordered domains, our data suggest a general model for temperature-induced protein 408 assembly and thermo-adaptation of the translational HSR. 409 The helicase domain of Ded1p consists of two globular RecA domains that each adopt a 410 Rossmann fold. The secondary structure of these folds is largely maintained upon temperature 411 increase (Iserman et al., 2020), suggesting that the helicase domain does not undergo 412 extensive denaturation. Here, using nanoDSF, a method that is sensitive to tertiary structure 413 changes, we identified a heat-induced structural change in the helicase domain that coincides 414 with assembly ( Figure 3A-B We previously showed that Ded1p assembly is linked to regulating the HSR (Iserman et al., 426 2020). Given the conservation and thermal adaptation of the HSR, we tested whether the 427 mechanism of Ded1p assembly is also adapted to the thermal niche of a given species. Using 428 in vitro condensation experiments with various fungal Ded1p orthologs, we were able to show 429 that the structural stability of Ded1p correlates with a species' growth temperature ( Table 1). 430 In fact, all tested orthologs contain a single conserved tryptophan residue in their helicase 431 domain (253W in Sc Ded1p) and exhibit similar spectroscopic shifts that presumably 432 represents a similar tertiary structure change (Figure 3A and Supplemental Figure 3C). 433 Importantly, the thermostability of Sc Ded1p is smaller compared to the rest of the proteome, 434 and it changes its tertiary structure well below the lethal temperature of the organism (Iserman  435   et  This region contains a highly conserved motif (RDYR), which is a potential candidate for 473 interacting with the helicase domain. While the C-terminal IDR does not seem to affect the 474 thermostability of Ded1p much (Figure 3B), future work will focus on investigating the role of 475 this motif in regulating Ded1p's function. Interestingly, the Ded1p Alphafold model also 476 predicts with low confidence, an interaction between the N-terminal IDR and the second 477   Our results are consistent with these studies but suggests that IDRs are not essential for heat-500 induced assembly of Ded1p (Figure 8A-B). Rather, condensate assembly by Ded1p requires 501 a concerted action of the structured helicase domain and IDRs. This synergy between folded 502 and disordered domains provides an explanation for the observed co-evolution of these 503 domains in Sc Ded1p to set the assembly temperature (Figure 6).  (Table 3) and these amino acid coordinates were used to design of truncation variants 601 and chimeric proteins in this paper (Table 3-4). The N-terminal and C-terminal regions that 602 flank the Ded1p helicase domain of are predicted to lack defined secondary structure (Buchan 603 and Jones, 2019) and to be intrinsically disordered (Dosztányi et al., 2005). 604   (Table  616 3), the orthologous amino acid sequences were aligned to the sequence of S. cerevisiae 617 Ded1p. 618

C. globosum helicase domain variants 619
To study the thermo-adaptation of Cg Ded1p and Tt Ded1p, helicase domain variants were 620 designed in which amino acids of Cg Ded1p were exchanged with the corresponding amino 621 acid of Tt Ded1p (Table 5). These constructs contain the structured domain only and not the 622 intrinsically disordered regions (Table 5). To identify candidate sites, the amino acid 623 sequences of both Cg (mesophile) and Tt (thermophile) were aligned with orthologous 624 sequences from other fungi from the Sordariomycetes class: Thermothelomyces thermophilus 625 (thermophile, Uniprot ID G2Q8V8), Chaetomium thermophilum (thermophile, uniprot ID 626 G0SG53) and Neurospora crassa (mesophile, uniprot ID Q9P6U9) (Supplemental Figure  627   5B). Increasing numbers of mutations were introduced into the helicase domain of Cg Ded1p 628 ( Table 5) were recorded, averaged and analyzed using the manufacturers software. For temperature 690 experiments, the temperature was increased from 20°C to 60°C with 1°C increments and 30 s 691 equilibration time at each temperature. For temperature experiments, the mean hydrodynamic 692 radii were plotted as a function of temperature, typically between 30°C £ T £ 55°C and mean 693 the rH < 500 nm to visualize variant-specific differences in the assembly temperature (Tonset). 694 The assembly temperature (Tonset) was derived from fitting the data to equation 1, internal reference #1735), the helicase core region (residues 82-535, internal reference 767 #1740) and the C-terminal IDR (residues 533-604, internal reference #2000) of Ded1p were 768 cloned into a modified pET-vector that carries an N-terminal TEV protease cleavable His6-tag. 769 Proteins were expressed overnight at 20°C in E. coli using minimal medium that was 770 supplemented with 15 N ammonium chloride and/ or 13 C6 glucose (for NMR visible proteins) or 771 using LB medium (for NMR invisible proteins). The cells were collected by centrifugation 772 (20°C, 20 minutes, 6.000 g), lysed by sonication in buffer NMR1 (25 mM NaPO4 pH 7.4, 1 M 773 NaCl, 1 mM DTT, 10 mM imidazole) that was supplemented with 0.1% Triton, 1 U/mL DNAse1 774 and 0.1 mg/mL lysozyme, after which the cell debris was removed by a second centrifugation 775 step (4°C, 30 minutes, 35.000 g). The supernatant was loaded onto a gravity flow Ni-NTA 776 column, the column was washed with 5 column volumes buffer NMR1, 10 column volumes of 777 buffer NMR2 (25 mM NaPO4 pH 7.4, 2 M NaCl, 1 mM DTT) and the bound protein was eluted 778 with buffer NMR3 (25 mM NaPO4 pH 7.4, 1 M NaCl, 1 mM DTT, 300 mM imidazole). Overnight, 779 the eluted proteins were dialyzed into buffer NMR4 (20 mM HEPES pH 7.3, 500 mM NaCl, 1 780 mM DTT) in the presence of 1 mg TEV protease. The cleaved affinity tag was removed by a 781 second gravity flow Ni-NTA column using buffer NMR4, after which the protein was applied to 782