The Intrinsically Disordered Region of Coronins Fine-tunes Oligomerization and Actin Polymerization

Coronins are highly conserved actin-binding proteins (ABPs) in the eukaryotic kingdom for polymerizing actin cytoskeleton. The biochemical activity of coronins is primarily mediated by the structural N-terminal β-propeller and the C-terminal helical coiled-coil (CC) domains, but less is known about the function of a middle nonconserved region, the “unique region (UR)”. The coronin UR is an intrinsically disordered region (IDR). Herein, we demonstrate that the low complexity of the UR is a conserved signature of the coronin protein family, and the UR/IDR exhibits a striking evolutionary correlated pattern associated with sequence length. By analyzing the role of the IDR in coronins via coarse-grained simulations, we reveal that evolutionary selection of IDR length is coupled with the oligomerization of IDR-containing proteins (IDPs) to provide optimal functional output. By integrating biochemical and cell biology experiments and protein engineering, we found that the IDR regulates Crn1 biochemical activity, both in vivo and in vitro, by fine-tuning CC domain oligomerization and maintaining Crn1 in a tetrameric state. The IDR-guided optimization of Crn1 oligomerization is critical for Arp2/3-mediated actin polymerization.


Evolutionarily conservation of IDR in coronin family proteins 254
The function of the IDR in fine-tuning CC domain oligomerization motivated us to ask whether and how 255 this mechanism is evolutionarily conserved. We analyzed the IDR at the N-terminus of the CC domain 256 across coronin family proteins. We collected 392 coronin homologs from 200 species, including species 257 from all major phyla, based on a previously reported phylogenetic analysis of coronin proteins 66 , and gen-258 erated a taxonomic tree from these homologs ( Figure 5). We defined their IDR by identifying each unique 259 region between the conserved β-propeller and CC domains. Each IDR was also analyzed by IUPs (website: 260 https://iupred2a.elte.hu/) to validate low-complexity sequences. We found that the presence of the IDR is 261 evolutionarily conserved among coronin family proteins, although the IDR is conserved in neither sequence 262 nor in length, which is a common feature of the IDR 67,68 . Coronins in only three out of 200 species, 263 T.vaginalis (protozoan), A.anophagefferens (protozoan), and B.natans (unicellular eukaryotic alga), do not 264 contain an IDR. Then, we manually mapped the analyzed IDR into the coronin taxonomic tree (named IDR-265 MAP hereafter), in which the IDR length is shown in scale ( Figure 5). Strikingly, we found that although 266 the IDRs vary in general, they follow a clustered pattern in which the IDR length within each taxonomic 267 clade is similar but clearly different from that of other clades. The coronins of most of the protozoa (includ-268 ing SAR, Excavata and Amoebozoa) contain a short IDR (< 50 aa), and coronins from only Trypanoso-269 matidae of Excavata and Aconoidasida of Alveolata contain a long IDR (between 50-100 aa or >100 aa). 270 However, in the fungal kingdom, all coronins under Ascomycota (branch marked with a black star), includ-271 ing S. cerevisiae, carry a long IDR (> 150 aa). In contrast, coronins in most metazoans, especially mammals, 272 have a short IDR (< 90 aa), except for Nematoda, in which the IDR is generally longer (> 150 aa). Our 273 coronin IDR-MAP suggests a potential evolutionary selection mechanism that determined IDR length. 274

The evolutionary interplay between IDR length and the stability of CC oligomers 275
The evolutionarily conserved presence of the IDR in coronin proteins (197 from 200 species) motivated us 276 to ask whether and why the IDR is necessary but varies in length among homologs. We were unable to 277 perform a global investigation of coronin homologs at the same depth. Here, we investigated two well-278 studied human coronin homologs, Coronin 1A and Coronin 1C 2,24,36,38 . We compared the in vivo localiza-279 tion of full-length Coronin 1A with that of Coronin 1C (1A-FL and 1C-FL) and their respective truncated The importance of the IDR for human and yeast coronins motivated us to further ask why coronin proteins 290 within distinct evolutionary clades maintain IDRs of different lengths. We first tested whether a shorter 291 IDR would be sufficient for yeast Crn1 functions by maintaining a short, random 40-aa middle fragment of 292 Crn1-IDR (Crn1ΔIDR-S, Figure 7A). Crn1ΔIDR-S was expressed at a protein level similar to that of WT 293  Figure 2A). We also ex-305 amined whether the short IDR of Crn1ΔIDR-S would be sufficient to modulate the oligomerization state of 306 Crn1-CC. We applied the same computational approach used for Crn1-ΔN ( Figure 2C) and Crn1-CC (Fig-307 ure 2D) to Crn1-S-IDR-CC. The results showed that the potential energy per chain decreased as the oligo-308 meric number increased in a hydrophobic interaction-dependent manner throughout the scope of investiga-309 tion ( Figure 7G). The fact that the potential energy per chain decreased further after the tetrameric state was 310 reached suggests that a short IDR lacks the capability to stabilize the CC domain in a tetrameric state, as 311 Crn1-FL can ( Figure 2C). 312 11 We next sought to understand how the IDR tunes the interaction between adjacent helical chains and con-313 sequently modulates the CC domain into optimal oligomeric states. First, using coarse-grained simulations, 314 we investigated how different IDRs influence the oligomerization of the mouse Coronin 1A CC domain 315 (named MmCC), which forms a stable parallel homotrimer in crystals and in solution (PDB ID: 2akf) 43 . 316 We compared the energy landscape of MmCC when fused with its own IDR (aa 403-429, MmIDR-MmCC), 317 short Crn1-S-IDR (aa 484-523, ScSIDR-MmCC), and Crn1-IDR (aa 401-604, ScIDR-MmCC) ( Figure 7H). 318 Intriguingly, all three different recombinants showed no further decrease in energy after the trimeric state 319 was reached, indicating that none of the three IDRs altered the oligomeric state of MmCC (Figures 7H; 320 that the IDR modulates the packing of Crn1-CC helices, which are unstable due to nonideal interacting 12 residues, but has a less pronounced influence on the well-packed MmCC. Such evolved CC domain se-344 quences and packing patterns of coronin family proteins may thereby contribute to evolutionary selection 345 of IDR length. 346

347
Oligomerization regulates the biochemical activities of ABPs in actin assembly 348 The initiation, reorganization, and depolymerization of the actin cytoskeleton are biochemically orches-349 trated by a diverse array of ABPs, including actin nucleation proteins (e.g., Arp2/3 complex, formin), G-350 ABPs (e.g., profilin), elongation-promoting factor (Ena/VASP), crossing proteins (e.g., fimbrin), and actin 351 depolymerization proteins (e.g., ADF, cofilin) 72-74 . Emerging evidence shows that inter-or intramolecular 352 interactions of ABPs, including homo-and heterooligomeric interactions and lower-order to higher-order 353 molecular assembly, are fundamental mechanisms in dynamic regulation of the biochemical activities of 354 ABPs, such as during signal transduction, and thereby actin cytoskeleton polymerization and organization 355 63,75-97 . For example, during plant immune activation, intermolecular interactions between plant formin di-356 mers on the plasma membrane enhance formin activities in actin nucleation, which plays a critical role in 357 remodeling the actin cytoskeleton of plant cells upon bacterial infection 63,87,93 . In mammals,multivalent 358 interactions between WASP and the Arp2/3 complex also activate actin nucleation during T-cell signal 359 transduction 94-96 . Mammalian Ena and VASP are also known to cluster as tetrameters to enhance actin 360 elongation activity by creating a tetravalent G-actin-binding (GAB) domain through a CC domain 78-80 . 361 After nucleation, F-actin stabilization also requires lower-order protein oligomerization, such as CC-medi-362 ated dimeric interactions between tropomyosins, which stabilize F-actin and exclude the binding of other 363 ABPs to F-actin 81-83 . In addition, dimerization of vinculin through its C-terminal tail is critical for F-actin 364 filament bundling and engages multiple binding partners at focal adhesion sites [84][85][86]97 . This oligomerization 365 state-dependent regulation of ABPs through interactions with different binding partners in an oligomeriza-366 tion state-dependent manner suggests sophisticated regulatory mechanisms underlying actin remodeling. 367 The coronin C-terminal domain is predicted to form an oligomeric α-helical CC 10,36,38,[40][41][42]60 . Here, we 368 reported the importance of coronin protein oligomerization for fine-tuning its inhibition in Arp2/3 complex-369 mediated actin nucleation and Crn1-mediated actin bundling, in which the optimal tetrameric state of yeast 370 Crn1 was maintained through coupling the CC domain with an IDR. Because it lacks a well-defined hydro-371 phobic residue pairing pattern within the heptad repeats 69-71,98,99 , Crn1-CC exhibited a heterogenic higher-372 order oligomeric state (n> 10) on its own, but Crn1-CC could be optimized by the IDR to finetune the 373 oligomerization status towards a better functional tetramer. Precise characterization of protein oligomer 374 status in vivo is technically challenging. 375 13 IDR tunes the higher-order assembly and functionality of coronins with evolutionary selection 376 Based on the number of N-terminal β-propeller domain, coronins were initially divided into short and 377 long coronins 5 . Short coronins are subdivided into type I (e.g., Coronin 1A, 1B and 1C) and type II (e.g., 378 Coronin 2A and Coronin 2B) coronins in metazoans and an 'unclassified' class in nonmetazoans (e.g., 379 Crn1 of S. cerevisiae, coronin of Dictyostelium) 66 , whereas long coronins are also classified as type III 380 (e.g., POD1) coronins. Short coronins are universal, and coronin homologs contain three domains: the N-381 terminal structural β-propeller domain, which contains five WD-repeat motifs 100-102 ; the 36,38,[40][41][42]60 ; and a region between the β-propeller domain and CC domain that was named the 383 "unique region (UR)" due to its large variations in length and sequence along evolution, the signatures of 384 the IDR 103 . While the IDRs of S. cerevisiae Crn1 and D. melanogaster Dpod1 have shown microtubule-385 binding abilities 29,44,104 , other functions of the IDR remain poorly understood. Here, we carried out a fo-386 cused analysis of 392 short coronins from 200 species. Our results revealed a previously unknown func-387 tion of the IDR, which maintains the optimal tetrameric state of Crn1 by suppressing higher-order CC do-388 main oligomerization. IDR-containing proteins (IDPs) work differently than fully structured proteins 389 (e.g., molecular recognition and assembly, protein modification, and spacing chains) 105 . With multivalent 390 interactions, some IDRs can regulate homo-and heterooligomeric interactions, particularly for dynamic 391 ensembles of macromolecular assemblies 106,107 . We have recently summarized a unique feature of the 392 IDR of endocytosis proteins, in which we observed a striking correlation between IDR length and tempo-393 rally regulated protein recruitment for endocytosis progression. Endocytic proteins that arrived earlier 394 have a longer IDR than the later actin polymerization-and scission-related proteins 103 , suggesting poten-395 tial evolutionary selection of IDR length depending on functional necessity. Recently, mammalian Eps15 396 and Fcho1/2, early proteins that arrive at the endocytic site, were found to undergo liquid phase separation 397 through weak interactions to optimize endocytosis initiation due to the presence of long IDRs 108 . How-398 ever, the functions of IDRs are diverse and do not necessary facilitate higher-order assembly. For exam-399 ple, certain IDRs may function only through conformational switching between static and disorder states 400 [109][110][111] . Several ABPs are generally well folded and contain relatively shorter IDRs or loops, such as the 401 capping protein fimbrin, the Arp2/3 complex, and cofilin 103 . A 27-aa IDR of yeast Sac6 regulates confor-402 mational flexibility between the N-terminal EF-hand domain and actin-binding domain 1 (ABD1) in a 403 phosphorylation-dependent manner 112,113 . Although the IDRs of fimbrin homologs vary in sequence and 404 length, the presence of an IDR and phosphorylation site is evolutionarily conserved among eukaryotic 405 species 112 . Here, we report another previously unknown mechanism of IDR in yeast Crn1, which opti-406 mizes CC domain-mediated oligomerization to maintain appropriate Crn1 function in F-actin crosslinking 407 and inhibition of the Arp2/3 complex. Using de novo-engineered Crn1 homo-oligomers spanning from a 408 14 dimer to a pentamer, both genetic interaction experiments in vivo and actin-based biochemical experi-409 ments in vitro, including F-actin crosslinking and inhibition of Arp2/3 complex nucleation, demonstrated 410 that the functions of Crn1 in the tetrameric state are optimal. 411 The CC is a common motif in many IDPs, and IDRs are often located next to the protein oligomerization 412 domain, such as CC or prion-prone domains [114][115][116] . The CC domain contains multiple seven-residue heptad 413 motifs and packs into parallel or antiparallel helical bundles through hydrophobic interactions 57,69,70 . The 414 stability of α-helix assembly is the balanced result of interhelical interactions and depends on multiple 415 determinants, including hydrophobic residues at the 'a' and 'd' positions, which constitute the hydrophobic 416 core. In addition, diverse packing geometries that do not necessarily follow classic heptad patterns exist 69-417 71,117-125 . We observed that the hydrophobic interaction between coronin CC helices is the dominant factor 418 that contributes to the interhelical interactions and CC domain oligomerization. Crn1-CC contains two non-419 hydrophobic residues at the 'a' and 'd' positions that generate nonideal heptad repeats ( ing long IDR can modulate the energy landscape of helix packing to prevent higher-order assemblies. Most 423 likely, coronins in higher species (e.g., mouse, human), in contrast, have evolved ideal hydrophobic resi-424 dues within the heptad, which exhibit robust interhelical interactions without needing a long IDR to main-425 tain an optimal energy level for a low-level oligomeric state. The evolutionarily correlated selection of IDR 426 length in the coronin protein family is striking. For an ideal helical assembly such as the CC domain of 427 murine coronin1A, which contains optimal interacting heptad pairs, length is not critical for modulating the 428 energy landscape of the CC domain. In contrast, the situation with nonideal helical assemblies, such as 429 Crn1-CC, is different. The profound evolution of coronin IDRs with varied lengths is partially driven by 430 how well the neighboring helical chains are packed. Of course, exhaustive tests could not be performed. 431 Therefore, we cannot exclude other factors, such as amino acid composition, that may contribute to evolu-432 tionary selection of the IDR based on its role in regulating CC domain oligomerization. Nevertheless, the 433 IDRs of yeast Crn1, human Coronin 1A, and Coronin 1C are indispensable for their functional association 434 with and regulation of the actin cytoskeleton, which at least provides conformational flexibility between 435 the N-terminal β-propeller and C-terminal CC domains. Our future endeavors will focus on identifying the 436 diverse regulatory mechanisms of the IDR, which are molecular grammar-and sequence-dependent. 437

RESOURCE AVAILABILITY 646
Further information and requests for resources and reagents should be directed to and will be fulfilled by 647 Yansong Miao (yansongm@ntu.edu.sg). 648

Materials availability 649
Yeast strains and plasmids created in this study are available upon request from the Lead Contact. and Coronin 1C plasmids, respectively. MEF cells were transiently transfected with plasmids above using 664 Lipofectamine 3000 (Invitrogen, USA) following the manufacturer's protocol and grown overnight on a 665 round glass-bottom dish (Thermo Fisher) at 37°C in a CO2 incubator for protein expression. Transfected 666 cells were identified by GFP fluorescence under microscope. 667

Yeast live-cell imaging 702
Yeast strains were cultured in the SC liquid medium with 2% dextrose without tryptophan overnight and 703 re-inoculated into a new medium to a staring OD600=0.2. Cells were allowed to grow for additional 4 hours 704 before imaging. Cells were immobilized on concanavalin A (1mg/ml, Sigma)-coated circular coverslip 705 (Marienfeld Superior) and imaged at 25°C by a wide-field microscope Leica Dmi8 (Leica Microsystems) 706 equipped with ORCA-Flash 4.0 LT scientific CMOS camera (Hamamatsu Photonics, Japan) and Leica 707 ×100 oil immersion objective lens (NA 1.4) using Metamorph software (Molecular Devices). Images were 708 acquired as a z-axis stack with a step size of 0.25µm for a total of 31 frames. Middle focal panel images 709 were used for representatives and intensity analysis. 710 28

Yeast growth assay 711
For the yeast spotting assay, each strain was inoculated into YPD liquid medium and cultured at 25°C for 712 overnight. The next day, the statured culture was inoculated to a fresh YPD medium with a starting 713 OD600=0.2 and cultured for additional 3 hours. An additional reinoculation was performed starting from 714 OD600=0.2 and cultured for another 3 hours before being diluted to OD600=0.1 for spotting assay with ten-715 fold serial dilutions in YPD medium. 4 μL culture of each dilution was spotted on the YPD agar plate. 716 Plates were incubated at the tested temperatures for 48 hours and scanned by a Perfection V600 Photo 717 scanner (Epson) with 600 dpi. Images were converted to grayscale and inverted in Photoshop. 718

Mammalian cell imaging 738
To image subcellular localization of GFP-tagged Coronin1A or Coronin1C protein variants and filamentous 739 F-actin in MEF cells, cells were washed twice with prewarmed phosphate-buffered saline buffer (PBS, pH 740 7.4), and fixed with 4% paraformaldehyde (PFA) solution in PBS for 20 min at room temperature, and fixed 741 29 cells were permeabilized with 0.1% Triton X-100 (Bio-Rad) in PBS for 5min. After washing twice with 742 PBS, Alexa Fluor™ 647 Phalloidin (Invitrogen) (1:40 dilution in PBS with 1% bovine serum albumin) was 743 applied to cells for 20 min at room temperature in the dark area, and cells were washed three times with 744 PBS. Samples were imaged by a wide-field microscope Leica DMi8 (Leica Microsystems) equipped with 745 ORCA-Flash 4.0 LT scientific CMOS camera (Hamamatsu Photonics, Japan) with Leica ×100 oil immer-746 sion objective lens (NA 1.4) controlled by Metamorph software (Molecular Devices). Z-axis scanning im-747 ages were acquired with a step size of 0.25 µm for a total of 31 frames. An image of the middle focal plane 748 was shown as the representative image. Images were deconvolved by Huygens Essential deconvolution 749 software (Scientific Volume Imaging, Netherlands). 750

Rabbit skeletal muscle actin (RMA) purification and labeling 784
To obtain monomeric ATP-bound RMA for actin cosedimentation assay, transmission electron microscopy, 785 and TIRF microscopy. 2g rabbit muscle acetone powder (Pel-Freez, LLC) was dissolved in a 200 mL cold 786 G-buffer (2 mM Tris, pH 8.0, 0.2 mM ATP, 0.5 mM DTT and 0.1 mM CaCl2) and stirred at 4°C overnight. 787 The mixture was filtered with a cheese cloth to remove muscle powder, then the actin-dissolved solution 788 was further centrifuged at 2600 x g, 4°C (Type 45 Ti rotor, Beckman Coulter) for 30 min to collect super-789 natant. Actin in supernatant was then polymerized with slowly stirring for 1 hour at 4°C by adding KCl and 790 MgCl2 solution to a final concentration of 50 mM and 2 mM, respectively. To remove tropomyosin and 791 other actin binding proteins, fine KCl powder was slowly added to reach a final concentration of 0.6 M and 792 the solution was kept stirring for another 30 min. The solution was then centrifuged at 14,000 x g (Type 45  793 Ti rotor, Beckman Coulter) for 3 hours at 4°C to collect the filamentous actin pellet. Pellet was then rinsed 794 with cold G-buffer and homogenized with a homogenizer in 7 mL of cold G-buffer followed by a short 795 time of sonication. The sample was then dialyzed in 2 L G-buffer at 4°C for 48 hours to induce depolymer-796 ization (G-buffer was changed every 12 hours). After buffer exchange, the sample was centrifuged at 797 167,000 x g (SW 55 Ti swinging-bucket rotor, Beckman Coulter) at 4°C for 2.5 hours, 5 mL of supernatant 798 was collected and loaded to a Sephacryl S-300 HR column (GE healthcare) pre-balanced with G-buffer. 799 Peak fractions were collected and combined, then 0.01% (final) sodium azide (Sigma) was added to inhibit 800 fungi contamination and kept at 4°C. Actin concentration was measured by reading OD290 with a Nanodrop 801 2000 (Thermo Scientific). 802 31 To label actin with Oregon Green™ 488 Iodoacetamide (Invitrogen) or NHS-dPEG®4-biotin (Sigma), the 803 same steps were followed as RMA purification until pelleted filamentous actin was homogenized and soni-804 cated. After this, the sample was dialyzed in a 1 L G-buffer at 4°C for overnight. The next day, the sample 805 was changed to 1 L G-buffer without DTT and dialyzed for 4 hours at 4°C (buffer changed once). Oregon 806 Green™ 488 Iodoacetamide or NHS-dPEG®4-biotin was dissolved in dimethylformamide to a final con-807 centration of 10 mM. Before labeling, actin concentration was measured by reading OD290 with a Nanodrop 808 2000 (Thermo Scientific). Actin was firstly diluted with an equal volume of cold 2X labeling buffer (50 809 mM Imidazole, pH 7.5, 200 mM KCl, 0.6 mM ATP and 4 mM MgCl2) and further diluted to final 23 µM 810 with cold 1X labeling buffer, then 10-fold molar excess of Oregon Green™ 488 Iodoacetamide or NHS-811 dPEG®4-biotin was added dropwise while very gently vortexing. The mixture was covered with aluminum 812 foil and rotated at 4°C for overnight. The next morning, labeled filamentous actin was centrifuged at 813 167,000 x g (Type 50.2 rotor, Beckman Coulter) for 3 hours at 4°C. Pellets were collected and homogenized 814 in 4 mL G-buffer, waited on ice for 1 hour, and homogenized again. Actin was then dialyzed in 1 L G-815 buffer at 4°C for 48 hours to induce depolymerization (dark, G-buffer changed every 12 hours). After buffer 816 exchange, actin was centrifuged at 436,000 x g (TLA100 rotor, Beckman Coulter) at 4°C for 1 hour. The 817 supernatant was collected and further purified by Sephacryl S-300 HR column (GE healthcare) pre-bal-818 anced with G-buffer. Peak fractions were collected and combined, then dialyzed in a 500 mL G-buffer with 819 50% (v/v) glycerol at 4°C overnight to reduce volume. Small aliquots were frozen in liquid nitrogen and 820 stored in a -80°C freezer. 821

Yeast protein expression and purification 822
To generate plasmid for overexpression C-terminal 6xHis-tagged yCrn1-FL protein in budding yeasts, 823 DNA sequences encoding Crn1 full length was amplified using pRS305-Crn1-FL-mRuby2-4*myc±500 as 824 a template and inserted into the pYeast Pro vector (pGAL-ORF-3×StreptagII-9×His) 112 . Plasmids were 825 transformed into strain YMY2043 112 using lithium acetate, and yeasts were spread on an SC agar plate with 826 2% dextrose without uracil to select positive transformants. 827 To obtain large-scale yeast culture for protein purification, we followed a published protocol 62 , 10-20 pos-828 itive transformants were inoculated into 10 mL SC medium with 2% raffinose without leucine and grow at 829 30°C with vigorous shaking for around 2 days until saturation (OD600 2 to 3). Then statured culture was 830 scaled to 100 mL using SC medium with 2% raffinose (MP biomedicals) without leucine and grew at 30°C 831 with vigorous shaking until saturation again. The culture was then transferred to 1.9 L of fresh SC medium 832 with 2% raffinose without leucine in a 5-L flask and kept growing at 30°C. When the culture reached 833 32 saturation after 36 to 48 hours, 160 mL of 30% (w/v) galactose (Sigma) and 240 mL of 10X YP (10% w/v 834 yeast extract, 20% w/v peptone) were added to the flask making the final concentration of 2% galactose 835 and 1X YP, in which protein was induced to express for 12 to 16 hours with vigorous shaking at 30°C. 836 Yeasts were collected by centrifugation at 6,000 x g (rotor JA10, Beckman Coulter) for 15 min at 4°C, and 837 washed twice with sterile double-distilled water. Cells were then resuspended with 20% volume of sterile, 838 double-distilled water, the mixture was frozen into small balls by dripping into liquid nitrogen, which was Healthcare). After affinity binding, the column was washed with binding buffer followed by gradient elu-847 tion from 20 mM to 500 mM Imidazole using elution buffer. Peak fractions were collected and checked by 848 SDS-PAGE, fractions were combined and dialyzed in 2 L protein buffer (50 mM Tris, pH 8.0, 150 mM 849 NaCl) at 4°C for overnight, then concentrated with 50-kDa cut-off concentrators (Merk Millipore). Protein 850 concentration was measured by the BCA protein assay kit (ThermoFisher). For storage, the protein was 851 aliquoted into a small volume, frozen in liquid N2, and stored in -80°C freezer. 852

Specimen preparation and transmission electron microscopy 853
Purified RMA was polymerized for 1 hour in F-buffer (2 mM Tris, pH 8.0, 50 mM KCl, 1 mM MgCl2, 1 854 mM EGTA, 0.2 mM ATP, 0.5 mM DTT, and 0.1 mM CaCl2). 1 µM of each purified different Crn1 protein 855 variant was incubated with 2 µM of pre-polymerized F-actin for 30 min. Samples were then applied to glow 856 discharged carbon-coated copper grid (200 mess, Electron Microscopy Sciences) and waited for 2 min. 857 Extra volume was removed by filter paper, and grids were negatively stained with 1% uranyl acetate (Elec-858 tron Microscopy Sciences) for 1 min. Grids were air-dried and examined at 120kV by an FEI Tecnai 12 859 TEM equipped with an Ultrascan 1000 CCD camera (Gatan, Inc). 860

Far-UV circular dichroism 861
Purified recombinant Crn1-CC and Crn1-IDR proteins were dialyzed against 50 mM sodium phosphate 862 buffer (pH 7.4, 37.7 mM Na2HPO4, 12.3 mM NaH2PO4) at 4°C for overnight. Protein was diluted to a final 863 concentration within the optimal range for the detector. 300 µL of the sample was loaded into a 1 mm path 864 33 length quartz cuvette (Hellma Analytics), spectra were recorded on a Chirascan TM Circular Dichroism Spec-865 trometer (Applied Photophysics) equipped with a temperature controller at 20°C and supplied with constant 866 N2 flushing. Spectra were acquired from 260 nm to 190 nm (step size: 1 nm) with an integration time of 867 1sec at each wavelength, and baseline was corrected by a sample with buffer alone. Raw data in machine 868 units θ (mdeg) were converted to Mean Residue Ellipticity [θ] (degrees cm 2 dmol -1 residue -1 ) by equation: 869 [θ] = θ x (0.1 x MRW)/(P x conc), where MRW=protein weight (daltons) / number of residues, P=patch 870 length (cm) and conc=protein concentration (mg/ml). 871

Coronin sequence analysis and the taxonomic tree 923
The murine Coronin1A (NCBI accession NP_034028.1) N-terminal sequence (1-402aa) was input as a 924 query sequence to NCBI blastp against non-redundant protein sequence database to identify coronin hom-925 ologues for the 200 species that contain coronin family members, which were previously reported in 66 . 926 35 We firstly excluded the "hypothetical/unnamed/predicted/uncharacterized/putative" candidates. In the re-927 maining list of coronin candidates, we defined a cut-off for "total alignment score" of 350, allowing us to 928 identify the coronin homologues containing around N-terminal 400aa, which is the core and conserved 929 coronin N-terminal domain. Thereby, we have identified 392 coronin homologues from 200 species. For 930 each identified coronin homologue, DeepCoil2 (https://toolkit.tuebingen.mpg.de/tools/deepcoil2) 54 was 931 used to predict coiled coil (CC) region, from which 11 (one from B.bigemina, one from P.berghei, one from 932 P.chabaudi, one from P.knowlesi, one from P.reichenowi, one from P.Vivax, one from P.yoelii, one from 933 T.annulate, one from A.aegypti, and two from M.lucifugus) out of 392 coronin homologues do not have 934 predicted CC. The unique regions (IDR) were identified between N-terminal and CC which were used to 935 generate IDR-MAP. To build the taxonomic tree of the 200 species, taxid of each species was input to 936 NCBI taxonomy tool-common tree, and generated tree was modified by iTOL (https://itol.embl.de/) 126 . 937 The length of the predicted IDR was then integrated as a bar diagram outside the taxonomy tree, resulting 938 in an IDR-MAP. 939

Coarse-grained (CG) model 941
The CG model represents every amino acid residue with one bead at its Cα atom position, following Ra- where is the van der Waals radius of residue i 142 . 975

Langevin Dynamics (LD) Simulation 976
The CG model was implemented for Langevin dynamics (LD) simulations with the GROMACS 5.1.2 pack-977 age 143 . All simulations were carried out in the NVT ensemble (constant atom number, simulation box size, 978 and temperature). Simulations were performed at 300K with a friction coefficient of 50/ps. A simulation 979 time step of 0.01ps was used. At least 1000 ns LD runs were conducted. Structures for Crn1 coiled-coil and 980 IDRs were modeled by the software spdbv (http://www.expasy.org/spdbv/). The structure for Coro1A 981 coiled-coil was from Protein Data Bank (PDB ID: 2akf). The initial structure for each simulation was con-982 structed with PyMOL 144 . 983

Data analysis 984
The potential energy per chain was used for evaluating the stability of different oligomers, and energy 985 contributions from electrostatic and hydrophobic interactions were further analyzed. For a certain kind of 986 species, e.g., murine Coronin1A coiled-coil (MmCC), its monomer was taken as the reference state whose 987 potential energy per chain was shifted to zero. 988 All the contact maps followed the workflow demonstrated in Figure  the contact between the regions where each wholesome motif locates at two chains. Each wholesome motif 992 was predicted from the sequence according to the coiled-coil packing pattern. Each motif-averaged contact 993 map, a 7×7 matrix also, is the mean matrix for all the motif boxes. The motif-averaged contact map repre-994 sents the average contact probability between different pairs of motif sites. As none of the contact proba-995 bility values are larger than 0.5, the maximum value on the scale of contact probability is changed from 0 996 to 0.5 for better contrast. Following the approach of Ryan et al. 145 , the inter-residue distance cut-off for a 997 contact to be formed is as defined above. In addition to the original motif-averaged probability map, a 998 motif contact difference map (Figure 7-figure supplement 2E, F; Figure 7I and J) was constructed to present 999 the contact probability differences at different contact pairs for two proteins (e.g. ScIDR-MmCC and 1000 MmCC in Figure 7-figure supplement 3B). The motif contact difference map indicates the difference in 1001 each contact pair between protein A and protein B, denoted as "motif contact difference map (A minus B)". 1002 Finally, the Euclidean distance was calculated as the score to measure the difference between two contact 1003 probability maps. 1004 38 All the analysis was based on the last 500 ns of each simulation. GROMACS tools and MATLAB (v. 1005 R2020b; The MathWorks, Natick, MA) were used to analyze simulation trajectories. Protein 3D structures 1006 were rendered by PyMOL. 1007

mRuby2-patch intensity analysis 1009
As illustrated in Figure 3-figure supplement 1A, to measure mRuby2 signal intensity at the actin patch, a 1010 rectangular box at 36x8 pixels was drawn crossing the Crn1-localized patches. The box was placed perpen-1011 dicularly to the mother cell membrane, with a patch in the middle. Cytosol region was defined by proximal 1012 8x8 box within the cell, patch region was defined by 8x8 box in the middle, and background region was 1013 defined by distal 8x8 box outside of the cell. The average intensity of each box was measured as signal 1014 intensity, termed as cytosol intensity (C), patch intensity (P), and background intensity (B), separately. 1015 Patch to total ratio was calculated use the equation: (P-B)/ [(P-B) + (C-B)]. 1016

Actin cosedimentation analysis 1017
The SDS-PAGE gel was scanned by the scanner and converted to an 8-bit gray value image in ImageJ. A 1018 rectangle area covering the entire band was selected, area size and a mean gray value (mgv) were obtained 1019 from ImageJ, whereas a same-sized rectangle box from the background area was also selected for normal-1020 ization. Band density was calculated as area x [(255-sample mgv) -(255-background mgv)]. F-actin bound 1021 Crn1 protein variants in high-speed actin cosedimentation assay were analyzed as previously described 146 . 1022 To calibrate measured signal intensity by protein quantity, we firstly generated a standard curve by plotting 1023 the band density versus protein mass after measuring the gel of the total fraction. The measured band density 1024 of F-actin bound Crn1 from pellet fraction was converted to Crn1 amount (mass) using the standard curve 1025 after subtracting the Crn1 band density in the control experiments without F-actin. Data points were curve 1026 fitted using the Hill equation in GraphPad Prism 9. To analyze actin bundling in low-speed cosedimentation 1027 assay, both pellet and supernatant fractions of actin were used for analysis: at each Crn1 concentration, 1028 actin band densities from the pellet (P) and supernatant (S) were measured, percentage of actin in the pellet 1029 (%) was calculated as P/(P+S). 1030
D 75kDa 100kDa C r n 1 Δ C C -T r i C r n 1 Δ C C -T e t C r n 1 Δ C C -P e n t C r n 1 Δ C C -D i  Representative images of the minor population of MEF cells overexpressing full-length Coronin 1A-GFP, Coronin 1C-GFP, and empty vector control (total 1A-FL, n=78; 1C-FL, n=33; vector, n=21 cells). Zoomed images were generated from white dashed boxes. Scale bar: 20 µm (left), 5 µm (zoom).     Figure 7H varies with number of helices in coarse grained simulations. Potential energy per chain was decomposed into hydrophobic term and electrostatic term for analysis of contribution. (D) Motif-averaged contact maps between heptads (abcdefg) in CC region from two adjacent chains of MmCC (PDB ID:2akf), and three recombinants of MmCC and IDRs: MmIDR-MmCC, ScSIDR-MmCC, and ScIDR-MmCC, which derived Figure  7I and (E-F) below. (E-F) Motif contact difference map between MmIDR-MmCC and MmCC (E), and between ScSIDR-MmCC and MmCC (F). Red and blue in motif contact difference map indicate increase and decrease in contact probability, respectively. (G) Motif-averaged contact maps between heptads (abcdefg) in CC region from two adjacent chains of Crn1-CC and Crn1-ΔN, which derived Figure 7J.  Step 1 Step 2 Step 3