Acetyl-CoA carboxylase catalyses the ATP-dependent carboxylation of acetyl-CoA, a rate-limiting step in fatty acid biosynthesis1,2. Eukaryotic acetyl-CoA carboxylases are large, homodimeric multienzymes. Human acetyl-CoA carboxylase occurs in two isoforms: the metabolic, cytosolic ACC1, and ACC2, which is anchored to the outer mitochondrial membrane and controls fatty acid β-oxidation1,3. ACC1 is regulated by a complex interplay of phosphorylation, binding of allosteric regulators and protein–protein interactions, which is further linked to filament formation1,4,5,6,7,8. These filaments were discovered in vitro and in vivo 50 years ago7,9,10, but the structural basis of ACC1 polymerization and regulation remains unknown. Here, we identify distinct activated and inhibited ACC1 filament forms. We obtained cryo-electron microscopy structures of an activated filament that is allosterically induced by citrate (ACC–citrate), and an inactivated filament form that results from binding of the BRCT domains of the breast cancer type 1 susceptibility protein (BRCA1). While non-polymeric ACC1 is highly dynamic, filament formation locks ACC1 into different catalytically competent or incompetent conformational states. This unique mechanism of enzyme regulation via large-scale conformational changes observed in ACC1 has potential uses in engineering of switchable biosynthetic systems. Dissecting the regulation of acetyl-CoA carboxylase opens new paths towards counteracting upregulation of fatty acid biosynthesis in disease.
Access optionsAccess options
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank T. Sharpe at the Biophysics facility, A. Schmidt at the Proteomics Core Facility, and the Imaging Core Facility, in particular A. Ferrand, of the Biozentrum Basel for protein characterization and imaging support, and EMBL Heidelberg for providing the pETG-10A vector. We thank sciCORE at University of Basel for support with high performance computing. A.H. is supported by a Fellowship for Excellence from the Biozentrum Basel International PhD program. M.H. was supported by a Novartis Excellence Fellowship. This work was supported by Swiss National Science Foundation grants 138262, 159696 and 164074.Reviewer information
Nature thanks R. Haselkorn, J. Kollman, M. St. Maurice and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
Negative stain electron microscopy micrographs of phosphorylated and dephosphorylated ACC in presence of citrate and BRCT. Addition of citrate to dephosphorylated ACC induces ACC–citrate filament formation, whereas when added to phosphorylated ACC, citrate results in aberrant ACC filament and ring formation. Addition of BRCT domains to phosphorylated ACC induces formation of ACC–BRCT filaments, whereas no effect can be observed when adding BRCT domains to dephosphorylated ACC, and ACC remains in its dimeric form. In the absence of citrate or BRCT domains, phosphorylated and dephosphorylated ACC are in the dimeric, flexible form. Scale is identical across all images.
Extended Data Fig. 2 Processing of electron microscopy data for ACC–citrate filaments and validation.
a, Flowchart of data processing showing initial and optimized raw micrographs, 2D classes, 3D classes and refinement. Initial cryo-EM grids showed a meshwork of ACC–citrate filaments, exemplifying their flexible nature. After optimization, ACC–citrate filaments attach to carbon and protrude into holes. Some interaction between filaments can still be seen at the edge of the holes; however, single ACC–citrate filaments can clearly be recognized. 2D classification and ab initio reconstruction were done in cryoSPARC, all other steps of processing were conducted in Relion. b, Overview of map quality for the BT, CT and CDN domains. Protein is shown in colour, according to the scheme in Fig. 1a, with transparent electron microscopy map. c, FSC curves for masked and unmasked as well high-resolution phase-randomized reconstructions and final corrected FSC curve. Map coloured according to local resolution, colour scale is provided. All electron microscopy maps are shown at contour level of 0.0172.
a, Flowchart of data processing showing a raw micrograph, 2D classes, 3D classes and refinement. b, Overview of map quality for the CDC2, CT and CDN domains. Protein is shown in colour according to scheme in Fig. 1a, with transparent electron microscopy map. Maps are shown at contour level of 0.009. c, FSC curves for masked and unmasked as well high-resolution phase-randomized reconstructions and final corrected FSC curve. Maps coloured according to local resolution, colour scale is provided. Electron microscopy maps are shown at contour level of 0.019 and 0.105 for mask 1 and mask 2, respectively.
a, Overview map of ACC–citrate filament obtained after post processing in Relion (this map type is used for the other figures throughout the manuscript) at an overall resolution of 5.4 Å judged by FSC using the 0.143 threshold criterion. Labels indicate local resolution estimates for a box size of 40 pixel around the indicated positions, calculated using localfsc in Chimera using either the unfiltered half-maps directly (left value) or the unfiltered half-maps after local averaging as input (right value). Local resolution differentially improves after local averaging. Map is shown at contour level of 0.0172. b, Map around a helix in the CT domain in the ACC–citrate filament with density of unfiltered half-map as obtained directly from refinement in Relion. c, Image of the same helix in the CT domain with density of the same unfiltered half-map but after local averaging for the CT domain. Here, additional B factor sharpening by −120 Å−2 was applied before local averaging for visualizing additional detail (additional sharpening was not applied in a for localfsc comparison). Maps in b and c are shown at contour level of 0.012. d, Overview map of ACC–BRCT filament, resolution values as obtained in a are indicated. Map is shown at contour level of 0.019. e, Map around a helix in the CT domain in the ACC–BRCT filament with density of unfiltered half-map as obtained directly from refinement in Relion. f, Image of the same helix in the CT domain with density of the same unfiltered half-map but after local averaging for the CT domain. Here, additional B factor sharpening by −40 Å−2 was applied before local averaging for visualizing additional detail (additional sharpening was not applied in d for localfsc comparison). Maps in d and f are shown at contour level of 0.013.
Extended Data Fig. 5 Conformational variability of ACC dimers and density of the covalently linked biotin cofactor.
a, Negative stain electron microscopy 2D class averages of dephosphorylated human ACC in the absence of citrate show the protein in a variety of conformations without considerable populations of closed dimers. b, Negative stain image of ACC–citrate filaments obtained from dephosphorylated ACC in a buffer with 10 mM citrate. Arrows indicate rarely observed, residual, non-polymerized ACC (further analysed in c). c, Negative stain electron microscopy 2D class averages of residual, non-polymerized ACC dimers observed in the presence of citrate on micrographs of polymerized ACC–citrate filaments (see b). ACC-like classes are marked in red. A variety of elongated conformations can be observed. d, Streptavidin (SA) shift assay to determine biotinylation level of ACC. ACC and streptavidin were mixed in different ratios and the shift upon SDS-resistant binding of streptavidin to biotin was observed via SDS–PAGE. At higher excess of streptavidin, no unbound ACC is observed, indicating complete biotinylation of ACC. Two degradation products of ACC (indicated by asterisks) are observed, one of which also shows a band shift (*-SA) and thus contains the biotinylated site. Owing to the tetrameric nature of streptavidin, higher-order complexes are formed, which can also be observed on the gel. An uncropped image of the gel is shown in Supplementary Fig. 1. e, f, Density of the covalently linked biotin cofactor in the two active sites of the CT domain dimer. The main chain Cα position of the biotinylated lysine (residue 786) is indicated. Due to limited resolution, the cofactor was not modelled; its orientation is shown schematically. For clarity, parts of BCCP are not shown. The map is shown at contour level of 0.0238.
a, The intermolecular interface in ACC–citrate filaments is shown in cartoon representation with the transparent electron microscopy map shown in grey. Local two-fold symmetry is indicated, and domains of the lower dimer are labelled. The map is shown at contour level of 0.0189. b, Close-up of the interface as indicated in a. ACC–citrate is shown in colour as cartoon representation. ScACC is shown as a cartoon in grey and superimposed for one side of the interface. For the other side, an additional surface representation is shown for ACC–citrate. The loop between Nα4 and Nα5 of the four-helix bundle of ACC–citrate CDN domain binds in the cradle formed by Lα2, Lα4 and β1. This loop is substantially shorter in ScACC, demonstrating incompatibility of ScACC with formation of the interface. c, Same depiction as in b, but the surface of the top ACC–citrate dimer is shown and ScACC, shown in grey, is superimposed on the bottom dimer. The extended loop in ScACC between Lα1 and Lα2 is not compatible with filament formation. d, Alignment of 20 ACC CD sequences of metazoan and fungal organisms. Residue numbers according to human ACC are indicated as well as the helices, the loops and the strand labelled in b and c. Darker colour indicates increased conservation. Pairwise identity over all aligned sequences is 61.5%, pairwise identity over metazoan and fungal sequences is 97.0% and 54.0%, respectively. Accession numbers: Homo sapiens, Q13085; Bos taurus, Q9TTS3; Canis lupus familiaris, E2RL01; Capra hircus, XP_017919660; Danio rerio, F1QH12; Drosophila melanogaster, Q7JV23; Felis catus, XP_011287256; Gallus gallus, P11029; Gorilla gorilla gorilla, XP_018881836; Mus musculus, Q5SWU9; Rattus norvegicus, P11497; Sus scrofa, D2D0D8; Aspergillus nidulans, AN6126.2; Candida glabrata, Q6FKK8; Candida albicans, C4YNG3; Chaetomium thermophilum, G0S3L5; Kluveromyces lactis, Q6CL34; Saccharomyces cerevisiae, Q00955; Schizosaccharomyces pombe, P78820; Trichophyton verrucosum, D4DIV5.
Extended Data Fig. 7 Impact of palmitoyl-CoA addition on ACC–citrate filaments and architecture of ACC–BRCT.
a, Negative stain electron microscopy micrographs of ACC–citrate filaments treated with increasing concentrations of palmitoyl-CoA. At a 1:1 molar ratio of ACC–citrate monomer to palmitoyl-CoA, filaments show no differences to ACC–citrate filaments. At 1:10 molar ratio, ACC–citratepalm filaments are observed. At 1:100 molar ratio, filaments dissolve. b, Top, domain organization of human ACC. Bottom left, enlarged negative stain electron micrograph of ACC–citrate filament with surface representation of the model coloured according to domains. Bottom right, electron micrograph of a ACC–citratepalm filament and interpretation by a plausible model derived from ACC–citrate filaments by disrupting the BC domain dimers and flipping out of the BC domain. c, Surface representation of ACC–BRCT with components of a single node coloured as in Fig. 3a. Domains of three molecules (A, B and B−1) add parts to the node. d, Same view as in c, but the domains are coloured according to the domain colour scheme in b. e, Same view as in c, with the CDC2 domains coloured according to domain colour scheme. These domains constitute the connecting arms between adjacent nodes. f, Surface representation of two consecutive dimers within one helix strand. Left, view without BRCT domains; the phosphosite loops are labelled. Right, view with dimeric BRCT domains establishing the connections between two dimers. g, Enlarged view of the phosphosite loop-BRCT interaction area, illustrating minimal contacts between the two CDC1 domains and between the filament strands and the BRCT domains. The interaction is governed by binding of the phosphosite loop to the dimeric BRCT.
a, Surface representation of dimers of ACC–BRCT, C. thermophilum (Ct) ACC CD–CT, and ACC–citrate, in the same relative orientation and coloured according to the sequence scheme shown below. b, CT-based overlay of the three structures, illustrating the rotations (indicated by arrows) of the CDC2 domains of ACC–citrate and CtACC relative to CDC2 of ACC–BRCT. CD–CT of ACC–BRCT, ACC–citrate and CtACC are shown in full colour, light grey and dark grey, respectively. Helix C2α1 is labelled. c, CDC2-based overlay of the three structures, representing the displacement (indicated) of CDC1 of ACC–citrate and CtACC relative to CDC1 of ACC–BRCT. Colouring as in b. CDL was omitted for clarity, and helix C1α1 is labelled. d, CDL-based overlay of the three structures, illustrating the displacement of the CDN of ACC–citrate and CtACC relative to CDN of ACC–BRCT. Colouring as in b. The four-helix bundle of helices Nα3–Nα6 is labelled. The range of displacements of the ends of the bundle is indicated by arrows. e, BC, BT and CDN domains of ACC–BRCT filament together with the electron microscopy map at a low contour level to visualize less well-ordered regions and to illustrate the placement of the B-domain cap of the BC domain in the clamp-like CDN domain. f, Overlay of CDN, BT and BCCP domains from ACC–BRCT and ACC–citrate, revealing a conserved conformation. g, CDN, BT and BCCP domains in the ACC–BRCT filament are shown together with the electron microscopy map at low contour level to visualize the poorly ordered BCCP domain. Maps in e and g are shown at contour level of 0.007. h, The SEC–MALS elution profiles show the molecular mass (right axis) and the scattering intensity (Rayleigh ratio) at the 90° detector (left axis) of BRCT domains bound to the indicated ACC peptides. Elution for BRCT and the BRCT–ACC_p1 complex correspond to a mostly monomeric species with a dimeric subpopulation in fast equilibrium with an average molecular mass of 31.8 kDa and 34.8 kDa, respectively. The elution of BRCT–ACC_p2 with a molecular mass of 51.1 kDa indicates a strong increase in the population of dimeric BRCT-peptide species.
This file contains Supplementary Fig. 1: Uncropped gel image related to Extended Data Figure 5d. Uncropped gel directly as obtained by scanning shows sizemarker with molecular masses indicated. The cropped region is indicated by a black square in dashed lines. The gel contains three molecular mass markers and two sets of the same samples, the right set contains half of the amount of protein of the left set. The contrast was enhanced for the gel shown in Fig. 5d for better visibility, contrast was applied uniformly over the whole gel.
Overview of the ACC–citrate filament as well as zoom-in onto an individual, excised ACC dimer, which is in a closed, catalytically active form. The protomers in the filament as well as the domains in the dimer are indicated. Colour scheme according to Fig. 1a.
Overview of the ACC–BRCT filament indicating the two-stranded architecture, followed by highlighting of an individual ACC dimer in the filament as well as the BRCT domains. Zoom-in shows an individual excised ACC dimer, which is in the extended form, with bound BRCT dimer. The protomers in the filament as well as the domains in the dimer are indicated. Colour scheme according to Fig. 1a.