Molecular architecture of LSM14 interactions involved in the assembly of mRNA silencing complexes

Abstract

The LSM domain-containing protein LSM14/Rap55 plays a role in mRNA decapping, translational repression, and RNA granule (P-body) assembly. How LSM14 interacts with the mRNA silencing machinery, including the eIF4E-binding protein 4E-T and the DEAD-box helicase DDX6, is poorly understood. Here we report the crystal structure of the LSM domain of LSM14 bound to a highly conserved C-terminal fragment of 4E-T. The 4E-T C-terminus forms a bi-partite motif that wraps around the N-terminal LSM domain of LSM14. We also determined the crystal structure of LSM14 bound to the C-terminal RecA-like domain of DDX6. LSM14 binds DDX6 via a unique non-contiguous motif with distinct directionality as compared to other DDX6-interacting proteins. Together with mutational and proteomic studies, the LSM14-DDX6 structure reveals that LSM14 has adopted a divergent mode of binding DDX6 in order to support the formation of mRNA silencing complexes and P-body assembly.

Keywords: P‐bodies; SLIMs; mRNA decapping; protein–protein interaction networks; translational repression.

Figure 1. Structure of the N‐terminal LSM domain of LSM14 in complex with a conserved C‐terminal fragment of 4E‐T

1. Schematic diagram of LSM14. The N‐terminal LSM domain and FDF, FFD, and TFG motifs are indicated (not to scale).

2. Schematic diagram of full‐length 4E‐T and 4E‐T fragments used for MBP pull‐down experiments in Fig 2B. Coordinates are indicated on the left and right of each fragment. The indicated regions (not to scale) are as follows: eIF4E, binding site for eIF4E; CHD, cup homology domain; LSM14, LSM14‐interacting site.

3. Crystal structure of the N‐terminal LSM domain of human LSM14 in complex with a conserved C‐terminal 4E‐T fragment. The diagram depicts the likely solution structure obtained by combining two fragments of 4E‐T_C that undergo crystallization‐induced domain swapping, as indicated by dashed line. Left: Cartoon representation of the complex. Right: LSM domain of LSM14 shown in surface representation and 4E‐T_C residues critical for complex formation are shown as sticks. Secondary structure elements are labeled according to (D) and (E).

4. Sequence alignment of conserved amino acids within the LSM domains of human (Hs) LSM14, Drosophila melanogaster (Dm) TraI, yeast (Sc) SCD6, Caenorhabditis elegans (Ce) CAR1, and human (Hs) EDC3. Secondary structure elements with corresponding numbering are indicated above the sequence.

5. Sequence alignment of conserved amino acids within the C‐terminal motifs of human (Hs), Xenopus laevis (Xl), zebrafish (Dr), and D. melanogaster (Dm) 4E‐T proteins. Secondary structure elements with corresponding numbering are indicated above the sequence.

Figure EV1. Structural analysis of the LSM14_LSM–4E‐T_C complex

1. Sequence alignment of conserved amino acids within the C‐terminal and middle motifs of human (Hs), Xenopus laevis (Xl), zebrafish (Dr), and Drosophila melanogaster (Dm) 4E‐T proteins.

2. Crystal structure of the N‐terminal LSM domain of LSM14 in complex with a conserved C‐terminal 4E‐T fragment reveals a tetrameric complex with 2:2 stoichiometry. Two perpendicular views shown in cartoon representation. Each LSM14 molecule (blue) is simultaneously bound by two 4E‐T molecules (green).

3. Analysis of purified LSM14_LSM–4E‐T_C complex by size exclusion chromatography coupled to MALS. The molar mass distribution (left ordinate, black line) indicates a molar mass of 12.9 kDa, which corresponds to a 1:1 complex in solution.

4. Structural comparison of the LSM domains of human LSM14 (blue), Drosophila TraI (yellow), and human EDC3 (cyan). The structures were superimposed using the DALI server (Holm & Laakso, 2016) and are shown in identical orientation.

5. Structural comparison of the LSM domains of human LSM14 (blue), human EDC3 (cyan), and human SmD3 (gray, PDB ID: 1D3B‐A). The structures were superimposed using the DALI server (Holm & Laakso, 2016) and are shown in identical orientation.

6. ITC binding isotherms of 500 μM 4E‐T_C peptide (left) and a W958A mutant (right) titrated into 50 μM LSM14_LSM. Data were fitted to a single‐binding site model, and the dissociation constant (K _d) was determined based on three independent experiments. Data analysis, fitting, and K _d calculation were performed using Origin7.

7. Representative Western blot analysis of expressed λNHA‐tagged proteins showing comparable expression of the respective proteins. Proteins were resolved by SDS–PAGE and probed with HA and actin antibodies.

Figure 2. LSM14–4E‐T complex formation is mediated by extensive hydrophobic interactions

1. Close‐up views of the interface between the LSM14 domain (light blue) and 4E‐T peptide (green). Interacting side chains of LSM14 and 4E‐T are shown as sticks and labeled by single letter code. Dashed lines indicate hydrogen‐bonding interactions, and secondary structures elements are numbered as in Fig 1D and E.

2. SDS–PAGE analysis of input (lower image) and MBP pull‐down experiments (upper image) using recombinant MBP‐LSM14 protein (MBP‐LSM14_LSM, residues 1–84) immobilized on amylose‐beads and incubated with recombinant GST‐tagged C‐terminal 4E‐T fragments. MW, molecular weight markers.

3. SDS–PAGE analysis of input (lower image) and pull‐down experiments (upper image) using recombinant MBP‐LSM14 protein (MBP‐LSM14_LSM, residues 1–84) immobilized on amylose‐beads and incubated with recombinant wild‐type or mutant GST‐tagged C‐terminal 4E‐T (4E‐T_C, residues 954–985) protein constructs. MUT5, 4E‐T_C L955A/W958A/F959A/V978A/L981A.

4. SDS–PAGE analysis of pull‐down experiments using recombinant wild‐type or mutant MBP‐LSM14_LSM constructs immobilized on amylose‐beads and incubated with recombinant wild‐type GST‐tagged 4E‐T_C fragment. MBP‐LSM14_LSM variants are indicated above the gel.

5. Immunoprecipitation (IP) of wild‐type and mutant FLAG‐LSM14A proteins from benzonase‐treated HeLa cell extracts using anti‐FLAG antibody. Immunoprecipitated complexes were separated by SDS–PAGE and probed with antibodies against the indicated proteins.

6. Luciferase reporter assays using HeLa cells co‐transfected with plasmids coding for RL‐5BoxB and FL reporters and plasmids expressing λNHA fusions of the wild‐type (WT) LSM14_LSM domain or a Y22E/I29E mutant. A plasmid expressing the silencing domain of GW182 was used as a positive control (Zipprich et al, 2009; Fabian et al, 2011). λNHA‐LacZ served as a negative control. Activity of RL was normalized to expression of FL. Values represent relative RL activities normalized to FL, with expression in the presence of λNHA‐LacZ set as 100%. Values represent means (±SEM) from triplicate experiments.

Source data are available online for this figure.

Figure 3. The LSM14 FDF and TFG motifs interact with the C‐terminal RecA‐like domain of DDX6

1. Schematic diagram of the domain organization in human DDX6. N‐ and C‐terminal RecA (DDX6_N and DDX6_C) domains are labeled as such. Proteins that interact with DDX6_C are denoted.

2. Schematic diagram of the domain organization of human LSM14. LSM14 constructs used for immunoprecipitations and pull‐down experiments in (C, D) are indicated.

3. Immunoprecipitation (IP) of wild‐type and mutant FLAG‐LSM14A proteins from benzonase‐treated HeLa cell extracts using anti‐FLAG antibody. Immunoprecipitated complexes were separated by SDS–PAGE and probed with antibodies against indicated proteins.

4. SDS–PAGE analysis of input (lower) and pull‐down experiments (upper) using recombinant wild‐type MBP‐DDX6_C immobilized on amylose‐beads and incubated with indicated GST‐tagged LSM14 fragments as described in (B).

Source data are available online for this figure.

Figure 4. Structure of the C‐terminal RecA‐like domain of DDX6 in complex with a LSM14 fragment

1. Sequence alignment of amino acids within a region spanning the FDF, FFD, and TFG motifs in human (Hs) LSM14, Drosophila melanogaster (Dm) TraI, Caenorhabditis elegans (Ce) CAR1, and yeast (Sc) SCD6. Secondary structure elements with corresponding numbering are indicated above the sequence. Invariant residues are colored dark blue, while conservative substitutions are depicted in shades of light blue.

2. Left: Crystal structure of the DDX6_C (orange) in complex with a LSM14 (light blue) fragment containing the FDF, FFD, and TFG motifs. The unstructured region of the LSM14 encompassing the FFD motif is denoted with blue dashed line. Secondary structure elements are labeled according to (A), and motifs involved in interaction are indicated. Right: Surface representation of DDX6_C. The bound LSM14 peptide (light blue) is shown as a cartoon representation, and residues from LSM14 mediating interaction with DDX6_C are depicted in stick format.

3. Close‐up view of the interface between the DDX6_C (orange) and LSM14 (light blue). Side chains of interacting amino acid residues are shown in stick format. Hydrogen‐bonding interactions are indicated with black dashed lines, and secondary structure elements are numbered as in (A).

Figure 5. The DDX6–LSM14 interaction is necessary for P‐body assembly

1. SDS–PAGE analysis of input (lower image) and pull‐down experiment (upper image) using recombinant MBP‐DDX6_C protein immobilized on amylose‐beads and incubated with recombinant GST‐LSM14_FDF‐TFG (residues 290–397) protein constructs. Substitutions in GST‐LSM14_FDF‐TFG variants are indicated. MUT4, F292A/F296A/F298A/F305A; MUT6, F292A/F296A/F298A/F305A/T395A/F396A.

2. Immunoprecipitation (IP) of wild‐type and mutant FLAG‐LSM14A proteins from benzonase‐treated HeLa cell extracts using anti‐FLAG antibody. Immunoprecipitated complexes were separated by SDS–PAGE and probed with antibodies against indicated proteins.

3. Confocal fluorescence micrographs of fixed HeLa cells expressing FLAG‐tagged wild‐type or mutant LSM14. Cells were stained with anti‐FLAG (green) and anti‐DDX6 (red) antibodies. The merged images show the FLAG signal in green and the DDX6 signal in red.

Source data are available online for this figure.

Figure EV2. LSM14‐dependent RNP granule assembly

1. Confocal fluorescence micrographs of fixed HeLa cells expressing FLAG‐fusions of full‐length LSM14 (WT) and mutant proteins. Cells were stained with anti‐FLAG (green) and anti‐DCP1 (red) antibodies. While WT LSM14 leads to formation of distinct cytoplasmic foci (P‐bodies), indicated deletion and substitution variants of LSM14 impair assembly of P‐bodies.

2. Confocal fluorescence micrographs of arsenite‐induced stress granule formation in fixed HeLa cells expressing FLAG‐fusions of full‐length LSM14 (WT) and mutant proteins. Cells were stained with anti‐FLAG (green) and anti‐HUR (HUR, Hu protein R, red) antibodies. All tested LSM14 mutants are recruited to stress granules independently of their ability to bind DDX6 or EDC4.

3. Confocal fluorescence micrographs of fixed HeLa cells expressing FLAG‐fusions of full‐length LSM14 (WT) and mutant proteins. Cells were stained with anti‐FLAG (green) and anti‐DDX6 (red) antibodies. While WT LSM14 leads to formation of distinct cytoplasmic foci (P‐bodies), inversion of the FDF and TFG motifs in LSM14 impairs P‐body assembly.

Figure 6. LSM14 motifs support formation of protein interaction network including EDC4

1. Dot plot depicting high‐confidence protein interactions identified by affinity purification of FLAG‐LSM14 and FLAG‐LSM14_ΔFFD in HeLa cells. An average of two independent experiments for each tagged variant is presented. Node color represents the absolute spectral count sum, the node edge color corresponds to the SAINTexpress Bayesian FDR value (BFDR), and the node size displays the relative abundance of a given prey comparing FLAG‐LSM14 to FLAG‐LSM14_ΔFFD samples. Avg‐Spec denotes the spectral counts for each indicated prey. Reduction of EDC4 interaction in FLAG‐LSM14 ΔFFD is indicated with red arrow.

2. Left: Schematic representation of wild‐type and FFD mutant (FFD_MUT). Right: Immunoprecipitation (IP) of WT, ΔFFD, and FFD_MUT FLAG‐LSM14A proteins from benzonase‐treated HeLa cell extracts using anti‐FLAG antibody. Immunoprecipitated complexes were separated by SDS–PAGE and probed by Western blotting using antibodies against the indicated proteins.

3. Schematic diagram of LSM14 and PATL1 proteins, as well as the TFG_PATL1‐FDF_LSM14 construct in which the PATL1 TFG motif (residues 27–42) has been fused upstream to the LSM14 FDF motif (residues 291–313).

4. SDS–PAGE analysis of a pull‐down experiment using recombinant MBP‐DDX6_C protein immobilized on amylose‐beads and incubated with recombinant GST‐TFG_PATL1‐FDF_LSM14 protein construct.

5. Immunoprecipitation (IP) of WT, ΔTFG, and TFG‐FDF_MUT FLAG‐LSM14A constructs. In the TFG‐FDF_MUT construct, the LSM14 TFG motif has been deleted and the equivalent motif from PATL1 was placed upstream of the FDF motif. Proteins were precipitated from benzonase‐treated HeLa cell extracts using anti‐FLAG antibody, separated by SDS–PAGE and probed with antibodies against the indicated proteins.

Source data are available online for this figure.

Figure EV3. Hydrophobic surfaces on DDX6_C mediate interactions with conserved sequence motifs in EDC3, PATL1, 4E‐T, and LSM14

1. Side‐by‐side structural comparison of human DDX6_C or its yeast homolog Dhh1 (orange) in complexes with LSM14 (blue), EDC3 (cyan, PDB ID 2WAY), yeast (Sc) Pat1 (salmon, PDB ID 4BRW), and 4E‐T (pink, PDB ID 5ANR). Structures are shown in identical orientations, and conserved sequence motifs are indicated. Unstructured portion of the LSM14 polypeptide chain is depicted as dashed line. Conserved sequence motifs from the DDX6‐interacting factors occupy identical patches on the surface of DDX6, yet the topology of these motifs is different in LSM14.

2. Sequence alignment of regions encompassing the FDF (red box), FFD (black box), and TFG (green boxes) motifs in human (Hs) LSM14, yeast (Sc) SCD6, human EDC3, and 4E‐T and yeast Pat1.

3. Structural model of the Pat1‐LSM14 chimeric protein used in this study, colored as in (B). Left: Structure of LSM14_FDF‐TFG in complex with DDX6_C. Middle: Model of yeast Pat1 FDF‐TFG motifs (salmon, PDB ID 4BRW) bound to human DDX6_C. Right: Model of a Pat1‐LSM14 chimeric protein in complex with DDX6_C. Fusion of a Pat1‐fragment containing the TFG motif upstream of a LSM14‐fragment containing the FDF motif is indicated by black dashed line. Motifs mediating interaction with DDX6 are indicated.

Figure EV4. ITC analysis of DDX6_C binding by LSM14, PATL1, EDC3, and 4E‐T

ITC binding isotherms of 500 μM LSM14_FDF‐TFG (upper left), PATL1_TFG‐FDF (upper middle), PATL1_TFGLSM14_FDF (upper right), LSM14_FDF‐TFG F396A (lower left), MBP‐EDC3_TFG‐FDF (lower middle), and 4E‐T_WFS‐IEL (lower right) titrated into 50 μM DDX6_C. Data were fitted to a single‐binding site model, and the dissociation constants (K _d) were determined based on three independent experiments. Data analysis, fitting, and K _d calculation were performed using Origin7.

Figure EV5. Structural comparison of the LSM domains of LSM14 and EDC3 in the context of 4E‐T binding

1. Multiple sequence alignment of the LSM domains of human (Hs) LSM14, Drosophila (Dm) Tral, zebrafish (Dr) LSM14, human (Hs) and Drosophila (Dm) EDC3. Residues of EDC3 that would impair 4E‐T binding by steric clashes are colored in red.

2. Structural comparison of the LSM14–4E‐T complex with a model of the EDC3–4E‐T complex. Left: Zoomed‐in view of two main interactions interfaces between LSM14 (light blue) and 4E‐T (green). Middle: Model of the LSM domain of EDC3 (cyan, PDB ID 2VC8) superimposed with the LSM14‐binding motifs of 4E‐T, shown in identical orientations as LSM14. Right: Overlay of the LSM14 and EDC3 LSM domains, shown in the respective orientations. 4E‐T residues involved in complex formation are shown as sticks and labeled by single letter code. Residues from LSM14 and EDC3 involved in complex formation or intramolecular interactions are shown as sticks and labeled by single letter code. EDC3 residues clashing with 4E‐T are colored in red. Hydrogen‐bonding interactions mediating divergent structural features in the LSM14 and EDC3 LSM domains are indicated with black dashed lines. Secondary structure elements are numbered as in Fig 2A.