(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



A MademoiseLLE domain binding platform links the key RNA transporter to endosomes [1]

['Senthil-Kumar Devan', 'Institute Of Microbiology', 'Heinrich Heine University Düsseldorf', 'Cluster Of Excellence On Plant Sciences', 'Düsseldorf', 'Stephan Schott-Verdugo', 'John Von Neumann Institute For Computing', 'Nic', 'Jülich Supercomputing Centre', 'Jsc']

Date: 2022-08 

Spatiotemporal expression can be achieved by transport and translation of mRNAs at defined subcellular sites. An emerging mechanism mediating mRNA trafficking is microtubule-dependent co-transport on shuttling endosomes. Although progress has been made in identifying various components of the endosomal mRNA transport machinery, a mechanistic understanding of how these RNA-binding proteins are connected to endosomes is still lacking. Here, we demonstrate that a flexible MademoiseLLE (MLLE) domain platform within RNA-binding protein Rrm4 of Ustilago maydis is crucial for endosomal attachment. Our structure/function analysis uncovered three MLLE domains at the C-terminus of Rrm4 with a functionally defined hierarchy. MLLE3 recognises two PAM2-like sequences of the adaptor protein Upa1 and is essential for endosomal shuttling of Rrm4. MLLE1 and MLLE2 are most likely accessory domains exhibiting a variable binding mode for interaction with currently unknown partners. Thus, endosomal attachment of the mRNA transporter is orchestrated by a sophisticated MLLE domain binding platform.

Eukaryotic cells rely on sophisticated intracellular logistics. Macromolecules like mRNA must be transported to defined subcellular destinations for local translation. This is mediated by active transport along the cytoskeleton. Endosomes are carrier vehicles that shuttle along microtubules by the action of molecular motors. It is currently unclear how mRNAs are attached mechanistically to these membranous units during transport. We study the model microorganism Ustilago maydis where numerous components of endosomal mRNA transport have already been identified. Previously, we found that the key RNA-binding protein Rrm4 interacts with the endosomal adaptor protein Upa1. Here, we perform a structure/function analysis and discovered that Rrm4 contains not one but three different versions of a protein-protein interaction domain, called the MademoiseLLE domain, to facilitate the attachment with transport endosomes. Importantly, they function with a strict hierarchy with one essential domain and the others play accessory roles. This is currently, the most detailed mechanistic description of how an RNA-binding protein and its bound cargo mRNAs are attached to endosomes. The usage of three similar protein-protein interaction domains forming a complex binding platform with a defined hierarchy might be operational also in other unknown protein-protein interactions.

Funding: The work was funded by grants from the Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy EXC-2048/1 - Project ID 39068111 to MF; Project-ID 267205415 – SFB 1208 to MF (project A09), HG (project A03), and LS (project A01). The Center for Structural Studies was funded by the Deutsche Forschungsgemeinschaft (DFG Grant number 417919780; INST 208/740-1 FUGG; INST 208/761-1 FUGG to S.H.J S). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files. The MLLE2 structure was deposited at the worldwide protein data bank under the accession code 7PZE. We uploaded the data to the Small Angle Scattering Biological Data Bank (SASBDB) with the accession codes SASDMS5(G-Rrm4) and SASDMT5 (H-Rrm4-NT4). https://www.sasbdb.org/data/SASDMS5/ https://www.sasbdb.org/data/SASDMT5/ .

Mutations in the C-terminal MLLE domain of Rrm4 result in the loss of Rrm4 motility, suggesting that the link to endosomes is disrupted [ 15 ]. Consistently, the C-terminus of Rrm4 recognises the PAM2L sequence of the adaptor protein Upa1 [ 16 ], suggesting that the interaction of MLLE domains with PAM2L sequences is responsible for its endosome association. This study combines structural biology with fungal genetics to demonstrate that the C-terminal half of Rrm4 has three divergent MLLE domains with a flexible arrangement and each domain contributes differentially to the endosomal attachment.

The MLLE domain was first identified as a conserved domain at the C-terminus of the human cytoplasmic poly(A)-binding protein 1 (PABP1C) [ 24 , 25 ]. Solution and crystal structures of PABC domains from PABP1C and ubiquitin ligase UBR5 showed that they are structurally conserved [ 26 , 27 ]. The domain is about 70 amino acids in length and consists of five bundled α-helices. Interaction with the PAM2-binding motif (consensus sequence xxLNxxAxEFxP) is characterised by the central α-helix 3 with the sequence KITGMLLE and mediated by two adjacent hydrophobic pockets [ 28 ], with the binding of the Phe residue of the PAM2 motif being the major determinant for this interaction [ 29 ]. Besides human PABPC1, there are currently only two additional proteins with MLLE domains described: the ubiquitin ligase UBR5 functioning, for example, during microRNA-mediated gene silencing [ 30 ] and Rrm4-type RNA-binding proteins from fungi ( Fig 1B ) [ 31 ].

( A ) Schematic representation of protein variants drawn to scale (bar, 200 amino acids, number of amino acids indicated next to protein bars) using the following colouring: dark green, RNA recognition motif (RRM); orange, MLLE Rrm4 domains; dark blue, MLLE Pab1 ; light blue PAM2; light orange PAM2L sequence (PL1–2) Ankyrin repeats (5xANK), FYVE domain, and RING domain of Upa1 are given in dark grey. ( B ) Sequence alignment of previously determined MLLE domains showing the degree of similarity to the three Rrm4-MLLE domains and the positions (Hs—Homo sapiens, Ta—Triticum aestivum, La—Leishmania major, Sc—Saccharomyces cerevisiae, Tc—Trypanosoma cruzi, Rn—Rattus norvegicus, Um—Ustilago maydis, PABPC1, Pab1 –poly [A]-binding protein, UBR5—E3 ubiquitin-protein ligase). Accession number and sequence coverage are listed in S1 Table . Multiple sequence alignment was performed by ClustalW. ( C) Identification and modelling of C-terminal MLLE domains of Rrm4. The iterative process is depicted graphically. The best-identified template for each run, and the region of that template that aligns with Rrm4, are displayed (see also S1A Fig for the templates used for the final models). The structural models obtained are shown for the span of the first identified template and are coloured according to their per-residue TopScore, where the scale from 0 to 1 indicates a low to high local structural error.

Rrm4 is the key RNA-binding protein of the transport process that recognises defined sets of cargo mRNAs via its three N-terminal RRMs ( Fig 1A ) [ 19 ]. Rrm4 and bound cargo mRNAs are linked to endosomes by Upa1, containing a FYVE zinc finger for interaction with PI 3 P lipids (phosphatidylinositol 3-phosphate; Fig 1A ) [ 16 , 20 ]. The adaptor protein Upa1 contains a PAM2 motif (poly[A]-binding protein interacting motif 2) [ 21 – 23 ] and two PAM2-like (PAM2L) sequences. These motifs are crucial for interaction with MademoiseLLE (MLLE) domains of the poly(A)-binding protein Pab1 and Rrm4, respectively ( Fig 1A ) [ 16 ].

Among the best-studied examples of membrane-coupled mRNA transport is the endosomal mRNA transport in the corn pathogen Ustilago maydis [ 5 , 13 , 14 ]. Extensive peripheral movement of mRNAs is needed for efficient unipolar growth of infectious hyphae. These hyphae grow highly polarised by expanding at the growing tip and inserting regularly spaced septa at the basal pole. Loss of mRNA distribution causes aberrant bipolar growth [ 6 , 15 , 16 ]. Key vehicles of cargo mRNAs are Rab5a-positive endosomes that shuttle along microtubules by the concerted action of plus-end directed kinesin-3 and minus-end directed cytoplasmic dynein [ 6 ]. Important cargo mRNAs are, for example, all four septin mRNAs. Their local translation during transport is essential to form heteromeric septin complexes on the surface of transport endosomes. Endosomes deliver these complexes to the hyphal tip, forming a defined gradient of septin filaments at the growing pole [ 17 – 19 ].

mRNA localisation and local translation are essential for spatiotemporal control of protein expression. An important mechanism to achieve localised translation is the active transport of mRNAs along the cytoskeleton [ 1 – 3 ]. Mainly, long-distance transport of mRNA is mediated by motor-dependent movement along microtubules. Transport endosomes are important carriers that move messenger ribonucleoprotein complexes (mRNPs), consisting of RNA-binding proteins and cargo mRNAs on their cytoplasmic surface [ 1 , 4 , 5 ]. This process is evolutionarily conserved in fungi, plants, and animals [ 5 – 11 ]. In endosperm cells of developing rice seeds, cargo mRNAs are transported to the cortical endoplasmic reticulum (ER) by the action of the two RNA recognition motif (RRM)-containing proteins RBP-P and RBP-L. These form a quaternary complex with membrane trafficking factor NSF (N-ethylmaleimide-sensitive factor) and small GTPase Rab5a on the endosomal surface [ 12 ]. In neurons, mRNA transport has been linked to early and late endosomes as well as lysosomal vesicles. Especially, local translation of mRNAs encoding mitochondrial proteins on the surface of late endosomes is needed for mitochondrial function. Importantly, this trafficking process has been associated with the neuronal Charcot Marie-Tooth disease [ 8 ]. Annexin 11, a factor implicated in amyotrophic lateral sclerosis (ALS), was found as an mRNP linker on motile lysosomal vesicles [ 9 ]. Also, the five-membered FERRY complex was recently identified connecting mRNAs encoding mitochondrial proteins to neuronal endosomes by interaction with the active form of Rab5 [ 10 , 11 ].

Results

Iterative structural modelling predicts three MLLE domains at the C-terminus of Rrm4 To generate structural models of the MLLE domains present in Rrm4, we focused on the C-terminal part of the protein (residues 421 to 792). This excluded the three N-terminal RRMs but included the previously predicted two C-terminal MLLE domains (Fig 1A and 1B) [31]. Subjecting this Rrm4 sequence region to iterative comparative modelling with TopModel (Fig 1C) [32] revealed, as expected, the previously identified two regions with homology for MLLE domains located at residues 571–629 and 712–791 (denoted MLLE2 and MLLE3; Fig 1C) [31]. Unexpectedly, using the TopModel workflow with its efficient template selection capabilities [32], we identified an additional de novo predicted MLLE domain located at residues 451–529 (denoted MLLE1; Figs 1B, 1C and S1A). Although the sequence identity between templates and their respective Rrm4 sequence stretches was only 17 to 32% (Figs 1B and S1A), the generated MLLE domain models had a high predicted local structural quality, as assessed by TopScore (Fig 1C) [33]. The generated models were also verified by the current deep neural network modelling approaches AlphaFold2 and RoseTTAFold (S1B Fig) [34,35], further indicating that the C-terminal half of Rrm4 has three MLLE domains instead of the previously identified two. All of these MLLE domains might be relevant for the interaction with Upa1.

X-ray analysis of the second MLLE domain confirms the predicted structural models To verify the structural models further, we expressed and purified an N-terminally truncated version of the Rrm4 carrying the three MLLE domains in Escherichia coli (S2A and S2B Fig; version H-Rrm4-NT4 carrying an N-terminal hexa-histidine-tag; Materials and methods) [16]. Size exclusion chromatography combined with Multi-angle light scattering (MALS) indicated that the protein was homogenous and did not form aggregates (S2C Fig). We thus set out to crystallize the protein for X-ray diffraction analysis (see Material and methods). Testing 2016 different conditions, crystals were only obtained in individual cases after at least 7 days of incubation. A complete dataset was collected from a single crystal diffracting to 2.6 Å resolution and a P4 3 2 1 2 symmetry. Data and refinement statistics are given in S2 Table. Surprisingly, the unit cell dimensions were small and, with a Matthews coefficient assuming 50% solvent content, only 128 amino acids would fit into the asymmetric unit of the crystal. Hence, the unit cell had an insufficient size to cover H-Rrm4-NT4, which contains 380 amino acids. Using the predicted models of MLLE1-3 as templates for molecular replacement, only MLLE2 gave a clear solution, showing after refinement that two copies of MLLE2 (residues 567–630) were present in the asymmetric unit. For comparison, previously, two copies of the MLLE domain in the asymmetric unit were reported in crystals of MLLE of UBR5 [36]. The structural data indicated that the protein was truncated from both termini during crystallisation, resulting in a shortened version of the H-Rrm4-NT4 protein that formed stable crystals (see Material and methods). Both MLLE2Rrm4 copies adopted the same overall fold as seen by the RMSD of 0.29 Å over 59 C-alpha atoms. The MLLE2Rrm4 crystal structure displayed high similarity with the MLLE domain of the ubiquitin ligase UBR5 (MLLEUBR5; PDB code 3NTW, RMSD of 0.97 Å over 56 amino acids) [36] and the MLLE domain of PABPC1 (PDB code 3KUS, RMSD of 1.34 Å over 61 amino acids) [37]. The MLLE2Rrm4 domain consisted of four helices (designated α2–5; Fig 2A) arranged as a right-handed superhelix similar to MLLEUBR5. In comparison to the MLLE domain of PABPC1, the first short helix was absent in both MLLE2Rrm4 and MLLEUBR5 structures. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Rrm4 contains a C-terminal tripartite MLLE binding platform. (A) Crystal structure of the MLLE2 domain is highlighted in orange. The four helices are indicated by α2–5 according to the 5 helix nomenclature found in MLLE domains (28). Note that the first short helix α1 is missing. Arg573 and Glu591 are highlighted in the sticks. These sides chains would interfere with the binding of the canonical Phe of PAM2 type motifs. (B) Structural alignment of the MLLE2 model generated by TopModel and the X-ray crystal structure of this domain (grey or orange, respectively). The all-atom RMSD is 0.69 Å, resulting mostly from different rotamers of solvent-exposed sidechains. (C) Comparison of peptide-binding sites after structural alignment of the models of Rrm4 MLLE domains (orange shades) and the canonical MLLE domain of HsPABPC1 (blue; PDB ID 3KUS) and manually placing the PAM2 motif of PAIP2 (lilac). In the interaction of MLLEPABPC1 with PAM2 of PAIP2, Phe118 of PAM2 is the major determinant for binding and present in all the PAM2 motifs except LARP4a and b ([28,29]; S3A Fig). Of the identified Rrm4 MLLE domains, only MLLE3 retains all sidechains that favour the binding of this characteristic Phe; particularly, Gly736 should allow the Phe to bind into a pocket. MLLE1 and MLLE2 have Ser471 and Arg573 instead of Gly in this position, suggesting that Phe binding would be sterically hindered in these interfaces. (D) Left panel Experimental data curve for GST-Rrm4 is shown in black dots with grey error bars, the EOM fit as a red line (χ2 = 1.289). The intensity is displayed as a function of momentum transfers. Right panel Selected model of the EOM analysis from GST-Rrm4 with a R g of 8.75 nm, a D max of 23.99 nm with a volume fraction of ~0.25. (E) left panel Experimental data curve for H-Rrm4NT4 is shown in black dots with grey error bars, the EOM fit as the red line (χ2 = 1.262). The intensity is displayed as a function of momentum transfers. right panel Selected model of the EOM analysis from H-Rrm4NT4 with a R g of 5.10 nm, a D max of 16.43 nm, and a volume fraction of ~0.75. The MLLE subdomains are shown in cartoon representation (MLLE1 in light orange, MLLE2 in orange, MLLE 3 in dark orange, and the GST in dark grey) and the missing amino acids as grey spheres (all other models and the SAXS data are available in S2E Fig). https://doi.org/10.1371/journal.pgen.1010269.g002 When comparing the obtained crystal structure with the MLLE2 Rrm4 model generated by TopModel, the average RMSD was 0.69 Å over the backbone atoms, close to the uncertainty of the atomic coordinates of the experimental structure (Fig 2B). Importantly, this confirmed our structural model of MLLE2Rrm4 and strongly suggested that the modelled MLLE1Rrm4 and MLLE3Rrm4 domains should be of equally high quality. We compared the predicted models of MLLE1-3Rrm4 with the known structure of the human PABPC1 focusing on the well-described PAM2 peptide-binding pocket. This revealed that MLLE3Rrm4 maintained a characteristic Gly residue at position 736 that binds the conserved Phe residue of the PAM2 motifs, a major binding determinant in PABPC1 and UBR5 (Fig 2C) [29,36]. However, the binding interfaces of MLLE1Rrm4 and MLLE2Rrm4 were altered compared to the ‘canonical’ binding site in PABPC1 and UBR5 (Fig 2C). Instead of Gly, MLLE1Rrm4 and MLLE2Rrm4 had a Ser and Arg in the corresponding positions 471 and 573. The notion that MLLE1Rrm4 and MLLE2Rrm4 may differ from canonical MLLE domains was also supported by the lower sequence identity of MLLE1Rrm4 and MLLE2Rrm4 when compared to previously characterised MLLE domains (Figs 1B; S1A). In summary, structural modelling revealed the presence of three MLLE domains at the C-terminus of Rrm4. Furthermore, the structure of the MLLE2Rrm4 domain was successfully verified by X-ray crystallographic analysis. MLLE1Rrm4 and MLLE2Rrm4 are divergent in the key region of PAM2 binding, suggesting that these domains might employ a different binding mode or show a different binding specificity.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1010269

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/