Reanalysis of cryo-EM data reveals ALK-cytokine assemblies with both 2:1 and 2:2 stoichiometries

Jan Felix; Steven De Munck; J. Fernando Bazan; Savvas N. Savvides

doi:10.1371/journal.pbio.3003124

Abstract

Activation of Anaplastic lymphoma kinase (ALK) and leukocyte tyrosine kinase (LTK) by their cognate cytokines ALKAL2 and ALKAL1 plays important roles in development, metabolism, and cancer. Recent structural studies revealed ALK/LTK-cytokine assemblies with distinct stoichiometries. Structures of ALK-ALKAL2 and LTK-ALKAL1 complexes with 2:1 stoichiometry determined by X-ray crystallography contrasted the 2:2 ALK-ALKAL2 complexes determined by cryo-EM and X-ray crystallography. Here, we show based on reanalysis of the cryo-EM data deposited in EMPIAR-10930 that over half of the ALK-ALKAL2 particles in the dataset are classified into 2D and 3D classes obeying a 2:1 stoichiometry besides the originally reported structure displaying 2:2 stoichiometry. Unlike particles representing the 2:2 ALK-ALKAL2 complex, particles for the 2:1 ALK-ALKAL2 complex suffer severely from preferred orientations that resulted in cryo-EM maps displaying strong anisotropy. Here, we show that extensive particle orientation rebalancing in cryoSPARC followed by 3D refinement with Blush regularization in RELION constitutes an effective strategy for avoiding map artifacts relating to preferred particle orientations and report a 3D reconstruction of the 2:1 ALK-ALKAL2 complex to 3.2 Å resolution from EMPIAR-10930. This new cryo-EM structure together with the crystal structures of ALK-ALKAL2 and LTK-ALKAL1 complexes with 2:1 stoichiometry reconciles a common receptor dimerization mode for ALK and LTK and provides direct evidence for the presence of an ALK-ALKAL2 complex with 2:1 stoichiometry next to the reported 2:2 stoichiometric assembly in the EMPIAR-10930 dataset. Finally, our analysis emphasizes the importance of public deposition of raw cryo-EM data to allow reanalysis and interpretation.

Citation: Felix J, De Munck S, Bazan JF, Savvides SN (2025) Reanalysis of cryo-EM data reveals ALK-cytokine assemblies with both 2:1 and 2:2 stoichiometries. PLoS Biol 23(4): e3003124. https://doi.org/10.1371/journal.pbio.3003124

Academic Editor: Yan Zhang, Zhejiang University School of Medicine, CHINA

Received: January 11, 2025; Accepted: March 23, 2025; Published: April 10, 2025

Copyright: © 2025 Felix et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The cryo-EM map and accompanying structural model for the 2:1 ALKTG-EGFL-ALKAL2 complex have been deposited in the EMDB/PDB with accession codes: EMD-51087/9g5i. The particle .star file obtained after 3D refinement with Blush Regularization in RELION 5 is added to this manuscript as a Supplementary Data file (S1 File) and can be used to extract particles after running CTF estimation on the Motion Corrected and Dose Weighted (*_DW.mrc) micrographs available from EMPIAR-10930 (https://www.ebi.ac.uk/empiar/EMPIAR-10930/).

Funding: SNS acknowledges research support from the Flanders Institute for Biotechnology (“Vlaams Instituut voor Biotechnologie”, VIB, grant number C0101) and the FWO (“Fonds voor Wetenschappelijk Onderzoek – Vlaanderen”, grant number G0B4918N). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: SDM and SNS are listed as co-inventors on a patent filing EP 21184332.1 with the European Patent Office.

Abbreviations: ADP, atomic displacement parameter; ALK, Anaplastic lymphoma kinase; CSF-1R, colony stimulating factor 1 receptor; EGFR, epidermal growth factor receptor; FGFR, fibroblast growth factor receptor; Flt3R, Flt3 receptor; FSC, Fourier Shell Correlation; IR, insulin receptor; LTK, leukocyte tyrosine kinase; PDGFR, platelet-derived growth factor receptor; RTK, receptor tyrosine kinase; VEGFR, vascular endothelial growth factor receptor

Introduction and results

Anaplastic lymphoma kinase (ALK) and the related leukocyte tyrosine kinase (LTK) are receptor tyrosine kinases (RTKs) that are activated by binding their cognate cytokines, ALKAL1 and ALKAL2 (also called FAM150A/B or AUGβ/α) [1–3]. Their signaling outputs elicit a range of pleiotropic activities in development and metabolism, and their dysregulated intracellular kinases are keenly pursued as therapeutic drug targets in various cancers [4]. However, the field had long been hampered by a paucity of insights into the intricate folds of ALK and LTK and their cytokine-bound complexes. A triad of recent studies [5–7] have aimed to provide the missing structural and mechanistic framework for the cytokine-mediated activation of ALK family receptors but diverged in their proposed assemblies.

Our study by De Munck and colleagues [5] provided representative assemblies for the entire ALK family by reporting crystal structures of both the ligand-binding extracellular regions of human ALK and LTK (ALK_TG/LTK_TG) in complex with ALKAL2 and ALKAL1, respectively (PDB entries 7nwz and 7nx0). These structures revealed how a single cytokine molecule nucleates homodimeric receptor assembly via distinct site 1 and 2 interaction interfaces, creating a key site 3 receptor-receptor contact (S1 Fig). In accordance with the evolutionary relatedness [8] and close structural similarity between ALK and LTK, key receptor-receptor and receptor-ligand interactions are conserved in the resulting 2:1 stoichiometric ALK-ALKAL2 and LTK-ALKAL1 assemblies [5], which are pseudo-symmetric due to the asymmetric nature of the ALKAL2/1 ligands. These structures thus suggest an asymmetric mode of ALK/LTK receptor dimerization and activation akin to fibroblast growth factor receptor (FGFR) signaling [9], and contrasting with symmetric receptor dimerization as observed for the RTKs Kit [10,11], colony stimulating factor 1 receptor (CSF-1R) [12], Flt3 receptor (Flt3R) [13], epidermal growth factor receptor (EGFR) [14], platelet-derived growth factor receptor (PDGFR) [15], and vascular endothelial growth factor receptor (VEGFR) [16].

In contrast, Li and colleagues [6] and Reshetnyak and colleagues [7] have focused only on ALK-ALKAL2 complexes comprising the extracellular ligand-binding fragment of ALK and the membrane-proximal EGFL domains (ALK_TG-EGFL), and reported symmetric assemblies obeying a 2:2 stoichiometry obtained via X-ray crystallography using an ALK_TG-EGFL-ALKAL2 fusion construct (PDB entry 7LS0) or separate ALK_TG-EGFL and full-length ALKAL2 via electron cryo-microscopy (cryo-EM, PDB entry 7n00), respectively (S1B Fig). In this side-by-side architecture, the main cytokine-receptor interface corresponds to the high affinity site 1 also reported by De Munck and colleagues [5]. An additional ALK-ALKAL2′ interaction, covering 250 Å² of buried surface area, was proposed to be responsible for bridging the two 1:1 ALK-ALKAL2 subcomplexes into an assembly that could support signaling. Intrigued by this 2:2 stoichiometry differing from the 2:1 stoichiometry revealed by our crystal structures of truncated ALK_TG in complex with ALKAL2 (and LTK_TG with ALKAL1), we closely examined the cryo-EM data and analysis reported by Reshetnyak and colleagues [7] as deposited in the EMPIAR [17] data base (entry EMPIAR-10930).

At the outset, we identified a number of 3D-classes (classes 1, 3, 7, 8, and 9 illustrated in Extended Data Fig 4 in Reshetnyak and colleagues [7]) resembling the compact 2:1 ALK_TG/LTK_TG-cytokine complexes we had reported [5]. To investigate this apparent particle heterogeneity, we leveraged the raw cryo-EM data and uncleaned particle stack as deposited by Reshetnyak and colleagues [7] (access code EMPIAR -10930). An initial data processing strategy following the methods described in Reshetnyak and colleagues [7], including 8× binning of the data (Materials and methods) revealed that 6 of the 10 most populated classes after initial 2D classification do not correspond to classes shown in the final 2D classification reported in Reshetnyak and colleagues [7] (Fig 1A). This prompted us to pursue extensive data processing (Materials and methods, S2 Fig) of particles present in both sets of 2D classes, namely the ones that correspond to clear 2-fold symmetry (Fig 1A, orange squares) and the ones that lack apparent symmetry (Fig 1A, blue squares). However, the final refined map of the C1 symmetric particles displayed severe preferential particle orientations leading to a map that is smeared along the Z-direction, a common problem in cryo-EM requiring sample and data collection optimization. Although the resulting map clearly displays two ALK_TG-EGFL extracellular domains and one ALKAL2 copy in a similar 2:1 assembly as observed in the crystal structure of the ALK_TG-ALKAL2 complex (S2 Fig), artifacts resulting from severe anisotropy hampered confident model building and prohibited detailed structural interpretations.

Download:

Fig 1. Reanalysis of cryo-EM data from EMPIAR entry EMPIAR-10930 deposited by Reshetnyak and colleagues, Nature (2021) [

7]. (A) Ten most populated 2D class averages after performing 2D classification on 11.8 million selected particles. Classes corresponding to 2:1 ALK–ALKAL2 particles are highlighted in blue, while classes corresponding to 2:2 ALK–ALKAL2 particles are highlighted in orange. (B) Ab initio model generation in cryoSPARC [18] results in classes corresponding to both a 2:1 ALK–ALKAL2 and 2:2 ALK–ALKAL2 assembly. Hetero-refinement of the 2:1 and 2:2 ALK–ALKAL2 classes results in the two models displaying 2:1 and 2:2 ALK–ALKAL2 assemblies, respectively. The ALK–ALKAL2 complexes with 2:1 (PDB 7nwz) and 2:2 (PDB 7n00) stoichiometries are shown fitted inside the maps, with ALK colored pink/gray and ALKAL2 colored blue/purple. (C) 3D refinement with Blush regularization [19] in RELION v5 of a trimmed particle set after extensive automatic rebalancing of particle orientations as implemented in cryoSPARC 4.5 (see also Materials and methods and S1 Table). A front view is shown of a map sharpened in RELION using a B-factor of −100 Å², with ALK colored pink/gray and ALKAL2 colored blue. Front, side, and top views are shown of a transparent map, sharpened using deepEMhancer, with fitted structural model. (D and E) Zoomed-in insets illustrating the quality of the cryo-EM map corresponding to the two major interaction interfaces (site 1 and 2) in the ALK:ALKAL2 complex. Key interacting residues are annotated and shown as sticks. (F) Local resolution estimation performed in RELION v5 of the final map for the 2:1 ALK-ALKAL 2 complex. (G) Fourier shell correlation (FSC) plot corresponding to the final map of the 2:1 ALK:ALKAL2 complex. The corrected FSC (purple lines) is calculated after performing correction by noise substitution [20], and the resolution at FSC = 0.143 is annotated via a dotted purple line. A map-to model FSC curve calculated using the fitted model and an unsharpened, unfiltered full map is shown (gray line) with the resolution at FSC = 0.5 annotated via a dotted gray line.

https://doi.org/10.1371/journal.pbio.3003124.g001

To overcome the apparent anisotropy presented in the 3D reconstructions toward a 2:1 ALK_TG-EGFL–ALKAL2 complex, we reasoned that combining exhaustive automatic particle orientation rebalancing (available in cryoSPARC [18], v4.5) to exclude particles from over-populated direction bins, followed by 3D refinement with Blush regularization [19] in RELION v5, might result in more isotropic 3D reconstructions. Blush regularization uses a pre-trained denoising convolutional neural network on pairs of half-set reconstructions as a smoothness prior and is shown to remove artifactual densities resulting from uneven angular distributions or streaky features in overfitted maps [19]. This new regularization approach to resolving vexing cryo-EM structural data was successfully employed to reveal the distinct binding modes of the anti-CRISPR repressor Aca2 with DNA and RNA [21].

Starting from a set of selected 3D classes (1.4 million particles) after 3D hetero-refinement (Materials and methods), we divided these 3D classes in two sets based on the absence or presence of apparent 2-fold symmetry. Such grouping resulted in 3D classes resembling either the C1-symmetric 2:1 ALK_TG-ALKAL2 crystal structure reported by De Munck and colleagues [5] (876,806 particles, ~60% of the total), or the deposited C2-symmetric 2:2 ALK_TG-EGFL-ALKAL2 cryo-EM map reported by Reshetnyak and colleagues [7] (523,467 particles, ~40% of the total) (Fig 2A).

Download:

Fig 2. Final processing pipeline of cryo-EM dataset EMPIAR-10930.

(A) Final processing steps were performed in CryoSPARC [18] v4.5 or RELION [22,23] v5. Post-processing was performed in RELION or using deepEMhancer [24]. (B and C) Conical FSC summary plots corresponding to the final maps of the 2:1 (B) and 2:2 (C) ALK:ALKAL2 complexes, generated via “Orientation Diagnostics” in cryoSPARC v4.5. The resolution at FSC = 0.143 is annotated via dotted blue lines. (D) Azimuth plot showing the distribution of orientations over Azimuth (x-axis) and Elevation (y-axis) angles for the particle set corresponding to the final map of the 2:1 ALK-ALKAL2 complex, obtained after Orientation Rebalancing in cryoSPARC v4.5 and 3D refinement with Blush regularization [23] in RELION v5. (E) Display of the 20 most populated classes after 2D classification of the final particle stack corresponding to the 2:1 ALK-ALKAL2 complex. New side and top views not observed in initial 2D classification runs are shown as blue squares.

https://doi.org/10.1371/journal.pbio.3003124.g002

Firstly, we wondered whether our approach would lead to a cryo-EM map of similar resolution and information content for the C2 symmetric ALK_TG-EGFL-ALKAL2 complex with 2:2 stoichiometry as reported by Reshetnyak and colleagues [7]. Indeed, further data processing in cryoSPARC resulted in a 2.3 Å resolution map for the 2:2 ALK_TG-EGFL-ALKAL2 complex (Materials and methods). The resulting map has a comparable resolution as the deposited map by Reshetnyak and colleagues [7] (EMD-24095, 2.3 Å) and, importantly, does not contain any particles from 2D classes corresponding to a 2:1 ALK_TG-EGFL-ALKAL2 stoichiometry (Fig 2A). We note that, similarly as the cryoSPARC refined map reported in Reshetnyak and colleagues [7], map signal for the EGFL domains is less defined than for the RELION refined map obtained by Reshetnyak and colleagues [7] at slightly lower resolution.

We then pursued analysis of the particle set with apparent resemblance to ALK_TG-EGFL-ALKAL2 complexes with 2:1 stoichiometry (Materials and methods). The combined use of Orientation Rebalancing in CryoSPARC and 3D refinement in RELION v5 with enabled Blush regularization, resulted in a final map at 3.2-Å resolution (Figs 1, 2A, S1 Table) displaying an impressive improvement in isotropy, as evidenced by the 10-fold increase in the cFAR score (Fig 2B), and overall interpretability (S3 Fig). Most importantly, post-processing in RELION yielded a sharpened map with clear main chain connectivity and signal for most side chains, allowing complete model building of a 2:1 ALK_TG-EGFL-ALKAL2 complex that includes the membrane-proximal EGFL-like domains of ALK. While the majority of the 2:1 ALK_TG-EGFL-ALKAL2 cryo-EM map has a local resolution to better than 3.5-Å resolution (Fig 1F), the C-terminal tips of the EGFL domains are only resolved at 4–4.5 Å resolution pointing to their inherent flexibility.

To verify that more than half of the ALK-ALKAL2 particles present in the EMPIAR-10930 dataset indeed correspond to particles with a 2:1 stoichiometry, we pursued three independent heterogeneous 3D refinement runs in cryoSPARC using the full set of 11.8 million particles obtained after removing junk classes during the first 2D classification run (Materials and methods), and utilizing the final 2:1 and 2:2 ALK-ALKAL2 complex maps blurred to 20 Å as reference input volumes (S4 Fig). These three runs, each started using different generated random seeds, resulted in 53.5%, 53.1%, and 52.7% particles corresponding to the 2:1 ALK-ALKAL2 3D class, and 46.5%, 46.9%, and 47.3% to the 2:2 ALK-ALKAL2 3D class, thus corroborating the results observed during initial 2D classification (S4 Fig).

We next wondered whether 3D refinement with Blush regularization would be sufficient to obtain an isotropic map from the particle set without Orientation Rebalancing. Starting from the same particle set before the first Orientation Rebalancing step, followed by Ab-Initio model generation (3 classes), hetero-refinement and a final 3D refinement with Blush Regularization of the highest resolution class, resulted in a map at higher resolution (2.7 Å) but with lower isotropy (cFAR score of 0.1) than the map obtained by combining Orientation Rebalancing and 3D refinement with Blush regularization (Fig 3A and 3B). We note that 3D refinement with Blush regularization of the non-Orientation Rebalanced particle set (cFAR score of 0.1) is superior in improving map isotropy than NU-refinement without or with Orientation Rebalancing, which yields maps with cFAR scores of 0.01 and 0.08, respectively (Fig 3B). Furthermore, we also performed a 3D refinement in RELION starting from the Orientation Rebalanced particle set, but without enabling Blush regularization (Fig 3A and 3B). This refinement resulted in a more isotropic 3D reconstruction (cFAR score of 0.25) than NU-refinement in CryoSPARC after Orientation Rebalancing but stalled at a lower resolution of 4.3 Å, prohibiting de novo model building. Thus, enabling Blush regularization during 3D refinement of the Orientation Rebalanced particle set seems to both result in increased map isotropy as well as resolution. Taken together, our analysis demonstrates that the combination of extensive Orientation Rebalancing and 3D refinement with Blush regularization may serve as a suitable strategy to overcome severe preferred particle orientations In Silico in datasets that contain a full but extremely unbalanced range of orientations.

Download:

Fig 3. Comparison of map (an)isotropy following data processing strategies with or without Orientation Rebalancing and/or 3D refinement with enabled Blush Regularization.

(A) Processing workflows (1–5) of cryo-EM dataset EMPIAR-10930 starting from 876,806 selected particles as outlined in Methods. Workflow 1 does not include Orientation Rebalancing (CryoSPARC v4.5) nor 3D refinement with Blush regularization (RELION v5). Workflow 2 is the same as 1 but with a final 3D refinement with Blush regularization in RELION v5 instead of NU-refinement in CryoSPARC v4.5. Workflow 3 utilizes Orientation Rebalancing in CryoSPARC v4.5 followed by NU-refinement in CryoSPARC. Workflow 4 includes Orientation Rebalancing in CryoSPARC v4.5 followed by regular 3D refinement in RELION (without Blush regularization), while workflow 5 combines Orientation Rebalancing in CryoSPARC with 3D refinement with enabled Blush regularization in RELION. (B) Conical FSC summary plots for the final 3D refinements obtained in processing workflows 1, 2, 3, 4, and 5, generated via “Orientation Diagnostics” in cryoSPARC v4.5. The resolution at FSC = 0.143 is annotated in blue.

https://doi.org/10.1371/journal.pbio.3003124.g003

Discussion

The new structure of ALK_TG-EGFL-ALKAL2 with 2:1 stoichiometry presented here is a critical contribution to the present collection of ALK/LTK-cytokine complexes revealed by X-ray crystallography and cryo-EM. The elucidation of an ALK_TG-EGFL-ALKAL2 complex with 2:1 stoichiometry based on a dataset that was initially reported to only support ALK_TG-EGFL-ALKAL2 complexes with 2:2 stoichiometry underscores the power of modern cryo-EM to harness the conformational and stoichiometric diversity harbored in protein samples. Furthermore, our analysis emphasizes the importance of public deposition of raw (cryo-EM) data to allow reanalysis and interpretation. Most importantly, though, this new structure is now the most complete ALK-ALKAL2 complex with 2:1 stoichiometry to date and agrees closely with the ALK_TG-ALKAL2 with 2:1 stoichiometry reported by De Munck and colleagues [5] featuring site 1 (ALK-ALKAL2), site 2 (ALK’-ALKAL2), and site 3 (ALK-ALK’) interaction interfaces (Fig 1). Accordingly, the root-mean-square deviation (r.m.s.d.) between these two structures is 1.3 Å over 675 aligned Cα-atoms (S6A Fig). The site 2 interface present in the 2:1 ALK:ALKAL2 complex, but not in the 2:2 ALK-ALKAL2 complex, involves the α1 helix of ALKAL2 with a key histidine residue (H100) interacting with S737, Y739, and L760 of ALK’ (Fig 1E).

The membrane proximal EGFL domain of ALK engaged in the “site 1” interaction adopts a conformation identical to ALK found in the 2:2 ALK_TG-EGFL-ALKAL2 complex (r.m.s.d. = 0.554 Å over 345 aligned Cα-atoms, S6C Fig). Likewise, a 1:1 ALK_TG-EGFL-ALKAL2 binary complex extracted from the 2:2 ALK_TG-EGFL-ALKAL2 structure matches the “site 1” interface found in the 2:1 ALK_TG-EGFL-ALKAL2 cryo-EM model (r.m.s.d. = 0.631 Å over 400 aligned Cα-atoms, S6B Fig). Due to the asymmetric nature of the ALKAL2 monomer, the EGFL domain of the opposing ALK protomer (ALK’) in the 2:1 ALK_TG-EGFL-ALKAL2 complex does not make any direct contacts at the “site 2” interaction with ALKAL2 (Fig 1). As a result, the ALK’ EGFL domain is less resolved at its extremity due to increased flexibility, resulting in the absence of density for the minor β-hairpin in the cryo-EM map (Figs 1 and S6). Moreover, the major β-hairpin of ALK’ has an outwards tilted conformation compared with the EGFL domain of ALK engaged in the “site 1” interaction (r.m.s.d.= 1.717 Å over 331 aligned Cα-atoms, S6D Fig). Despite this asymmetry, the C-termini of the EGFL domains of the two ALK copies in the 2:1 ALK_TG-EGFL-ALKAL2 complex obtained from EMPIAR-10930 are spaced ~30 Å apart (Fig 1), a distance that would allow facile juxtapositioning of receptor chains and TM helices as already illustrated for many RTK—cytokine assemblies featuring cytokine-mediated receptor dimerization such as KIT–SCF [10,11], CSF-1R–CSF-1 [12], Flt3–Flt3R [13], VEGF-VEGFR [16], PDGF-PDGFR [15], EGF–EGFR [14], FGF23–FGFR–αKlotho–HS [9], HER2–HER3–NRG1β [25], and Insulin bound to insulin receptor (IR) family receptors [26,27]. The presence of two distinct stoichiometries of the ALK-ALKAL2 complex among the particles imaged in the EMPIAR-10930 dataset is intriguing and their role in ALK signaling is currently unclear. One hypothesis could be that increasing local cytokine concentrations could shift an ALK-ALKAL2 assembly from a 2:1 to a 2:2 stoichiometry, resulting in an increase in the distance between the membrane proximal EGFL domains from 30 Å to 90 Å, possibly leading to different signaling outcomes (S5 Fig). Interestingly, designed ligands for erythropoietin receptor aimed at tuning the distance between receptor monomers upon dimerization result in modulation of erythropoietin receptor signaling and phosphorylation efficiency of downstream adaptors [28].

Full-length wild-type human ALKAL2 can form a dimer in solution via Cys66 in the predicted α-helix of its flexible N-terminal tail, yet monomeric ALKAL2 lacking this tail is as active as its dimeric counterpart [2]. While our study by De Munck and colleagues [5] and the study by Reshetnyak and colleagues [7] used monomeric or mutated (C66Y) ALKAL2, respectively, both 2:1 and 2:2 ALK-ALKAL2 assemblies present in EMPIAR-10930 are structurally compatible with dimeric ALKAL2. In the 2:2 ALK-ALKAL2 assembly, the additional 2 × 68 N-terminal ALKAL2 residues not built in the cryo-EM structure would need to be tacked underneath or along the length of the complex parallel to the cell membrane. On the other hand, in the 2:1 ALK-ALKAL2 assembly, the N-terminal tail would have ample space to flare outwards, allowing its facile linkage to another 2:1 ALK-ALKAL2 complex.

Finally, structural, biophysical, and cellular studies, including full-length ALK-ALKAL2 complexes, will be required to fully define the possible roles of distinct ALK/LTK-cytokine stoichiometries.

Materials and methods

Reanalysis of cryo-EM data as reported in EMPIAR-10930

Initial data analysis was carried out in CryoSPARC [18] v3.3.1, and additional steps were performed in either cryoSPARC [18] v4.5 or RELION [22,23] v5, specifically for the use of Orientation Rebalancing (“Rebalance Orientations” job) and 3D refinement including Blush regularization [19], respectively. Dose-weighted and motion-corrected micrographs were downloaded from https://www.ebi.ac.uk/empiar/EMPIAR-10930/ and were CTF-corrected using patch CTF estimation. After importing the uncleaned particle stack (~18 million particles) from EMPIAR-10930 and extracting the corresponding particles (8× binned), several iterative cycles of reference-free 2D classification were performed to get rid of “junk” particles, resulting in a “cleaned” set comprising 11,849,911 (~11.8 million) particles. This number of particles approaches the total number of particles (~12.4 million) used by Reshetnyak and colleagues [7]. Particle unbinning followed by reference-free 2D classification revealed that almost 2/3 of the 10 most populated classes do not correspond to classes shown in the final 2D classification represented in Reshetnyak and colleagues [7] (Fig 1A). Ab initio models generated on a selected subset of most populated classes (4× binned, 1.5 million particles, 3 Ab Initio classes, C1 symmetry) followed by a round of heterorefinement revealed one model that clearly corresponds to the 2:2 ALK–ALKAL2 structure reported by Reshetnyak and colleagues[7], as well as a model resembling the 2:1 ALK–ALKAL2 structure presented in De Munck and colleagues [5] (Figs 1B and S2). Next, the three Ab Initio classes were used together with the cleaned dataset of ~11.8 million particles (unbinned), as input for heterorefinement (C1, 9 classes). Classes 1 and 2 of this heterorefinement were grouped together, and selected particles (~1.2 million) were used for 3D classification without alignment (8 classes). Class 5 after 3D classification was selected (334,654 particles) and used as an input for a final Non-Uniform (NU) refinement [29], While the final map shows the presence of two ALK copies and one ALKAL2 copy (with its characteristic three central α-helices), the map suffers from the severe preferential particle orientations of the 2:1 ALK–ALKAL2 particles, as evidenced by the paucity of different views in the 2D classes as well as the Azimuth versus Elevation plot shown in S2 Fig. Indeed, Orientation Diagnostics, as implemented in the newest version of cryoSPARC [18] (v4.5), pointed to extreme anisotropy corresponding to a cFAR score of 0.03 (S2 Fig). Accordingly, the anisotropy present in the obtained NU-refined map of the 2:1 ALK_TG-EGFL–ALKAL2 complex prohibited further model building and refinement.

Next, starting from the heterorefinement of 11.849.911 particles, highest resolution 3D classes were grouped in two pools based on the presence or absence of 2-fold symmetry. Further processing of the particles with 2-fold symmetry (523,467 particles) included one additional round of 2D classification to remove remaining junk classes, ab-Initio model generation (2 classes), hetero-refinement and 3D NU-refinement of the highest resolution class while applying C2 symmetry, and resulted in a 2.3 Å resolution map for the 2:2 ALK_TG-EGFL-ALKAL2 complex (Fig 2). Further processing of the particles without apparent symmetry (876,806 particles) by 2D classification showed that 2D classes mainly consisted of side views of the 2:1 ALK-ALKAL2 complex, with a lack of discernible top views (Fig 2). We next used Orientation Rebalancing in cryoSPARC v4.5 to trim overpopulated views in the particle stack using a rebalance percentile threshold of 60, resulting in 599.954 remaining particles. These particles were used as an input for an Ab Initio refinement job (C1, 3 classes) followed by heterorefinement (C1, 3 classes). The highest resolution 3D class (class 1) was used for NU-refinement, followed by another round of Orientation Rebalancing, now using a more stringent rebalance percentile threshold of 5. Retained particles (142.986) where used for a second NU-refinement and subsequently exported to RELION v5 for a final round of 3D refinement with enabled Blush regularization. A final 2D classification was performed on the final particle stack (142.986 particles) to demonstrate the appearance of new side and top views not seen in previous 2D classification runs (Fig 2). Orientation Diagnostics jobs were run on obtained intermediate 3D reconstructions as well as the final map to monitor improvements in map isotropy. The final obtained map has a Fourier Shell Correlation (FSC, computed from independently refined half-maps) resolution of 3.2 Å at the 0.143 threshold (Fig 1, S1 Table) and was post-processed either in RELION using a sharpening B-factor of −100 Å² or by using deepEMhancer [24].

To obtain the ratio of particles in the dataset corresponding to 2:1 or 2:2 ALK-ALKAL2 complexes, a 3D heterorefinement run (CryoSPARC v4.5) was performed in triplicate, using the final maps for the 2:1 and 2:2 ALK-ALKAL2 complexes blurred to 20 Å and cleaned particle stack of 11,849,911 particles as input. Each run was initialized using a different random seed (run 1: 1740971936, run 2: 1817490782, run 3: 150017633).

Model building and refinement

Two copies of ALK_TG-EGFL and one copy of ALKAL2 were extracted from PDB ID 7n00, and rigid-body fitted in the sharpened map of the 2:1 ALK_TG-EGFL-ALKAL2 complex (B-factor: −100 Å²) using USCF Chimera [30]. The resulting rigid-body fitted model was subjected to a cycle of manual building in Coot [31], aided by both the regularly sharpened map and a map sharpened using DeepEMhancer [24]. Next, the manually built and corrected model was real-space refined against the regularly sharpened map in Phenix v1.19.2-4158 [32], using global minimization, local grid search, atomic displacement parameter refinement, secondary structure restraints, and Ramachandran restraints. Several cycles of manual building in Coot were followed by real-space refinement in Phenix. After the last cycle of manual building in Coot, a nonbonded-weight parameter of 200 was used during refinement in Phenix. The resulting model was subjected to a final refinement against the two half-maps in REFMAC-Servalcat [33] run within CCP-EM suite v1.6 [34], with automatic weighting applied. The final model has a map-model FSC of 3.20 Å at the 0.5 threshold, calculated using the unfiltered, unsharpened full map (Fig 1). A summary of cryo-EM data collection, processing, refinement, and validation statistics can be found in S1 Table.

Supporting information

S1 Fig. Comparison of 2:1 ALK/LTK and 2:2 ALK receptor-ligand complexes.

(A) Cartoon views of structures of ALK-ALKAL2 and LTK-ALKAL1 complexes with 2:1 stoichiometry as determined by De Munck and colleagues [5] (PDB entries 7nwz and 7nx0) and (B) of ALK-ALKAL2 with 2:2 stoichiometry as determined by Reshetnyak and colleagues [7] and Li and colleagues [6] (PDB entries 7n00 and 7ls0). Putative positions of the EGFL domains absent in (A) are shown as dotted outlines.

https://doi.org/10.1371/journal.pbio.3003124.s001

(TIF)

S2 Fig. Initial processing workflow of cryo-EM dataset EMPIAR-10930.

(A) All initial processing steps were performed in CryoSPARC [18] v3.3.1. (B) Fourier Shell Correlation (FSC) plot corresponding to the final map obtained after non-uniform refinement (C1 symmetry). Curves are shown after applying no mask (blue), a loose mask (green), or a tight mask (red) to the two half-maps before calculating FSCs. The corrected FSC (purple) is calculated using the tight mask with correction by noise substitution [20], and the resolution at FSC = 0.143 is annotated via a dotted purple line. (C) Conical FSC summary plot generated via “Orientation Diagnostics” in cryoSPARC v4.5. (D) Left: Azimuth plot showing the distribution of orientations over Azimuth (x-axis) and Elevation (y-axis) angles for the particle set corresponding to the NU-refined map shown in (A). Right: Plot showing Relative Signal amount vs. Viewing Direction.

https://doi.org/10.1371/journal.pbio.3003124.s002

(TIF)

S3 Fig. Representative EM map regions for site 1, site 2, and site 3 interfaces of the 2:1 ALK-ALKAL2 complex.

(A) Front (upper view) and top (lower view) views are shown of a transparent map of the 2:1 ALK-ALKAL2 complex with fitted structural model, with ALK colored pink/gray and ALKAL2 colored blue. Insets shown in (B), (C), and (D) are annotated with black squares. (B and C) Insets showing zooms of the site 1 ALK-ALKAL2 (B), site 2 ALK’-ALKAL2 (C), and site 3 ALK-ALK’ (D) interfaces, with the EM map displayed as a dark blue mesh. The fitted 2:1 ALK-ALKAL2 model is shown as a cartoon with selected annotated residues shown as sticks (ALK colored pink/gray and ALKAL2 colored blue).

https://doi.org/10.1371/journal.pbio.3003124.s003

(TIF)

S4 Fig. 3D heterorefinement using a set of 11,849,911 particles and the final 2:1 and 2:2 ALK-ALKAL2 maps blurred to 20 Å as input.

All initial processing steps were performed in CryoSPARC [18] v3.3.1. The three 3D heterorefinement replicate runs, each ran with a different random seed, were performed using CryoSPARC v4.5.

https://doi.org/10.1371/journal.pbio.3003124.s004

(TIF)

S5 Fig. Side-by-side comparison of 2:1 and 2:2 ALK-ALKAL2 complexes present in the EMPIAR-10930 cryo-EM dataset.

Models with pdb code 9g5i (this study) and 7n00 (Reshetnyak et al., 2021) [7] are shown as cartoons with ALK/ALK’ colored pink/gray and ALKAL2 colored blue in the 2:1 ALK-ALKAL2 complex, or blue/purple in the 2:2 ALK-ALKAL2 complex. The distance between the C-termini of membrane-proximal EGFL domains is annotated.

https://doi.org/10.1371/journal.pbio.3003124.s005

(TIF)

S6 Fig. Structural alignments of ALK-ALKAL2 X-ray and cryo-EM models.

(A) Structural alignment of the 2:1 ALK-ALKAL2 X-ray structure (PDB 7nwz) with the cryo-EM structure of the 2:1 ALK-ALKAL2 complex obtained in this study (PDB 9G5I). (B) Structural alignment of a 1:1 ALK-ALKAL2 complex, extracted from the cryo-EM structure of the 2:2 ALK-ALKAL2 complex (PDB 7n00), with the cryo-EM structure of the 2:1 ALK-ALKAL2 complex obtained in this study. (C) Structural alignment of one copy of ALK, extracted from the cryo-EM structure of the 2:2 ALK-ALKAL2 complex (PDB 7n00), with the first ALK molecule (pink) present in the cryo-EM structure of the 2:1 ALK-ALKAL2 complex obtained in this study. (D) Structural alignment of one copy of ALK, extracted from the cryo-EM structure of the 2:2 ALK-ALKAL2 complex (PDB 7n00), with the second ALK molecule (ALK’, gray) present in the cryo-EM structure of the 2:1 ALK-ALKAL2 complex obtained in this study.

https://doi.org/10.1371/journal.pbio.3003124.s006

(TIF)

S1 Table. Cryo-EM data collection, processing, refinement, and validation statistics.

https://doi.org/10.1371/journal.pbio.3003124.s007

(DOCX)

S1 File. Particle coordinates .star file, obtained after final 3D refinement with Blush Regularization in RELION 5.

https://doi.org/10.1371/journal.pbio.3003124.s008

(ZIP)

References

1. Guan J, Umapathy G, Yamazaki Y, Wolfstetter G, Mendoza P, Pfeifer K, et al. FAM150A and FAM150B are activating ligands for anaplastic lymphoma kinase. Elife. 2015;4:e09811. pmid:26418745
- View Article
- PubMed/NCBI
- Google Scholar
2. Reshetnyak AV, Mohanty J, Tomé F, Puleo DE, Plotnikov AN, Ahmed M, et al. Identification of a biologically active fragment of ALK and LTK-Ligand 2 (augmentor-α). Proc Natl Acad Sci USA. 2018;115(33):8340–5. pmid:30061385
- View Article
- PubMed/NCBI
- Google Scholar
3. Reshetnyak AV, et al. Augmentor alpha and beta (FAM150) are ligands of the receptor tyrosine kinases ALK and LTK: hierarchy and specificity of ligand-receptor interactions. Proc Natl Acad Sci U S A. 2015;112(52):15862–7.
- View Article
- Google Scholar
4. Hallberg B, Palmer RH. Mechanistic insight into ALK receptor tyrosine kinase in human cancer biology. Nat Rev Cancer. 2013;13(10):685–700. pmid:24060861
- View Article
- PubMed/NCBI
- Google Scholar
5. De Munck S, Provost M, Kurikawa M, Omori I, Mukohyama J, Felix J, et al. Structural basis of cytokine-mediated activation of ALK family receptors. Nature. 2021;600(7887):143–7. pmid:34646012
- View Article
- PubMed/NCBI
- Google Scholar
6. Li T, Stayrook SE, Tsutsui Y, Zhang J, Wang Y, Li H, et al. Structural basis for ligand reception by anaplastic lymphoma kinase. Nature. 2021;600(7887):148–52. pmid:34819665
- View Article
- PubMed/NCBI
- Google Scholar
7. Reshetnyak AV, Rossi P, Myasnikov AG, Sowaileh M, Mohanty J, Nourse A, et al. Mechanism for the activation of the anaplastic lymphoma kinase receptor. Nature. 2021;600(7887):153–7. pmid:34819673
- View Article
- PubMed/NCBI
- Google Scholar
8. Dornburg A, Wang Z, Wang J, Mo ES, López-Giráldez F, Townsend JP. Comparative genomics within and across bilaterians illuminates the evolutionary history of ALK and LTK proto-oncogene origination and diversification. Genome Biol Evol. 2021;13(1):evaa228. pmid:33196781
- View Article
- PubMed/NCBI
- Google Scholar
9. Chen L, Fu L, Sun J, Huang Z, Fang M, Zinkle A, et al. Structural basis for FGF hormone signalling. Nature. 2023;618(7966):862–70. pmid:37286607
- View Article
- PubMed/NCBI
- Google Scholar
10. Yuzawa S, Opatowsky Y, Zhang Z, Mandiyan V, Lax I, Schlessinger J. Structural basis for activation of the receptor tyrosine kinase KIT by stem cell factor. Cell. 2007;130(2):323–34. pmid:17662946
- View Article
- PubMed/NCBI
- Google Scholar
11. Krimmer S.G., Bertoletti N, Suzuki Y, Katic L, Mohanty J, Shu S, et al.Cryo-EM analyses of KIT and oncogenic mutants reveal structural oncogenic plasticity and a target for therapeutic intervention. Proc Natl Acad Sci U S A, 2023;120(13):e2300054120.
- View Article
- Google Scholar
12. Felix J, De Munck S, Verstraete K, Meuris L, Callewaert N, Elegheert J, et al. Structure and assembly mechanism of the signaling complex mediated by human CSF-1. Structure. 2015;23(9):1621–31. pmid:26235028
- View Article
- PubMed/NCBI
- Google Scholar
13. Verstraete K, Vandriessche G, Januar M, Elegheert J, Shkumatov AV, Desfosses A, et al. Structural insights into the extracellular assembly of the hematopoietic Flt3 signaling complex. Blood. 2011;118(1):60–8. pmid:21389326
- View Article
- PubMed/NCBI
- Google Scholar
14. Huang Y, Ognjenovic J, Karandur D, Miller K, Merk A, Subramaniam S, et al. A molecular mechanism for the generation of ligand-dependent differential outputs by the epidermal growth factor receptor. Elife. 2021;10:e73218. pmid:34846302
- View Article
- PubMed/NCBI
- Google Scholar
15. Chen P-H, Unger V, He X. Structure of full-length human PDGFRβ bound to its activating ligand PDGF-B as determined by negative-stain electron microscopy. J Mol Biol. 2015;427(24):3921–34. pmid:26463591
- View Article
- PubMed/NCBI
- Google Scholar
16. Markovic-Mueller S, Stuttfeld E, Asthana M, Weinert T, Bliven S, Goldie KN, et al. Structure of the full-length VEGFR-1 extracellular domain in complex with VEGF-A. Structure. 2017;25(2):341–52. pmid:28111021
- View Article
- PubMed/NCBI
- Google Scholar
17.. A. Iudin, PK Korir, S. Somasundharam, S. Weyand, C. Cattavitello, N. Fonseca, et al. EMPIAR: the Electron Microscopy Public Image Archive. Nucleic Acids Res. 2022;51(D1): D1503–DD1511.
- View Article
- Google Scholar
18.. Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods. 2017;14(3):290–6. pmid:28165473
- View Article
- PubMed/NCBI
- Google Scholar
19.. Kimanius D, et al. Data-driven regularization lowers the size barrier of cryo-EM structure determination. Nat Methods. 2024.
- View Article
- Google Scholar
20. Chen S, McMullan G, Faruqi AR, Murshudov GN, Short JM, Scheres SHW, et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy. 2013;135:24–35. pmid:23872039
- View Article
- PubMed/NCBI
- Google Scholar
21.. Birkholz N, Kamata K, Feussner M, Wilkinson ME, Cuba Samaniego C, Migur A, et al. Phage anti-CRISPR control by an RNA- and DNA-binding helix-turn-helix protein. Nature. 2024;631(8021):670–7. pmid:38987591
- View Article
- PubMed/NCBI
- Google Scholar
22.. Scheres SHW. RELION: implementation of a Bayesian approach to cryo-EM structure determination. J Struct Biol. 2012;180(3):519–30. pmid:23000701
- View Article
- PubMed/NCBI
- Google Scholar
23.. Kimanius D, Dong L, Sharov G, Nakane T, Scheres SHW. New tools for automated cryo-EM single-particle analysis in RELION-4.0. Biochem J. 2021;478(24):4169–85. pmid:34783343
- View Article
- PubMed/NCBI
- Google Scholar
24.. Sanchez-Garcia R, Gomez-Blanco J, Cuervo A, Carazo JM, Sorzano COS, Vargas J. DeepEMhancer: a deep learning solution for cryo-EM volume post-processing. Commun Biol. 2021;4(1):874. pmid:34267316
- View Article
- PubMed/NCBI
- Google Scholar
25.. Diwanji D, Trenker R, Thaker TM, Wang F, Agard DA, Verba KA, et al. Structures of the HER2-HER3-NRG1β complex reveal a dynamic dimer interface. Nature. 2021;600(7888):339–43. pmid:34759323
- View Article
- PubMed/NCBI
- Google Scholar
26.. Uchikawa E, Choi E, Shang G, Yu H, Bai X-C. Activation mechanism of the insulin receptor revealed by cryo-EM structure of the fully liganded receptor-ligand complex. Elife. 2019;8:e48630. pmid:31436533
- View Article
- PubMed/NCBI
- Google Scholar
27.. Wang L, Hall C, Li J, Choi E, Bai X-C. Structural basis of the alkaline pH-dependent activation of insulin receptor-related receptor. Nat Struct Mol Biol. 2023;30(5):661–9. pmid:37055497
- View Article
- PubMed/NCBI
- Google Scholar
28.. Mohan K, Ueda G, Kim AR, Jude KM, Fallas JA, Guo Y, et al. Topological control of cytokine receptor signaling induces differential effects in hematopoiesis. Science. 2019;364(6442):eaav7532. pmid:31123111
- View Article
- PubMed/NCBI
- Google Scholar
29.. Punjani A, Zhang H, Fleet DJ. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat Methods. 2020;17(12):1214–21. pmid:33257830
- View Article
- PubMed/NCBI
- Google Scholar
30.. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, et al. UCSF Chimera--a visualization system for exploratory research and analysis. J Comput Chem. 2004;25(13):1605–12. pmid:15264254
- View Article
- PubMed/NCBI
- Google Scholar
31.. Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60(Pt 12 Pt 1):2126–32. pmid:15572765
- View Article
- PubMed/NCBI
- Google Scholar
32.. Liebschner D, Afonine PV, Baker ML, Bunkóczi G, Chen VB, Croll TI, et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol. 2019;75(Pt 10):861–77. pmid:31588918
- View Article
- PubMed/NCBI
- Google Scholar
33.. Yamashita K, Palmer CM, Burnley T, Murshudov GN. Cryo-EM single-particle structure refinement and map calculation using Servalcat. Acta Crystallogr D Struct Biol. 2021;77(Pt 10):1282–91. pmid:34605431
- View Article
- PubMed/NCBI
- Google Scholar
34.. Burnley T, Palmer CM, Winn M. Recent developments in the CCP-EM software suite. Acta Crystallogr D Struct Biol. 2017;73(Pt 6):469–77. pmid:28580908
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. Guan J, Umapathy G, Yamazaki Y, Wolfstetter G, Mendoza P, Pfeifer K, et al. FAM150A and FAM150B are activating ligands for anaplastic lymphoma kinase. Elife. 2015;4:e09811. pmid:26418745
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. Reshetnyak AV, Mohanty J, Tomé F, Puleo DE, Plotnikov AN, Ahmed M, et al. Identification of a biologically active fragment of ALK and LTK-Ligand 2 (augmentor-α). Proc Natl Acad Sci USA. 2018;115(33):8340–5. pmid:30061385
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Reshetnyak AV, et al. Augmentor alpha and beta (FAM150) are ligands of the receptor tyrosine kinases ALK and LTK: hierarchy and specificity of ligand-receptor interactions. Proc Natl Acad Sci U S A. 2015;112(52):15862–7.
View Article
Google Scholar

[10] View Article

[11] Google Scholar

[ref4] 4. Hallberg B, Palmer RH. Mechanistic insight into ALK receptor tyrosine kinase in human cancer biology. Nat Rev Cancer. 2013;13(10):685–700. pmid:24060861
View Article
PubMed/NCBI
Google Scholar

[13] View Article

[14] PubMed/NCBI

[15] Google Scholar

[ref5] 5. De Munck S, Provost M, Kurikawa M, Omori I, Mukohyama J, Felix J, et al. Structural basis of cytokine-mediated activation of ALK family receptors. Nature. 2021;600(7887):143–7. pmid:34646012
View Article
PubMed/NCBI
Google Scholar

[17] View Article

[18] PubMed/NCBI

[19] Google Scholar

[ref6] 6. Li T, Stayrook SE, Tsutsui Y, Zhang J, Wang Y, Li H, et al. Structural basis for ligand reception by anaplastic lymphoma kinase. Nature. 2021;600(7887):148–52. pmid:34819665
View Article
PubMed/NCBI
Google Scholar

[21] View Article

[22] PubMed/NCBI

[23] Google Scholar

[ref7] 7. Reshetnyak AV, Rossi P, Myasnikov AG, Sowaileh M, Mohanty J, Nourse A, et al. Mechanism for the activation of the anaplastic lymphoma kinase receptor. Nature. 2021;600(7887):153–7. pmid:34819673
View Article
PubMed/NCBI
Google Scholar

[25] View Article

[26] PubMed/NCBI

[27] Google Scholar

[ref8] 8. Dornburg A, Wang Z, Wang J, Mo ES, López-Giráldez F, Townsend JP. Comparative genomics within and across bilaterians illuminates the evolutionary history of ALK and LTK proto-oncogene origination and diversification. Genome Biol Evol. 2021;13(1):evaa228. pmid:33196781
View Article
PubMed/NCBI
Google Scholar

[29] View Article

[30] PubMed/NCBI

[31] Google Scholar

[ref9] 9. Chen L, Fu L, Sun J, Huang Z, Fang M, Zinkle A, et al. Structural basis for FGF hormone signalling. Nature. 2023;618(7966):862–70. pmid:37286607
View Article
PubMed/NCBI
Google Scholar

[33] View Article

[34] PubMed/NCBI

[35] Google Scholar

[ref10] 10. Yuzawa S, Opatowsky Y, Zhang Z, Mandiyan V, Lax I, Schlessinger J. Structural basis for activation of the receptor tyrosine kinase KIT by stem cell factor. Cell. 2007;130(2):323–34. pmid:17662946
View Article
PubMed/NCBI
Google Scholar

[37] View Article

[38] PubMed/NCBI

[39] Google Scholar

[ref11] 11. Krimmer S.G., Bertoletti N, Suzuki Y, Katic L, Mohanty J, Shu S, et al.Cryo-EM analyses of KIT and oncogenic mutants reveal structural oncogenic plasticity and a target for therapeutic intervention. Proc Natl Acad Sci U S A, 2023;120(13):e2300054120.
View Article
Google Scholar

[41] View Article

[42] Google Scholar

[ref12] 12. Felix J, De Munck S, Verstraete K, Meuris L, Callewaert N, Elegheert J, et al. Structure and assembly mechanism of the signaling complex mediated by human CSF-1. Structure. 2015;23(9):1621–31. pmid:26235028
View Article
PubMed/NCBI
Google Scholar

[44] View Article

[45] PubMed/NCBI

[46] Google Scholar

[ref13] 13. Verstraete K, Vandriessche G, Januar M, Elegheert J, Shkumatov AV, Desfosses A, et al. Structural insights into the extracellular assembly of the hematopoietic Flt3 signaling complex. Blood. 2011;118(1):60–8. pmid:21389326
View Article
PubMed/NCBI
Google Scholar

[48] View Article

[49] PubMed/NCBI

[50] Google Scholar

[ref14] 14. Huang Y, Ognjenovic J, Karandur D, Miller K, Merk A, Subramaniam S, et al. A molecular mechanism for the generation of ligand-dependent differential outputs by the epidermal growth factor receptor. Elife. 2021;10:e73218. pmid:34846302
View Article
PubMed/NCBI
Google Scholar

[52] View Article

[53] PubMed/NCBI

[54] Google Scholar

[ref15] 15. Chen P-H, Unger V, He X. Structure of full-length human PDGFRβ bound to its activating ligand PDGF-B as determined by negative-stain electron microscopy. J Mol Biol. 2015;427(24):3921–34. pmid:26463591
View Article
PubMed/NCBI
Google Scholar

[56] View Article

[57] PubMed/NCBI

[58] Google Scholar

[ref16] 16. Markovic-Mueller S, Stuttfeld E, Asthana M, Weinert T, Bliven S, Goldie KN, et al. Structure of the full-length VEGFR-1 extracellular domain in complex with VEGF-A. Structure. 2017;25(2):341–52. pmid:28111021
View Article
PubMed/NCBI
Google Scholar

[60] View Article

[61] PubMed/NCBI

[62] Google Scholar

[ref17.] 17.. A. Iudin, PK Korir, S. Somasundharam, S. Weyand, C. Cattavitello, N. Fonseca, et al. EMPIAR: the Electron Microscopy Public Image Archive. Nucleic Acids Res. 2022;51(D1): D1503–DD1511.
View Article
Google Scholar

[64] View Article

[65] Google Scholar

[ref18.] 18.. Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods. 2017;14(3):290–6. pmid:28165473
View Article
PubMed/NCBI
Google Scholar

[67] View Article

[68] PubMed/NCBI

[69] Google Scholar

[ref19.] 19.. Kimanius D, et al. Data-driven regularization lowers the size barrier of cryo-EM structure determination. Nat Methods. 2024.
View Article
Google Scholar

[71] View Article

[72] Google Scholar

[ref20] 20. Chen S, McMullan G, Faruqi AR, Murshudov GN, Short JM, Scheres SHW, et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy. 2013;135:24–35. pmid:23872039
View Article
PubMed/NCBI
Google Scholar

[74] View Article

[75] PubMed/NCBI

[76] Google Scholar

[ref21.] 21.. Birkholz N, Kamata K, Feussner M, Wilkinson ME, Cuba Samaniego C, Migur A, et al. Phage anti-CRISPR control by an RNA- and DNA-binding helix-turn-helix protein. Nature. 2024;631(8021):670–7. pmid:38987591
View Article
PubMed/NCBI
Google Scholar

[78] View Article

[79] PubMed/NCBI

[80] Google Scholar

[ref22.] 22.. Scheres SHW. RELION: implementation of a Bayesian approach to cryo-EM structure determination. J Struct Biol. 2012;180(3):519–30. pmid:23000701
View Article
PubMed/NCBI
Google Scholar

[82] View Article

[83] PubMed/NCBI

[84] Google Scholar

[ref23.] 23.. Kimanius D, Dong L, Sharov G, Nakane T, Scheres SHW. New tools for automated cryo-EM single-particle analysis in RELION-4.0. Biochem J. 2021;478(24):4169–85. pmid:34783343
View Article
PubMed/NCBI
Google Scholar

[86] View Article

[87] PubMed/NCBI

[88] Google Scholar

[ref24.] 24.. Sanchez-Garcia R, Gomez-Blanco J, Cuervo A, Carazo JM, Sorzano COS, Vargas J. DeepEMhancer: a deep learning solution for cryo-EM volume post-processing. Commun Biol. 2021;4(1):874. pmid:34267316
View Article
PubMed/NCBI
Google Scholar

[90] View Article

[91] PubMed/NCBI

[92] Google Scholar

[ref25.] 25.. Diwanji D, Trenker R, Thaker TM, Wang F, Agard DA, Verba KA, et al. Structures of the HER2-HER3-NRG1β complex reveal a dynamic dimer interface. Nature. 2021;600(7888):339–43. pmid:34759323
View Article
PubMed/NCBI
Google Scholar

[94] View Article

[95] PubMed/NCBI

[96] Google Scholar

[ref26.] 26.. Uchikawa E, Choi E, Shang G, Yu H, Bai X-C. Activation mechanism of the insulin receptor revealed by cryo-EM structure of the fully liganded receptor-ligand complex. Elife. 2019;8:e48630. pmid:31436533
View Article
PubMed/NCBI
Google Scholar

[98] View Article

[99] PubMed/NCBI

[100] Google Scholar

[ref27.] 27.. Wang L, Hall C, Li J, Choi E, Bai X-C. Structural basis of the alkaline pH-dependent activation of insulin receptor-related receptor. Nat Struct Mol Biol. 2023;30(5):661–9. pmid:37055497
View Article
PubMed/NCBI
Google Scholar

[102] View Article

[103] PubMed/NCBI

[104] Google Scholar

[ref28.] 28.. Mohan K, Ueda G, Kim AR, Jude KM, Fallas JA, Guo Y, et al. Topological control of cytokine receptor signaling induces differential effects in hematopoiesis. Science. 2019;364(6442):eaav7532. pmid:31123111
View Article
PubMed/NCBI
Google Scholar

[106] View Article

[107] PubMed/NCBI

[108] Google Scholar

[ref29.] 29.. Punjani A, Zhang H, Fleet DJ. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat Methods. 2020;17(12):1214–21. pmid:33257830
View Article
PubMed/NCBI
Google Scholar

[110] View Article

[111] PubMed/NCBI

[112] Google Scholar

[ref30.] 30.. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, et al. UCSF Chimera--a visualization system for exploratory research and analysis. J Comput Chem. 2004;25(13):1605–12. pmid:15264254
View Article
PubMed/NCBI
Google Scholar

[114] View Article

[115] PubMed/NCBI

[116] Google Scholar

[ref31.] 31.. Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60(Pt 12 Pt 1):2126–32. pmid:15572765
View Article
PubMed/NCBI
Google Scholar

[118] View Article

[119] PubMed/NCBI

[120] Google Scholar

[ref32.] 32.. Liebschner D, Afonine PV, Baker ML, Bunkóczi G, Chen VB, Croll TI, et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol. 2019;75(Pt 10):861–77. pmid:31588918
View Article
PubMed/NCBI
Google Scholar

[122] View Article

[123] PubMed/NCBI

[124] Google Scholar

[ref33.] 33.. Yamashita K, Palmer CM, Burnley T, Murshudov GN. Cryo-EM single-particle structure refinement and map calculation using Servalcat. Acta Crystallogr D Struct Biol. 2021;77(Pt 10):1282–91. pmid:34605431
View Article
PubMed/NCBI
Google Scholar

[126] View Article

[127] PubMed/NCBI

[128] Google Scholar

[ref34.] 34.. Burnley T, Palmer CM, Winn M. Recent developments in the CCP-EM software suite. Acta Crystallogr D Struct Biol. 2017;73(Pt 6):469–77. pmid:28580908
View Article
PubMed/NCBI
Google Scholar

[130] View Article

[131] PubMed/NCBI

[132] Google Scholar

Figures

Abstract

Introduction and results

Discussion

Materials and methods

Reanalysis of cryo-EM data as reported in EMPIAR-10930

Model building and refinement

Supporting information

S1 Fig. Comparison of 2:1 ALK/LTK and 2:2 ALK receptor-ligand complexes.

S2 Fig. Initial processing workflow of cryo-EM dataset EMPIAR-10930.

S3 Fig. Representative EM map regions for site 1, site 2, and site 3 interfaces of the 2:1 ALK-ALKAL2 complex.

S4 Fig. 3D heterorefinement using a set of 11,849,911 particles and the final 2:1 and 2:2 ALK-ALKAL2 maps blurred to 20 Å as input.

S5 Fig. Side-by-side comparison of 2:1 and 2:2 ALK-ALKAL2 complexes present in the EMPIAR-10930 cryo-EM dataset.

S6 Fig. Structural alignments of ALK-ALKAL2 X-ray and cryo-EM models.

S1 Table. Cryo-EM data collection, processing, refinement, and validation statistics.

S1 File. Particle coordinates .star file, obtained after final 3D refinement with Blush Regularization in RELION 5.

References