We present crystal structures of a trimeric coiled-coil domain found in echinoderm microtubule (MT)-associated protein (EMAP)-like (EML) proteins. This trimerization domain (TD) mediates self-association and interactions between a subset of EML proteins. MT-association of EML proteins requires the TD and an adjacent basic region.

INTRODUCTION

Microtubules (MTs) are polymers of tubulin heterodimers that form a dynamic network as part of the eukaryotic cytoskeleton. They carry out many essential cellular processes including forming an intracellular transport system for RNAs, proteins and organelles and regulating cell motility and changes in cell shape [1,2]. In mitosis, MTs form the spindle apparatus that segregates chromosomes into two equal sets. The dynamic nature of MTs, their stochastic growth and shrinkage, is crucial for their function and is regulated by MT-associated proteins (MAPs) [3].

The present work is focused on the echinoderm MT-associated protein (EMAP)-like (EML) family of MAPs. The archetypal member of this family was identified as the most abundant MAP in dividing echinoderm eggs and embryos [4]. Further members of the EML family associate with MTs and are essential for the proper formation of both the interphase MT network and the mitotic spindle [58]. EML proteins affect MT dynamics and may act as scaffold proteins to localize mitotic kinases to MTs [5,9,10]. Drosophila and Caenorhabditis elegans each have a single EML homologue, whereas six EML proteins are present in humans of which EML1–4 have molecular masses in the 70–120 kDa range whereas EML5 and 6 are in excess of 200 kDa. The EML proteins contain WD40 repeats and in EML1–4 these fall within a ∼70 kDa core region that begins with a conserved ∼60 amino acid sequence that has been termed the ‘hydrophobic EML protein’ (HELP) motif [5]. We recently determined the crystal structure of the core HELP–WD region of human EML1, which forms a tandem atypical β-propeller (TAPE) domain that binds soluble tubulin [11]. Most splice variants of EML1–4 are also predicted to contain a coiled-coil in the N-terminal region.

Rearrangements in the short arm of chromosome 2 leading to genetic fusions between EML4 and the gene encoding anaplastic lymphoma kinase (ALK) occur in ∼5% of cases of non-small cell lung cancer (NSCLC) [12]. Multiple variants of the EML4–ALK fusion have been identified in NSCLC resulting from translocations at different points within the EML4 gene [13]. A similar genetic fusion has also been reported between EML1 and the gene encoding another tyrosine kinase Abelson 1 (ABL1) in T-cell acute lymphoblastic leukaemia [14]. These fusions have potent transforming activity due to constitutive activation of the tyrosine kinase but confer addiction to the oncogene and inhibition of the tyrosine kinase induces apoptosis in transformed cells [15,16]. The transforming activity of EML4–ALK variant 1 is dependent on the N-terminal region and this is proposed to be due to oligomerization of ALK [12]. Indeed, this hypothesis appears to be confirmed by the identification of EML4–ALK variant 5, which encodes just the N-terminal region up to the end of the predicted coiled-coil [17].

In the present study, we determined the crystal structure of the N-terminal region of EML proteins that is required to drive the transforming capability of EML4–ALK fusions and reveal a trimeric state of oligomerization. We then investigated the role of this coiled-coil in MT binding in both EML proteins and oncogenic EML–kinase fusions.

MATERIALS AND METHODS

Gene manipulation

Human EML cDNAs were cloned by PCR from human cDNA (Clontech). Products encoding EML4 6–64 and EML2 isoform 2 11–60 were amplified by PCR and cloned into a modified version of pET30. Constructs were made comprising EML1 residues 167–815 (TAPE domain), 1–174 (N-terminus), 23–78 [trimerization domain (TD)] and 80–815 (Δ-TD). The EML1 AAAA mutant contained the mutations L59A and D61A. Cloning of EML1–ABL1 and EML4–ALK variants was described previously [11]. Site-directed mutagenesis was carried out by the Quikchange procedure (Stratagene).

Protein expression and purification

Histidine-tagged EML2 11–60 and EML4 6–64 I38M were expressed in B834 cells grown in Luria broth and SelenoMet Medium (Molecular Dimensions) respectively. The proteins were purified from clarified lysate in 50 mM HEPES, pH 7.5, 500 mM NaCl using 5 ml of a nickel-charged chelating Sepharose column and eluted on a gradient from 50 to 500 mM imidazole. Histidine-tags were removed by digestion with tobacco etch virus (TEV) protease and the protein passed through a second chelating Sepharose column. Finally, the proteins were gel filtrated into 10 mM HEPES, pH 7.5, 80 mM NaCl, 2 mM DTT on a Superdex 75 16/60 column (GE Healthcare).

Crystal structure determination

Crystals of EML4 6–64 I38M and EML2 11–60 were obtained by the hanging drop vapour diffusion method using 100 mM Tris/HCl, pH 8.5, 30% PEG 4000, 200 mM MgCl2 and 100 mM Bis-Tris-propane, pH 6.5, 20% PEG 3350, 200 mM potassium thiocyanate respectively in the reservoir and were flash frozen in the same solutions with the addition of 20% glycerol and 5 mM DTT. Drops were formed from ∼15 mg/ml protein solution mixed 1:1 with reservoir solution and incubated at 18°C for ∼2 days.

Single-wavelength anomalous dispersion data to a resolution of 2.9 Å (1 Å=0.1 nm) was collected from an EML4 6–64 I38M selenomethionine derivative crystal at Diamond I02. Diffraction images were indexed using imosflm [18] and scaled using scala [19]. An initial map and model were obtained using PHENIX to locate selenium sites [20] revealing one trimer in the asymmetric unit. Manual rebuilding was carried out using COOT [21]. The 2.1 Å data was collected from a native EML2 11–60 crystal at Diamond I04. PHENIX was used to solve the structure of EML2 11–60 by molecular replacement phasing with the EML4 6–64 I38M structure, to automatically build the two trimers in the asymmetric unit and for refinement.

In vitro MT-binding assays

Tubulin was prepared from pig brain as described previously [22]. Labelled tubulin was from Cytoskeleton Inc. YFP–EML1 proteins for use in in vitro MT-binding assays were expressed as FLAG–Strep2 fusions in human embryonic kidney (HEK)293F suspension culture and purified from clarified cell lysate on Strep-tactin Sepharose (Qiagen) in TBS and eluted using 3 mM desthiobiotin (IBA), followed by incubation with GST-tagged 3C protease to cleave off the tag and then passed over a glutathione–sepharose column to remove the protease. Finally, the protein was separated from co-purifying tubulin by gel filtration on a Superose 6 column (GE Healthcare) in 20 mM Tris/HCl, pH 7.5, 80 mM NaCl, 2 mM DTT before dialysis into MRB80 (80 mM PIPES, pH 6.8, with KOH, 1 mM EGTA, 4 mM MgCl2)+50 mM KCl and concentration to 20 μM. MT seeds were assembled from tubulin, biotin–tubulin and Hilyte647–tubulin at a molar ratio of 25:1:2 in the presence of 1 mM GMP–CPP (Guanosine-5′-[(α,β)-methyleno] triphosphate) in MRB80 for 1 h at 37°C, diluted 20-fold with MRB80+2 μM Taxol and stored at 21°C. A 100 μm deep flow chamber was made from a slide and an HCl-treated coverslip using double-sided Scotch tape (3M) and passivated with poly(L-lysine)–PEG–50% biotin (Susos AG). Seeds were attached to this surface using streptavidin and blocked with 1 mg/ml κ-casein (Sigma). A reaction mix containing 12 μM tubulin, 1 μM rhodamine–tubulin, 50 mM KCl, 1 mM GTP, 0.6 mg/ml κ-casein, 0.2% methyl cellulose, 4 mM DTT, 0.2 mg/ml catalase, 0.4 mg/ml glucose oxidase, 50 mM glucose in MRB80, supplemented with 500 nM YFP–EML1 protein or buffer was clarified for 8 min at 190000 g in an airfuge (Beckman), the supernatant added to the flow chamber and sealed with candle wax. For MT-binding experiments in the absence of free tubulin the reaction mixture was prepared as above but lacking tubulin and assays carried out using GMP–CPP stabilized MTs.

MTs were observed on an Olympus TIRF (total internal reflection fluorescence) system with a 100× numerical aperture 1.49 objective, 488, 561 and 640 nm laser lines and a Hamamatsu ImageEM-1k back-illuminated EM–CCD (charge coupled device) camera under the control of xcellence software (Olympus). To quantify MT-bound YFP–EML1 proteins the average YFP-fluorescence intensity along a 3 pixel wide line along the MT was measured using ImageJ.

Cell culture and transfection

HeLa, HEK293 and U2OS cells were grown in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% heat-inactivated FBS, 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C in a 5% CO2 atmosphere. EML constructs, EML1–ABL1 and EML4–ALK variants, were cloned into pcDNA3 or pcDNA3.1-hygro (Invitrogen) with N-terminal YFP- or FLAG-tags. HeLa and HEK293 cells were transiently transfected with these plasmids using Fugene HD reagent (Promega) and U2OS cells using Lipofectamine 2000 (Invitrogen).

HEK293F cells were grown in suspension culture in Freestyle 293 medium (Gibco) in a 5% CO2 atmosphere and transfected with Strep2–YFP–EML1 constructs in a modified version of pMAX (Lonza) at a density of 1×106 cells/ml using 2 μg/ml polyethylenimine.

NCI-H2228 cells were provided by Dr S Gray (Institute of Molecular Medicine, St James's Hospital, Dublin, Ireland); NCI-H3122 cells were obtained from the NCI Tumor Repository. Cells were cultured in Roswell Park Memorial Institute (RPMI) medium 1640, 2 mM glutamax and 10% foetal calf serum (Gibco).

Fixed and live cell immunofluorescence (IF) microscopy

For fixed cell microscopy, HeLa, H2228 and H3122 cells grown on acid-etched coverslips were fixed with ice-cold methanol and processed for IF microscopy as described previously [23], with the modification that cells were blocked for 30 min in 1× PBS supplemented with 1% BSA and 0.2% Triton X-100. Primary antibodies used were 1 μg/ml anti-GFP ab6556 (Abcam), 0.3 μg/ml anti-α-tubulin (Sigma) and 1:200 anti-ALK D5F3 (Cell Signaling). Secondary antibodies were 1 μg/ml Alexa Fluor 488 conjugated goat anti-rabbit and Alexa Fluor 594 conjugated goat anti-mouse IgGs (Invitrogen). Cells were imaged using a Leica SP5 laser scanning confocal microscope. Deconvolution of image stacks was carried out using Huygens Essential software and the z-stacks were assembled in ImageJ. Quantitative image analysis was performed with LAS-AF software (Leica) or ImageJ.

Western blot analysis

Primary antibodies for Western blotting were mouse anti-FLAG M2 (Sigma) and rabbit anti-GFP ab6556 (Abcam). Secondary antibodies were horseradish peroxidase conjugated anti-mouse or anti-rabbit IgGs (GE Healthcare).

RESULTS

A conserved region in the EML N-terminus forms a trimerization domain

Proteins of the EML family contain regions N-terminal to the conserved TAPE core that are poorly conserved, often subject to splice variation and predicted to be mostly disordered. However, there is an island of conserved primary sequence within this region that is consistent with a coiled-coil (Figure 1A). To investigate the structural basis of self-association, we crystallized protein fragments corresponding to this part of human EML2 and EML4 proteins. The structure was determined by single-wavelength anomalous dispersion phasing using crystals of a selenomethionine-labelled EML4 6–64 I38M mutant (Table 1; Figure 1B; Supplementary Figure S1A). Only residues 14–44 were visible in the electron density map but most strikingly they formed a trimeric coiled-coil. We then obtained 2.1 Å diffraction data using crystals of EML2 isoform 2 residues 11–60 and solved its structure by molecular replacement phasing (Table 1; Figure 1C; Supplementary Figure S1B).

The N-termini of EML proteins contain a trimerization domain

Figure 1
The N-termini of EML proteins contain a trimerization domain

(A) Schematic diagram of domain organization in EML1. (B) A view along the three-fold axis of the EML4 TD. In (BD), side chains forming the hydrophobic core of the trimer are shown in yellow and those forming salt-bridges are shown in light blue. (C) A view along the three-fold axis of the EML2 TD. (D) The structure of the EML2 TD shown as a cartoon representation (pink) with key residues in two of the protomers shown as stick representations. Gln29 is shown in grey. Below, the arrows linking the pair of sequences indicate salt-bridge pairing in adjacent protomers. (E) A sequence alignment of the TD region of human EML proteins. Residues are coloured as in (BD) and boxes indicate the regions present in the EML2 and EML4 TD structures. Conservation is shown underneath, identical residues and highly conserved residues are marked by (*) and (:) respectively. Gln29 and the ALAD motif are underlined.

Figure 1
The N-termini of EML proteins contain a trimerization domain

(A) Schematic diagram of domain organization in EML1. (B) A view along the three-fold axis of the EML4 TD. In (BD), side chains forming the hydrophobic core of the trimer are shown in yellow and those forming salt-bridges are shown in light blue. (C) A view along the three-fold axis of the EML2 TD. (D) The structure of the EML2 TD shown as a cartoon representation (pink) with key residues in two of the protomers shown as stick representations. Gln29 is shown in grey. Below, the arrows linking the pair of sequences indicate salt-bridge pairing in adjacent protomers. (E) A sequence alignment of the TD region of human EML proteins. Residues are coloured as in (BD) and boxes indicate the regions present in the EML2 and EML4 TD structures. Conservation is shown underneath, identical residues and highly conserved residues are marked by (*) and (:) respectively. Gln29 and the ALAD motif are underlined.

The EML TD forms an amphipathic α-helix with a series of nine conserved hydrophobic residues (predominantly leucine and valine) along the inner face whose side chains pack together into the core of the trimer (coloured yellow in Figure 1). A conserved glutamine (Gln29 in EML2, coloured grey in Figure 1D) on the inner face makes hydrogen bond interactions with main-chain carbonyls in the core, perhaps serving to maintain the register of the coiled-coil. Interspersed between the hydrophobic residues in the sequence are conserved charged residues whose side chains are positioned to form salt-bridges with those of the adjacent protomers around the outside of the trimer (coloured light blue in Figure 1). The pattern of conserved residues indicates that human EML1–4 are all trimeric, as are their respective oncogenic fusion proteins (Figure 1E).

The trimerization domain is critical for EML self-association

The most conserved surface patch in the TD is composed of several amino acids (aa) with short side chains (Ala-Leu-Ala-Asp, ALAD motif), which constrict the shape of the coiled-coil (Figure 2A). We investigated whether the TD or disruption of the ALAD motif affect self-association in a representative member of the EML family, EML1. We tested the ability of various YFP-tagged EML1 constructs to co-immunoprecipitate with FLAG-tagged full-length EML1 from lysates of co-transfected HEK293 cells (Figure 2B). Full-length EML1 (FL), the isolated N-terminus (1–174) and a construct corresponding to residues 23–78 (TD) co-immunoprecipitated with full-length EML1 whereas the TAPE region did not. A double mutation (L59A, D61A) in the most conserved motif in the TD (AAAA) did not disrupt self-association but deletion of the whole conserved region (Δ-TD) abrogated self-association completely.

Oligomerization of EML1 depends on the TD

Figure 2
Oligomerization of EML1 depends on the TD

(A) Sequence conservation mapped on to the surface of the EML2 TD structure; residues that are identical among human EML proteins are coloured red and highly conserved residues are coloured orange. (B) Immunoblot analysis of EML1 self-association. HEK293 cells were co-transfected with FLAG–EML1 and various YFP–EML1 constructs as indicated (top two panels). Proteins were immunoprecipitated using anti-FLAG M2 (Sigma; lower two panels). Cell extracts were prepared as previously described [26] and analysed by SDS/PAGE and Western blotting using anti-FLAG M2 and anti-GFP antibodies. (C) EML3 association with EML1-3. U2OS cells were co-transfected with YFP–EML3 and FLAG–EML1-3 as indicated. Immunoprecipitation and analysis was carried out as in (B).

Figure 2
Oligomerization of EML1 depends on the TD

(A) Sequence conservation mapped on to the surface of the EML2 TD structure; residues that are identical among human EML proteins are coloured red and highly conserved residues are coloured orange. (B) Immunoblot analysis of EML1 self-association. HEK293 cells were co-transfected with FLAG–EML1 and various YFP–EML1 constructs as indicated (top two panels). Proteins were immunoprecipitated using anti-FLAG M2 (Sigma; lower two panels). Cell extracts were prepared as previously described [26] and analysed by SDS/PAGE and Western blotting using anti-FLAG M2 and anti-GFP antibodies. (C) EML3 association with EML1-3. U2OS cells were co-transfected with YFP–EML3 and FLAG–EML1-3 as indicated. Immunoprecipitation and analysis was carried out as in (B).

The key residues involved in self-association within the TD are imperfectly conserved and EML3 has the most divergent TD (Figure 1E). This observation prompted us to investigate whether EML3 could self-associate and whether associations between EML3 and other members of the EML family could happen. We tested the ability of YFP–EML3 to co-immunoprecipitate with FLAG–EML1, FLAG–EML2 or FLAG–EML3 from lysates of co-transfected HEK293 cells (Figure 2C). YFP–EML3 co-immunoprecipitated with FLAG–EML3 and, more weakly, with FLAG–EML2, but not with YFP–EML1. This is consistent with the degree of conservation within the 17 key residues involved in self-association: EML1 and EML2, 12 identical, five conservatively substituted; EML1 and EML3, six identical, five conservatively substituted, five non-conserved; EML2 and EML3, nine identical, three conservatively substituted, five non-conserved.

The trimerization domain is critical for MT binding

We confirmed co-localization of YFP-tagged human EML1 to the interphase MT network of transiently transfected HeLa cells, as has been previously observed for endogenous EML proteins (Figure 3) [58]. Overexpression of a tagged protein might affect its regulation and distribution on MTs and we shall not comment on these features of the data. However, this approach enabled us to screen the MT-binding properties of multiple YFP–EML1 constructs to identify which domains were needed for this activity (Figure 3A). The co-localization of YFP–EML1 proteins with MTs was visualized (Figure 3B) and quantified (Figure 3C).

MT association of EML1

Figure 3
MT association of EML1

(A) Schematic illustrations of EML1 constructs. (B) Immunofluorescence microscopy images of interphase HeLa cells transfected with YFP-tagged EML1 constructs (z-projections; lower image) and of a magnified single MT (one focal plane; upper image). Cells are stained with Hoechst 33258 (blue, DNA) and anti-GFP (green) and anti-α-tubulin (red, MTs) antibodies. Scale bar, 10 μm. (C) Representative intensity profiles along a line crossing 2–3 MTs showing co-localization of YFP-fusions with MTs. R (Pearson's correlation coefficient) measures the correlation between the YFP and α-tubulin signals and is given as the mean of single line-scan measurements in 10 different cells for each construct±S.D.

Figure 3
MT association of EML1

(A) Schematic illustrations of EML1 constructs. (B) Immunofluorescence microscopy images of interphase HeLa cells transfected with YFP-tagged EML1 constructs (z-projections; lower image) and of a magnified single MT (one focal plane; upper image). Cells are stained with Hoechst 33258 (blue, DNA) and anti-GFP (green) and anti-α-tubulin (red, MTs) antibodies. Scale bar, 10 μm. (C) Representative intensity profiles along a line crossing 2–3 MTs showing co-localization of YFP-fusions with MTs. R (Pearson's correlation coefficient) measures the correlation between the YFP and α-tubulin signals and is given as the mean of single line-scan measurements in 10 different cells for each construct±S.D.

The isolated N-terminal region (aa 1–174), which is not conserved at the sequence level for most of its length but includes the TD, strongly associated with MTs. However, neither the putatively disordered region lying between the TD and TAPE domains (74–174) nor the TD alone (23–78) nor the isolated TAPE domain was sufficient to associate with MTs in cells. Deletion of the TD from EML1 considerably reduced its interaction with MTs in cells, whereas mutation of the ALAD motif within the TD decreased MT co-localization modestly. The overexpressed EML1 proteins might interact directly with MTs, through association with endogenous EML1 or other EMLs or indeed through associations with other MAPs. To distinguish between these different modes of interaction, direct binding of recombinantly expressed and purified YFP–EML1 protein constructs to purified MTs was measured in vitro by TIRF microscopy (Figure 4). Here, direct association of full-length EML1 and the N-terminal region with MTs was measured, whereas the isolated TAPE domain did not associate (Figures 4A and 4C); full-length EML1 with a mutated ALAD motif exhibited reduced binding compared with wild-type and the deletion of the TD abolished MT interaction altogether (Figures 4B and 4D).

Association of EML1 with MTs assembled from purified tubulin

Figure 4
Association of EML1 with MTs assembled from purified tubulin

(A) Representative TIRF microscopy fields of view showing Hilyte640-labelled GMP–CPP-stabilized MT seeds (left), XRhodamine-labelled dynamic MTs extensions (centre) and the binding to those MTs of 500 nM YFP–EML1 proteins as indicated in the presence of 12 μM free tubulin (right). Note even decoration with no preference for nucleotide state of tubulin. (B) TIRF images of Hilyte640-labelled GMP–CPP-stabilized MTs (left) and their decoration by 500 nM YFP-tagged EML1 mutants in the absence of free tubulin (right). Scale bars 10 μm. (C) Statistical box plots describing quantification of MT-binding of YFP–EML1 proteins in vitro by TIRF microscopy in the presence of 500 nM YFP-tagged EML1 domains and 12 μM free tubulin. n=25–38 MTs. (D) Statistical box plots describing quantification of MT-binding of 500 nM YFP-tagged EML1 mutants without free tubulin. n=30–201 MTs. Boxes in (C and D) show twenty-fifth, fiftieth and seventy-fifth percentile, whiskers show tenth and ninetieth percentile of data. a.u., arbitrary units.

Figure 4
Association of EML1 with MTs assembled from purified tubulin

(A) Representative TIRF microscopy fields of view showing Hilyte640-labelled GMP–CPP-stabilized MT seeds (left), XRhodamine-labelled dynamic MTs extensions (centre) and the binding to those MTs of 500 nM YFP–EML1 proteins as indicated in the presence of 12 μM free tubulin (right). Note even decoration with no preference for nucleotide state of tubulin. (B) TIRF images of Hilyte640-labelled GMP–CPP-stabilized MTs (left) and their decoration by 500 nM YFP-tagged EML1 mutants in the absence of free tubulin (right). Scale bars 10 μm. (C) Statistical box plots describing quantification of MT-binding of YFP–EML1 proteins in vitro by TIRF microscopy in the presence of 500 nM YFP-tagged EML1 domains and 12 μM free tubulin. n=25–38 MTs. (D) Statistical box plots describing quantification of MT-binding of 500 nM YFP-tagged EML1 mutants without free tubulin. n=30–201 MTs. Boxes in (C and D) show twenty-fifth, fiftieth and seventy-fifth percentile, whiskers show tenth and ninetieth percentile of data. a.u., arbitrary units.

Taken together, the results from the cell-based and in vitro MT-binding assays demonstrate that the N-terminal region of EML1 confers MT binding and that the TD is required. However, the TD is not sufficient for MT association, which requires other sequences in the N-terminal region.

EML4–ALK variant 3 localizes to MTs

Having identified the N-terminus as a MT-binding region of EML proteins and recognizing that it is intact in many of the fusion oncoproteins, we supposed that the fusion oncoproteins might be targeted to MTs. We examined HeLa cells transfected with YFP-tagged EML–tyrosine kinase fusions (Figures 5A–5C) and found that this was not generally the case: EML4–ALK variants 1 and 2 and EML1–ABL1 form cytoplasmic aggregates and EML4–ALK variant 5a is diffusely localized in the cytoplasm, whereas only EML4–ALK variant 3a was localized to MTs. Next, we investigated by immunofluorescence microscopy the localization of the endogenous EML4–ALK fusion proteins in two patient-derived NSCLC cell lines, H3122 and H2228 (Figures 5D–5F). The variant 1 protein present in H3122 cells displayed a patchy appearance in the cytoplasm consistent with a protein that is largely aggregated but the variant 3b protein in H2228 cells was localized to the MT network.

Localization of EML4–ALK and EML1–ABL1 fusion proteins

Figure 5
Localization of EML4–ALK and EML1–ABL1 fusion proteins

(A) Schematic illustrations of EML–tyrosine kinase fusion constructs. (B) Immunofluorescence microscopy images of interphase HeLa cells transfected with YFP-tagged EML–tyrosine kinase fusion constructs as indicated, (lower image) and of a magnified single MT (upper image). Scale bar, 10 μm. (C) Co-localization of each YFP-fusion with MTs was quantified as for Figure 3C. (D) Schematic illustrations of EML–tyrosine kinase fusion constructs in H3122 and H2228. (E) Immunofluorescence microscopy images of interphase NSCLC cells (lower image) and of a magnified single MT (upper image). Scale bar, 10 μm. (F) Co-localization of ALK with MTs was quantified as for Figure 3C.

Figure 5
Localization of EML4–ALK and EML1–ABL1 fusion proteins

(A) Schematic illustrations of EML–tyrosine kinase fusion constructs. (B) Immunofluorescence microscopy images of interphase HeLa cells transfected with YFP-tagged EML–tyrosine kinase fusion constructs as indicated, (lower image) and of a magnified single MT (upper image). Scale bar, 10 μm. (C) Co-localization of each YFP-fusion with MTs was quantified as for Figure 3C. (D) Schematic illustrations of EML–tyrosine kinase fusion constructs in H3122 and H2228. (E) Immunofluorescence microscopy images of interphase NSCLC cells (lower image) and of a magnified single MT (upper image). Scale bar, 10 μm. (F) Co-localization of ALK with MTs was quantified as for Figure 3C.

Taken together, these results show that the N-terminal domain of EML4 is able to confer MT localization on EML4–ALK variant 3, but not when a truncated part of the EML4 TAPE domain is present as in variants 1 and 2.

DISCUSSION

Association of the EML1 N-terminal region with MTs requires both the TD and the ∼90 residue region between the TD and TAPE domains. This latter, putatively disordered, inter-domain region is not conserved at the sequence level but contains many basic residues. The TD presumably enhances binding of EML proteins to MTs through an avidity effect and may contribute directly to the interaction through a conserved surface region that is disrupted by the AAAA mutation.

EML oncogenic fusions such as EML4–ALK and EML1–ABL1 are naturally occurring and exist in a context in which the MT-binding properties of the EML N-terminus are isolated from the TAPE region. In EML4–ALK variant 3a/b the whole of the EML4 N-terminal region and evidently its MT-binding activity, is transferred to the fusion protein (Figure 6). EML4–ALK variant 5 has a TD, but lacks the basic region and is therefore incapable of binding to MTs. The other EML4–ALK variants and EML1–ABL1 contain a complete N-terminus encompassing the TD and basic region and might be expected to localize to MTs, but do not. In these cases, it appears that the presence of a partial TAPE domain interferes with MT binding, perhaps through protein misfolding leading to aggregation into complexes with Hsp90 that may sterically block MT association. Indeed, the microscopy studies of several of the EML–kinase fusions, including endogenous EML4–ALK variant 1 in H3122 cells, were consistent with aggregation of these proteins. Furthermore, we have previously shown that EML4–ALK variants 1 and 2 are more sensitive to Hsp90 inhibitors than variants 3 and 5 [11].

Models summarizing the properties of EML proteins and EML4–ALK variants

Figure 6
Models summarizing the properties of EML proteins and EML4–ALK variants

(A) Cartoon illustrating trimerization and MT-binding in EML proteins. (B) Cartoon illustrating trimerization and MT binding in EML4–ALK variant 3. The constituent β-propellers of the EML TAPE domain are shown in teal and green and the Blade 12-N subdomain is shown in orange [11].

Figure 6
Models summarizing the properties of EML proteins and EML4–ALK variants

(A) Cartoon illustrating trimerization and MT-binding in EML proteins. (B) Cartoon illustrating trimerization and MT binding in EML4–ALK variant 3. The constituent β-propellers of the EML TAPE domain are shown in teal and green and the Blade 12-N subdomain is shown in orange [11].

Table 1
Summary of crystallographic analysis
 EML4 TD I38M EML2 TD 
Lattice parameters   
Space group P43 21P21 21 21 
 a (Å) 67.49 47.87 
 b (Å) 67.49 51.14 
 c (Å) 50.60 101.69 
 α (°) 90.00 90.00 
 β (°)  90.00 90.00 
 γ (°) 90.00 90.00 
Data collection   
X-ray source Diamond I02 Diamond I04 
Wavelength (Å) 0.9796 0.9795 
Resolution range (Å) 67.49-2.90 50.86-2.15 
 (3.06-2.90)* (2.32-2.15) 
No. unique reflections 2864 (394) 14593 (1058) 
Completeness (%) 100.0 (100.0) 99.8 (99.9) 
Redundancy 13.4 (14.4) 7.0 (7.3) 
Rmerge (%) 13 (125) 8.5 (63.8) 
Rpim (%) 4.1 (33.2) 3.8 (27.2) 
II 13.3 (3.1) 14.2 (3.2) 
Phasing (figures of merit)   
Before density modification 0.39 – 
After density modification 0.74 – 
Refinement   
Resolution range (Å) 47.7-2.90 50.9-2.15 
Rwork/Rfree (%) 26.87/29.15 20.78/25.15 
No. protein molecules 
No. amino acids 83 254 
No. water molecules 102 
No. hetero molecules 
Mean B-factors (Å2  
Protein 88.75 43.30 
Water – 46.05 
Hetero – 50.31 
RMSDs   
Bond lengths (Å) 0.006 0.006 
Bond angles (°) 1.02 0.930 
MolProbity analysis   
All-atom clash-score 17.12 8.97 
Poor rotamers (%) 8.8 5.1 
Ramachandran outliers (%) 0.0 0.0 
Ramachandran favoured (%) 100.0 99.6 
MolProbity score 1.99 1.93 
 EML4 TD I38M EML2 TD 
Lattice parameters   
Space group P43 21P21 21 21 
 a (Å) 67.49 47.87 
 b (Å) 67.49 51.14 
 c (Å) 50.60 101.69 
 α (°) 90.00 90.00 
 β (°)  90.00 90.00 
 γ (°) 90.00 90.00 
Data collection   
X-ray source Diamond I02 Diamond I04 
Wavelength (Å) 0.9796 0.9795 
Resolution range (Å) 67.49-2.90 50.86-2.15 
 (3.06-2.90)* (2.32-2.15) 
No. unique reflections 2864 (394) 14593 (1058) 
Completeness (%) 100.0 (100.0) 99.8 (99.9) 
Redundancy 13.4 (14.4) 7.0 (7.3) 
Rmerge (%) 13 (125) 8.5 (63.8) 
Rpim (%) 4.1 (33.2) 3.8 (27.2) 
II 13.3 (3.1) 14.2 (3.2) 
Phasing (figures of merit)   
Before density modification 0.39 – 
After density modification 0.74 – 
Refinement   
Resolution range (Å) 47.7-2.90 50.9-2.15 
Rwork/Rfree (%) 26.87/29.15 20.78/25.15 
No. protein molecules 
No. amino acids 83 254 
No. water molecules 102 
No. hetero molecules 
Mean B-factors (Å2  
Protein 88.75 43.30 
Water – 46.05 
Hetero – 50.31 
RMSDs   
Bond lengths (Å) 0.006 0.006 
Bond angles (°) 1.02 0.930 
MolProbity analysis   
All-atom clash-score 17.12 8.97 
Poor rotamers (%) 8.8 5.1 
Ramachandran outliers (%) 0.0 0.0 
Ramachandran favoured (%) 100.0 99.6 
MolProbity score 1.99 1.93 
*

Values in parentheses are for the highest-resolution shell.

The EML2 and EML4 TD structures presented here reveal that EML proteins and EML4–ALK oncoproteins self-associate as trimers through a specific and stable coiled-coil interface. The more widely expressed isoform 1 of EML2 does not possess a TD region and is therefore unlikely to interact directly with other EMLs but a TD is present in the brain/spinal cord variant isoform 2 studied in [24]. EML3 can also self-associate and can interact with EML2 isoform 2, but not with EML1. The functional relevance of interactions between EML family members is not clear, but the potential for these interactions should be borne in mind during functional studies.

It is notable that the more divergent human paralogues, EML5 and EML6, lack coiled-coil regions but have three copies of the TAPE region each. Thus, the quaternary structure of all six human EML proteins may feature three TAPE regions (Figure 6A). In oncogenic ALK-fusions oligomerization conferred by the fusion partner mimics the native oligomerization-dependent activation mechanism of receptor tyrosine kinases and engenders constitutive activation of ALK [25]. The use of rationally-designed peptide and peptidomimetic inhibitors is under investigation for the disruption of several coiled-coil interactions that are implicated in disease and clinically useful therapeutics have been developed using this approach [2729]. The TD structures may enable the development of protein–protein interaction inhibitors targeting the trimerization interface of the EML4 TD, another possible avenue towards therapeutic intervention in EML4–ALK NSCLC and one that would be applicable to all variants.

Abbreviations

     
  • aa

    amino acids

  •  
  • ABL1

    Abelson 1

  •  
  • ALK

    anaplastic lymphoma kinase

  •  
  • EMAP

    echinoderm microtubule-associated protein

  •  
  • EML

    echinoderm microtubule-associated protein-like

  •  
  • HEK

    human embryonic kidney

  •  
  • HELP

    hydrophobic EML protein

  •  
  • MAP

    microtubule-associated protein

  •  
  • MT

    microtubule

  •  
  • NSLC

    non-small cell lung cancer

  •  
  • TAPE

    tandem atypical β-propeller

  •  
  • TD

    trimerization domain

  •  
  • TIRF

    total internal reflection fluorescence

AUTHOR CONTRIBUTION

Mark Richards made the DNA constructs, protein samples and crystals and determined the crystal structures. Laura O'Regan carried out immunofluorescence microscopy and analysed the resulting data, whereas immunoprecipitation experiments were performed and analysed by Laura O'Regan and Jessica Montgomery. Daniel Roth and Anne Straube carried out in vitro MT-binding experiments and analysed the resulting data. Mark Richards, Richard Bayliss, Anne Straube and Andrew Fry wrote the manuscript.

We thank the beamline support staff at DLS I02 and I04 for assistance with data collection and the Centre for Core Biotechnology Services (University of Leicester) for imaging support.

FUNDING

This work was supported by the Royal Society University Research Fellowship [grant number UF090021 (to R.B.)]; the Cancer Research UK programme [grant number C24461/A12772 (to R.B.)];the Marie Curie Cancer Care programme [(to A.S.)]; the Lister Institute Research Prize [(to A.S.)]; the Worldwide Cancer Research [grant number 13-0042 (to A.M.F.)]; and the Wellcome Trust [grant number 082828 (to A.M.F.)].

References

References
1
Hirokawa
N.
Noda
Y.
Tanaka
Y.
Niwa
S.
Kinesin superfamily motor proteins and intracellular transport
Nat. Rev. Mol. Cell Biol.
2009
, vol. 
10
 (pg. 
682
-
696
)
[PubMed]
2
Kaverina
I.
Straube
A.
Regulation of cell migration by dynamic microtubules
Semin. Cell Dev. Biol.
2011
, vol. 
22
 (pg. 
968
-
974
)
[PubMed]
3
van der Vaart
B.
Akhmanova
A.
Straube
A.
Regulation of microtubule dynamic instability
Biochem. Soc. Trans.
2009
, vol. 
37
 (pg. 
1007
-
1013
)
[PubMed]
4
Suprenant
K.A.
Dean
K.
McKee
J.
Hake
S.
EMAP, an echinoderm microtubule-associated protein found in microtubule-ribosome complexes
J. Cell Sci.
1993
, vol. 
104
 (pg. 
445
-
450
)
[PubMed]
5
Eichenmuller
B.
Everley
P.
Palange
J.
Lepley
D.
Suprenant
K.A.
The human EMAP-like protein-70 (ELP70) is a microtubule destabilizer that localizes to the mitotic apparatus
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
1301
-
1309
)
[PubMed]
6
Houtman
S.H.
Rutteman
M.
De Zeeuw
C.I.
French
P.J.
Echinoderm microtubule-associated protein like protein 4, a member of the echinoderm microtubule-associated protein family, stabilizes microtubules
Neuroscience
2007
, vol. 
144
 (pg. 
1373
-
1382
)
[PubMed]
7
Pollmann
M.
Parwaresch
R.
Adam-Klages
S.
Kruse
M.L.
Buck
F.
Heidebrecht
H.J.
Human EML4, a novel member of the EMAP family, is essential for microtubule formation
Exp. Cell. Res.
2006
, vol. 
312
 (pg. 
3241
-
3251
)
[PubMed]
8
Tegha-Dunghu
J.
Neumann
B.
Reber
S.
Krause
R.
Erfle
H.
Walter
T.
Held
M.
Rogers
P.
Hupfeld
K.
Ruppert
T.
, et al. 
EML3 is a nuclear microtubule-binding protein required for the correct alignment of chromosomes in metaphase
J. Cell Sci.
2008
, vol. 
121
 (pg. 
1718
-
1726
)
[PubMed]
9
Brisch
E.
Daggett
M.A.
Suprenant
K.A.
Cell cycle-dependent phosphorylation of the 77 kDa echinoderm microtubule-associated protein (EMAP) in vivo and association with the p34cdc2 kinase
J. Cell Sci.
1996
, vol. 
109
 
Pt 12
(pg. 
2885
-
2893
)
[PubMed]
10
Hamill
D.R.
Howell
B.
Cassimeris
L.
Suprenant
K.A.
Purification of a WD repeat protein, EMAP, that promotes microtubule dynamics through an inhibition of rescue
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
9285
-
9291
)
[PubMed]
11
Richards
M.W.
Law
E.W.
Rennalls
L.P.
Busacca
S.
O'Regan
L.
Fry
A.M.
Fennell
D.A.
Bayliss
R.
Crystal structure of EML1 reveals the basis for Hsp90 dependence of oncogenic EML4-ALK by disruption of an atypical beta-propeller domain
Proc. Natl. Acad. Sci. U.S.A.
2014
, vol. 
111
 (pg. 
5195
-
5200
)
[PubMed]
12
Soda
M.
Choi
Y.L.
Enomoto
M.
Takada
S.
Yamashita
Y.
Ishikawa
S.
Fujiwara
S.
Watanabe
H.
Kurashina
K.
Hatanaka
H.
, et al. 
Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer
Nature
2007
, vol. 
448
 (pg. 
561
-
566
)
[PubMed]
13
Sasaki
T.
Rodig
S.J.
Chirieac
L.R.
Janne
P.A.
The biology and treatment of EML4-ALK non-small cell lung cancer
Eur. J. Cancer.
2010
, vol. 
46
 (pg. 
1773
-
1780
)
[PubMed]
14
De Keersmaecker
K.
Graux
C.
Odero
M.D.
Mentens
N.
Somers
R.
Maertens
J.
Wlodarska
I.
Vandenberghe
P.
Hagemeijer
A.
Marynen
P.
Cools
J.
Fusion of EML1 to ABL1 in T-cell acute lymphoblastic leukemia with cryptic t(9;14)(q34;q32)
Blood
2005
, vol. 
105
 (pg. 
4849
-
4852
)
[PubMed]
15
Choi
Y.L.
Takeuchi
K.
Soda
M.
Inamura
K.
Togashi
Y.
Hatano
S.
Enomoto
M.
Hamada
T.
Haruta
H.
Watanabe
H.
, et al. 
Identification of novel isoforms of the EML4-ALK transforming gene in non-small cell lung cancer
Cancer Res.
2008
, vol. 
68
 (pg. 
4971
-
4976
)
[PubMed]
16
Takezawa
K.
Okamoto
I.
Nishio
K.
Janne
P.A.
Nakagawa
K.
Role of ERK-BIM and STAT3-survivin signaling pathways in ALK inhibitor-induced apoptosis in EML4-ALK-positive lung cancer
Clin. Cancer Res.
2011
, vol. 
17
 (pg. 
2140
-
2148
)
[PubMed]
17
Takeuchi
K.
Choi
Y.L.
Soda
M.
Inamura
K.
Togashi
Y.
Hatano
S.
Enomoto
M.
Takada
S.
Yamashita
Y.
Satoh
Y.
, et al. 
Multiplex reverse transcription-PCR screening for EML4-ALK fusion transcripts
Clin. Cancer Res.
2008
, vol. 
14
 (pg. 
6618
-
6624
)
[PubMed]
18
Battye
T.G.
Kontogiannis
L.
Johnson
O.
Powell
H.R.
Leslie
A.G.
iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
271
-
281
)
[PubMed]
19
Collaborative Computational Project, Number 4
The CCP4 suite: programs for protein crystallography
Acta Crystallogr. D Biol. Crystallogr.
1994
, vol. 
50
 (pg. 
760
-
763
)
[PubMed]
20
Adams
P.D.
Grosse-Kunstleve
R.W.
Hung
L.W.
Ioerger
T.R.
McCoy
A.J.
Moriarty
N.W.
Read
R.J.
Sacchettini
J.C.
Sauter
N.K.
Terwilliger
T.C.
PHENIX: building new software for automated crystallographic structure determination
Acta Crystallogr. D Biol. Crystallogr.
2002
, vol. 
58
 (pg. 
1948
-
1954
)
[PubMed]
21
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2126
-
2132
)
[PubMed]
22
Gell
C.
Friel
C.T.
Borgonovo
B.
Drechsel
D.N.
Hyman
A.A.
Howard
J.
Purification of tubulin from porcine brain
Methods Mol. Biol.
2011
, vol. 
777
 (pg. 
15
-
28
)
[PubMed]
23
O'Regan
L.
Fry
A.M.
The Nek6 and Nek7 protein kinases are required for robust mitotic spindle formation and cytokinesis
Mol. Cell. Biol.
2009
, vol. 
29
 (pg. 
3975
-
3990
)
[PubMed]
24
Lepley
D.M.
Palange
J.M.
Suprenant
K.A.
Sequence and expression patterns of a human EMAP-related protein-2 (HuEMAP-2)
Gene
1999
, vol. 
237
 (pg. 
343
-
349
)
[PubMed]
25
Roskoski
R.
Jr
Anaplastic lymphoma kinase (ALK): structure, oncogenic activation, and pharmacological inhibition
Pharmacol. Res.
2013
, vol. 
68
 (pg. 
68
-
94
)
[PubMed]
26
Hames
R.S.
Crookes
R.E.
Straatman
K.R.
Merdes
A.
Hayes
M.J.
Faragher
A.J.
Fry
A.M.
Dynamic recruitment of Nek2 kinase to the centrosome involves microtubules, PCM-1, and localized proteasomal degradation
Mol. Biol. Cell.
2005
, vol. 
16
 (pg. 
1711
-
1724
)
[PubMed]
27
Wild
C.T.
Shugars
D.C.
Greenwell
T.K.
McDanal
C.B.
Matthews
T.J.
Peptides corresponding to a predictive alpha-helical domain of human immunodeficiency virus type 1 gp41 are potent inhibitors of virus infection
Proc. Natl. Acad. U.S.A.
1994
, vol. 
91
 (pg. 
9770
-
9774
)
28
Dixon
A.S.
Pendley
S.S.
Bruno
B.J.
Woessner
D.W.
Shimpi
A.A.
Cheatham
T.E.
III
Lim
C.S.
Disruption of Bcr-Abl coiled-coil oligomerization by design
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
27751
-
27760
)
[PubMed]
29
Rao
T.
Ruiz-Gómez
G.
Hill
T.A.
Hoang
H.N.
Fairlie
D.P.
Mason
J.M.
Truncated and helix-constrained peptides with high affinity and specificity for the coiled-coil of AP-1
PLoS One
2013
, vol. 
8
 (pg. 
e59415
-
e59415
)
[PubMed]

Author notes

1

These authors contributed equally to this work.

The structural co-ordinates reported for EML2 11-60 and EML4 6-64 I38M appear in the PDB under codes 4CGB and 4CGC respectively.

Supplementary data