2.1. Materials

For protein expression, Luria–Bertani (LB) medium, ampicillin, isopropyl β-D-1-thiogalactopyranoside (IPTG), Tris, NaCl, EDTA, DTT and glutathione were obtained from Sigma–Aldrich, Darmstadt, Germany, Escherichia coli BL21 (DE3) cells were acquired from Novagen, Darmstadt, Germany, protease-inhibitor tablets were purchased from Roche, DNase I was purchased from Sigma–Aldrich, Darmstadt, Germany and Pre-Scission Protease (PSS) was obtained from Cytiva, Uppsala, Sweden.

For protein purification, Glutathione Sepharose 4B resin and an ÄKTA go system were obtained from Cytiva, Uppsala, Sweden. The HiLoad 16/600 Superdex 200 prep-grade column was from GE Healthcare, Barcelona, Spain. Protein fractions were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). Protein concentration was carried out using 30 kDa concentrators from Millipore, Madrid, Spain.

For protein crystallization, HEPES and polyethylene glycol 4000 were purchased from Sigma–Aldrich, Darmstadt, Germany, isopropyl alcohol was purchased from Quimipur, Madrid, Spain, glycerol was purchased from Fisher Chemical, Madrid, Spain, liquid N₂ was supplied by Linde Gas, Valencia, Spain, 96-well crystallization plates were purchased from Sigma–Aldrich, Darmstadt, Germany and the Milli-Q Type I Ultrapure Water Production System was obtained from Merck Millipore, Darmstadt, Germany.

2.2. Expression of human fascin1

Human fascin1 fused to glutathione S-transferase (GST) was expressed according to the protocol established by Huang et al. (2018) with some modifications. A starter of 5 ml LB medium supplemented with ampicillin was inoculated with 5 µl of an E. coli BL21 (DE3) culture previously transformed with pGEX-6P-2 plasmid and grown overnight at 37°C. The next day, the starter was transferred to 1 l LB medium supplemented with ampicillin and incubated at 37°C until the optical density at 600 nm (OD₆₀₀) reached a value of between 0.6 and 0.8. The culture was then induced with 0.5 mM IPTG and incubated overnight at 25°C. The cells were subsequently harvested by centrifugation at 3300g for 15 min. The cell pellets were resuspended in 10 ml PBS buffer (140 mm NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄ pH 7.4), flash-frozen with liquid N₂ and stored at −80°C.

2.3. Purification of human fascin1

Purification of GST-human fascin1 was carried out following the protocol previously reported by Sedeh et al. (2010) with some variations. Frozen PBS-resuspended cells were thawed and one protease-inhibitor tablet and 5 µl DNase were added. The cells were disrupted by sonication using three cycles of 2 min each, alternating 2 s on and 2 s off, with 2 min resting on ice. The lysate was clarified by ultracentrifugation at 803 000g at 4°C for 30 min. The supernatant was then added to Glutathione Sepharose 4B resin previously equilibrated with PBS buffer. The fusion protein was eluted with 10 mM glutathione, 50 mM Tris pH 8.0 and dialyzed overnight against 50 mM Tris, 150 mM NaCl, 1 mM EDTA, 1 mM DTT pH 7.0. Human fascin1 was liberated from the GST tag by a 4 h cleavage with Pre-Scission Protease (PSS, Cytiva) and subsequent Glutathione Sepharose affinity chromatography. The purity of all protein fractions throughout the purification process was monitored using SDS–PAGE (Supplementary Figs. S1a and S1b). Purified human fascin1 was finally dialyzed extensively against 20 mM HEPES pH 7.5, 100 mM NaCl, flash-frozen with liquid N₂ and stored at −80°C.

An additional purification step was required to obtain the protein microcrystals (see Section 2.5). This was performed using size-exclusion chromatography on a HiLoad 16/600 Superdex 200 prep-grade column with an elution buffer composed of 20 mM HEPES pH 7.5, 100 mM NaCl (Supplementary Fig. S1c). The pure protein was concentrated to 10 mg ml⁻¹, flash-frozen with liquid N₂ and stored at −80°C.

2.4. Crystallization of human fascin1

Crystallization of the human fascin1 protein was carried out using the sitting-drop vapor-diffusion method in 24-well plates. In outline, 1 µl protein solution at 15–20 mg ml⁻¹ was mixed with 1 µl precipitant solution composed of 0.1 M HEPES pH 7.5, 20% polyethylene glycol (PEG) 4000, 2% 2-propanol. Protein droplets were equilibrated against 500 µl precipitant solution in the reservoir at 20°C. Prior to data collection, the crystals were soaked in a cryoprotectant solution consisting of the precipitant solution supplemented with 20% glycerol, looped, cryocooled in liquid N₂, stored in a dry-shipper and shipped to the synchrotron-radiation facility for data collection.

2.5. Data collection and structure determination

X-ray diffraction data collection was performed on the BL-13 XALOC beamline at the ALBA synchrotron-radiation source, Barcelona, Spain using a wavelength of 0.98 Å and a PILATUS X 6M detector. Indexing of the collected data sets was performed with XDS (Kabsch, 2010), and scaling and merging were performed with AIMLESS (Evans & Murshudov, 2013) from the CCP4 suite (Agirre et al., 2023). A 5% fraction of reflections was included in the generated R_free set. Phasing was performed by employing molecular replacement with MOLREP (Vagin & Teplyakov, 2010) using the crystal structure of human fascin1 with PDB code 3llp (Chen et al., 2010) as the search model. The obtained model was refined with REFMAC5 (Murshudov et al., 2011), with iterative cycles of automated refinement and manual rebuilding in Coot (Emsley et al., 2010). Noncrystallographic symmetry (NCS) restraints were applied, and translation–libration–screw (TLS) refinement was implemented from the outset, with each β-trefoil domain defined as an independent TLS group to account for structural modularity and domain flexibility. This approach was used to improve the geometry while minimizing overfitting. Water molecules were automatically upgraded after each refinement cycle. The final refined structure was validated using the wwwPDB Validation Service and submitted to the PDB for deposition as PDB entry 9fn6. All data-collection and refinement statistics are summarized in Table 1. All figures showing the protein structure presented in this work were generated with PyMOL (version 3.1.3; Schrödinger).

2.6. Microcrystallization assays

Microcrystallization of human fascin1 was performed by optimizing the crystallization conditions used to obtain the large crystals described in Section 2.4. A range of crystallization conditions were assayed by varying the concentration and molecular weight of PEG, the protein concentration and the protein:precipitant ratio, as well as the incorporation of additives along with the use of seeding. Crystallization was carried out using the batch method, where the protein solution is quickly added to and mixed with the precipitant solution. Seeding was employed to increase the crystal density and size homogeneity while reducing the crystallization time. The seeds were obtained by fragmenting one-day-old large crystals (100–200 µm) grown at 18°C using the hanging-drop vapor-diffusion method under the aforementioned crystallization conditions. The best needle-shaped microcrystals, measuring approximately 25–30 µm in their longest dimension, were obtained overnight using a precipitant solution consisting of 0.1 M HEPES pH 7.5, 30% PEG 4000, 2% 2-propanol, 50 mM Li₂SO₄, with 1 µl of seed stock added per 100 µl of precipitant. Crystallization was performed at a protein concentration of 10 mg ml⁻¹, using a 1:3 protein:precipitant ratio and incubating at 18°C. The seed stock was prepared by crushing large crystals (100–200 µm) grown over 24 h at 18°C by hanging-drop vapor diffusion using a similar precipitant solution (0.1 M HEPES pH 7.5, 20% PEG 4000, 2% 2-propanol, 50 mM Li₂SO₄) with a protein concentration of 10 mg ml⁻¹ and a 1:3 protein:precipitant ratio. All crystal samples were analyzed using high-resolution optical microscopy.

3.1. Crystal structure of the full-length human fascin1 protein

Pure protein from the last purification step was subjected to crystallization. The best human fascin1 crystals (100 µm; Fig. 2a) were obtained in 0.1 M HEPES pH 7.0, 20% PEG 4000, 2% propanol using the hanging-drop vapor-diffusion method. The crystals belonged to space group I121, with unit-cell parameters a = 114.57, b = 70.80, c = 122.21 Å, α = 90.00, β = 92.78, γ = 90.00°. The crystals were observed to diffract beyond 2.0 Å resolution (Fig. 2b). A total of 2400 images were collected and successfully indexed, integrated and merged. Two molecules of human fascin1, chains A and B, were found in the asymmetric unit, yielding a Matthews coefficient (V_M) of 2.23 A³ Da⁻¹ and a solvent content of 44.98%. The structure of human fascin1 was solved by molecular replacement using PDB entry 3llp (Chen et al., 2010), from which the solvent molecules had been previously removed, as a search model. The structure was refined to a final resolution of 2.2 Å, with R_work and R_free values of 23.4% and 30.4%, respectively. The final data-collection and refinement statistics are given in Table 1.

Figure 2 Crystallization and diffraction of the human fascin1 protein. (a) Large human fascin1 crystals obtained by the sitting-drop vapor-diffusion method. (b) A representative diffraction pattern from the crystals obtained in (a).

All 493 residues of human fascin1 were manually inspected, and loop regions lacking initial density were built de novo in Coot (Emsley et al., 2010) based on the electron-density maps. The model spans from the N-terminus to the C-terminus without interruptions, except for the first seven N-terminal residues, which could not be modeled due to the absence of interpretable density. The resulting structure shows the typical fascin1 fold containing four β-trefoil domains (residues 8–139 for β-trefoil domain 1, residues 140–260 for β-trefoil domain 2, residues 261–381 for β-trefoil domain 3 and residues 382–493 for β-trefoil domain 4). Fig. 3(a) illustrates the molecule of human fascin1 corresponding to chain A in the asymmetric unit. The stereochemistry of the structure was excellent, with root-mean-square deviations (r.m.s.d.s) from the ideal of 0.012 Å for bond lengths and less than 1.823° for bond angles (Table 1). The Ramachandran diagram places most residues within the favored (91%) and allowed (7%) regions, and only 2% of the residues fall into the outlier region (Table 1). The resulting experimental maps revealed the presence of 308 water molecules. The high quality of our structure can also be assessed from the 2mF_o − DF_c electron-density map of the entire protein (Fig. 3b), as well as those for the ABS1 region (Fig. 3c for chain A and Supplementary Fig. S2a for chain B) and the regions of the protein that were not modeled in previous crystal structures (Figs. 3d and 3e for chain A and Supplementary Figs. S2b and S2c for chain B).

Figure 3 Crystal structure of the full-length human fascin1 protein. (a) Cartoon representation of molecule A in the asymmetric unit. β-Trefoil domain 1 is shown in magenta, β-trefoil domain 2 is shown in orange, β-trefoil domain 3 is shown in green and β-trefoil domain 4 is shown in blue. (b) 2mFo − DFc electron-density map contoured at 1σ of the whole protein. The dashed black box highlights the ABS1 region. (c) 2mFo − DFc electron-density map contoured at 1σ of the residues comprising ABS1 of chain A highlighted in (b). (d) 2mFo − DFc electron-density map contoured at 1σ of residues 49–56 in β-trefoil domain 1. (e) 2mFo − DFc electron-density map contoured at 1σ of residues 274–280 in β-trefoil domain 3.

3.2. Comparison of the human fascin1 structure with related structures

Currently, three structures of human fascin1 in its wild-type form have been deposited in the PDB [PDB entries 1dfc (Sedeh et al., 2010), 3llp (Chen et al., 2010) and 3p53 (Jansen et al., 2011)]. However, due to the exceptionally high flexibility reported in the literature for this protein, several regions in β-trefoil domains 1 and 3 were often missing in one of the two molecules within the asymmetric unit. The crystal structure presented in this study is, to the best of our knowledge, the first full-length structure of human fascin1 in which all residues of the two molecules of the asymmetric unit have been modeled fully. As an example, Figs. 3(d) and 3(e) highlight two highly flexible regions, residues 49–56 in β-trefoil domain 1 and residues 274–280 in β-trefoil domain 3, which have been fully modeled for the first time in the two molecules of the asymmetric unit of our structure.

To further investigate the flexibility of human fascin1, we carried out a structural alignment of the two molecules in the asymmetric unit and observed that even though the r.m.s.d. value indicates that the two molecules are similar (1.13 Å), a visual inspection of the superimposition shows that there are significant differences between them (Fig. 4). These differences are mostly found in loop regions, especially in the regions comprising residues 49–60 in β-trefoil domain 1, residues 274–281 in β-trefoil domain 3 and residues 395–405 in β-trefoil domain 4, which have been modeled in a different conformation in each molecule of the asymmetric unit (Fig. 4). It is important to mention that the electron density around the residues of these two regions is clearly visible and the backbone was well modeled. In addition to this, we also compared the four individual β-trefoil domains with each other for the two molecules found in the asymmetric unit (Supplementary Table S1). Supplementary Fig. S3 shows a comparison of the individual β-trefoil domains with each other and Supplementary Table S1 collects the overall r.m.s.d.s from this comparison. Interestingly, β-trefoil domains 1 and 3, which have been reported to be somewhat interconnected (Lamb & Tootle, 2020; Huang et al., 2018; Sedeh et al., 2010; Chen et al., 2010; Jansen et al., 2011), show the lowest r.m.s.d., indicating that these two domains of human fascin1 are structurally similar compared with domains 2 and 4.

Figure 4 Structural comparison. Superimposition of the two molecules A (β-trefoil domain 1 in magenta, β-trefoil domain 2 in orange, β-trefoil domain 3 in green and β-trefoil domain 4 in blue) and B (β-trefoil domain 1 in pink, β-trefoil domain 2 in gray, β-trefoil domain 3 in yellow and β-trefoil domain 4 in purple) in the asymmetric unit of the human fascin1 structure presented in this study. Regions 1, 2 and 3, consisting of residues 49–60 in β-trefoil domain 1, 274–281 in β-trefoil domain 3 and 395–405 in β-trefoil domain 4, respectively, are enclosed in dashed boxes. A closer view of each region is also provided.

Further evaluation of the human fascin1 structure was carried out by comparing it with previously reported crystal structures in its free form [PDB entries 1dfc (Sedeh et al., 2010), 3llp (Chen et al., 2010) and 3p53 (Jansen et al., 2011)]. Overall, all human fascin1 structures aligned well with each other, with r.m.s.d values for all atoms of between 0.65 and 1.70 Å and an average value of 1.23 Å. Supplementary Fig. S4 shows the superimposition of our structure with all free human fascin1 structures. As expected, the larger structural differences are mainly found in the loop regions and solvent-exposed areas, as reflected by the higher r.m.s.d. values. However, there are several regions in which significant conformational changes are observed (residues 49–59, 156–162, 274–280, 298–305 and 397–404). Supplementary Fig. S4 also shows a closer view of these flexible regions. This highlights the importance of our full-length structure in obtaining a complete picture of the protein folding, which was missing in other related structures.

In addition to these local differences within the individual β-trefoil domains, high variability is also observed when comparing their relative orientations. This is illustrated in Fig. 5, where the different structures available for free fascin1 are compared. In this figure, the F2 domains of all structures have been superposed in order to highlight the `breathing' dynamics of the protein. As can be observed, the orientation of β-trefoil domain 1 and most of β-trefoil domain 3 varies significantly, even though the internal structures of the domains remain mostly invariable, as described before. Interestingly, while molecule A in our structure lies within the conformational dynamics previously observed in structures of free wild-type fascin1 [PDB entries 3llp (Chen et al., 2010) and 3p53 (Jansen et al., 2011)] and the inactive S39D mutant (PDB entry 4gov; Yang et al., 2013), molecule B captures larger conformational departures, clearly illustrating the high plasticity of free fascin1 (Fig. 5).

Figure 5 Structural variability within crystal structures of the free human fascin1 protein. Ribbon representations of molecule A (left panel) and molecule B (right panel) from various crystal structures of human fascin1. Structures shown include wild-type fascin1 from the present study (yellow for molecule A, green for molecule B), wild-type fascin1 from PDB entry 3llp (Chen et al., 2010; black), wild-type fascin1 from PDB entry 3p53 (Jansen et al., 2011; red) and the inactive S39D mutant from PDB entry 4gov (Yang et al., 2013; blue). To highlight conformational differences and domain flexibility, all structures were superposed based on β-trefoil domain 2. The comparisons reveal variations in domain orientations that underline the structural plasticity of fascin1 and its functional regulation.

3.3. Analysis of salt-bridge interactions in human fascin1

Fascin1 is a highly charged protein with a total of 134 (27.1%) charged residues across the protein, including 77 that are positively charged (15.6%) and 57 that are negatively charged (11.6%). Using PyMOL with a minimum distance cutoff of 2.2 Å and a maximum distance cutoff of 5 Å, we have identified all salt bridges across chains A and B (Supplementary Table S3). The number of salt bridges differs slightly between the two molecules in the asymmetric unit, with molecule A having 103 and molecule B having 101 (Table 2). The most significant differences between the two molecules lie in the distribution of their interactions throughout the protein. While slight variations are observed across β-trefoil domains 3 and 4, β-trefoil domains 1 and 2 show the greatest differences in salt-bridge distribution (Table 2 and Fig. 6). More importantly, among all salt bridges, we have found several β-trefoil interdomain interactions which could be responsible for the stability of the protein (Table 2 and Fig. 6). In molecule A, β-trefoil domain 1 is connected to β-trefoil domain 2 via Asp97, which forms salt bridges with Arg185 and Arg224 in β-trefoil domain 2 (Table 2). The link between β-trefoil domain 2 and β-trefoil domain 3 is mediated by an interaction between Arg167 and Asp290 (Table 2). Likewise, β-trefoil domain 3 is connected to β-trefoil domain 4 through a series of salt bridges: Lys303 with Asp457, Asp342 with Lys464, Arg344 with Asp420, and Arg343 with Asp450 (Table 2). In molecule B all of these interactions are preserved, except for the Lys303–Asp457 and Asp97–Arg185 interactions (Table 2). In contrast, an interaction between residues Lys434 and Asp337 is observed (Table 2). In both molecules, no salt-bridge interactions were found between β-trefoil domains 1 and 4.

Figure 6 Salt-bridge interactions in human fascin1. Ball-and-stick representation of all salt bridges found in molecules A (left panel) and B (right panel) in the asymmetric unit of the structure reported in this study. Positive residues (Arg and Lys) are shown in blue and negative residues (Glu and Asp) are shown in red. Salt bridges are shown as black dashed lines. Residues involved in interdomain interactions are shown as spheres. Fascin1 molecules are shown in a cartoon representation using the same color code as in Fig. 4.

We extended our analysis of salt-bridge interactions to previously reported crystal structures of fascin1 in its free form [PDB entries 1dfc (Sedeh et al., 2010), 3llp (Chen et al., 2010) and 3p53 (Jansen et al., 2011)]. All salt bridges across chains A and B in each crystal structure are collected in Supplementary Tables S4–S6. A comprehensive summary of the salt-bridge interactions observed across these structures is provided in Table 2. Our analysis revealed several differences when compared with our structure, particularly in the distribution of salt-bridge interactions. The unliganded structures exhibit the following numbers of salt bridges (Table 2): PDB entry 1dfc, 81 in chain A, 94 in chain B; PDB entry 3llp, 91 in chain A, 99 in chain B; PDB entry 3p53, 97 in chain A, 101 in chain B. While the overall number of salt bridges is comparable across all free fascin1 structures, the most significant differences lie in their distribution throughout the protein (Supplementary Tables S3–S6, Fig. 6 and Supplementary Fig. S5). Also, we observed marked differences in interdomain interactions (Table 2, Supplementary Fig. S5) so that only four interdomain salt bridges are conserved across all structures. Two of these bridges connect β-trefoil domains 3 and 4 (Arg343–Asp450 and Arg344–Asp420), while the remaining two link β-trefoil domain 2 to β-trefoil domain 1 (Arg224–Asp97) and β-trefoil domain 3 (Arg167–Asp290). Moreover, β-trefoil domain 3 engages in the highest number of salt-bridge interactions overall, suggesting its prominent role in stabilizing the protein structure. These findings highlight that despite the crystallization conditions being identical for all structures, differences in the spatial arrangement of salt bridges, especially interdomain salt bridges, could have implications for the functional dynamics of the protein.

3.4. Preliminary experiments on microcrystals of human fascin1

Our next step in human fascin1 research will be the study of its structural dynamics and allosteric properties. For this purpose, we will use time-resolved serial crystallography (TR-SX) experiments both at X-ray free-electron lasers (XFELs) and synchrotrons. Due to the nature of TR-SX experiments, crystal samples contain large amounts of microcrystals in the range of a few micrometres that are highly homogeneous in size. To this end, we optimized the human fascin1 crystallization conditions to obtain the required microcrystals. The best preliminary microcrystals were grown using the batch method with seeding (Fig. 7a) under optimization of the crystallization conditions in which large crystals were grown (see Section 2.6). They were needle-like microcrystals of about 25–30 µm in size (Fig. 7b) and with an ideal density for TR-SX experiments. These crystals were tested at a synchrotron and diffracted to ∼2.5 Å resolution, supporting their suitability for future time-resolved experiments.

Figure 7 Microcrystals of human fascin1. (a) Schematic diagram of the microcrystallization process of the human fascin1 protein. (b) Microcrystals of human fascin1 were about 25–30 µm in size.