The N-terminal tail of the hydrophobin SC16 is not required for rodlet formation

Hydrophobins are small proteins that are secreted by fungi, accumulate at interfaces, modify surface hydrophobicity, and self-assemble into large amyloid-like structures. These unusual properties make hydrophobins an attractive target for commercial applications as green emulsifiers and surface modifying agents. Hydrophobins have diverse sequences and tertiary structures, and depending on the hydrophobin, different regions of their structure have been proposed to be required for self-assembly. To provide insight into the assembly process, we determined the first crystal structure of a class I hydrophobin, SC16. Based on the crystal structure, we identified a putative intermolecular contact that may be important for rodlet assembly and was formed in part by the N-terminal tail of SC16. Surprisingly, removal of the N-terminal tail did not influence the self-assembly kinetics of SC16 or the morphology of its rodlets. These results suggest that other regions of this hydrophobin class are required for rodlet formation and indicate that the N-terminal tail of SC16 is amenable to modification so that functionalized hydrophobin assemblies can be created.

phylogenetic analysis of hydrophobin sequences, class I sequences can be further separated into those originating from the Ascomycota (class IA) and Basidiomycota (class IB) fungi 6,20 .
Class IB hydrophobins have a greater degree of sequence conservation than class IA hydrophobins, suggesting that they may follow a common assembly mechanism. Current models of rodlet assembly have been developed by studying the class IA hydrophobin EAS (from Neurospora crassa) and involve L 3 undergoing a structural transition from a disordered loop to a β-strand, which then oligomerizes with L 3 of other EAS molecules into amyloidlike rodlets 21 . However, this model is not compatible with the nuclear magnetic resonance (NMR)-derived structure of the class IB hydrophobin SC16 (from Schizophyllum commune), since in SC16 L 3 only consists of four residues which are structured as a β-turn. This makes it unlikely that L 3 of SC16 can undergo large conformational changes during rodlet assembly, meaning that rodlet assembly likely involves other regions of SC16.
To investigate possible assembly mechanisms of SC16 we used X-ray crystallography to determine its atomicresolution crystal structure, with the rationale that crystal contacts may provide insight into how SC16 monomers may associate with each other. We noted a large crystal contact was formed that involved the N-terminal tail of SC16. We then used NMR spectroscopy to investigate the dynamics of this tail, and functional assays to investigate how this region of SC16 influences rodlet assembly.

Results and discussion
Crystallization of SC16 identifies a possible biological contact. Protein crystallography inherently relies on protein association, and proteins will often crystallize in biologically meaningful ways because of favourable energetics 22 . Therefore, we set out to crystallize SC16 in order to gain insight into which regions of the molecule may form extensive intermolecular contacts and hence may be involved in the self-association that occurs during rodlet self-assembly. After optimizing initial conditions, SC16 crystallized as a cluster of plates over 1-2 weeks (Supplemental Fig. 1). Virtually indistinguishable atomic-resolution structures were determined for two crystals of SC16 that existed in two different space groups: C222 1 ( Fig. 1 light blue; 2.0 Å resolution) and P2 1 2 1 2 (Fig. 1b dark blue; 2.2 Å resolution). These structures superpose to a root mean squared deviation (RMSD) of 0.1 Å and had refinement statistics consistent with good quality models ( Table 1). The SC16 crystal structure consists of four disulphide bonds which are characteristic of hydrophobins 18 , a β-barrel core consisting of four β-strands (β 1 : 39Tyr-46Asp, β 2 : 69Leu-76Pro, β 3 : 92Val-98Tyr, β 4 : 102Leu-107Cys), and three loops (L 1 -L 3 ) that connect the β-strands. An α-helix was present in L 1 from 50Lys-59Leu. Residues near the N-terminus and L 2 (C222 1 : 79Val-85Asn, P2 1 2 1 2: 78Ser-85Asn) did not have interpretable electron density, consistent with these regions being flexible. Overall, the structure of SC16 obtained from crystallization is consistent with the NMR-derived ensemble of SC16 structures 20 (Protein Data Bank, PDB ID:2NBH, RMSD of 1.39 Å and 1.37 Å for the C222 1 and P2 1 2 1 2 crystal forms, respectively, Fig. 1b).
We then used the EPPIC (Evolutionary Protein-Protein Interface Classification) webserver 23 to probe for large protein-protein interfaces in the crystal lattice that may represent biologically relevant associations of SC16. EPPIC detected a large intermolecular interface that was present in both SC16 crystals and had a surface area of 607 Å 2 and 589 Å 2 for the C222 1 and P2 1 2 1 2 space groups, respectively (Fig. 1c). This surface involves both the N-and C-termini of SC16 interacting with L 1 , L 2 , and the β-barrel of a second symmetry-related SC16 molecule, with 44Thr, 72Leu, 89Ala, and 110Ile being flagged by EPPIC as participating in close contacts.
In particular, the N-terminal tail adopts an extended structure from 27Gly-32Ser with partial β-strand character that associates with the β 1 strand over residues 42Asn-46Asp of a symmetry related SC16 molecule (Fig. 1c). Although the buried surface area (~ 600 Å 2 ) is smaller than the 800 Å 2 expected for a definitive biological contact 24 , it is larger than a typical crystal contact, involves a very small protein, and is found in two different space groups. Repeating units of this interface also result in a linear association of SC16 monomers that suggests a possible assembly pattern within rodlets. NMR spectroscopy identifies the termini and L 2 of SC16 as dynamic. In both crystal structures of SC16, sections of the N-terminus and L 2 were poorly defined. To assess the flexibility of SC16 in solution we used NMR spectroscopy to determine which regions of the protein are dynamic. Through analysis of triple resonance NMR spectra and with the aid of deposited NMR assignments for SC16 (BioMagRes Bank accession 25976) 20 we were able to assign 89% of the amide resonances of SC16. To assess the dynamics of SC16 on the ns-ps timescale under native conditions, T 1 (longitudinal) relaxation, T 2 (transverse) relaxation, and heteronuclear nuclear Overhauser effect (NOE) experiments were collected and analyzed using the relaxation module of CcpNmr Analysis (Fig. 2). Values of T 1 were constant over the entire protein with an average value of 808 ± 66 ms, while T 2 and NOE values both plateaued over the ranges of 31Lys-79Val and 89Ala-112Ile, with values of 82 ± 18 ms and 0.748 ± 0.014, respectively. T 2 values increased and NOE values decreased for the N-and C-termini and L 2 (80Ile-88Ser), identifying these as dynamic regions that may not be resolvable by X-ray diffraction. Importantly, this suggests that 27Gly-32Ser are dynamic enough to undergo conformational changes and be important for rodlet assembly of SC16. Based on the dynamic nature of the N-terminus and its involvement in a large, possibly biological, crystal contact, we generated a variant of SC16 that lacked the N-terminus up to 31Lys (SC16ΔN). SC16ΔN was expressed and purified in the same manner as SC16 (Supplemental Figs. 2 and 3).
The N-terminal tail of SC16 is dispensable for rodlet assembly. Thioflavin T (ThT) is a fluorescent dye commonly used to quantify amyloid formation and hydrophobin assembly, since fluorescence emission intensity is linearly related to fibril concentration 25 . Thus, ThT fluorescence assays were used to quantitatively measure the assembly of both SC16 and SC16ΔN when agitated to create an air:water interface. SC16ΔN assembly was monitored in buffers of pH 5.5-8.5 containing 10 mM-1 M NaCl (Supplemental Fig. 4). Among these conditions, pH 8.5 and 1 M NaCl generated maximal ThT fluorescence upon SC16ΔN agitation. In this buffer www.nature.com/scientificreports/ condition, the assembly kinetics of SC16 and SC16ΔN were indistinguishable (Fig. 3), suggesting that residues 18Thr-30Pro do not play a significant role in the rate of SC16 self-assembly. We further characterized the morphology of SC16ΔN rodlets with atomic force microscopy. SC16ΔN was deposited onto freshly cleaved highly oriented pyrolytic graphite (HOPG) and allowed to assemble for either 20 min or overnight (Fig. 4 with additional fields shown in Supplemental Fig. 5). Rodlets of a uniform width (~ 10 Å) were observed and were often paired together. The rodlets could either remain dispersed (Fig. 4, top) or form laterally associated bundles (Fig. 4, bottom), which is consistent with what has been observed previously for SC16 20 and rodlets from other hydrophobins 21,26 . Interestingly, the longest rodlets often form from multiple shorter rodlets closely organized end-to-end.
Understanding the mechanism of and the sequences responsible for hydrophobin assembly is important for developing commercial applications of these proteins. For example, mutation of residues important for assembly could tailor hydrophobin assembly to different buffer conditions. Conversely, regions that are not important for self-assembly can be modified to add new functionality to hydrophobins. The mechanistic origins of hydrophobin rodlet assembly are unclear and they seem to vary depending on the hydrophobin studied. For example the class IA hydrophobin EAS requires sequences in L 3 for assembly 21 , while sequences in L 1 are dispensable 27 . Conversely in the hydrophobin DewA (from Aspergillus nidulans), L 2 is predicted to be required for self-assembly and rodlet formation 26 .
In this study we used X-ray crystallography to determine the structure of SC16 and investigated a potential mode of its self-assembly. We found that although the N-terminal tail of SC16 associates with neighbouring molecules (coloured differently) in the C222 1 SC16 crystal. The inset highlights a putative biological contact spanning 606.80 Å 2 that involves portions of L 1 , L 2 , and β-barrel structure from one molecule (coloured cyan and orange) and the N-and C-termini of a second molecule (coloured light blue and red). Residues flagged by EPPIC as participating in close contacts are highlighted orange or red, labelled, and drawn as sticks. The second inset illustrates the association between the N-terminal tail of one SC16 molecule with the β 1 strand of another. Putative hydrogen bonds are indicated with yellow lines. www.nature.com/scientificreports/ molecules in the crystalline state, it is dynamic in solution and does not significantly influence the self-assembly rate or morphology of SC16 rodlets. This suggests that other regions of SC16 (such as L 1 and L 2 ) are likely essential for rodlet formation. These results outline an opportunity for hydrophobin functionalization, since modification of the N-terminal tail of SC16 should be a viable strategy to create functionalized SC16 variants that are still able to self-assemble onto surfaces.

Methods
Plasmid construction. pET21-derived plasmids coding for SC16 with the N-terminal signal sequence removed (18Thr-114Leu of the hyd1 gene of S. commune; NCBI ID: EFI94929) with upstream sequences coding for a hexahistidine tag, the B1 domain of protein G, and a thrombin protease recognition sequence (H 6 -GB1-SC16) are the same as those used previously 28 . A second expression plasmid was constructed using restriction enzyme-based procedures that coded for SC16 with 13 residues from the N-terminal tail removed (residues 31Lys-114Leu of SC16; H 6 -GB1-SC16ΔN). The validity of all plasmids was verified by sequencing (Eurofins Genomics).
Protein expression and purification. SC16 29 , and purified as above. After purification samples were either used immediately or frozen at − 20 °C for long-term storage.    www.nature.com/scientificreports/ reaction volumes were 100 μL and all conditions contained 50 μg/mL protein, 20 μM filtered ThT. Each condition was replicated 6 times and measured in a sealed 96-well black-bottomed plate. A SpecraMax M3 plate reader was used to monitor the ThT fluorescence (λ ex = 430 nm, λ cutoff = 455 nm, and λ em = 480 nm) while shaking the reactions for 30 s every 2 min.
Atomic force microscopy. SC16ΔN was dialysed against dH 2 O overnight at 4 °C. Hydrophobin drops (50 µL at 5 µg/mL) were placed onto freshly cleaved 3 mm HOPG discs and allowed to dry overnight. Alternatively, 50 µL drops containing 10 μg/mL of hydrophobins were allowed to sit on the HOPG for 20 min. Excess solution was wicked away and HOPG was allowed to dry overnight at room temperature. Two samples were produced for each condition and at least 3 fields of view were collected for each sample. Images were collected on a NanoWizard II Ultra from JPK using Tap300AI-G tips (Budget Sensors).