Toward Artificial Mussel‐Glue Proteins: Differentiating Sequence Modules for Adhesion and Switchable Cohesion

Abstract Artificial mussel‐glue proteins with pH‐triggered cohesion control were synthesized by extending the tyrosinase activated polymerization of peptides to sequences with specific modules for cohesion control. The high propensity of these sequence sections to adopt β‐sheets is suppressed by switch defects. This allows enzymatic activation and polymerization to proceed undisturbed. The β‐sheet formation is regained after polymerization by changing the pH from 5.5 to 6.8, thereby triggering O→N acyl transfer rearrangements that activate the cohesion mechanism. The resulting artificial mussel glue proteins exhibit rapid adsorption on alumina surfaces. The coatings resist harsh hypersaline conditions, and reach remarkable adhesive energies of 2.64 mJ m−2 on silica at pH 6.8. In in situ switch experiments, the minor pH change increases the adhesive properties of a coating by 300 % and nanoindentation confirms the cohesion mechanism to improve bulk stiffness by around 200 %.

3.2 Preparation of AbPPO4 tyrosinase from common mushrooms .
All buffers and aqueous solutions were prepared with Milli-Q water. Sodium citrate buffer was used for all experiments except for CD experiments where sodium phosphate buffer was used.

Enzymatic assay and reactions
The enzyme AbPPO4 was prepared as described earlier (cf. section 3.2). [1] Lyophilisates of the enzyme from sodium citrate buffer (50 mM, pH 6.8) were stored at -20 °C and dissolved in Milli-Q water prior to use. Enzyme solutions were stored at -20 °C as well. L-Tyrosine (≥98%) was obtained from Sigma Aldrich (Seelze, Germany) and L(+)-ascorbic acid (≥99%) was purchased from Carl Roth GmbH (Karlsruhe, Germany). 0.7 nmol/U ascorbic acid was used as antioxidant additive to prevent possible oxidations in the material solutions and stored at -20 °C either as solutions or as lyophilized powders.

Instrumentation
Ultraviolet-visible spectroscopy (UV-Vis). The enzymatic assay was carried out in a UV-Vis EonC Microplate Spectrophotometer with cuvette port (BioTek, Bad Friedrichshall, Germany) using quartz cuvettes.
High-performance liquid chromatography (HPLC). Analytical HPLC was performed on a Shimadzu (Japan) system using a SCL-10A vp system controller, a SPD-M10A vp diode array detector, a LC-10AD vp liquid chromatograph pump unit and a CTO-
Transmission electron microscopy (TEM). 10µL of 0.5 µM of samples were spotted on a carbon film 300 mesh, copper grid.
TEM imaging was performance using Talos TEM from Field electron and ion (FEI).
Atomic force microscopy (AFM). For atomic force microscopy, the samples were spin-coated (3000 rpm) from solution (0.5-0.05 µM) on freshly cleaved Mica substrates. AFM imaging was done in tapping mode under a silicon nitride cantilever using Veeco Nanoscope VIII Multimode AFM. Commercial silicon tips (Type SCANASYST-AIR) were used with a tip radius 2-12 nm, employing a spring constants 0.4 Nm -1 at a resonance frequency of 50-90 kHz.
For adhesion measurements, clean, plasma activated (0.2 mbar, 100W, 1 min; 440-G, TePla, Wettenberg, Germany) glass slides were coated with the polymers and an AFM (MFP 3D, Asylum Research, Oxford Instruments, California, USA) were used a soft colloidal probe was chosen for the adhesion measurements, because this results in a significantly larger contact area than for hard colloidal probes (e.g. glass, silica). Adhesion experiments were performed with a tipless cantilever (NSC 35, Mikromasch Europe, Wetzlar, Germany) equipped with a Polydimethylsiloxane (PDMS) soft colloidal probe (diameter 23.6 µm, E-modulus ~ 2.2 MPa). Prior to use the cantilever was O2 plasma treated (0.2 mbar, 100 W, 10 sec; 440-G, TePla, Wettenberg, Germany) to activate the bead surface and establish reproducible conditions. Determination of the spring constant of the cantilever was achieved by measurement of thermal noise [2] (k=15.4 N/m) and by pressing the cantilever against a nondeformable surface lever sensitivity [3] was determined for the tipless cantilever. Since lever sensitivity with a soft colloidal probe cannot be calculated by the same method, thermal noise was used to assess this value. The sensitivity of the soft colloidal probe cantilever is in good agreement with the tipless one. Nanoindentation. Nanoindentation studies were done using a TriboIndenter TI-950 (Hysitron-Bruker, MN, USA) equipped with a standard 2D transducer and a Berkovich tip. The tip was calibrated for the required contact depths using a standard PMMA sample (E=5.13 GPa). A cyclic load function composed of 10 cycles with an increment of 10 N/cycle and a max. load of 100 N was used for measurements. The cyclic load function was used to make sure that the extracted indentation curves were not affected by the silicon wafer substrate or any inhomogeneity on the surface of the films. The Oliver-Pharr method [4] was used for calculation of the elastic modulus and hardness of the samples. Scanning probe microscopy was used to measure the thickness of the samples (tpH 5.5 = 8 m and tpH 6.8 = 100 m), which were more than 10 times thicker than the maximum contact depths.

Peptide synthesis
Peptides were synthesized following standard ABI-Fastmoc protocol (single coupling with capping) with NMP as solvent using standard Fmoc-amino acid derivatives. As solid support, Fmoc Rink Amide resin (loading 0.74 mmol/g, 0.1 mmol, 100-200 mesh) was used. Synthesis was performed on an automated ABI 433a peptide synthesizer (Applied Biosystems, Foster City, USA). Fmoc-amino acid coupling was facilitated by HBTU/DIPEA. After final Fmoc removal, the resin was transferred to a 10 mL syringe reactor and subsequently washed with dichloromethane. Peptides were cleaved from the solid support with a mixture of 95:4:1 vol.% TFA/H2O/TES for 3 h, which resulted in fully deprotected peptide. In the case of peptides whose C-terminal is a Boc-AA, the cleavage from the solid support (3 ml/10µmol) was carried out with a mixture of 80:15:5 vol.% TFA/TFMSA/mcresol for 2 h. In both cases, the resin was filtered, washed with TFA and the collected supernatants were concentrated in vacuo.
The product was isolated by precipitation with diethyl ether and subsequent centrifugation. Purified products were obtained by lyophilization from Milli-Q water or from Milli-Q water with 0.1% HCOOH for peptides that contains switch segments.

Synthesis of the switch segment -Val(Boc)Thr-
The coupling of the Fmoc-Val-OH onto the unprotected hydroxyl side chain functionality of the prior attached Boc-Thr-OH was accomplished in a syringe reactor in NMP. Coupling was facilitated by Fmoc-Val-OH/DIC/NMI 10/10/7.5 eq. and repetitive coupling cycles were carried out to force the reaction to completion (at least 3×2 h and 1×12 h). The resin was washed with NMP several times after every coupling step. The last coupling was followed by a capping step (10% DIPEA, 10% Ac2O in NMP, 2×10 min) and Fmoc deprotection. Afterwards the resin was washed with NMP, and peptide synthesis was continued as described above.

Preparation of AbPPO4 tyrosinase from common mushrooms
The gene encoding AbPPO4 was PCR-amplified from cDNA derived from an A. bisporus fruiting body at growth stage 5 [5] and cloned into the expression vector pGEX-6P-1 (GE Healthcare Europe, Freiburg, Germany). The resulting construct encoding glutathione S-transferase [6] N-terminally fused to AbPPO4 was expressed in E. coli BL21(DE3) grown in LB media supplemented with 2 mM MgSO4, 500 mM NaCl, 1x mineral stock M [7] , 1x sugar stock 5052, 100 mg l -1 Na-ampicillin and 0.5 mM CuSO4 at 20 °C for approximately 40 h. Cells were lysed by high-pressure extrusion and non-target proteins were removed by affinity chromatography on Glutathione Sepharose (GE Healthcare). The fusion partner GST was removed by proteolysis with GST-tagged picornain 3C which was afterwards removed along with the cleaved-off GST by a second round of affinity chromatography on the same column material. The resulting latent AbPPO4 was activated by limited proteolysis with proteinase K and the active AbPPO4 was purified by size exclusion chromatography on a Superdex 200 Increase column (GE Healthcare).

AbPPO4 activity assay
Prior to use of enzyme, an activity assay was performed using UV spectroscopy based on the method of Duckworth and Coleman. [8] The absorbance from the oxidation of tyrosine to Dopaquinone is monitored at 280 nm over a period of 20 min at 25 °C using a 3 mL quartz cuvette. The assay solution contained 1 mL of sodium citrate buffer (50 mM, pH 6.8), 1 mL of tyrosine solution (1 mM in Milli-Q water), 0.9 mL of Milli-Q water and 0.1 mL of AbPPO4 enzyme solution (in 50 mM sodium citrate buffer, pH 6.8). Enzyme solution was added immediately before starting the measurement. The activity was calculated using the average slope of the 3 minute interval ΔA280 with the maximum slope of the absorbance-time curve according to equation S1. volumetric enzyme activity .

Enzymatic unimer activation and polymerization reactions
Standard enzymatic activation reactions were performed using a substrate concentration of 0.25 µmol/mL (from 1.0 mM stock solutions) and 100 U/mL of AbPPO4 tyrosinase in a sodium citrate buffer solution (17 mM, pH 5.5) at 25 °C. AbPPO4 was mixed with 0.7 nmol/U ascorbic acid as an additive for the enzyme prior to addition to the substrate solution. In polymerization reactions for GPC, TEM, AFM, nanoindentation and QCM experiments, 0.75 µmol/mL substrate concentration and 50 U/mL AbPPO4 were used.

SDS PAGE
Samples for gel electrophoresis as well as the protein ladder were diluted with Milli-Q water to a volume of 20 µL and subsequently 5 µL of lane marker were added and mixed. The gel cassettes were clamped into the electrode assembly and the assembly and the tank were filled with approximately 800 mL of running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS in Milli-Q water). Samples of 20 µL were loaded into the wells of the gel using a 10-100 µL pipette and runs were performed at 140 V until the lane marker reached the lower end of the gel. Staining was performed using silver stain (Thermo Fisher Scientific, USA) according to the manufacturer's standard protocol.

Quartz crystal microbalance
The piezoelectric sensor crystals coated with 100 nm aluminum oxide (QSX 309, Biolin Scientific, Sweden) were cleaned with    Figure S7b). The signals corresponding to the dimer were found at m/z 1790 -1796 ( Figure S7b, and S7c inset). In the resulting fragmentation spectrum 75% of the corresponding y-and b-ions for the dimer could be found (Table S1). None of the signals indicated the presence of lysinyldopa adducts, which could potentially occur in a much slower alternative reaction pathway. [9] Instead, fragment 1 and fragment 2 were observed, resulting from S-Cβ bond cleavage of the cysteinyldopa linked [C/Y]-(VT)1 9 dimer ( Figure S8).

Switch peptide integrity monitored by IR
The evidence of the rearrangement of the switched peptides to their native backbone were collected by infrared analyses. The presence of the switch ester carbonyl bands (Osec-ester band at ν ≈ 1748 cm -1 ) of the depsipeptides are visible and stable ( Figure S9b). After restoring the native backbone by minor pH changes from pH 5.5 to pH 6.8, the ester carbonyl of the depsipeptide disappears and the carbonyl vibration band of the ester group is no longer observed in the infrared spectra ( Figure   S9c). In polymers case the presence/absence of the carbonyl band of the ester group by pH control is also observed ( Figure   S9d).

S14
nitrogen and stored at -20 °C until SDS PAGE measurement. 15 µL of each reaction sample and an AbPPO4 tyrosinase reference were loaded onto the gel for measurements ( Figure S10 and S11).
Depending on the length of the cohesion module, strong interferences with the enzymatic activation and polymerization process were evidenced. Due to the high tendency of (VT) domains to form β-sheets, the unimers and polymerization products was difficult. While [C/Y]-(VT)1 could be rapidly activated by AbPPO4 and polymerization leads to an intense band at ~35 kDa, [C/Y]-(VT)2 directly forms gels during polymerization, and the peptide [C/Y]-(VT)3 shows gel formation prior to polymerization unable to perform the enzymatic oxidation ( Figure S10). On the other hand, the unimers containing depsipeptide connectivities suppress the aggregation tendency of the (VT)n domains and the polymerization process was used according to the protocol given in the section 3.4. Figure S11 shows immediate formation of multimers ranging from ~10 kDa to ~35 kDa in apparent molecular weight. After 15 min reaction time the average molecular weight shifts to slightly higher values and an intense band at ~35 kDa is observed for poly([C/Y]-(VT) 5 Ψ11,15 ). Further, no significant changes in molecular weight distribution are observed. This indicates a rapid polymerization process that leads to high yields after several minutes at an AbPPO4 concentration of 50 U/mL. Additionally, stained area is visible at the bottom of the wells, which is not the case for the enzyme reference or the protein ladder. This indicates formation of high molecular weight material that cannot be resolved in the polymer matrix due to its molecular weight cutoff.

UV/vis measurements: pH control and enzyme activity
It is known that tyrosinase enzyme has the highest activity at pH 6-7 in aqueous buffer solution. However, to prevent the high tendency of (VT) domains to form β-sheets, the enzymatic oxidation process has to carry out under mildly acid conditions. Therefore, a different enzyme oxidation process was carried out at different acid pH to determine the highest activity of the tyrosinase in these new conditions ( Figure S12b). In buffered solution at pH 5.5 the enzyme proved to practically instantaneously oxidize the tyrosine residues of all [C/Y]-(VT)n Ψ unimers to Dopa and Dopa quinone as shown by UV/vis spectra ( Figure S12c).
Only minor differences in the initial activation rates can be found as the rates increase consistently from   Figure S13 and Table 2).

Secondary interactions and switch behavior monitored by circular dichroism  Depsipeptides
A 250 µM solution of each depsipeptide switch in phosphate buffer 10 mM and at pH 5.5 and pH 7.4 was measured ( Figure S14). The measurements of the depsipeptides in acidic conditions show the absence of -sheet formation ( Figure S14a). However, after the adjustment of the pH to 7.4, the O→N-acyl transfer rearrangement took place in the switch segments of the monomer constructs, which restores the native peptide backbone and the (VT)n domains regain their -sheet formation tendency. This evidence is showed by the presence of the typical Cotton bands for -sheets at -214 nm and +193 nm in the CD spectra ( Figure S14b As in the depsipeptide measurements, at pH 5.5 no evidence of -sheet formation was found. However, at pH 7.4 as the length of the (VT)n segments in the polymerized depsipeptides increases, more pronounced -sheet formation is evident in the CD spectra ( Figure S15). Moreover, the -sheet signal observed in the polymers measurements has a more pronounced intensity than its corresponding monomer, showing the presence of a higher -sheet content in the system ( Figure S16).

Microscopy studies
Microscopy analysis was performed in order to visualize nanostructures generated by the formation of -sheet in the switched polymers.   Mica substrates at 100 rpm followed by 2500 rpm to dry the substrate. In this case, also a well-defined fibrillar aggregates occurs for switched polymers confirming the disturbing effects of the depsi-structure defects ( Figure S18).

QCM-D experiments on aluminum oxide surface
For QCM measurements sample concentration (0.75 µmol/mL) was reduced and samples were diluted after reaction prior to measurement with degassed Milli-Q water 1:21 v/v (0.03 µmol/mL). Therefore, the citrate buffer with the appropriate pH for the specific experiment (pH 5.5 or pH 7.4, 17 mM) was diluted to 0.8 mM as well for use in equilibration and rinsing steps. All measurements were performed at 100 µL/min flow according to protocol 3.5 on an aluminium oxide coated sensor (QSX 309, Biolin Scientific, Sweden).

peptide([C/Y]-(VT)5  ) control
As a control experiment QCM-D measurement of the pure depsipeptide was carried out. Therefore, a solution of [C/Y]-(VT)5  in buffer 17 mM at pH 5.5 (0.75 µmol/mL) was prepared, the sample of 1.5 mL was diluted to 33 mL with degassed Milli-Q water and the sensor was incubated for 5 h followed by buffer rinsing. The observed adsorption is rather weak, but Δf remained constant during buffer rinsing ( Figure S19), indicating no washing-off of the formed coating. The sample of 1.5 mL was diluted to 33 mL with degassed Milli-Q water and the sensor was incubated for 4 h followed by buffer rinsing at pH 5.5. The observed adsorption is rather weak, but Δf remained constant during buffer rinsing ( Figure S20a, b), indicating again no washing-off of the formed coating.

SUPPORTING INFORMATION
S23 

Rinsing of polymer coating
To test for coating stability, the poly([C/Y]-(VT)5  ) coated sensor was rinsed with 599 mM NaCl solution and 4.2 M hypersaline solution for 1 h. The used hypersaline solution was modeled after salt concentrations of Dead Sea [12] water and contained MgCl2•6H2O masses amounts to 727 ng/cm² (2.4%) for NaCl rinsing and 673 ng/cm² (7.2%) for hypersaline rinsing compared to the initial coating, which demonstrates the coating is highly stable against salinity ( Figure S21). The sample of 1.5 mL was diluted to 33 mL with degassed Milli-Q water and the sensor was incubated for 4 h followed by buffer rinsing at pH 5.5. The observed adsorption is rather weak, but Δf remained constant during buffer rinsing ( Figure S22a), the formed coating was not washed-off.

SUPPORTING INFORMATION
S25 

Rinsing of polymer coating
Coating stability of the poly([C/Y]-(VT)4  ) coated sensor was tested by rinsing with 599 mM NaCl and with hypersaline solution (4.2 M) for 1 h. The calculated difference in adsorbed masses amounts to 538 ng/cm² (10%) for NaCl and to 499 ng/cm² (11%) for hypersaline solution compared to the initial coating, which demonstrates high resistance to salinity ( Figure S23). To gain the adhesive data, the raw data set was transformed to force versus deformation curves by taking the spring constant, lever sensitivity, and the lever deflection in contact into account. [3,13] From these curves, the adhesion force (maximum negative force during retraction out of contact) is calculated and converted to the work of adhesion per unit area according to the Johnson-Kendall-Roberts (JKR) model. [14] (eq. S2) where is the effective radius of the probe and the substrate, is the radius of the probe and is the radius of the substrate. Because of the geometry (sphere-plane) the radius of the substrate is infinite, hence .

S27
A smooth and stable polymer layer was observed at pH 5.5. The adhesion values increase with the load force from 100 nN to 500 nN.
As demonstrated in Figure S24a and b the adhesion values of poly([C/Y]-(VT)5  ) at pH 6.8 drop after several measurements with soft colloidal probe and then remain constant.
To determine the adhesion values, a huge contact area between the thin polymer layer and the PDMS probe is necessary. Due to the non-covalent bonding of the polymer layer to the substrate, it is possible that some polymer is collected by the probe with increasing number of contacts, which may cause change of adhesion values after several contacts. This behavior is demonstrated in Figure S24c, dark blue line represents the first measurements at pH 6.8 while the dotted dark blue line shows the retraction of the soft colloidal probe after several contacts. We assume that some polymer material sticks to the PDMS bead and a combination of cohesion and adhesion is measured. Nevertheless, the calculated values are at least two times higher for pH 6.8 in comparison to the lower pH.

S28  Switchable adhesion in situ experiments
For in situ adhesion measurements a glass cover was coated with the poly([C/Y]-(VT)5 Ψ 11,15 ) at pH 5.5 for 2 h of coating time with the same procedure as described above. Two 4 points force maps were done on the surface at different positions. After each force map, the cantilever was rinsed with an excessive amount of water to prevent contamination of the cantilever with the polymer chains in order to access the value of adhesion itself. After rinsing, force curves were recorded on the cleaned glass at the buffer solution. The corresponding work of adhesion on the polymer layer measured at pH 5.5 is W adh-pH5.5 =0.60±0.19 mJ/m 2 for 500 nN (Figure S 25a). To demonstrate that the switch in the polymer backbone takes place due to a pH change from pH 5.5 to pH 6.8, 12 mL of pH 6.8 buffer solution was pumped through the cell using microfluidic pump with the flow rate of 3 mL/min. The reference was recorded at the pH 6.8 buffer solution to ensure the reliability of the measurements. After an equilibration time of approx. 40 minutes at pH 6.8, 4 force maps were recorded on the surface at different positions, following the same procedure used for pH 5.5. Considering the same number of points (n=8), the value of adhesion force after pH changing, increased dramatically showing a work of adhesion measured at pH 6.8 of W adh-pH6.8 =1.80±0.25mJ/m 2 for 500 nN. The results emphasize that the change in adhesion in the Figure S25 is due to a switching of the polymer. As evident in previous measurements, the probe collects some adhesive polymer in the duration of the repetitive measurements, causing higher noise and scattering of adhesion values with increased number of contacts. If the probe surface is partially modified by adhesive polymers throughout the measurements of several force curves, a mixture between cohesion and adhesion was measured that is not straightforward to interpret. Calculation of adhesion forces considering the stable values from the first half of the measurement cycles (n=8 of 16) avoids misinterpretation of adhesion values. However, the average work of adhesion considering the complete set of measurements at pH 6.8 that includes the pick-up bias values (n=16 of 16) provides not significantly reduced values of W adh-pH6.8 =1.54±0.37mJ/m 2 for 500 nN. It is important to mention that during all experiments no significant change in the adhesion values and force curves shape for the reference was found. One of the sample was dissolved in a water solution at pH 5.5 (0.75 µmol/mL) and 10 µL of the solution were deposited on the silicon wafer surface by drop casting.


Film preparation for poly([C/Y]-(VT) 5 Ψ*11,15 ) at pH 6.8 The other sample was dissolved in a water solution at pH 6.8 (0.75 µmol/mL). After 3h, 10 µL of the solution were deposited on the silicon wafer surface by the drop casting method.