Single-Molecule Force Spectroscopy Reveals Adhesion-by-Demand in Statherin at the Protein–Hydroxyapatite Interface

Achieving strong adhesion in wet environments remains a technological challenge in biomedical applications demanding biocompatibility. Attention for adhesive motifs meeting such demands has largely been focused on marine organisms. However, bioadhesion to inorganic surfaces is also present in the human body, in the hard tissues of teeth and bones, and is mediated through serines (S). The specific amino acid sequence DpSpSEEKC has been previously suggested to be responsible for the strong binding abilities of the protein statherin to hydroxyapatite, where pS denotes phosphorylated serine. Notably, similar sequences are present in the non-collagenous bone protein osteopontin (OPN) and the mussel foot protein 5 (Mefp5). OPN has previously been shown to promote fracture toughness and physiological damage formation. Here, we investigated the adhesion strength of the motif D(pS)(pS)EEKC on substrates of hydroxyapatite, TiO2, and mica using atomic force microscopy (AFM) single-molecule force spectroscopy (SMFS). Specifically, we investigated the dependence of adhesion force on phosphorylation of serines by comparing findings with the unphosphorylated variant DSSEEKC. Our results show that high adhesion forces of over 1 nN on hydroxyapatite and on TiO2 are only present for the phosphorylated variant D(pS)(pS)EEKC. This warrants further exploitation of this motif or similar residues in technological applications. Further, the dependence of adhesion force on phosphorylation suggests that biological systems potentially employ an adhesion-by-demand mechanism via expression of enzymes that up- or down-regulate phosphorylation, to increase or decrease adhesion forces, respectively.


S1 Classification of force distance (FD) curves.
The classification of retraction force distance (FD) curves allows the differentiation of nonspecific from specific interactions and ensures that the pull-off forces belongs to the attached sample. The FD curves were classified in three different categories: (A) no adhesion event, (B) non-specific adhesion events in the contact region (< 60 nm) of the AFM tip and surface, or (C) specific adhesion events (distance > 60 nm from contact region). 1,2 Figure S1. Classification of three different adhesion events: (a) no adhesion event, (b) nonspecific adhesion events in the contact region (< 60 nm) of the AFM tip and surface, or (c) specific adhesion events (distance > 60 nm from contact region).

S2 Investigation of the influence in length of the maleimide-PEG-NHS linker system
To investigate the influence in length of the linker system, two maleimide-PEG-NHS linker with 27 ethylene glycol units and with 162 ethylene glycol units were compared to each other.
Therefore, DpSpSEEKC was tethered on both linker systems to evaluate the best conditions for adhesion measurements on the molecular level. The FD curve of the PEG linker (27 ethylene glycol units) in Figure S2a shows no typical linker stretching contrary to the PEG linker (162 ethylene glycol units) ( Figure S2b). Consequently, the linker stretching of the shorter PEG-linker is not visible caused by overlapping of nonspecific adhesion events with specific adhesion events. The adhesion forces seem correct in magnitude, but to investigate single adhesion events sufficiently it is necessary to separate them.

S3 Investigating the adhesion of every single functionalization step
To verify the functionalization method per se, three AFM tips per functionalization step were measured on different substrates (naked tip, amino tip, linker tip). The pull-off forces of the naked tips at different dwell times are significantly lower than of the amino functionalized tips.
These tips in turn show significantly higher adhesion forces (specific binding event) than the linker functionalized tips due to electrostatic interactions between the positively charged amino tip and negatively charged surfaces. In the absence of an attached peptide sequence (only linker functionalized) and in the absence of all (only naked tip) the force-distance curves show essentially no hysteresis between extension and retraction curves.  The used AFM chips (Bruker MSNL-10) have a maximal tip radius of 12 nm. We are able to calculate the maximum number of anchored molecules on the apex of the tip with a spherical cap model containing rmax = 12 nm and the cap height hmax = 1 nm (see Figure S4a). means, if more than one molecule for a given cantilever is interacting with the substrate during a pull-off, multiple adhesion-rupture events are recorded with similar forces. This is present in about 10% of the recorded force curves (see Figure S4b). (c) Force plotted against the corresponding loading rates (LR) on logarithmic abscissa with fit.

S6 SMFS measurement of DSSEEKC and
The loading rate is the product of pulling velocity and effective spring constant (keff).
(d) Schematic illustration of statherin where its phosphorylated serines on the N-terminus unveil only a minor beneficial effect due to lack of calcium ions in muscovite.

S7 Determination of Bond Length (xb) and Bond Dissociation Energy (∆Eb)
SMFS measurements with DpSpSEEKC and DSSEEKC were performed on HAP, TiO2 and mica by varying the loading rate, which is the product of pulling velocity and effective spring constant (keff). The rupture force was plotted logarithmically against the corresponding loading rate (pN s -1 ) (see where v is the natural molecular vibration frequency in vacuum (≈ 10 10 s -1 ).
The determined bond lengths and bond dissociation energies of DpSpSEEKC and DSSEEKC on CDHAP, TiO2 and mica were shown in Figure S7 and Table S7.   9 describing the extension of the peptide sequence. Figure S8. Calculation of the effective spring constant keff consisting of the cantilever stiffness kc and the stiffness of the PEG-linker-sample system

S9 Calculation of PEG-peptide extension
In order to describe the PEG-peptide stretching in detail for calculating kL, we were using the worm-like-chain (WLC) model 9

Statistical Analysis Figure 2c:
For the assessment of normality a Kolmogorov-Smirnov test was performed on data for DpSpSEEKC with different dwell times measured on CDHAP (see Table S10a).  Table S10b). The adhesion force measured at 4 s and 8 s are similar. The null hypothesis was tested for equal sample-sample distributions.

Statistical Analysis Figure 3:
For the assessment of normality a Kolmogorov-Smirnov test was performed on data for DpSpSEEKC and DSSEEKC at 4 s dwell time and 1000 nm s -1 measured on CDHAP (see Table S10c).

Statistical Analysis Figure 4:
For the assessment of normality a Kolmogorov-Smirnov test was performed on data for DpSpSEEKC and DSSEEKC at 4 s dwell time and 1000 nm s -1 measured on TiO2 (see Table   S10d).  Figure S6:

Statistical Analysis
For the assessment of normality a Kolmogorov-Smirnov test was performed on data for DpSpSEEKC and DSSEEKC at 4 s dwell time and 1000 nm s -1 measured on mica (see Table   S9e).

Statistical Analysis Figure 5 and S3:
The statistical analysis of Figure 5 is included in S3. In Figure 5 the adhesion mean values of DpSpSEEKC and dopamine were compared in a Mann-Whitney-U test for two samples. In Figure S3 the whole data was compared in a Kruskal-Wallis test with post-hoc test. For the assessment of normality a Kolmogorov-Smirnov test was performed on data for different samples with 4 s dwell time measured on TiO2 (see Table S10f).  Table   S9g). The adhesion force measured with the naked tip and linker tip are similar. The null hypothesis was tested for equal sample-sample distributions.

S11 Hydroxyapatite substrate preparation
The substrates were in the form of disks 8 and 14 mm in diameter and 2 mm height composed of calcium deficient hydroxyapatite (CDHAP), which is the final product of calcium phosphate bone cements. CDHAP was formed after the hydrolysis of α-TCP (α-tricalcium phosphate) according to the reaction: The preparation of α-TCP, which is the solid phase of cement, was based on the solid-state reaction between calcium carbonate (CaCO3) and calcium pyrophosphate (Ca2P2O7): Equimolar quantities of the reactants were mixed and placed in a ball milling (Pulverisette-6 Fritsch, Germany) and mixed using 1 cm agate spheres at 450 rpm for 20 minutes and 3 repetitions. Next the mixture was transferred in a 100 ml alumina crucible (Coors Ceramics, USA), placed in a programmable furnace (Nabertherm, Germany) and heated to 1400 o C for 48 hours. Next the crucible quenched at room temperature and the product was ground using a ball milling at 500 rpm for 15 minutes and 3 repetitions. The obtained fine powder was composed of α-TCP and stored at room temperature in a desiccator under vacuum.
For the preparation of the cement paste, α-TCP was mixed with an aqueous solution of 4% w/v Na2HPO4, which was the liquid phase, with a ratio of 0.32 ml/gr until the mixture was homogenized. Then, with the help of a spatula, the cement was placed in silicone molds and remained under 100 % humidity environment for 2h hrs to achieve the initial hardening (setting) of the cement. Next the disks placed in plastic containers containing Ringer's solution (Fresenius Kabi, Lactated Ringer) and remained at a temperature of 37 o C for 21 days to complete the hardening. Finally, the aged cements washed with distilled water, dried at 80 o C and characterized by XRD in order to verify the formation of calcium deficient hydroxyapatite ( Figure S11). Figure S11. XRD graphs of substrates and JCPDS card for calcium deficient hydroxyapatite

S12 TiO2 coated silicon wafer substrate preparation
Silicon wafers (Si-Mat, type-Orient P/Bor 100; Thickness 525 ± 25 μm) were cut into 0.5 x 0.5 cm pieces. First, they were cleaned by a Boekel UV-cleaner for 15 min. Afterwards, the wafers were cleaned chemically according to the cleaning procedure in Table S11a. Table S11a. Cleaning procedure for silicon wafers cleaning step procedure 1 wafer 2 x 5 min washed with water + ultrasonic bath Before spin-coating of TiO2 particles (Aeroxide TiO2 P25 rutile + anatase, Evonik), the silicon wafers were heated at 150 °C for 60 s to remove moisture. Spin-coating was conducted immediately after cooling down the wafer by argon flow, with a rotation speed at 2000 rpm for 50 s. The coated wafers were kept under inert atmosphere before SMFS measurements.

S12 Ellipsometry of TiO2 coated silicon wafer
The thickness of the TiO2 coated silicon wafer was determined by ellipsometry (see Table   S12b). 50 different spots on the blank wafer and on the coated wafer were measured. The results show an increase of layer thickness after the TiO2 coating. Additionally the layer thickness of the coated wafer is very homogeneously distributed.