Direct Immobilization of Engineered Nanobodies on Gold Sensors

Single-domain antibodies, known as nanobodies, have great potential as biorecognition elements for sensors because of their small size, affinity, specificity, and robustness. However, facile and efficient methods of nanobody immobilization are sought that retain their maximum functionality. Herein, we describe the direct immobilization of nanobodies on gold sensors by exploiting a modified cysteine strategically positioned at the C-terminal end of the nanobody. The experimental data based on secondary ion mass spectrometry, circular dichroism, and surface plasmon resonance, taken together with a detailed computational work (molecular dynamics simulations), support the formation of stable and well-oriented nanobody monolayers. Furthermore, the nanobody structure and activity is preserved, wherein the nanobody is immobilized at a high density (approximately 1 nanobody per 13 nm2). The strategy for the spontaneous nanobody self-assembly is simple and effective and possesses exceptional potential to be used in numerous sensing platforms, ranging from clinical diagnosis to environmental monitoring.


Gold substrates
Gold sensor chips used in the SPR experiments were acquired from Reichert Technologies -Ametek Inc (USA) and consisted of polycrystalline gold surfaces (50 nm) on glass substrates sized 1 cm x 1 cm. For ellipsometry, contact angle and ToF-SIMS studies, gold (100 nm thickness and rms roughness < 2,5 nm) on silicon <100> wafers pre-coated with titanium were acquired from George Albert PVD (Germany) and cut into 1 cm x 1 cm pieces using a diamond scriber.

Substrate treatment
Before functionalization, the gold substrates were submerged for 10 min in a strong oxidizing piranha solution (70% H2SO4, 30% H2O2) to remove organic residues (Caution: Piranha solution reacts violently with all organic compounds and should be handled with care). Subsequently, the gold substrates were rinsed with copious amount of Ultra High Pure (UHP) water and HPLC ethanol and lastly dried under argon.

NbVCAM1 self-assembly monolayers (SAMs)
The lyophilized NbVCAM1 was kept at -20 o C until further use. Freshly piranha cleaned gold chips were incubated with a solution of 1 µM NbVCAM1 diluted in 1x PBS for 24 h at room temperature on a moving plate. As controls, gold chips were immersed in either 1x PBS or HPLC ethanol solutions under the same conditions.

Contact angle
The advancing and receding contact angles were obtained with the instrument OneAttension using the sessile drop analysis mode. The drop volume reached 4-10 µL at a rate of 0.5 µL/sec. Two to three measurements per chip (duplicates) were performed in different chip locations.

Ellipsometry
Ellipsometry measurements were obtained with a J.A. Woollam alpha-SE instrument using gold on silicon wafers. Data analysis used the Cauchy model that considers three layers: Ambient/Monolayer/Substrate. The refractive index was fixed at 1.5. Each chip was measured before and after functionalization. Data was fitted with the software CompleteEASE, with a defined resolution of 0.1 eV. Four measurements per chip (duplicates) were performed in different chip locations.

Time-of-flight-secondary ion mass spectroscopy (ToF-SIMS) and three dimensional Orbitrap secondary ion mass spectroscopy (3D OrbiSIMS)
ToF-SIMS spectra were acquired using a ToF IV (IONTOF GmbH) instrument with 25 keV Bi3 + primary ion beam raster over 500 × 500 µm area. Additional high lateral resolution ToF-SIMS imaging was acquired using 3D OrbiSIMS instrument with 25 keV Bi3 + primary ion beam and delayed extraction. Two 256 × 256 pixel images over area of 100 × 100 µm were acquired on two replicates of each sample type. Measurements were performed in both positive and negative mode. Positive mode spectra were calibrated to: CH3 + , C7H7 + , Au3 + . Negative mode spectra were calibrated to: CH -, CN -, CNO -, Au3 -. Two measurements were taken for each sample and each polarity.

S-3
3D OrbiSIMS measurements were acquired using 20 keV Ar3000 + as primary ion beam. The current of the primary beam was 220 pA. Each spectrum was acquired from an area of 300 × 300 µm using random raster mode and the crater size was 381.9 × 381.9 µm. The spectra were collected in positive and negative mode, in mass ranges 50-750 m/z and 150-2250 m/z. Target potential was set to +57.5 V for positive mode and -57.5 V in negative mode.
Two separate areas were analyzed on each sample and two replicates of each sample type were analyzed. Each measurement lasted 30 scans, the total ion dose per measurement was 1.6 × 10 10

Circular dichroism (CD)
The circular dichroism was performed on a NbVCAM1 SAM surface and with NbVCAM1 in solution, allowing confirmation of its conformation. Studies of the NbVCAM1 in 1xPBS were performed in a Jasco J-1500, using a nanobody concentration of 1 mg/ml, at room temperature. CD analysis of the NbVCAM1 SAM surface were performed using a Chirascan plus. Piranha cleaned quartz slides were incubated overnight with a 4% solution of mercapto-trimethoxysilane (MPTES) in IPA, allowing thiol functionalization. Following rinsing with IPA to remove the excess of MPTES, the slides were immersed in 15 mM copper perchlorate solution in H2O for 15 minutes to provide a Cu 1+ ion surface. Finally, the slides were incubated with NbVCAM1 0.1 mg/ml in 10 mM phosphate buffer, pH 7.9 for 1 h, and rinsed with the same buffer. Three slides were loaded into a quartz cuvette.
All measurements were performed with a 10 mM phosphate buffer pH 7.9.

Simulation model for nanobody adsorption
In order to study and analyze the adsorption of NbVCAM1 on a model gold surface, the NAMD2.12 3 software was used with periodic boundary conditions, the TIP3P water model, the CHARMM27 force-field with a 12 Å cut-off for short-range potentials, and smooth particle mesh Ewald summation for the electrostatics. Visual molecular dynamics (VMD) software version 1.9.1 was employed to analyze the results. 4 The NbVCAM1 structure was obtained from Phyre 2 (protein homology/analogy recognition engine) software, 5 that predicted the structure according to the amino-acid sequence.

S-5
The predicted structure considered 127 residues (amino-acids), not including the C-terminal termination -LEY linker and cysteine-alkyne linker, which were both added with NAMD2.12 software. This is due to the linker -LEY being an addition to the natural sequence, added artificially, as well as the cysteine-alkyne-linker (EPL). 6 The predicted structure has a total net charge of +2e and a disulfide bridge which was kept throughout the simulations.
The residues belonging to the binding site were identified at the N-terminal by the open source platform for ligand detection Fpocket 7 .
The simulations were performed in a NaCl solution and with a slab of the gold (81 Å x 86 Å x 14 Å) that consisted of Au atoms. The close packed gold surface of Au, 8 as already reported elsewhere, 9 has been created as a facecentered cubic crystal (fcc) with lattice parameter 4.078 Å. CHARM-METAL 10,11 has been used for gold forcefield parameters. The gold atoms were kept immobile during the simulation.
The simulations start with the protein above the solid surface, with a minimum protein-gold separation of 20 Å, so that the protein is free to diffuse before it contacts the neutral gold surface. In order to not bias the adsorption process, the protein starts in different orientations in different trajectories, as illustrated in Figure S3. In P1, the N-C axis is normal to the surface with the C-terminal facing away from the surface, and in P2 it faces towards the surface. In P3, the N-C axis lies parallel to the surface. In all these starting configurations, the simulation box is then solvated with the TIP3P waters, neutralized by addition of one Clion, and then brought to 150 mM NaCl concentration. shown, but the water is not for clarity.
The C-terminal modified cysteine is important due to the possibility for a thiolate bond forming with the surface, anchoring the nanobody in a favorable orientation for antigen binding. While these classical MD simulations do S-6 not attempt to create the thiolate bond during the simulation, it is still important to understand whether its formation is favourable due to the initial physical adsorption of the nanobody to the surface. At least three trajectories from each initial position (P1, P2, P3) were performed. The system minimization was performed sequentially in two steps, first allowing water and ion movements, and then freeing the protein. The system is then heated to 310 K temperature over 30 ps, followed by 270 ps equilibration at constant temperature. Finally, the production simulations were performed for 100 ns with a time-step of 2 fs. Periodic Boundary conditions and NVT ensemble were applied in the simulations together with the smooth Particle Mesh Ewald (SPME) 12 for the Coulomb interactions. For ionizable residues the most probable charge states at pH 7 were chosen and no additional restrictions on momentum were applied.

Simulation model for nanobody anchored by a thiolate bond to gold
So far it is not possible to simulate a bond formation event such as the thiolate bond between the nanobody and gold surface, hence this bond had to be created with the force field parameters described below (Table S1), which were added to the parameters and topology files.
Where the parameter is the number of atoms in the protein structure and ⃗ ( 1 ) is the position of the ℎ atom at a given time . The RMSD calculation treats two protein structures to be compared as two rigid bodies (no internal flexibility allowed), then overlaps (aligns) using translations and rotations. In this case, the nanobody

S-7
NbVCAM1 is compared with itself, between the final and initial defined structures. Herein, the RMSD results were applied to the alpha carbons that composed every residue of the nanobody or oligopeptide.
For RMSF, the RMSD is calculated for each nanobody's residue. It reflects each residue's mobility during the MD trajectory, by reporting an amplitude of residue movement (fluctuation) from the average position (in the aligned structures) over the total length of the MD trajectory. The time average for the atoms belonging to the same residue were calculated from the formula Where ⃗ ( ) is the position of the atom in residue at the time , is the number of atoms in the residue, and 〈 ⃗ 〉 is the time average over the trajectory.

Surface plasmon resonance (SPR)
SPR experiments allowed monitoring in real-time the NbVCAM1 immobilization onto the surface, followed by antigen binding. Experiments were performed in a semi-automatic Reichert Technologies SPR at the set temperature of 25 ºC. Initial traces were stabilized before each experiment at a flow rate of 100 µl/min with the running buffer (RB, 1xPBS). Blank injections (with RB) were 10 to 15 min long and were performed before each experiment to remove any impurities and after each experiment to stabilise the final response. Each injection started with a burst flow of 1500 µl/min for 10 sec of hVCAM1 (0.27 ), followed by another injection at a flow rate of 8 µl/min for 30 min. The rising step started by injecting the RB at a flow rate of 8 µl/min for 20 min and then changing it to 100 µl/min.