SPR experimental setup

For all SPR experiments, the running buffer consisting of 10 mM HEPES, 150 mM NaCl, 0.01% (v/v) P20, 1% (v/v) glycerol, 0.01% (v/v) BSA and 2% (v/v) DMSO at pH 7.4 was used. The TCF(7–30) peptide (GGGDDLGANDELISFKDEGEQEEK-K-biotin-amide) peptide was obtained from Biosyntan GmbH. Experiments were conducted at room temperature on a Biacore T200 (GE Healthcare, Uppsala, Sweden). The peptide was immobilized onto a Series S sensor chip SA (10352113, Cytiva, Uppsala, Sweden). Following the manufacturer’s protocol, the sensor was first prepared, and the peptide was injected at a concentration of 10 nM at a flow rate of 10 µL/min. Immobilization was achieved via biotin-streptavidin interaction to a level of 5, 25 and 50 RU in three different flow channels respectively. Finally, all flow channels including the reference channel were passivated with free biotin at 500 nM. Binding analysis was monitored on all four flow channels with sample association time of one minute and dissociation time of four minutes at a flow rate of 50 µL/min. Regeneration (i.e., removal of β-catenin from ligand) of the sensor surface was achieved by injecting 50 mM glycine (pH 9.5) for three minutes at a flow rate of 100 µL/min. Even though this step was not necessary for SPR, it was implemented in order to match the assay setup from FM.

FM experimental setup

FM experiments were conducted using a prototype FM reader from the “Fermi” generation developed by lino Biotech AG (Supplementary Figure 10 S1 File). The flow channel shape was hexagonal (Supplementary Figure 9 A&C S1 File) and the parallel region measured 12.0 x 1.9 x 0.1 mm. Only 18 molograms were recorded per measurement in order to save computing power for faster sampling rates of the intensities of the single molograms. Thus, the lower part of the double flow chamber was used. In all FM experiments, the same running buffer as in SPR was used. Samples in glass vials were injected from the autosampler (Alias, Spark Holland) and transferred with a syringe pump (Precision Syringe Drive/4, Hamilton) to the flow channel for 17 seconds at a flow rate of 400 µL/min (sample delivery step). This purged the chamber and reduced the influence of concentration gradients due to Taylor dispersion in the delivery tubing, which had an internal diameter of 250 µm and a length of 80 cm. During the association phase, the sample was passed through the flow chamber for one minute at a flow rate of 50 µL/min. To prevent the recording of the dilutive end of the sample, a fast flow rate of 500 µL/min was applied for two minutes to record the dissociation phase, followed by additional two minutes at a flow rate of 50 µL/min. For the regeneration, regeneration solution was injected from glass vials into the autosampler loop. The regeneration solution was passed through the flow chamber for three minutes at a flow rate of 100 µL/min.

Chip fabrication

A sensor chip of FM measured 9 x 18 mm. The sensing area per mologram was 400 x 400 µm with 54 molograms (6 x 9 array) being present on one sensor chip [21]. The center to center spacing of molograms in flow direction was 690 µm and 750 µm perpendicular to the flow. The sensors used in FM were manufactured and treated following published manufacturing protocols [16,18,21] with the difference that the photoprotective PhSNPPOC group was replaced by a more hydrophilic PySNPPOC group. To activate the ridges using reactive immersion lithography, a phase mask was placed on the surface with a layer of 0.1% (v/v) hydroxyl amine in DMSO applied in between the mask and the sensor surface. The chip was exposed at 405 nm at a dose of 4000 mJ/cm² with a custom-built setup [17]. Following exposure, the chip was incubated in 1 mM methyltetrazine-4-polyethyleneglycol-N-Hydroxysuccinimide (MeTz-PEG4-NHS) dissolved in HBST (10 mM HEPES, 150 mM NaCl, 0.05% (v/v) Tween20, pH 8.0) for 1 hour. After incubation, the chip was washed sequentially with 50:50 H₂O/DMSO, DMSO, Isopropanol and H₂O. Next, the entire chip surface, including the grooves, was activated by flood illumination with a dose of 18 J/cm². The chip was washed sequentially with 50:50 MQ H₂O/DMSO, DMSO, Isopropanol and MQ H₂O. As a final step, the chip was incubated with 10 mM tetrazine (Tz)-PEG5-NHS in HBST (10 mM HEPES, 150 mM NaCl, 0.05% (v/v) Tween20, pH 8.0) for 1 hour and washed with isopropanol and H₂O.

Peptide immobilization in FM

The initial sensor surface for FM consisted of MTz chemical linkers in the ridges and Tz chemical linkers in the grooves. Trans-cyclooctene (TCO) labelled TCF4(7–30) peptides were immobilized onto the FM sensor surface via click chemistry. First, mutant TCF4(7–30) was manually pipetted (100 uL) at 10 µM (in PBS) into an open flow chamber (Supplementary Figure 9 B&C in S1 File) on the sensor. After aspirating the mutant TCF4(7–30) peptide, the binding TCF4(7–30) peptide was injected manually (100 uL) at 20 µM (in PBS) concentration. For all sensors produced for the FM experiments, an immobilization level of approximately 3 pg/mm² of the TCF4(7–30) binding peptide was used.

Proteins and peptides

Human β-catenin was procured from Proteros Biostructures (Variant: β-catenin (human) (Hs134–670)+TCF4(Hs1–53)). The β-catenin binding peptide TCF4(7–30) (TCO-PEG3-C-GGGDDLGANDELISFKDEGEQEEK), the non-binding (D16A/E17A) mutant version TCF4(7–30) (TCO-PEG3-CGGGDDLGANAALISFKDEGEQEEK) and TCF4(7–51) (TCO-PEG3-C- GGGDDLGANDELISFKDEGEQEE KSSENSSAERDLADVKSSLVNE) were obtained from LifeTein, LLC.

Binding analysis

The equilibrium dissociation constant (K_D) and half maximal inhibitory concentration (IC₅₀) were determined from binding signals collected in direct binding and competition experiments, respectively. To achieve this, various concentrations of β-catenin and the strong binding competing peptide TCF4(7–51) [42] were injected onto the SPR and FM sensors. β-catenin was titrated up to 1000 nM (7 concentration points with a dilution factor of 2 with one blank injection run at the beginning) in direct binding experiments. In the competition experiments, the β-catenin was kept at constant concentration of 250 nM and the competing peptide TCF4(7–51) was titrated up to 1000 nM (0, 50, 100, 125, 150, 175, 200, 250, 500, and 1000 nM).

Cell culture

RKO cells were obtained from the American Type Culture Collection (ATCC, CRL-2577). The cells were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% fetal bovine serum and 100 μg/ml penicillin-streptomycin (all from Gibco) under humidified atmosphere containing 5% CO2 at 37°C.

Cell lysis

The pelleted cells were resuspended in PBST (10 mM phosphate buffer, 2.7 mM KCl, 137 mM NaCl and 0.05% (v/v) P20, pH 7.4) to a concentration of 20 Million cells per mL. To this suspension, cOmplete^TM protease inhibitor (Roche) and the detergent C12E8 were added to a final concentration of 1%. The cells were lysed for one hour at 4°C while being vertically rotated at 5 rpm. For storage, the lysate was clarified by centrifugation at 14’000 g for 15 minutes at room temperature. The supernatant was removed, and the resulting cell lysate pellet was stored at −20°C.

Data analysis

The SPR data was visualized and analyzed using GraphPad Prism 10.0.0 (Boston, Massachusetts USA, www.graphpad.com). For SPR analysis, signals recorded on the active surfaces were referenced with signals from the reference channel. The signal responses observed in equilibrium were normalized to 100% and plotted against the concentration of protein or peptide, respectively. This allowed the calculation of K_D (specific binding with hill slope) and IC₅₀ (log(inhibitor) vs. normalized response -- Variable slope) values using non-linear regression fit with the following formula:

(1)

Where R_eq being the response in equilibrium, C the concentration of the injected protein (K_D) or peptide (IC₅₀), h the hill slope [43] and X being the fitted K_D or IC₅₀, respectively. The FM data were analyzed using custom made software (lino Biotech AG). The software utilized the python lmfit library [44] to fit the binding signal in equilibrium with respect to protein and peptide concentration. The processed data was then visualized in GraphPad prism.

Click chemistry mediated backfilling of the molosensor surface

For the immobilization of the TCO labelled peptides TCF4(7–30) (specific for β-catenin) as well as the mutated, non-binding version TCF4(7–30) the TCO click chemistry [45] on the sensor surface was used. Here, the sensor surface previously modified with tetrazine (Tz) in grooves and methyltetrazine (MTz) in ridges was applied to immobilize native and mutated peptide, respectively. The mologram notation follows the convention [ridges|grooves]. The response in FM has a unit of pg/mm² and refers to “coherent mass density”. The word “coherent” is important as only the mass difference between ridges and grooves is detected and not the total mass density bound to the surface. The conversion factor between coherent mass density and mass density is known as analyte efficiency [19]. Using the aforementioned peptides, a backfilled (affinity matched) [18] sensor surface [TCF4(7–30)|mutated TCF4(7–30)] as described in previous studies (Fig 2) was generated. A backfilled sensor further improves correction for nonspecific binding by matching biochemical properties of ridges and grooves. This design ensured that non-specific binding towards the peptide was likely to occur uniformly and the sensor will have a flat baseline upon the injection of cell lysate. Importantly, while backfilling is not always necessary, it becomes critical for large ligands (due to their greater exposed surface area) and more crowded complex media. The non-specific binding towards the sensor surface itself is referenced out in any case. To create a backfilled sensor, the two peptides were immobilized at similarly matched ligand density on the ridges and grooves. This approach minimized the initial baseline drift, thereby enhancing sensitivity by improving the signal-to-noise (S/N) ratio, and by promoting an equal distribution of non-specific binders on ridges and grooves. Initially, the sensor surface chemistry consisted of MTz in the ridges and Tz in the grooves (Fig 2, step 1). In the first step, the non-binding TCF4(7–30) mutant was injected onto the sensor. Since the reaction with Tz is faster than with MTz [46], the peptide predominantly immobilizes in the grooves, creating a mass difference between ridges and grooves and resulting in an increase in signal (Fig 2, step 2). Once the Tz linkers in the grooves were depleted, the peptide continued immobilizing in the ridges. As a result, the coherent mass difference between ridges and grooves decreased, leading to a reduction in signal (Fig 2, step 3). In the next step, the protein-specific peptide TCF4(7–30) was injected. This peptide could only bind free MTz linkers in the ridges (Fig 2, step 4) until the sensor surface became fully saturated (Fig 2, step 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Backfilling the FM sensor surface with click chemistry.

Left: The time trace sensogram collected while backfilling of FM sensor with TCF4(7-30) peptide and its mutant. The median trace out of 18 molograms is shown. The coherent mass density increase is displayed on the y-axis. The x-axis shows the time dimension. Right: Scheme of backfilling process on the sensor surface. Initial surface chemistry; MTz in ridges and Tz in grooves (Step 1). In step 2, the non-binding mutant peptide was injected and primarily reacted with Tz in grooves. In step 3, since all Tz is depleted in grooves, the peptide only immobilized in ridges. In step 4, the binding peptide was only immobilized in the ridges until the signal decreased back to baseline. Step 5 depicts fully backfilled sensor surface. Figure was created with BioRender.

https://doi.org/10.1371/journal.pone.0333554.g002

TCF4(7–30) was successfully immobilized with a ligand density of approximately 3 pg/mm², and the sensor was backfilled with mutated TCF4(7–30) covalently on the FM sensor surface using TCO – Tz click chemistry (Fig 2). The increase of 3 pg/mm² for the TCF4(7–30) peptide corresponds to a coherent ligand density of roughly 1.2 fmol/mm² (using a molecular mass of 2500 g/mol for the peptide). To get the real ligand density present on the sensor from the coherent ligand density an assumption for the analyte efficiency is required, it was shown previously that the analyte efficiency is between 10–20% [17,26]. Therefore, the ligand density on the surface is around 5–10 fmol/mm2. In contrast, to immobilize TCF(7–30) on the SPR sensor, the biotin-streptavidin approach was applied. The interaction between biotinylated TCF4(7–30) and the streptavidin on the SPR sensor is highly affine, effectively mimicking covalent binding, and is routinely used as an immobilization strategy in the SPR technique [13,47–49]. Furthermore, the morphology of the two sensor architectures differed in the arrangement of the binding sites in the sensing volume. In FM, the binding sites are rather arranged in a two-dimensional, planar configuration with a layer thickness of 10–20 nm. In SPR, however, for the employed sensors the binding sites are attached via streptavidin to a dextran matrix with a layer thickness of 100 nm [12]. Thus, in the SPR experiments ligand densities approximately one order of magnitude higher can be achieved. This increases the sensitivity towards the analyte, especially for small molecule studies. While SPR currently offers an advantage in detection of small molecule and fragments due to its three-dimensional sensor surface architecture, the two-dimensional planar configuration of FM still provides robust performance for interactions where such high mass sensitivity is not required. FM can also become well-suited for a broader range of analytes including small molecules exemplifying weak or covalent binders.

Focal molography allows quantification of protein-peptide interaction in buffer and complex media

Having established robust ligand immobilization strategies for both FM and SPR, the next step was to quantitatively characterize the binding affinity and binding specificity of the monitored β-catenin–TCF4(7–30) interaction. By comparing the equilibrium binding parameters derived from both platforms, and performing experiments in both simple buffer and complex cell lysate environments, we (i) determined the binding affinity constant (K_D) in both buffer and cell lysate, (ii) confirmed the specificity of the observed interactions in competition experiments, and (iii) assessed the reproducibility of FM-based binding data both within and across sensors by performing three individual runs on the same sensor and over three different sensors.

To determine the binding affinity constant (K_D value) binding experiments with β-catenin were performed on the SPR sensor, functionalized with biotinylated TCF4(7–30) at three different immobilization densities, as well as on the FM sensor, functionalized with TCO-TCF4(7–30) on the ridges and backfilled with mutant TCO-TCF4(7–30), in buffer (Fig 3A) and cell lysate containing 1 Mio cells/mL (Fig 3B). To assess the half maximum inhibition concentration (IC₅₀), β-catenin in solution was supplemented with the free competing peptide TCF4(7–51), which targets the same binding pocket on the protein as TCF4(7–30) immobilized on the sensor surface. Experiments were conducted in both buffer (Fig 3C) and cell lysate containing 1 Mio cells/mL (Fig 3D). The concentrations of the injected protein and free peptide are listed in the method section. Time traces of all experiments are provided in the Supplementary Materials (SPR, Supplementary Figure 3; FM, Supplementary Figure 1 and 2 in S1 File) and exemplary time traces for the dose response and competition experiment are shown in Fig 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. FM and SPR measurements of β-catenin/TCF4 interactions and β-catenin competition with a free peptide.

Normalized sigmoidal dose response (y-axis) of β-catenin (x-axis) response versus concentration monitored in buffer (A) and cell lysate (B). Normalized sigmoidal dose response (y-axis) of TCF4(7-51) (x-axis) response in solution with β-catenin versus concentration monitored in buffer (C) and cell lysate (D). For FM experiments, a representative sensor containing 18 molograms, the normalized equilibrium response (R_eq) for each injection of protein or peptide is plotted from 3 individual experiments. For SPR experiments, the normalized equilibrium response (R_eq) for each injection of protein or peptide from 3 experiments on the same sensor containing 3 different immobilization densities (5, 25, 50 RU, which under the assumption 1 pg/mm² = 1 RU, corresponds to 2, 10, 20 fmol/mm²) is shown. The 95% confidence interval band is shown as dotted line for each experiment. The normalized sigmoidal dose response curve obtained from SPR experiments in buffer are shown in all plots for comparison to FM data. The sigmoidal dose response curves were fitted with the non-linear regression model in Graphpad Prism 10.4.1 (log(inhibitor) vs. normalized response – Variable slope).

https://doi.org/10.1371/journal.pone.0333554.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Exemplary time traces of beta-catenin titration over TCF4(7-30) peptide surface and corresponding competition experiments with a free TCF4(7-51) peptide collected with FM.

(A) Dose response of beta-catenin (up to 1000 nM) to determine binding affinity constant; (B) Competition of beta-catenin by a free peptide TCF(7-51) titrated in dose response (up to 1000 nM) at 250 nM beta-catenin.

https://doi.org/10.1371/journal.pone.0333554.g004

For the data obtained in SPR experiments in buffer (Fig 3), with non-linear regression fit (using formula (1)), a K_D of 96–121 nM (in the 95% confidence interval (CI)) was determined. Using formula (1) for the calculation of IC₅₀, the measured values ranged from 95 to 115 nM (within the 95% CI). This data is in a good agreement with the findings presented by Yu et al. [42]. From the attempts of experiments with cell lysate on the SPR sensor, no information could be gathered about the interaction affinity of β-catenin towards the TCF4(7–30) immobilized on the surface due to massive, non-specific background interactions (Supplementary Figure 6 S1 File). In contrast, when injecting the RKO cell lysate on the peptide functionalized FM sensor surface, no response was observed (Supplementary Figure 7 in S1 File). The RKO cell line was selected since it was showing the lowest basal β-catenin level compared to other cell lines (Supplementary Method 1; Supplementary Figure 8 in S1 File).

For FM, the normalized equilibrium response (R_eq) obtained from each injection of protein or peptide in buffer and cell lysate is shown across the 18 molograms for a representative sensor is shown (Fig 4). These values were plotted against the injected concentrations of protein or peptide respectively, in order to compare the sigmoidal curve shape of SPR and FM (Fig 4). A clear similarity in the sigmoidal curves between SPR and FM experiments could be observed, indicating that the binding mode in both measurement systems were comparable. However, the Hill slope of the dose-response observed in FM was higher (1.88 ± 0.07, for 3 Hill slopes) in comparison to SPR (1.21, only one Hill slope) for experiments conducted in buffer (Fig 3A). This may indicate a higher surface density of ligands available for binding [50] in FM compared to SPR. Moreover, the planar nature of the FM surface contrasts with the three-dimensional dextran matrix used in SPR, which can significantly affect ligand orientation and accessibility. The dextran matrix may create steric hindrance or alter ligand orientation, influencing binding dynamics and reducing apparent ligand density and accessibility, thereby affecting the Hill slope observed in SPR [51–54]. This phenomenon aligns with findings from molecular dynamics simulations, which have shown that ligand orientation and conformational changes can significantly influence receptor activity and binding affinity [51,55]. Additionally, the SPR system might exhibit avidity effects due to multivalent interactions facilitated by the dextran matrix, contributing to lower observed Hill slopes. A modeling effort considering these interactions could provide a more comprehensive understanding of surface architecture impacts on binding [51,56]. While a higher Hill slope in FM indicates increased ligand density, other factors such as cooperative binding and conformational changes should also be considered. Cooperative binding could increase the Hill slope in FM if binding of one ligand increases the affinity for subsequent ligand binding. Similarly, conformational changes upon ligand binding may alter effective ligand density. This is supported by studies showing structural changes upon binding, which can alter ligand accessibility and binding site presentation [51,53]. The effect decreased when comparing the curves of SPR (1.21) in buffer and FM (1.6 ± 0.06) in cell lysate (Fig 3B). That could be explained by the presence of cell lysate components partially blocking the FM sensor surface and changing the conformation and the mode of action between the protein and ligand, thus decreasing the Hill slope. Such conformational changes are known to affect binding dynamics, as indicated by several studies [51–55]. The specificity of the β-catenin/TCF4(7–30) interaction was demonstrated in the competition experiments. The results underlined the outstanding performance of FM to detect specific interaction in a complex environment. The sigmoidal curve response plots for all sensors can be found in Supplementary Figures 4 and 5 in S1 File.

In order to evaluate the reproducibility of the binding affinity and half maximum inhibition constants determined in FM experiments, each of the 18 molograms on each chip in each experiment (run) were analyzed and the results were visualized with boxplots (Figs 5–6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Overview of binding affinities (K_D) and half maximum inhibitory concentrations (IC₅₀) obtained in FM measurements performed in buffer.

The experiments were in triplicates on 3 different sensors. The boxplot analysis represents the indivudual binding affinities (A, K_D) and half maximum inhibitory concentrations (B, IC₅₀) calculated from 18 indivudual molograms with software provided by Lino Biotec. The box contains 50% of the values and the median value is represented as line, while the whiskers contain the upper and lower 25% of all molograms of an experiment. (C) Median K_D and IC₅₀ values calculated from 18 molograms as well as the avarage values calculated from three experiments with standart deviation.

https://doi.org/10.1371/journal.pone.0333554.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Overview of binding affinities (K_D) and half maximum inhibitory concentrations (IC₅₀) obtained in FM measurements performed in cell lysate (1 Mio cells/mL).

The experiments were in triplicates on 3 different sensors. The boxplot analysis represents the indivudual binding affinities (A, K_D) and half maximum inhibitory concentrations (B, IC₅₀) calculated from 18 indivudual molograms with software provided by Lino Biotec. The box contains 50% of the values and the median value is represented as line, while the whiskers contain the upper and lower 25% of all molograms of an experiment. (C) Median K_D and IC₅₀ values calculated from 18 molograms as well as the avarage values calculated from three experiments with standart deviation.

https://doi.org/10.1371/journal.pone.0333554.g006

The K_D values determined in buffer from FM experiments ranged from 68 to 160 nM, while SPR analysis yielded values from 96 to 121 nM, demonstrating good comparability between both assays. Furthermore, the FM competition experiments with the free peptide resulted in IC₅₀ values ranging from 110 to 200 nM whereas SPR yielded values ranging from 95 to 115 nM further confirming reproducibility and reliability of the FM measurements.

This consistency was observed not only across different molograms in the same experiment but also across multiple experiments on the same sensor as well as across different sensors. However, the remaining deviations could be attributed to differences in flow profiles within the FM measurement chamber. The FM measurement chamber is significantly lager, especially in the direction perpendicular to the flow (12.0 x 1.9 x 0.1 mm) compared to SPR (2.1 x 0.5x 0.05 mm) [57]. Consequently, when samples were injected into the FM chamber, border effects may have impacted the distribution and velocity of the samples across individual molograms. This could be resolved in the future by miniaturizing the sample chamber of FM. Moreover, although similar immobilization densities while functionalizing the FM sensors with the peptides were targeted, differences could be due to variations in the density at the level of chemical linkers (MTz and Tz) during the chip fabrication process. These differences can influence the final ligand densities immobilized on the sensor surface. This is not only true for the different sensors, but also for the different molograms on the same sensors, as the quality of the molograms varies due to reactive immersion lithography process that was not performed in a displacement Talbot mode for the chips used in this study [58]. Nonetheless, these effects were only insignificantly influencing the observed binding affinities and half-inhibition concentrations (K_D and IC₅₀ values).

Next, the comparability of the binding data for the protein-peptide interaction were analyzed in cell lysate (Fig 6).

The determined K_D values for the protein-peptide interaction in cell lysate using FM ranged from 100 to 140 nM, while SPR analysis yielded 96–121 nM for the same interaction in buffer. These results demonstrate that FM can reliably measure this interaction in a crude sample, in presence of non-specific binders found in cell lysate on the chip matrix. This assay would not be straightforward to implement with SPR, highlighting the power of FM’s background subtraction, without intensive and time-consuming assay optimization. Furthermore, FM experiments in cell lysate revealed an IC₅₀ between 140–170 nM in excellent agreement with the results obtained using SPR in buffer (95–115 nM). Also, this assay would involve considerable optimization on an SPR device. In summary, FM enabled the comparable and reproducible measurement of binding affinities and half maximum inhibition concentrations for protein-peptide interactions in buffer and cell lysate whereas SPR was limited to measurements in buffer.

Slight deviations of binding affinities and inhibition concentrations between individual molograms on the same sensor (within a factor of 2) could again be attributed to variations in the flow profile within the measurement chamber. Likely, not all molograms received the exact same concentration of samples due to border effects in the measurement chamber, hence small variations in the calculated K_D and IC₅₀ values were expected and observed. Furthermore, initial coupling of the incident light beam to the sensor chip (waveguide) is different when changing the sensors or experiments [27]. Important in this case would be to ensure similar initial coupling conditions when comparing binding results. Additionally, a slight decrease in binding affinity (higher K_D values) in cell lysate was observed compared to the experiments performed in buffer. This observation could be explained by β-catenin or free TCF4(7–51) peptide interacting with components of the cell lysate, leading to competing interactions and lowered effective concentrations.

One key finding of our study was that molecules randomly adsorbed from the cell lysate onto the FM sensor surface did not contribute to the recorded signal during the experiments. This highlights a significant advantage of FM diffractive sensor over classical refractometric sensors like SPR, where randomly adsorbed molecules cause a refractive index change, leading to a signal increase that can obscure the specific binding signal of the analyte to the ligand, even after subtraction of the signal measured in the reference flow channel.

In contrast, FM effectively eliminates this issue. On a backfilled FM sensor, randomly adsorbed molecules distribute equally between ridges and grooves due to the highly similar chemical environments and the sub-micron proximity of reference and signal regions. Since the FM signal only increases when mass accumulates unequally between ridges and grooves, no signal is generated from non-specific binding. Specific interactions, such as the binding of β-catenin to the peptide on the sensor surface, result in an unequal distribution of mass between ridges and grooves, thereby producing a measurable signal only upon specific binding.