Oriented Antibody Coupling to an Antifouling Polymer Using Glycan Remodeling for Biosensing by Particle Motion

Biosensors based on immobilized antibodies require molecular strategies that (i) couple the antibodies in a stable fashion while maintaining the conformation and functionality, (ii) give outward orientation of the paratope regions of the antibodies for good accessibility to analyte molecules in the biofluid, and (iii) surround the antibodies by antibiofouling molecules. Here, we demonstrate a method to achieve oriented coupling of antibodies to an antifouling poly(l-lysine)-grafted-poly(ethylene glycol) (PLL-g-PEG) substrate, using glycan remodeling to create antibody–DNA conjugates. The coupling, orientation, and functionality of the antibodies were studied using two analysis methods with single-molecule resolution, namely single-molecule localization microscopy and continuous biosensing by particle motion. The biosensing functionality of the glycan-remodeled antibodies was demonstrated in a sandwich immunosensor for procalcitonin. The results show that glycan-remodeled antibodies enable oriented immobilization and biosensing functionality with low nonspecific binding on antifouling polymer substrates.


Modification of trastuzumab
The protocol for glycan remodeling was provided by Synaffix.The protocol was executed as received, without optimizations.Initially, trastuzumab was modified and analyzed to master the protocol, optimize analytical protocols, and optimize DBCO-ssDNA ratio for click-coupling.Trastuzumab is a human IgG1 kappa antibody that is used in the treatment of breast cancer due to its interfering properties with HER2 [1].The antibody was used because of its easy availability and similarity in antibody type to the anti-procalcitonin-antibody used in the paper, which is also an IgG1 antibody.Both trastuzumab and anti-procalcitonin-antibody 13B9 contain glycosylation sites only in the Fc domain.
LC-MS analyses of the different modification steps of trastuzumab (fig.S1) show similar results as the modification of anti-PCT 13B9 (fig.2).The mass spectrum of unmodified reduced Abs showed the presence of glycans at the heavy chains (HC), with masses ranging from 50.550 Da to 50.940 Da (fig.S1A).After glycan removal using endoglycosidase SH, the MS spectrum shows a reduction in mass of the HC, showing peaks corresponding to the presence of only one N-acetylglucosamine (49.352Da) or one N-acetylglucosamine with a fucose (49.499Da) (fig.S1B).Following attachment of GalNac6N 3 , the mass of the main peaks shifted with 203 Da to 49.554 Da and 49.701 Da for the glycans with and without fucose present respectively, corresponding to the mass addition of GalNac6N 3 of which the azides are reduced by DTT (fig.S1C).The MS spectrum also shows small peaks at the masses of HC's without GalNac6N 3 present, indicating that GalNac6N 3 was not attached to all glycosylation sites.To determine the yield of glycan attachment, the peak area's with and without GalNac6N 3 were compared.Glycan remodeling yield was calculated following equation S1: in which I GalNac6N3 is the peak area of the HC with GalNac6N 3 attached and I 0 is the peak area of the HC without GalNac6N 3 .The glycan remodeling yield is approximately 81%.
Gel-electrophoresis was performed of reduced Abs to determine the Ab-DNA conjugation yield of Trastuzumab (Fig. S2).The gel shows the Abs at the different stages of the modification together with controls containing the enzymes without Abs.All controls (lanes 3, 5, and 7) show no signal, indicating no interference of the enzymes or DNA in the gels.In comparison to the unmodified Ab (lane 1), the Abs shows a small reduction in size after glycan removal (lane 2), and no change in size after GalNac6N 3 attachment (lane 4), as the change in mass is too small to be observable in gel electrophoresis.After attachment of DBCO-ssDNA, a band appears at ~73 kDa, indicating successful attachment of DNA to a fraction of the Abs (lane 8).Non-specific attachment of DBCO-ssDNA to the Ab does not occur, as this band at ~73 kDa is not observed when intact Abs are incubated with DBCO-ssDNA (lane 6).
Next, the Ab-DNA conjugation yield was optimized by incubating antibodies with various equivalents of DBCO-ssDNA and analyzing the conjugates using SDS-PAGE (fig S3).DBCO-ssDNA to Ab equivalents ranging from 2 to 10 were used.The Ab-DNA conjugation yield was determined using the relative intensity of the two bands of unconjugated Abs and Ab-DNA conjugate at ~50 kDa (I Ab ) and ~73 kDa (I Ab-DNA conjugate ) respectively following equation S2: A maximum conjugation yield between 60% and 64% was reached at 4 equivalents of DBCO-ssDNA, and did not increase with additional DBCO-ssDNA.

Modification of anti-PCT Antibody 13B9
Modification efficiencies of the anti-procalcitonin antibody 13B9 are discussed in the results section.Figure 2A3 shows the mass spectrum of Abs to which GalNac6N 3 is attached.This sample was measured in quintuplicate, for which the peak areas at 49.648 and 49.444 Da, shown in table S1, were measured and used to determine the modification yield using equation S1.The modification yield (+SEM) of anti-PCT antibody 13B9 is 80.6 ± 0.5%.
The conjugation yield of anti-PCT antibody 13B9 to DBCO-ssDNA was analyzed using gel electrophoresis, for which results are shown in figure 2B.Gel electrophoresis was repeated to accurately determine the conjugation yield, for which the results are shown in table S1.The conjugation yield (+SEM, calculated using equation S2) of anti-PCT-antibody 13B9 to DBCO-ssDNA is 33 ± 2%.
Table S1.Modification yield of anti-PCT antibody 13B9.Relative peak area of the masses 49.648 and 49.444 measured in the mass spectra of reduced, remodeled anti-PCT antibody 13B9.The peaks correspond to the antibody without and with GalNac6N 3 , respectively.The modification yield was calculated using equation S1.

Orientation of coupled antibodies
Antibodies coupled to silica particles (1 µm diameter) were imaged in 3D using single-molecule DNA-PAINT microscopy.For each particle the number of localizations in the green and red fluorescence channels were counted.Particles on the edges of the FOV were excluded from analysis because at the edges the inhomogeneous laser intensities resulted in significant differences in the observed number of localizations.The XZ views of particles used to construct fig. 3D

Procalcitonin BPM sensor
BPM using Ab-DNA conjugates functionalized on a PLL-g-PEG coating on a polystyrene substrate was compared to physical adsorption of Abs on a glass surface.Figure S5 shows the diffusivity histograms of both functionalization strategies with serial additions of PCT to the sensor.The data shown in the red panel in Figure S5 is the same as for Figure 4C.

Figure S1 .
Figure S1.Trastuzumab glycan remodeling at different stages; Mass spectra.Deconvoluted mass spectra of heavy chain (HC) fragments of Abs at different stages of the glycan remodeling.Glycan structures are indicated next to the most prominent peaks.The inserts show the chromatograms of the corresponding mass spectra.A) HC of reduced trastuzumab with intact glycosylation sites.B) HC of reduced trastuzumab after glycosylation.C) HC of reduced trastuzumab with GalNac6N 3 coupled to the glycosylation site.The peak intensity at 49.701 Da is 4.6•10 6 and at 49.497 Da is 1.1•10 6 .The mass spectra are taken from the components with an LC retention time between 6 and 8 minutes.
are shown in fig.S4.Particles functionalized with biotinylated Abs show 77.8 ± 9.1 % (±SD) green localizations and particles functionalized with Ab-DNA conjugates show 99.5 ± 0.6 % (±SD) green localizations.The experiments show small particle-to-particle variations in a single measurement.Different experiments resulted in different densities of particles on the substrate.Due to the limited availability of the equipment, only a few experiments could be performed.

Figure S4 .
Figure S4.XZ views of particles with two biofunctionalization methods: biotinylated Abs (top) and Ab-DNA conjugates (bottom).Green localizations indicate protein M binding to Fab sites and red localizations indicate protein G binding to Fc sites.
The sensor with physically adsorbed antibodies shows that many particles are already bound without PCT, indicating non-specific interactions.With increasing PCT concentrations, the fraction of bound particles rapidly increases and at 333 pM almost all particles are stuck.There is an abrupt change of freely moving particles at 111 pM to almost all particles stuck at 333 pM, indicating that non-specific binding of PCT might play a role.The sensor functionalized with PLL-g-PEG and Abs conjugates shows a much lower signal in the absence of PCT, indicating that there are less non-specific interactions.Furthermore, the sensor response increases more gradually and over a larger concentration range, which suggests that nonspecific interactions are reduced with this coupling methodology.

Figure S5 .
Figure S5.Diffusivity histograms for two substrate functionalizations; physically adsorbed Abs to a glass substrate (top, blue) and Ab-DNA conjugates coupled to PLL-g-PEG on a polystyrene substrate (bottom, red).Diffusivity histograms are shown for increasing PCT concentrations with serial additions of PCT.

Table S2 . Conjugation yield of anti-PCT antibody 13B9.
Relative intensities of the bands detected at ±50 kDa and ±75 kDa, corresponding to HC and HC-DNA conjugate respectively, in various gel electrophoresis experiments used to determine the percentage of HC antibody fragments that are conjugated to DNA.The conjugation yield was calculated using equation S2.