3.1. Functionalization of Gold Surfaces
In view of the detection of small molecules by SPRi detection, we aimed at an amplification of the SPRi signal by the use of gold nanoparticles (AuNPs). Those nanoparticles have two main effects: besides an amplification phenomenon due to their mass, we may also expect a coupling between the localized surface plasmon of the AuNPs and the surface plasmons of the biosensor itself [30]. The latter effect is present as soon as the AuNPs are sufficiently close to the sensor surface (distance lower than 10 nm) [31,32]. However, this amplification requires successful functionalization on both gold surfaces with anti-fouling molecules in order to avoid non-specific adsorption. 
PEG molecules are commonly used to decrease protein spontaneous adsorptions on surfaces and thus increase signal to noise ratios. This functionalization approach was successfully used for the detection of proteins by aptamer microarrays [21,22]. We adapted this method to the case of sandwich assays with split-aptamers grafted to the gold surface of SPRi prisms and nanoparticles. Two different lengths of thiolated PEG molecules have been tested for self-assembling monolayer (SAM) formation on gold: short and long ones of respectively 300 Da (PEG300) and 2 kDa (PEG2000). As can be seen from Figure 1, the longer PEG co-grafted with the split-aptamer sequences gave impressive results. While injecting gold nanoparticles grafted with Split-APT and PEG2000, no signal was observed on the gold surface of the SPRi prisms (Au curve in yellow). Furthermore, no signal was observed also on spots grafted with negative control sequences CN8 and PEG2000 (CN8 curve in grey). The lack of non‑specific interactions and SPRi signal on both cases insures that any SPRi signal observed on other spots effectively corresponds to specific interactions.
Figure 1  Reflectivity shifts observed upon injection of gold nanoparticles grafted with Split-APT sequences on various spots of the Surface Plasmon Resonance imaging (SPRi) biosensor surface without the target adenosine present in the solution. Signal increase is observed only on Split-APT8 spots through hybridization of the hairpin stems of the split‑aptamers. APT8 and APT4 spots do not present signal shift due to a folding of the complete aptamer on the surface while Split-APT4 sequences present stems (four bases) too short to hybridize with the Split-APT sequences grafted on the gold nanoparticles. Lack of non-specific signal is also confirmed on control spots with sequences CN8 and Split-APT or on pure gold.   The same experiments were realized with PEG300 on one or both gold surfaces (prisms or AuNPs). These trials were always leading to increased non-specific signals compared to PEG2000 grafting. Thus, the grafting with PEG300 was not considered for the following experiments, only PEG2000 co‑grafting of the spotsand the AuNPs were considered. In conclusion, it seems that the hydrophilic behavior brought by the long PEG chains avoid any non-specific interactions and justifies the use of this original grafting strategy for sandwich assays with AuNP SPRi signal amplification.