The covalent coupling of heparin to titanium dioxide was realised in this study by spacers with different chain lengths like APMS, Di- and Triamino-APMS (Figure 1). The hypothesis was, that the use of longer spacer chains should result in a higher mobility of immobilised heparin and hence a higher biological potency of the drug. Coupling of heparin to the spacer modified surfaces was performed following activation of carboxylic groups with EDC and NHS, which then react with primary amino groups of the spacer molecules (Figure 1a), whereas the secondary amino groups (R-NH-R') are not able to form amide bonds with carboxyl groups. The successful modification of TiO2 surfaces with coupling agent and heparin was demonstrated by measuring zeta-potential changes in aqueous solution (Figure 2) since both the spacer as well as the drug led to a change of the potential due to charged amino and sulphate groups in the modification.
Although this reaction took place for all three coupling agents, the amount of immobilised spacer and drug was significantly different in the order APMS > Diamino-APMS > Triamino-APMS. This phenomenon can be explained, since the larger molecules like Triamino-APMS are sterically more hindered on the titania surface than the smaller ones. Furthermore, a higher number of secondary amino groups result in a stronger intermolecular electrostatic repulsion. Both effects lead to less pronounced deposition of spacer molecules on the surface. Comparing the values of Table 1 with those of the literature, where a monolayer of NH2 groups is described with 2–3 NH2/nm2, the here accomplished aminations are in the range of multilayers . Furthermore, the number of amino groups on Ti discs was multiple higher than those on the powders (Table 1). However, calculating the surface area of the discs, ideally plane surfaces were supposed, whereas the high micro roughness of the real surface and hence a higher actual surface area was not considered.
A similar behaviour was found for the amount of immobilised heparin which subsequently decreased with increasing the length of the spacer (Table 2). This correlates with the highest density of surface bound primary amino groups of APMS that can form bonds with carboxylic groups of the heparin. If the amount of attached heparin is compared with literature values for heparin modified polymers [15-17] (Table 3), the here obtained values are much lower. This is probably due to the fact, that polymers can bind heparin over their whole volume what can be considered as a heparinisation in the range of many multilayers. Although the amount of immobilised heparin was lower for Triamino-APMS, this coupling agent showed the highest biological potency of the covalently attached drug (Figure 3), which confirmed our hypothesis that longer spacer chains will maintain the biological activity of the material due to a higher flexibility of the immobilised heparin.
The influence of the secondary amino groups on the stability of heparin immobilisation was investigated by measuring the hydrolysis rate of the drug under in vitro simulated physiological conditions (37°C, PBS, 70 shakes per minute) (Figure 4). In an aqueous medium, these secondary amino groups are protonated to R2NH2 +, which makes them able to interact with negative sulphate, carboxylate and hydroxyl groups of the heparin via electrostatic interactions. This ionic bonding is weaker than the covalent one and should have only an additional effect. It was shown that spacers like Triamino-APMS with many secondary amino groups had the highest hydrolysis rate of the drug during the first 100 h, compared to Diamino-APMS and APMS. This effect is up to heparin molecules that are only bound electrostatically, which makes the hydrolysis easier and faster. After this period, the hydrolysis profile inverses and the drug hydrolysed slower from Triamino-APMS than from the other ones. The reason probably is, that the remaining heparin molecules are both covalently and ionically bound on the Triamino-APMS compared to pure APMS that can only bind the drug covalently.
The primary reaction after incorporating an implant into the body is the formation of a biofilm composed of adsorbed matrix proteins. The quantitative composition of this protein film and conformational changes during adsorption are thought to be the key factor controlling the attachment of cells and the biological response to the implant material . In case of materials in contact with blood, it is also known that the adsorption of various proteins combined with structural changes of the protein layer, e.g. fibrinogen can initiate blood clothing. The adsorption of fibrinogen to heparinised TiO2 surfaces was measured in this study using the QCM-D technique. The QCM-D method offers two advantages, firstly, the adsorption can be followed in real-time due to the mass increase of the quartz crystal and secondly, it is possible to analyse conformational changes due to the viscoelastic properties of the adsorbed protein layer. Therefore, the gold surfaces of the quartz crystals were modified in a first step with a thin (approx. 100 nm) TiO2 layer using the PVD technique which were then coated with the coupling agents APMS/Di-/Triamino-APMS and heparin similar to the methods described for titanium substrates. The concentration of the extracellular plasma protein fibrinogen (in PBS, pH = 7.4) used was 50.0 μg/mL. Although this was much lower compared to that of human blood (3 mg/mL), this lower concentration allows single fibrinogen molecules to interact with the surface, before the surface area around the protein is blocked by newly adsorbed fibrinogen molecules. The resulting two effects are firstly a tight bound layer of fibrinogen which contains less water, whereby the effect of water in the protein layer during the QCM-D measurment is reduced. Secondly, lower protein concentrations (μg/ml) advance surface effects on the protein adsorption and thus the QCM-D measurements provide more information regarding differently functionalised surfaces for protein adsorption .
Figure 6 shows a correlation between ΔD and Δf for the differently heparinised surfaces which is helpful for analysing and comparing the viscoelastic properties of protein films on different surfaces since Δf and ΔD did not have the same time dependency such that time as parameter can be eliminated due to this approach. In various experiments it was shown that a low ΔD/Δf is an indication for rigid adsorbed films . The results demonstrate that the ΔD/Δf values for surfaces modified with APMS and heparin nearly built up a line with only a small increase at higher frequencies. In contrast, the ratios of the heparinised surfaces by means of Diamino-APMS and Triamino-APMS are generally higher. This behaviour indicates structural changes within the adsorbed protein layer .
Lower ΔD/Δf ratios for the differently heparinised surfaces indicate a more dense package of the protein layer onto the surfaces which is either due to conformational changes of the adsorbed protein layer (denaturation) and/or a high affinity of the binding sites of the protein to the substrate. Fibrinogen is a long rod-like molecule with three different globular sub-domains (E, D and αC domains). Under physiological conditions the overall charge of fibrinogen is negative. The D and E domains are negatively charged but the αC domains are positively charged. The αC domain acts as a pioneer in the surface binding process, but it does not bind strongly to the surface . It seems certain, that the differences for the ΔD/Δf ratios are partially caused by electrostatic interactions between the domains of fibrinogen with the heparin and respectively the different spacer molecules that are bound onto the substrate.
Indeed, it could be figured out through zeta-potential measurements that surfaces from substrates modified with APMS, Di- and Triamino-APMS and the respective subsequent heparinised surfaces (Figure 2) had positive and accordingly negative charges that were all very similar. But even if these overall charges of the respective functionalised surfaces are all in the same range, it is likely, that the secondary amino groups of the spacer molecules Di- and Triamino-APMS interact with the positively charged αC domains of fibrinogen while adsorbing. On the one hand, this could have an influence on the adsorption kinetics of the protein and explain, why the Δf curves of the haparinised surfaces by means of Di- and Triamino-APMS are similar and arise faster compared with that of APMS. On the other hand, this electrostatic interaction could influence conformational changes in the adsorbed protein layer and therefore be responsible for the highest ΔD/Δf ratio for Triamino-APMS.
It is also likely that sterical reasons are responsible for the different ΔD/Δf ratios from Figure 6. Protein layers could be packed denser on surfaces that were modified with the spacer APMS with the shortest molecule chain rather than on Di- and Triamino-APMS modified surfaces, since APMS molecules underlie a lower sterical hindrance among each other. This effect could result in a subsequent denser functionalisation with heparin which finally allows a stronger interaction of fibrinogen with the drug and therefore a denser package, which results in lower ΔD/Δf ratios compared to the heparinisation with longer spacer molecules like Di- and Triamino-APMS. Since a densification of protein films is normally accompanied by conformational changes, it is likely that these changes are less pronounced on the Triamino-APMS surfaces compared to the shorter coupling agents.