2. Can Covalent Poly(Amidoamine) Dendrimers Efficiently Bind Heparin as Protamine Replacers?
As mentioned in the previous section, protamine is the only clinically approved agent to be employed in heparin anticoagulant activity reversal during chirurgical and medical practices. Derived from shellfish, this small arginine-rich cationic protein causes adverse reactions in nearly 10% of patients, and up to 2.6% of cardiac surgeries experience serious complications due to protamine suboptimal administration [30]. Since covalent dendrimers can mimic many aspects of protein behavior [31], we reasoned that these hyperbranched molecules, and particularly the poly(amidoamine) (aka PAMAM) dendrimers originally reported by Don Tomalia in 1985 [32], which feature positively charged groups on their outer surface, could work well as heparin ligands. Thus, we began our journey in the quest of possible protamine replacers by studying heparin binding by ethylenediamine (EDA)-core PAMAMs with different branching generations (G0–G6) by means of a combined experimental/computational approach [33].
The investigation started with an experimental heparin binding competition assay (HBCA) exploiting Mallard Blue (MB), a deep-blue colored, positively charged (+5) synthetic dye based on an arginine-functionalized thionine. This compound was specifically developed by our group for ultra-high-affinity sensing of clinically-relevant heparin levels in both physiological solutions and serum [34]. The dye reports on heparin levels by quantitatively binding to the polyanion; this, in turn, results in a significant change in its UV−visible spectroscopic profile. Based on this evidence, we conceived the HBCA as follows: First, two solutions containing heparin and MB are mixed and then the resulting system is gradually titrated with solutions containing protamine and PAMAM dendrimers of varying generations. In the presence of increasing concentration of each possible heparin binder, the dye is gradually displaced from heparin and the absorbance intensity of free MB increases accordingly, as shown in Figure 2a. Curve fitting of the data shown in this figure allowed us to calculate the main binding parameters, i.e., the effective heparin binder concentration require to displace 50% of the MB dye from the polyanion (EC50), and the charge excess, that is the number of binder positive charges per heparin negative charge (CE50) required to attain 50% displacement of the MB dye. These values are listed in Table A1.
From Figure 2a and Table A1 we see that the smallest dendrimer of the series (G0) is clearly not able to displace MB from the polysaccharide. On the other hand, as the dendrimer generation grows, the corresponding EC50 value decreases (from 10.10 µM of G1 to 0.22 µM of G6) by virtue of the increasing dendrimer surface charge leading to higher heparin/binder affinity. However, in terms of a given dendrimer’s ability to exploit each individual positive charge in heparin binding (CE50), the observed behavior is not monotonic, in that both the smallest (G1) and the biggest (G6) dendrimers are characterized by the worst (i.e., highest) EC50 value, while the other PAMAMs (G2–G4) exhibit better heparin binding performance in this respect (Table A1). An utterly analogous trend is observed when considering the ligand dose required to bind 100 IU of heparin (last column in Table A1). The touchstone molecule, protamine, is characterized by an EC50 value of 2.34 µM, a CE50 of 0.52, and a corresponding dose of 0.32 mg/100 heparin IU (Table A1). Therefore, a global reconsideration of all values listed in Table A1 leads to the conclusion that the most promising protamine replacer among the entire PAMAM dendrimer series considered is G2, not only because its heparin binding parameter set (EC50 = 2.55 µM and CE50 = 0.38) and dosage required (0.25 mg/100 heparin IU) most favorably compare with those characterizing the small arginine-based protein but also because low generation dendrimers are generally endowed with lower in vivo toxicity effects [35].
This experimental evidence was rationalized using computer-based molecular dynamics (MD) simulations (see Figure 2b–d). To the purpose, we exploited the concept of effective free energy of binding ∆Gbind,eff, that is the specific energetic contribution to each heparin/binder complex formation afforded only by those dendrimer (or protamine) charges in constant and productive interaction with the polyanion. The analysis of each heparin/binder nanoassembly MD simulation allowed us to precisely identify and quantify these dendrimer (protamine) positively-charged groups (Neff), estimate their individual contribution to the overall binding energy, and cumulatively express this as ∆Gbind,eff, as shown in Table A2. Yet, to be able to compare simulation data both among themselves and with experimental CE50 values, we further introduced the concept of effective-charge-normalized binding free energy, i.e., ∆Gbind,eff/Neff, as shown in the last column of Table A2. As a result of these in silico experiments, it appears evident that although 6 out of 8 positive charges of the small G1 dendrimer are productively engaged in heparin binding, their overall effective binding contribution is quite low, just −0.19 kcal/mol (Table A2). On the other hand, the largest G6 dendrimer, despite its plethora of positively-charged amine termini (+256), is able to organize only 45 of them for best productive heparin binding (Figure 2d), most likely because of its intrinsic high molecular rigidity. Accordingly, this big hyperbranched molecule is a sub-optimal polyanion binder (∆Gbind,eff/Neff = −0.40 kcal/mol, Table A2). The G3 and G4 PAMAMs behave quite similarly to each other, with 15 and 16 positive charges in contact with heparin, respectively, ultimately resulting in comparable, intermediate effective-charge-normalized affinity values (−1.06 and −0.91 kcal/mol, respectively, Table A2). Finally, the last G2 dendrimer (Figure 2c) is not only able to exploit 13/16 terminal groups in complexing heparin but it also does so in the most effective way (∆Gbind,eff/Neff = −1.30 kcal/mol, Table A2), in agreement with the corresponding experimental CE50 value (Table A1). If compared with the predicted values for protamine (Neff = 12 over 24 available charges, and ∆Gbind,eff/Neff = −0.33 kcal/mol, Table A2), it appears that the in silico analysis also identifies the EDA-core G2 PAMAM dendrimer as the preferred polyanion binder in the search of an alternative heparin anticoagulant activity reversal agent.
With this double information at hand, we next verified whether the G2 PAMAM could still operate heparin binding in the presence of serum, and compared the relevant results with those obtained using protamine in the same environment. Very pleasingly we found that, under these substantially more challenging conditions, while protamine worsened its performance (EC50,serum = 3.51 µM, CE50,serum = 0.79, and dose = 0.49 mg/100 heparin IU) the G2 dendrimer slightly enhanced its binding affinity for the polyanion (EC50,serum = 2.15 µM, CE50,serum = 0.32, and dose = 0.21 mg/100 heparin IU). Notwithstanding these highly encouraging results we ultimately reasoned that, even if low generation PAMAM dendrimers are often overlooked for biomedical applications despite their good toxicological profiles [35], a substantially more degradable, less expensive, and easier-to produce system could be more desirable for safe and effectual production under good-manufacturing-practice (GMP) quantities required for the translation of a new protamine replacer to the clinic. Bearing this belief in mind, we then plunged ourselves into the sea of self-assembly nanoscale multivalent surfaces in the hope of finding an alternative, even more responsive, synthetic, protamine-mimic heparin binder.