The nature of dendrimer–protein interactions has been investigated in numerous publications ^{[1][2][3][4][5][6][7][8][9][10][11]}. In general, the binding of dendrimers with proteins is driven by the electrostatic interactions between dendrimers and oppositely charged amino acid residues. However, other forces are also involved in complex formation. These are the hydrophobic interactions of non-polar groups of dendrimers and hydrophobic parts of amino acids, hydrogen bonds and van der Waals interactions. The contribution of each type of interaction strongly depends on the dendrimer characteristics.

The charged cationic and anionic dendrimers are generally believed to interact with proteins through electrostatic forces ^[1][3][12], while neutral dendrimers tend to form complexes through hydrophobic interactions and hydrogen bonds, as was exemplified by PAMAM-OH and PPI dendrimers modified with terminal maltose groups ^[13][14][15]. Nevertheless, some authors report that hydrophobic forces and hydrogen bonding are driving forces for interaction even for charged dendrimers. In ^[4], the hydrophobicity has been determined to play a major role in interactions between dendrimers with various terminal groups (cationic primary amine group, anionic sulfate group, a highly hydrophobic benzoate group, and the monosaccharide α-D-mannose) with intrinsically disordered protein NUPR1. However, molecular dynamic (MD) simulations point out that hydrogen bonds and electrostatic forces played a further key role, modulating the hydrophobic interactions. In another work ^[16], hydrophilic, H-bonding and van der Waals interactions have been recognized as the driving forces of the PAMAM G4.5 interaction with trypsin and trypsin inhibitor.

It should be noted that electrostatic interactions are also observed for similarly charged dendrimers and protein. The protein charge determined by zeta-potential measurements corresponds to the net charge of the molecule. In fact, the heterogeneity of charge localization always exists. Although the net charge of the molecule could be positive, there are oppositely charged domains, which offer the opportunity for complexation with polycations, and vice versa. This behavior has been previously reported for interactions of similarly charged proteins and linear polyelectrolytes ^[17][18][19]. This phenomenon was investigated in more detail in ^[5]. The authors determined that PAMAM G5.5 dendrimers and BSA possessing the positive charge at pH 5.0 interact through electrostatic forces due to the negatively charged structural domain I of BSA. At physiological pH 7.5, both BSA and dendrimers have a negative charge. In this case, binding occurs via interaction of dendrimers with positively charged domain III. In the work ^[6], the researchers demonstrated an effective interaction of cationic pyridylphenylene dendrimers with positively charged prion protein. The complex formation proceeds effectively with high binding constants between positively charged dendrimers and negatively charged regions of the protein surface, as was determined by MD simulations.

Nevertheless, in general, the complex formation is determined by a combination of molecular interactions. The contribution of each type of interaction may be assessed through a careful evaluation of thermodynamic parameters of binding, which could be derived from the results of isothermal titration calorimetry (ITC) experiments along with computational methods ^{[4][5][6][16]}.

The dendrimer–protein complex stability is another important factor determining the potential application of dendrimer–protein interactions in nanomedicine. Tentatively, this feature results from high affinity of dendrimers to proteins and high values of association constants arising from multipoint contacts. Nevertheless, the mechanism of interaction and driving force are also of particular importance for complex stability. Thus, the hydrophobic interactions have been shown to play a crucial role in the resistance of complexes toward disruptive action of competitive compounds and aid in the complex stability under physiological conditions (0.14 M NaCl). In ^[20], the cooperation of electrostatic and hydrophobic interaction has been shown to improve the stability of polyion complexes. The presence of aromatic groups, together with the carboxylate ones at the terminal functionalities of the dendrimer, allowed a more specific and favorable binding of dendrimers with NUPR protein ^[4]. The hydrophobic interactions are responsible for the stability of the complexes toward the dissociative action of low-molecular-weight electrolytes such as sodium chloride. While electrostatic interactions are known to become weaker under high ionic strength, the hydrophobic forces ensure the complex stability under a wide range of ionic strength ^[6]. The hydrophobic forces contribute to the insusceptibility of the dendrimer–protein complexes to competitive interactions with charged polyelectrolytes.

Binding of dendrimers may cause changes in the protein secondary structure, which could be evident from circular dichroism (CD) measurements. In most cases, a decrease in α-helix content accompanied by a slight increase in β-sheets and random coil structure is observed ^{[5][7][8][11][16]}. The dendrimers may also induce the alteration of tertiary and quaternary structures of well-folded proteins ^[8] or partial protein destabilization ^[16].

The influence of dendrimers on the secondary structure of proteins strongly depends on the dendrimer/protein molar ratio. The electrostatically driven binding of a hybrid carbosilane–viologen–phosphorous dendrimer with human serum albumin (HSA) did not induce conformational changes in protein molecules up to 2-fold molar excess of dendrimers, while the addition 5- and 10-fold molar excess significantly altered the shape of albumin CD spectrum ^[11]. The same trend was observed in other studies, indicating the necessity of large dendrimer excess to induce the changes in protein secondary structure ^[5][6].

Besides the dendrimer/protein ratio, the ability to trigger the protein conformational changes is also governed by the dendrimer structure and nature of terminal groups, thus determining the mechanism and efficacy of interaction. The changes in dendrimer structure vary their impact on protein molecules. For example, PAMAM dendrimers of the third and fifth generations substituted with sugar residues did not alter the secondary structure of albumin, while PEG-modified cationic PAMAM dendrimers showed strong binding to albumin, affecting its secondary structure and conformation ^[21][22]. The effect of PPI dendrimers on insulin also depended on the nature of the end groups of the dendrimer (unmodified or modified with guanidine or urea) ^[23]. However, all dendrimers changed the secondary structure and thermal stability of the protein. Thus, the nature of the terminal groups of the dendrimer plays an important role in their interaction with proteins.

In general, the conformational changes in protein occur in case of tight binding accompanied with high values of association constants, which are usually characteristic of electrostatically driven interactions ^[8]. In ^[11], the authors revealed the pronounced changes in HSA secondary structure upon the interaction with carbosilane–viologen–phosphorous dendrimers bearing 12 inner and 24 outer charges and interacting with high-binding constants even at low dendrimer concentration, while no conformational changes were observed for HSA interaction with the similar dendrimer possessing 24 total positive charges and weaker binding constant at the same dendrimer concentration. The results suggest that alteration in protein structure is a consequence of high strength and affinity of interaction.

Quantitatively, the strength of an interaction is described by the association constant. The association constant can be calculated by different experimental techniques. The most widely used are ITC and the fluorescence quenching method. The fluorescence quenching method is based on the presence of tryptophan residues in protein molecules which are extremely sensitive fluorophores ^[24]. Upon the complex formation, the intrinsic fluorescence is quenched due to interaction with a ligand. The construction of the plot of relative fluorescence intensity on the concentration of the ligand allows one to estimate the efficacy of binding as a Stern–Volmer constant. To exclude the unspecific decrease in fluorescence intensity due to optical properties of the sample, one should consider the possibility of overlapping absorbance spectra of protein and ligand, which should be carefully checked prior to measurements. In ITC experiments, the association constant is determined based on thermodynamic parameters derived from binding isotherm. However, an accurate determination of a binding constant is very comprehensive and requires implementation of a complex binding model since the heat released during the calorimetric titration of a protein by a ligand is a sum of different processes including protein–ligand binding, protein–protein interactions, conformational rearrangements in protein structure, etc.

The affinity of interaction assessed based on binding constant increases with an increase in the number of groups involved in complex formation. In most cases, this correlates with the dendrimer generation number and the affinity of interaction increases with the generation ^{[4][5][6][11]}. Functionalities at the dendrimer periphery also modulate the affinity for proteins and could be a valuable tool for controllable tuning of the complex properties ^[25].

However, the ability of dendrimers to alter the protein conformation is a basis for a valuable inherent property of dendrimers—antiamyloid activity ^[2]. Thus, the dendrimers can interfere with the ordered beta-sheet structure formation. Briefly, the addition of the dendrimers to amyloidogenic, usually intrinsically disordered proteins leads to formation of stable complexes which are not prone to aggregation. Despite the minor unfolding of protein structure, the tight interactions with dendrimer stabilize the protein molecule and retard the amyloid fibril formation. The antiamyloid properties of dendrimers are discussed in more detail below.

Molecular dynamic (MD) simulations are widely used to study protein–dendrimer interactions ^{[6][10][26][27][28][29][30][31]}. This is a valuable technique supporting the experimental results and providing data which cannot be obtained experimentally. MD helps to identify the binding sites, number and location of amino acid residues involved in association; predict the complex structure and conformational changes of protein; and assess the contribution of different molecular interactions in complex formations. MD simulations are usually applied simultaneously with biophysical techniques, providing a complementary view on the complexation details.

For example, computer simulations of HSA interactions with PAMAM dendrimers revealed the participation of protons of secondary amino groups located in the inner sphere of dendrimer together with terminal amino groups ^[32]. This could be explained by the flexibility of dendrimer structural units able to adopt different conformations, which offers possibilities for interaction of inner groups. Computer simulations demonstrated two important impacts of PAMAM dendrimers binding with alpha-chymotrypsinogen A: the interaction enhances the conformational stability of protein and diminishes the protein–protein interactions due to wrapping by dendrimer molecules ^[33].

MD simulations were used to study the influence of dendrimer counterions on the interaction with proteins ^[34][35]. The study utilized PAMAM dendrimers modified with guanidine moieties bearing Cl⁻, SO₄²⁻ and H₂PO₄⁻ groups as counterions and alpha-chymotrypsinogen A. The interaction efficiency was measured by the preferential interaction coefficient G, which is associated with the free energy of protein transfer from water to an aqueous salt solution and can be used to assess the affinity of the interaction of salt ions with the protein surface. Upon 0.18 M concentration, coefficient G equaled 1.0, 2.7 and 2.3 for Cl⁻, SO₄²⁻ and H₂PO₄⁻, correspondingly. This means that Cl⁻ anions promote the dendrimers’ interaction with the protein, while SO₄²⁻ and H₂PO₄⁻ inhibit binding. For Cl-containing dendrimers, the interaction proceeds through hydrogen bonding of guanidine groups of dendrimer with negatively charged amino acids of protein along with cation–π interaction with aromatic amino acids. The simulation data also revealed the possibility of simultaneous cooperative interaction of several guanidine groups with the protein surface. This results in higher strength of contacts in comparison with single functional group binding. However, switching the counterion to either sulfate or phosphate inhibits the appearance of such multiple interactions.

In another work ^[36], MD allows one to explore the effect of prevention of the cytokine response mediated by the TLR4-MD-2-lipopolysaccharide complex due to dendrimer binding with lymphocyte antigen MD2. The authors highlighted the importance of cooperative electrostatic interactions of dendrimer glucosamine moieties with the amino acid residues of the MD-2 protein located near the hydrophobic pocket.