Protein–dendron conjugates for DNA binding: understanding the effect of the protein core on multivalency

Abstract

In this work we have used molecular dynamics simulations to explore the binding between two different protein–dendron conjugates and double stranded DNA. Different protein cores protect the binding site from the external environment, but can also influence the multivalent recognition between the spermine ligands of the dendron and the DNA.

---

Protein–polymer conjugates are an interesting class of biohybrid materials with potential biopharmaceutical applications.^1,2 A wide range of methodologies has been applied to conjugate proteins with different types of polymers which have either linear or branched topology.³ Linear polymers attached onto protein surfaces have previously been shown to enhance or to modify the properties of such conjugates, for example by improving the circular half-life of the conjugate or by enhancing tissue and intracellular targeting.⁴ We are particularly interested in developing branched polymers, dendrimers and dendrons that can be conjugated with proteins in a precise manner.^5,6 In this framework, dendritic macromolecules offer several advantages, for example, a precisely defined branching structure with a high density of binding ligands which can establish multiple interactions leading to enhanced binding⁷—the principle known as multivalency.⁸ However, due to the complexity of such conjugate molecules, what kind of parameters can affect the multivalency in these conjugates and how the protein part can influence the binding properties of the polymer have not previously been characterized in detail.

We have previously studied a series of different Newkome-type dendrons with spermine functionalized surfaces capable of binding DNA with high affinity.⁹ These dendrons were further modified to allow their conjugation on protein surfaces and were shown to act as high-affinity DNA adhesion patches.^5,6 An amphiphilic class I hydrophobin protein (HFBI) functionalized with a second-generation spermine dendron was biocompatible and could promote gene transfection in vitro. Recently, we have developed multivalent dendrons that can be degraded with mild UV light irradiation, which triggers the release of dendron bound biomolecules.¹⁰ Finally, we reported an HFBI and anti-Human Epidermal growth factor Receptor 2 (HER2) single-chain Fragment variable (scFv) antibody functionalized with a photoresponsive Newkome-type dendron as a potential temporary DNA binder with enhanced DNA release properties.¹¹ However, due to the complex nature of multivalent interactions, a clear comprehension of the binding event was challenging to obtain only by experimental methods.

The most important features of these dendrons are their directionality and multivalent charged surface. The rigidity/flexibility of the dendritic scaffold also plays an important role in the multivalent binding with a charged oligonucleotide.¹² Interestingly, we have recently demonstrated that even if UV-degradable dendrons result to be more rigid than the Newkome-type polyamide dendrons (both functionalized with spermine ligands), the presence of the photolabile linker (shown in blue, Fig. 1a) induces a higher uniformity spreading in the interactions between spermines and the DNA.¹³ Structural “rigidity” and “flexibility” are difficult properties to define and are closely related to the molecular architecture and the surrounding environment.¹⁴

Fig. 1 Strategy for the temporary adhesion of proteins to DNA and preparation of protein–dendron conjugates. (a) Optically degradable dendrons used for protein conjugation and DNA binding. (b) Symbols key. (c) Constructed protein–dendron conjugate. Maleimide in the core of the dendron is used to react with a single free thiol on the protein surface. (d) Schematic illustration of the DNA binding process. (e) Sequence used for the molecular DNA model.

Here we use molecular dynamics modeling as a tool to access the binding site and to analyze in detail the multivalent molecular recognition between the protein–dendron conjugates and nucleic acids. In particular we were interested in comparing how the different protein cores, a small and neutral hydrophobin (HFBI) and a large and charged bovine serum albumin (BSA), affect the ability of the dendron to use its spermine ligands actively to bind the phosphate groups of the DNA. It is in fact intuitive that different protein cores (e.g., size, flexibility, net and local charges, etc.) may have diverse impact on the multivalent recognition between the dendrons and nucleic acids. With this aim we bound the first and second generation UV-degradable dendrons (pllG1 and pllG2) to HFBI and BSA. The corresponding HFBI– and BSA–pllG1 and –pllG2 molecular models were constructed and simulated in explicit water. In order to understand how the ions of the solution affect the multivalent binding between the conjugates and DNA (Fig. 2), two salt concentrations of 9.4 mM and 150 mM NaCl were also studied. The dendrons were grafted to the proteins following the same methodology adopted during the experiments by our group (extensive details regarding the grafting procedure can be found in the original experimental papers).^5,6 In the BSA-based conjugates the dendrons were attached to the free Cysteine-34 present in the BSA amino acid chain. On the other hand, the N terminus of HFBI was modified by site-directed mutagenesis to add a free cysteine residue necessary to graft pllG1 and pllG2 to the protein (Fig. 2a). For the DNA model we used a 21 base pair long DNA model containing a mixture of bases (Fig. 1e).^11–13

Fig. 2 Molecular models of (a) HFBI–pllG2 and (b) BSA–pllG2 conjugates in contact with DNA at 150 mM NaCl solution. In the dendron model, the core (COR) is represented in yellow (and also the small peptide in HFBI–pllG2), the repetitive (REP) units in green and the spermine (SPM) residues in red. HFBI, BSA and DNA are represented as pink, purple and black ribbons respectively. Oxygen atoms of water are colored in cyan and Na⁺ and Cl⁻ ions in purple and green respectively.

In order to study the global interactions between the protein–dendron conjugates and DNA, we calculated the free energies of binding following the well known MM/PBSA approach. Table 1 presents the energetics of the systems where ΔG_bind is composed of an enthalpic (ΔH_bind) and an entropic (−TΔS_bind) term and defined as:

ΔG_bind = ΔH_bind − TΔS_bind (1)

Table 1 Free energies of binding (ΔG_bind) of protein–dendron conjugates in complex with DNA. Free energies of binding are composed of an enthalpic (ΔH_bind) and an entropic term (−TΔS_bind)

9.4 mM [NaCl]^a	ΔHbind^b	−TΔSbind^b	ΔGbind^b
HFBI–pllG1	−11.4 ± 2.0	+3.7 ± 0.2	−7.7
HFBI–pllG2	−12.5 ± 1.0	+4.7 ± 0.2	−7.8
BSA–pllG1	−8.6 ± 0.7	+6.4 ± 0.4	−2.2
BSA–pllG2	−11.0 ± 0.5	+3.1 ± 0.2	−7.9

---

150 mM [NaCl]^a	ΔHbind^b	−TΔSbind^b	ΔGbind^b
a Energies are reported for high (150 mM) and low (9.4 mM) salt concentrations and expressed in kcal mol^−1. b The errors represent the standard deviations of the energetic values over the 200 MD snapshots used for the analysis.
HFBI–pllG1	−11.2 ± 1.0	+6.9 ± 0.3	−4.3
HFBI–pllG2	−11.3 ± 0.4	+3.7 ± 0.3	−7.6
BSA–pllG1	−11.0 ± 0.6	+7.6 ± 0.4	−3.4
BSA–pllG2	−10.6 ± 0.3	+2.8 ± 0.2	−7.8

---

Details about the creation of the molecular systems and the computational procedures adopted are available in the ESI.† The simulation procedure adopted in this study certainly presents limitations arising mainly from the use of the MM/PBSA approach for energy estimation (approximate description of thermodynamics) and to the use of a single simulation which could induce potential configuration-dependency in the results. However, due to the highly statistical nature of the dendrons, subject of this study, and their strong symmetry, this approach was considered to be well-representative of what on average happens in solution during the conjugate–DNA binding.

The multivalent binding between charged molecules is controlled by a delicate balance between electrostatic attraction between the spermine's and the DNA's opposite charges. Usually at 150 mM NaCl such systems express higher affinity for DNA (more negative ΔG_bind) for higher generation dendrons than for the smaller ones. This is due to the disturbing action of the ions in solution. However, the pllG2 dendron has enough spermine ligands that it is able to use some of the free ligands to “protect” the central part of the binding site, ensuring stable binding despite the destabilizing action of ions in solution. On the other hand, pllG1 which has only three spermine ligands needs all of them for the binding and is not able use free ligands to protect the binding site.^12,13 The same behavior is also observed with the protein–dendron conjugates (Table 1).

The binding data for the conjugates at 9.4 mM NaCl show interesting differences when compared to our previous results on free dendrons without a protein conjugate. Detailed DNA binding results for the same UV-degradable dendrons without the protein conjugate are available in our previously published work.¹³ Usually at low salt concentration both first and second generation dendrons are able to bind DNA with a high affinity because all of the spermine ligands are attracted only by the phosphate groups in the DNA double strand. Indeed, in such conditions very few disturbing ions are present in solution—that is why the first generation dendrons are able to reach ΔG_bind values in the same range as the second generation ones.^11–13 This is true also for HFBI–dendron conjugates, but not for the BSA-based ones. The binding data at 150 mM NaCl can be perfectly explained as electrostatic interferences generated by salt ions in solution. On the other hand, the dramatic loss in ΔG_bind reported for BSA–pllG1 at 9.4 mM NaCl (ΔG_bind falls to −2.2 kcal mol⁻¹ with respect to the −7.9 kcal mol⁻¹ value of the BSA–pllG2 second generation conjugate) appears to have a different origin.

Fig. 2 shows the notable difference in size between HFBI and BSA proteins. The multivalent recognition between dendrons and DNA double strands is strongly influenced by the presence of salt, however other larger molecules can also interfere with the binding event. This is the case for the BSA–dendron conjugates. BSA has an overall negative charge of −15 and its dimensions are larger (∼66 kDa) than those of the smaller (∼8 kDa) and globally neutral HFBI (Fig. 2). Therefore it is easy to argue that BSA can disturb the multivalent recognition between dendrons and DNA owing to multiple intramolecular interactions with pllG1 and pllG2, playing an important role in the DNA binding event.

Fig. 3 shows the binding between DNA and HFBI–pllG1 (Fig. 3a) and BSA–pllG1 (Fig. 3b). Consistent with our previous studies on dendrons and their protein conjugates, the affinity between DNA and HFBI-based conjugates is not dependent on the generation of the dendron at 9.4 mM salt concentration. In fact, ΔG_bind for HFBI–pllG1 and HFBI–pllG2 is −7.7 and −7.8 kcal mol⁻¹ respectively (Table 1). These results demonstrate that the smaller hydrophobin core does not interfere with the attraction between the spermines and DNA.

Fig. 3 Binding between (a) HFBI–pllG1 or (b) BSA–pllG1 and DNA at 9.4 mM NaCl. CEN, REP and SPM residues of the dendrons are colored in pink, green and red respectively. HFBI, BSA and DNA are represented as purple, cyan and dark ribbons. Water and ions are omitted for clarity. (c) Detail of the binding site: intramolecular interactions between SPM 3 and the Thr 48 of the BSA protein—formation of a 3.01 Å long hydrogen bond.

Fig. 3c shows that the low ΔG_bind of the BSA–pllG1 conjugate (ΔG_bind = −2.2 kcal mol⁻¹) is a result of eventual intramolecular interactions between the charged spermines and the BSA protein. BSA has many local interaction spots which potentially can interact with the dendron disturbing the binding with DNA. Fig. 3c shows the presence of possible active interactions between the N atoms of the spermines and certain protein amino acids. For example, SPM 3 creates a H-bond 3.01 Å long with the OH group of the Thr 48. This is a single example of the multiple dynamic intramolecular interferences (interactions) existing between BSA and the multivalent surface of the dendrons. In the case of Fig. 3, one of the three spermine ligands is completely detracted from the binding with DNA (Table 1: low binding enthalpy despite of a high entropic penalty, that is due to the loss of degrees of freedom induced by the H-bond). The disturbing effect of BSA on the dendron–DNA binding is thus comparable to that generated by the presence of a high salt concentration in solution—salt ions are not the unique factor which can influence the multivalent binding with DNA.

The same phenomenon is present also for the second generation BSA–pllG2 conjugate, but the interaction site between the dendron and the protein is different. Fig. 4 illustrates the interaction site between BSA and spermine ligands in BSA–pllG2. Here (Fig. 4b) the glutamic acid 78 interacts with SPM 1 of pllG2 and generates a H-bond with a length of 2.86 Å.

Fig. 4 (a) Binding between BSA–pllG2 and DNA at 9.4 mM NaCl. CEN, REP and SPM residues of the dendrons are colored in pink, yellow and red respectively. Water and ions are omitted for clarity. (b) Detail of the binding site: intramolecular interactions between SPM 1 and the Glu 78 of the BSA protein—formation of a H-bond (length: 2.86 Å).

In this case, however, the binding affinity of BSA–pllG2 does not suffer from the intramolecular interactions with BSA. Data in Table 1 show that BSA–pllG2 can fully compensate this interference and the ΔG_bind remains high at both salt concentrations (−7.8 and −7.9 kcal mol⁻¹ at 150 and 9.4 mM NaCl respectively). Given the analogy existing between the interference created by the presence of a large protein and the one generated by salt in solution, both these effects can be counted as a “response to the external environment”. We already demonstrated how these kinds of dendrons are able (at second generation and above) to sacrifice some of the external SPM ligands to protect the binding site from salt ions.¹³ The same happens in the case of intramolecular interactions with the protein core—BSA–pllG2 has enough SPM ligands that, even if some of them are subtracted from the DNA binding because of the attractive interactions with the protein amino acids, it is still able to generate a strong and stable binding with DNA.

In order to quantify the intramolecular interactions between the dendrons and their protein-conjugate, we calculated the average interaction energy (ΔE_intra) between each SPM ligand and HFBI and BSA protein cores.

The intramolecular energies (ΔE_intra) reported in Table 2 are related to the interactions between proteins and the SPM ligands only. They are calculated as the sum of gas-phase non-bond interactions and corrected for solvation.^12–14 Moreover, ΔE_intra are normalized per SPM unit—they measure the average attraction efficiency lost by each spermine due to intramolecular interactions with the protein core. Data show that the interactions between HFBI and the spermines are more than one order of magnitude lower than the ones generated by BSA. In fact both pllG1 and pllG2 dendrons do not feel the presence of such a small protein core during their multivalent binding with DNA. Table 2 shows also that, in general, the interactions between the first generation dendrons and the protein core are stronger at 9.4 than at 150 mM NaCl. This can be explained by the presence of ions that, at high salt concentration screen the spermine ligands not only from the interactions with DNA, but also with the protein core. If the dendron is not disturbed by a higher salt concentration, it is more strongly disturbed by the presence of a large and charged protein core. Finally, even if the interactions of the second generation conjugates with BSA are stronger than in the case of HFBI, the ΔG_bind values reported in Table 1 are substantially invariant identifying a “modulation” between protein–dendron intramolecular interactions and disturbing action by solution ions. Interestingly, the effect of the ionic strength and the intramolecular-protein interactions can be both considered as “external factors” influencing the multivalent recognition between the dendron and the DNA during the binding.

Table 2 Interaction energies (ΔE_intra) between SPMs of dendrons and HFBI and BSA proteins

Hybrid	9.4 mM [NaCl]^a	150 mM [NaCl]^a
a Interaction energies are reported for low (9.4 mM) and high (150 mM) salt concentrations and are expressed in kcal mol^−1. Values are normalized per spermine unit. The errors represent the standard deviations of the energetic values over the 200 MD snapshots used for the analysis.
HFBI–pllG1	−0.2 ± 0.0	−0.1 ± 0.0
HFBI–pllG2	−0.2 ± 0.0	−0.1 ± 0.0
BSA–pllG1	−3.2 ± 0.2	−1.4 ± 0.1
BSA–pllG2	−1.1 ± 0.1	−1.4 ± 0.1

---

In conclusion, we have used molecular dynamics modeling to explore the multivalent recognition between HFBI and BSA–UV-degradable dendron conjugates and DNA. Analysis of the modeling data highlighted the important role played by the different protein cores in the binding event. The dendrons do not feel the presence of the smaller HFBI core during the binding to DNA. On the contrary, BSA creates important interactions with some of the spermine ligands of the dendrons. While the second generation conjugate has enough SPMs to compensate these interferences, the affinity of pllG1 for DNA is dramatically compromised. This study presents a new point of view into the factors which control the multivalent recognition between molecules and in particular the behavior of protein–dendron conjugates. These data offer clear indication about how structural modifications can generate intramolecular interactions which are also capable of influencing strongly the multivalency of this type of binding agent. Understanding such delicate binding interactions is a goal of considerable importance for the preparation of effective protein–polymer conjugates for biomedical applications.

References

1. G. G. Kochendoerfer, S.-Y. Chen, F. Mao, S. Cressman, S. Traviglia, H. Y. Shao, C. L. Hunter, D. W. Low, E. N. Cagle, M. Carnivali, V. Gueriguian, P. J. Keogh, H. Porter, S. M. Stratton, M. C. Wiedeke, J. Wilken, J. Tang, J. J. Levy, L. P. Miranda, M. M. Crnogorac, S. Kalbag, P. Botti, J. Shindler-Horvat, L. Savatski, J. W. Adamson, A. Kung, S. B. H. Kent and J. A. Bradburne, Science, 2003, 299, 884–887.
2. Z. L. Ding, R. B. Fong, C. J. Long, P. S. Stayton and A. S. Hoffman, Nature, 2001, 411, 59–62.
3. (a) K. L. Heredia and H. D. Maynard, Org. Biomol. Chem., 2007, 5, 45–53; (b) B. S. Lele, H. Murata, K. Matyjaszewski and A. J. Russell, Biomacromolecules, 2005, 6, 3380–3387.
4. (a) G. G. Kochendoerfer, Curr. Opin. Chem. Biol., 2005, 9, 555–560; (b) F. M. Veronese, C. Monfardini, P. Caliceti, O. Schiavon, M. D. Scrawen and D. Beer, J. Controlled Release, 1996, 40, 199–209; (c) R. Weissleder, K. Kelly, E. Y. Sun, T. Shtatland and L. Josephson, Nat. Biotechnol., 2005, 23, 1418–1423.
5. M. A. Kostiainen, G. R. Szilvay, J. Lehtinen, D. K. Smith, M. B. Linder, A. Urtti and O. Ikkala, ACS Nano, 2007, 1, 103–113.
6. M. A. Kostiainen, G. R. Szilvay, D. K. Smith, M. B. Linder and O. Ikkala, Angew. Chem., Int. Ed., 2006, 45, 3538–3542.
7. D. K. Smith, A. R. Hirst, C. S. Love, J. G. Hardy, S. V. Brignell and B. Huang, Prog. Polym. Sci., 2005, 30, 220–293.
8. (a) M. Mammen, S.-K. Choi and G. M. Whitesides, Angew. Chem., Int. Ed., 1998, 37, 2754–2794; (b) A. Mulder, J. Huskens and D. N. Reinhoudt, Org. Biomol. Chem., 2004, 2, 3409–3424.
9. M. A. Kostiainen, J. G. Hardy and D. K. Smith, Angew. Chem., Int. Ed., 2005, 44, 2556–2559.
10. (a) G. Doni, M. A. Kostiainen, A. Danani and G. M. Pavan, Nano Lett., 2011, 11, 723–728; (b) M. A. Kostiainen, O. Kasyutich, J. J. L. M. Cornelissen and R. J. M. Nolte, Nat. Chem., 2010, 2, 394–399; (c) M. A. Kostiainen, D. K. Smith and O. Ikkala, Angew. Chem., Int. Ed., 2007, 46, 7600–7604.
11. M. A. Kostiainen, J. Kotimaa, M.-L. Laukkanen and G. M. Pavan, Chem.–Eur. J., 2010, 16, 6912–6918.
12. (a) G. M. Pavan, A. Danani, S. Pricl and D. K. Smith, J. Am. Chem. Soc., 2009, 131, 9686–9694; (b) G. M. Pavan and A. Danani, Phys. Chem. Chem. Phys., 2010, 12, 13914–13917.
13. G. M. Pavan, M. A. Kostiainen and A. Danani, J. Phys. Chem. B, 2010, 114, 5686–5693.
14. (a) G. M. Pavan, L. Albertazzi and A. Danani, J. Phys. Chem. B, 2010, 114, 2667–2675; (b) L. B. Jensen, K. Mortensen, G. M. Pavan, M. R. Kasimova, D. K. Jensen, V. Gadzhyeva, H. M. Nielsen and C. Foged, Biomacromolecules, 2010, 11, 3571–3577; (c) M. Shema-Mizrachi, G. M. Pavan, E. Levin, A. Danani and N. G. Lemchoff, J. Am. Chem. Soc., 2011, 133, 14359–14367.

---

Footnote

† Electronic supplementary information (ESI) available: Details of the construction of the molecular models, complete computational procedure for simulations and for energetic analysis. See DOI: 10.1039/c1ra00472g

---