Understanding the Interaction of Polyelectrolyte Architectures with Proteins and Biosystems

Abstract The counterions neutralizing the charges on polyelectrolytes such as DNA or heparin may dissociate in water and greatly influence the interaction of such polyelectrolytes with biomolecules, particularly proteins. In this Review we give an overview of studies on the interaction of proteins with polyelectrolytes and how this knowledge can be used for medical applications. Counterion release was identified as the main driving force for the binding of proteins to polyelectrolytes: Patches of positive charge become multivalent counterions of the polyelectrolyte and lead to the release of counterions from the polyelectrolyte and a concomitant increase in entropy. This is shown from investigations on the interaction of proteins with natural and synthetic polyelectrolytes. Special emphasis is paid to sulfated dendritic polyglycerols (dPGS). The Review demonstrates that we are moving to a better understanding of charge–charge interactions in systems of biological relevance. Research along these lines will aid and promote the design of synthetic polyelectrolytes for medical applications.


Introduction
Polyelectrolytes consist of long linear or branched macromolecules that contain charged units.W hen dispersed in water or as ufficiently polar solvent, the counterions balancing the charge of the polyelectrolyte will partially dissociate. Therefore,t he properties of polyelectrolytes in solution will be greatly determined by their counterions.P olyelectrolytes are ubiquitous in biological systems and play acentral role in almost all biochemical processes. [1] DNAisperhaps the beststudied natural polyelectrolyte and in the preceding 50 years most work has been directed towards ad etailed understanding of the interaction between DNAa nd proteins related to DNAr epair proteins or transcription factors. [2][3][4][5][6][7][8][9][10][11] Thet hermodynamics of the binding of DNAo rRNA to proteins has been shown to be dominated by charge-charge interactions, and the biological activity of natural polyelectrolytes such as DNAi si ntimately related to their highly charged molecular structures. [2,[12][13][14][15] Heparin provides another example of anatural polyelectrolyte with four charges per repeating unit that has been studied intensively during the last 30 years. [16][17][18][19] Synthetic polyelectrolytes,o nt he other hand, have become valuable tools for various medical purposes during the past 20 years.T hus,c omplexes of synthetic polyelectrolytes with DNAa re now used as nonviral vectors for gene delivery. [20,21] Research along this line has been aimed at welldefined complexes with optimized efficiency.M ore recently, block copolymers containing cationic sequences have been used for this purpose,and transfection using polycations is an active field of polymer science these days. [22][23][24][25] Other polymer architectures used so far include nanogels, [26][27][28][29] which consist of crosslinked polyelectrolytes. [30,31] These systems have also become another highly useful device for gene delivery [21,25,26,32,33] as well as for the defined uptake and delivery of proteins and drugs in general. [23,25] An equally fascinating and rather recent development is the use of polyelectrolytes as drugs themselves. [34] Here, sulfated dendritic polyglycerol sulfate (dPGS), which consists of ad endritic poly(glycerol) scaffold with each sulfated end group bearing an egative charge,h as become afocus of our research. [35] First designed as areplacement for heparin, [34] dPGS has been used for av ariety of biomedical purposes that range from tumor targeting to anti-inflammatory treatment. [34,36] Previous studies suggest that the interaction of dPGS with various proteins and cell-surface molecules proceeds in as pecific way.T hus,D ernedde et al. [36] surmised that dPGS can block the cell adhesion molecules (CAMs) L-and P-selectin on leukocytes and activated endothelial cells,r espectively,w hich are central to The counterions neutralizing the charges on polyelectrolytes such as DNAo rh eparin may dissociate in water and greatly influence the interaction of such polyelectrolytes with biomolecules,p articularly proteins.I nt his Review we give an overview of studies on the interaction of proteins with polyelectrolytes and howthis knowledge can be used for medical applications.Counterion release was identified as the main driving force for the binding of proteins to polyelectrolytes: Patches of positive charge become multivalent counterions of the polyelectrolyte and lead to the release of counterions from the polyelectrolyte and aconcomitant increase in entropy. This is shown from investigations on the interaction of proteins with natural and synthetic polyelectrolytes.S pecial emphasis is paid to sulfated dendritic polyglycerols (dPGS). The Review demonstrates that we are moving to abetter understanding of charge-charge interactions in systems of biological relevance.R esearcha long these lines will aid and promote the design of synthetic polyelectrolytes for medical applications.
inflammatory processes,t hrough as elective charge-charge interaction. Hence,d PGS seems to act as am acromolecular inhibitor that may mimic naturally occurring ligands. Prompted by the success of dPGS as an anti-inflammatory compound, anumber of structures have been synthesized that contain dPGS as ab uilding block for various biological processes in which inflammation plays ac entral role:N anogels based on dPGS with different degrees of flexibility have been shown to possess antiviral properties. [37] dPGS has also been used as abuilding block for micellar structures that can be used for targeting tumor cells. [38] Furthermore,i ts interaction with neural microglia has been the subject of several studies. [39,40] Substituted polyglycerols bearing positive charges have been introduced as potent antibleeding agents with excellent anticoagulant reversal activity upon binding the polyanion heparin. [41][42][43] Summing up all the research done to date,itisfair to state that al arge number of charged polymeric systems and potential drugs have been synthesized recently and the possible medical applications of these systems are hard to overlook. [23,25] However,o nly as mall subset of polyelectrolyte systems has reached the stage of clinical trials.T he problems at hand are:S uch polymeric drugs must remain active in the complex environment of cells or am ultitude of proteins in the blood stream. Ideally,the drug should interact only with ac hosen target structure in ah ighly specific manner.Unspecific interaction with blood proteins should be avoided. At this moment, we clearly lack ag eneral understanding of these systems,w hich would allow us to design them in as traightforward manner to circumvent these problems.
To make progress in this field we need aq uantitative understanding of the interaction of polyelectrolytes with proteins in general. In this Review we discuss recent work along these lines and how the analysis and the modeling of the interaction of polyelectrolytes with proteins can be used for ar ational design of charged polymeric drugs.T he central hypothesis of the present discussion is that this interaction is largely dominated by counterion release.F igure 1s hows this process in aschematic fashion:Weconsider the interaction of aprotein carrying surface charges with ahighly charged linear polyelectrolyte.Afraction of the counterions around the polyelectrolyte is "condensed", that is,c losely bound to the macroion. [1] Proteins,i ng eneral, are polyampholytes,w hich carry patches of negative and positive charge on their surface. Most proteins bear an overall negative charge under physiological conditions.H owever,t he patches bearing ap ositive charge remain and can interact with negatively charged polyelectrolytes such as DNAo rh eparin. In this way,t he proteins become multivalent counterions of the polyelectrolyte,t hereby releasing ac oncomitant number of its monovalent counterions.T he gain in entropy thus achieved is the main driving force. [2] Detailed considerations to be discussed further below demonstrate that this counterion release force is operative even under ap hysiological salt concentration of 150 mm.
Inspired by earlier work on the interaction of DNAwith various proteins, [2,10,13,14,[46][47][48][49] we recently reconsidered the problem of counterion release by as eries of thermodynamic studies related to the interaction of polyelectrolytes with proteins.F irst, the interaction of human serum albumin (HAS) with short-chain poly(acrylic acid) in aqueous solution was studied by ac ombination of calorimetry and molecular dynamics simulations. [50] We also studied the binding of dPGS of different generations [51] to lysozyme [45,52] and to HSA. [53] Recently,t his work has been continued to include aq uantitative discussion of the role of water in the binding process. [54] In addition to this,wedemonstrated that MD simulations can reproduce the experimental binding constant of L-selectin to second-generation dPGS with surprising accuracy and be rationalized in terms of counterion release. [45] Thelatter result could hence furnish aq uantitative proof of earlier conjectures [36] on the use of dPGS as an anti-inflammatory drug. In this way we have acquired arather advanced understanding of the interaction of dPGS with various systems of medical relevance.
Here we survey this work and how it can be applied for ab etter understanding and design of polyelectrolytes for medical purposes.Special emphasis will be laid on biomedical applications of dPGS and related systems.I ti so rganized in terms of the matrix of chemical systems and biochemical problems with increasing complexity shown in Figure 2. Hence,t he Review is subdivided in three parts as follows: 1. In the next section, we shall discuss the current understanding of the interaction of proteins with linear polyelectrolytes. [2,10,11,45,49,52,53] This section will survey the formation of complexes of proteins with DNA, which presents the best-studied case in this field. At ah igher level of biological complexity,p roblems related to drug delivery, [32,60,61] gene transfection, und ultimately gene therapy will be discussed. Glycosaminoglycans (GAGs) such as heparin or heparan sulfate,w hich consist of disaccharide units that may be sulfated, present another important class of natural, highly charged polyelectrolytes. [18,19,62] GAGs are important components of the extracellular matrix of cells.I th as been recognized for along time that the interaction with proteins is driven by electrostatic forces and counterion release. [63][64][65] This fact is underscored by more recent studies [66][67][68] and will be discussed as well. 2. Then ext level of complexity is given by dendritic and hyperbranched polyelectrolytes,c harged networks,a nd polyelectrolyte brushes.Here,collective effects caused by the polymer architecture and their consequences for the interaction with proteins will be discussed. In particular, the counterion dPGS belongs in this section, which also contains the consequences for virus binding [69] and inactivation as well as for the diagnostics and therapeutic use of dPGS in anti-inflammation. [36,70] This section will also highlight the interaction of proteins with polyelectrolytes of higher complexity,s uch as charged networks. [21,27,31,32,37,56,71] This work can be rationalized with recent theoretical studies on these systems. [72][73][74][75][76] The interactions of charged polyglycerols with cellular systems will also be discussed at this point. [36,[77][78][79][80][81] 3. Finally,asection devoted to complex polyelectrolyte architectures will highlight polyelectrolyte systems with higher complexity,s uch as micelles and designed polymeric structures that act as anticoagulants. [25,38] These systems are far more difficult to understand in aquantitative fashion. However, there has been some progress towards medical applications recently,w hich will be discussed here.T hus,m icelles coated by al ayer of dPGS have been used as adrug in tumor targeting. [38] Moreover, there have been recent successful developments of cationic polyelectrolyte drugs with heparin-reversal activity in blood. [41,42] Thee ntire discussion will highlight the general importance of electrostatic factors for the self-assembly and biological activity of charged polymeric systems.A ss hown in Section 2, the driving forces are now rather well understood. Hence,this knowledge can now be used for the rational design and modeling of more complicated systems,a s discussed in Sections 3a nd 4. Steps in this direction will be discussed for all systems considered in Figure 2.

Theory
Up to now,the interaction of linear polyelectrolytes with proteins has been considered for two classes of problems, namely,the 1) interaction of natural polyelectrolytes (mainly DNA) with various proteins,a nd 2) interaction of synthetic polyelectrolytes with proteins.T here is great number of studies devoted to the latter systems,s ince the classical investigations of Bungenberg de Jong in the 1930s (cf.t he review of this work in Refs. [82,83]). However,inmany cases, mixing of synthetic polyelectrolytes with various proteins leads to phase separation (complex coacervates), which constitutes ap roblem of its own, [82,[84][85][86][87] and which may lead to ac omplex phase behavior. [88,89] DNA, on the other hand, forms well-defined 1:1c omplexes with proteins,s uch as polymerases, [7,[90][91][92][93][94] that can be understood in terms of achemical equilibrium. This fact was recognized along time ago [2,12] and counterion release has been singled out as the main driving force for binding.T he basic argument can be understood as follows:A sd epicted in Figure 1, there is ac ertain fraction of counterions that are condensed to the linear polyelectrolyte.T he fraction of condensed counterions can be estimated from ar elationship described by Manning: [12,95] If b is the distance between two charges along the linear polyelectrolyte,acharge parameter x can be defined through Equation (1).
Here, l B is the Bjerrum length (l B ¼ k B T 4pee 0 ; e:d ielectric constant of the medium, k B :B oltzmann constant, e 0 :permittivity of the vacuum, T:a bsolute temperature). If x > 1, afraction 1À1/x of the counterions will be condensed onto the linear chain, that is,strongly correlated with the polyelectrolyte.I ti si mportant to note that this fraction does not contribute to the osmotic pressure of the system. ForD NA, this fraction amounts to about 70 %o fa ll counterions.T he condensed counterions can be regarded as ap hase that may be characterized by a" surface concentration" c ci ,w hich for DNAi softhe order of 1m. [12] If we consider the interaction of such ah ighly charged polyelectrolyte with ap rotein, these condensed counterions must be treated as areaction partner and, thus,contribute to the stoichiometry of the reaction. [2,12] Hence,t he complexation of ap rotein P with an anionic polyelectrolyte PE to acomplex PEP is defined by Equation (2). [2] Here, Dn ci denotes the number of cations of type M + that have been released during the course of the binding reaction. Them easured equilibrium constant K b can be formulated in Figure 1. Interaction of proteins with highly charged polyelectrolytes, for example, DNA, by counterion release:Proteins bear negative (red) and positive charges (blue) on their surface. Above the isoelectric point, the overall surface charge is negative, but the positive patches remain. The polyelectrolyte bears alarge number of charges that will lead to counterion condensation, that is, acertain fraction of the counterions are highly correlated with the polyelectrolyte, as shown here. Upon binding of the protein to the polyelectrolyte, ap ositive patch on the surface of the protein becomes atrivalent counterion of the polyelectrolyte. Thus, three counterions condensed on the polyelectrolyte are released upon binding. The free energy of binding will, therefore, be dominatedb ythe entropic gain through the release of the counterions. [12,44,45] Fort he sake of clarity,only the condensed counterions are shown here. However, all the charges on the protein and the polyelectrolyte are balanced by an equal number of counterions.

Angewandte Chemie
Reviews 3886 www.angewandte.org terms of molar concentrations,a nd its relationship to the thermodynamic constant K T related to the activities of the components is given by Equation (3). [2] Here, g P , g PE , g PEP ,and g M denote the activity coefficients of the protein, the polyelectrolyte,the complex, and the free ions,r espectively (see also the discussion of this problem in Ref. [54]). Since the concentration [M + ] of monovalent cations is much larger than the concentrations of the polyelectrolyte and the protein, [M + ] equals,t oan excellent approximation, the concentration of added salt c s . First, the activity coefficient of the ions can be disregarded, since we deal mostly with small concentrations of the ions.Moreover,it can be shown that the second term on the right-hand side of Equation (3) related to the activity coefficients give as mall but non-negligible contribution for linear polyelectrolytes that scales with ln(c s ). [44,96] This term contains the Debye-Hückel interactions of the various parts of the complex. For complexes of proteins with spherical polyelectrolytes,a ll contributions from activity coefficients may be shown to be small and negligible to af irst approximation. [54] Hence,t o ag ood approximation, Equation (3) can be simplified to ðÞ À Dn ci lnc s ð3aÞ Here, K b (1m)i sthe binding constant extrapolated to one molar salt concentration. Thus,the stoichiometric coefficient Dn ci is given to agood approximation by ÀdlnK b =dlnc s ,that is, by the negative slope of the plots of the log of the measured equilibrium constant K b against log c s . Only at very low ion concentrations of the order of 1mm and less will the data deviate from linearity because of an on-negligible repulsive Debye-Hückel interaction. [54,97] Many years ago,T anford argued that Equation (3) needs to be supplemented by at erm that takes into account the number Dw of released or bound water molecules during the course of complex formation. [98] Thus,aterm scaling such as À(n i /n w )Dw should be included in Equation (3). Here, n i and n w denote the molar number of ions and of water molecules in the system, respectively.However, n i is typically of the order of 10 À2 to 10 À1 ,w hereas n w is 55.6. Hence,t his term, which reflects the change of hydration during complex formation, is small and can be dismissed for low ion concentrations n i . [2,98] This term comes into play for high ion concentrations in excess of 1m. [99][100][101] In this case,plots of log K b versus log c s are no longer linear. This problem has been studied in aseries of Figure 2. Interaction of polyelectrolytes with biosystemsa tdifferent levels of complexity:Linear polyelectrolytes may be assembled into networks [55,56] and branched systems. Ultimately,they may become building blocks for systems with higher complexity,f or example, micelles with core-shells tructures. Complexity on the biological side starts with single protein molecules that can interact with polyelectrolyte systemsw ith various architectures. On this level, the therapeuticactivity of polyelectrolytes can often be traced back to ab locking of proteins by asuitable polyelectrolyte system. [31,36,39,40,42,45] Cells present the next level of complexity and their interaction with charged polymeric systems must be understood when considering these systems for,for example, drug delivery or gene transfection. [20,22,57,58,59] Organs present the highest level of complexity and the understanding of their interaction with synthetic polyelectrolyte systemsi sinits infancy.However,cationic polyelectrolytes with suitable architecturesh ave recently been introduced as agents with anticoagulant reversal activity in blood. [41][42][43] The entire matrix of systems and problems gives agood overview of the possible medical problemst owhich synthetic polyelectrolytes may provide solutions.
important investigations by Bergqvist, Ladbury,a nd coworkers. [99,100,102,103] Here,p lots according to Equation (3) indeed exhibit amarked curvature,which can be explained in terms of am odel taking into account the release of water molecules during binding. [98,104,105] Amuch-refined discussion of the release of water was presented by Record and coworkers, [106,107] who demonstrated that Dw is intimately related to the preferential adsorption of the ions on the surface of the biomolecule (cf.R ef. [106] and further references therein). Themodel of vander Meulen et al. [107] predicts that Dw vanishes if there is no preferential adsorption of the co-or counterions.T he analysis of experimental data on the binding of proteins to DNAl ed vander Meulen et al. to the conclusion that Dw is small if salts in the middle of the Hofmeister series, [108] for example,N aCl or KCl, are used. Hence, Dw will be small, and nearly all complexes of DNAw ith proteins have been modeled by Equation (3a). [2,10,13,14,[46][47][48][49] Equation (3a) can be used to analyze the measured binding constant K b further by splitting it up through extrapolation to a1m salt concentration. Thus, K b now consists of areference part K b (1m)a nd at erm depending solely on the release of counterions. [10,11] Equation (4) gives Equation (5), where DG res is the residual of the Gibbs free energy of binding derived from K b (1m), whereas DG ci denotes the part related to counterion release. [11] Extrapolation of the measured K b value to a1 m salt concentration according to Equation (5) leads to DG res ,that is, DG res = ÀRTln(K b (1m)), and in turn to DG ci . Here,t he quantity DG res denotes all contributions to the free energy of binding which are not from counterion release,s uch as direct electrostatic interaction, [54] hydrogen bonding,o rs alt bridges as well as other effects.I nt his way,t he salt concentration of 1m constitutes areference state.
From the above approximations,c ounterion release is afully entropic effect and we obtain Equation (6). [54] DG ci ffiÀTDS ci ffi RTDn ci ln c ci c s ð6Þ Here, DS ci denotes change in entropy of the counterions, which can be calculated from the surface concentration c ci introduced above. [45,52,[109][110][111] Thequantity c ci can be estimated for linear polyelectrolytes from x as prescribed by Manning [12] or it can be deduced from molecular dynamics simulations,as shown recently. [45] Moreover,w ith the total binding entropy DS b being known, the residual part DS res can be calculated with Equation (7). [11,54] It is evident from Equations (1)-(7) that acomprehensive thermodynamic analysis of the binding of proteins to polyelectrolytes can be achieved.
We now turn to an important tool that has been pivotal for thermodynamic analysis:I nt he last two decades isothermal titration calorimetry (ITC) has become the central tool for the analysis of complex formation in natural and synthetic systems. [112,113] ITC measures directly the heat evolved upon complex formation with high precision. Thus,b yu sing ITC, the K b value of DNAwith ag reat variety of proteins can be determined with high accuracy; [112,114,115] al arge number of such studies have now been carried out. [13,47,90,91,[116][117][118][119][120][121][122][123] It is fair to state that most of the quantitative knowledge on the interaction of polyelectrolytes with proteins has been acquired by ITC experiments and this method holds great promise for further understanding of these systems,i n particular when applied to the design of pharmaceutical systems. [112,124]

Enthalpy-Entropy Compensation
Investigations by ITC and application of Equation (4) have led to ag reat amount of precise thermodynamic data. Here,studies of the dependence of the binding constant K b on temperature revealed as trong enthalpy-entropy compensation, that is,m ost of the measured binding enthalpy is balanced by ac oncomitant entropic contribution. This enthalpy-entropy compensation (EEC) [13,49,[125][126][127][128] has been ac ontroversial subject for quite some time. [129][130][131][132] However, Grunwald and Steel [133] pointed out many years ago that the EEC is the natural consequence of the rearrangement of solvent molecules around as olute.T his idea was reviewed carefully more recently by Liu and Guo. [134] Moreover,L i et al. showed that the EEC is ar eal effect with as ound experimental basis. [135] This is in full agreement with recent experimental studies of the Whitesides group that explain the EEC by the reformation of the water network around the complex. [132] Jen-Jacobson and co-workers showed that EEC exists in systems of biological relevance. [13,14,49,136] Synthetic systems have been studied with equal intensity, [134] and from the vast amount of literature we only cite the very recent investigation by Schçnbeck and Holm [137] on the EEC for complexes of cyclodextrin with various host molecules. Dragan et al. have recently suggested that EEC may be related to the release or uptake of water. [11] In Section 3w e will discuss our recent studies on the EEC for the interaction of dPGS with various proteins,w hich come to the same conclusions. [52][53][54] Summing up this survey of theoretical and experimental work that now extends over 50 years, [125] it is clear that EEC is au biquitous phenomenon that has been firmly established by agreat number of experimental studies.

Interaction of DNA with Proteins
In this section, the above conclusion will be compared with experimental findings.H ere we start with as urvey on studies carried out on natural systems.F or al ong time [2] the interaction of DNAwith various proteins has been analyzed in terms of Equation (3). Theapplication of Equation (3a) to the formation of protein/DNAc omplexes has been analyzed for awide variety of systems by Privalov,Dragan, and Crane-Robinson. [10] In all cases,s traight lines were obtained indeed by application of Equation (3a). Moreover,t he number of released ions Dn ci is found to be strictly correlated to the number of ionic contacts seen between DNAand the protein in crystal structures (cf.the discussion in Ref. [10]). It should be noted that ion-specific effects may change Dn ci slightly and should be considered carefully. [10] Furthermore,the binding of DNAt op roteins may lead to changes of the secondary structure and the partial refolding of proteins.T his point has been discussed in detail by Privalov et al. [10,138] and by Jen-Jacobson et al. [13,49] Dragan et al. [11] have used Equation (5) to split the measured DG b into the part corresponding to counterion release and ar esidual part. As imilar analysis has been applied to the binding of DNAtoproteins by other research groups as well. In particular,Dragan et al. could demonstrate that the EEC is an entirely non-electrostatic phenomenon: Plotting the measured binding enthalpy DH b against the residual entropy DS res resulted in aperfect master curve [11] for some 30 DNA/protein complexes.The authors concluded that the EEC observed in these systems must, hence,b ed ue to hydration, that is,t he release or uptake of water. Thes ame master curve was found by us for the system dPGS/ lysozyme, [54] which will be discussed further in Section 3.

Interaction of RNA with Proteins
There are much fewer thermodynamic studies on the interaction of RNAw ith various proteins that consider explicitly the dependence on ionic strength in terms of Equation (3). Maiti and co-workers presented ac omprehensive study of the interaction of HIV-1 TARR NA and Tatderived arginine-rich peptides by various techniques,i ncluding ITC. [139] Plots of the binding constant according to Equation (3a) are linear and show that one ion is released upon binding.Samatanga et al. investigated the interaction of single-stranded RNAwith various proteins containing various RNA-recognition motifs by ITC. [140] Use of Equation (3a) demonstrated that the ionic interaction is small for these systems and DG res is mainly dominated by hydrogen bonding. It is interesting that both studies found the interaction of RNAw ith the respective proteins to be mainly driving by enthalpy.C ababie et al. recently presented ac arefully conducted thermodynamic study of the interaction of the NS3 helicase with single-stranded RNAb yu sing fluorescence titration. [141] Equation (3a) was shown to give agood description of the measured binding constants.T ypically, Dn ci was found to be five and rather independent of the ions used for adjusting c s .

Glycosaminoglycans (GAG) as Highly Charged Polyelectrolytes
Glycosaminoglycans (GAGs) such as heparin consist of oligosaccharide units that may be sulfated. [16,18,19] Animal tissues contain multiple sulfated glycosaminoglycans,such as heparan sulfate (HS), heparin, chondroitin sulfate (CS), dermatan sulfate (DS), and keratan sulfate (KS), which can be distinguished by their sugar constituents and sulfation pattern. [19] Figure 3d isplays the repeating unit of heparin, which is the most-studied GAG. In general, GAGs exhibit variations of the molecular structure,a nd the degree of sulfation may change.T hus,F igure 3s hows only the most abundant repeating unit (see the discussion of this point in Ref. [19]). Heparin has four charges per disaccharide repeat unit and is one of the most highly charged biopolymers. Heparin acts as an anticoagulant [42] and can interact with various proteins. [17,62,142,143] Moreover,hydrogels consisting of heparin and modified GAGunits are capable of sequestering proteins and, in particular,cytokines that may prevent wound healing. [30,56,144] Heparan sulfate (HS), which is slightly less sulfated than heparin, is located in the extracellular matrix and serves as aprimary receptor for many pathogens,such as bacteria and viruses.H Sw as shown to be involved in the infection by many viruses through facilitating their internalization or interaction with secondary receptors. [37,69,79,80,145] Thus,i ti sn ow clear that attachment of many viruses to cells involves electrostatic interactions with HS. [145] Therefore, an umber of sulfated molecules have been investigated as inhibitors (cf.t he discussion of Table 2i nR ef. [145]). In general, GAGs have been tested for sequestering or the defined delivery of cytokines and growth factors. [19] A thorough and quantitative understanding of the interaction of GAGs with proteins is ac entral problem in biomedical research.
Since the early work of Olson et al. [63] and of Mascotti and Lohman [64] it is well-established that electrostatic interaction and counterion release play ac entral role in the binding of proteins to heparin. [83,[146][147][148][149][150][151][152] Thep revalence of electrostatic interactions beween proteins and heparin has been corroborated by aconsiderable number of investigations; [68,83,151,153,154] asurvey of the older literature may be found in the review by Seyrek and Dubin from 2010. [65] Thus,l inear plots of log K b versus log c s are found in an umber of investigations. [65] The number of quantitative studies employing Equation (3), however, is rather small given the obvious importance of GAGs as biomaterials. [19,155,156] It is important to note that electrostatic interactions with heparin are already used in medical applications.T hus,p rotamine,w hich is ah ighly cationic polypeptide,i su sed to neutralize an overdose of heparin. [157] Ad etailed discussion of this application will be given in Section 4.  Less quantitative work has been carried out using ITC on the interaction of synthetic linear polyelectrolytes with proteins.C areful work by Dubin and co-workers,h owever, has shown that charge-charge interactions are central for the understanding of the complex formation between proteins and various polyelectrolytes. [83,[158][159][160][161][162] Af irst investigation of the linear polyelectrolyte poly(allylamine hydrochloride) with BSA by Ball et al. [163] using ITC demonstrated that the driving force for complex formation is entropic.E quation (3a) was used repeatedly to model the interaction, and in many cases ag ood linearity was found, at least at higher ionic strength (see the discussion of this problem in Ref. [109]) Recently,acareful investigation of this problem was presented by Lounis et al., [164,165] who demonstrated that Equation (3a) provides an excellent description of experimental data for the interaction of linear and dendrigraft poly(lysine) with synthetic anionic polyelectrolytes.
We have recently analyzed the interaction of human serum albumin (HSA) with short chains of poly(acrylic acid) (PAA) in aqueous solution as afunction of the ionic strength and temperature. [50] Thel ow molecular weight of PA A prevented the formation of complex coacervates,a nd ITC could be used for afully quantitative analysis of DG b . Figure 4 shows that Equation (3) is valid for higher ionic strengths, whereas low salt concentrations led to deviations,asdiscussed above.T he simplicity of this systems allowed us to perform molecular dynamics (MD) simulations of this binding process. Figure 4b displays as imulation snapshot of ac omplex between HSA and poly(acrylic acid). Thesimulations suggest that the linear polyelectrolyte is bound in the Sudlow II site, which is to be expected from earlier studies of HSA. Moreover,t he number of released counterions could be obtained from the simulations.This number can be compared to the experimental result obtained through application of Equation (3a) (Figure 4a). We found three counterions to be released in the binding process from simulations as well as from the experiment. [50] Thus,t he good agreement between theory and experiment corroborates the analysis of binding in terms of Equation (3a). [50] It is important to check whether agiven protein is changed upon interacting with apolyelectrolyte.Inthis case,apart of the caloric signal would be due to ap artial denaturation or ar efolding of the protein. Fort he case of DNAi nteracting with various proteins,t his problem has been investigated by Privalov et al. [138] (see also the discussion in Ref. [11]) and by Jen-Jacobson et al. [13] Fort he system HSA/PAA discussed above,the resulting complexes have been analyzed by smallangle neutron scattering (SANS). [166] No significant changes in the overall structure of HSA could be detected by this method. CD spectroscopy is an excellent tool to reveal possible changes in the secondary structure of complexes. [167][168][169][170] Thus,i fH SA interacts with dendrimers having partially hydrophobic moieties,there is asignificant loss of ahelices. [170] Te sts on the secondary structure of ap rotein in ac omplex with ag iven polyelectrolyte are therefore mandatory.
From the results obtained for al arge number of natural and synthetic systems,one can state that the analysis of DG b in terms of Equation (3) has led to as emiquantitative understanding of the interaction of polyelectrolytes with proteins. Thee ffect of counterion release can be separated from the other factors by use of Equation (3a), which provides the first step towards the quantitative understanding of DG b . ITC measurements have turned out to be central for these studies and MD simulations will allow us to acquire am olecular understanding of the thermodynamic data.

Biotechnological and Medical Applications of Linear Synthetic Polyelectrolytes
An important application of charge-charge interaction is gene delivery by nonviral vectors. [22,23,[171][172][173] Here,c ationic polyelectrolytes are used to compact DNAa nd RNAb y formation of so-called polyplexes. [20] Themicelles and aggregates formed by this interaction may then form more complex supramolecular structures. [174] Polyethyleneimine (PEI) has been the cationic polyelectrolyte of choice. [175,176] Concerns about the inherent toxicity of PEI has led to an enormous number of studies that have tried to improve gene delivery by The polyelectrolyte is bound to the Sudlow II site of HSA. [50] designed block copolymers,w hich has recently been review by Kataoka and co-workers [25] and by Reineke and coworkers. [177] Charged dendrimers have also been used for this purpose. [178] An interesting application is the delivery of proteins through as uitable packaging by block copolymers with charged blocks.Here,weonly cite recent work on block copolymers that deliver the CRISPR/Cas9 system [179] and the nanoformulation of the brain-derived neurotrophic factor (BDNF) by ab lock copolymer containing ap oly(glutamic acid) block. [60] In the latter case,c ationic patches on the BDNF interact electrostatically with the negatively charged block, and the resulting supramolecular structures then lead to abetter delivery of the BDNF.
It is interesting to note that linear polyelectrolytes may act as synthetic chaperones thus,g uiding proteins to adopt the correct tertiary structure.T his was shown by Semenyuk et al. in as eries of careful studies. [180][181][182] Thec omplexes of the proteins with polyelectrolytes such as polystyrene sulfonic acid also stabilized the structure of the proteins against aggregation in av ery efficient manner.F urthermore,t he complexes were stable at temperatures where the free protein would be denatured. This complexation of proteins with linear polyelectrolytes hence holds the promise for further biotechnological applications.
At otally different problem of medical relevance arises when considering the interaction of short-chain polyelectrolytes and small charged molecules such as phenylacetic acid with HSA. These substances adhere strongly to HSA and are, therefore,d ifficult to remove by ac onventional dialysis. Patients with chronic kidney disease have high concentrations of such uremic toxins,w hich may lead to ahigher cardiovascular morbidity. [166,183,184] ITC is ac entral tool for analyzing the interaction of such toxins with HSA. [166] Here,s hort polyelectrolytes may serve as models for the so-called middle molecules that present uremic toxins stemming from degraded proteins (cf.R ef. [166]). Thei nteraction with HSA is mainly depends on counterion release,a ss hown above. [50] Small toxins such as phenylacetic acid or indoxyl sulfate, however, interact mainly with the hydrophobic sites of HSA and exhibit ar ather high binding constant. [166] Removal of uremic toxins is,h ence,acentral task of clinical nephrology and an improved thermodynamic understanding of their interaction with HSA is absolutely necessary.

Charged Networks, Dendritic and Hyperbranched
Polyelectrolytes, and Polyelectrolyte Brushes

Dendritic and Hyperbranched Polyelectrolytes
Thep revious section hasd emonstrated that the interaction of proteins with linear polyelectrolytes can be largely understood and modeled. In an ext step,w ec onsider more complicated structuresand start withbranched anddendriticpolyelectrolytes. Highly charged dendrimers have been the subject of intense research since the first pioneering theoretical study by Welch and Muthukumar [186] in 1998. Charged dendrimers have been studied for gene transfection for al ong time, [172,178,[187][188][189][190][191] and discussed for drug delivery in general. [192][193][194] We have recently investigated charged dendritic polyglycerols.F igure 5g ives asurvey of these systems and the main results achieved so far. Figure 5a displays the chemical structure of the polyanionic dendritic polyglycerol sulfate (dPGS). Thes caffold consists of the highly hydrophilic polyglycerol, on to which sulfate groups are appended. These systems based on hyperbranched polyglycerol were made for the first time in 2004 [185] and used for various medical purposes. [34] In afirst step,for abetter understanding of the interaction of dPGS with proteins and more complicated biological systems (see Figure 2), we have studied the spatial structure of these dendrimers.F irst, MD simulations were used to explore the interaction of the highly charged systems with their counterions. [51] Figure 5b displays at ypical simulation snapshot of ad PGS dendrimer of the 2nd generation. The segments of the scaffold and the end groups were modeled in acoarse-grained fashion. It is clear that these systems present rather dense structures,where the charged groups are located mainly at the outside.S imulation can serve to define an approximate surface of the dendritic structure that may be compared with measured hydrodynamic radii. [51] Thecounterions are highly correlated to the macroion and form ad ense layer on the surface of the dendrimer. Hence,asurface concentration c ci may be defined in the same way as already discussed in conjunction with linear polyelectrolytes [see the discussion of Equation (1)].T his surface concentration is on the order of 1m for ad PGS of second generation and rises considerably for higher generations.
It is important to understand that correlation of the counterions with the highly charged dendrimer proceed on am esoscopic level, in which molecular details play am inor role.C oarse-grained simulations may hence lead to ab etter understanding of the counterion release mechanism, but cannot reveal details of interactions related to,f or example, hydrogen bonding.However,MDsimulations can be directly compared to the hydrodynamic radius,a nd the effective charge determined experimentally. [51] These data agree with the simulations within the limits of error.I np articular,t he effective surface charge levels off with increasing number of generations,w hile the bare charge increases exponentially. Thus,t hese systems exhibit the charge renormalization expected for highly charged spherical macroions. [51] This charge renormalization must be kept in mind when comparing the interaction of charged dendrimers of different generations with proteins.E vidently,t hese systems are expected to interact with proteins through counterion release in the same way as already discussed for the linear systems above.F urthermore,as alt concentration of 1m will lead to av anishing contribution of the counterion release [see Eq. (6)] and provide ag ood reference state.T he results of these coarse-grained simulations have been checked and fully corroborated by atomistic simulations with explicit water. [195] In asecond step,MDsimulations turned out to be highly revealing when studying the interaction of dPGS with proteins. [45,53] Figure 5c displays the ITC diagrams for the interaction of as econd-generation dPGS with lysozyme in aqueous solution. Ap arameter of the different curves is the ionic strength in these solutions,which ranges from 10 mm to ap hysiological concentration of 150 mm.T he weakening of the interaction with increasing ionic strength is directly apparent, and the inset of Figure 5c shows that the logarithm of the binding constant scales linearly with the salt concentration c s in solution, as predicted by Equation (6). Theslope of these lines leads directly to the number of released counterions (three), in good approximation to that already discussed in conjunction with Equation (6). Moreover,t he free energy of binding DG b is nearly independent of temperature,w hich is followed by as trong compensation of the enthalpy and the entropy of binding.This particular point will be discussed further in Section 3.5.
MD simulations now lead to data that can be directly compared to experiments:F igure 5d displays at ypical snapshot of ac omplex in which four lysozyme molecules are bound to at hird-generation PGS.F irst of all, the interaction of the protein with ad PGS molecule is quantitatively obtained by steered Langevin simulations:H ere the centers of gravity of the dPGS and the protein are kept at af ixed distance and the force between the two molecules is averaged. By integration over the distance,weobtain apotential of the mean force,t he maximum of which is the free energy of binding DG b . Moreover,t he number of released counterions and the average number of bound proteins can directly be obtained from these simulations and compared to experiments.Acomparison with experimental data showed an excellent agreement. [45,50,53] Then umber of released counterions derived from the simulations compare very well with the experimental data (cf.the discussion of Figure 5c). Moreover, it was demonstrated that the free energies derived from simulations can be directly compared to experimental data. Here again, good agreement is found. [45] Hence,M Ds imulations provide an excellent tool for the quantitative understanding of the interaction of polyelectrolytes with proteins.
Thes ame combination of ITC and MD simulations was recently applied to complexes formed between secondgeneration dPGS and HSA. [53] Thesame features as discussed for the dPGS/lysozyme system are found here as well:Awelldefined 1:1complex is formed and counterion release is found to be the main driving force.Again, astrong EEC is found by ITC measurements as af unction of temperature.T he experimental binding constant again agrees with the simulated one within the limits of error. TheC Ds pectra of the complex measured up to 310 Kshowed no significant change Figure 5. Sulfated dendritic polyglycerol (dPGS) and its interactionw ith proteins. a) Chemical structure of dPGS. The scaffold consists of ahighly hydrophilic dendritic or hyperbranched structure, with each end group carrying asulfate group. [185] b) Snapshot of the coarse-grained structure of asecond-generation dPGS. Red beads mark the terminal sulfate groups of the dendritic structure, and yellow beads mark its scaffold. The counterions are displayed as green beads. [51] c) Interaction of asecond-generation dPGS with lysozyme measured by ITC at different ionic strengths. The incremental heat per injection is plotted against the molar ratio of lysozyme to dPGS in aqueous solution. The inset displays the log of the resulting binding constant as afunction of the log of the salt concentration according to Equation (3). [45] d) Coarse-grained MD simulationso fthe interaction of athird-generationd PGS with lysozyme. The snapshot shows acomplex of the central dPGS molecule with four lysozymes. [45] when compared to the spectra recorded at room temperature. This finding is in contrast to complexes formed by aP AMAM dendrimer with HAS,where amajor reduction of the a-helix content was found because of partial unfolding. [170] TheM Ds imulations of dPGS interacting with proteins [45,52,53] were carried out using only implicit water, that is,a ll simulations have assumed water to be as tructureless medium with ag iven dielectric constant. Here the question arises in what way water is involved in the process of binding. This problem has recently been elucidated further by reconsidering the measured binding constant of the dPGS/ lysozyme [52] system in terms of Equations (5)- (7). [54] Figure 6a shows atypical plot of the measured binding constant as af unction of the salt concentration c s according to Equation (3a). Thes trict linearity of this plot allows us to determine the number of released counterions with good accuracy.M oreover,t he binding constant K b (1m)c ould be extrapolated with equal precision and used for the breakdown of the measured data, according to Equation (5), into to apart (DG ci )s olely due to counterion release and ar esidual part (DG res )d ue to specific interactions,s uch as salt bridges and hydrogen bonding.Atthe same time,the enthalpy of binding was largely balanced by an entropic term of comparable magnitude.F igure 6b displays ap lot of the enthalpy of binding versus the residual entropy of binding DS res multiplied by T [see the discussion of Eq. (7)].A ll data collapse on as ingle master curve that shows that the breakdown of the free energies of binding according to Equation (5) provides an excellent approximation for the data. This master curve is given by Equation (8).
Thei ntercept of À21.3 kJ mol À1 is,h ence,t he average value of DG res for the present system, and the slope very near to unity shows that there is anearly full compensation of the enthalpy by entropy.
It is interesting to note that this master curve obtained for the dPGS/lysozyme [52] system shown in Figure 6b)v irtually coincides with the master curve found by Dragan et al. for some 30 systems in which DNAi nteracts with various proteins (dashed line in Figure 6b). Thes lope of this master curve is slightly higher than 1( 1.09 vs.1 .017 for the dPGS/ lysozyme system) and DG res is slightly smaller. Despite these small differences,both investigations agree that the binding of proteins to DNAl eads to am arked EEC with an on-zero value of DG res . Dragan et al. [11] explained the marked EEC by the uptake or release of water during binding.Our findings [54] underscore this idea and demonstrate,i na ddition, that the binding of dPGS to proteins may be directly compared and modeled as the binding of DNAt ovarious proteins.

Sulfated Polyglycerol as an Anti-inflammatory Drug
As already mentioned in the Introduction, dPGS has as trong anti-inflammatory effect. [34,36,196] Figure 7s hows the mode of action of dPGS:The recruitment of leukocytes to the sites of inflammation is an important step in the pathogenesis of acute and inflammatory diseases,w hich include hypersensitivity reactions and autoimmune diseases.This process is orchestrated by gradients of cytokines and chemokines and by distinct expression and activation of several family members of adhesion molecules,i ncluding selectins and integrins. Dernedde et al. demonstrated that dPGS binds to the positively charged amino acid residues (arginines) close to the carbohydrate binding pocket of L-and P-selectin with high affinity in the nanomolar range;n ob inding takes place with E-selectin due to the absence of basic residues. [36] As discussed above,t hese findings were recently directly proven by MD simulation of the interaction of these selectins: Figure 6. Thermodynamic analysis of the binding of lysozyme to asecond-generation dPGS. [54] Top: Ap lot of log K b versus log c s as suggested by Equations (4) and (6). There is aperfectly linear relation in this double-logarithmic plot, in which the slope gives the numbero f released counterions,a sd iscussedi nconjunction with Equation (3a). The linear relationship is used to extrapolate the binding constant DG b at asalt concentration of 1 m. K b (1 m)i srelated to DG res ,the residual of the Gibbs free energy of binding according to Equation (5,) and reflects all contributionst oDG b not related to counterion release. Bottom:Enthalpy-entropy compensation for the data obtained on the system dPGS-G2/lysozyme. The enthalpy DH b is plotted against TDS res = TDS b ÀTDS ci according to Equation (7). The solid line denotes the fit by Equation (8). The dashed line shows the master curve derived by Dragan et al. [11] for awide variety of systems in which DNA interacts with proteins.
Counterion release was found to be the main driving force for binding,a nd the experimental binding constant of at hirdgeneration dPGS with L-selectin [78] could be reproduced by the simulation in anearby quantitative fashion. [45] It is important to note that ion-specific effects may play an important role as well. Thus,W einhart et al. analyzed the interaction of several polyglycerol-based anions with Lselectin. [197] Thes trength of interaction increased in the order carboxylate (no inhibition) < phosphate < phosphonate % sulfonate < bisphosphonate < sulfate.Hence,the electrostatic effect alone cannot be solely responsible for the strength of binding.F urthermore,P aulus et al. studied the effect of dPGS-branching on the inhibition of inflammatory processes. [191] It was found that ad PGS with ad egree of branching of 60 %h ad ah igher binding strength than as ulfated, perfect dendrimer characterized by ad egree of branching of 100 %. More recently,biodegradable dPGS was shown to exhibit promising features for anti-inflammatory applications,t hus replacing heparin. [196] All the results obtained so far clearly reveal charge-charge interactions to be the major driving force for binding.
In the meantime,s everal inflammation and tumor-relevant proteins were identified as nanomolar binders for dPGS, such as interleukin-1 (IL-1), L-selectin, P-selectin, interleukin-6 (IL-6), [198] lectin-type oxidized low-density lipoprotein receptor 1(LOX-1), [199] and the complement factors C1q and C5a. [200] Thebinding of dPGS is rather unspecific and does not necessarily depend on au nique protein structure.T his is in contrast to species-specific inhibitors that target ligandreceptor interactions with peptides,p roteins,o ra ntibodies. Thus,t argeting by dPGS is much less sensitive due to evolutiondriven variations.

Complement Pathway
Thei mmune system of vertebrates consists of ac ombination of complex mechanisms that must be tightly regulated to act prompt and properly. [201] Here,t he innate immune system is the unspecific first line of defense against invading microorganisms and must act broadly to detect and eliminate pathogens. Thec omplement system consists of an umber of soluble blood proteins that are activated through ap roteolytic cascade mechanism. Inhibition at distinct checkpoints paralyze the complement activation and are of importance in several pathologies characterized by dysregulated excessive activation, with sepsis being the most prominent disease. [202,203] Recently,S ilberreis et al. identified that dPGS targets the three different pathways of the complement cascade and that charge-charge interaction plays an important role to balance the activation (Figure 8). [200] It was shown that dPGS binding to the complement factors C3 and C5 inhibits further processing and subsequent release of the anaphylatoxins C3a and C5a. In addition, charge-dependent sequestration limits the anaphylatoxin function. Highly charged polyelectrolytes such as heparin and heparan sulfate were shown to act similarly, [204] but by far not as effective as the synthetic polymer dPGS that binds the anaphylatoxin C3a with low micromolar affinity and C5a with nanomolar affinity. [200] Thus,d PGS may be ap romising candidate for ad rug that counteracts an overshooting complement activation in sepsis and other diseases such as rheumatoid arthritis. [205][206][207]

Interaction of dPGS with Cellular Systems
Theeffect of dPGS on neural cells was also investigated in models of endotoxemia caused by lipopolysaccharide (LPS) in primary neural cultures and in animals. [198] Figure 9s hows the main findings in aschematic fashion:dPGS can reduce the negative impact of cytokines on neural brain cells through attenuation of the hyperactivity of microglia and lipocalin-2 Figure 7. The anti-inflammatory effect of dPGS. As shown by Dernedde et al., [36] dPGS inhibits an overwhelming inflammatory response and reduces the extravasation of leukocytes.dPGS targets the adhesion molecules L-and P-selectin, while no binding to E-selectin is observed.T he same finding was made in our recent study by MD simulations. [45] Thus, dPGS acts by preventingleukocyte extravasation through the binding of the selectins. Moreover,binding to complementf actors C3 and C5 inhibits the formation of the proinflammatory anaphylatoxins. Here, the reduction of the C5a level decreases further leukocyte activation and recruitment. As aresult, the adhesion cascade is balanced and contributes to initiate the healing process. [196] Angewandte Chemie Reviews 3894 www.angewandte.org release from astrocytes.E nhanced microglia activation caused astrocyte activation, and dPGS was ap owerful modulator of the cross-talk between the microglia and astrocytes.d PGS directly bound to IL6, thereby preventing the binding of cytokine to its receptors and reduced the propagation of neuroinflammation. dPGS was internalized both by microglia and astrocytesinaconcentration-and timedependent manner.
Aside from strong neuroglia activation by LPS,A b 42 oligomers can also activate neuroglia but to al esser extent. [40] Them echanism of the dPGS action involved ad irect binding of the Aß 42 oligomers to dPGS,t hus interfering with the formation of Ab fibrils. [40] Thetreatment with dPGS prevents the deleterious effects of oligomeric Ab on dendritic spines at the excitatory synapses in the hippocampus and normalizes the neuroglia activity in this brain structure (Figure 9). Taken together,t hese studies suggest that dPGS is avalid candidate for therapeutic interventions in neurodegenerative disorders implicating neuroinflammation of the central nervous system.
Summing up the previous work related to dPGS and its application to various systems and medical problems,i t becomes evident that the marked localization of the counterions on the surface of these dendritic structures provides the key for the understanding of the results:M Ds imulations together with experiments [51] demonstrate that the high charge density on the surface leads to asurface concentration of counterions of the order of 1m [see the discussion of Equation (3)].E lectrostatic interaction with proteins and more complicated systems will release apart of these surfacebound ions into the bulk solution with ar educed ion concentration. In cells,this concentration is 150 mm, whereas the extracellular matrix is characterized by even lower salt concentrations.B inding will be brought about by entropic forces that work even under physiological salt concentrations. Evidently,this counterion release force is only one part of the free energy,o ther factors,s uch as the release of water molecules and hydrogen bonding,will come into play as well.

Polyelectrolyte Brushes
If long linear polyelectrolyte chains are appended to planar or curved surfaces,apolyelectrolyte brush Figure 8. Complementpathway:dPGS interfereswith the three pathways of complementa ctivationa nd reduces formation of the membrane attack complex (MAC), which is apore that is inserted into the cytoplasmic membrane and thereby leads to cell death. Reduced activity of the C3 and C5 convertase results in IC 50 values of 60 nm (lectin pathway), 300 nm (classical pathway), and 900 nm (alternative pathway). [200] Figure 9. Modulatory effects of dPGS in neuroinflammation caused by Aß oligomers. a) Exposure of microglia to Ab oligomers causes the activation of microglia and loss of dendritic spines in the hippocampal excitatory neurons. Hyperactivemicroglia activate astrocytes and these glial cells (reactivea strocytes) produce excessive amounts of lipocalin 2( LCN2). LCN2 in combination with cytokines released from hyperactive microglia contribute to the impairmento fsynaptic functions. b) dPGS attenuates microglia hyperactivity,binds to Ab 42 and normalizes the number and function of dendritic spines. [40] results. [110,[209][210][211][212] Theb rush limited is reached when the average distance between the grafted chains on the surface is smaller than their dimensions in solution. [209] Thei nteraction of these polyelectrolyte brushes with proteins has been the subject of al arge number of studies,w hich have been reviewed recently. [109] Hence,abrief discussion of this problem will suffice here.F igure 10 displays schematically the adsorption of proteins on spherical polyelectrolyte brushes.F or al ow ionic strength in solution, 95-98 %o ft he counterions are confined within the brush layer. [110,213,214] This confinement will lead to ah igh osmotic pressure within the brush layer and ac oncomitantly strong stretching of the polyelectrolyte chains. [214] Theuptake of proteins will lead to ap artial release of these counterions,w hich is the main driving force for adsorption. [111,215] At high ionic strength, on the other hand, the limit of asalted brush is attained. [211,213,214] at this limit, proteins can hardly adsorb on the brush layer. Moreover,a dsorbed proteins will be released when going from al ow ionic to ah igh ionic strength. [216,217] The interaction of proteins with such adense polyelectrolyte layer can, hence,beunderstood in terms of the counterion release force discussed above in Section 2.1. FTIR spectroscopic studies revealed that there is hardly any change in the secondary structure of the adsorbed proteins. [218,219] Thesame conclusion could be drawn from the activity of adsorbed enzymes [220,221] and from spectroscopic studies of the green Figure 10. Uptake of proteins by aspherical polyelectrolyte brush (SPB). [208] Top: The polyelectrolyte brushes consist of asolid polystyrene core (gray sphere) with aradius R h,core between 50 and 100 nm. Onto its surface are grafted long chains of polyelectrolytes, for example, poly(acrylic acid). Red spheres on the PAA chains represent the negative charge of the acidic residues, while blue spheres represent the positive counterions. Nearly all of the counterions of the brushes are confined within the brush layer (osmotic brush). The protein molecules are represented by green spheres. Their uptake will lead to the release of aconcomitant number of counterions. Adsorptiono fproteins by polyelectrolyte brushes is hence mainly entropy-driven. [109,208] Bottom:a )The Gibbs free energy of binding DG b of HSA to aspherical polyelectrolyte brush carrying long chains of poly(acrylic acid) (black squares) compared to the results for HSA binding to dPGS and of HSA interacting with linear chains of poly(acrylic acid). In all cases, the ITC-determined DG b exhibits only aw eak dependence on temperature, which is followed by astrong enthalpy-entropy compensation (EEC) shown in (b) for HSA interacting with aSPB:Both DH b as well as TDS b vary strongly with temperature, whereas DG c stays nearly constant because of the EEC. [208] Angewandte Chemie Reviews 3896 www.angewandte.org fluorescent protein adsorbed on as pherical polyelectrolyte brush. [217] It is interesting to note that the adsorption of proteins on spherical polyelectrolytes is accompanied by am arked enthalpy-entropy compensation, exactly in the way discussed for the dPGS/lysozyme system (see the discussion of Figures 5  and of 10 b,c). Ar ecent study [208] of the adsorption of human serum albumin on as pherical polyelectrolyte brush by ITC has revealed that the free energy of binding depends very little on the temperature,while the enthalpy and the entropy of adsorption vary linearly with temperature (Figure 10 c). [208] Figure 10 b,cs uggests that the strong enthalpy-entropy compensation is ag eneral feature that always occurs when polyelectrolytes interact with proteins-from the complexes of DNAw ith proteins [11,49,91] to the binding of proteins to synthetic polyelectrolytes. [54,109,208] Afully quantitative theory of this effect, however, is still lacking.

Charged Networks
Networks bearing charges have been aclassical subject of polymer science and the first quantitative theory dates back to the classical paper of Michaeli and Katchalski from 1955. [222] More recently,c harged networks have been the subject of as eries of comprehensive theoretical studies by Košovan, Holm, and co-workers. [76,[223][224][225] It is fair to state that we now have acquired avery good physical modeling of these systems that helps us to understand their interaction with proteins. Figure 11 shows the main feature of charged networks exemplified for charged core-shell particles: [73,74] Thec ounterions are fully confined within the network and the total number of co-and counterions within the network is determined through the Donnan equilibrium. TheD onnan potential determines the leading term for the interaction of charged entities such as proteins with the network. The decisive parameter for protein uptake is the difference in the ionic strength inside and outside the network and the overall charge of the proteins. [74] There is alarge number of experimental studies related to the uptake of proteins by charged networks,w hich started with as eries of investigations by Kabanov,Z ezin et al. [55,226] Later studies include the work of Cohen-Stuart and coworkers. [227] Am ore detailed discussion of these investigations is beyond the scope of the present Review.Here we only mention the studies by Yigit et al., [73,74] who investigated the uptake of various proteins by charged core-shell microgels and compared the findings to their theoretical model. In particular,O berle et al. [74] were able to show that this model can even predict the results of the competitive adsorption of two different proteins,t hus demonstrating the power of ap urely analytical model. Moreover,t he difference in the free energy between the free and the adsorbed state of aprotein can be used in Dynamic Density Functional Theory (DDFT) to model the kinetics of protein uptake into an etwork. [75] DDFT is also capable of describing non-monotonous effects in competitive adsorption [228] ("Vroman effect"; cf.the discussion in Ref. [212]).
In as eries of papers,W erner and co-workers develop biocompatible highly charged hydrogels that can be used to adsorb and hence act as medical aids. [27,31,56,71,229] Glycosaminoglycan (GAG) based hydrogels with avaried GAGcontent and GAGs ulfation pattern were prepared and applied to sequester cytokines.Cytokines are small proteins with various isoelectric points.Hydrogels containing GAGs with different sulfation patterns have been shown to adsorb cytokines, chemokines,and growth factors. [31,56] Thus,networks containing defined GAGs equences can be employed, for example, for healing of chronically inflamed wounds by sequestering various cytokines.Areview of this application and others has recently been provided by Werner and co-workers. [31]

Virus Inhibition by Nanogels
As mentioned in Section 2.5, heparan sulfate (HS) moieties are located in the extracellular matrix and the glycocalyx. They are involved in the infection of many viruses through interaction with secondary receptors. [37,69,79,80,145] In general, viruses attach to and ultimately enter cells using multivalent interactions of viral ligands with receptors localized on the cell surface.Hence,nanoparticles of suitable size and that are highly charged can be used as multivalent Figure 11. Modeling the competitive adsorption of proteins onto charged networks as exemplified by charged core-shell microgels. [73,74] The shell consists of acharged network built up of hydrophilic chains. The network containsn egatively charged monomeru nits, which lead to acharge density c g . The concentrationo fthe counterions and the co-ions within the network are regulated by the Donnan potential. The proteins are modeled by charged spheres with charge numbers z 1 and z 2 ,r espectively, whereas the overall radii are given by R 1 and R 2 , respectively.The uptake of proteins is governed by the interaction of the charged proteins with the Donnan potential of the network. [73] The model can consider the competitive adsorptiono fseveral proteins onto the network. [74] Here, two different proteins with effective charges z 1 and z 2 undergo competitive adsorption to the charged core-shell particle. The model leads to af ully quantitative understanding of the experimental results with four different proteins. [74] receptors that compete with HS and thus block the docking of viruses on the cell surface. [37] Sulfated nanogels with asize of 100-200 nm to match the virus size were synthesized and tested as antiviral agents.Flexibility of the cores turned out to be important because it resulted in amore effective shielding of the surface of the virus.Dey et al. [37] demonstrated that, for example,H SV-1 viruses are blocked by charge-charge interactions:T he positively charged glycoproteins on the virus surface normally adhere to the negatively charged HS. Highly charged dPGS microgels can suppress this interaction by adhering to the virus and thus prevent the uptake of the virus by the cell. Thus,c harge-charge interactions,m ost probably by counterion release,s eems to be central for ac learer understanding of virus uptake and inhibition.

Complex Polyelectrolyte Architectures
In the last section of this Review,wenow turn to systems with higher complexity.H ere we deal with rather large polymeric structures that have been generated through the formation of covalent bonds or by self-assembly,for example, micelles.T hese systems have been designed for special purposes,s uch as drug delivery,a nd must match an umber of requirements:Low toxicity should be combined with high efficiency for targeting,f or example,t umor cells.T he polymeric scaffold with as ize of 10-100 nm should be degradable for full clearance afterwards.T he synthesis and analysis of complex architectures fulfilling these conditions certainly presents ag reat challenge,a nd the number of systems near to clinical use is still small. Here we choose two major problems in which polymeric systems have been applied successfully so far, namely drug delivery and anticoagulant reversal.

Drug Delivery
Micelles based on block copolymers with acharged block play ac entral role in this field. If the charged segments are characterized by acharge parameter x > 1, counterion release will again be am ajor driving force for self-assembly. [88,230,231] Polymersomes present another example for complex polymeric carrier systems. [232] Much of this work has been reviewed recently by Kataoka and co-workers, [25] and so the present discussion of carrier systems will be focused more on recent studies using dPGS micelles.
Ideal polymeric drug carriers should, of course,fulfil two requirements:T he micelles should be nontoxic and not interact with blood proteins.Astrong adsorption of various blood proteins may lead to prompt immune reactions and opsonization (cf.t he discussion in Refs. [233][234][235]). This problem has been addressed in many systems by ad ense coating of poly(ethylene glycol) chains.M oreover,t he micelles should carry their payload, for example,a na nticancer drug, directly to the cancerous tissue in ah ighly specific manner. This requires concepts for targeting micelles and presents an important problem for present research (cf. the discussion of this point by Cabral et al. [25] ).

Micelles for Tumor Targeting
Dendritic dPGS-based polymer micelle and the dPGS dendritic copolymer are highly potent candidates for the targeted delivery of poorly water-soluble drugs.T he extraordinary potential of such dPGS copolymer micelle formulations was first demonstrated by Zhong et al., [38] who used ad isulfide-bridged, cleavable dPGS-SS-PCL copolymer micelle ( Figure 12) for the encapsulation of poorly watersoluble dyes.T his study provided the first demonstration of tumor-targeted delivery and drug release for an intrinsic tumor-affine polymer. [38] To prove the applicability of dPGS copolymer micelle formulations of doxorubicin in vivo, Zhong et al. first investigated the elimination of doxorubicin from the blood in mice.B oth cleavable and no-cleavable micellar formulations delayed the elimination of doxorubicin from the blood (Figure 12 b). Af actor of 10 increase in the bioavailability after as ingle parental application of doxorubicin was shown.
Further proof of concept was provided by the treatment of established human mammary MCF-7 xenografts in nude mice.F or this purpose,t he hydrophobic anticancer drug doxorubicin was encapsulated within both the cleavable and noncleavable dPGS PCL copolymer micelle.T he growth inhibition of human MCF-7 mammary carcinoma cells in vitro was demonstrated with both formulations.W ed emonstrated that both cleavable and noncleavable dPGS copolymer micellar formulations of doxorubicin may increase the survival of tumor-bearing mice compared to vehicle-or doxorubicin-treated controls.H owever,s table long-term survival in 100 %o fi mplanted tumors was only achieved by repeated treatment with the doxorubicin-loaded cleavable dPGS-SS-PCL micelles,w hich can release more antitumor drug specifically inside the cells of the tumor tissue.

Polycation-Based Therapeutics for Polyanion Neutralization in Blood
Heparin-based anticoagulant drugs (unfractionated heparin (UFH), low molecular weight heparins (LMWHs), and fondaparinux) are widely prescribed for prophylaxis and the treatment of thromboembolic disorders,a swella si ns urgeries. [236,237] Despite its widespread use in clinics,am ajor limitation of this class of drug is as ide effect of bleeding, which necessitates the need for antidotes which can neutralize their anticoagulant activity. [42,238] To date,p rotamine is the only clinically approved antidote for UFH;however,itisnot effective against all heparins. [42,238] Protamine is ah ighly cationic polypeptide that interacts electrostatically with negatively charged heparins to form stable complexes, thereby providing antidote activity. [157] Its cationic charge density and binding strength is not sufficient to generate as table complex with LMWH or fondaparinux because of their low molecular weight and low degree of sulfonation (see Section 2).
To overcome these deficiencies,the Kizhakkedathu group recently developed an UHRA, as ynthetic nontoxic macromolecular heparin antidote capable of neutralizing all clin-Angewandte Chemie Reviews 3898 www.angewandte.org ically available heparin-based anticoagulants. [41,43,[239][240][241][242] Figure 13 as hows the chemical structure of the UHRA and its way of interacting with the antithrombin/heparin complex (Figure 13 b). TheU HRA consists of ac ore of HPG and tertiary amine based heparin binding groups that acquire cationic charges at ap hysiological pH value.T his core is protected by as hell of methoxypolyethylene glycol (mPEG) chains (brush layer). Unlike the naked cationic charges in protamine,t he shielded dense cationic charge within the UHRA prevents its ionic interaction with endogenous anionic macromolecules in blood such as proteins (e.g.f ibrinogen, coagulation factors) and cells (platelets,red blood cells). The mPEG brush layer offers sufficient entropic penalty to the incoming polyanions as ar esult of brush compression;t hus, only those highly charged polyanions such as heparins can overcome such ab arrier,t hereby providing selectivity to UHRA. [41] Thus,c harge-charge interactions and probably counterion release play amajor role in these processes.

Conclusion
Thes urvey of investigations on natural and synthetic polyelectrolytes demonstrates that their interaction with proteins is largely dominated by charge-charge interactions. Section 2s hows that this interaction can be described by counterion release embodied in Equation (3). As imilarly clear picture emerges from the studies done on dendritic polyelectrolytes,b rushes,a nd networks summarized in Section 3. Specifically, systems based on dendritic polyglycerol sulfate (dPGS) are already applied in animal models for medical purposes,for example as anti-inflammatory drugs, and the better understanding of the interaction of dPGS with proteins now achieved will certainly pave the way for many further applications.The situation is less clear for the more complex polyelectrolyte architectures discussed in Section 4. However,a ll the results obtained so far demonstrate the importance of charge-charge interactions, most probably related to counterion release.Moreover,all the investigations discussed here clearly reveal the importance of temperature as ad ecisive variable:I na ll the cases studied so far, as trong enthalpy-entropy compensation is observed. Further work is needed on this phenomenon and to explore its importance in living systems.T he entire survey,h owever, clearly demonstrates that am uch better understanding of charge-charge interactions is the key for the design of drugs based on polyelectrolytes.
At this point, we now suggest further work along the following lines:F or ag iven architecture of ap olyelectrolyte,t wo parameters are decisive:1)ionic strength and 2) temperature.Hence,ameaningful study of the interaction of ap olyelectrolyte system must always vary these two parameters.I np articular, the investigation of potential drugs based on polyelectrolytes must always include experiments at 37 8 8C, which may lead to distinctly different results to the ones conducted at room temperature.Calorimetric studies carried out as afunction of the salt concentration and temperature should be used to reveal and to design the strength and specificity of the interaction. Finally,t he huge potential of MD simulations must be explored further. Here,acombination of simulations with studies on single molecules may be an ew and very interesting avenue. [243][244][245][246] Taken together,s ynthetic polyelectrolytes and systems derived therefrom are certainly highly promising candidates for the development of drugs. Figure 13. Design of UHRA and its interaction with heparin. a) Structure of UHRA and heparin binding groups (R). The chemical structure consists of an HPG-based core and mPEG 350 brush layer with hexamethylated tris(2-aminoethylamine) as HBGs arranged in amultivalent fashion. b) The mechanism of the antidote action of UHRA. The antithrombin (AT) bound heparin complex responsible for the anticoagulant activity of polyanionic heparin is dissociated by interaction with the UHRA because of its high binding affinity.This process restores the generation of thrombin. [41,239] Angewandte Chemie Reviews 3900 www.angewandte.org