Home Investigation of the structure of regulatory proteins interacting with glycosaminoglycans by combining NMR spectroscopy and molecular modeling – the beginning of a wonderful friendship
Article Open Access

Investigation of the structure of regulatory proteins interacting with glycosaminoglycans by combining NMR spectroscopy and molecular modeling – the beginning of a wonderful friendship

  • Georg Künze ORCID logo , Daniel Huster ORCID logo EMAIL logo and Sergey A. Samsonov
Published/Copyright: April 21, 2021
Download Article (PDF)
Biological Chemistry
From the journal Biological Chemistry Volume 402 Issue 11

Abstract

The interaction of regulatory proteins with extracellular matrix or cell surface-anchored glycosaminoglycans (GAGs) plays important roles in molecular recognition, wound healing, growth, inflammation and many other processes. In spite of their high biological relevance, protein-GAG complexes are significantly underrepresented in structural databases because standard tools for structure determination experience difficulties in studying these complexes. Co-crystallization with subsequent X-ray analysis is hampered by the high flexibility of GAGs. NMR spectroscopy experiences difficulties related to the periodic nature of the GAGs and the sparse proton network between protein and GAG with distances that typically exceed the detection limit of nuclear Overhauser enhancement spectroscopy. In contrast, computer modeling tools have advanced over the last years delivering specific protein-GAG docking approaches successfully complemented with molecular dynamics (MD)-based analysis. Especially the combination of NMR spectroscopy in solution providing sparse structural constraints with molecular docking and MD simulations represents a useful synergy of forces to describe the structure of protein-GAG complexes. Here we review recent methodological progress in this field and bring up examples where the combination of new NMR methods along with cutting-edge modeling has yielded detailed structural information on complexes of highly relevant cytokines with GAGs.

Keywords: CXCL12; CXCL14; interleukin-10; molecular dynamics simulation; protein-GAG complex; solution NMR spectroscopy

Introduction

The interaction of proteins with glycosaminoglycans (GAGs) is of fundamental importance in biology. In particular, many recognition functions are mediated by cell surface bound GAGs. Thus, the interaction of cytokines and chemokines, which represent the particular focus of this review, as well as other regulatory proteins with the cell surface and subsequent diffusion to the relevant receptors is established. Also, virus entry into cells is related to the interaction with cell surface GAGs as impressively shown in a recent cryo electron microscopy structure of eastern equine encephalitis virus in complex with heparin (Chen et al. 2020).

GAGs are negatively charged linear polysaccharides composed of repeating disaccharide units, in which one residue is an amino sugar [N-acetyl-d-galactosamine (GalNAc) or N-acetyl-d-glucosamine (GlcNAc)] and the other one is an uronic acid residue [d-glucuronic acid (GlcA) or l-iduronic acid (IdoA)] or d-galactose (Gal) (Esko et al. 2009; Schiller and Huster 2012). Six GAG classes are distinguished based on monosaccharide composition and glycosidic linkage: heparan sulfate (HS), heparin (HP), chondroitin sulfate (CS), dermatan sulfate (DS), keratan sulfate (KS), and hyaluronic acid (HA) (Figure 1). In addition, both degree and pattern of sulfation vary between different types of GAGs as well as within the same GAG chain, a fact that is known as sulfation code and influences GAG structure and activity (Gama et al. 2006). Except for HA, which is not sulfated and synthesized as free polymer (Maytin 2016), all GAGs are initially linked to core proteins forming proteoglycans. The mammalian genome contains 43 proteoglycan-encoding genes, which can lead to an even larger number of alternatively-spliced protein variants (Iozzo and Schaefer 2015). Proteoglycans are abundantly present on the surface of animal cells and an integral part of the extracellular matrix (ECM), which serves as scaffold material in connective tissues. Thus, GAGs contribute to maintaining the structural integrity of cells and tissues. In addition, they bind chemokines, cytokines, growth factors, and enzymes, which underlies their function in crucial physiological processes like tissue development (Schwartz and Domowicz 2018), angiogenesis (Kastana et al. 2019), coagulation (Bourin and Lindahl 1993), and immunity (Collins and Troeberg 2019), as well as their involvement in pathological situations, e.g. viral infection (Tiwari et al. 2015), cancer (Afratis et al. 2012), or neurodegenerative diseases (Maïza et al. 2018). Therefore, elucidating the basic principles of interactions of specific GAG motifs with proteins and protein-protein complexes is critical for understanding the structure-activity relationships of GAGs and the mechanisms underlying these important biological processes. Such knowledge has the potential to contribute to new biomedical applications of GAGs, e.g. as anticoagulants (Pavao 2002), delivery systems for therapeutics (Ossipov 2019; Passi and Vigetti 2019), coating materials for implants (Bierbaum et al. 2012; Biran and Pond 2017), and components in three-dimensional scaffolds for tissue engineering (Farrugia et al. 2018; López-Ruiz et al. 2019; Yang et al. 2020).

Figure 1: Repeating disaccharide units of GAGs. Hyaluronan (HA), chondroitin sulfate (CS), keratan sulfate (KS), dermatan sulfate (DS), heparan sulfate (HS), and heparin (HP). Functional groups that can also be sulfated are indicated with a red box. The glycosidic linkage carbon atoms are numbered and labeled in red.
Figure 1:

Repeating disaccharide units of GAGs. Hyaluronan (HA), chondroitin sulfate (CS), keratan sulfate (KS), dermatan sulfate (DS), heparan sulfate (HS), and heparin (HP). Functional groups that can also be sulfated are indicated with a red box. The glycosidic linkage carbon atoms are numbered and labeled in red.

Several biochemical and biophysical methods have been applied to protein-GAG systems including affinity chromatography, analytical ultracentrifugation, electrophoretic mobility shift assays, mass spectrometry, isothermal titration calorimetry, and surface plasmon resonance (reviewed in Yang and Chi 2017). Recent progress in chemoenzymatic synthesis of structurally defined GAGs (Zhang et al. 2020; Lu et al. 2018) and in carbohydrate microarray technology (Zong et al. 2017; Rogers et al. 2011; Yang et al. 2017) have allowed screening the interaction of a target protein with a large number of GAG ligands, which has undoubtedly accelerated protein-GAG interaction studies. However, while all these methods have their specific scopes in terms of analyzing the influence of chain length (i.e., degree of polymerization, dp), charge density, monosaccharide composition, and sulfation pattern on GAG binding, they usually cannot provide atomically detailed structural information about protein-GAG interaction modes. At the same time, X-ray crystallography and cryo-electron microscopy (EM) usually have difficulties to resolve GAG molecules due to their high flexibility. Consequently, only a limited number of co-crystal structures of protein-GAG systems, usually with fully sulfated HP ligands instead of more physiologically relevant HS or CS ligands (less than 100 in the protein data bank, PDB), have been determined (Samsonov and Pisabarro 2016). By contrast, the extensive molecular dynamics of GAGs is not problematic for NMR spectroscopy, which can provide unique high-quality information about protein-GAG complexes. However, NMR spectroscopy suffers from the highly repetitive nature of GAGs rendering assignments to a given sugar ring difficult and from the fact that binding of GAGs to proteins is mediated by the protonless sulfate groups preventing detection of nuclear Overhauser effects for distance constraints.

These limitations suggest that experiments alone are often not sufficient for comprehensive insights into protein-GAG complexes at atomic level (Almond 2018). In particular, the GAG heterogeneity resulting from clustering of disaccharides with a specific sulfation pattern along the polymer chains and uronic acid epimerization hampers the collection of sufficient experimental data. This limitation can be overcome by the use of oligosaccharides of defined sequences and length in computational approaches aiming at deciphering molecular recognition features of GAGs. Along these lines, molecular modeling approaches could not only be complementary to experiments, but also provide crucial details inaccessible by experiments. This may also rationalize further experimental steps. Therefore, modeling can significantly contribute to the process of optimizing the analysis of protein-GAG complexes. Carbohydrates in general and GAGs in particular represent a far more complex and diverse class of biomolecules in comparison to proteins and nucleic acids which makes them more difficult for both experimental and theoretical analysis (DeMarco and Woods 2008). Together with the dramatic increase of the available computational power, substantial progress has recently been made in the development of theoretical approaches to characterize these systems: numerous carbohydrate-specific methods were developed aided by new force fields and energy functions (Griffith et al. 2017; Kerzmann et al. 2008; Kirschner et al. 2008) proving to be useful for detailed characterization of the structure-function relationship and the recognition in protein-GAG complexes.

So far, numerous proteins were analyzed and described computationally in terms of their interactions with GAGs: FGFs (Angulo et al. 2004; Bojarski et al. 2019; Faham et al. 1996; Muñoz-García et al. 2013, 2014; Nieto et al. 2013), VEGF (Uciechowska-Kaczmarzyk et al. 2018), endostatin (Ricard-Blum et al. 2004), IL-8 (Gandhi and Mancera 2011; Hofmann et al. 2015; Joseph et al. 2015; Möbius et al. 2013; Nordsieck et al. 2012; Pichert et al. 2012; Schlorke et al. 2012), RANTES (Singh et al. 2015), IL-10 (Künze et al. 2014, 2016), CXCL14 (Penk et al. 2019), PECAM-1 (Gandhi et al. 2008), CXCL12 (Panitz et al. 2016; Sapay et al. 2011), langerin (Munoz-Garcia et al. 2015), BMPs (Gandhi and Mancera 2012; Hintze et al. 2014), sclerostin (Veverka et al. 2009; Salbach-Hirsch et al. 2015), integrin (Ballut et al. 2013), LOX (Vallet et al. 2018), cathepsin proteases (Bojarski et al. 2019; Sage et al. 2013), PCPE-1 (Potthoff et al. 2019). Some of these studies are purely theoretical; others are conducted as combination of computer modeling with experimental methods such as NMR, surface plasmon resonance, mass spectrometry and diverse biochemical and cell assays.

Previously, common biochemical and biophysical approaches to the characterization of protein-GAG interactions (Yang and Chi 2017), the elucidation of the solution conformation of GAGs by NMR (Pomin 2014a), as well as common NMR methods for the investigation of protein-GAG complexes (Pomin and Wang 2018a), including strategies for making selectively isotopically labeled GAGs and proteins, have been reviewed. Most interaction studies have been conducted with fragments of HP and HS, and more than 400 HP-binding proteins have been identified (Ori et al. 2011). This has led to the concept of a HP interactome (Pomin and Mulloy 2015), i.e., a group of proteins with a diverse array of functions that all interact with HP. Among those proteins, chemokines and other cytokines, adhesion proteins, coagulation factors, growth factors, and growth factor receptors have drawn much attention and current knowledge about these systems has been summarized (Pomin 2014b, Proudfoot et al. 2017). Other types of GAGs, e.g., CS, DS, and KS, were, however, underrepresented in those studies. More recently, new developments in the chemoenzymatic synthesis of structurally defined GAG oligosaccharides (Xu et al. 2018, Zhang et al. 2020) and GAG microarray technology for the study of GAG interactomics (Pomin and Wang 2018b) have been discussed.

In our review, we present recent achievements in the investigation of protein-GAG interactions obtained by using a combination of NMR and molecular modeling. We focus on GAG-binding cytokines with immunologically relevant functions studied within the collaborative research center Transregio 67 “Functional biomaterials for controlling healing processes in bone and skin tissue – from material to clinic”. First, the current methodology and new advancements in NMR and computational modeling studies of GAGs, as well as their interplay are introduced, before several most significant examples resulting from the combination of NMR experiments with computer modeling approaches are discussed in detail. A summary of the workflow to obtain structural insight for GAG/protein complexes from a combined NMR/MD approach and an overview over the parameters that can be obtained from the combined methods is given in Figure 2.

Figure 2: Overview of the combined modeling and NMR approaches used in studying protein-GAG interactions. (A) Experimental and computational system preparation. (B) Prediction of the GAG binding region. (C) Protein-GAG complex structure determination. (D) Extraction of the atomic details on protein-GAG interactions specificity, their dynamic and energetic properties. Further details about the methodology are provided in the Methods section of this review.
Figure 2:

Overview of the combined modeling and NMR approaches used in studying protein-GAG interactions. (A) Experimental and computational system preparation. (B) Prediction of the GAG binding region. (C) Protein-GAG complex structure determination. (D) Extraction of the atomic details on protein-GAG interactions specificity, their dynamic and energetic properties. Further details about the methodology are provided in the Methods section of this review.

Methods

NMR techniques to study protein-GAG complexes

NMR spectroscopy offers a powerful toolbox to study protein-GAG complexes. It is capable of identifying chemical sites in GAGs and proteins that are involved in binding, and can describe GAG-induced changes in protein structure and dynamics and how those relate to biological function. Furthermore, the GAG structure is amenable to NMR methods. NMR techniques that are commonly applied to the study of protein-GAG systems include chemical shift perturbation, saturation transfer difference, transferred nuclear Overhauser effect (NOE) spectroscopy, and measurement of NMR effects induced by labeling of proteins or GAGs with paramagnetic probes. Particularly, paramagnetic NMR data provide a complementary source of data to local or ambiguous structural information derived from NOEs and chemical shift perturbations, and, thus, offer unique opportunities to guide molecular modeling. Recent paramagnetic NMR studies on protein-GAG systems are discussed below. When a protein-GAG complex has large size and/or low solubility, solid-state NMR techniques provide further options.

Chemical shift perturbations

Chemical shift perturbation (CSP) (Figure 2B) is perhaps the most commonly used NMR method to investigate how proteins bind GAGs (Pichert et al. 2012; García-Mayoral et al. 2013; Sepuru and Rajarathnam 2019) and is also widely used to study interactions of proteins with each other (Göbl et al. 2014) and with other types of ligands (Aguirre et al. 2015). The experiment is typically carried out with 15N-labeled protein to which increasing amounts of GAG are titrated. A 15N-edited heteronuclear single quantum coherence (HSQC) spectrum is recorded, from which the backbone amide hydrogen and nitrogen chemical shifts of each amino acid residue (except proline) can be obtained. The location of the GAG binding site can be inferred from the largest chemical shift changes, which are usually the consequence of the close proximity between protein and ligand. The magnitude of the CSP is quantified with an empirical relationship (Δδ = [(Δδ H )2 + (Δδ N /5)2]1/2) and can also be used to determine the binding affinity of the protein-ligand complex.

Saturation transfer difference

The saturation transfer difference (STD) experiment (Figure 2D) can identify the specific groups on the GAG that are involved in the interaction with a protein. It is performed at excess concentration of ligand and relies on the fact that for weakly interacting ligands (10−8 mol/L < K D < 10−3 mol/L) there is a rapid exchange between the bound and the free state which potentiates the observed STD signal. The method involves recording an NMR spectrum of the ligand obtained by selectively saturating the protein (on-resonance spectrum), and subtracting it from a spectrum recorded without protein saturation (off-resonance spectrum) (Walpole et al. 2019). Saturation is spread across the protein and transferred to the bound ligand via spin diffusion through the nuclear Overhauser effect resulting in a decrease in the signal intensity of ligand hydrogens that are in close distance to the protein surface. Thus, in the difference spectrum only signals of hydrogens that receive saturation are visible, whereas signals of ligands, which do not bind to the protein and receive no saturation, are canceled out. Saturation transfer is distance-dependent, and, from the observed STD effects, ligand binding epitopes can be deduced. This has been used to investigate GAG specificity of several GAG-binding proteins (Munoz-Garcia et al. 2015; Yu et al. 2014). Moreover, when the structure of the protein-ligand complex is known, complete relaxation and conformational exchange matrix (CORCEMA) theory can be applied to simulate STD NMR spectra and extract the influence of the binding site geometry on the saturation transfer (Jayalakshmi and Krishna 2002). This approach has been used to validate protein-GAG structural models by quantitatively comparing experimental and predicted STD effects (Angulo and Nieto 2011; García-Jiménez et al. 2019). An illustration for the determination of STD effects in HP bound to the protein IL-10 is given in Figure 3A.

Figure 3: Protein-binding regions and structure of HP interacting with IL-10 studied by STD and trNOE NMR experiments. (A) 2D 1H-13C HSQC STD and reference spectra of HP tetrasaccharide (dp4) recorded at a saturation time of 3 s and a 50-fold excess of ligand using natural abundance of 13C. The strength of the STD effect of individual protons is indicated in the HP structure model on the right-hand side. (B) Left: NOE cross peaks involving proton H1 of the internal IdoA,2S residue of HP dp4 in the IL-10-bound state. No NOEs were observed in the absence of IL-10. trNOEs are depicted schematically by gray arrows in the chemical structure below the spectrum. Right: HP dp4 structure models obtained from trNOE data by simulated annealing MD simulations. Residues are labeled with capital letters starting at the 4,5Δ-uronic acid ring at the non-reducing end. Functional groups of ring A and hydrogen atoms are not shown for clarity. The NMR spectra in (A) and the image in (B) were reproduced with permission from Künze et al. (2014). © Oxford University Press.
Figure 3:

Protein-binding regions and structure of HP interacting with IL-10 studied by STD and trNOE NMR experiments. (A) 2D 1H-13C HSQC STD and reference spectra of HP tetrasaccharide (dp4) recorded at a saturation time of 3 s and a 50-fold excess of ligand using natural abundance of 13C. The strength of the STD effect of individual protons is indicated in the HP structure model on the right-hand side. (B) Left: NOE cross peaks involving proton H1 of the internal IdoA,2S residue of HP dp4 in the IL-10-bound state. No NOEs were observed in the absence of IL-10. trNOEs are depicted schematically by gray arrows in the chemical structure below the spectrum. Right: HP dp4 structure models obtained from trNOE data by simulated annealing MD simulations. Residues are labeled with capital letters starting at the 4,5Δ-uronic acid ring at the non-reducing end. Functional groups of ring A and hydrogen atoms are not shown for clarity. The NMR spectra in (A) and the image in (B) were reproduced with permission from Künze et al. (2014). © Oxford University Press.

Transferred-NOE

Information on the bound conformation of GAGs can be obtained through the analysis of transferred nuclear Overhauser effects (trNOEs) (Gao et al. 2016, 2018; Künze et al. 2014) (Figure 2C). This method relies on a change in the molecular tumbling rate of GAG oligosaccharides upon binding to protein. The rotational correlation time of small- and medium-sized GAGs (disaccharides [dp2] to hexasaccharides [dp6]) lies in the picosecond range but approaches several nanoseconds for the protein-GAG complex. Consequently, the sign of the associated NOEs also changes: it is negative and usually weak for free GAGs but positive and reasonably strong for protein-bound GAGs. For successful measurement of trNOEs, the kinetic on/off rates of the GAG-protein interaction must be fast compared to the spin relaxation rates, so that NOEs build up by the GAG ligand in its bound state can still be detected when the GAG gets dissociated from the protein. Through careful adjustment of the experimental conditions such as temperature, magnetic field strength, and ionic strength, the NOEs of the free-state GAG can be minimized, therefore allowing conversion of trNOEs into hydrogen pair distances of the bound state, which can be used as restraints in modeling. Exemplary trNOE spectra and the resulting structures of HP bound to IL-10 are shown in Figure 3B.

Paramagnetic NMR effects

Increasingly more NMR studies of protein-GAG systems make use of paramagnetic labels (Deshauer et al. 2015; Köhling et al. 2016; Künze et al. 2016; Morgan and Wang 2013; Moure et al. 2018; Park et al. 2016). NMR methods involving paramagnetic labels enable the detection of even weak protein-GAG interactions and characterization of their structural details. In addition, the position and orientation of GAG ligands in protein-GAG complexes can be determined more precisely. All paramagnetic labels cause line-broadening of the NMR signal due to paramagnetic relaxation enhancement (PRE) (Figure 2B). For nitroxide radicals and metal ions with a symmetric g-tensor (Mn2+, Cu2+, Gd3+), the predominant contribution to the PRE is dipole-dipole relaxation, which exhibits a 1/r 6 distance dependence, therefore permitting PRE data to be used quantitatively as distance restraints. Metal ions with an asymmetric g-tensor such as lanthanides additionally produce distance- and orientation-dependent changes of the chemical shifts proportional to 1/r 3, referred to as pseudocontact shifts (PCSs, Figure 2C), and induce partial alignment of the GAG or protein in the magnetic field so that residual dipolar couplings (RDCs) can be observed (Otting 2010). PCSs and RDCs offer additional valuable structure information about the binding modes of GAGs and proteins.

GAGs can be functionalized at their reducing ends with paramagnetic labels. For example, Deshauer and coworkers introduced TEMPO ((2,2,6,6-tetramethylpiperidin-1-yl)oxidanyl) into CS dp6 by reductive amination at the anomeric carbon, which allowed the detection of PRE effects in the chemokine CCL5 (RANTES) (Deshauer et al. 2015). Using two types of TEMPO-labeled CS dp6 ligands (denoted CS444 and CS644), which differed in the position of the sulfate group at the non-reducing GalNAc residue, varying PRE patterns were observed. This indicated that the GAG binding orientation is highly dependent on the sulfation pattern and that multiple binding modes can coexist in case of CS644. Morgan and Wang used TEMPO-labeled HP dp6 and measured residue-specific PREs in decorin-binding protein A (DBPA), a GAG-binding cell surface lipoprotein from Borrelia burgdorferi (Morgan and Wang 2013). The PRE data provided confirmation that the reducing end of the HP fragment is located close to a BXBB (B, basic amino acid; X, hydrophobic amino acid) motif in the linker connecting helices 1 and 2 in DBPA. A different synthetic strategy was chosen by Köhling et al., who attached a metal ion-EDTA tag to the reducing end of a nonasulfated HA dp4 by copper(I)-catalyzed cycloaddition (Köhling et al. 2016). The EDTA tag was loaded with either Mn2+ or Cu2+ ions, which induced pronounced PREs in the protein IL-10.

These approaches are complemented by strategies where the paramagnetic tag is attached to the protein side (Joss and Häussinger 2019). Kurzbach and coworkers applied site-directed labeling of cysteines with MTSL to the protein osteopontin (OPN) (Kurzbach et al. 2014). PRE data in combination with 15N relaxation data indicated an expansion of the central core region and increased flexibility of OPN in the presence of its natural HP ligand consistent with an “unfolding-upon-binding” event. This partial unfolding of OPN has been suggested to be primarily a consequence of electrostatic interactions between HP and charged patches along the protein backbone. Other NMR studies took advantage of the unique magnetic properties of lanthanide ions, which could be introduced into GAG binding proteins with the help of genetically encoded lanthanide binding peptide (LBP) tags. Gao et al. engineered a LBT tag into the immunoglobulin (Ig) 1 domains of Roundabout 1 (Robo1) (Gao et al. 2016) and Leukocyte common antigen-related (LAR) protein (Gao et al. 2018), respectively. PCSs could be observed for Robo1 and its HS ligand as pronounced diagonal shifts in their NMR peak positions when comparing the superimposed spectra recorded with paramagnetic Tm3+ and diamagnetic Lu3+. A model of the Robo1-HS complex was generated using the PCS and all other NMR data in a constrained docking calculation with HADDOCK. Künze et al. introduced a LBP tag to the C-terminus of IL-10. This proved to be a sufficiently rigid attachment such that pronounced PCSs and RDCs could be observed for a total of four different lanthanide ions (Künze et al. 2016).

Recently, a novel methodology for the site-specific nitroxide spin labeling of glycoproteins has been demonstrated by Moure et al. (2018). The strategy relies on the presence of a single N-glycosylation site, which is present in many proteins or can be engineered by mutational elimination of all but one glycosylation site. Recombinant expression in HEK293S (GnT1-) cells, which are deficient in N-acetylglucosaminyltransferase 1, results in a high mannose-type N-glycan which can be trimmed by enzymatic digestion with endoglycosidase F1 to leave a single GlcNAc residue. This can in turn be modified by enzymatic addition of a GalNAz residue bearing an azido group that is subject to reaction with an alkyne-carrying TEMPO spin label using copper(I)-catalyzed click chemistry. This methodology was successfully applied to the two-domain construct of Robo1 and allowed the measurement of PREs for Robo1 and an interacting HS dp4 ligand (Moure et al. 2018).

Solid-state NMR methods

Although protein-ligand interaction is a classical field of solution NMR applications, its use is restricted to complex sizes that do not exceed molecular weights of a few tens of kDa. In situations where protein-ligand complexes do not tumble isotropically as in the case of membrane proteins, solid-state NMR techniques represent the only choice (Yang et al. 2018). So far, however, no such investigations have been reported for protein-GAG complexes. However, GAGs play an important role in inducing amyloid formation. Amyloids are fibrillary aggregates of small peptides or proteins showing a canonical cross-β structure with high relevance especially for neuronal diseases (Tycko 2006). The interaction of GAGs with amyloid β (Aβ) fibrils has been studied using solid-state NMR (Stewart et al. 2016). While specific technology such as magic-angle spinning and high power 1H decoupling has to be applied in order to achieve reasonable resolution in 13C NMR (Huster 2005), the GAG-protein interaction largely relies on CSP measurements (Madine et al. 2012). Using 13C-labeled GAG, information on the molecular dynamics of a HP analog bound to Aβ fibrils has been obtained and GAG-protein molecular contacts could be measured (Stewart et al. 2016).

As GAGs and in particular proteoglycans are important components of the ECM of various tissues, solid-state NMR has helped understanding in particular the dynamic features of these polysaccharides in cartilage (Naji et al. 2000), bone (Wise et al. 2007), and artificially engineered biological tissue (Scheidt et al. 2010).

Investigation of protein-GAG complexes by molecular modeling

Many challenges in modeling protein-GAG interactions still remain despite the aforementioned successes achieved in theoretical studies. First of all, the available structural data on protein-GAG complexes in the PDB are limited to fewer than a hundred structures, which practically disallows application of machine learning techniques that are very effective creating novel tools for a particular class of molecules (Kinnings et al. 2011). Furthermore, GAGs are highly charged molecules and, therefore, electrostatics-driven and solvent-mediated interactions should be treated properly (Samsonov and Pisabarro 2016; Sankaranarayanan et al. 2018; Teyra et al. 2011). Similarly, proper structural data on the divalent ions in protein-GAG complex interfaces could be crucial for the use of in silico approaches (Kogut et al. 2021). GAG periodicity, linearity, dynamic flexibility and as a consequence often equivalent disposition of the functional groups in relation to the GAG ends renders it difficult to calculate the energetically most favorable GAG binding pose (Forster and Mulloy 2006; Sankaranarayanan et al. 2018) resulting in the co-existence of energetically comparable multiple binding modes (Atkovska et al. 2014; Joseph et al. 2015; Penk et al. 2019; Potthoff et al. 2019). In particular, antiparallel orientations of the GAG molecule in the binding site, referred as binding polarity, corresponding to comparable binding free energy values are very challenging to distinguish (Bojarski and Samsonov 2021). In addition, the fact that GAGs in the ECM are long heterogeneous polymers hampers the application of computational approaches developed for short peptides or small molecules. This motivated the development of coarse-grained models of GAGs (Bathe et al. 2005; Jana et al. 2018; Kolesnikov et al. 2014; Samsonov et al. 2015, 2019a; Sattelle et al. 2013; 2015). A coarse-grained approach allowed, for example, to characterize spatiotemporal organization of full-size proteoglycans in the ECM (Sattelle et al. 2015). Another important issue for the specificity of GAG interactions (Canales et al. 2006), bioactivity (Guerrini et al. 2008) and the thermodynamics of protein-GAG interactions (Samsonov and Pisabarro 2013) is the pucker conformational space of IdoA derivatives, which are components of HP and HS. Because its proper sampling is achieved only on μs time scales (Bojarski et al. 2019; Sattelle et al. 2010, 2015), rigorous treatment of IdoA rings is computationally expensive. Finally, there is a central question of how specific protein-GAG interactions are, and if computational methods are capable to discern this specificity. For several systems, alterations in the GAG type and sulfation pattern changes their protein binding specifically, whereas for others, only net GAG charge affects the strength of such interactions rendering them purely electrostatics-driven and rather unspecific. However, the very concept of the protein-GAG interaction specificity is not trivial and often ambiguous.

In general, the modeling pipeline to study protein-GAG interactions could be presented as following: (i) GAG binding regions prediction (Figure 2B); (ii) GAG binding poses calculations by molecular docking (Figure 2C); (iii) conformational and energetic analysis of the protein-GAG complexes by molecular dynamics (MD) simulations (Figure 2D). NMR data can be beneficially implemented at any of the steps listed within this modeling pipeline. First, the binding region predictions could be restricted only to the patches on the protein surface including the residues for which changes in the chemical shifts upon a GAG binding are significant. This would decrease the volume in which docking sampling is performed and, therefore, both speed up the calculations and allow for the use of more rigorous sampling parameters. At the second step, only those docking poses that fit best to CSPs and STD data, which can be used as ambiguous contact restraints, could be analyzed further by the MD approaches, while the rest of the binding poses could be discarded. NOE- and PRE-derived atom pair distance restraints as well as PCSs and RDCs could be also treated explicitly as additional score terms in scoring functions that compare experimental data and predicted values calculated from multiple trial structures. Finally, at the third step, the computed structures and their free energy characteristics could be again compared with the NMR data to find out the best matching solutions.

Since protein-GAG interactions are mainly electrostatically-driven (Imberty et al. 2007; Samsonov and Pisabarro 2016), potential GAG binding regions may be predicted based on protein sequential motifs (Hileman et al. 1998) or spatial patches of positively charged residues on the protein surface (Bitomsky and Wade 1999; Kuhn et al. 1990) (Figure 2B). However, such predictions can be ineffective if the same protein binds GAGs in different binding sites for the same (Murphy et al. 2007) or for different GAG types (Sage et al. 2013), or if a single point mutation essentially influences the localization of a GAG binding site (Nordsieck et al. 2012). The study performed on the complete dataset of protein-GAG structures available in the PDB suggested that calculation of electrostatic potential isosurfaces by Poisson-Boltzmann surface area (PBSA) method represents a powerful approach for the prediction of GAG binding regions (Samsonov and Pisabarro 2016). Nevertheless, in case the binding site of a GAG on a protein is known, it is still challenging to dock a GAG (Figure 2C). The reasons for this are: (i) positively charged residues binding GAGs have long side-chains contributing to the higher number of essential degrees of freedom to be considered in modeling the interactions (Möbius et al. 2013); (ii) GAG ligands are poorly complementary to their protein receptors (Taroni et al. 2000); (iii) water molecules play a crucial role for GAG binding and therefore should not be neglected (Samsonov et al. 2011); (iv) GAGs are usually very long (Gama et al. 2006; Rabenstein 2002), while reliable docking is feasible only for oligomers of limited length. Fortunately, there is evidence that oligomeric GAGs represent essential protein binding units (García-Jiménez et al. 2017; Samsonov and Pisabarro 2016; Stewart et al. 2016). (v) There are substantially fewer specific docking tools developed for GAGs than for other classes of biomolecules. Evaluations of 14 docking programs performance for protein-GAG systems showed that only Autodock 3 (AD3) (Morris et al. 1999) yielded good-quality results (Samsonov and Pisabarro 2016; Uciechowska-Kaczmarzyk et al. 2019), though it potentially yields wrong conformations of the glycosidic linkages (Nivedha et al. 2014). Although heparin parameters are explicitly available at the ClusPro server (Kozakov et al. 2017), this software did not overperform other programs. The application of targeted MD (dynamic molecular docking: DMD) was shown to be beneficial and comparable to AD3 in terms of its performance for docking GAGs since it takes into account full flexibility of both receptor and ligand molecules in explicit solvent (Samsonov et al. 2014). In their work, Griffith et al. (Griffith et al. 2017) developed GAG-Dock, a GAG-specific method which allowed to successfully dock several protein-GAG systems, which however, did not represent complicated targets for other docking approaches either (Samsonov and Pisabarro 2016; Uciechowska-Kaczmarzyk et al. 2019). The docking limitation of the length of GAG ligands was overcome by the fragment-based approach for GAG docking, in which GAG dp3 were docked and then assembled into a longer GAG chain (Samsonov et al. 2019b). Recently, a novel methodology originally proposed to dock protein-protein complexes using replica exchange MD with repulsive scaling (Siebenmorgen et al. 2020) was successfully applied to a dataset of protein-GAG complexes (Maszota-Zieleniak et al. 2021). This approach, where effective pairwise van der Waals radii are increased in different Hamiltonian replicas, allows for full flexibility of both receptor side chains and ligand, provides a reliable scoring scheme based on molecular mechanics Poisson–Boltzmann surface area (MM-PBSA) free energy calculations (Kollman et al. 2000; Srinivasan et al. 1998) and is independent of the GAG ligand length rendering it feasible to evaluate the role of GAG binding sub-sites and prediction of GAG minimal binding units.

Considering all the docking challenges, implementation of experimental data-based restraints is beneficial (Hofmann et al. 2015; Seyfried et al. 2007). To summarize, more specific docking approaches to treat GAG containing systems should be developed by mutual integration of basic physico-chemical principles, established and novel methodologies, and experimental data available for these systems. In the third step of protein-GAG modeling, MD is applied to further study the complexes obtained from the experiment or molecular docking (Figure 2D). An excellent review on the applicability of different force fields to carbohydrates in general and GAGs, in particular, can be found elsewhere (Xiong et al. 2015). However, it is important to keep in mind that due to the highly charged nature of GAGs, the performance of standard MD-based approaches and free energy calculation techniques should be always re-evaluated and optimal parameter sets should be derived and calibrated, respectively. Again, the best way to do this is to compare and to combine modeling with the experimental data such as from NMR. The examples of how this was accomplished in our studies for IL-10, CXCL12, and CXCL14 cytokines are provided in the next section.

Examples

IL-10

Interleukin-10 (IL-10) is a key regulatory cytokine in the immune system with both pro- and anti-inflammatory functions (Sabat et al. 2010). It inhibits the synthesis of inflammatory cytokines (Fiorentino et al. 1989), but has also stimulatory effects on certain immune cells (Levy and Brouet 1994). Because of its broad array of functions, deregulation of IL-10 activity is associated with imbalanced immune reactions e.g. forms of inflammation such as enteritis (Kühn et al. 1993), asthma (Tournoy et al. 2000), and allergy (Grünig et al. 1997), or cases of autoimmunity such as systemic lupus erythematodes (Llorente et al. 1995). From a structural perspective, IL-10 forms a homodimer, which is created by intertwining the α-helical bundles of two identical protein chains (Zdanov et al. 1995).

When secreted from cells, IL-10 can interact with components of the ECM. Binding of IL-10 to GAGs was demonstrated experimentally (Salek-Ardakani et al. 2000). In that study, sulfated cell-surface GAGs were necessary to trigger the full mitogenic activity of IL-10 on monocytes and macrophages, whereas soluble GAGs inhibited the IL-10-induced expression of CD16 and CD64 by these cells (Salek-Ardakani et al. 2000). The molecular mechanisms underlying these effects are unclear. As starting point to better understand the biological consequences of IL-10’s interaction with GAGs, we characterized the binding properties of GAGs and determined their binding region in IL-10 by NMR and modeling (Gehrcke and Pisabarro 2015; Köhling et al. 2016; Künze et al. 2014, 2016).

Using STD NMR, we first determined which chemical groups of a series of HP and CS dp2 molecules are proximal to the IL-10 surface in binding (Figure 3A). We then extended this approach to a natural abundance 13C-edited 2D STD HSQC experiment which allowed the detection and unambiguous assignment of STD signals for a longer HP dp4 molecule (Künze et al. 2014). The dissociation constants of HP and CS dp2 molecules with varying patterns of sulfation, and of differently long HP ligands (dp2–dp10) were determined using the initial growth rate approach of the STD amplification factor (Angulo and Nieto 2011). Binding affinity increased with GAG chain length and number of sulfate groups, but no predominant role of one type of O- or N-sulfation was observed. Thus, while this result points to the crucial role of electrostatic forces in IL-10-GAG interaction, all sulfate groups in small HP and CS dp2 ligands seem equally important in establishing those interactions.

We obtained structural information for HP dp4 bound to IL-10 through the analysis of trNOEs (Figure 3B), and compared bound and free-state conformations (Künze et al. 2014). The free HP dp4 showed only a handful of weak NOEs, and distance data were obtained from ROESY experiments. The presence of IL-10 led to the appearance of multiple strong cross-peaks with negative amplitude in the NOESY spectrum of HP dp4, which clearly indicated an interaction with IL-10, because the effective correlation time of the ligand had obviously increased to produce negative NOE signals. Analysis of trNOEs provided a set of hydrogen pair distance data, which was used to create structural models of HP dp4 by simulated annealing MD calculations. Comparison of the structural models of HP dp4 revealed that the overall backbone conformation is slightly changed in the presence of IL-10. The conformational equilibrium between 1C4 chair and 2SO skew-boat conformation of the internal IdoA residue is also slightly shifted but maintained between the free and bound form. This result is not unexpected as in the high-affinity complex between bFGF and HP (Faham et al. 1996) the backbone differences relative to the free-state HP structure (Mikhailov et al. 1996) are also rather small and provide no indication about the interaction strength. A longer HP dp6 molecule was also tested, but this ligand produced negative NOE signals in both the free and bound state, which prohibited its structural characterization by the trNOE approach.

The GAG binding region in IL-10 could be determined by NMR and computational approaches (Gehrcke and Pisabarro 2015; Köhling et al. 2016; Künze et al. 2016) (Figure 4). First, 1H/15N CSPs of IL-10 in the presence of different types of GAGs (HA, CS, DS, HP, and nonasulfated HA) were measured to narrow down the binding region (Figure 4A and D). The CSPs induced by these GAGs were localized to the N-terminus, helix B, loop BC, loop DE, and the C-terminus. In the 3D structure, these residues cluster in or around the central crevice formed between the two dimer subunits of IL-10, which pointed to this region as the GAG binding interface. However, since the CSP data provided a rather low-resolution binding model, we sought to obtain additional structural information about the mode of interaction between IL-10 and GAGs. To this end, we performed measurements with a lanthanide binding tag and collected PCS NMR data for IL-10 and a HP dp4 ligand (Figure 4B and D). The C-terminal α-helix of IL-10 was extended by a lanthanide binding peptide sequence (Nitz et al. 2003), which enabled rigid attachment of lanthanides to IL-10 as proved by the large magnitudes of PCSs and RDCs. A total of 537 PCSs for IL-10 and 67 PCSs for HP dp4 were collected and served as restraints in HP docking using Xplor-NIH (Schwieters et al. 2003). Best docking poses were then refined by MD in NAMD (Phillips et al. 2005) using the CHARMM force field for HP dp4 (MacKerell et al. 1998), which confirmed stability of the structures obtained by docking and provided a model for GAG binding. In the simulations, HP dp4 interacted with a cluster of basic amino acid residues on helix D and E including K99, R102, R104, R106, R107, K117, and K119. Noteworthy, residues 101–108 (LRMRLRRC) within the identified cluster match a typical Cardin-Weintraub consensus sequence for a HP-binding motif (Cardin and Weintraub 1989). Binding of GAGs to this region in IL-10 was also predicted by in silico experiments (Gehrcke and Pisabarro 2015). Gehrcke and Pisabarro developed models of IL-10 complexed with HP dp4, HP dp6, CS4 dp6, CS6 dp6, HA dp4, HA dp6 using the DMD approach (Samsonov et al. 2014). Analysis of these models suggested two different GAG binding orientations on the surface of the protein. The simulations performed for R107A mutant proposed that this particular residue is especially essential for IL-10’s interactions with GAGs. From the methodological point of view, this work of Gehrcke et al. further contributed to the evaluation of the DMD approach performance for protein-GAG systems demonstrating its advantages over classical docking approaches.

Figure 4: Location of the GAG binding site in IL-10 determined by CSP analysis, paramagnetic NMR, and molecular modeling. (A) Left: 1H-15N HSQC NMR spectra of IL-10 with increasing concentrations of HP tetrasaccharide (dp4). CSPs of selected amino acid residues are highlighted. Right: 1H/15N CSPs of IL-10 at 200 mol% HP dp4 plotted vs. the IL-10 residue number. The α-helix regions of IL-10 are indicted in green. (B) Detection of PCSs in IL-10 and its HP dp4 ligand induced by a protein-attached lanthanide binding peptide. Left: Sections of 1H-15N HSQC spectra of IL-10 and 1H-13C HSQC spectra of HP dp4 in the presence of diamagnetic (Lu3+) or paramagnetic lanthanides (Tb3+, Tm3+, Dy3+). Right: Position and Δχ-tensor of Tb3+ calculated from PCS data of IL-10. The Δχ-tensor is represented as isosurfaces corresponding to PCS values of ±0.75 and ±0.20 ppm, respectively. (C) Detection of PREs in IL-10 induced by a metal-EDTA-labeled nonasulfated HA (9S-HA-EDTA) dp4 ligand. Sections of 1H-15N HSQC spectra of IL-10 in the presence of 9S-HA-EDTA that was loaded with diamagnetic Ca2+ (black) or paramagnetic Mn2+ (red). NMR signals that were quenched by Mn2+ are labeled with their corresponding amino acid residue. The chemical structure of the linker-EDTA moiety that was conjugated to 9S-HA is shown above the spectrum. (D) IL-10-GAG binding models derived from CSP, PCS, and PRE NMR analyses. Left: Mapping of CSPs onto the IL-10 backbone structure was used to narrow down the GAG binding region. The initial GAG binding model was refined by PCS-restrained docking of HP to IL-10 (middle), and further confirmed by determination of the metal-ion position of Mn2+-labeled 9S-HA ligand (right). Data shown in this Figure were originally published in Künze et al. (2016). © The American Society for Biochemistry and Molecular Biology.
Figure 4:

Location of the GAG binding site in IL-10 determined by CSP analysis, paramagnetic NMR, and molecular modeling. (A) Left: 1H-15N HSQC NMR spectra of IL-10 with increasing concentrations of HP tetrasaccharide (dp4). CSPs of selected amino acid residues are highlighted. Right: 1H/15N CSPs of IL-10 at 200 mol% HP dp4 plotted vs. the IL-10 residue number. The α-helix regions of IL-10 are indicted in green. (B) Detection of PCSs in IL-10 and its HP dp4 ligand induced by a protein-attached lanthanide binding peptide. Left: Sections of 1H-15N HSQC spectra of IL-10 and 1H-13C HSQC spectra of HP dp4 in the presence of diamagnetic (Lu3+) or paramagnetic lanthanides (Tb3+, Tm3+, Dy3+). Right: Position and Δχ-tensor of Tb3+ calculated from PCS data of IL-10. The Δχ-tensor is represented as isosurfaces corresponding to PCS values of ±0.75 and ±0.20 ppm, respectively. (C) Detection of PREs in IL-10 induced by a metal-EDTA-labeled nonasulfated HA (9S-HA-EDTA) dp4 ligand. Sections of 1H-15N HSQC spectra of IL-10 in the presence of 9S-HA-EDTA that was loaded with diamagnetic Ca2+ (black) or paramagnetic Mn2+ (red). NMR signals that were quenched by Mn2+ are labeled with their corresponding amino acid residue. The chemical structure of the linker-EDTA moiety that was conjugated to 9S-HA is shown above the spectrum. (D) IL-10-GAG binding models derived from CSP, PCS, and PRE NMR analyses. Left: Mapping of CSPs onto the IL-10 backbone structure was used to narrow down the GAG binding region. The initial GAG binding model was refined by PCS-restrained docking of HP to IL-10 (middle), and further confirmed by determination of the metal-ion position of Mn2+-labeled 9S-HA ligand (right). Data shown in this Figure were originally published in Künze et al. (2016). © The American Society for Biochemistry and Molecular Biology.

To further explore the applicability of paramagnetic NMR techniques, we also studied the IL-10-GAG system using a paramagnetically labeled GAG (Figure 4C and D). An EDTA moiety was attached to nonasulfated HA dp4 via click chemistry and loaded with Mn2+ and Cu2+ which produced pronounced PREs in IL-10 (Köhling et al. 2016). The position of the paramagnetic center could be determined with high precision from the PRE data and served as starting point for model construction. The binding model of sulfated HA dp4 was very similar to the model of HP dp4 proposed by (Gehrcke and Pisabarro 2015), and agreed well with the GAG binding region determined by CSP and PCS NMR analyses (Künze et al. 2016), which demonstrates in a convincing way the complementary nature of NMR and modeling approaches.

CXCL12

Stromal cell-derived factor 1α (CXCL12) is a CXC-type chemokine that is involved in a broad array of physiological processes including the chemotaxis and extravasion of neutrophils into sites of infection (Kucia et al. 2004), mobilization and directed migration of stem cells (Kucia et al. 2005), HIV-1 infection (Arenzana-Seisdedos 2015), and cancer (Teicher and Fricker 2010). CXCL12 shares a common structure with other chemokines composed of a three-stranded β-sheet followed by an α-helix, which is stabilized by typically two conserved disulfide bonds (Miller and Mayo 2017). The ability of chemokines like CXCL12 to interact with cell surface GAGs is essential to their function of directing cell migration (Proudfoot 2006), and has motivated a number of structural studies on the interaction of CXCL12 with HP and HS oligosaccharides (Amara et al. 1999; Laguri et al. 2011; Murphy et al. 2007; Sadir et al. 2004; Ziarek et al. 2013). These investigations led to the identification of two HP-binding regions in CXCL12 with putatively varying affinities (Murphy et al. 2007; Ziarek et al. 2013). A high-affinity HP binding region (HAHBR) is localized to residues in the first β-strand (K24, H25, K27) and residues flanking that region in the 3D structure (R41, K43, R47). In the crystal structure of CXCL12 with HP dp2 a second putatively low affinity HP binding region (LAHBR) was observed (Murphy et al. 2007). We complemented these earlier studies by investigating the interaction of CXCL12 with other GAGs including CS, HA, desulfated HP (deHP), HP, and an artificial persulfated HA (pHA) which could have application as biomaterial (Tiwari and Bahadur 2019). Interaction data for these GAGs also help to resolve inconsistent conclusions obtained previously about the binding orientation of GAGs in complex with CXCL12 (Laguri et al. 2011; Sadir et al. 2004; Ziarek et al. 2013).

We monitored binding of HP, pHA, CS, and HA to CXCL12 by 1H-15N HSQC NMR, and obtained atomically detailed information on that interaction by comparison with docking and MD simulation (Panitz et al. 2016) (Figure 5). For NMR measurements, we used a high protein concentration (500 μM) to assure that CXCL12 formed a dimer (Veldkamp et al. 2005). Hence, chemical shift changes of CXCL12 reflected primarily GAG binding, whereas contributions from CXCL12 self-association were deemed insignificant. HP dp6 caused the largest CSPs, followed by pHA dp4, and CS dp6. HA dp4 failed to induce significant CSPs showing that it did not bind CXCL12. The CSPs induced by HP dp6 were mostly confined to the β1 strand (A21, V23, K24, H25, K27) and the β2 strand (A40, R41, L42, K43, N45), consistent with these residues forming the HAHBR in CXCL12. Additionally, large CSPs were localized to the short 310 helix before β1 (V18, R20, A21), the loop after β3 (K54), and the C-terminal α-helix (W57, K64, A65). In the 3D structure, these residues form a cluster that corresponds to the LAHBR observed in the crystal structure of CXCL12 with HP dp2 (Murphy et al. 2007). Interestingly, large CSPs were also observed for the N-terminal CXCR4 binding motif (R12, F14, S16), suggesting that the HP interface partially overlaps with the receptor binding site of CXCL12.

Figure 5: NMR and molecular modeling studies of CXLC12 interacting with GAGs. (A) X-ray structure (PDB ID: 2NWG) of CXCL12 dimer (in surface) in complex with two HP disaccharides (in sticks). High and low affinity HP binding sites (HAHBR and LAHRB, respectively) are labeled. The arrow shows the potential direction of a longer HP binding pose connecting two binding sites. (B) Positively charged amino acid residues, R and K, are shown in blue and yellow surface, respectively. (C) CXCL12 (in surface) in complex with HP dp6 (in sticks) as predicted by the modeling. (D) 1H/15N CSPs of CXCL12 in the presence of HP dp6 plotted versus the CXCL12 residue number. The blocks of amino acid residues corresponding to HAHBR and LAHBR are highlighted by boxes. (E) Sections of 1H-15N HSQC NMR spectra of CXCL12 with increasing concentrations of HP dp6 for the amino acid residues that showed the highest CSPs upon HP binding. (F) The same amino acid residues are shown in red on the CXCL12 surface (transparent, the secondary structure is in cartoon representation) and labeled. The graph in (D) and the spectra in (E) were reproduced with permission from Panitz et al. (2016). © Oxford University Press.
Figure 5:

NMR and molecular modeling studies of CXLC12 interacting with GAGs. (A) X-ray structure (PDB ID: 2NWG) of CXCL12 dimer (in surface) in complex with two HP disaccharides (in sticks). High and low affinity HP binding sites (HAHBR and LAHRB, respectively) are labeled. The arrow shows the potential direction of a longer HP binding pose connecting two binding sites. (B) Positively charged amino acid residues, R and K, are shown in blue and yellow surface, respectively. (C) CXCL12 (in surface) in complex with HP dp6 (in sticks) as predicted by the modeling. (D) 1H/15N CSPs of CXCL12 in the presence of HP dp6 plotted versus the CXCL12 residue number. The blocks of amino acid residues corresponding to HAHBR and LAHBR are highlighted by boxes. (E) Sections of 1H-15N HSQC NMR spectra of CXCL12 with increasing concentrations of HP dp6 for the amino acid residues that showed the highest CSPs upon HP binding. (F) The same amino acid residues are shown in red on the CXCL12 surface (transparent, the secondary structure is in cartoon representation) and labeled. The graph in (D) and the spectra in (E) were reproduced with permission from Panitz et al. (2016). © Oxford University Press.

The GAGs pHA dp4 and CS6 dp6 induced similar patterns of CSPs relative to HP dp6 indicating that these GAGs interact with CXCL12 primarily via the HAHBR. However, while HP and pHA led to CSPs in the CXCR4 binding motif, CS caused CSPs of residues next to the CXCR4 motif (V18, A19, R20), suggesting that these GAGs have different modes of interaction with this area in CXCL12.

Due to the fact that in the CXCL12 crystal structure two HP dp2 molecules bound to two different binding sites (i.e. HAHBR and LAHBR) are observed (Murphy et al. 2007), it was particularly interesting to find out if the modeling protocols can handle such a system properly and yield results that are in agreement with X-ray and NMR. Most of the docking programs performed well for HP dp2 in both sites (Samsonov and Pisabarro 2016). In addition to the GAGs used in the NMR experiments, we systematically modeled the whole series of dp2, dp4, dp6, dp8 of HA, pHA, HP and deHP to understand how glycosidic linkage type, monosaccharide composition, length, and sulfation can influence GAG binding to CXCL12. We were able to rank binding affinity of the GAGs similarly to the one in NMR experiments, predicted and properly ranked the two binding sites in terms of the most contributing residues. The increase of the GAG chain length led to more favorable binding, while sulfated GAGs bound stronger than unsulfated GAGs, which can be attributed to the net electrostatics effect in this system. At the same time and similar to the NMR data, the calculations showed that despite a higher charge of pHA relative to HP, both GAGs bind with the similar strength. Also, deHP was predicted to bind CXCL12 stronger than HA despite their equal net charges. These results suggest the importance of the glycosidic linkage, sulfation pattern, and monosaccharide components of GAGs for binding, which means also that these interactions are not only driven by the net charge of the complex components but also reveal certain specificity. Furthermore, docking analysis suggested that a long GAG would bind first to the HAHBR and then could be elongated into LAHBR. This finding is methodologically very important since it opens the way to predict the structures of longer GAGs by docking their shorter fragments. Previously, this observation was used in “grafting” procedure for other carbohydrates by the Woods group (Grant et al. 2016) and served as a basis for developing a fragment-based docking approach for GAGs (Samsonov et al. 2019b). The data obtained in this study confirm the relevance of the used docking and MD-based approaches by agreement with the NMR data both for identification of binding sites and ranking of different GAG molecules by the calculated free binding energy (Panitz et al. 2016).

CXCL14

CXCL14, chemokine (C-X-C motif) ligand 14, is a member of the CXC chemokine family. It is expressed at high levels in epithelial tissues and mucosae (Frederick et al. 2000). Current knowledge suggests that CXCL14 plays roles in the immune homeostasis. It is a chemoattractant and activator of monocytes (Sleeman et al. 2000), macrophages (Kurth et al. 2001), and immature dendritic cells (Shellenberger et al. 2004), and can stimulate the migration of activated natural killer cells (Starnes et al. 2006). CXCL14 orchestrates the infiltration of dendritic cells into the epidermis (Shurin et al. 2005), where they differentiate into Langerhans cells. CXCL14 also exhibits broad antimicrobial activity, which contributes to maintaining a healthy skin (Maerki et al. 2009).

CXCL14 is highly positively charged, and HP can block binding of CXCL14 to human monocytic cells (Tanegashima et al. 2010), suggesting an interaction occurs between CXCL14 and GAGs. To deeper understand the binding specificity of CXCL14 for different types of GAGs, we studied this system by 1H-15N HSQC NMR and molecular modeling (Penk et al. 2019) (Figure 6). The CSP method was again used to experimentally determine the GAG binding region of CXCL14. The 1H-15N HSQC spectrum of CXCL14 showed gradual chemical shift changes upon addition of GAGs which indicated that complexes of CXCL14 with GAGs were in fast exchange on the NMR time scale. DS, CS-A/C, and CS-D with lengths dp6 or dp10 caused similar patterns of CSPs. The regions of CXCL14 that were most affected by those GAGs were the C-terminal α-helix, the turn connecting β2 and β3 strands (residues Y44-Q47), and residues N-terminal to the first β-strand (R7, G9, K11-S15). By contrast, a HP dp6 molecule led to a different pattern of CSPs. Residues in the terminal α-helix experienced less pronounced CSPs compared to DS and CS, whereas additional changes occurred for residues in the N-terminus and in β1 and β2 strands. Furthermore, the signals of some residues of CXCL14 (e.g., I12, I36, T37) were even shifted into opposite directions in the spectrum by HP versus DS and CS. The most direct explanation of these results is that the HP binding site of CXCL14 is different to that of DS, CS-D, and CS-A/C. Based on the CSP magnitudes induced by different GAGs of equal chain length, we estimated the order of their apparent binding affinities with CXCL14: HP > DS ∼ CS-D > CS-A/C > HA.

Figure 6: NMR and molecular modeling reveal distinct binding modes for HP (top) and CS-D (bottom) with CXCL14. (A) 1H/15N CSPs of CXCL14 in the presence of HP and CS-D dp6 plotted versus the CXCL14 residue number. (B) The amino acid residues on the CXCL14 surface with the highest contributions to binding according to the NMR experiment are shown in red and labeled. (C) CXCL14 (in surface) in complex with HP and CS-D dp6 (in sticks) as predicted by modeling. (D) Schematic representation of the binding poses obtained by modeling: the arrow reflects the direction of the calculated representative GAG binding poses from the reducing to the non-reducing end. The graphs in (A) were reprinted with permission from Penk et al. (2019). © Oxford University Press.
Figure 6:

NMR and molecular modeling reveal distinct binding modes for HP (top) and CS-D (bottom) with CXCL14. (A) 1H/15N CSPs of CXCL14 in the presence of HP and CS-D dp6 plotted versus the CXCL14 residue number. (B) The amino acid residues on the CXCL14 surface with the highest contributions to binding according to the NMR experiment are shown in red and labeled. (C) CXCL14 (in surface) in complex with HP and CS-D dp6 (in sticks) as predicted by modeling. (D) Schematic representation of the binding poses obtained by modeling: the arrow reflects the direction of the calculated representative GAG binding poses from the reducing to the non-reducing end. The graphs in (A) were reprinted with permission from Penk et al. (2019). © Oxford University Press.

We applied our standard protein-GAG complex modeling pipeline to complement and to explain the NMR findings in atomic details. Since CS-A/C dp6 was a heterogeneous mixture of CS with different sulfation patterns, we used eight different dp6 combinations with particular sulfation pattern to model this GAG. Three similar clusters of solutions were observed for all GAGs, while two of them could be interpreted as elongation of each other and corresponded to the most favorable binding energies. These predictions were corroborated by the NMR data according to which GAGs would occupy the region including these two binding sites. The clear overlap of the predicted GAG structures with the most positive electrostatic potential surfaces on the protein molecule suggested that electrostatics in this system represent a driving force for binding. At the same time, the populations of the respective clusters obtained for different GAGs reveal the specificity of the interactions: HP and DS are more probable to occupy both binding sites, while CS-A/C and CS-D prefer just one of them. This again agreed with the experimentally obtained CSP pattern found for these GAG types. The GAGs were ranked in terms of their calculated affinity to CXCL14 as following: HP > DS ∼ CS-D > CS-A/C > HA, which was in full agreement with the NMR results. Furthermore, we analyzed the polarity of binding in terms of the reducing/non-reducing end orientations in the binding pose that was previously suggested to be important for IL-8 and PCPE-1 interactions with GAGs (Pichert et al. 2012; Potthoff et al. 2019). It turned out that for HP and CS-D, the dominant orientations in the adjacent binding sites allow for binding a longer GAG, which would potentially connect these two binding sites, while this is not the case for DS or CS-A/C. This finding supports the specificity of interactions in the system. Also, per residue binding free energy decomposition further reveals the specificity of GAG binding depending on their types. For the series HP, DS, CS-D, CS-A/C per residue binding contribution is shifted from the residues in N-loop towards the C-terminal α-helix, which perfectly reflects the corresponding NMR data. In particular, R72 was found to be crucial for CS-D, CS-A/C, DS but not for HP binding both theoretically and experimentally. The flexibility of N-loop (residues 6–17) for the complexes with different GAGs also shows significant difference for HP (more rigid), DS and CS-D (more flexible).

Conclusions

In spite of some progress over the last decades, the structural biology of protein-GAG complexes remains an underexplored area with only a few tens of structures of such complexes in the PDB. The last years have experienced a lot of progress both in the NMR as well as the computer modeling field. In NMR spectroscopy, in particular paramagnetic probes are increasingly used to obtain structural constraints of longer range that exceed the NOE distance (Köhling et al. 2016; Künze et al. 2016; Moure et al. 2018). Computer modeling has progressed to improve the predictive power of docking approaches combined with subsequent MD simulations-based analysis help understanding the molecular dynamics and free energy properties of the GAG-protein complexes. Also, NMR spectroscopy has unprecedented power to study molecular dynamics in detail (Sekhar and Kay 2019). This great potential of both NMR and in silico approaches has not yet fully been explored and will substantially contribute to the field in the near future. As it is demonstrated in this review on the examples for particular cytokines, overall, the quality of models of GAG-protein complexes with various classes of GAGs determined from the combination of NMR and computational analysis is quite high but the experimental efforts to provide well-resolved and precise structural constraints still represents challenges. As many protein-GAG complexes are too small for efficient cryo-EM studies, X-ray analysis and NMR spectroscopy will remain the only effective experimental tools in structural biology of protein-GAG interactions. Along with cutting-edge computer modeling, a powerful combination of forces will provide the most efficient approach to describe protein-GAG complexes in atomic detail.

Corresponding author: Daniel Huster, Institute for Medical Physics and Biophysics, University of Leipzig, Härtelstr. 16-18, D-04107 Leipzig, Germany, E-mail:

Funding source: Deutsche Forschungsgemeinschaft

Award Identifier / Grant number: SFB-TRR 67, A06

Funding source: ASCRS Research Foundation

Funding source: National Science Centre of Poland

Award Identifier / Grant number: UMO-2018/30/E/ST4/00037

  1. Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

  2. Research funding: This research was funded by the German Research Foundation (DFG, Grant Number 59397982, TRR-SFB 67, subproject A06) and by the National Science Centre of Poland (UMO-2018/30/E/ST4/00037).

  3. Conflict of interest statement: The authors declare no conflicts of interest regarding this article.

References

Afratis, N., Gialeli, C., Nikitovic, D., Tsegenidis, T., Karousou, E., Theocharis, A.D., Pavão, M.S., Tzanakakis, G.N., and Karamanos, N.K. (2012). Glycosaminoglycans: key players in cancer cell biology and treatment. FEBS J. 279: 1177–1197, https://doi.org/10.1111/j.1742-4658.2012.08529.x.Search in Google Scholar PubMed

Aguirre, C., Cala, O., and Krimm, I. (2015). Overview of probing protein-ligand interactions using NMR. Curr. Protoc. Protein Sci. 81, https://doi.org/10.1002/0471140864.ps1718s81.Search in Google Scholar PubMed

Almond, A. (2018). Multiscale modeling of glycosaminoglycan structure and dynamics: current methods and challenges. Curr. Opin. Struct. Biol. 50: 58–64, https://doi.org/10.1016/j.sbi.2017.11.008.Search in Google Scholar PubMed

Amara, A., Lorthioir, O., Valenzuela, A., Magerus, A., Thelen, M., Montes, M., Virelizier, J.L., Delepierre, M., Baleux, F., Lortat-Jacob, H., and Arenzana-Seisdedos, F. (1999). Stromal cell-derived factor-1α associates with heparan sulfates through the first β-strand of the chemokine. J. Biol. Chem. 274: 23916–23925, https://doi.org/10.1074/jbc.274.34.23916.Search in Google Scholar PubMed

Angulo, J. and Nieto, P.M. (2011). STD-NMR: application to transient interactions between biomolecules-a quantitative approach. Eur. Biophys. J. 40: 1357–1369. https://doi.org/10.1007/s00249-011-0749-5.Search in Google Scholar PubMed

Angulo, J., Ojeda, R., de Paz, J.L., Lucas, R., Nieto, P.M., Lozano, R.M., Redondo-Horcajo, M., Gimenez-Gallego, G., and Martin-Lomas, M. (2004). The activation of fibroblast growth factors (FGFs) by glycosaminoglycans: influence of the sulfation pattern on the biological activity of FGF-1. Chembiochem 5: 55–61, https://doi.org/10.1002/cbic.200300696.Search in Google Scholar PubMed

Arenzana-Seisdedos, F. (2015). SDF-1/CXCL12: a chemokine in the life cycle of HIV. Front. Immunol. 6: 256, https://doi.org/10.3389/fimmu.2015.00256.Search in Google Scholar PubMed PubMed Central

Atkovska, K., Samsonov, S.A., Paszkowski-Rogacz, M., and Pisabarro, M.T. (2014). Multipose binding in molecular docking. Int. J. Mol. Sci. 15: 2622–2645, https://doi.org/10.3390/ijms15022622.Search in Google Scholar PubMed PubMed Central

Ballut, L., Sapay, N., Chautard, E., Imberty, A., and Ricard-Blum, S. (2013). Mapping of heparin/heparan sulfate binding sites on ανβ3 integrin by molecular docking. J. Mol. Recogn. 26: 76–85, https://doi.org/10.1002/jmr.2250.Search in Google Scholar PubMed

Bathe, M., Rutledge, G.C., Grodzinsky, A.J., and Tidor, B. (2005). A coarse-grained molecular model for glycosaminoglycans: application to chondroitin, chondroitin sulfate, and hyaluronic acid. Biophys. J. 88: 3870–3887, https://doi.org/10.1529/biophysj.104.058800.Search in Google Scholar PubMed PubMed Central

Bierbaum, S., Hintze, V., and Scharnweber, D. (2012). Functionalization of biomaterial surfaces using artificial extracellular matrices. Biomatter 2: 132–141, https://doi.org/10.4161/biom.20921.Search in Google Scholar PubMed PubMed Central

Biran, R. and Pond, D. (2017). Heparin coatings for improving blood compatibility of medical devices. Adv. Drug Deliv. Rev. 112: 12–23, https://doi.org/10.1016/j.addr.2016.12.002.Search in Google Scholar PubMed

Bitomsky, W. and Wade, R.C. (1999). Docking of glycosaminoglycans to heparin-binding proteins: validation for aFGF, bFGF, and antithrombin and application to IL-8. J. Am. Chem. Soc. 121: 3004–3013, https://doi.org/10.1021/ja983319g.Search in Google Scholar

Bojarski, K.K., Sieradzan, A.K., and Samsonov, S.A. (2019). Molecular dynamics insights into protein-glycosaminoglycan systems from microsecond-scale simulations. Biopolymers 110: e23252, https://doi.org/10.1002/bip.23252.Search in Google Scholar PubMed

Bojarski, K.K. and Samsonov, S.A. (2021). Role of oligosaccharide chain polarity in protein-glycosaminoglycan interactions. J. Chem. Inf. Model. 61: 455–466, https://doi.org/10.1021/acs.jcim.0c01402.Search in Google Scholar PubMed

Bourin, M.C. and Lindahl, U. (1993). Glycosaminoglycans and the regulation of blood coagulation. Biochem. J. 289: 313–330, https://doi.org/10.1042/bj2890313.Search in Google Scholar PubMed PubMed Central

Canales, A., Lozano, R., Lopez-Mendez, B., Angulo, J., Ojeda, R., Nieto, P.M., Martin-Lomas, M., Gimenez-Gallego, G., and Jimenez-Barbero, J. (2006). Solution NMR structure of a human FGF-1 monomer, activated by a hexasaccharide heparin-analogue. FEBS J. 273: 4716–4727, https://doi.org/10.1111/j.1742-4658.2006.05474.x.Search in Google Scholar PubMed

Cardin, A.D. and Weintraub, H.J.R. (1989). Molecular modeling of protein-glycosaminoglycan interactions. Arteriosclerosis 9: 21–32, https://doi.org/10.1161/01.atv.9.1.21.Search in Google Scholar PubMed

Chen, C.L., Hasan, S.S., Klose, T., Sun, Y., Buda, G., Sun, C., Klimstra, W.B., and Rossmann, M.G. (2020). Cryo-EM structure of eastern equine encephalitis virus in complex with heparan sulfate analogues. Proc. Natl. Acad. Sci. U. S. A. 117: 8890–8899, https://doi.org/10.1073/pnas.1910670117.Search in Google Scholar PubMed PubMed Central

Collins, L.E. and Troeberg, L. (2019). Heparan sulfate as a regulator of inflammation and immunity. J. Leukoc. Biol. 105: 81–92, https://doi.org/10.1002/jlb.3ru0618-246r.Search in Google Scholar PubMed

DeMarco, M.L. and Woods, R.J. (2008). Structural glycobiology: a game of snakes and ladders. Glycobiology 18: 426–440, https://doi.org/10.1093/glycob/cwn026.Search in Google Scholar

Deshauer, C., Morgan, A.M., Ryan, E.O., Handel, T.M., Prestegard, J.H., and Wang, X. (2015). Interactions of the chemokine CCL5/RANTES with medium-sized chondroitin sulfate ligands. Structure 23: 1066–1077, https://doi.org/10.1016/j.str.2015.03.024.Search in Google Scholar

Esko, J.D., Kimata, K., and Lindahl, U. (2009). Proteoglycans and sulfated glycosaminoglycans. In: Varki, A., Cummings, R.D., Esko, J.D., Freeze, H.H., Stanley, P., Bertozzi, C.R., Hart, G.W., and Etzler, M.E. (Eds.), Essentials of glycobiology. New York: Cold Spring Harbor Laboratory Press.Search in Google Scholar

Faham, S., Hileman, R.E., Fromm, J.R., Linhardt, R.J., and Rees, D.C. (1996). Heparin structure and interactions with basic fibroblast growth factor. Science 271: 1116–1120, https://doi.org/10.1126/science.271.5252.1116.Search in Google Scholar

Farrugia, B.L., Lord, M.S., Whitelock, J.M., and Melrose, J. (2018). Harnessing chondroitin sulphate in composite scaffolds to direct progenitor and stem cell function for tissue repair. Biomater. Sci. 6: 947–957, https://doi.org/10.1039/c7bm01158j.Search in Google Scholar

Fiorentino, D.F., Bond, M.W., and Mosmann, T.R. (1989). Two types of mouse T helper cell. IV. Th2 clones secrete a factor that inhibits cytokine production by Th1 clones. J. Exp. Med. 170: 2081–2095, https://doi.org/10.1084/jem.170.6.2081.Search in Google Scholar

Forster, M. and Mulloy, B. (2006). Computational approaches to the identification of heparin-binding sites on the surfaces of proteins. Biochem. Soc. Trans. 34: 431–434, https://doi.org/10.1042/bst0340431.Search in Google Scholar

Frederick, M.J., Henderson, Y., Xu, X., Deavers, M.T., Sahin, A.A., Wu, H., Lewis, D.E., El-Naggar, A.K., and Clayman, G.L. (2000). In vivo expression of the novel CXC chemokine BRAK in normal and cancerous human tissue. Am. J. Pathol. 156: 1937–50, https://doi.org/10.1016/s0002-9440(10)65067-5.Search in Google Scholar

Gama, C.I., Tully, S.E., Sotogaku, N., Clark, P.M., Rawat, M., Vaidehi, N., Goddard, W.A., Nishi, A., and Hsieh-Wilson, L.C. (2006). Sulfation patterns of glycosaminoglycans encode molecular recognition and activity. Nat. Cell Biol. 2: 467–473, https://doi.org/10.1038/nchembio810.Search in Google Scholar PubMed

Gandhi, N.S., Coombe, D.R., and Mancera, R.L. (2008). Platelet endothelial cell adhesion molecule 1 (PECAM-1) and its interactions with glycosaminoglycans: 1. Molecular modeling studies. Biochemistry 47: 4851–4862, https://doi.org/10.1021/bi702455e.Search in Google Scholar PubMed

Gandhi, N.S. and Mancera, R.L. (2011). Molecular dynamics simulations of CXCL-8 and its interactions with a receptor peptide, heparin fragments, and sulfated linked cyclitols. J. Chem. Inf. Model. 51: 335–358, https://doi.org/10.1021/ci1003366.Search in Google Scholar PubMed

Gandhi, N.S. and Mancera, R.L. (2012). Prediction of heparin binding sites in bone morphogenetic proteins (BMPs). Biochim. Biophys. Acta 1824: 1374–1381, https://doi.org/10.1016/j.bbapap.2012.07.002.Search in Google Scholar PubMed

Gao, Q., Chen, C.Y., Zong, C., Wang, S., Ramiah, A., Prabhakar, P., Morris, L.C., Boons, G.J., Moremen, K.W., and Prestegard, J.H. (2016). Structural aspects of heparan sulfate binding to robo1-Ig1-2. ACS Chem. Biol. 11: 3106–3113, https://doi.org/10.1021/acschembio.6b00692.Search in Google Scholar PubMed PubMed Central

Gao, Q., Yang, J.Y., Moremen, K.W., Flanagan, J.G., and Prestegard, J.H. (2018). Structural characterization of a heparan sulfate pentamer interacting with LAR-Ig1-2. Biochemistry 57: 2189–2199, https://doi.org/10.1021/acs.biochem.8b00241.Search in Google Scholar PubMed PubMed Central

García-Jiménez, J., de Paz, J.L., and Nieto, P.M. (2019). Langerin-heparin interaction: analysis of the binding to the non-lectin site. Nat. Prod. Commun., https://doi.org/10.1177/1934578X19851597 (Epub ahead of print).Search in Google Scholar

García-Jiménez, M.J., Gil-Caballero, S., Canales, Á., Jiménez-Barbero, J., de Paz, J.L., and Nieto, P.M. (2017). Interactions between a heparin trisaccharide library and FGF-1 analyzed by NMR methods. Int. J. Mol. Sci. 18: 1293, https://doi.org/10.3390/ijms18061293.Search in Google Scholar PubMed PubMed Central

García-Mayoral, M.F., Canales, A., Díaz, D., López-Prados, J., Moussaoui, M., de Paz, J.L., Angulo, J., Nieto, P.M., Jiménez-Barbero, J., Boix, E., and Bruix, M. (2013). Insights into the glycosaminoglycan-mediated cytotoxic mechanism of eosinophil cationic protein revealed by NMR. ACS Chem. Biol. 8: 144–151, https://doi.org/10.1021/cb300386v.Search in Google Scholar PubMed

Gehrcke, J.P. and Pisabarro, M.T. (2015). Identification and characterization of a glycosaminoglycan binding site on interleukin-10 via molecular simulation methods. J. Mol. Graph. Model. 62: 97–104, https://doi.org/10.1016/j.jmgm.2015.09.003.Search in Google Scholar PubMed

Göbl, C., Madl, T., Simon, B., and Sattler, M. (2014). NMR approaches for structural analysis of multidomain proteins and complexes in solution. Prog. Nucl. Magn. Reson. Spectrosc. 80: 26–63, https://doi.org/10.1016/j.pnmrs.2014.05.003.Search in Google Scholar PubMed

Grant, O.C., Tessier, M.B., Meche, L., Mahal, L.K., Foley, B.L., and Woods, R.J. (2016). Combining 3D structure with glycan array data provides insight into the origin of glycan specificity. Glycobiology 26: 772–783, https://doi.org/10.1093/glycob/cww020.Search in Google Scholar PubMed PubMed Central

Griffith, A.R., Rogers, C.J., Miller, G.M., Abrol, R., Hsieh-Wilson, L.C., and Goddard, W.A.III. (2017). Predicting glycosaminoglycan surface protein interactions and implications for studying axonal growth. Proc. Natl. Acad. Sci. U. S. A. 114: 13697–13702, https://doi.org/10.1073/pnas.1715093115.Search in Google Scholar PubMed PubMed Central

Grünig, G., Corry, D.B., Leach, M.W., Seymour, B.W., Kurup, V.P., and Rennick, D.M. (1997). Interleukin-10 is a natural suppressor of cytokine production and inflammation in a murine model of allergic bronchopulmonary aspergillosis. J. Exp. Med. 185: 1089–1099, https://doi.org/10.1084/jem.185.6.1089.Search in Google Scholar

Guerrini, M., Guglieri, S., Casu, B., Torri, G., Mourier, P., Boudier, C., and Viskov, C. (2008). Antithrombin-binding octasaccharides and role of extensions of the active pentasaccharide sequence in the specificity and strength of interaction. Evidence for very high affinity induced by an unusual glucuronic acid residue. J. Biol. Chem. 283: 26662–26675, https://doi.org/10.1074/jbc.m801102200.Search in Google Scholar

Hileman, R.E., Fromm, J.R., Weiler, J.M., and Linhardt, R.J. (1998). Glycosaminoglycan-protein interactions: definition of consensus sites in glycosaminoglycan binding proteins. Bioessays 20: 156–167, https://doi.org/10.1002/(sici)1521-1878(199802)20:2<156::aid-bies8>3.0.co;2-r.10.1002/(SICI)1521-1878(199802)20:2<156::AID-BIES8>3.0.CO;2-RSearch in Google Scholar

Hintze, V., Samsonov, S.A., Anselmi, M., Moeller, S., Becher, J., Schnabelrauch, M., Scharnweber, D., and Pisabarro, M.T. (2014). Sulfated glycosaminoglycans exploit the conformational plasticity of bone morphogenetic protein-2 (BMP-2) and alter the interaction profile with its receptor. Biomacromolecules 15: 3083–3092, https://doi.org/10.1021/bm5006855.Search in Google Scholar

Hofmann, T., Samsonov, S.A., Pichert, A., Lemmnitzer, K., Schiller, J., Huster, D., Pisabarro, M.T., von Bergen, M., and Kalkhof, S. (2015). Structural analysis of the interleukin-8/glycosaminoglycan interactions by amide hydrogen/deuterium exchange mass spectrometry. Methods 89: 45–53, https://doi.org/10.1016/j.ymeth.2015.02.011.Search in Google Scholar

Huster, D. (2005). Investigations of the structure and dynamics of membrane-associated peptides by magic angle spinning NMR. Prog. Nucl. Magn. Reson. Spectrosc. 46: 79–107, https://doi.org/10.1016/j.pnmrs.2005.01.001.Search in Google Scholar

Imberty, A., Lortat-Jacob, H., and Perez, S. (2007). Structural view of glycosaminoglycan-protein interactions. Carbohydr. Res. 342: 430–439, https://doi.org/10.1016/j.carres.2006.12.019.Search in Google Scholar

Iozzo, R.V. and Schaefer, L. (2015). Proteoglycan form and function: a comprehensive nomenclature of proteoglycans. Matrix Biol. 42: 11–55, https://doi.org/10.1016/j.matbio.2015.02.003.Search in Google Scholar

Jana, P., Radhakrishna, M., Khatua, S., and Kanvah, S. (2018). A "turn-off" red-emitting fluorophore for nanomolar detection of heparin. Phys. Chem. Chem. Phys. 20: 1326313270, https://doi.org/10.1039/c7cp06300h.Search in Google Scholar

Jayalakshmi, V. and Krishna, N.R. (2002). Complete relaxation and conformational exchange matrix (CORCEMA) analysis of intermolecular saturation transfer effects in reversibly forming ligand-receptor complexes. J. Magn. Reson. 155: 106–118, https://doi.org/10.1006/jmre.2001.2499.Search in Google Scholar

Joseph, P.R., Mosier, P.D., Desai, U.R., and Rajarathnam, K. (2015). Solution NMR characterization of chemokine CXCL8/IL-8 monomer and dimer binding to glycosaminoglycans: structural plasticity mediates differential binding interactions. Biochem. J. 472: 121–133, https://doi.org/10.1042/bj20150059.Search in Google Scholar

Joss, D. and Häussinger, D. (2019). Design and applications of lanthanide chelating tags for pseudocontact shift NMR spectroscopy with biomacromolecules. Prog. Nucl. Magn. Reson. Spectrosc. 114-115: 284–312, https://doi.org/10.1016/j.pnmrs.2019.08.002.Search in Google Scholar PubMed

Kastana, P., Choleva, E., Poimenidi, E., Karamanos, N., Sugahara, K., and Papadimitriou, E. (2019). Insight into the role of chondroitin sulfate E in angiogenesis. FEBS J. 286: 2921–2936, https://doi.org/10.1111/febs.14830.Search in Google Scholar PubMed

Kerzmann, A., Fuhrmann, J., Kohlbacher, O., and Neumann, D. (2008). BALLDock/SLICK: a new method for protein-carbohydrate docking. J. Chem. Inf. Model. 48: 1616–1625, https://doi.org/10.1021/ci800103u.Search in Google Scholar PubMed

Kinnings, S.L., Liu, N., Tonge, P.J., Jackson, R.M., Xie, L., and Bourne, P.E. (2011). A machine learning-based method to improve docking scoring functions and its application to drug repurposing. J. Chem. Inf. Model. 51: 408–419, https://doi.org/10.1021/ci100369f.Search in Google Scholar PubMed PubMed Central

Kirschner, K.N., Yongye, A.B., Tschampel, S.M., González-Outeiriño, J., Daniels, C.R., Foley, B.L., and Woods, R.J. (2008). GLYCAM06: a generalizable biomolecular force field. Carbohydr. J. Comput. Chem. 29: 622–655, https://doi.org/10.1002/jcc.20820.Search in Google Scholar PubMed PubMed Central

Kogut, M.M., Maszota-Zieleniak, M., Marcisz, M., and Samsonov, S.A. (2021). Computational insights into the role of calcium ions in protein-glycosaminoglycan systems. Phys. Chem. Chem. Phys. 23: 3519–3530, https://doi.org/10.1039/d0cp05438k.Search in Google Scholar PubMed

Köhling, S., Künze, G., Lemmnitzer, K., Bermudez, M., Wolber, G., Schiller, J., Huster, D., and Rademann, J. (2016). Chemoenzymatic synthesis of nonasulfated tetrahyaluronan with a paramagnetic tag for studying its complex with interleukin-10. Chemistry 22: 5563–5574, https://doi.org/10.1002/chem.201504459.Search in Google Scholar PubMed

Kolesnikov, A.L., Budkov, Y.A., and Nogovitsyn, E.A. (2014). Coarse-grained model of glycosaminoglycans in aqueous salt solutions. A field-theoretical approach. J. Phys. Chem. B 118: 13037–13049, https://doi.org/10.1021/jp503749a.Search in Google Scholar PubMed

Kollman, P.A., Massova, I., Reyes, C., Kuhn, B., Huo, S., Chong, L., Lee, M., Lee, T., Duan, Y., Wang, W., et al.. (2000). Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Acc. Chem. Res. 33: 889–897, https://doi.org/10.1021/ar000033j.Search in Google Scholar PubMed

Kozakov, D., Hall, D.R., Xia, B., Porter, K.A., Padhorny, D., Yueh, C., Beglov, D., and Vajda, S. (2017). The ClusPro web server for protein-protein docking. Nat. Protoc. 12: 255–278, https://doi.org/10.1038/nprot.2016.169.Search in Google Scholar

Kucia, M., Jankowski, K., Reca, R., Wysoczynski, M., Bandura, L., Allendorf, D.J., Zhang, J., Ratajczak, J., and Ratajczak, M.Z. (2004). CXCR4-SDF-1 signalling, locomotion, chemotaxis and adhesion. J. Mol. Histol. 35: 233–245, https://doi.org/10.1023/b:hijo.0000032355.66152.b8.10.1023/B:HIJO.0000032355.66152.b8Search in Google Scholar

Kucia, M., Ratajczak, J., and Ratajczak, M.Z. (2005). Bone marrow as a source of circulating CXCR4+ tissue-committed stem cells. Biol. Cell 97: 133–146, https://doi.org/10.1042/bc20040069.Search in Google Scholar

Kuhn, L.A., Griffin, J.H., Fisher, C.L., Greengard, J.S., Bouma, B.N., España, F., and Tainer, J.A. (1990). Elucidating the structural chemistry of glycosaminoglycan recognition by protein C inhibitor. Proc. Natl. Acad. Sci. U. S. A. 87: 8506–8510, https://doi.org/10.1073/pnas.87.21.8506.Search in Google Scholar

Kühn, R., Löhler, J., Rennick, D., Rajewsky, K., and Müller, W. (1993). Interleukin-10-deficient mice develop chronic enterocolitis. Cell 75: 263–274, https://doi.org/10.1016/0092-8674(93)80068-p.Search in Google Scholar

Künze, G., Gehrcke, J.P., Pisabarro, M.T., and Huster, D. (2014). NMR characterization of the binding properties and conformation of glycosaminoglycans interacting with interleukin-10. Glycobiology 24: 1036–1049, https://doi.org/10.1093/glycob/cwu069.Search in Google Scholar PubMed

Künze, G., Köhling, S., Vogel, A., Rademann, J., and Huster, D. (2016). Identification of the glycosaminoglycan binding site of interleukin-10 by NMR spectroscopy. J. Biol. Chem. 291: 3100–3113, https://doi.org/10.1074/jbc.m115.681759.Search in Google Scholar PubMed PubMed Central

Kurth, I., Willimann, K., Schaerli, P., Hunziker, T., Clark-Lewis, I., and Moser, B. (2001). Monocyte selectivity and tissue localization suggests a role for breast and kidney-expressed chemokine (BRAK) in macrophage development. J. Exp. Med. 194: 855–861, https://doi.org/10.1084/jem.194.6.855.Search in Google Scholar PubMed PubMed Central

Kurzbach, D., Schwarz, T.C., Platzer, G., Höfler, S., Hinderberger, D., and Konrat, R. (2014). Compensatory adaptations of structural dynamics in an intrinsically disordered protein complex. Angew. Chem. Int. Ed. 53: 3840–3843, https://doi.org/10.1002/anie.201308389.Search in Google Scholar PubMed

Laguri, C., Sapay, N., Simorre, J.P., Brutscher, B., Imberty, A., Gans, P., and Lortat-Jacob, H. (2011). 13C-labeled heparan sulfate analogue as a tool to study protein/heparan sulfate interactions by NMR spectroscopy: application to the CXCL12α chemokine. J. Am. Chem. Soc. 133: 9642–9645, https://doi.org/10.1021/ja201753e.Search in Google Scholar PubMed

Levy, Y. and Brouet, J.C. (1994). Interleukin-10 prevents spontaneous death of germinal center B cells by induction of the bcl-2 protein. J. Clin. Invest. 93: 424–428, https://doi.org/10.1172/jci116977.Search in Google Scholar PubMed PubMed Central

Llorente, L., Zou, W., Levy, Y., Richaud-Patin, Y., Wijdenes, J., Alcocer-Varela, J., Morel-Fourrier, B., Brouet, J.C., Alarcon-Segovia, D., Galanaud, P., and Emilie, D. (1995). Role of interleukin 10 in the B lymphocyte hyperactivity and autoantibody production of human systemic lupus erythematosus. J. Exp. Med. 181: 839–844, https://doi.org/10.1084/jem.181.3.839.Search in Google Scholar PubMed PubMed Central

López-Ruiz, E., Jiménez, G., Álvarez de Cienfuegos, L., Antic, C., Sabata, R., Marchal, J.A., and Gálvez-Martín, P. (2019). Advances of hyaluronic acid in stem cell therapy and tissue engineering, including current clinical trials. Eur. Cell. Mater. 37: 186–213, https://doi.org/10.22203/ecm.v037a12.Search in Google Scholar

Lu, W., Zong, C., Chopra, P., Pepi, L.E., Xu, Y., Amster, I.J., Liu, J., and Boons, G.J. (2018). Controlled chemoenzymatic synthesis of heparan sulfate oligosaccharides. Angew. Chem. Int. Ed. 57: 5340–5344, https://doi.org/10.1002/anie.201800387.Search in Google Scholar PubMed PubMed Central

MacKerell, A.D., Bashford, D., Bellott, M., Dunbrack, R.L., Evanseck, J.D., Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., et al.. (1998). All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102: 3586–3616, https://doi.org/10.1021/jp973084f.Search in Google Scholar PubMed

Madine, J., Pandya, M.J., Hicks, M.R., Rodger, A., Yates, E.A., Radford, S.E., and Middleton, D.A. (2012). Site-specific identification of an Aβ fibril-heparin interaction site by using solid-state NMR spectroscopy. Angew. Chem. Int. Ed. 51: 13140–13143, https://doi.org/10.1002/anie.201204459.Search in Google Scholar PubMed PubMed Central

Maerki, C., Meuter, S., Liebi, M., Mühlemann, K., Frederick, M.J., Yawalkar, N., Moser, B., and Wolf, M. (2009). Potent and broad-spectrum antimicrobial activity of CXCL14 suggests an immediate role in skin infections. J. Immunol. 182: 507–514, https://doi.org/10.4049/jimmunol.182.1.507.Search in Google Scholar PubMed

Maïza, A., Chantepie, S., Vera, C., Fifre, A., Huynh, M.B., Stettler, O., Ouidja, M.O., and Papy-Garcia, D. (2018). The role of heparan sulfates in protein aggregation and their potential impact on neurodegeneration. FEBS Lett. 592: 3806–3818, https://doi.org/10.1002/1873-3468.13082.Search in Google Scholar PubMed

Maszota-Zieleniak, M., Marcisz, M., Kogut, M.M., Siebenmorgen, T., Zacharias, M., and Samsonov, S.A. (in press). Evaluation of replica exchange with repulsive scaling approach for docking glycosaminoglycans. J. Comput. Chem. 10.1002/jcc.26496Search in Google Scholar PubMed

Maytin, E.V. (2016). Hyaluronan: more than just a wrinkle filler. Glycobiology 26: 553–559, https://doi.org/10.1093/glycob/cww033.Search in Google Scholar PubMed PubMed Central

Mikhailov, D., Mayo, K.H., Vlahov, I.R., Toida, T., Pervin, A., and Linhardt, R.J. (1996). NMR solution conformation of heparin-derived tetrasaccharide. Biochem. J. 318: 93–102, https://doi.org/10.1042/bj3180093.Search in Google Scholar

Miller, M.C. and Mayo, K.H. (2017). Chemokines from a structural perspective. Int. J. Mol. Sci. 18: 2088, https://doi.org/10.3390/ijms18102088.Search in Google Scholar

Möbius, K., Nordsieck, K., Pichert, A., Samsonov, S.A., Thomas, L., Schiller, J., Kalkhof, S., Pisabarro, M.T., Beck-Sickinger, A.G., and Huster, D. (2013). Investigation of lysine side chain interactions of interleukin-8 with heparin and other glycosaminoglycans studied by a methylation-NMR approach. Glycobiology 23: 1260–1269, https://doi.org/10.1093/glycob/cwt062.Search in Google Scholar

Morgan, A. and Wang, X. (2013). The novel heparin-binding motif in decorin-binding protein A from strain B31 of Borrelia burgdorferi explains the higher binding affinity. Biochemistry 52: 8237–8245, https://doi.org/10.1021/bi401376u.Search in Google Scholar

Morris, G., Goodsell, D., Halliday, R., Huey, R., Hart, W., Belew, R., and Olson, A. (1999). Automated docking using a Lamarckian genetic algorithm and an empirical binding free function. J. Comput. Chem. 19: 1639–1662.10.1002/(SICI)1096-987X(19981115)19:14<1639::AID-JCC10>3.0.CO;2-BSearch in Google Scholar

Moure, M.J., Eletsky, A., Gao, Q., Morris, L.C., Yang, J.Y., Chapla, D., Zhao, Y., Zong, C., Amster, I.J., Moremen, K.W., Boons, G.J., and Prestegard, J.H. (2018). Paramagnetic tag for glycosylation sites in glycoproteins: structural constraints on heparan sulfate binding to Robo1. ACS Chem. Biol. 13: 2560–2567, https://doi.org/10.1021/acschembio.8b00511.Search in Google Scholar

Munoz-Garcia, J.C., Chabrol, E., Vivès, R.R., Thomas, A., de Paz, J.L., Rojo, J., Imberty, A., Fieschi, F., Nieto, P.M., and Angulo, J. (2015). Langerin-heparin interaction: two binding sites for small and large ligands as revealed by a combination of NMR spectroscopy and cross-linking mapping experiments. J. Am. Chem. Soc. 137: 4100–4110, https://doi.org/10.1021/ja511529x.Search in Google Scholar

Muñoz-García, J.C., Garcia-Jiménez, M.J., Carrero, P., Canales, A., Jimenez-Barbero, J., Martin-Lomas, M., Imberty, A., de Paz, J.L., Angulo, J., Lortat-Jacob, H., and Nieto, P.M. (2014). Importance of the polarity of the glycosaminoglycan chain on the interaction with FGF-1. Glycobiology 24: 1004–1009, https://doi.org/10.1093/glycob/cwu071.Search in Google Scholar

Muñoz-García, J.C., Solera, C., Carrero, P., de Paz, J.L., Angulo, J., and Nieto, P.M. (2013). 3D structure of a heparin mimetic analogue of a FGF-1 activator. A NMR and molecular modelling study. Org. Biomol. Chem. 11: 8269–8275, https://doi.org/10.1039/c3ob41789a.Search in Google Scholar

Murphy, J.W., Cho, Y., Sachpatzidis, A., Fan, C., Hodsdon, M.E., and Lolis, E. (2007). Structural and functional basis of CXCL12 (stromal cell-derived factor-1 alpha) binding to heparin. J. Biol. Chem. 282: 10018–10027, https://doi.org/10.1074/jbc.m608796200.Search in Google Scholar

Naji, L., Kaufmann, J., Huster, D., Schiller, J., and Arnold, K. (2000). 13C NMR relaxation study on cartilage and cartilage components. The origin of 13C NMR spectra of cartilage. Carbohydr. Res. 327: 439–446, https://doi.org/10.1016/s0008-6215(00)00064-1.Search in Google Scholar

Nieto, P.M., Canales, A., Fernández, I.S., Santillana, E., González-Corrochano, R., Redondo-Horcajo, M., Cañada, F.J., Nieto, P., Martin-Lomas, M., Gimenez-Gallego, G., and Jiménez-Barbero, J. (2013). Heparin modulates the mitogenic activity of fibroblast growth factor by inducing dimerization of its receptor. a 3D view by using NMR. Chembiochem 14: 1732–1744, https://doi.org/10.1002/cbic.201300313.Search in Google Scholar PubMed

Nitz, M., Franz, K.J., Maglathlin, R.L., and Imperiali, B. (2003). A powerful combinatorial screen to identify high-affinity terbium(III)-binding peptides. Chembiochem 4: 272–276, https://doi.org/10.1002/cbic.200390047.Search in Google Scholar PubMed

Nivedha, A.K., Makeneni, S., Foley, B.L., Tessier, M.B., and Woods, R.J. (2014). Importance of ligand conformational energies in carbohydrate docking: sorting the wheat from the chaff. J. Comput. Chem. 35: 526–539, https://doi.org/10.1002/jcc.23517.Search in Google Scholar PubMed PubMed Central

Nordsieck, K., Pichert, A., Samsonov, S.A., Thomas, L., Berger, C., Pisabarro, M.T., Huster, D., and Beck-Sickinger, A.G. (2012). Residue 75 of interleukin-8 is crucial for its interactions with glycosaminoglycans. Chembiochem 13: 2558–2566, https://doi.org/10.1002/cbic.201200467.Search in Google Scholar PubMed

Ossipov, D.A. (2019). Hyaluronan-based delivery of therapeutic oligonucleotides for treatment of human diseases. Expert. Opin. Drug Deliv. 16: 621–637, https://doi.org/10.1080/17425247.2019.1617693.Search in Google Scholar PubMed

Otting, G. (2010). Protein NMR using paramagnetic ions. Annu. Rev. Biophys. 39: 387–405, https://doi.org/10.1146/annurev.biophys.093008.131321.Search in Google Scholar PubMed

Panitz, N., Theisgen, S., Samsonov, S.A., Gehrcke, J.P., Baumann, L., Bellmann-Sickert, K., Köhling, S., Pisabarro, M.T., Rademann, J., Huster, D., and Beck-Sickinger, A.G. (2016). The structural investigation of glycosaminoglycan binding to CXCL12 displays distinct interaction sites. Glycobiology 26: 1209–1221, https://doi.org/10.1093/glycob/cww059.Search in Google Scholar PubMed

Park, Y., Jowitt, T.A., Day, A.J., and Prestegard, J.H. (2016). Nuclear magnetic resonance insight into the multiple glycosaminoglycan binding modes of the link module from human TSG-6. Biochemistry 55: 262–276, https://doi.org/10.1021/acs.biochem.5b01148.Search in Google Scholar PubMed PubMed Central

Passi, A. and Vigetti, D. (2019). Hyaluronan as tunable drug delivery system. Adv. Drug Deliv. Rev. 146: 83–96, https://doi.org/10.1016/j.addr.2019.08.006.Search in Google Scholar PubMed

Pavão, M.S. (2002). Structure and anticoagulant properties of sulfated glycosaminoglycans from primitive Chordates. An. Acad. Bras. Cienc. 74: 105–112, https://doi.org/10.1590/s0001-37652002000100007.Search in Google Scholar PubMed

Penk, A., Baumann, L., Huster, D., and Samsonov, S.A. (2019). NMR and molecular modeling reveal specificity of the interactions between CXCL14 and glycosaminoglycans. Glycobiology 29: 715–725, https://doi.org/10.1093/glycob/cwz047.Search in Google Scholar PubMed

Phillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel, R.D., Kalé, L., and Schulten, K. (2005). Scalable molecular dynamics with NAMD. J. Comput. Chem. 26: 1781–1802, https://doi.org/10.1002/jcc.20289.Search in Google Scholar PubMed PubMed Central

Pichert, A., Samsonov, S.A., Theisgen, S., Thomas, L., Baumann, L., Schiller, J., Beck-Sickinger, A.G., Huster, D., and Pisabarro, M.T. (2012). Characterization of the interaction of interleukin-8 with hyaluronan, chondroitin sulfate, dermatan sulfate and their sulfated derivatives by spectroscopy and molecular modeling. Glycobiology 22: 134–145, https://doi.org/10.1093/glycob/cwr120.Search in Google Scholar PubMed PubMed Central

Pomin, V.H. (2014a). Solution NMR conformation of glycosaminoglycans. Prog. Biophys. Mol. Biol. 114: 61–68, https://doi.org/10.1016/j.pbiomolbio.2014.01.001.Search in Google Scholar PubMed

Pomin, V.H. (2014b). Biological findings from the recent NMR-based studies of glycosaminoglycan-protein interactions. Glycobiology 24: 991–1003, https://doi.org/10.1093/glycob/cwu065.Search in Google Scholar PubMed

Pomin, V.H. and Mulloy, B. (2015). Current structural biology of the heparin interactome. Curr. Opin. Struct. Biol. 34: 17–25, https://doi.org/10.1016/j.sbi.2015.05.007.Search in Google Scholar PubMed

Pomin, V.H. and Wang, X. (2018a). Glycosaminoglycan-protein interactions by nuclear magnetic resonance (NMR) spectroscopy. Molecules 23: 2314, https://doi.org/10.3390/molecules23092314.Search in Google Scholar PubMed PubMed Central

Pomin, V.H. and Wang, Xu, W. (2018b). Synthetic oligosaccharide libraries and microarray technology: a powerful combination for the success of current glycosaminoglycan interactomics. ChemMedChem 13: 1860–7179, https://doi.org/10.1002/cmdc.201800185.Search in Google Scholar

Potthoff, J., Bojarski, K.K., Kohut, G., Lipska, A.G., Liwo, A., Kessler, E., Ricard-Blum, S., and Samsonov, S.A. (2019). Analysis of procollagen C-proteinase enhancer-1/glycosaminoglycan binding sites and of the potential role of calcium ions in the interaction. Int. J. Mol. Sci. 20: 5021, https://doi.org/10.3390/ijms20205021.Search in Google Scholar PubMed PubMed Central

Proudfoot, A.E. (2006). The biological relevance of chemokine-proteoglycan interactions. Biochem. Soc. Trans. 34: 422–426, https://doi.org/10.1042/bst0340422.Search in Google Scholar PubMed

Proudfoot, A.E.I., Johnson, Z., Bonvin, P., and Handel, T.M. (2017). Glycosaminoglycan interactions with chemokines add complexity to a complex system. Pharmaceuticals 10: 70, https://doi.org/10.3390/ph10030070.Search in Google Scholar PubMed PubMed Central

Rabenstein, D.L. (2002). Heparin and heparan sulfate: structure and function. Nat. Prod. Rep. 19: 312–331, https://doi.org/10.1039/b100916h.Search in Google Scholar PubMed

Ricard-Blum, S., Féraud, O., Lortat-Jacob, H., Rencurosi, A., Fukai, N., Dkhissi, F., Vittet, D., Imberty, A., Olsen, B.R., and van der Rest, M. (2004). Characterization of endostatin binding to heparin and heparan sulfate by surface plasmon resonance and molecular modeling: role of divalent cations. J. Biol. Chem. 279: 2927–2936, https://doi.org/10.1074/jbc.m309868200.Search in Google Scholar PubMed

Rogers, C.J., Clark, P.M., Tully, S.E., Abrol, R., Garcia, K.C., Goddard, W.A.III, and Hsieh-Wilson, L.C. (2011). Elucidating glycosaminoglycan-protein-protein interactions using carbohydrate microarray and computational approaches. Proc. Natl. Acad. Sci. U. S. A. 108: 9747–9752, https://doi.org/10.1073/pnas.1102962108.Search in Google Scholar PubMed PubMed Central

Sabat, R., Grütz, G., Warszawska, K., Kirsch, S., Witte, E., Wolk, K., and Geginat, J. (2010). Biology of interleukin-10. Cytokine Growth Factor Rev. 21: 331–344, https://doi.org/10.1016/j.cytogfr.2010.09.002.Search in Google Scholar PubMed

Sadir, R., Imberty, A., Baleux, F., and Lortat-Jacob, H. (2004). Heparan sulfate/heparin oligosaccharides protect stromal cell-derived factor-1 (SDF-1)/CXCL12 against proteolysis induced by CD26/dipeptidyl peptidase IV. J. Biol. Chem. 279: 43854–43860, https://doi.org/10.1074/jbc.m405392200.Search in Google Scholar PubMed

Sage, J., Mallèvre, F., Barbarin-Costes, F., Samsonov, S.A., Gehrcke, J.P., Pisabarro, M.T., Perrier, E., Schnebert, S., Roget, A., Livache, T., Nizard, C., Lalmanach, G., and Lecaille, F. (2013). Binding of chondroitin 4-sulfate to cathepsin S regulates its enzymatic activity. Biochemistry 52: 6487–6498, https://doi.org/10.1021/bi400925g.Search in Google Scholar PubMed

Salbach-Hirsch, J., Samsonov, S.A., Hintze, V., Hofbauer, C., Picke, A.K., Rauner, M., Gehrcke, J.P., Moeller, S., Schnabelrauch, M., Scharnweber, D., Pisabarro, M. T., and Hofbauer, L.C. (2015). Structural and functional insights into sclerostin-glycosaminoglycan interactions in bone. Biomaterials 67: 335–345, https://doi.org/10.1016/j.biomaterials.2015.07.021.Search in Google Scholar PubMed

Salek-Ardakani, S., Arrand, J.R., Shaw, D., and Mackett, M. (2000). Heparin and heparan sulfate bind interleukin-10 and modulate its activity. Blood 96: 1879–1888, https://doi.org/10.1182/blood.v96.5.1879.h8001879_1879_1888.Search in Google Scholar

Samsonov, S.A., Bichmann, L., and Pisabarro, M.T. (2015). Coarse-grained model of glycosaminoglycans. J. Chem. Inf. Model. 55: 114–124, https://doi.org/10.1021/ci500669w.Search in Google Scholar PubMed

Samsonov, S.A., Gehrcke, J.P., and Pisabarro, M.T. (2014). Flexibility and explicit solvent in molecular-dynamics-based docking of protein-glycosaminoglycan systems. J. Chem. Inf. Model. 54: 582–592, https://doi.org/10.1021/ci4006047.Search in Google Scholar PubMed

Samsonov, S.A., Lubecka, E.A., Bojarski, K.K., Ganzynkowicz, R., and Liwo, A. (2019a). Local and long range potentials for heparin-protein systems for coarse-grained simulations. Biopolymers 110: e23269, https://doi.org/10.1002/bip.23269.Search in Google Scholar PubMed

Samsonov, S.A. and Pisabarro, M.T. (2013). Importance of IdoA and IdoA(2S) ring conformations in computational studies of glycosaminoglycan-protein interactions. Carbohydr. Res. 381: 133–137, https://doi.org/10.1016/j.carres.2013.09.005.Search in Google Scholar PubMed

Samsonov, S.A. and Pisabarro, M.T. (2016). Computational analysis of interactions in structurally available protein-glycosaminoglycan complexes. Glycobiology 26: 850–861, https://doi.org/10.1093/glycob/cww055.Search in Google Scholar PubMed

Samsonov, S.A., Teyra, J., and Pisabarro, M.T. (2011). Docking glycosaminoglycans to proteins: analysis of solvent inclusion. J. Comput. Aided Mol. Des. 25: 477–489, https://doi.org/10.1007/s10822-011-9433-1.Search in Google Scholar PubMed PubMed Central

Samsonov, S.A., Zacharias, M., and Chauvot de Beauchene, I. (2019b). Modeling large protein-glycosaminoglycan complexes using a fragment-based approach. J. Comput. Chem. 40: 1429–1439, https://doi.org/10.1002/jcc.25797.Search in Google Scholar PubMed

Sankaranarayanan, N.V., Nagarajan, B., and Desai, U.R. (2018). So you think computational approaches to understanding glycosaminoglycan-protein interactions are too dry and too rigid? Think again!. Curr. Opin. Struct. Biol. 50: 91–100, https://doi.org/10.1016/j.sbi.2017.12.004.Search in Google Scholar PubMed PubMed Central

Sapay, N., Cabannes, E., Petitou, M., and Imberty, A. (2011). Molecular modeling of the interaction between heparan sulfate and cellular growth factors: bringing pieces together. Glycobiology 21: 1181–1193, https://doi.org/10.1093/glycob/cwr052.Search in Google Scholar PubMed

Sattelle, B.M., Shakeri, J., and Almond, A. (2013). Does microsecond sugar ring flexing encode 3D-shape and bioactivity in the heparanome?. Biomacromolecules 14: 1149–1159, https://doi.org/10.1021/bm400067g.Search in Google Scholar PubMed

Sattelle, B.M., Shakeri, J., Cliff, M.J., and Almond, A. (2015). Proteoglycans and their heterogeneous glycosaminoglycans at the atomic scale. Biomacromolecules 16: 951–961, https://doi.org/10.1021/bm5018386.Search in Google Scholar

Sattelle, B.M., Shakeri, J., Roberts, I.S., and Almond, A. (2010). A 3D-structural model of unsulfated chondroitin from high-field NMR: 4-sulfation has little effect on backbone conformation. Carbohydr. Res. 345: 291–302, https://doi.org/10.1016/j.carres.2009.11.013.Search in Google Scholar

Scheidt, H.A., Schibur, S., Magalhaes, A., de Azevedo, E.R., Bonagamba, T.J., Pascui, O., Schulz, R., Reichert, D., and Huster, D. (2010). The mobility of chondroitin sulfate in articular and artificial cartilage characterized by 13C magic-angle spinning NMR spectroscopy. Biopolymers 93: 520–532, https://doi.org/10.1002/bip.21386.Search in Google Scholar

Schiller, J. and Huster, D. (2012). New methods to study the composition and structure of the extracellular matrix in natural and bioengineered tissues. Biomatter 2: 115–131, https://doi.org/10.4161/biom.20866.Search in Google Scholar

Schlorke, D., Thomas, L., Samsonov, S.A., Huster, D., Arnhold, J., and Pichert, A. (2012). The influence of glycosaminoglycans on IL-8-mediated functions of neutrophils. Carbohydr. Res. 356: 196–203, https://doi.org/10.1016/j.carres.2012.02.025.Search in Google Scholar

Schwartz, N.B. and Domowicz, M.S. (2018). Proteoglycans in brain development and pathogenesis. FEBS Lett. 592: 3791–3805, https://doi.org/10.1002/1873-3468.13026.Search in Google Scholar

Schwieters, C.D., Kuszewski, J.J., Tjandra, N., and Clore, G.M. (2003). The Xplor-NIH NMR molecular structure determination package. J. Magn. Reson. 160: 65–73, https://doi.org/10.1016/s1090-7807(02)00014-9.Search in Google Scholar

Sekhar, A. and Kay, L.E. (2019). An NMR view of protein dynamics in health and disease. Annu. Rev. Biophys. 48: 297–319, https://doi.org/10.1146/annurev-biophys-052118-115647.Search in Google Scholar PubMed

Sepuru, K.M. and Rajarathnam, K. (2019). Structural basis of chemokine interactions with heparan sulfate, chondroitin sulfate, and dermatan sulfate. J. Biol. Chem. 294: 15650–15661, https://doi.org/10.1074/jbc.ra119.009879.Search in Google Scholar

Seyfried, N.T., Atwood, J.A.III, Yongye, A., Almond, A., Day, A.J., Orlando, R., and Woods, R.J. (2007). Fourier transform mass spectrometry to monitor hyaluronan-protein interactions: use of hydrogen/deuterium amide exchange. Rapid Commun. Mass Spectrom. 21: 121–131, https://doi.org/10.1002/rcm.2817.Search in Google Scholar PubMed PubMed Central

Shellenberger, T.D., Wang, M., Gujrati, M., Jayakumar, A., Strieter, R.M., Burdick, M.D., Ioannides, C.G., Efferson, C.L., El-Naggar, A.K., Roberts, D., et al.. (2004). BRAK/CXCL14 is a potent inhibitor of angiogenesis and a chemotactic factor for immature dendritic cells. Cancer Res 64: 8262–8270, https://doi.org/10.1158/0008-5472.can-04-2056.Search in Google Scholar PubMed

Shurin, G.V., Ferris, R.L., Tourkova, I.L., Perez, L., Lokshin, A., Balkir, L., Collins, B., Chatta, G.S., and Shurin, M.R. (2005). Loss of new chemokine CXCL14 in tumor tissue is associated with low infiltration by dendritic cells (DC), while restoration of human CXCL14 expression in tumor cells causes attraction of DC both in vitro and in vivo. J. Immunol. 174: 5490–5498, https://doi.org/10.4049/jimmunol.174.9.5490.Search in Google Scholar PubMed

Siebenmorgen, T., Engelhard, M., and Zacharias, M. (2020). Prediction of protein-protein complexes using replica exchange with repulsive scaling. J. Comput. Chem. 41: 1436–1447, https://doi.org/10.1002/jcc.26187.Search in Google Scholar PubMed

Singh, A., Kett, W.C., Severin, I.C., Agyekum, I., Duan, J., Amster, I.J., Proudfoot, A.E., Coombe, D.R., and Woods, R.J. (2015). The interaction of heparin tetrasaccharides with chemokine CCL5 is modulated by sulfation pattern and pH. J. Biol. Chem. 290: 15421–15436, https://doi.org/10.1074/jbc.m115.655845.Search in Google Scholar PubMed PubMed Central

Sleeman, M.A., Fraser, J.K., Murison, J.G., Kelly, S.L., Prestidge, R.L., Palmer, D.J., Watson, J.D., and Kumble, K.D. (2000). B cell- and monocyte-activating chemokine (BMAC), a novel non-ELR α-chemokine. Int. Immunol. 12: 677–689, https://doi.org/10.1093/intimm/12.5.677.Search in Google Scholar PubMed

Srinivasan, J., Cheatham, T.E.III, Cieplak, P., Kollman, P.A., and Case, D.A. (1998). Continuum solvent studies of the stability of DNA, RNA, and phosphoramidate–DNA helices. J. Am. Chem. Soc. 120: 9401–9409, https://doi.org/10.1021/ja981844+.10.1021/ja981844+Search in Google Scholar

Starnes, T., Rasila, K.K., Robertson, M.J., Brahmi, Z., Dahl, R., Christopherson, K., and Hromas, R. (2006). The chemokine CXCL14 (BRAK) stimulates activated NK cell migration: implications for the downregulation of CXCL14 in malignancy. Exp. Hematol. 34: 1101–1105, https://doi.org/10.1016/j.exphem.2006.05.015.Search in Google Scholar PubMed

Stewart, K.L., Hughes, E., Yates, E.A., Akien, G.R., Huang, T.Y., Lima, M.A., Rudd, T.R., Guerrini, M., Hung, S.C., Radford, S.E., and Middleton, D.A. (2016). Atomic details of the interactions of glycosaminoglycans with amyloid-β fibrils. J. Am. Chem. Soc. 138: 8328–8331, https://doi.org/10.1021/jacs.6b02816.Search in Google Scholar PubMed

Tanegashima, K., Suzuki, K., Nakayama, Y., and Hara, T. (2010). Antibody-assisted enhancement of biological activities of CXCL14 in human monocytic leukemia-derived THP-1 cells and high fat diet-induced obese mice. Exp. Cell Res. 316: 1263–1270, https://doi.org/10.1016/j.yexcr.2010.01.016.Search in Google Scholar PubMed

Taroni, C., Jones, S., and Thornton, J.M. (2000). Analysis and prediction of carbohydrate binding sites. Protein Eng. 13: 89–98, https://doi.org/10.1093/protein/13.2.89.Search in Google Scholar PubMed

Teicher, B.A. and Fricker, S.P. (2010). CXCL12 (SDF-1)/CXCR4 pathway in cancer. Clin. Cancer Res. 16: 2927–2931, https://doi.org/10.1158/1078-0432.ccr-09-2329.Search in Google Scholar PubMed

Teyra, J., Samsonov, S.A., Schreiber, S., and Pisabarro, M.T. (2011). SCOWLP update: 3D classification of protein-protein, -peptide, -saccharide and -nucleic acid interactions, and structure-based binding inferences across folds. BMC Bioinf. 12: 398, https://doi.org/10.1186/1471-2105-12-398.Search in Google Scholar PubMed PubMed Central

Tiwari, S. and Bahadur, P. (2019). Modified hyaluronic acid based materials for biomedical applications. Int. J. Biol. Macromol. 121: 556–571, https://doi.org/10.1016/j.ijbiomac.2018.10.049.Search in Google Scholar PubMed

Tiwari, V., Tarbutton, M.S., and Shukla, D. (2015). Diversity of heparan sulfate and HSV entry: basic understanding and treatment strategies. Molecules 20: 2707–2727, https://doi.org/10.3390/molecules20022707.Search in Google Scholar PubMed PubMed Central

Tournoy, K.G., Kips, J.C., and Pauwels, R.A. (2000). Endogenous interleukin-10 suppresses allergen-induced airway inflammation and nonspecific airway responsiveness. Clin. Exp. Allergy 30: 775–783, https://doi.org/10.1046/j.1365-2222.2000.00838.x.Search in Google Scholar PubMed

Tycko, R. (2006). Molecular structure of amyloid fibrils: insights from solid-state NMR. Q. Rev. Biophys. 39: 1–55, https://doi.org/10.1017/s0033583506004173.Search in Google Scholar PubMed

Uciechowska-Kaczmarzyk, U., Babik, S., Zsila, F., Bojarski, K.K., Beke-Somfai, T., and Samsonov, S.A. (2018). Molecular dynamics-based model of VEGF-A and its heparin interactions. J. Mol. Graph. Model. 82: 157–166, https://doi.org/10.1016/j.jmgm.2018.04.015.Search in Google Scholar PubMed

Uciechowska-Kaczmarzyk, U., Chauvot de Beauchene, I., and Samsonov, S.A. (2019). Docking software performance in protein-glycosaminoglycan systems. J. Mol. Graph. Model. 90: 42–50, https://doi.org/10.1016/j.jmgm.2019.04.001.Search in Google Scholar PubMed

Vallet, S.D., Miele, A.E., Uciechowska-Kaczmarzyk, U., Liwo, A., Duclos, B., Samsonov, S.A., and Ricard-Blum, S. (2018). Insights into the structure and dynamics of lysyl oxidase propeptide, a flexible protein with numerous partners. Sci. Rep. 8: 11768, https://doi.org/10.1038/s41598-018-30190-6.Search in Google Scholar PubMed PubMed Central

Veldkamp, C.T., Peterson, F.C., Pelzek, A.J., and Volkman, B.F. (2005). The monomer-dimer equilibrium of stromal cell-derived factor-1 (CXCL 12) is altered by pH, phosphate, sulfate, and heparin. Protein Sci. 14: 1071–1081, https://doi.org/10.1110/ps.041219505.Search in Google Scholar PubMed PubMed Central

Veverka, V., Henry, A.J., Slocombe, P.M., Ventom, A., Mulloy, B., Muskett, F.W., Muzylak, M., Greenslade, K., Moore, A., Zhang, Gong, J., et al.. (2009). Characterization of the structural features and interactions of sclerostin: molecular insight into a key regulator of Wnt-mediated bone formation. J. Biol. Chem., 284: 10890–10900, https://doi.org/10.1074/jbc.m807994200.Search in Google Scholar PubMed PubMed Central

Walpole, S., Monaco, S., Nepravishta, R., and Angulo, J. (2019). STD NMR as a technique for ligand screening and structural studies. Methods Enzymol. 615: 423–451, https://doi.org/10.1016/bs.mie.2018.08.018.Search in Google Scholar PubMed

Wise, E.R., Maltsev, S., Davies, M.E., Duer, M.J., Jaeger, C., Loveridge, N., Murray, R.C., and Reid, D.G. (2007). The organic-mineral interface in bone is predominantly polysaccharide. Chem. Mater. 19: 5055–5057, https://doi.org/10.1021/cm702054c.Search in Google Scholar

Xiong, X., Chen, Z., Cossins, B.P., Xu, Z., Shao, Q., Ding, K., Zhu, W., and Shi, J. (2015). Force fields and scoring functions for carbohydrate simulation. Carbohydr. Res. 401: 73–81, https://doi.org/10.1016/j.carres.2014.10.028.Search in Google Scholar PubMed

Xu, D., Arnold, K., and Liu, J. (2018). Using structurally defined oligosaccharides to understand the interactions between proteins and heparan sulfate. Curr. Opin. Struct. Biol. 50: 155–161, https://doi.org/10.1016/j.sbi.2018.04.003.Search in Google Scholar PubMed PubMed Central

Yang, J. and Chi, L. (2017). Characterization of structural motifs for interactions between glycosaminoglycans and proteins. Carbohydr. Res. 452: 54–63, https://doi.org/10.1016/j.carres.2017.10.008.Search in Google Scholar PubMed

Yang, J., Hsieh, P.H., Liu, X., Zhou, W., Zhang, X., Zhao, J., Xu, Y., Zhang, F., Linhardt, R.J., and Liu, J. (2017). Construction and characterisation of a heparan sulphate heptasaccharide microarray. Chem. Commun. 53: 1743–1746, https://doi.org/10.1039/c6cc08204a.Search in Google Scholar PubMed PubMed Central

Yang, J., Shen, M., Wen, H., Luo, Y., Huang, R., Rong, L., and Xie, J. (2020). Recent advance in delivery system and tissue engineering applications of chondroitin sulfate. Carbohydr. Polym. 230: 115650, https://doi.org/10.1016/j.carbpol.2019.115650.Search in Google Scholar PubMed

Yang, Z., Han, S., Keller, M., Kaiser, A., Bender, B.J., Bosse, M., Burkert, K., Kögler, L.M., Bernhardt, G., Plank, N., et al.. (2018). Structural basis of ligand binding modes at the neuropeptide Y Y1 receptor. Nature 556: 520–524, https://doi.org/10.1038/s41586-018-0046-x.Search in Google Scholar PubMed PubMed Central

Yu, F., Roy, S., Arevalo, E., Schaeck, J., Wang, J., Holte, K., Duffner, J., Gunay, N.S., Capila, I., and Kaundinya, G.V. (2014). Characterization of heparin-protein interaction by saturation transfer difference (STD) NMR. Anal. Bioanal. Chem. 406: 3079–3089, https://doi.org/10.1007/s00216-014-7729-4.Search in Google Scholar PubMed

Zdanov, A., Schalk-Hihi, C., Gustchina, A., Tsang, M., Weatherbee, J., and Wlodawer, A. (1995). Crystal structure of interleukin-10 reveals the functional dimer with an unexpected topological similarity to interferon γ. Structure 3: 591–601, https://doi.org/10.1016/s0969-2126(01)00193-9.Search in Google Scholar

Zhang, X., Lin, L., Huang, H., and Linhardt, R.J. (2020). Chemoenzymatic synthesis of glycosaminoglycans. Acc. Chem. Res. 53: 335–346, https://doi.org/10.1021/acs.accounts.9b00420.Search in Google Scholar PubMed

Ziarek, J.J., Veldkamp, C.T., Zhang, F., Murray, N.J., Kartz, G.A., Liang, X., Su, J., Baker, J.E., Linhardt, R.J., and Volkman, B.F. (2013). Heparin oligosaccharides inhibit chemokine (CXC motif) ligand 12 (CXCL12) cardioprotection by binding orthogonal to the dimerization interface, promoting oligomerization, and competing with the chemokine (CXC motif) receptor 4 (CXCR4) N terminus. J. Biol. Chem. 288: 737–746, https://doi.org/10.1074/jbc.m112.394064.Search in Google Scholar PubMed PubMed Central

Zong, C., Venot, A., Li, X., Lu, W., Xiao, W., Wilkes, J.L., Salanga, C.L., Handel, T.M., Wang, L., Wolfert, M.A., and Boons, G.J. (2017). Heparan sulfate microarray reveals that heparan sulfate-protein binding exhibits different ligand requirements. J. Am. Chem. Soc. 139: 9534–9543, https://doi.org/10.1021/jacs.7b01399.Search in Google Scholar PubMed PubMed Central

Received: 2021-01-21
Accepted: 2021-04-09
Published Online: 2021-04-21
Published in Print: 2021-10-26

© 2021 Georg Künze et al., published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Frontmatter
  2. Highlight: Extracellular Matrix Engineering for Advanced Therapies
  3. 12 years more than just skin and bones: the CRC Transregio 67
  4. Improvement of wound healing by the development of ECM-inspired biomaterial coatings and controlled protein release
  5. Modulation of macrophage functions by ECM-inspired wound dressings – a promising therapeutic approach for chronic wounds
  6. Fibrillar biopolymer-based scaffolds to study macrophage-fibroblast crosstalk in wound repair
  7. Collagen/glycosaminoglycan-based matrices for controlling skin cell responses
  8. Investigation of the structure of regulatory proteins interacting with glycosaminoglycans by combining NMR spectroscopy and molecular modeling – the beginning of a wonderful friendship
  9. Biodegradable macromers for implant bulk and surface engineering
  10. Insights into structure, affinity, specificity, and function of GAG-protein interactions through the chemoenzymatic preparation of defined sulfated oligohyaluronans
  11. Chemically modified glycosaminoglycan derivatives as building blocks for biomaterial coatings and hydrogels
  12. Men who stare at bone: multimodal monitoring of bone healing
  13. New insights into the role of glycosaminoglycans in the endosteal bone microenvironment
  14. Identification of intracellular glycosaminoglycan-interacting proteins by affinity purification mass spectrometry
  15. Structural insights into the modulation of PDGF/PDGFR-β complexation by hyaluronan derivatives
  16. Tuning the network charge of biohybrid hydrogel matrices to modulate the release of SDF-1
  17. Impact of binding mode of low-sulfated hyaluronan to 3D collagen matrices on its osteoinductive effect for human bone marrow stromal cells
https://doi.org/10.1515/hsz-2021-0119
Keywords for this article
CXCL12; CXCL14; interleukin-10; molecular dynamics simulation; protein-GAG complex; solution NMR spectroscopy
Creative Commons
BY 4.0

Articles in the same Issue

  1. Frontmatter
  2. Highlight: Extracellular Matrix Engineering for Advanced Therapies
  3. 12 years more than just skin and bones: the CRC Transregio 67
  4. Improvement of wound healing by the development of ECM-inspired biomaterial coatings and controlled protein release
  5. Modulation of macrophage functions by ECM-inspired wound dressings – a promising therapeutic approach for chronic wounds
  6. Fibrillar biopolymer-based scaffolds to study macrophage-fibroblast crosstalk in wound repair
  7. Collagen/glycosaminoglycan-based matrices for controlling skin cell responses
  8. Investigation of the structure of regulatory proteins interacting with glycosaminoglycans by combining NMR spectroscopy and molecular modeling – the beginning of a wonderful friendship
  9. Biodegradable macromers for implant bulk and surface engineering
  10. Insights into structure, affinity, specificity, and function of GAG-protein interactions through the chemoenzymatic preparation of defined sulfated oligohyaluronans
  11. Chemically modified glycosaminoglycan derivatives as building blocks for biomaterial coatings and hydrogels
  12. Men who stare at bone: multimodal monitoring of bone healing
  13. New insights into the role of glycosaminoglycans in the endosteal bone microenvironment
  14. Identification of intracellular glycosaminoglycan-interacting proteins by affinity purification mass spectrometry
  15. Structural insights into the modulation of PDGF/PDGFR-β complexation by hyaluronan derivatives
  16. Tuning the network charge of biohybrid hydrogel matrices to modulate the release of SDF-1
  17. Impact of binding mode of low-sulfated hyaluronan to 3D collagen matrices on its osteoinductive effect for human bone marrow stromal cells
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 26.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/hsz-2021-0119/html
Scroll to top button