Tuning cellular responses to BMP-2 with material surfaces

Abstract

Bone morphogenetic protein 2 (BMP-2) has been known for decades as a strong osteoinductive factor and for clinical applications is combined solely with collagen as carrier material. The growing concerns regarding side effects and the importance of BMP-2 in several developmental and physiological processes have raised the need to improve the design of materials by controlling BMP-2 presentation. Inspired by the natural cell environment, new material surfaces have been engineered and tailored to provide both physical and chemical cues that regulate BMP-2 activity. Here we describe surfaces designed to present BMP-2 to cells in a spatially and temporally controlled manner. This is achieved by trapping BMP-2 using physicochemical interactions, either covalently grafted or combined with other extracellular matrix components. In the near future, we anticipate that material science and biology will integrate and further develop tools for in vitro studies and potentially bring some of them toward in vivo applications.

1. Introduction

Bone morphogenetic protein 2 (BMP-2) is a multifunctional growth factor belonging to the transforming growth factor β (TGF-β) superfamily. It was identified in the 1970s as an essential molecule for de novo bone formation in adult animals [1, 2]. Indeed BMP-2 is one of the strongest osteoinductive factors known so far: it initiates the differentiation of mesenchymal stem cells (MSCs) into osteoblasts and chondrocytes in vivo and in vitro [3], as well as the transdifferentiation of muscle cells into bone cells [4, 5].

In view of its osteogenic potential, the clinical use of recombinant human BMP-2 (rhBMP-2), first purified in 1988 by Wang et al. [6], was approved in 2002 by the Food and Drug Administration (FDA) and validated by the European Medicines Agencies. To date, the only FDA-approved material carrier is an absorbable collagen sponge to which a high amount of rhBMP-2 is applied (up to 2.1 mg/level) [7], due to its poor affinity for collagen [8]. In clinical trials, it has been reported that up to 23% of patients suffered complications, such as hematomas and swelling in the neck and throat regions [9], dysphagia and a heightened risk of cancer [10]. In Europe, while the clinical use of rhBMP-2 as an adjunct to standard care has been approved, the increasing number adverse event reports and the growing socio-economic need for bone repair therapies raise the important question of how to develop effective materials which allow the control of the biological responses to BMP-2.

In the last decade, several studies have shown the possibility to deliver BMP-2 from various carrier materials [11–13], especially polymeric materials and ceramics. Since in vitro tests were promising and pre-clinical studies are currently being performed, it is likely that future medical devices containing new formulations of BMP-2 will be approved. However, it is still challenging to achieve a controlled presentation of BMP-2, while retaining its activity and minimizing the amount of protein applied locally. Standard biological studies stimulate cells with BMP-2 added to the culture media. In these cases high amounts of the growth factor are needed because of the limited lifetime of BMP-2 in solution. Additionally, this condition does not represent the natural cellular environment, since BMP-2, like other growth factors, is sequestered in the extracellular matrix (ECM) and released upon matrix degradation [14, 15]. Thus advanced biomaterials which take into consideration the physical and chemical complexity of the extracellular environmental are being developed. These materials could serve as a tool for biologists to unravel novel biological properties of BMP-2 which could not be explored so far with standard culture methods [4].

A timeline showing a few of the most important findings on BMP-2 in biological and material sciences is shown in Figure 1. It is noteworthy that approval for the in vivo use of BMP-2 (2003) took place before the development of advanced materials able to control and reduce BMP-2 release and before the discovery of new biological functions of BMP-2 such as its influence on the whole human body. Hence, there is now a great need to build an integrative approach including material science, chemistry, engineering, biochemistry and cell biology, and bridge the gap between these different disciplines. From the materials side, researchers could bring innovations in the design of materials for BMP-2 presentation by providing functionalization strategies and characterization methods as well as by developing new tools for the spatial control of BMP-2 delivery using micro- and nanotechnology approaches. From the biochemical and biological standpoint, researcher could provide new tools to produce BMP proteins, engineered mutants or tagged molecules.

Figure 1.

Time-line showing few of the most important findings on BMP-2 in biology (in red) and in material sciences (in blue). Fundamental biological discoveries such as the influence of BMP-2 in the whole human body and the development of advanced materials able to modulate the physicochemical presentation of BMP-2 followed the approval for the in vivo use of BMP-2 in 2002.

In this review, we first summarize the emerging functions of BMP-2 in cell biology and the resulting signaling responses at the interface between cells and their environment. We then present recent developments on engineered surfaces that aim at mimicking the presentation of BMP-2 in its natural environment. Finally, we discuss how specific properties of materials may help in optimizing existing systems or may bring new ideas for the design of innovative delivery systems.

2. Cell responses to soluble BMP-2

Although BMP-2 signaling has historically been linked to bone, the growing number of known BMPs functions in different tissues brought the biology community to coin a new term for all bone morphogenetic proteins: “body morphogenetic proteins” [16]. Figure 2 schematically illustrates the major steps for the BMP-mediated induction of osteogenic differentiation in bone progenitor cells and myoblasts, which transdifferentiate into osteoblasts upon BMP-2 stimulation [4]. BMP-2, like other members of the TGF-β superfamily, signals upon binding to two types of cell transmembrane serine/threonine kinase receptors, the BMP type I (BMPRI) and type II (BMPRII) receptors. The binding of BMP-2 to BMPRI results in the phosphorylation of SMAD1/5/8, which forms a complex with co-SMAD (SMAD4) and translocates to the nucleus [17]. For transcriptional signaling, this shuttling leads to a subsequent expression of transcription factors such as Id-1 and BMP-2 responsive element, typical markers of osteogenic differentiation [18]. At later time points, alkaline phosphatase (ALP) is expressed after several days, and mineralized matrix deposition is detected after several weeks of culture [3]. Besides the SMAD pathway, gene transcription is induced by BMP-2 via non-SMAD signaling as BMP induces the MAPK pathway, which leads to the expression of ALP, osteopontin and collagen I (for details about signaling, see review [19]). Regarding the non-transcriptional signaling mediated by BMP-2, recent studies have shown that BMPRs might control cytoskeletal rearrangements involved in cell migration [20–23]. The regulation of BMP signaling takes place at several levels, from receptor complex formation to crosstalk with other pathways, as will be discussed in the next paragraphs.

Figure 2.

Schematic representation of the major steps in the differentiation of bone progenitor cells and myoblasts over time. Note that the crosstalk with other signaling pathways and other relevant markers are omitted for simplicity. The relative size of all molecules is not drawn to scale.

2.1. Modulation of BMP-2 signaling at the cell surface

2.1.1. BMP receptor complex formation

Studies from P. Knaus group have indicated that BMP receptors present a distinct mode of oligomerization and activation [24]. For the formation of a functional signaling receptor complex, BMP-2 binds to BMPRI, which is either already organized in a receptor complex with BMPRII, or recruits BMPRII. These modes of oligomerization result in the activation of different signaling pathways: binding of BMP-2 to a pre-formed complex induces the classical SMAD signaling pathway, while ligand-induced oligomerization induces the non-SMAD pathway. So far, these events have been analyzed by applying biochemical separation of detergent-resistant membranes and co-immunoprecipitation methods [25]. There is still little information regarding the spatial arrangement of BMPRs at the nanoscale and the localization of the different complexes in distinct cellular compartments. Only recently was the spatial distribution of BMPRIb and BMPRII visualized using high-resolution imaging techniques. Using two-color Stimulated Depleted Emission (STED) microscopy (Figure 3A), single BMPRII appear to arrange sparsely, whereas BMPRI assemble in larger clusters comprised of multiple receptors [26]. When BMP-2 was added to the cell culture media, the BMPRII associated with the larger BMPRI assemblies at the cell periphery.

Figure 3.

A, Confocal microscopy (right) and STED microscopy (left) images of BMPRIb (in green) and BMPRII (in red). In the absence of BMP-2, the two different receptors rarely co-localized (upper white arrowhead) and BMPRII did not cluster (lower arrowhead). When cells were exposed to BMP-2, BMPRII associated with the larger BMPRIb assemblies. This different behavior could not be appreciated with confocal microscopy. Image adapted from [26]. B, Example of colocalization (indicated by arrows) of BMPRI and BMPRII (red) with αvβ5 integrins (green) detected by confocal microscopy. Images adapted from [45].

The lateral mobility of BMPRI and BMPRII is also very distinct, as shown by single particle tracking experiments: in fact BMPRI is very confined, both in presence or absence of the ligand, whereas the mobility of BMPRII can be either confined or free diffusing [27]. The preformed complex, which triggers the SMAD-dependent pathway, does not require the confined movement of BMPRI, while the non-SMAD seems to be highly dependent on the localization of BMPRI in membrane microdomains. Thus, non-SMAD signaling might require more stable complexes, possibly to allow interaction with other protein complexes, e.g. those involved in signaling to the cytoskeleton. To determine BMPR localization, the successful expression of tagged receptors has been possible for overexpression of human influenza hemagglutinin (HA)-tagged BMPRII [28] and it remains very challenging for BMPRI because of its low expression level. Tools are currently lacking in order to combine high-resolution approaches with studies on the dynamics of receptor complex formation and to identify the physical determinants of receptor mobility.

2.1.2. Receptor-ligand internalization

For BMP-mediated signaling the receptor complexes are internalized in two possible ways: (i) caveolae pits are formed for BMPRI and recruited BMPRII complex and activate non-SMAD pathways; (ii) clathrin-dependent internalization is required for the preformed receptor complex resulting in the activation of the SMAD pathway [29]. Different points of discussion have been raised regarding clathrin-dependent internalization of the ligand-receptor complex in growth factor signaling. The first point is whether receptor internalization is required for signaling. For tyrosine kinase receptors, such as vascular epidermal growth factor receptor 2 (VEGFR2) and epidermal growth factor receptor, clathrin-mediated endocytosis is important in regulating receptor recycling to modulate the amplitude of biological response [30–32]. VEGFR2 internalization is required for the activation of ERK1/2 signaling but dispensable for other signaling pathways [33]. For serine/threonine kinase receptors, such as BMPRs, recent studies combining confocal and atomic force microscopy (AFM) have indicated that BMP-2 signaling might already start in domains of the plasma membrane outside of clathrin-coated pits, where BMP-2 molecules bind to BMPRIa, which then phosphorylates and triggers SMAD signaling [34]. Regarding downstream signaling events, the treatment of cells with endocytosis inhibitors does not affect SMAD phosphorylation, while the downstream signal propagation is hindered [29, 35, 36]. Conversely, inhibition of BMP-2 endocytosis by an epigenetic approach actually elevates transcriptional responses [37]. Additionally, dynamin inhibition impairs osteogenic differentiation but does not block completely the transcriptional activation of several other genes, suggesting the presence of alternative SMAD-dependent signaling cascades which are independent of endocytosis [38].

These biochemical approaches to inhibit endocytosis lead to another point of discussion related to growth factor internalization. As of today, it remains elusive whether the ligand has to remain bound to the receptors and become internalized via the clathrin-mediated pathway, or if it would be sufficient to have trafficking of the activated receptors, regardless of ligand internalization. In 1997 Jortikka et al. [36] reported that bonds with carrier materials should not be tight nor in covalent form to allow endocytosis of BMP-2. However, recent studies demonstrated that anchorage of the growth factor to the ECM or to a surface still conveys signaling by prolonged activation of receptors and differential phosphorylation [39, 40]. Thus, ligand-receptor interaction at the cell membrane might be sufficient to obtain a sustained signaling response. It remains to be elucidated if a mechanical component causes deformation of the membrane and affects internalization signaling or if co-recruitment of other adhesion receptors such as integrins might occur in these cases where the BMP-2 molecules cannot be internalized.

2.2. BMP-2 signaling in a cell adhesion context

2.2.1. Osteogenic and adhesion signaling crosstalk

Extracellular factors orchestrate the commitment and differentiation of many cell types; in turn, a concerted action of adhesive and growth factor signals regulates adhesion and motility, which are mediated by interactions with the physical and biochemical cues from the environment. The signaling crosstalk between BMP-dependent and integrin-mediated pathways has been explored towards the modulation of both osteogenic differentiation and adhesion to the ECM [41]. Regarding the participation of integrin signaling in the transcription of genes for osteogenic differentiation, the collagen-binding integrins α₁β₁ and α₂β₁ regulate BMP-induced differentiation by acting downstream of BMPRI [42, 43]. Moreover, following binding to collagen, FAK phosphorylation is necessary for the transcriptional activity of SMAD6 but not for the translocation of SMAD1 [44]. α_v integrin also regulates BMP-dependent osteogenic differentiation [45], and in particular osteoblastic response to CYR61, a bone activator that increases the level of BMP-2 and activtes the α_vβ₃ integrin/ILK/ERK signaling pathway [46].

For the regulation of adhesion, as of today only few studies have shown the impact of BMP signaling on integrins and integrin-mediated structures. Lai et al. [45] reported that during 4 days stimulation of osteoblasts with BMP-2 in the media, the expression of α_v integrins is increased, BMPRs colocalize with α_v and β₁ integrins in focal adhesions (Figure 3B) and coprecipitate with these receptors. However, the colocalization pattern with vinculin, a structural protein present in focal adhesions, could not be confirmed by recent studies using high-resolution microscopy [26]. In osteoblasts, BMP-2 enhances the formation of focal adhesions and stress fibers by increasing α₅ and β₁ integrin expression, and triggers migration events by enhancing the incorporation of β₁ integrin into lipid rafts [47, 48].

In this context, there are still several key questions that remain unanswered and might add further complexity to the entire picture encompassing BMP and adhesion signaling. First, it should be elucidated where binding sites for integrins and BMPR are located relative to each other within the extracellular matrix. As a consequence, there is the need for a deep understanding of how BMPRs and integrins are spatially organized at the plasma membrane to allow both physical interactions and signaling crosstalk. Finally, it should be determined how multiple pathways modulating adhesion dynamics are regulated spatio-temporally.

2.2.2. Effects of BMP-2 on cytoskeleton assembly and cell migration

The evidence that BMP signaling is involved in the crosstalk with other pathways has brought to attention new functions of BMP-2, which are not necessarily related to its transcriptional signaling pathways. For example, BMP-2 signaling is involved in wound healing and cancer invasiveness by acting on actin cytoskeleton dynamics [49–51]. Upon BMP-binding to the BMPR complex, LIMK1 dissociates from BMPRII and phosphorylates cofilin [52]. The activation of LIMK1 by BMP-2 initiates the signaling to the cytoskeleton in a PI3K-dependent manner; a concomitant activity of Cdc42 is however required [21].

Hiepen et al. [22] have recently shown that a regulatory subunit of PI3K is essential in directed cell migration mediated by BMP-2 at the leading edge of migrating cells. BMP-2 also induces the activation of the p38/MK2/Hsp25 pathway at cortical actin protrusions in migrating cells [23]. To further add complexity, other signaling pathways independent from LIMK1 activation have been identified, where actomyosin assembly is mediated by ROCK1 kinase downstream of Rho GTPases and myosin light chain kinase [53]. Taken together, these studies clearly indicate that BMP-2 participates in the regulation of cell protrusion formation and migration, acting on multiple parallel pathways involved in actin reorganization. However, as for the interaction of the receptors at the plasma membrane, the spatio-temporal aspects of such regulation of signaling to the cytoskeleton still remain unclear.

These new and intriguing functions of BMP-2 are also relevant for the design of biomaterials/implants for the delivery of BMP-2, adhesion being the first step at the interface between cells and artificial materials. In turn, many answers to these open questions might come in the near future with the aid of material science approaches which allow control over the presentation of BMP-2 to cells.

3. Mimicking the BMP-2 microenvironment with material surfaces

Several growth factors are present in tissues in a matrix-bound form and released upon matrix degradation [14, 15]. The mode of presentation of BMP-2 at the interface with cells might be crucial in modulating its biological activity. For this reason, material surfaces applied to biological studies should mimic the physicochemical properties of the native ECM, to facilitate and allow predictions of cellular responses. In particular, using materials that enable the control of the amount of BMP-2 on their surface and its local distribution might help in determining the spatio-temporal regulation of BMP-2 signaling pathways.

In comparison with soluble BMP-2 (Figure 4A), the presentation of the growth factor on material surfaces could be tailored to achieve controlled immobilization and/or release of the protein from the surface (Figure 4B). This might lead to different signaling kinetics as well as the activation of alternative signaling pathways. Additionally, modifications in surface chemistry which allow the spatial control of BMP-2 (Figure 4C) could support the quantitative analysis of signaling events. Finally, surfaces where BMP-2 is presented together with ECM components (Figure 4D) could maintain or even enhance the biological activity of BMP-2 while possessing adhesive properties to allow the growth and colonization of cells.

Figure 4.

Schematic representation of different material surfaces approaches to control the presentation of BMP-2 at the interface with cells. A, BMP-2 is added to the cell culture media, which represents the standard stimulation way. B, BMP-2 either entrapped by electrostatic interactions (left) or chemically bound to the material surface (right). C, Surface patterning of BMP-2 for the spatial control of BMP-2 presentation. As an example, gradients of matrix-bound BMP-2 are schematically shown. D, Copresentation of BMP-2 and ECM components.

3.1. Temporal control of BMP-2 activity with material surfaces

In the design of materials aiming at achieving a time-controlled presentation of BMP-2, the growth factor can be immobilized on surfaces either by physical entrapment (i.e. electrostatic interaction, hydrophobic effect, hydrogen-bonds) which allows a slow release and internalization of the molecule, or by immobilization through a chemical linker or through biotin-Streptavidin (SAv) binding, which leads to a sustained presentation of BMP-2 (Figure 4B).

3.1.1. Physical entrapment of BMP-2

The formation of Layer-by-Layer (LbL) polyelectrolyte multilayer films is a method that allows the entrapment of BMP-2 over a long period of time (Figure 5A). LbL films are made of poly(L-lysine) (PLL) and Hyaluronic Acid (HA), which can be stabilized by covalent crosslinking with 1-ethyl-3-(-dimethylaminopropyl) carbodiimide (EDC). The films are post-loaded with BMP-2 by simple diffusion and retain the growth factor for at least 9 days [54]. The amount of retained BMP-2 can be tuned by varying film thickness and the initial concentration of BMP-2 in solution. For instance, a maximal value of 1.42 ± 0.26 μg/cm² can be trapped in 1.4 µm thick (PLL/HA) films when the initial concentration of BMP-2 in solution is 20 µg/mL. More recently, it was shown that the crosslinking extent of the film allows the control of the amount of BMP-2 remaining in the film after a burst release [55]. This burst release depends on the crosslinking extent (7-11% for the highly cross-linked film in comparison to 62-77% for the low crosslinked films). The final amount of BMP-2 retained in the film varied (between 4 and 14 µg/cm²) when the initial concentration was 100 µg/mL. A BMP-2 adsorbed amount of 800 ng/cm² was sufficient to trigger SMAD phosphorylation after 4 hours and ALP activity at 5 days in C2C12 cells [56]. In addition, BMP-2 loaded on soft films induced adhesion and spreading, in contrast to BMP-2 added in solution. Cells also formed focal adhesions in response to matrix-bound BMP-2, suggesting a possible crosstalk between BMP receptors and adhesion receptors (e.g. integrins) [56]. It should be noted that for this type of films a direct comparison of the surface concentration of BMP-2 and soluble concentrations is difficult due to the difference in dimensionality (matrix-bound versus soluble) and molecular diffusion.

Figure 5.

Examples of material surfaces applied for the control BMP-2 effect on cells. Top: schematic representation of the material design. Down: cellular response to the substrates A, Electrostatic entrapment on BMP-2 on polyelectrolyte multilayer films. C2C12 cells were plated on LbL soft films containing BMP-2 and stained for actin (red) and nucleus (DAPI, in blue). Figure adapted from [56]. B, Immobilization of b-BMP2 on Streptavidin gradient. Immunofluorescence images showing the nuclear translocation of the osteogenic marker Osterix, in cells grown on the BMP-2 modified surfaces. Image adapted from [62]. C, BMP-2 immobilized to gold nanoparticle arrays produced by block copolymer micellar nanolithography. The histogram shows a comparison of SMAD1/5/8 phosphorylation levels and kinetics in cells stimulated with 1 ng of BMP-2 either added to the culture media or bound to the nanoparticles. Image adapted from [40]. D, Heparin binding peptides immobilized on a SAM captures endogenous heparin and BMP-2 from serum. The histogram shows that hMSCs area significantly increases in cells adhering to the functionalized surfaces. Image adapted from [85].

The use of temperature-sensitive polymers is another manner to electrostatically entrap BMP-2 which is already applied in vivo [57]. The polymers can be formulated in aqueous buffers at a low temperature but become insoluble when delivered to the physiological milieu. A library of temperature-sensitive polymers has been created [58], however only a few of them were able to retain BMP-2 for more than 5 days after the in vivo injection.

Entrapment by LbL techniques may be easily adapted for in vivo applications and some promising results have already been obtained. Indeed, hydrolytically degradable LbL coating of implants [59] was used to entrap both BMP-2 and VEGF and induced de novo bone formation in 4-9 weeks. Interestingly, such surface coatings can be dried and sterilized, all the while preserving BMP-2 bioactivity [55]. Clinical applications of physical entrapment-based materials can be expected in the near future.

Since the physical entrapment-based techniques are quite versatile and do not require expensive equipment, they could represent an alternative surface material to study the temporal dependence of BMP-mediated signaling. In addition, the parameters of the microenvironment, such as stiffness or growth factors presentation, can be tuned in order to analyze their effects on the BMP-2 pathway. However the nature of adhesive interactions between cells and LbL films should be clarified in order to be able to distinguish between the mere contribution of BMP-2 to signaling from the possible contribution of adhesive receptors (e.g. integrins, HA receptors), which may induce secondary signaling pathways.

3.1.2. Chemical binding of BMP-2

Biotin-Streptavidin (SAv) is the strongest non-covalent bond, which can be used to immobilize a protein on a surface following its biotinylation. This method provides not only a stable binding but also a versatile platform on which it is possible to immobilize different biotinylated compounds [60]. The drawback consists in the need of two grafting steps, i.e. fist biotin moieties on the surface and then SAv, before growth factor immobilization. Amino-biotinylated BMP-2 added to culture media exhibits an increase in bioactivity, in contrast to carboxyl-biotinylated BMP-2 [61]. BMP-2 was amino-biotinylated and grafted on a poly(methyl methacrylate) (PMMA) thin film presenting a gradient of SAv in the range of 1.4 to 2.3 pmol/cm² [62] (Figure 5B). While the SAv concentration was measured by surface plasmon resonance, the binding of a BMP-2 dimer to a single SAv could be only estimated, based on the comparable size of the two proteins. However, this assumption does not consider variations in protein solubility due to aggregate formation, and the presence of non-bound biotin molecules which could change the 1:1 ratio between BMP-2 and SAv. With such an approach, a dose-dependent osteogenic response was measured on the same substrate over a period of 6 days. Neutravidin was used to immobilize biotinylated BMP-2 (b-BMP2) on biotinylated fibronectin (b-FN) [63] for studies on SMAD-dependent signaling and cell migration. By means of Quartz Crystal Microbalance with dissipation monitoring (QCM-D) and ELISA assays, and the amount of immobilized b-BMP-2 was detected on the surface for at least 6 days. While biotinylation is a relatively straightforward method to link proteins, it is however not site-specific and might negatively affect the biological activity of b-BMP2, when biotins are close to the BMPR binding sites. Moreover the immobilization through the small biotin moiety (~1 nm) might constrain BMP-2 spatial conformation, further inhibiting its recognition by cellular receptors.

For decades the use of covalently immobilized growth factors has been a matter of debate because of its negative impact on receptor binding and complex formation, as well as on the internalization of the protein, as discussed in section 2.1.2. To achieve covalent binding of growth factors to supporting materials, several approaches have been developed and the use of bifunctional linkers, which target either the amino- or the carboxy-groups of the protein, is the most commonly used. Such linkers are either pre-coupled to the growth factor and then immobilized on the surface, or are at first immobilized onto the surface and then the growth factor is immobilized in a second step [64, 65]. While the former has the advantage of involving fewer preparation steps, the latter appears to be advantageous to avoid protein denaturation due to unspecific interactions with the material surface [66]. Additionally, the use of molecular linkers, which confer a certain degree of flexibility to the tethered growth factor, may have an impact on the mobility and accessibility of the protein for receptor binding, without loss due to diffusion. BMP-2 has been immobilized covalently to gold surfaces via a self-assembled monolayer (SAM) consisting of 11-mercaptoundecanoyl N-hydroxysuccinimide ester which binds to the free amine residues of the protein and remains bioactive for a period of 6 days without being internalized [39].

Besides the difficulties in performing and controlling the different steps for the covalent immobilization, as well as in tailoring the immobilization strategies to the specific growth factor, a remaining challenge is to control the exact number of immobilized molecules. Thus, alternative approaches such as protein modification by expression of artificial domains or peptide tags, e.g. his-tags, have been also developed [67]. So far, the biological activity of BMP-2 is often affected by such a modification in comparison with the native protein.

3.2. Surface patterning for the spatial control of BMP-2 presentation

To achieve control over the distribution and amount of proteins presented on materials, various strategies for surface patterning at different length scales have been developed over the last two decades. A few examples showing the patterning of BMP-2 from sub-millimeter down to nanoscale are described in the following paragraphs. These approaches may help in the future in improving the design of biomaterials as well as in deciphering BMP-2 signaling pathways (Figure 4C).

3.2.1. Sub-millimeter patterning of BMP-2

During morphogenesis, an essential long range BMP-2 gradient is formed along the ventral to dorsal axis [68]. In vitro mimicry of long-range gradients or spatially organized tissues may help deciphering the pathways of BMP-2 signaling underlying tissue formation and spatial organization. By taking advantage of the natural affinity of FN for BMP-2 (see part 3.3), Miller et al. created millimeter-sized BMP-2 patterns by printing the growth factor as a “bioink” on fibrin [69]. This technique is versatile as it is possible to form patterns of various sizes and shapes, as well as gradients 1.5 mm long with different amounts of BMP-2 (from ~0.02 to ~2.245 μg/cm²) that are deposited by overprinting BMP-2 at the same location. These BMP-2 patterns were shown to be bioactive, as assessed by ALP expression in two different cell types, namely C2C12 myoblasts and mesenchymal fibroblasts.

Another strategy consists in using microfluidics in combination with LbL technology to create millimeter-sized gradients of matrix-bound BMP-2 [70]. To this end, a microfluidic chamber was set in contact with a PLL/HA film and a BMP-2 gradient in solution was generated via passive flow pumping. As the amount of BMP-2 adsorbed onto the film directly depends on the BMP-2 concentration in solution in the channel [54], a 40 mm-long gradient of matrix-bound BMP-2, ranging from 0.04 µg/cm² to 2 µg/cm², was thus generated. BMP-2 remained bioactive after 3 days as assessed by ALP activity in C2C12 myoblasts. This matrix-bound BMP-2 enabled the generation of a spatially controlled osteogenic differentiation, confined to the patterned area and dependent on the amount of BMP-2. Such patterns may be further used to create microtissues for studies on the effects of specific gene mutations or drugs on the formation and maintenance of bone tissues.

3.2.2. Micrometer-sized patterns of BMP-2 on surfaces

BMP-2 patterned at the micron scale allows studies on single-cell responses. To this end, by using microcontact printing with a poly(dimethylsiloxane) stamp, Hauff et al. created 25 µm-wide patterned stripes of FN onto which BMP-2 was then immobilized by biotin-neutravidin binding [63]. These patterns are stable for at least one day and b-BMP2 is not released from the stripes. Because of the discrete localization of BMP-2 molecules on the stripes, the amount of the immobilized protein on the surface is relatively high (0.52 μg/cm²). The grafted BMP-2 triggers SMAD1/5/8 phosphorylation and inhibits myotube formation in C2C12 cells. Interestingly, in comparison with samples where BMP-2 was added to the culture medium, SMAD phosphorylation is prolonged over a period of 90 min, leading to a sustained localization of the SMAD complex in the nucleus. In this regard, it remains to be elucidated whether the prolonged SMAD-signaling might impact other BMP-mediated pathways. These patterned stripes served also as platform to study directed cell migration: while the migration velocity seems independent of the immobilization of BMP-2 on the patterned stripes, cells do not show any preference for a direction on the immobilized BMP-2. These results suggest it is not the binding of BMP-2 to the extracellular matrix, but rather the presentation of the proteins in gradients that might be therefore necessary to guide migration. So far, continuous surface chemical gradients of BMP-2 have been applied to study the effects of different amounts of surface-immobilized BMP-2 on cell differentiation [62]. However, such gradients might not be steep enough to induce migratory responses. This still leaves the challenge of creating surfaces that could serve as platforms to decipher the haptotactic function of BMP-2 gradients and to study possible differences with chemotactic gradients in BMP-induced migration signaling.

To uncouple total surface density from localized density of BMP-2, microcontact printing or dip-pen nanolithography were used to produce circular micropatterned islands of BMP-2 having a diameter of 4-5 µm [71]. The latter technique is based on the use of an AFM tip to deposit molecules on the surface as an ink droplet while varying the spacing between the islands. For the chemical binding of BMP-2 to the surface, either a thiolated biotin linker or a thiolated biotin lipid layer was first placed on gold-coated substrates using the micropatterning approaches. Following incubation with SAv, b-BMP2 was immobilized on the patterned regions and remained bioactive. Cell differentiation was comparable to non-patterned BMP-2 on the surface, when taking into account the estimated total surface density of the protein. When considering the impact of the local density of BMP-2 on cell response, these studies suggest that BMPR oligomerization might be favored when the growth factor is presented in discrete regions, thus leading to more efficient signaling, but this remains to be elucidated.

3.2.3. Nanoscale surface patterning of BMP-2

Materials which allow the control of cell responses at the nanoscale are of special interest, being at the length scale of BMP-2 and BMPRI and II interactions. Nanoscale modifications of surfaces carrying BMP-2 have been applied to study the influence of substrate modifications on osteogenic differentiation by changing their physicochemical properties [72], or for determining the effect of the surface density of BMP-2 on cell signaling [40]. In the first case, nanogrooves and nanodots ranging between 150-300 nm and 460 nm in size, respectively, consisting of polyurethane acrylate and coated with poly(glycidyl methacrylate), were functionalized with BMP-2 peptides. The presence of nanoscale features on the surface improves calcium deposition and the expression of osteogenic markers, which are even enhanced in presence of BMP-2 peptides. Better tuning of the nanostructure size to allow the formation of focal adhesions [73] and quantifying the amount of BMP-2 peptides immobilized on the surface may help to further use these new nanotopography tools to study BMP-mediated signaling

To achieve control over BMP-2 surface density, gold nanostructured substrates produced by block copolymer micellar nanolithography were recently applied as a substrate for the immobilization of BMP-2 using a bifunctional linker, as described in 3.1.2, now coupled to gold nanoparticles (Figure 5C) [40]. The coupling of BMP-2 heterodimers to every single nanoparticle on the surface was detected and quantified at the single molecule level by AFM, thus making it possible to experimentally determine the amount of immobilized growth factor on the surface. Additionally, with this nanopatterning technique it is possible to vary the amount of immobilized BMP-2 by varying the interparticle spacing and to achieve the controlled immobilization of amounts which are below the lowest value reported previously (31 ng/cm²) [74]. Interestingly, the bioactivity of the immobilized protein shows a characteristic regulation of SMAD phosphorylation levels and kinetics, which differs from those triggered by BMP-2 added to the cell culture medium. In fact, when BMP-2 is immobilized on the surface, regardless of the amount used (ranging from 0.2 to 3.3 ng/cm²), SMAD phosphorylation onset is delayed but then is still maintained over a long period of time (180 min). Additionally, while the lowest amount of BMP-2 added to the culture media is not sufficient to activate the SMAD complex, the corresponding concentration immobilized on the surface leads to a remarkable SMAD phosphorylation. This study indicates that the sustained presentation rather than the amount of BMP-2 regulates SMAD-signaling, suggesting a different temporal regulation of BMP-mediated signaling pathways when the growth factor cannot be internalized. One hypothesis is that the immobilization might affect lateral receptor mobility and oligomerization on the one hand. On the other hand, when the receptors cannot be internalized in a complex with the ligands, the number activated receptors and their internalization rates might be different than those in presence of BMP-2 in the media.

3.3. Materials inspired by the interaction of BMP-2 with ECM components

One of the ECM functions is to serve as a reservoir of growth factors via a large variety of interactions (for example electrostastic, hydrogen-bonds, hydrophobic, Van der Waals). This type of interaction is important for growth factor release in soluble phase, orientation and therefore signaling. ECM presents epitopes which bind growth factors to limit their diffusion and maintain their activity locally. Therefore, the incorporation of BMP-2 binding sites of the ECM on materials would permit BMP-2 sequestration in a non-covalent manner [75] (Figure 4D).

3.3.1. Modulation of the activity of BMP-2 bound to glycosaminoglycans

Glycosaminogycans (GAG) are major polysaccharide components of the ECM. These biopolymers can be divided into four groups: HA, the only not sulfated, heparin sulfate (HS), chondroitin sulfate (CS) and dermatan sulfate (DS). GAGs bind growth factors with a low binding constant [76, 77], mainly due to electrostatic interactions. It has been shown that BMP-2-GAG binding could either up- or downregulate BMP-2 cellular activity [78–80]. Ruppert et al. [14] demonstrated that the BMP-2 homodimer has an heparin-binding site at its N-terminus. The binding seems to be due to the interactions between the basic residues of the Hp-binding site and the sulfate groups presented on GAGs. In particular, it has been demonstrated by surface plasmon resonance that GAGs alter the binding between BMP-2 and its receptor IA in a sulfation-dependent manner [81].

Hp can be used as a material coating to present BMP-2 to cells. For example titanium substrates modified with Hp to present BMP-2 promote osteoblast function, osteointegration, and bone regeneration in vitro and in vivo [82, 83]. Resorbable polymer (poly(L-lactic acid) and poly(ε-caprolactone) films, covalently functionalized with oriented Hp, linked via reductive amination, immobilize BMP-2 and improve cell attachment and proliferation [84]. However, the surface functionalization with Hp is only qualitative: with these materials it is not possible to achieve a precise quantification of both GAG and growth factor and to characterize the BMP-2 release during cell culture. Moreover, changes in mechanical properties after Hp coating might also influence cell behavior by changing cell-substrate forces and activating cytoskeleton rearrangements.

A different way to exploit the use of surfaces functionalized with Hp to bind growth factors has been proposed by Hudalla et al. [85] (Figure 5D). Here a SAM presenting Hp-binding peptides and RGD peptides was used to specifically bind the endogenous Hp, which is complexed with the growth factors present in the cell medium. Thanks to the inert SAM background, these surfaces avoid the non-specific binding of other components of the serum and reduce the need of high non-physiological concentrations of growth factors. Human MSCs plated on SAM substrates show an enhancement of the BMP signaling pathway, and therefore an enhanced cellular proliferation and osteogenic differentiation. Hp and HS have been extensively used in 3D scaffolds due to their synergistic effect on BMP-2 activity. Other reviews extensively report the use of Hp for drug delivery and for in vivo applications [75, 86]. Even though promising for clinical applications, these approaches do not provide structural and stoichiometric information of the GAG/BMP-2 binding. The combined effects of GAGs on BMP-2 cellular responses could lead to the hypothesis that HS proteoglycans function as co-receptors for BMP-2. Thus, it is crucial to get clear information on the GAGs structural and conformational modifications after BMP-2 binding. Functionalized surfaces together with surface-sensitive techniques could provide useful tools for answering this question.

A detailed surface-based study was proposed by the Svedhem group [87] using CS, an important structural ECM component. CS, covalently attached to supported lipid bilayers, binds BMP-2 and cells spread in response to BMP-2. Although the bioactivity of BMP-2 in these conditions was not verified, this type of model assembly opens new possibilities for the study of BMP-2 interactions with biopolymers in controlled environments.

HA is a GAG which possesses highly interesting physical and mechanical properties. By interacting with water molecules, HA provides the tissue with the ability to resist compression stresses [88]. HA alone has a positive effect on cell proliferation and upregulates osteogenic markers [89]. Several studies have described that BMP-2 can be trapped in HA crosslinked gels [8, 90–92], and its retention can be improved using different strategies. For instance, Kisiel et al. [93] pre-complexed BMP-2 with DS or with Hp to increase its affinity for HA. Thus it is possible to load three times more pre-complexed BMP-2 in HA hydrogels than free BMP-2. The retention of the pre-complexed BMP-2 is significantly higher than free-BMP-2 on HA gels after 30 days. Alternatively, HA can be chemically modified to better retain BMP-2. For instance, bisphosphonates can be grafted onto HA, which leads to a 8-fold increase in the retention capacity for BMP-2 in comparison with pure HA gels [94]. HA can also be associated in LbL with positively-charged polypeptides to create thin self-assembled films that can be deposited on material surfaces, as described in 3.1.1 [54]. This presentation mode maintains the biological activity of BMP-2, as confirmed by SMAD phosphorylation. Interestingly, the phosphorylation signal is increased in cells cultured on the matrix-bound BMP-2 soft films in comparison with the stiff ones [56]. Indeed these biomimetic substrates combine both physical and chemical cues, thereby opening new possibilities to investigate the importance of BMP-2 in mechanotransduction. In fact, BMP signaling appears to be closely connected to mechanotransduction pathways at several levels. During embryonic development , for example, both BMP-2 gradients and mechanical signals such as tissue stiffness and compressive forces contribute to tissue polarity and patterning [95], although a deep understanding of the exact mechanisms is still missing.

3.3.2. Co-presentation of BMP-2 and cell binding motifs

Many efforts have been taken to engineer the environment so that it is supportive of both adhesion and differentiation in a controlled manner. However, the presentation of multiple and defined cues at the cell-material interface is still a challenge and so far the main focus has been on the effects on long-term responses and in vivo applications, whereas information on the signaling pathways and crosstalk is still missing. An emerging approach is the co-presentation of integrin-binding motifs and BMP-2 (Figure 4D). Here, adhesion peptides such as RGD or collagen peptides are immobilized on the material surface to induce integrin-mediated adhesion [96–99].

Several studies from the Hubbell group have demonstrated that BMP-2 binds to ECM proteins like FN [100], tenascin C [101], fibrinogen, but not to collagen I [97, 102]. In particular, FNIII_12-14 binds BMP-2 and other growth factors in a promiscuous manner, with a K_D in the nanomolar range and without affecting the biological activity of the growth factors [100]. Engineered substrates made with fibrin molecule carrying a peptide containing FNIII_12-14 permit a greater retention of the factors with respect to normal fibrin matrices. Fibrin matrix itself and its heparin-binding domain could promiscuously bind several growth factors, including BMP-2 [102]. Fibrin-synthetic matrices presenting both the fibrin heparin-binding domain inside a polymeric scaffold and growth factors, like fibroblast growth factor-2 (FGF-2) and Platelet-derived growth factor-2 (PDGF-2), have been successfully tested in vivo. The proximity between the RGD motif present in FNIII_9-10 and the growth factor binding site on FNIII_12-14 serves as rationale for the use of such peptides to allow synergy with BMP-2 and potentiate bone formation [100]. The synergistic interaction between immobilized collagen I and BMP-2 in osteogenic differentiation of MSCs has recently been investigated using a microcontact printing platform [103].

To achieve more defined responses, the immobilization of adhesive motifs and BMP-derived peptides on material surfaces have been also performed. The immobilization of BMP-peptides has been applied to various materials, including polymers and hydrogels, but here we will focus on two examples where the molecules have been grafted onto 2D surfaces. Zouani et al. 2010 [104] grafted RGD and BMP-2 mimetic peptides on polyethylene terephthalate to enhance osteogenic differentiation. The impact on osteogenic differentiation of the co-presentation of RGD and BMP-bioactive peptides carrying an azide group has been also investigated at concentration gradients on self-assembled monolayers generated by UVO treatment [105]. Osteopontin and BMP-2-derived motifs have been also immobilized by engineering a cysteine residue and 12-aminoacid stretch switch tag to address the C-terminus of the peptides [106]. These strategies rely on the use of BMP-derived peptides based on the sequence of the knuckle epitope of a BMP-2 monomer comprising the low affinity site for binding to BMPRII [107]. However, this is in contrast with biochemical studies showing that two knuckle epitopes should be present on one BMP-2 molecule in order to achieve receptor activation, since depletion of a single epitope results in complete loss of ALP activation [108]. This leaves the question whether the surface immobilization strategies might unveil otherwise masked activities of the BMP-2 molecule which are not possible to investigate with BMP-2 in solution.

4. Concluding remarks and perspectives

Recently new aspects in BMP-mediated signaling have been unraveled, pointing out the need to design and develop new approaches for BMP-2 delivery. In view of future clinical applications, some critical questions regarding BMP-2 presentation and functions remain to be solved in order to provide innovative solutions for bone tissue engineering. It is therefore important to engineer materials that can present BMP-2 in a spatially and temporally controlled manner.

In this review, we have shown that several technical solutions have now been developed to present BMP-2 in a controlled manner to cells, using either covalent grafting, physical entrapment or interactions with ECM components, which precisely tune the activity of BMP-2 and control its orientation. The use of two-dimensional surfaces offers the advantage of being controllable with surface-sensitive techniques and compatible with high resolution microscopy. Some of the technical approaches here described, such as physical entrapment of BMP-2 and GAG-based materials, might be soon applied to scaffolds for tissue engineering applications. In vitro studies with BMP-2-presenting surfaces could allow the deciphering of hidden biological functions of BMP-2. For instance, materials on which BMP-2 and ECM ligands (adhesion ligand and/or GAGs) are co-presented in a spatially controlled manner could provide important information on the crosstalk between adhesion (e.g. integrins) and BMP-2 signaling pathways. Super-resolution microscopy techniques could be helpful to clarify the interactions at the cell membrane between BMP-2 and its receptors, explaining the dynamics of receptor recruitment and mobility, as well as the architecture of receptor complexes. By combining the spatio-temporal control over BMP-2 presentation on surfaces and high-resolution imaging techniques it should also be possible to elucidate the regulation of BMP-2 receptor endocytosis and its impact on signaling pathways. Certainly there is a need to develop labeling strategies to track BMP-2, as well as BMPRs, without affecting their biological activity and signaling kinetics. Recent attempts have shown that BMP-2 activity is significantly slower when fluorophores are coupled to the growth factors [109]. Thus, the development of new biochemical tools becomes essential: for example, the conjugation of BMP-2 to various types of linkers should be in a site-specific manner, to permit the control of its orientation, once grafted on surfaces, and to improve the bioactivity of covalently-grafted BMP-2. Moreover, biochemical and structural studies at the molecular level could also help in improving our knowledge of the mechanisms of BMP-2 binding to GAGs and to ECM proteins, which is largely incomplete at present.

In conclusion, innovative solutions in bone regenerative medicine are needed to repair critical bone defects. Surface materials with controlled delivery and presentation of BMP-2 can be used to direct cell signaling for bone repair. In the future, through a joint effort from material and biological sciences, it should be possible to further improve the presentation of BMP-2 at the cell surface. The knowledge gained from in vitro studies, using well-defined materials platforms, may open new ways for regenerative therapies.

Abbreviations

2(3)D

Two(Three) dimensional

ALP

Alkaline phosphatase

BMP-2

Bone morphogenetic protein

b-BMP2

biotinylated BMP-2

rhBMP-2

Recombinant human BMP-2

BMPRI(II)

BMP type I (II) receptors

Cdc42

Cell division control protein 42 homolog

CS

Chondroitin sulfate

C2C12

Mouse myoblast cell line

ECM

Extracellular Matrix

ELISA

Enzyme-linked immunosorbent assay

ERK

Extracellular signal-regulated kinases

FAK

focal adhesion kinase

FN

Fibronectin

GAGs

Glycosaminoglycans

GTPase

guanosine triphosphate hydrolase

HA

Hyaluronic acid

Hp

Heparin

HS

Heparan sulfate

Id-1

Inhibitor of DNA binding 1

ILK

integrin-linked kinase

LbL

Layer-by-Layer

LIMK1

LIM domain kinase 1

MAPK

Mitogen-activated protein kinase

PI3K

Phosphoinositide 3-kinase

PLL

Poly-L-lysine

RGD

Arginine, Glycine, Aspartic Acid

Rho

Ras homolog gene family

ROCK

Rho-associated protein kinase

SAM

Self assembly monolayer

SAv

Streptavidin

SMAD

small mothers against decapentaplegic

TGF-β

Transforming growth factor beta

VEGFR

Vascular epidermal growth factor