Hydrolytically Labile Linkers Regulate Release and Activity of Human Bone Morphogenetic Protein-6.

Release of growth factors while simultaneously maintaining their full biological activity over a period of days to weeks is an important issue in controlled drug delivery and in tissue engineering. In addition, the selected strategy to immobilize growth factors largely determines their biological activity. Silica surfaces derivatized with glycidyloxy propyl trimethoxysilane and poly(glycidyl methacrylate) brushes yielded epoxide-functionalized surfaces onto which human bone morphogenetic protein-6 (hBMP-6) was immobilized giving stable secondary amine bonds. The biological activity of hBMP-6 was unleashed by hydrolysis of the surface siloxane and ester bonds. We demonstrate that this type of labile bonding strategy can be applied to biomaterial surfaces with relatively simple and biocompatible chemistry, such as siloxane, ester, and imine bonds. Our data indicates that the use of differential hydrolytically labile linkers is a versatile method for functionalization of biomaterials with a variety of growth factors providing control over their biological activity.

Introduction

Growth factors (GFs) are endogenously present either bound to the extracellular matrix (ECM)[1] or as soluble molecules secreted by cells. Bound GFs, as opposed to their soluble analogues, provide control over their local concentration and protection from enzymatic degradation.[2] Many growth factors are bound in an inactive conformation to the ECM. Upon proteolytic cleavage, these growth factors become active and are released in the extracellular space. An example of such a GF is the Bone Morphogenetic Protein-6 (n class="Gene">hBMP-6) growth factor, which is a member of the transforming growth factor-β (TGF-β) superfamily. class="Chemical">pan class="Gene">hBMP-6 is stored in the ECM as an inactive precursor molecule. Upon proteolytic cleavage, active, mature hBMP-6 is released from the degrading ECM which in turn stimulates cells to repair the lost tissue.[3] BMP-6 is known to regulate differentiation of various cells implicated in cartilage and bone formation during skeletal development and fracture repair.[4−6] This is exemplified by the hBMP-6 mediated induction of alkaline phosphatase activity in osteoblast precursor cells, a hallmark of osteogenic differentiation.[7]

Throughout the past decade, many efforts have been devoted to develop methods to immobilize GFs at biomaterial surfaces.[8−10] The main aim of these efforts is to mimic the endogenous GF delivery system and thereby improving the performance of biomaterials in tissue engineering and therapeutic applications.[8−10] For example, n class="Gene">hBMP-2 has been mixed with biomaterials to be released by diffuclass="Chemical">pan class="Chemical">sion and thereby activating cell differentiation.[11,12] Other approaches have been focusing on linking ECM elements, such as heparins,[13−15] or mimics thereof,[16−20] to biomaterials in order to position growth factors on their surfaces through heparin’s growth factor binding domains. Single-domain antibodies have also been employed to reversibly link growth factors to surfaces,[21] or alternatively, direct coupling of growth factors has also been pursued using noncovalent His-tag technology.[22] Covalent chemistry yields long-term presentation of growth factors,[23,24] which continuously engage in receptor binding and protein dimerization, and hamper internalization of growth factors.[25]

Here, we investigate the effect on cells when hydrolytically labile n class="Chemical">siloxane and class="Chemical">pan class="Chemical">imine bonds are used as a linkage between the surface and the GF. Such chemistry has been successfully applied for the controlled formation of for example reversible gels, polymers and functional surfaces.[26−28] Siloxane and imine bonds are attractive as they can be hydrolyzed under physiological conditions.[29−34] Klok and co-workers have studied the stability of trialkoxysilane-anchored brushes and found that brushes detached from the surface upon prolonged exposure to cell culture medium.[33b] Using reversible imine bonds, biologically active molecules were cross-linked in a hydrophilic polymer network.[29] Differences in the percentage of released small molecules as well as the duration of that release were strictly depended on the percentage of the entrapped compound in the solvent of the gel.[29] In addition, reversible imine bonds were used to fabricate various mixed surface patterns and gradients to control surface compositions of small molecules and their subsequent release.[31,32] Despite the potential of adopting reversible chemistry for constructing and functionalizing materials and surfaces, to date these types of reversible siloxane and imine bonds have not been explored for the delivery of GFs. We study the effect on cell response of releasing hBMP-6 through hydrolytical cleavage. Release can be slowed down by creating a polymer brush. The data suggests the released growth factors are bioactive.

Results and Disucssion

The Immobilization of hBMP-6 onto Epoxide-Coated Surfaces and Its Cell Signaling Capability

Monolayers of hBMP-6

To establish the method of hydrolytically relean class="Chemical">sing class="Chemical">pan class="Gene">hBMP-6, we started with model studies on freshly prepared epoxide-terminated glass slides. Deposition of (3-glycidyloxypropyl)trimethoxysilane (GPTMS, Figure a) successfully yielded monolayers, evident from the recorded advancing contact angle value of 55° ± 2° and the increase in film thickness of 0.9 nm ± 0.2 nm using ellipsometry, which is in agreement with literature (Table ).[33,34] hBMP-6 was covalently immobilized to epoxide-functionalized substrates by reactive microcontact printing (μCP)[35] (Figure a) through a nucleophilic ring-opening reaction, presumably with primary amine groups of the lysine residues that are localized at the periphery of the dimeric GF structure. Immobilization of hBMP-6 led to an increase of the advancing contact angle value from 55° ± 2° to 64° ± 5°, to the appearance of an N1s peak in the XPS spectrum indicating the presence of ca. 1% N and to a further increase in film thickness of 1 nm as observed by ellipsometry (Table ). Inspection of the monolayers using atomic force microscopy (AFM) (Figure b, c) revealed clear patterns of effectively immobilized hBMP-6 with on average nearly 70% of the total area of the pattern occupied with protein. A height of 0.5 nm between the protein patterns and the background was observed (Figure S1), which is lower than observed using ellipsometry presumably due to tip-induced compression of the proteins. These results indicate that hBMP-6 is immobilized in a moderately packed, single layer configuration, which amounts to ca. 120 ng/cm2.

Figure 1

(a) Schematic overview of the preparation of epoxide-terminated glass substrates with two linkers of different length that have a difference in hydrolytic release: (3-glycidyloxypropyl)trimethoxysilane (GPTMS) (top) and poly(glycidyl methacrylate) (PGMA) (bottom) and subsequent immobilization of hBMP-6 (accessible lysine residues are indicated in dark gray) by reactive microcontact printing (μCP) (atom transfer radical polymerization (ATRP)). (b) Atomic force microscopy (AFM) phase image of hBMP-6 micropatterned (10 μm diameter dots and 5 μm spacing) on an epoxide-terminated surface without using a passivated background. (c) Zoom in of the phase image. (d) Fluorescence microscopy image (inset shows the intensity profile) of immunostained μCP patterns of hBMP-6 (100 μm diameter dots and 100 μm spacing) on an epoxide-terminated surface using a background passivated by NH2-PEG (see the Supporting Information for details).

Table 1

Advancing (θa) and Receding (θb) Water Contact Angle Values, Ellipsometric Thicknesses, and N Fractions Determined by XPS of the Two Epoxide-Terminated Linkers before and after the Immobilization of hBMP-6

monolayer (full)	θa [deg]	θb [deg]	ellipsometric thickness [nm]	N [%] (XPS)
GPTMS	55 ± 2	40 ± 5	0.9 ± 0.2	–a
GPTMS + hBMP-6	64 ± 5	47 ± 2	1.9 ± 0.2	1.0 ± 0.3
PGMA	60 ± 1	52 ± 4	80 ± 3	–a
PGMA + hBMP-6	65 ± 1	41 ± 4	87 ± 2	0.7 ± 0.2

Not detected.

(a) Schematic overview of the preparation of n class="Chemical">epoxide-terminated glass substrates with two linkers of different length that have a difference in hydrolytic release: class="Chemical">pan class="Chemical">(3-glycidyloxypropyl)trimethoxysilane (GPTMS) (top) and poly(glycidyl methacrylate) (PGMA) (bottom) and subsequent immobilization of hBMP-6 (accessible lysine residues are indicated in dark gray) by reactive microcontact printing (μCP) (atom transfer radical polymerization (ATRP)). (b) Atomic force microscopy (AFM) phase image of hBMP-6 micropatterned (10 μm diameter dots and 5 μm spacing) on an epoxide-terminated surface without using a passivated background. (c) Zoom in of the phase image. (d) Fluorescence microscopy image (inset shows the intensity profile) of immunostained μCP patterns of hBMP-6 (100 μm diameter dots and 100 μm spacing) on an epoxide-terminated surface using a background passivated by NH2-PEG (see the Supporting Information for details).

Induction of Osteogenic Differentiation

Immobilized n class="Gene">hBMP-6 was also visualized uclass="Chemical">pan class="Chemical">sing immunofluorescence staining of the surface. The recorded fluorescence micrograph (Figure d) showed clearly the presence of hBMP-6 at the surface. This result indicates that the immobilized hBMP-6 had conceivably retained the GF secondary structure, which is essential for antibody binding. The covalent immobilization appeared to be specific to the epoxide SAM because in controls with preblocking the epoxide groups with BSA or incubation on a glass surface, the immunofluorescence staining was absent. To determine whether immobilized hBMP-6 remained biologically active on the surface, a mesenchymal precursor cell line (KS483) was seeded for 24 h on the surfaces. Cells adhered on glass in the absence of hBMP-6, while addition of hBMP-6 or immobilized hBMP-6 on GPTMS layers did not lead to significant changes in cell spreading kinetics, as expected (Figure S2). Actin cytoskeleton stains showed no significant differences in cell area and stress fiber formation (Figure S2). Confluency was reached after 4 days, and no cytotoxicity of the surfaces was observed (Figure S2). KS483 cells can differentiate into osteoblasts during a 1–3 week culture period under osteogenesis-inducing conditions and this can be shown by the induction of alkaline phosphatase (ALP) as early marker of osteogenic differentiation. Mature human and mouse BMP-6 share 96% amino acid identity, and human BMPs cross react with mouse BMP receptors. This cell model is therefore suitable to assess the ability of hBMP-6 to initiate differentiation into mineralizing osteoblasts.[21,36] KS483 cells were seeded on immobilized hBMP-6 coated surfaces and both an immunofluorescence hBMP-6 stain and an immunohistochemical ALP stain were clearly visible in contrast to control surfaces with immobilized polyethyleneglycol (Figure a, b). These results indicate that only in the areas where functional hBMP-6 was available, ALP production was induced. The biological activity of immobilized hBMP-6 and supplemented hBMP-6 was compared by normalizing the ALP activity in KS483 cells after seven days of culture to the DNA content of the total number of cells in contact with the substrate and expressed as relative induction values compared to a glass control (Figure c). Interestingly, cells seeded on immobilized hBMP-6 produced significantly more ALP when compared to all other cases (Figure c). Increase in induced ALP activity on immobilized hBMP-6 was 4-fold in comparison with controls on glass. This contrasts significantly with the 2-fold increase in induced ALP activity when the culture medium was supplemented with comparable amounts of hBMP-6. Controls where hBMP-6 was partly denatured (by heating at 90 °C for 2 min) prior to cell seeding showed a significant loss of ALP. Controls where cells were seeded on epoxide-terminated glass, in medium supplemented with hBMP-6, showed a significantly lower ALP activity when compared to immobilized hBMP-6. Yet, the ALP induction was significantly higher when compared to epoxide-terminated glass controls without supplemented hBMP-6 and glass controls with supplemented hBMP-6. Also, epoxide-terminated surfaces induced ALP activity by 2-fold versus glass controls indicating an osteo-inductive effect of the epoxide groups. Dose-dependent ALP activities were performed for immobilized and supplemented hBMP-6 (Figure d). When supplemented hBMP-6 was doubled to 200 ng/mL, a significant rise in ALP activity was noted, but the ALP activity plateaued when hBMP-6 was doubled again. A concentration of 200–400 ng/mL hBMP-6 is in excess.[37] Interestingly, the ALP activity of cells seeded on immobilized hBMP-6 compared favorably to the case when 200 ng/mL or higher of supplemented hBMP-6 was used. This indicates that a sufficient amount of active hBMP-6 is available on the surface (ca. 120 ng/cm2) to achieve the maximum ALP activity. The immobilized hBMP-6 dose was easily reduced by a fourth (ca. 28 ng/cm2) using a featured μCP stamp during the immobilization step. This reduced dose resulted in a significantly lower ALP activity when compared to an entirely covered surface, but still significantly higher when compared to 100 ng/mL of hBMP-6 that was supplied to the medium. Taken together, these results indicate that immobilized hBMP-6 is bioactive and effective to induce ALP activity in cells.

Figure 2

Study of early osteogenesis in KS483 cells by immobilized hBMP-6. (a) Fluorescence microscopy image of hBMP-6 immobilized (by immersion 2 h in this case) on half of the substrate after immunofluorescence staining (inset shows averaged intensity profiles). (b) Immunohistochemical staining of ALP, after 4 days by KS483 cells (inset shows intensity profile with high (white) intensity indicative of low ALP activity). This slide was also prepared by immobilization of hBMP-6 by immersion (2 h). (c) ALP activity normalized by the total DNA content of KS483 cells after 6 days of culture, expressed as relative induction to a glass control in the presence of soluble (s (100 ng/mL)), bound (b), denatured bound (db) hBMP-6. Immobilization of hBMP-6 using flat stamps. (d) Dose response of KS483 cells cultured for 7 days in the presence of increasing concentrations of hBMP-6 in solution and two different amounts of bound hBMP-6, i.e., a full layer (100%, ca. 120 ng/cm2) of hBMP-6 and a μCP patterned layer of hBMP-6 (23% of the coverage in comparison to the full layer (using a flat stamp, ca. 28 ng/cm2). *p < 0.05 and **p < 0.01.

Study of early osteogenen class="Chemical">sis in KS483 cells by immobilized class="Chemical">pan class="Gene">hBMP-6. (a) Fluorescence microscopy image of hBMP-6 immobilized (by immersion 2 h in this case) on half of the substrate after immunofluorescence staining (inset shows averaged intensity profiles). (b) Immunohistochemical staining of ALP, after 4 days by KS483 cells (inset shows intensity profile with high (white) intensity indicative of low ALP activity). This slide was also prepared by immobilization of hBMP-6 by immersion (2 h). (c) ALP activity normalized by the total DNA content of KS483 cells after 6 days of culture, expressed as relative induction to a glass control in the presence of soluble (s (100 ng/mL)), bound (b), denatured bound (db) hBMP-6. Immobilization of hBMP-6 using flat stamps. (d) Dose response of KS483 cells cultured for 7 days in the presence of increasing concentrations of hBMP-6 in solution and two different amounts of bound hBMP-6, i.e., a full layer (100%, ca. 120 ng/cm2) of hBMP-6 and a μCP patterned layer of hBMP-6 (23% of the coverage in comparison to the full layer (using a flat stamp, ca. 28 ng/cm2). *p < 0.05 and **p < 0.01.

Shielded Release of hBMP-6

We next attached n class="Gene">hBMP-6 to the surface by essentially class="Chemical">pan class="Chemical">similar chemistry, but now using a longer brush-type linker.[38] To this end, layers of poly(glycidyl methacrylate) (PGMA) brushes on glass were prepared (Figure a).[38−42] Briefly, the atom transfer radical polymerization (ATRP) initiator, 2-bromo-2-methylpropionyl bromide, was reacted with the terminal amine moieties of a (3-aminopropyl)triethoxysilane (APTES) monolayer (Figure a). Subsequently, the immobilized initiator was immersed in an oxygen-free solution of glycidyl methacrylate as the ATRP monomer for 2 h to yield a PGMA brush layer of 80 nm thickness (Table ). Immobilization of hBMP-6 onto the PGMA layers was confirmed using contact angle, ellipsometry and XPS measurements (Table ). Immunofluorescence stains of hBMP-6 on PGMA brushes were of similar intensity as compared to the hBMP-6 patterns on GPTMS surfaces (Figures d and S3) suggesting similar loading of hBMP-6 on either the GPTMS or PGMA coated surfaces. The release of hBMP-6 in culture medium was quantified using an enzyme-linked immunosorbent assay (ELISA) over a period of 15 days without cells seeded (Figure top). In the case of the short (GPTMS) linker, a released quantity of 100 ng/mL hBMP-6 was promptly reached within 24 h, which moderately increased and plateaued in the following 7 days. The concentration of 100 ng/mL of hBMP-6 is a therapeutic level reported in literature.[37,43] In the case of the long, PGMA linker, the release of hBMP-6 was much slower (by a factor 6–10) in the first 4 days and the therapeutic hBMP-6 level of 100 ng/mL was only reached at day 5, while the total amount of released hBMP-6 after 7 and 15 days was half of that on GPTMS. When the ELISA assay was performed when cells were seeded on the hBMP-6 coated surfaces, remarkable differences were observed. For cells seeded on layers of hBMP-6 on short (GPTMS) linkers, a much reduced, near negligible, amount of released hBMP-6 protein was observed during the first 5 to 6 days. When cells were seeded on layers of hBMP-6 on long (PGMA) linkers, the detected release profile was found close to the case without cells in the first 5 to 6 days. We also observed a reduced cell adhesion and spreading on PGMA layers in comparison with the GPTMS layers (Figure S2). A confluent cell layer was only reached after 7 days on the PGMA layers, while on GPTMS layers, confluency was reached at day 4. After 7 days, lower quantities of hBMP-6 were also detected in the medium on PGMA layers in the presence of cells. The observations indicate that when cells adhere earlier, as on GPTMS, less hBMP-6 escaped to the medium, and adhered cells with more bioactive hBMP-6 available, facilitates more interactions with cellular BMP receptors at the cell membrane. However, when cells adhere later, as on PGMA, slower released hBMP could still escape into the medium. These results confirm that release of hBMP-6 occurred and that the release kinetics could be decreased on brushes of PGMA, but for released hBMP-6 to be confined to cells, as required for registration by cells, cells need to be adhered to the surface.

Figure 3

Quantification of hBMP-6 in medium by ELISA (top) over a period of 15 days without (solid lines) and with (dashed lines) 5000 cells/cm2 on GPTMS and PGMA with initially μCP hBMP-6 (patterned stamps). Relative induction of ALP by KS483 cells after 4 days relative to a glass control (bottom). Black bars represent relative induction of cells seeded in contact with the films. White bars represent relative induction of cells seeded in transwell. **p < 0.01.

Quantification of n class="Gene">hBMP-6 in medium by ELISA (top) over a period of 15 days without (solid lines) and with (dashed lines) 5000 cells/cm2 on class="Chemical">pan class="Chemical">GPTMS and PGMA with initially μCP hBMP-6 (patterned stamps). Relative induction of ALP by KS483 cells after 4 days relative to a glass control (bottom). Black bars represent relative induction of cells seeded in contact with the films. White bars represent relative induction of cells seeded in transwell. **p < 0.01.

In a next experiment, KS483 cells were seeded for 4 days either in direct contact to or separated from the surface un class="Chemical">sing a transwell assay (Figure bottom). Low class="Chemical">pan class="Gene">ALP activity was observed on control surfaces (glass, GPTMS and PGMA), irrespective of the presence of the transwell insert. Remarkably, when cells were seeded in direct contact to or in transwell above hBMP-6 immobilized on PGMA layers, in both of these cases, a lower ALP activity was observed in comparison when cells were seeded in direct contact to immobilized hBMP-6 on GPTMS layers, while an intermediate ALP activity level was observed using GPTMS layers in the transwell assay. These observations confirm that cells seeded on GPTMS layers register more bioactive hBMP-6 after four days. This is presumably aided by adhered cells that are able to confine and register released hBMP-6 to produce ALP. This state of adhered cells and hence ability to confine released hBMP-6 is different on PGMA layers, where both release of hBMP-6 and cell adhesion is slower, but not to same extent slower.

The biological activity of released n class="Gene">hBMP-6 was further confirmed by measuring the luciferase activity of C2C12 cells that were stably transfected with a BMP-responclass="Chemical">pan class="Chemical">sive element (BRE) fused to the firefly luciferase reporter gene (BRE-Luc cells) (Figure a).[44] C2C12 is a model to study differentiation into bone cells.[44] To avoid effects related to material stiffness, differences in luciferase expression relative to background of the same surface in absence of hBMP-6 were used. Figure a shows that luciferase activity was delayed to day two, when cells were seeded on surfaces with immobilized hBMP-6 using the short (GPTMS) linker in comparison to soluble hBMP-6, whereas on the long (PGMA) linker, a significant rise in luciferase activity was observed only at day seven. In addition, mRNA expression levels of osteoblast-specific runt-related transcription factor 2 (RUNX2) and osteocalcin (OCN) genes were analyzed using quantitative PCR (Figure b, c).[45] Upregulation of RUNX2 was clearly observed already one day after cell seeding using either soluble hBMP-6 or hBMP-6 immobilized using the short linker. OCN expression was first observed after day three, which is in line with the delayed response of this gene during osteoblast differentiation.[45,46] The increase in expression of the same genes when cells were seeded on hBMP-6 coated surfaces using the longer PGMA linker was clearly delayed. Taken together, these results confirm that the activation of ALP production and the induction of osteoblastic specific gene expression were activated by hBMP-6. Attachment of hBMP-6 to a surface lessens its biological activity as compared to the soluble growth factor controls. The activity is unleashed by hydrolysis of the linker, which releases the hBMP-6 from the surface. These results are in agreement with the ELISA studies showing that in the case of the long linker, the release of hBMP-6 is delayed with 4 to 5 days before cells adhere and have a minimal dosage of released hBMP-6 present that can activate ALP production and gene transcription. Therefore, we believe that the biological activity of hBMP-6 growth factors anchored via siloxane-based linkers is related to their release from the surface, but cells need to be adhered to confine the released growth factors to the cells.

Figure 4

(a) Luminescence of C2C12 cells transfected with BRE-Luc, measured at different time points relative to a background using the same surface without hBMP-6. Flat stamps were used for immobilization of hBMP-6. Inset shows the luminescence results at day 7 from an independent experiment. Gene expression analysis at days 1, 3, and 7 of (b) RUNX2 and (c) OCN was determined via the 2ΔΔCt method using the expression on day 1 on glass as a reference. *p < 0.05 and **p < 0.01.

(a) Luminescence of C2C12 cells transfected with BRE-Luc, measured at different time points relative to a background un class="Chemical">sing the same surface without class="Chemical">pan class="Gene">hBMP-6. Flat stamps were used for immobilization of hBMP-6. Inset shows the luminescence results at day 7 from an independent experiment. Gene expression analysis at days 1, 3, and 7 of (b) RUNX2 and (c) OCN was determined via the 2ΔΔCt method using the expression on day 1 on glass as a reference. *p < 0.05 and **p < 0.01.

Growth Factor Activity of hBMP-6 on Biomaterials

To demonstrate that our release strategy is applicable to n class="Chemical">polymeric biomaterials, we next attempted to functionalize a representative biomaterial with growth factors adopting our findings in line with our model study on glass. Therefore, a class="Chemical">pan class="Chemical">siloxane-bound GPTMS linker was attached on plasma-oxidized (Ox) 100 μm thick films of commercially available biopolymers (Figure a and 5b) made of a poly(ether ester) copolymer (PA) comprising poly(ethylene oxide) (PEO) and poly(butylene terephthalate (PBT) onto which hBMP-6 was immobilized (Figure c, II). In a second set of experiments, hBMP-6 was immobilized directly onto this polymer surface after introduction of aldehyde groups by oxygen plasma treatment and the subsequent incubation with hBMP-6 to introduce imine bond formation (Figure c, I). A control surface with irreversibly bound GF was obtained by mildly reducing the imine bond (see the Supporting Information) to generate a stable, irreversible, secondary amine linkage under physiological conditions (Figure c, III). GPTMS-functionalization of PA was verified with contact angle measurements, XPS and FT-IR (Table S1, Figure S4). Upon oxidizing the PA surface the advancing contact angle decreased with respect to the native PA films, while upon coupling of GPTMS the contact angle increased. Upon reaction with GPTMS, the presence of elemental Si was detected in the XPS spectra, while deconvoluted XPS spectra of the C1s region (Figure S4) showed an increase of the peak at 286.35 eV, which can be attributed to ether bonds (C–O).[47] In addition, the peak centered at 288.52 eV, which can be mainly attributed to ester bonds (O–C=O) that are present in native PA films, was broadened during plasma treatment indicating other carbonyl-containing species such as aldehyde/ketone (C=O) groups, which typically have bands in the range of 287.1–288.1 eV. In the FT-IR spectrum an O-H stretching band appeared after plasma treatment of the native PA films (inset Figure S5a). This band was significantly reduced upon reaction with GPTMS indicating the conversion of O-H bonds into O-Si bonds. Also, the ratio of the intensities of the asymmetric C–O–C and the C=O stretching bands (Figure S5b) was inverted after the plasma treatment with a higher intensity for the carbonyl peak. Upon reaction with GPTMS, the ratio was inverted back with a slightly higher intensity for the C–O–C stretching band corresponding to the epoxide ring of the linker. Taken together, these results are indicative of successful activation of the PA films with groups that allow the immobilization of hBMP-6 as envisioned.

Figure 5

(a) Molecular structure of the block-copolymer (PA) consisting of a hydrophilic block of poly(ethylene glycol terephthalate) (PEGT) and a hydrophobic block of poly(butylene terephthalate) (PBT). (b) 100 μm thick PA films were prepared by casting a solution of 3.4 wt% PA in chloroform overnight and cured above 60 °C for several hours. (c) Schematic procedure of the surface functionalization of PA films after preactivation using O2 plasma, with GPTMS. hBMP-6 is subsequently immobilized via reactive μCP on the activated film via the reaction between the primary amine groups (accessible lysine residues in dark gray) and aldehyde groups forming an imine bond (I), which is further reduced in order to increase its stability toward hydrolysis (II) or on the GPTMS (III) with unreacted aldehyde groups also available to tether the growth factor as in I.

(a) Molecular structure of the block-n class="Chemical">copolymer (PA) conclass="Chemical">pan class="Chemical">sisting of a hydrophilic block of poly(ethylene glycol terephthalate) (PEGT) and a hydrophobic block of poly(butylene terephthalate) (PBT). (b) 100 μm thick PA films were prepared by casting a solution of 3.4 wt% PA in chloroform overnight and cured above 60 °C for several hours. (c) Schematic procedure of the surface functionalization of PA films after preactivation using O2 plasma, with GPTMS. hBMP-6 is subsequently immobilized via reactive μCP on the activated film via the reaction between the primary amine groups (accessible lysine residues in dark gray) and aldehyde groups forming an imine bond (I), which is further reduced in order to increase its stability toward hydrolysis (II) or on the GPTMS (III) with unreacted aldehyde groups also available to tether the growth factor as in I.

n class="Gene">hBMP-6 microarrays on the activated PA films were preclass="Chemical">pared via reactive μCP as described in the model study on glass and visualized by immunostaining (Figure S6). The results showed that class="Chemical">pan class="Gene">hBMP-6 was successfully anchored via the different immobilization strategies. For the GPTMS linker, a significant decay of fluorescence intensity was observed within two days, which is slower in comparison to the case of applying this linker on glass substrates, but faster than for immobilizing hBMP-6 via imines, while, after reduction using NaBH3CN, the fluorescence intensity remained nearly constant over the course of seven days. This indicated that hBMP-6 remained linked to the surface in this time period. Significant enhancement of cell adhesion was observed after the formation of a GPTMS linker on the PA films, irrespective of the presence of hBMP-6 (Figure S7d, h). Plasma treated PA films (PAOx) exhibited improved cell adhesion and spreading in comparison to native PA films (Figure S7a, b), which is in agreement with literature,[48] while immobilization of hBMP-6 did not change cell adhesion as expected (Figure S7e, f). Similar cell adhesion behavior was found on these films after reduction of the imine bond (Figure S7g). The biological activity of the released hBMP-6 from PA films using the different linkers was verified using C2C12 BRE-Luc cells (Figure ). The luciferase activity was increased when substantial quantities of hBMP-6 were released from the surface. For example, in the case of the GPTMS linker, a significant release of hBMP-6 after two days coincided with an increase in luciferase activity. Significant luciferase activity on imine films was only observed after five days, which indicates a slower release of hBMP-6. No luciferase activity was observed on surfaces where the imine bond was reduced into an irreversible secondary amine bond, which indicates effectively blocked release of hBMP-6 from the surface. These results confirm that the covalent attachment of hBMP-6 to a biomaterial surface was not biological active. This activity can subsequently be recovered after hydrolysis of the linker releasing the growth factor from the surface. By varying linker length (GPTMS versus PGMA using epoxide linkage) or linking strategy (imine versus secondary amine bonding), each with different susceptibility to hydrolysis and thus release from the growth factor from the surface, customizable bioactive surfaces can be engineered with spatio-temporal control over growth factor activation.

Figure 6

Luminescence of C2C12 BRE-Luc at different time points relative to the background. *p < 0.05 and **p < 0.01. Surfaces: hBMP-6 supplemented to medium over PAOx (PAOx-shBMP-6), hBMP-6 bound to GPTMS on PAOx (PAOx-GPTMS-bhBMP-6), hBMP-6 bound as imine on PAOx (PAOx-bhBMP-6), and hBMP-6 bound via an amine resulting of the reduction of the imine (PAOx-rbhBMP-6). Flat stamps were used for immobilization of hBMP-6.

Luminescence of C2C12 BRE-Luc at different time points relative to the background. *p < 0.05 and **p < 0.01. Surfaces: n class="Gene">hBMP-6 supplemented to medium over class="Chemical">pan class="Chemical">PAOx (PAOx-shBMP-6), hBMP-6 bound to GPTMS on PAOx (PAOx-GPTMS-bhBMP-6), hBMP-6 bound as imine on PAOx (PAOx-bhBMP-6), and hBMP-6 bound via an amine resulting of the reduction of the imine (PAOx-rbhBMP-6). Flat stamps were used for immobilization of hBMP-6.

Conclusions

We have demonstrated the use of hydrolyzable linkers, such as n class="Chemical">siloxane, class="Chemical">pan class="Chemical">ester, and imine, for immobilization and regulated release of hBMP-6 to model surfaces as well as biomaterials. Immobilized hBMP-6 is biologically inactive. Upon release from the surface through hydrolysis of the covalent bonds its biological activity is recovered. We established a relation between the hydrolytic susceptibility of the bonds and osteoblastic cell differentiation. The use of labile strategies for the covalent attachment of growth factors to a surface with susceptibility for hydrolysis provides the means to engineer biomaterial surfaces with control over the dose of the growth factor to determine the release of bioactive growth factors over prolonged periods of time. We anticipate that this concept can be used to improve the design of a new generation of growth factor delivery systems in controlled drug delivery and for tissue engineering purposes.