Preservation of FGF-2 bioactivity using heparin-based nanoparticles, and their delivery from electrospun chitosan fibers

Acta Biomaterialia 8 (2012) 1551–1559

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat

Preservation of FGF-2 bioactivity using heparin-based nanoparticles, and their delivery from electrospun chitosan fibers Fabio Zomer Volpato a,b, Jorge Almodóvar c, Kristin Erickson c, Ketul C. Popat d,e, Claudio Migliaresi a,b, Matt J. Kipper c,e,⇑ a

Department of Materials Engineering and Industrial Technologies, University of Trento, Via delle Regole 101, Trento 38123, Italy BIOtech Research Center, University of Trento, Via delle Regole 101, Trento 38123, Italy Department of Chemical and Biological Engineering, Colorado State University, Fort Collins, CO 80523-1370, USA d Department of Mechanical Engineering, Colorado State University, Fort Collins, CO 80523-1374, USA e School of Biomedical Engineering, Colorado State University, Fort Collins, CO 80523-1376, USA b c

a r t i c l e

i n f o

Article history: Received 22 September 2011 Received in revised form 28 November 2011 Accepted 15 December 2011 Available online 20 December 2011 Keywords: Growth factors Glycosaminoglycans Mesenchymal stem cells Layer-by-layer

a b s t r a c t Here we present a novel matrix-mimetic nanoassembly based on polysaccharides. Chitosan electrospun fiber networks are decorated with heparin-containing polyelectrolyte complex nanoparticles (PCNs) that present basic fibroblast growth factor (FGF-2), both stably adsorbed to the surfaces and released into solution. These FGF-2/PCN complexes can be released from the fibers with zero-order kinetics over a period of 30 days. Further modification of fibers with a single bilayer of polyelectrolyte multilayer (PEM) composed of N,N,N-trimethyl chitosan and heparin completely prevent release, and the FGF-2/PCN complexes are retained on the fibers for the duration of the release experiment (30 days). We also compare the mitogenic activity of these FGF-2/PCN complexes delivered in two different states: adsorbed to a surface and dissolved in solution. FGF-2/PCN complexes exhibit mitogenic activity with respect to ovine bone marrow-derived mesenchymal stem cells, even after being preconditioned by incubating for 27 days at 37 °C in solution. However, when the FGF-2/PCN complexes are adsorbed to chitosan and coated with PEMs, the mitogenic activity of the FGF-2 steadily decreases with increasing preconditioning time. This work demonstrates a new system for stabilizing and controlling the delivery of heparinbinding growth factors, using polysaccharide-based matrix-mimetic nanomaterials. This work also contributes to our understanding of the preferred mode of growth factor delivery from porous scaffolds. Ó 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction The use of growth factors to guide the differentiation of stem cells is a particularly promising strategy for engineering slow-healing tissues such as bone and cartilage, to treat a variety of injury and disease states [1]. Growth factors from the fibroblast growth factor (FGF) family and transforming growth factor-b (TGF-b) superfamily, which includes the bone morphogenetic proteins (BMPs), affect wound healing, tissue synthesis and mesenchymal stem cell (MSC) differentiation. For example, FGF-2 is involved in osteogenesis [2,3], chondrogenesis [4] and angiogenesis [5]. However, many therapeutic strategies based on growth factor delivery are impeded by the relative instability of growth factors on time scales associated with these biological processes. Members of the ⇑ Corresponding author at: Department of Chemical and Biological Engineering, Colorado State University, Fort Collins, CO 80523-1370, USA. Tel.: +1 970 491 0870; fax: +1 970 491 7369. E-mail address: [email protected] (M.J. Kipper).

FGF family and TGF-b superfamily have plasma half-lives on the order of minutes (1.5 min for FGF-2, and between 11 and 160 min for TGF-b1, for example) [6]. FGF-2 and BMP-2 have been demonstrated to lose their activity within 24 h and become completely degraded within 3 days when adsorbed to and released from mineral-based and ceramic scaffolds for bone tissue engineering [7,8]. Materials for skeletal tissue engineering that use growth factors should be developed that can present the growth factor in a structural and biochemical context similar to native tissue. This could mean presenting the growth factor bound to a surface or slowly releasing the growth factor into nearby tissue [9]. In mammalian tissues, polysaccharides are found in nanostructured proteoglycans, like aggrecan, which impart both biomechanical and biochemical function to the extracellular matrix (ECM) [10]. One of the most important of these biochemical functions is to serve as a reservoir for the binding and stabilization of growth factors. Heparin is a glycosaminoglycan that protects FGF-2 from proteolytic and chemical inactivation [11]. This stabilization likely results from the binding of FGF-2 to specific sulfation patterns in

1742-7061/$ - see front matter Ó 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2011.12.023

1552

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

heparin that also potentiate the interaction between the growth factor and the growth factor receptor [12]. The heparin-binding sequence of FGF-2 is homologous to a similar sequence in TGF-b1 [2]. Sulfated or sulfonated materials that might mimic this biochemical function of the ECM have been proposed for growth factor delivery. These include sulfonated silk fibroin [13], heparin-conjugated fibrin gels [14] and glycosaminoglycan-containing polyelectrolyte multilayers (PEMs) [15,16]. These materials have been designed either to present growth factors on a surface or to deliver growth factors by releasing them into solution. It is unclear whether one or the other of these particular strategies is the preferred mode of delivery. The present work has two goals. The first goal is to demonstrate an ECM-mimetic polysaccharide-based nanoassembly that can present both surface-bound and soluble FGF-2 that is stabilized by complexation with heparin-containing nanoparticles. The second goal is to compare the ability of these nanoparticles to preserve the mitogenic activity of FGF-2 in the surface-adsorbed and solution states. In addition to the binding and stabilization of growth factors, polysaccharides possess many other properties that make them attractive biomaterials. These properties include biodegradability, antimicrobial activity and the ability to support mammalian cell growth. Furthermore, polysaccharides can be readily processed by a number of techniques to produce materials that mimic both biophysical and biochemical features of the ECM. Previously we have studied glycosaminoglycan-containing PEMs, polyelectrolyte complex nanoparticles (PCNs) and electrospun fibers [16–21]. We have demonstrated that these polysaccharide-based nanostructures can bind FGF-2 and that FGF-2 mitogenic activity can be enhanced by incorporating it into heparin-containing PEMs [16]. Here we demonstrate the use of nanostructured chitosan-based electrospun fibers with high porosity and a large surface area for the binding, stabilization and controlled release of heparin-binding growth factors. Chitosan is an amine-containing polysaccharide. Its pendant amine is protonated in acidic medium, making chitosan a weak polycation. Chitosan has excellent film- and fiber-forming properties, but is difficult to electrospin due to its solution properties. Several reports have demonstrated successful electrospinning of chitosan by blending with other water-soluble polymers [22– 24]. Sangsanoh and Supaphol [25] reported the electrospinning of pure chitosan using a trifluoroacetic acid/dichloromethane (TFA/DCM) solvent. This solvent system likely facilitates the electrospinning because of the formation of ammonium–TFA salt with the amine groups in chitosan [26]. In this work, we develop chitosan fiber networks that can present growth factor-containing nanoparticles both on the surface and by releasing them into solution. Chitosan fiber networks are modified with chitosan–heparin PCNs containing of FGF-2 as a model heparin-binding growth factor. Zero-order release of FGF-2 from the fibers is demonstrated for up to 30 days. Release of FGF-2 is controlled by additional complexation of the adsorbed nanoparticles with polysaccharides. When fibers are further modified with a single bilayer of polysaccharide-based PEM, composed of N,N,N-trimethyl chitosan (TMC) and heparin, release of the FGF-2/PCN complexes is prevented for the duration of the analysis (30 days). The mitogenic activity of the FGF-2/PCN complexes is also evaluated, with respect to the proliferation of ovine bone-marrow derived MSCs. Complexation is found to preserve FGF-2 activity for 30 days.

2. Materials and methods 2.1. Materials Highly purified chitosan (80 kDa, 5% acetylated as determined by 1H NMR, Protasan UP B 90/20) was purchased from NovaMatrix

(Sandvika, Norway). Heparin sodium (from porcine intestinal mucosa, 12.5% sulfur) was purchased from Celsus Laboratories (Cincinnati, OH). TMC was synthesized following a procedure described by de Britto and Assis [27]. Recombinant human FGF-2 146 aa was purchased from R&D Systems (Minneapolis, MN). 5(6)-Carboxyfluorescein N-hydroxysuccinimide ester was purchased from Sigma–Aldrich (St. Louis, MO). Sodium bicarbonate and dimethyl sulfoxide (DMSO) were purchased from Fisher Scientific (Pittsburgh, PA). TFA, DCM, ammonium hydroxide, dimethyl sulfate, sodium hydroxide and sodium chloride were purchased from Acros Organics (Geel, Belgium). The following were purchased from HyClone (Logan, UT): fetal bovine serum (FBS), 0.25% trypsin with EDTA, low glucose Dulbecco’s modified Eagle’s medium, minimum essential medium alpha (a-MEM) (supplemented with L-glutamine, ribonucleosides and deoxyribonucleosides), and Dulbecco’s phosphate-buffered saline (PBS) without Ca2+ and Mg2+. The following were purchased from Gibco (Grand Island, NY): antibiotic-antimycotic (anti/anti), 1 M HEPES buffer solution and PBS with Ca2+ and Mg2+. Calcein-AM in DMSO (4 mM) was purchased from Invitrogen (Eugene, OR). 40 6-Diamidino-2-phenylindole-2HCl (DAPI) was purchased from Thermo Fisher Scientific (Rockford, IL). Human fibronectin was purchased from BD Biosciences (Bedford, MA). All of the polymers, growth factor and solvents were used without further purification. All aqueous solutions were prepared using ultrapure, 18.2 MX cm, water (DI water). Rhodamine-modified chitosan was prepared by dissolving 100 mg of chitosan in 10 ml of 0.1 M acetic acid, adding 10 ml of methanol to the solution, adding 3.25 ml of 2 mg ml1 rhodamine B isothiocyanate (Sigma, St. Louis, MO) in methanol, and allowing the solution to react overnight. Rhodamine-modified chitosan was purified via dialysis, freeze dried and stored at 4 °C protected from light until use. Fluorescein-labeled FGF-2 (FGF-2LB) was prepared by dissolving of FGF-2 in 0.1 M NaHCO3 (pH 8) at a concentration of 25 lg ml1. The solution was vigorously stirred in an ice bath to avoid protein denaturation. The dye solution was prepared by the dissolution of 5(6)-carboxyfluorescein N-hydroxysuccinimidyl ester in DMSO at a concentration of 10 mg ml1 [28]. Next, 10 ll of the dye solution was slowly added to 1 ml of protein solution and allowed to react for 4 h with vigorous stirring in an ice bath. The unreacted dye was removed by the dialysis of the labeled protein solution in PBS (pH 7.4) for 24 h, using 3 kDa MWCO dialysis cassettes (Slide-A-Lyzer, Thermo Scientific, PA). A fluorescence microplate reader (FLUOstar Omega, BMG Labtech, Durham, NC) was used to confirm labeling of the FGF-2LB and to determine the calibration curve used to quantify labeled FGF-2 in solution (kex = 485 nm and kem = 520 nm). 2.2. Electrospinning chitosan The electrospinning apparatus consisted of a high-voltage (130 kV) direct current power supply (GAMMA High Voltage Instruments, ES30P-10 W/DAM), a syringe pump (Harvard Apparatus, UK) and a rectangular brass collector covered with aluminum foil. The metal needle of the syringe, connected to the power supply, and the collection target were kept at a horizontal distance of 18 cm. Chitosan (or rhodamine-labeled chitosan) was dissolved in TFA:DCM (7:3) overnight at 7 wt.%, at room temperature. The condition was selected after preliminary analysis of conductivity and viscosity as well as evaluations on the stability and homogeneity of the solutions. Electrospinning of the 7 wt.% chitosan was achieved with a flow rate of 0.5 ml h1 and an applied voltage of 18 kV. Although chitosan is insoluble in neutral aqueous solutions, the presence of residual solvent or ammonium–TFA salts render the electrospun fibers soluble in water. To eliminate residual solvent, a procedure for fiber network stabilization has been

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

1553

developed by Almodóvar and Kipper [20]. Briefly, the fiber networks were gently sandwiched between two filter papers (Whatman 41, 20 lm, ashless, Whatman, NJ) and placed in a Büchner funnel. Then a 5 M solution of ammonium hydroxide was slowly pulled through the filter papers under weak vacuum, maintaining the fibers’ integrity. Finally, it was flushed with 4 l of DI water to remove the ammonium hydroxide.

0.22 lm PVDF syringe filters (Fisher Scientific, PA). DI water was used for rinsing steps. For this, 250 ml of each solution was pulled through the samples once, in the following sequence: (i) heparin, (ii) water rinse, (iii) TMC and (iv) water rinse. Almodóvar and Kipper have previously demonstrated the efficacy of this procedure for modifying electrospun chitosan fibers with polysaccharide-based PEMs [20].

2.3. PCN production

2.5. Microscopic evaluation of electrospun networks

The production of heparin–chitosan PCNs and their characterization (including yield of particles, zeta potential and hydrodynamic radius for different charge mixing ratios) has been previously described by Boddohi et al. [18,19]. Briefly, polysaccharide solutions (0.9 mg ml1 for chitosan and 0.95 mg ml1 for heparin) were prepared in 0.1 M acetate buffer, pH 5.0. At pH 5.0 chitosan has sufficient protonated amines to form complexes with sulfate and carboxylate groups of heparin. The solutions were filtered using 0.22 lm polyvinylidene fluoride syringe filters (PVDF, Fisher Scientific, PA). PCNs were prepared at room temperature by a one-shot addition using heparin as the starting solution for the production of anionic particles. The ratio of 4:1 (heparin:chitosan, on the saccharide basis) was chosen, taking into account previous studies [18,19]. Vigorous stirring for 3 h was maintained during the complex formation. After 3 h of stirring, the solution was allowed to settle overnight to remove aggregated particles. After settling, the solution containing dissolved PCNs was decanted and centrifuged (4500g, 15 min) to separate the particles from uncomplexed polymer using an Eppendorf 5804 centrifuge (Eppendorf, Westbury, NY). The supernatant was decanted off, and the particles were resuspended in DI water. PCN diameter was measured using dynamic light scattering (DLS). DLS was performed using a DynaPro Titan (Wyatt Technologies, Santa Barbara, CA) instrument using an 830 nm laser. All measurements consisted of 10 acquisitions, performed at a fixed angle of 90° at 25 °C. FGF-2 (or FGF-2LB) was complexed with PCNs, exploiting the ability that FGF-2 has to bind sulfated glycosaminoglycans, such as heparin. A saturated solution of FGF-2 with a concentration of 600 ng mg1 of PCNs was prepared in order to maximize binding of FGF-2 to the available sites in heparin. The solution was gently mixed for 30 min at room temperature to allow the reaction. The unbound FGF-2 was removed by dialysis against DI water for 48 h at room temperature using 20 kDa MWCO dialysis cassettes (Slide-A-Lyzer, Thermo Scientific, PA). This dialysis step is used to help ensure that the FGF-2 remaining on the PCNs is stably bound, and improves the likelihood that the FGF-2 will remain bound during subsequent aqueous processing steps. The concentrated FGF-2/ PCN complexes were diluted in DI water at the necessary concentrations depending on the experiment. FGF-2 adsorption on PCNs and entrapment efficiency were confirmed using the fluorescence microplate reader mentioned above.

Scanning electron microscopy (SEM) (Supra 40, Zeiss and JSM6500F, Jeol) was used to evaluate the morphology of the fiber networks. Samples were mounted on aluminum stubs and coated with 10 nm of gold by evaporation. Chitosan fiber networks were imaged using SEM before and after NH4OH treatment, after modification with FGF-2/PCN complexes and after modification with PEMs. SEM was also conducted on samples after FGF-2 release experiments (described below). Qualitative analysis of the modification of chitosan fiber networks was performed using confocal laser scanning microscopy (Zeiss LSM510, and A1, Nikon). Modification with FGF-2LB/PCN complexes was performed on a fiber network made from rhodamine-modified chitosan. An image (63) for the rhodaminelabeled fibers was obtained (kex 540 nm and kem 625 nm). Afterwards, an image for the FGF-2LB/PCN was obtained at the same location (kex 492 nm and kem 517 nm) and the two images were merged. For all other experiments, unlabeled chitosan was used, and only the fluorescence of the FGF-2LB was observed (20). The samples were imaged before and after the release experiments (described below) to assess the presence of FGF-2LB.

2.4. Modification of fiber networks with PCNs and PEMs Chitosan fibers were modified with FGF-2LB/PCN complexes and some samples were also subsequently modified with heparin–TMC PEMs (one bilayer). Modification of fibers was achieved using a procedure similar to the one used to stabilize the fiber networks. First, a 250 ml aqueous suspension of FGF-2LB/PCN complexes, at a concentration of 4 lg ml1, was slowly pulled through a 3 cm2 sample of fibers sandwiched between filter papers under a weak vacuum. The FGF-2LB/PCN complexes were electrostatically adsorbed on the chitosan fibers. Next, some of these samples were further modified with a single bilayer of heparin–TMC PEM. Briefly, TMC and heparin solutions with concentration of 0.01 M (on a saccharide basis) were prepared in DI water and filtered using

2.6. Release of FGF-2LB/PCN complexes from fiber networks with and without PEMs Release of FGF-2LB/PCN complexes from fiber networks was determined by fluorescence spectroscopy of the supernatant. Modified fiber samples (3 cm2) were placed in a Petri dish with 10 ml of DI water and incubated at 37 °C under continuous agitation (at 10 rpm). At regular intervals, 300 ll of solution was collected and stored frozen, and a similar amount of fresh DI water was added to the release medium. At the end of the release study all samples were thawed and the concentration of FGF-2LB in the supernatant was measured using the fluorescence microplate reader. The fluorescence data were compared to a calibration curve to report the amount of FGF-2LB in solution. The cumulative mass of FGF-2LB was determined by correcting for the sampling and supernatant replacement. The experimental conditions studied are summarized in Table 1. Each condition was studied in triplicate. 2.7. Harvest and culture expansion of ovine MSCs MSCs were harvested and cultured following the methods previously reported by Almodóvar et al. [16]. Briefly, bone marrow aspirates were obtained from female sheep (4–7 years old). Nucleated cells were separated from the bone marrow aspirate via centrifugation, and these cells were seeded on culture flasks. The non-adherent cells were removed and MSC colonies were allowed to develop for at least 7 days. The MSCs were then cryopreserved (in FBS with 5% DMSO) in liquid nitrogen until further use. 2.8. FGF-2 stability assay The stability of the FGF-2 was assessed by evaluating its mitogenic activity using MSCs. FGF-2 is a potent mitogen for ovine MSCs, when delivered at the appropriate concentration in solution. We have previously shown that the mitogenic activity of FGF-2 can

1554

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559 Table 1 Sample nomenclature for FGF-2 release kinetics experiments. Sample name

Condition

Fibers only FGF-2/PCN FGF-2/PCN + PEM

Fiber samples without PCNs or PEMs FGF-2LB/PCN complexes adsorbed onto fibers FGF-2LB/PCN complexes adsorbed onto fibers, further modified with one PEM bi-layer

Table 2 Sample nomenclature for FGF-2 activity assay. Sample name

Condition

PCN + PEM

PCNs adsorbed on chitosan-coated TCPS, further modified with one bilayer of PEM, were preconditioned for 1, 9, 18 or 27 days (DI water, 37 °C). Surfaces were then coated with fibronectin and MSCs and culture medium were added after the preconditioning period FGF-2/PCN complexes adsorbed on chitosan-coated TCPS, further modified with one PEM bilayer, were incubated for 1, 9, 18 or 27 days (DI water, 37 °C). Surfaces were then coated with fibronectin and MSCs and culture medium were added after the preconditioning period PCNs were preconditioned for 1, 9, 18 or 27 days (DI water, 37 °C) and then added to MSC cultures on fibronectin-coated TCPS FGF-2/PCN complexes were preconditioned for 1, 9, 18 or 27 days (DI water, 37 °C) and then added to MSC cultures on fibronectin-coated TCPS

FGF-2/ PCN + PEM PCN FGF-2/PCN

be enhanced when it is delivered to ovine MSCs by incorporating it into heparin-containing PEMs [16]. To evaluate whether the FGF-2 activity is preserved under the release conditions studied above, FGF-2/PCN complexes were preconditioned for 1, 9, 18 and 27 days at 37 °C in DI water, both in solution and in heparin–TMC PEMs. Negative control conditions were also studied. For these negative control conditions, PCNs without FGF-2 were also preconditioned for 1, 9, 18 and 27 days, both in solution and in heparin–TMC PEMs on tissue-culture polystyrene (TCPS). These conditions are summarized in Table 2. For these experiments unlabeled FGF-2 was used, to avoid any effects of the dye. In a positive control experiment (not shown), uncomplexed FGF-2 was found to completely lose its mitogenic activity within 48 h when preconditioned in DI water at 37 °C. Referring to the entries in Table 2, to prepare the PCN (and FGF2/PCN) conditions, PCNs (and FGF-2/PCN complexes) were preconditioned at 37 °C in DI water for 1, 9, 18 or 27 days. To prepare PCN + PEM (and FGF-2/PCN + PEM) samples, chitosan was first adsorbed to the surfaces of 24-well TCPS plates. These chitosanmodified surfaces were then modified with PCNs (or FGF-2/PCN complexes) and coated with heparin–TMC PEMs (one bilayer) according to methods we have previously described [16]. Eight of each type of sample were prepared and were preconditioned for 1, 9, 18 or 27 days (duplicate samples were prepared for each of four time points) under mild agitation. MSC proliferation experiments were used to test the FGF-2 activity of these preconditioned samples. In all of these experiments, MSCs were cultured in the wells of 24-well TCPS plates, coated with fibronectin (from 10 lg ml1 solutions). We have previously observed that MSCs seeded on chitosan-based PEMs do not adhere well, thus fibronectin is needed to improve cell adhesion [16]. For the PCN and FGF-2/PCN conditions, the particles that had been preconditioned for 1, 9, 18 or 27 days were added to the cultures (MSCs on fibronectin-coated TCPS) at a concentration equivalent to about 0.2 ng ml1 of FGF-2. This concentration is less than the optimally mitogenic dose of FGF-2 (1–10 ng ml1) for ovine MCSs when it is delivered in solution, as determined from our previous work [16]. For the PCN + PEM and the FGF-2/ PCN + PEM samples, the fibronectin coating was applied on top of the previously preconditioned PEMs. For all experiments, MSCs were seeded at 7000 cells cm2. This is the optimum cell density determined in our previous work for measuring ovine MSC proliferation over 4 days [16]. In that work, lower cell densities resulted in non-reproducible cell proliferation. When MSCs are plated at 7000 cells cm2, they can approach a confluent monolayer in 96 h, given the optimally mitogenic dose of FGF-2 (1 ng ml1)

in low-serum medium. These cells were cultured for 4 days using 1 ml of low-serum medium per well (a-MEM with 2.5% FBS, 1% anti/anti, 2.5% HEPES 1 M) at 37 °C and 5% CO2. After the 4 days of culture the MSCs were stained with calcein, fixed with glutaraldehyde, and counter stained with DAPI [16]. Fluorescence microscopy images were obtained using an Olympus IX70 epifluorescence microscope (Center Valley, PA) equipped with a QImaging Micropublisher camera and a filter wheel, with the wide ultraviolet and narrow blue filters for DAPI and calcein-AM, respectively. Images for cell counting were obtained using a 4 objective. Both green and blue channel images were collected from five fields of view from each well of the 24-well plates, representing approximately 25% of the total surface area. These images were processed using the ImageJ 1.41o software (National Institutes of Health, MD) to count the nuclei (DAPI stained). The blue channel of each image (nuclei stained with DAPI) was thresholded and automatically counted using the particle analyzer algorithm in the ImageJ software, to obtain cell numbers per area. 2.9. Statistical analysis All values are expressed as mean ± standard deviation. Comparisons between groups were performed via analysis of variance models with Tukey’s multiple comparison tests. Differences with p < 0.05 were considered statistically significant. 3. Results 3.1. Electrospinning of chitosan and modification of fibers with PCNs and PEMs Scanning electron micrographs of the electrospun chitosan networks are shown in Fig. 1A before and Fig. 1B after NH4OH treatment. The stabilization treatment modifies the fiber morphology in terms of homogeneity and fiber cross-section. Nevertheless, the main features, such as high surface area and open porosity, are still present. Fig. 1C shows a high-magnification SEM image of a fiber network after NH4OH treatment to show the smooth surface of the fibers. After NH4OH treatment, fibers have diameters ranging from several hundred nanometers to 1 lm. The hydrodynamic radii of the PCNs and FGF-2LB/PCN complexes were measured by DLS. Hydrodynamic radii of 270 and 302 nm were obtained before and after FGF-2LB adsorption, respectively. The FGF-2LB loading and entrapment efficiency of the FGF2LB/PCN complexes was determined by fluorescence of FGF-2LB/

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

1555

Fig. 1. Chitosan electrospun fiber network morphology (A) as prepared and after NH4OH treatment at (B) low and (C) high magnification. After NH4OH treatment the fiber networks are more compressed and have reduced porosity (compare A and B). After NH4OH treatment, the fibers have smooth, featureless surfaces (C).

Fig. 2. High-magnification SEM images depicting fiber roughness before release experiments (A) FGF-2LB/PCN and (B) FGF-2LB/PCN + PEM, and after release experiments (C) FGF-2LB/PCN and (D) FGF-2LB/PCN + PEM. Scale bars correspond to 1 lm. The inset in (A) is digitally zoomed 2 to show detail.

Fig. 3. (A) Confocal microscopy image using 63 objective of rhodamine-modified chitosan fiber network (red) with FGF-2LB/PCN complexes adsorbed (green). (B, C) Confocal microscopy images (using 20 objective) of FGF-2LB/PCN fibers and FGF-2LB/PCN + PEM fibers after 30 days of release at 37 °C.

PCN complexes shortly after production and purification through dialysis. A standard curve was prepared using known concentrations of FGF-2LB. After 48 h of dialysis to remove weakly bound FGF-2LB, the remaining amount of FGF-2LB loaded was 37.5 ng of FGF-2LB per mg of PCN (entrapment efficiency was 6.3%). These FGF-2LB/PCN complexes were electrostatically adsorbed to the fibers. High-magnification scanning electron micrographs of the

fiber networks are presented in Fig. 2. After drying to prepare the samples for SEM, the adsorbed FGF-2LB/PCN complexes appear as rough surface features on the fibers in Fig. 2A (inset). We note in Fig. 2A that PCNs are able to penetrate into the fiber network, and are not restricted to fibers close to the surface. After PEM deposition, these features are significantly reduced in the SEM image shown in Fig. 2B. Fig. 2C (FGF-2LB/PCN) and 2D (FGF-2LB/

1556

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

PCN + PEM) show fiber networks after 30 days of release at 37 °C in DI water. After release, the presence of particles on the FGF-2LB/ PCN sample is reduced (compare Fig. 2A to Fig. 2C). The PCNs are not as apparent in the PEM-modified samples either before (Fig. 2B) or after (Fig. 2D) release experiments. A high-magnification (63 objective) confocal micrograph of FGF-2LB/PCN complexes adsorbed onto rhodamine-modified chitosan fibers is shown in Fig. 3A. The FGF-2LB/PCN complexes penetrate the fiber network and adsorb well on the fibers, further confirming what was observed in the SEM images. Z-stacks and images taken from both sides of the fiber networks confirm uniform distribution of the PCNs throughout the fiber network (not shown). Fig. 3B and C show lower magnification (20 objective) confocal micrographs of fiber networks (without rhodamine) after the 30 days of release. Only very weak fluorescence is observed on the FGF-2LB/PCN sample (Fig. 3B), indicating significant release of the FGF-2LB during the release experiment. In contrast, intense fluorescence is observed on the FGF-2LB/PCN + PEM sample, indicating that significant amounts of FGF-2LB are retained on the fibers when the PEM is present.

3.2. Release of FGF-2LB/PCN complexes from chitosan fibers with and without PEM To evaluate the release of the FGF-2LB/PCN, a 30 day release study was conducted. Table 1 summarizes each of the conditions studied. The cumulative release of FGF-2LB from 3 cm2 samples of fiber networks is shown in Fig. 4. When the FGF-2LB/PCN complexes are adsorbed to the fiber networks in the absence of additional PEM coatings, the complexes are released over at least 27 days, with apparent zero-order kinetics. However, when a single bilayer of heparin–TMC PEM is adsorbed on top of the PCNs, there is no discernible release of FGF-2LB into the supernatant. The presence of the PEM effectively traps the FGF-2LB, preventing its release into solution. The PCN + PEM fibers did exhibit some non-zero fluorescence readings that tended to increase during the course of the release study, but these readings were no different from control samples containing no FGF-2LB (labeled ‘‘Fibers Only’’ in Fig. 4). These release profiles are consistent with the observation of very little fluorescence on the FGF-2/PCN sample and substantial fluorescence on the FGF-2/PCN + PEM sample after

Fig. 4. Cumulative release of FGF-2LB from chitosan fibers. Data represent the mean and standard deviation of triplicate experiments for each condition, using 3 cm2 of fibers for each experiment.

the release experiment (Fig. 3B and C). Large uncertainties in Fig. 4 arise in part because the release data are not readily normalized to the mass of each fiber sample; the masses of the triplicate 3 cm2 samples for each condition were very small and could not be accurately determined. 3.3. Biological activity of preconditioned FGF-2/PCN complexes The biological activity of the FGF-2/PCN complexes was evaluated by measuring MSC proliferation. These experiments are designed to determine whether the biological activity of FGF is preserved over 30 days when (i) FGF-2/PCN are released from surfaces and (ii) FGF-2/PCN are retained on surfaces by the use of PEMs. We have previously demonstrated that FGF-2 increases ovine MSC proliferation under low-serum conditions, while the absence of FGF-2 under low-serum conditions hinders proliferation [16]. In the current work, biological activity of FGF-2 in the PCNs was evaluated by preconditioning PCNs for 1, 9, 18 and 27 days, both in solution (FGF-2/PCN) and adsorbed to chitosan and coated with a heparin– TMC bilayer (FGF-2/PCN + PEM). Negative controls for both of these conditions were prepared similarly, but without the FGF-2 (PCN and PCN + PEM). Table 2 summarizes the nomenclature for these four experimental and control conditions. After the 1, 9, 18 or 27 day preconditioning, MSCs were cultured on fibronectincoated 24-well TCPS plates for 4 days, with PCNs or FGF-2/PCN complexes added to the culture medium. For the PCN + PEM and FGF-2/PCN + PEM conditions, the preconditioned surfaces were coated with fibronectin, and the MSCs were cultured directly on the surfaces. The FGF-2 loading in the PCNs was used to determine the amount of FGF-2/PCN complexes necessary to dose the MSCs with FGF-2 at approximately 0.2 ng ml1. This dose is lower than the optimally mitogenic dose of 1 ng ml1 FGF-2 that we had previously determined, for FGF-2 delivered in solution [16]. Fig. 5 shows normalized cell density (Fig. 5A and D) after 4 days of culture for all of the experimental conditions listed in Table 2, and representative fluorescence micrographs (Fig. 5B, C, E and F). The top row of Fig. 5 shows cells grown on preconditioned PCN + PEM and FGF-2/PCN + PEM surfaces, and the bottom row of Fig. 5 shows cells grown on fibronectin-coated TCPS, dosed with preconditioned PCNs and FGF-2/PCNs in solution. The cell densities in Fig. 5A and D are all normalized to MSCs grown on bare TCPS without FGF-2. After 4 days the bare TCPS displayed a fairly low cell density (4  104 cells cm2) compared to MSCs treated with the optimally mitogenic dose of soluble FGF-2 (1 ng ml1 FGF-2 on bare TCPS), which were confluent, with a cell density of 1.4  105 cells cm2 (not shown). TCPS coated with PCNs and PEMs showed poor cell growth in the absence of FGF-2, which is in agreement with our previous work [16], even though the PEMs are coated with the adhesive protein fibronectin (Fig. 5A, white bars, and Fig. 5B). We believe that this is due in part to reduced cell adhesion to the PEM-coated surface compared to MSCs cultured on uncoated TCPS (white bars in Fig. 5A are all <1). FGF-2/PCN complexes beneath the PEM bilayer did increase the proliferation (Fig. 5A, black bars, and Fig. 5C), however this difference was only statistically significant for the PEMs preconditioned for 1 day. In fact, the difference in cell density between the cells cultured on PCN + PEM and FGF-2/PCN + PEM conditions is reduced as the preconditioning time increases. When PCNs (without FGF-2) in solution are added to the MSC cultures on fibronectin-coated TCPS, poor cell proliferation is observed (Fig. 5D, white bars, and Fig. 5E). However, when FGF-2/PCN complexes are delivered in solution, FGF-2 retains its mitogenic activity over 30 days (27 days of preconditioning + 4 days of cell culture) (Fig. 5D, black bars, and Fig. 5F). The MSCs cultured at this condition are nearly confluent in the image shown in Fig. 5F. In summary, PCNs without FGF-2, either adsorbed to surfaces in

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

1557

Fig. 5. (A, D) MSC density for all experimental conditions listed in Table 2. Cells were seeded at 7000 cells cm2 and cultured in low-serum medium for 4 days. Different days indicate the amount of days that the tested samples were preconditioned before cell seeding. (B, C, E, F) Representative fluorescence microscopy images of MSCs stained with DAPI (nuclei) and calcein-AM (cytoplasm). (A–C) Cells cultured on TCPS coated with chitosan, PCNs (or FGF-2/PCN complexes), one bilayer of TMC–heparin PEM and fibronectin. (D–F) Cells cultured on fibronectin-coated TCPS with PCNs (or FGF-2/PCN complexes) delivered in solution. § in (A) indicates statistically different results for the two conditions. ⁄ in (D) indicates that the FGF-2/PCN-complexes preconditioned for 1, 18 and 27 days in solution result in cell densities that are statistically different from all other conditions in both (A) and (D) that correspond to the same preconditioning time (n = 3, p < 0.05).

polysaccharide PEMs or delivered in solution, tend to inhibit MSC proliferation (white bars in Fig. 5A and D, respectively). However, FGF-2/PCN complexes enhance cell proliferation. Whether delivered in solution or on the surface, the mitogenic activity of the FGF-2 apparently decreases with increasing preconditioning time, but the activity is preserved much longer when the FGF-2/PCN complexes are preconditioned in solution than when they are preconditioned in the adsorbed state. 4. Discussion and conclusions Several studies have developed materials for the delivery of FGF-2 based on sulfated or sulfonated synthetic or natural materials [13–16], showing promising results. Sulfated glycosaminoglycans, such as heparin and chondroitin sulfate, are especially promising as biomaterials because they can protect growth factors from both chemical and proteolytic degradation and enhance their activity by several mechanisms [10,11]. Of particular relevance to the current work, Kim et al. [29] recently demonstrated the release of bioactive FGF-2 for a period of 30 days, using electrospun polycaprolactone/gelatin blends containing heparin. They evaluated the biological activity using human umbilical vascular endothelial cells. In the absence of heparin they noticed a burst release, and the FGF-2 quickly became inactive. Thus, in their work, heparin served to both control the FGF-2 release and to preserve its biological activity. In the present work, we develop a polysaccharide-based nanoassembly using electrospun fibers as a high-surface-area porous structure, heparin-containing PCNs as a vehicle for FGF-2 and heparin-containing PEMs to control the FGF-2 release. The use of PCNs as a delivery vehicle is validated by the demonstrated ability of

PCNs to protect FGF-2 in solution over 30 days. Furthermore, the use of PCNs to carry the growth factors might enable one to tailor cocktails of multiple growth factors by combination of PCNs containing different heparin-binding growth factors. The FGF-2 loading of the PCNs (37.5 ng of FGF-2 per mg of PCNs) reported here represents the FGF-2 that remains stably bound to the PCNs after 48 h of dialysis. In recent work by Tang et al. [30], heparin/chitosan/poly(c-glutamic acid) nanoparticles bound between 130 and 255 ng of FGF-2 mg1 (3.5–7 times the amount of FGF-2 bound in this study). Their PCNs were collected by centrifugation for 20 min after FGF-2 binding. In their release study, approximately 85% of this bound FGF-2 was released from the PCNs within 48 h, when the PCNs were incubated at pH 7.4 [30]. After 48 h, there was very little measured release of the remaining of FGF-2 (up to 72 h) in their work. Hence our FGF-2 loading measured after 48 h of dialysis (37.5 ng of FGF-2 per mg of PCNs) is comparable to the high end of the range of FGF-2 stably bound in the work by Tang et al. (20–38 ng of FGF-2 per mg of PCNs). The adsorption of FGF-2LB/PCN complexes to fibers was confirmed by SEM and confocal images. In the SEM images, the complexes appear as small (<100 nm diameter) features on the surfaces of the fibers. In the confocal images, they appear as punctate fluorescent spots throughout the fiber network. In the SEM images, the apparent diameter of the complexes is smaller when they are adsorbed to the fibers and dried for SEM imaging, compared to the hydrodynamic radius that is observed by DLS when they are hydrated in aqueous solutions, as we have noted in previous reports [19,20]. Over a period of 27 days, FGF-2 and PCNs were constantly released from the fibers modified with FGF-2LB/PCN, while no significant release was observed on the FGF-2LB/PCN + PEM samples.

1558

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559

We hypothesize that the primary mechanism of FGF-2LB release from the fiber networks is by desorption of the PCNs. It is also possible that FGF-2LB desorbs from the complexes, though the results here do not indicate that this is the dominant mechanism of FGF2LB release. In a separate study, PCNs were prepared from rhodamine-labeled chitosan and the release of rhodamine from fiber networks was observed under otherwise identical conditions to those used above for FGF-2LB/PCN complexes. (Data not shown.) Furthermore, the SEM image in Fig. 2C shows fewer nanoparticle features on the surfaces of the fibers than the corresponding image prior to release in Fig. 2A. Also, the FGF-2/PCN complexes retain the activity of FGF-2 when preconditioned in solution for up to 27 days at 37 °C. This suggests that the FGF-2 is likely strongly and stably associated in these complexes, while FGF-2 preconditioned alone completely loses its mitogenic activity with respect to MSCs in 48 h. In this work, the biological evaluation of the FGF-2 activity confirmed its bioactivity by stimulating the proliferation of MSCs using both FGF-2/PCN complexes in solution and adsorbed within PEMs (Fig. 5). The large uncertainties on the data for the cell proliferation from the FGF-2/PCN complexes in solution in Fig. 5D arise because the cells in some of the wells became so confluent that cells began to lift off. Cells that lifted from the surface were not counted. When FGF-2/PCN complexes were adsorbed in heparin–TMC PEMs, a single PEM bilayer was sufficient to retain the complexes on the fiber networks for the duration of the release experiments. However, we also note that increasing the preconditioning time for the FGF-2/PCN + PEM condition resulted in a decrease in the biological activity of the FGF-2. This could be due to slow loss of the FGF-2 from the flat surfaces (either through protein or complex desorption) during preconditioning that was not detected in the analogous experiment conducted on the fibers, or due to slow denaturation of the FGF-2. In our previous work, we delivered FGF-2 to MSCs by adsorbing the FGF-2 to chitosan–heparin PEMs and culturing the MSCs on the surfaces for 4 days. In that work, we found that the MSCs exhibited reduced cell density and tended to reduce spreading, particularly on chitosan-terminated PEM surfaces. Cell densities were highest on the surfaces that were heparin-terminated and modified with FGF-2. In the present work, we also observed reduced cell numbers on the PEM-coated surfaces without FGF-2 (Fig. 5A and B). Interestingly, in this work we also show that delivery of heparin–chitosan PCNs in solution to MSCs on fibronectin-coated TCPS reduces the cell number (Fig. 5D and E). In contrast, FGF-2/PCN complexes promote MSC proliferation. This effect is stronger when the complexes are delivered in solution than when the complexes are adsorbed to a surface, and is maintained even when the complexes are incubated at 37 °C for over 30 days. This work did not investigate the mechanism whereby polysaccharide-based PEMs or PCNs seem to inhibit cell proliferation in the absence of FGF-2. We anticipate that the PCNs may interfere with cell-surface interactions. Future studies should also investigate the potential cytotoxicity of the PCNs. The polysaccharide-based nanoassemblies presented here combine structural and biochemical features of the extracellular matrix. These features could be used as scaffold materials to support the growth of primary cells, such as bone-marrow-derived MSCs, and also to present otherwise unstable heparin-binding growth factors in a context that mimics important features of the native biological context in which these growth factors are found. We demonstrate that the growth factor activity can be preserved for 30 days, and that its mode of delivery (either retained on the fibers’ surfaces or slowly released into solution) can be modulated. The time scale associated with this stabilization and release of growth factor is appropriate for stem cell differentiation, wound healing and tissue regeneration, suggesting that these materials

could be further developed for therapeutic applications in skeletal tissues. Acknowledgements The authors acknowledge the ‘‘Provincia Autonoma di Trento’’ (PAT) for funding the SmartBone project, which led to the conclusion of this work, and the National Science Foundation (DMR 0847641). We also acknowledge funding from a Core Infrastructure Grant for Microscope Imaging from Colorado State University. References [1] Li RH, Wozney JM. Delivering on the promise of bone morphogenetic proteins. Trends Biotechnol 2001;19:255–65. [2] Lee J-Y et al. Characterization of the surface immobilized synthetic heparin binding domain derived from human fibroblast growth factor-2 and its effect on osteoblast differentiation. J Biomed Mater Res A 2007;83A:970–9. [3] Hanada K, Dennis JE, Caplan AI. Stimulatory effects of basic fibroblast growth factor and bone morphogenetic protein-2 on osteogenic differentiation of rat bone marrow-derived mesenchymal stem cells. J Bone Miner Res 1997;12:1606–14. [4] Vincent T, Hermansson M, Bolton M, Wait R, Saklatvala J. Basic FGF mediates an immediate response of articular cartilage to mechanical injury. Proc Natl Acad Sci USA 2002;99:8259–64. [5] Zamora PO, Tsang R, Peña LA, Osaki S, Som P. Local delivery of basic fibroblast growth factor (bFGF) using adsorbed silyl-heparin, benzylbis(dimethylsilylmethyl)oxycarbamoyl-heparin. Bioconjug Chem 2002;13:920–6. [6] Tessmar JK, Göpferich AM. Matrices and scaffolds for protein delivery in tissue engineering. Adv Drug Deliv Rev 2007;59:274–91. [7] Ziegler J, Anger D, Krummenauer F, Breitig D, Fickert S, Guenther K-P. Biological activity of recombinant human growth factors released from biocompatible bone implants. J Biomed Mater Res A 2008;86A:89–97. [8] Ziegler J, Mayr-Wohlfart U, Kessler S, Breitig D, Gunther K-P. Adsorption and release properties of growth factors from biodegradable implants. J Biomed Mater Res 2002;59:422–8. [9] Uludag H, Gao T, Porter TJ, Friess W, Wozney JM. Delivery systems for BMPs: factors contributing to protein retention at an application site. J. Bone Jt. Surg. Am. 2001;83-A:S128–35. [10] Boddohi S, Kipper MJ. Engineering polysaccharide nanostructures. Adv Mater 2010;22:2998–3016. [11] Damon DH, Lobb RR, D’Amore PA, Wagner JA. Heparin potentiates the action of acidic fibroblast growth factor by prolonging its biological half-life. J Cell Physiol 1989;138:221–6. [12] Guimond S, Maccarana M, Olwin BB, Lindahl U, Rapraeger AC. Activating and inhibitory heparin sequences for FGF-2 (basic FGF). J Biol Chem 1993;268:23906–14. [13] Wenk E, Murphy AR, Kaplan DL, Meinel L, Merkle HP, Uebersax L. The use of sulfonated silk fibroin derivatives to control binding, delivery and potency of FGF-2 in tissue regeneration. Biomaterials 2010;31:1403–13. [14] Jeon O, Kang S-W, Lim H-W, Hyung Chung J, Kim B-S. Long-term and zeroorder release of basic fibroblast growth factor from heparin-conjugated poly(llactide-co-glycolide) nanospheres and fibrin gel. Biomaterials 2006;27:1598–607. [15] Macdonald ML, Rodriguez NM, Shah NJ, Hammond PT. Characterization of tunable FGF-2 releasing polyelectrolyte multilayers. Biomacromolecules 2010;11:2053–9. [16] Almodóvar J, Bacon S, Gogolski J, Kisiday JD, Kipper MJ. Polysaccharide-based polyelectrolyte multilayer surface coatings can enhance mesenchymal stem cell response to adsorbed growth factors. Biomacromolecules 2010;11:2629–39. [17] Boddohi S, Killingsworth CE, Kipper MJ. Polyelectrolyte multilayer assembly as a function of pH and ionic strength using the polysaccharides chitosan and heparin. Biomacromolecules 2008;9:2021–8. [18] Boddohi S, Moore N, Johnson PA, Kipper MJ. Polysaccharide-based polyelectrolyte complex nanoparticles from chitosan, heparin, and hyaluronan. Biomacromolecules 2009;10:1402–9. [19] Boddohi S, Almodóvar J, Zhang H, Johnson PA, Kipper MJ. Layer-by-layer assembly of polysaccharide-based nanostructured surfaces containing polyelectrolyte complex nanoparticles. Colloids Surf B 2010;77:60–8. [20] Almodóvar J, Kipper MJ. Coating electrospun chitosan nanofibers with polyelectrolyte multilayers using the polysaccharides heparin and N,N,Ntrimethyl chitosan. Macromol Biosci 2011;11:72–6. [21] Almodóvar J, Place L, Gogolski J, Erickson K, Kipper MJ. Layer-by-layer assembly of polysaccharide-based polyelectrolyte multilayers: a spectroscopic study of hydrophilicity, composition, and ion pairing. Biomacromolecules 2011;12:2755–65. [22] Lee KY, Jeong L, Kang YO, Lee SJ, Park WH. Electrospinning of polysaccharides for regenerative medicine. Adv Drug Deliv Rev 2009;61:1020–32. [23] Teng S-H, Wang P, Kim H-E. Blend fibers of chitosan–agarose by electrospinning. Mater Lett 2009;63:2510–2.

F. Zomer Volpato et al. / Acta Biomaterialia 8 (2012) 1551–1559 [24] Homayoni H, Ravandi SAH, Valizadeh M. Influence of the molecular weight of chitosan on the spinnability of chitosan/poly(vinyl alcohol) blend nanofibers. J Appl Polym Sci 2009;113:2507–13. [25] Sangsanoh P, Supaphol P. Stability improvement of electrospun chitosan nanofibrous membranes in neutral or weak basic aqueous solutions. Biomacromolecules 2006;7:2710–4. [26] Hasegawa M, Isogai A, Onabe F, Usuda M, Atalla RH. Characterization of cellulose–chitosan blend films. J Appl Polym Sci 1992;45:1873–9. [27] de Britto D, Assis OBB. A novel method for obtaining a quaternary salt of chitosan. Carbohydr Polym 2007;69:305–10.

1559

[28] Crouzier T, Ren K, Nicolas C, Roy C, Picart C. Layer-by-layer films as a biomimetic reservoir for rhBMP-2 delivery: controlled differentiation of myoblasts to osteoblasts. Small 2009;5:598–608. [29] Kim MS, Shin YM, Lee J-hye, Kim SI, Nam YS, Shin CS, et al. Release kinetics and in vitro bioactivity of basic fibroblast growth factor: effect of the thickness of fibrous matrices. Macromol Biosci 2011;11:122–30. [30] Tang D-W, Yu S-H, Ho Y-C, Mi F-L, Kuo P-L, Sung H-W. Chitosan/poly(cglutamic acid) nanoparticles for multi-functional delivery of fibroblast growth factor and heparin. Biomaterials 2010;31:9320–32.