Release of Nerve Growth Factor from HEMA Hydrogel-Coated

174

Biomacromolecules 2009, 10, 174–183

Release of Nerve Growth Factor from HEMA Hydrogel-Coated Substrates and Its Effect on the Differentiation of Neural Cells Shalin J. Jhaveri,†,‡ Matthew R. Hynd,§ Natalie Dowell-Mesfin,§ James N. Turner,§ William Shain,§ and Christopher K. Ober*,† Department of Materials Science and Engineering and Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853 and Wadsworth Center, NYS Department of Health, Albany, New York 12201 Received September 28, 2008

Local pharmacological intervention may be needed to ensure the long-term performance of neural prosthetic devices because of insertion-related neuron loss and reactive cell responses that form compact sheaths, leading to decreased device performance. We propose that local delivery of neurotrophins would enhance neuron survival and promote neuron sprouting toward device electrodes, thus providing improved electrode-neuron communication and device performance for recording and stimulating CNS activity. In this study, three different types of poly(2hydroxyethyl methacrylate) (pHEMA) hydrogels were developed and assessed for storage capacity and release rates of the neurotrophin, nerve growth factor (NGF). Additionally, a method was developed for routine coating of microfabricated neuroprosthetic devices with the different pHEMA hydrogels. Biological responses to hydrogeldelivered NGF from the devices were measured using primary cell cultures of dorsal root ganglion (DRG) neurons. Neuron process growth was used to assess biological responses to released NGF. When targeted media concentrations were the same, responses to bath-applied NGF and NGF released from pHEMA hydrogels were not significantly different. When NGF was released from lysine-conjugated pHEMA hydrogels, a significant increase in process growth was observed. Our studies demonstrate that pHEMA coatings can be used on neural devices consistent with the needs for local neurotrophin delivery in the brain.

Introduction Microfabricated neural prosthetic devices are in the early stages of design and development for chronic recording and stimulation of nervous system function.1–4 Interfacing the nervous system with such electronic devices for recording or stimulation purposes holds great promise for acquiring new knowledge of nervous system function as well as treatment following injury or disease. Examples include recording devices that can be used for brain-computer interfaces resulting in controlled robotic movements5–7 or computer cursor movement interfaces.8,9 Other approaches include deep brain stimulators that are being routinely used for treating Parkinson’s disease and movement disorders10–12 or cochlear implants that are being used to treat hearing loss.13,14 However, the long-term performance of microfabricated devices is, in many cases, compromised by neuron loss15 and reactive cell responses.15,16 Immediate and short-term responses are due to damage associated with device insertion,17–19 whereas long-term, sustained responses result from brain-device interactions.19,20 To make devices biocompatible and to aid nerve cell regrowth, researchers have used surface-modified devices21–23 or have employed the delivery of neurotrophins/anti-inflammatory drugs.24–28 In this article, neurotrophins were delivered from poly(2hydroxyethyl methacrylate) (pHEMA) hydrogel-coated substrates. pHEMA was chosen as our base matrix for delivery of neurotrophins because it is mechanically soft and is capable of controlled drug release.29–32 These hydrogels are FDA approved, * To whom correspondence should be addressed. Tel: (607) 255-8417. Fax: (607) 255-2365. E-mail: [email protected]. † Department of Materials Science and Engineering, Cornell University. ‡ Department of Chemistry and Chemical Biology, Cornell University. § NYS Department of Health.

they absorb water close to the levels of living tissues, they can be fine-tuned to possess mechanical properties close to those of brain tissue, and both their hydroxyl groups and the network mesh size can be readily modified.33–35 The fractional water content in living tissue is between 0.68 and 0.80,36 and pHEMA hydrogels cross-linked with ethylene glycol dimethacrylate (EGDMA) have been shown to have water content as high as 0.76.37 pHEMA hydrogels have a number of additional properties that may improve long-term device performance: enhanced biocompatibility, low toxicity, and successful use for drug delivery.38–42 Their hydrophilicity and porous nature make these materials extremely good candidates for controlled drugdelivery,29–32 especially those based on therapeutic proteins.43–47 Physical properties of these hydrogels, such as degree of swelling, can be controlled by cross-linking and the introduction of charged moieties. Finally, hydrogels in general are inert to normal biological processes, show resistance to degradation, withstand the heat of sterilization without damage, can be easily processed, and can be prepared in a variety of shapes and forms. To demonstrate the effectiveness of coated devices, we conducted an in vitro study of nerve growth factor (NGF) release from controlled cross-link density pHEMA hydrogel-coated substrates using dorsal root ganglia (DRG) neurons. NGF release can enhance neuronal survival near the site of the insertion of the device and promote neuronal sprouting toward the device. There have been no prior studies, to the best of our knowledge, that detail the effect of release and release rate of NGF from pHEMA hydrogels and their effects on the differentiation of neurons. By comparing neuron responses to bath-applied and hydrogel-released NGF, we show that by releasing NGF from carefully designed hydrogels, greater neural differentiation is possible. This could enhance the electrode-neuron communica-

10.1021/bm801101e CCC: $40.75  2009 American Chemical Society Published on Web 12/05/2008

NGF Release from HEMA Hydrogel-Coated Substrates

Biomacromolecules, Vol. 10, No. 1, 2009

175

Figure 1. Chemical structures of components used to make poly(2-hydroxyethyl methacrylate) hydrogels. (a) Initiator, 2,2-dimethoxy-2-phenyl acetophenone; (b) monomer, 2-hydroxyethyl methacrylate (HEMA); and (c) cross-linker, ethylene glycol dimethacrylate. Scheme 1. Reagents and Conditions Used for Lysine Conjugation to Poly(2-hydroxyethyl methacrylate) (pHEMA)

tion and lead to improved device performance in vivo. The utility of this method demonstrates not only the retention of the biological activity of loaded and released proteins from the hydrogels but also its effectiveness for release into the brain from inserted devices. Additionally, researchers in the past have used materials having a thickness of >1 mm for NGF delivery.24,25,48 However, there is likely more damage to the neural tissue during device insertion with an increase in the device size.17 In this article, we have used ∼300-µm-thick materials for NGF delivery and have demonstrated a method for enhancing the neural response without compromising the biological activity of the released NGF and the coating size. DRG neurons were used as our model target, because a subpopulation of these cells responds to NGF by rapid extending neuronal processes. We report here methods for hydrogel modification, device coating, and hydrogel protein loading and release. Our results indicate that controlled local neurotrophin release may be incorporated into intervention strategies for enhancing microfabricated device performance.

Materials and Methods Materials. 2-Hydroxyethyl methacrylate (HEMA), EGDMA, 2,2dimethoxy-2-phenyl acetophenone (Irgacure 651), NR,Nε-di-Fmoc-Llysine (lysine-fmoc), 4-(dimethylamino) pyridine (DMAP), 1,3dicyclohexylcarbodiimide (DCC), and bovine serum albumin conjugated with fluorescein isothiocyanate (BSA-FITC) were supplied by SigmaAldrich (St. Louis, MO). Texas-Red-labeled dextrans (Dex10K, MW 10 kDa; Dex70K, MW 70 kDa) and CBQCA protein quantitation kit were purchased from Invitrogen (Carlsbad, CA). Mouse NGF (2.5s subunit) was purchased from Roche Applied Science (Indianapolis, IN). Arch Punch (10 mm diameter) was purchased from McMaster-Carr (New Brunswick, NJ). BK-7 optical glass (7.6 cm diameter × 0.45

cm thick, transmittance 95% at 365 nm) was purchased from Esco Products (Oak Ridge, NJ). Micromachined neural prosthetic devices were obtained from NeuroNexus Technologies (Ann Arbor, MI). Phosphate-buffered saline (PBS; pH 7.4) was made using the following recipe: 0.016 M NaH2PO4, 0.08 M Na2HPO4, and 150 mM sodium chloride. All other chemicals were obtained from Sigma-Aldrich and were used as received unless otherwise noted. Methods. pHEMA Hydrogel Disks. Before use, HEMA was vacuum distilled at 80 °C to remove monomethyl ether hydroquinone inhibitor. A prepolymer solution consisting of 62.5% (w/w) HEMA, 0.003% (w/ w) EGDMA, and 0.006% (w/w) Irgacure 651 was prepared in distilled water. We synthesized hydrogel disks by pipetting the prepolymer solution between two Parafilm-covered optical glass slides separated by a 1-mm-thick spacer. The entire assembly was exposed to UV light (365 nm, 1.8 mW/cm2) using a hand-held UV lamp (Spectroline; Westbury, NY) for 10 min. Following polymerization, hydrogels were peeled off the Parafilm-covered glass surface, and smaller hydrogel disks were made using a 10 mm Arch punch. Subsequently, the disks were placed in distilled water for 2 days to remove unreacted monomer with periodic replacement of the water used for leaching; consistent with other published reports for the removal of unreacted reagents.49–53 Macroporous pHEMA-Sodium Chloride (pHEMA-NaCl) Hydrogel Disks. Large pore-size hydrogel disks were prepared in the same manner as the standard pHEMA hydrogels, except that HEMA was diluted to 40% (w/w) in a 0.6 M NaCl solution before polymerization. pHEMA and pHEMA-NaCl DeVice Coatings. Micromachined devices were pretreated with 3-methacryloxypropyl trimethoxysilane as previously described by Hynd et al.54 Silanized devices were individually inserted in sterile polyethylene (PE) tubing (o.d. ) 640 µm; i.d. ) 280 µm; Scientific Commodities, Lake Havasu City, AZ) containing either HEMA or HEMA-NaCl prepolymer solutions. Tubes were placed between two hand-held ultraviolet lamps, and the solution was polymerized by exposure to UV light for 4 min. Following polymerization, the outer surface of each tube was heated for 3 s using a heat

176


Jhaveri et al.

Figure 2. (a,d,g) Representative SEM images of the surface of pHEMA, (b,e,h) pHEMA-NaCl, and (c,f,i) pHEMA-Lys hydrogel samples. Images g-i are of cross-sections through hydrogel samples. Micrometer-sized pores can be seen in the SEM images of pHEMA-NaCl hydrogel, much smaller pores are observed in the pHEMA samples, and no visible pores are seen in the pHEMA-Lys samples. Scale Bar ) 5 µm.

gun, and the tube was placed in concentrated HCl (36-38%) for 1 min. Coated devices were then removed from the tubes by the use of tweezers and were placed in distilled water to remove unreacted monomer with periodic replacement of the water used for leaching. pHEMA-Lysine (pHEMA-Lys) Hydrogel Disks and DeVice Coatings. pHEMA hydrogel disks were prepared as described above and were then reacted with lysine-fmoc in the following manner to obtain lysine-conjugated pHEMA hydrogels. A pHEMA hydrogel disk was placed in a 15 mL vial, and lysine-fmoc was added in a molar ratio of 0.5:1 lysine/fmoc to hydroxyl groups on pHEMA. DCC and DMAP were added sequentially in a molar ratio of 1.5 and 0.05 mol with respect to lysine-fmoc. The mass of pHEMA used in these conjugations was based on the geometry of the hydrogel and the molar mass of HEMA. Dimethylformamide (DMF, 5 mL) was used as the solvent. The reaction vial was then placed on a shaker for 12 h at room temperature. After conjugation, the disk was removed from the reaction vial and placed in fresh DMF for 12 h to remove any unreacted reagents. We deprotected the fmoc group by transferring the hydrogel disk to 20% piperidine in DMF (v/v), and the mixture was stirred for 2 h at room temperature. Finally, the disk was placed in DMF for 2 h and transferred to distilled water until use. For the preparation of pHEMA-Lys-coated devices, pHEMA-coated devices were prepared as described above and were then inserted in a 2 mm i.d. PE tube for lysine modification. The tubes containing the devices were placed in a vial containing DMF. Hydrogels were secured in the 2 mm tubes following swelling in DMF. DCC, DMAP, and lysine-fmoc were added in the same molar ratio as that used for making pHEMA-Lys disks. After conjugation, tubes were removed from the reaction vial and placed in a new vial containing fresh DMF for 12 h to remove any unreacted reagents. We deprotected the fmoc group by transferring the tubes to a vial containing 20% piperidine in DMF (v/ v) and stirring the reaction for 2 h at room temperature. The tubes were then placed in DMF for 2 h and transferred to distilled water for 1 h. Finally, the tubes were removed from distilled water, and the devices were pulled out of the tube by the use of tweezers and were placed in distilled water until further use. Lysine coupling was

confirmed by three methods. In the first, the piperidine-fmoc adduct formed during conjugation was detected using a UV spectrophotometer (Lamda 10, PerkinElmer, Waltham, MA) at 301 nm. This provided indirect evidence of lysine conjugation. In the second, lysine was cleaved from lysine-modified hydrogel samples using trifluoroacetic acid (TFA), and electrospray ionization mass spectrometry (ESI-MS) was performed. This provided direct evidence of lysine conjugation. TFA hydrolysis was carried out in the following manner. The pHEMALys hydrogel disk were first placed in individual 15 mL vials, 1 mL of TFA was added, and the vials were shaken for 1 h. The TFA solution was then pipetted in new vials, the solution was concentrated using a rotary evaporator, and the vials were subsequently placed overnight in a vacuum oven at 40 °C. The concentrated solution was then analyzed using ESI-MS. In a third separate method, total lysine content of acid hydrolysis was estimated using the ninhydryin assay method.55 This provided a measurement of the total lysine conjugated and allowed us to calculate the efficiency of lysine conjugation. Scanning Electron Microscopy (SEM) Analysis of the Hydrogel Samples. Hydrogel samples were cut with a razor blade and dehydrated as previously described by Mikos et al.56 Samples were sputtered coated with gold/palladium for 30 s. The surfaces of the gold/palladium-coated hydrogels samples were observed under SEM (Leica 440) at 15 kV. Dynamic Mechanical Analysis. Mechanical properties were tested using DMA 2980 (TA Instruments) in tensile test geometry. The hydrogel samples (25 × 15 × 1 mm3) were submerged in deionized (DI) water for measurements. Tests were performed in controlled force mode. The preload force was 0.01 N and the ramp force was 0.01 N/min up to 0.2 N. We analyzed the data by calculating the Young’s modulus from the slope of the force curve. Hydrogel Loading. Initial experiments for characterizing hydrogel loading properties were performed using pHEMA hydrogel disks (diameter ) 1 cm; thickness ) 1 mm). The hydrogel disks were placed in 15 mL vials and loaded with Dex10K or Dex70K for 24 h with stirring. For coated microfabricated devices, we performed loading by placing each coated device in individual wells of a 96-well plate containing different concentrations of NGF in PBS for 24 h, and the plate was rocked to provide stirring. Samples were then removed from



177

Table 1. Physical Characteristics (Modulus, Cross-Link Density, and Average Pore Size) of Different pHEMA Hydrogels

hydrogel type

modulus (E) (kPa)

molecular weight between cross-links (Mc) (g/mol)

pHEMA pHEMA-lysine pHEMA-NaCl

440 ( 10 350 ( 7 93 ( 5

22 177 ( 515 26 486 ( 515 102 701 ( 5529

the loading solution, and excess solution was removed by gentle blotting with tissue paper. We determined the amount of NGF that was loaded in the devices by estimating the cumulative release after 48 h. No significant amount of NGF was released after that time, and hence that release amount was noted as the amount loaded in the hydrogels. Release Measurements. Hydrogel disks were placed in standard disposable cuvettes containing PBS for release studies. The readings were taken at predetermined times from 1 min to 48 h. For each reading, the release solution was pipetted and transferred to a new cuvette. After each reading the release solution was replaced by fresh PBS. We determined the released quantity by measuring the fluorescence intensity of the released labeled dextran. For experiments with hydrogel-coated devices loaded with NGF, release was quantified using the CBQCA protein quantitation kit, as directed by the manufacturer’s instructions, except that NGF was used to prepare protein standards. Also, because the reaction buffer required for the CBQCA assay was 0.1 M sodium borate buffer (pH 9.3), the release solution (PBS buffer) was extensively dialysed against borate buffer using the Slide-A-Lyzer MINI dialysis unit (Pierce Biotechnology; Rockford, IL). After a 2 h incubation, the solution was pipetted in standard disposable microcuvettes for measuring fluorescence intensity with a spectrofluorometer (SLM 8000c, SLM instruments, Urbana, IL) using a photon-counting method. The NGF protein standards were used for determining the exact quantity released. Dorsal Root Ganglion Cell Culture. DRG neurons were obtained from postnatal day 2 (P2) Sprague-Dawley rats, as previously described by Thompson et al.57 Wadsworth Center Institutional Animal Care and Use Committee (IACUC) approved all animal procedures. DRG cells were cultured in Neurobasal media (Invitrogen) containing 2% B27 supplement with 2.0 mM GlutaMAX at 37 °C on polylysine-coated 18 mm round coverslips in individual wells of a 12-well cell culture tray. Cells were plated at a density of 10 000 cells/well. To measure cell responses to hydrogel-coated devices, devices were preloaded with NGF to produce a final NGF concentration of 10 ng/mL, placed in individual wells of a 12-well plate, and incubated for 3 days. Thus, each well contained DRG cells and a single device. For bath-applied conditions, NGF was diluted in culture medium.

ν2,s

cross-link density × 105(mol/cm3)

average pore size (Å)

0.50 ( 0.01 0.38 ( 0.01 0.40 ( 0.003

5.82 ( 0.14 4.87 ( 0.01 1.25 ( 0.07

94.07 ( 1.10 113.17 ( 1.10 16000 ( 7000

Cell Fixation and Immunocytochemistry. Cells were fixed after the 3 day incubation by the use of the standard fixation procedure.58 Briefly, coverslips were rinsed in Ca2+- and Mg2+-free HEPES-buffered Hanks’ saline (HBHS) prewarmed to 37 °C for 2 min. Cells were fixed using 4% paraformaldehyde containing 0.5% Triton X-100 for 1 min at 4 °C then 4% paraformaldehyde for 15 min at 37 °C. Cells were rinsed in HBHS for 5 min and blocked in 6% BSA for 30 min at room temperature. Coverslips were rinsed once in HBHS and incubated in anti-β-III-tubulin primary antibody (1.0 mg/mL, rabbit polyclonal, 1:500; Sigma) for 1 h at 37 °C. Subsequently, they were rinsed three times in HBHS and incubated in secondary antibody (Texas Red antirat, 1:1000 dilution) for 45 min at 37 °C. Cell nuclei were stained using Hoechst 33342 (5.0 µL of a 5 mg/mL stock in 1 mL of distilled H2O). The coverslips were then mounted onto slides using a mounting media consisting of HBHS and glycerol saturated with n-propyl gallate.59 Imaging and Tracing of Dorsal Root Ganglion Cells. After the cell fixation and immunocytochemistry, fluorescently labeled cells were imaged using an Olympus wide-field microscope (Olympus, Center Valley, PA) equipped with a mercury vapor lamp with either a 10× or a 20× objectives. We analyzed images by using an automated 2D tracing program60 to determine total process lengths from randomly selected fields for each coverslip (10 fields/coverslip). Statistical Analysis. Statistical analyses were performed using Systat 11 (Systat Software). Analyses of variance (ANOVA) with Tukey HSD posthoc testing was used to analyze the data. p values of 40%, then heterogeneous hydrogels having undesirable micronsized pores would be obtained,61 and thus water contents of 37.5% (w/w) were routinely used. Such high water content heterogeneous hydrogels also have reduced mechanical strength62 and could be easily damaged during probe insertion in the brain. To maximize the protein/drug loading, the cross-linker concentration was kept at 0.006% (w/w) because the loading is inversely related to the cross-linking density. Increasing the cross-linker concentration to 0.018% decreased the loading capacity by 10-15%. pHEMA-Lysine (pHEMA-Lys) Hydrogels. Lysine-conjugated pHEMA hydrogels were prepared to produce a matrix with a positive charge at physiological pH.2 Fmoc-protected lysine was used to couple pHEMA via DCC coupling reaction. After ester formation, we deprotected Fmoc groups by treating hydrogels with piperidine (Scheme 1). The lysine groups were conjugated post-hydrogel-synthesis to (i) avoid the possible formation of unwanted oxidized species during UV photo-crosslinking and (ii) enable the use of the same process for making the lysine-free hydrogels. We verified pHEMA-lysine conjugation by monitoring piperidine-fmoc adduct formation using UV spectrophotometry at 301 nm (Lambda 10, Perkin-Elmer, MA). This adduct indicated that piperidine had successfully cleaved the fmoc group. The detection of lysine conjugation was carried out after the lysine group was cleaved using TFA and the presence of lysine conjugates was established using ESI-MS. A lysine peak at 147.1 Da (M + H) was clearly seen in the ESI-MS spectra (Supporting Information, Figure S1). To ensure that the lysine was covalently coupled and not just physically adsorbed in the hydrogel, a similar reaction was carried out without the addition of DCC. No lysine peak was seen in the ESI-MS spectra (Supporting Information, Figure S2) under these conditions. Finally, ∼0.410 g lysine/g pHEMA was determined to be coupled to the pHEMA hydrogels by the use of the procedure outlined by McCarley et al.55 Using this method, it was determined that the yield of the reaction was 70-80%. pHEMA-Sodium Chloride (pHEMA-NaCl) Hydrogels. In general, homogeneous hydrogels have smaller nanometer-scale pore sizes compared with heterogeneous hydrogels, which have a large, nonuniform pore size.61 Because a small pore size might be a major limitation to protein loading, we prepared heterogeneous hydrogels with larger pore sizes, on the order of a few micrometers, by synthesizing hydrogels in the presence of NaCl. By adding salt to the pregel solution, an enhanced phase separation between the aqueous and polymer phase is created, which can lead to the production of a macroporous gel structure.56 The large pore size of these heterogeneous hydrogels (assessed by SEM) was shown to influence both neurotrophin loading and release. Whereas heterogeneous hydrogels often have increased water content and reduced mechanical strength, it has been observed that by using 0.6 M NaCl as a diluent, heterogeneous hydrogels with sufficient mechanical strength may be produced.56 SEM images comparing the surface and cross-sections of the homogeneous pHEMA and heterogeneous pHEMA-NaCl hy-

Jhaveri et al.

Figure 4. Device coatings. (a) Photograph of a silicon-based microfabricated neural prosthetic device (shank length ) 3 mm, maximum width ) 123 µm, thickness ) 15 µm). (b) Brightfield micrograph of a pHEMA hydrogel-coated device post-lysine-modification (pHEMA-Lys) showing the tip of the device. Device was imaged in PBS. (c) Fluorescence microscope image of a device coated with fluoresceintagged pHEMA-Lys hydrogel. This representative micrograph illustrates the top view in the midportion of the device. (d) Demonstration of uniform device coatings. Thicknesses were measured from the midline of the device (n ) 3). The exact coating thickness varies as the device width changes along the shank of the device. The values represent mean ( standard deviation. The error bars for these values are shorter than the height of the symbols. For panels b and c, scale bar ) 100 µm.

drogels confirm that the latter hydrogels possess an irregular, porous surface with micrometer-sized pores. In contrast pHEMA



Table 2. Comparison of the Half Life (t1/2) of Release of NGF from the Different Hydrogelsa hydrogel type

t1/2 (min)

pHEMA pHEMA-lysine pHEMA-NaCl

77 ( 28 200 ( 33 31 ( 9

a There is a significant difference between the half life of NGF released from pHEMA-lysine hydrogels and that of NGF released from pHEMANaCl hydrogels. The half life was calculated from the release profiles of NGF from the different hydrogels. The release profiles for the first 4 h are shown in Fig 5. The values represent the mean ( standard deviation obtained from three readings.

and pHEMA-Lys hydrogels have a uniform smooth surface and possess no apparent larger diameter pores when imaged at the same magnification (Figure 2). In the latter case, cross-link density and the average pore size were assessed by dynamic mechanical analysis. The comparison of the pHEMA, pHEMANaCl, and pHEMA-Lys samples demonstrated that the two modifications promoted significant changes in the morphology of the hydrogel surface and bulk of the hydrogel. Dynamic Mechanical Analysis for Determining CrossLink Density, Modulus, and Pore Size. One of the reasons for the lack of biocompatibility between a silicon neural device and neural tissue is the high modulus of the silicon. Controlling the mechanical modulus of the hydrogels is therefore important to ensure that it is comparable to that of the brain tissue so that it can act as a mechanical buffer. The cross-link density, Fx shown in Table 1, was calculated using the equation given by Peppas et al.34

Fx )

1 νMc

(1)

where ν is the specific volume of pHEMA, estimated to be 0.785 cm3/g, and Mc is the effective molecular weight between crosslinks. Also shown in Table 1, pHEMA hydrogels have a Young’s modulus of ∼440 kPa in contrast with silicon, which has a modulus >170 GPa.63 Typical brain tissue has a modulus of ∼10 kPa,64 and hence the pHEMA hydrogels have modulus values much closer to that of brain tissue than to that of silicon. The pHEMA-NaCl hydrogels have a modulus of only ∼93 kPa, whereas the pHEMA-Lys hydrogels have a modulus of ∼350 kPa, slightly lower than the pHEMA hydrogel. Although, it is very important to have a modulus very close to that of the brain tissue, care must also be taken so that the hydrogel is not so soft that it fails upon insertion in the neural tissue. To verify this, the coated substrates were inserted in 4% agarose gels. None of the three coatings were delaminated upon multiple insertions inside the 4% agarose gel. Because considerably softer 0.6% agarose gel closely mimics the mechanical properties of in vivo brain tissue,65 these pHEMA hydrogel coatings should be robust enough to resist delamination upon insertion in the brain. The pore sizes of the pHEMA and pHEMA-Lys networks were estimated using the analysis given by Peppas et al.34 and are listed in Table 1. Briefly, the mesh size or the pore size was calculated using the equation

ξ)

( ) 2CnMc Mo

1/2

-1/3 lν2,s

(2)

where ξ is the mesh size, Cn is the Flory polymer characteristic ratio (6.9, poly(methyl methacrylate) value) as used by Peppas et al.,34 Mc is the effective molecular weight between cross-

179

links calculated using the equation derived from the rubber elasticity theory,66,67 Mo is the molecular weight of the repeating units making up the polymer chain, l is the carbon-carbon double-bond length (1.54 Å), and ν2,s is the equilibrium polymer volume fraction at maximum swelling. ν2,s was calculated using the procedure outlined by Pishko et al.68 In contrast, the average pore size of the pHEMA-NaCl was estimated from the SEM images using the ImageJ software.69 As can be seen, the pore size of pHEMA and pHEMA-Lys is ∼10 nm compared with the micrometer-sized pores for the pHEMA-NaCl hydrogels, which also explains the slower release of protein from pHEMA hydrogels and pHEMA-Lys hydrogels. Hydrogel Loading and Release with Dextrans. Initial loading and release studies were performed using dextrans (Dex) of two different molecular weights with standard pHEMA and pHEMA-NaCl hydrogel disks and hydrogel-coated devices. Fluorescently tagged dextrans were used for determining release rates by means of fluorescence measurements. Dex10K and Dex70K were chosen to be model macromolecules to increase the understanding of the effects of molecular size on hydrogel loading and release characteristics because they provide information that brackets the range of molecular sizes that could be used to load these hydrogel constructs. From the release studies, it was determined that the Dex70K could not be uniformly loaded within the pHEMA hydrogel matrix. Because of their larger pore size, pHEMA-NaCl hydrogels were capable of uniform loading of both the Dex10K and Dex70K. It was determined that both the pHEMA and pHEMA-NaCl hydrogels were capable of uniform loading of globular molecules with molecular weights of at least 26 kDa, which suggests that NGF (MW ≈ 26 kDa) could be uniformly loaded within its matrix, as shown later in this article. The release data of the dextrans from the hydrogels, the calculation of the effective diffusion coefficient values of the dextrans from the hydrogels, and the estimation of release of molecules with MWs of 26 kDa from the hydrogels by the use of design of experiment are provided in the Supporting Information. From these release studies, the half life, or the time at which 50% of the total amount (10 ng/ mL) of 26 kDa dextran is released from hydrogel coated devices, was calculated to be 85 ( 10 min for the pHEMA hydrogel and 30 ( 7 min for the pHEMA-NaCl hydrogel. Device Coating. Typically, ∼1 mm thick slabs of material have been used to study neurotrophin release.24,25 However, because a goal of this work is to reduce device-related tissue damage, it is essential to keep the total device size (microfabricated neural prosthetic device + hydrogel coatings) to dimensions that will not produce such damage (Szarowski et al.). Some of the common techniques for producing thin coatings are dip coating, spray coating, and brush coating. These methods could not produce uniform coatings in a reproducible manner. Therefore, we developed a molding method using photopolymerization of the hydrogel in a PE tube. The PE tubes (o.d. ) 640 µm; i.d. ) 280 µm) have an absorbance of 0.055 (transmittance ≈ 88%) at 365 nm; thus polymerization conditions would be similar to those developed for making hydrogel disks using the BK-7 optical glass. The devices were inserted in the PE tubing filled with pregel solution (all ingredients used to obtain the pHEMA hydrogel). Coatings were obtained by first forming the hydrogel through a photopolymerization step and then briefly heating the assembly to expand the PE tubing to facilitate the removal of the coated device from the PE tubing (Figure 3). The devices were then placed in DI water to remove unreacted reagents. This strategy provided coatings of a more reproducible, uniform thickness with a means to control coating

180


Jhaveri et al.

Figure 5. Release profile of NGF from pHEMA (2), pHEMA-lysine (O), and pHEMA-NaCl (9) for the first 240 min. * represents the values at which the release of NGF from pHEMA-lysine hydrogel is significantly different from the release from the pHEMA-NaCl hydrogel. The data were fitted so that there was no inflection at the different time points in the release profile. (n ) 3, *p < 0.05).

Figure 6. Representative images of DRG neurons from cultures with targeted NGF doses of 10 ng/mL. (a) No NGF (negative control); (b) bath-applied NGF (10 ng/mL); NGF released from (c) pHEMA hydrogel (loading concentration 6 µg/mL) and (d) pHEMA-NaCl hydrogel (loading concentration 2 µg/mL); (e) pHEMA-Lys hydrogel (loading concentration 20 µg/mL); and (f) results of automatic tracing obtained from neuron in image e. The green lines in the image are tracings of the neuronal process from which the length of the neuronal process can be calculated. Scale bar ) 400 µm.

thickness simply by varying the i.d. of the PE tubing. The tip of a hydrogel-coated device is shown in Figure 4b. For many insertion experiments, fluorescently labeled hydrogel was used to allow clear imaging after insertion (Figure 4c). When devices were coated using 280 µm i.d. PE tubing, the final o.d. of the coated devices was ∼315 µm thick because of swelling after equilibration in PBS. These pHEMA coatings were uniform (calculated from the midline of the device) along the entire shank (Figure 4d). Extended Release of Nerve Growth Factor from pHEMALysine Hydrogel-Coated Neural Prosthetic Devices. A CBQCA protein quantitation kit was used to determine the released quantity of NGF from hydrogel-coated microfabricated devices. This assay has been reported to be fairly accurate in determining protein quantities of

Release of Nerve Growth Factor from HEMA Hydrogel-Coated

Recommend Documents