Surface characterization and in vitro biocompatibility assessment of photosensitive polyimide films

Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Surface characterization and in vitro biocompatibility assessment of photosensitive polyimide films Sami Myllymaa a,∗ , Katja Myllymaa a,b , Hannu Korhonen a , Mikko J. Lammi c,d , Virpi Tiitu e , Reijo Lappalainen a a

Department of Physics, University of Kuopio, P.O. Box 1627, FI-70211 Kuopio, Finland Microsensor Laboratory, School of Engineering and Technology, Savonia University of Applied Sciences, P.O. Box 6, FI-70201 Kuopio, Finland Department of Applied Biotechnology, University of Kuopio, P.O. Box 1627, FIN-70211 Kuopio, Finland d Biocenter Kuopio, University of Kuopio, P.O. Box 1627, FI-70211 Kuopio, Finland e Department of Anatomy, University of Kuopio, P.O. Box 1627, FIN-70211 Kuopio, Finland b c

a r t i c l e

i n f o

Article history: Received 23 July 2009 Received in revised form 23 November 2009 Accepted 15 December 2009 Available online 22 December 2009 Keywords: Photosensitive polyimide Biocompatibility Fibroblast Surface characterization Zeta potential

a b s t r a c t Polyimide (PI) is a commonly used polymer in microelectronics. Recently, numerous PI-based flexible neural interfaces have been developed for reducing mechanical mismatch between rigid implant and soft neural tissue. Most approaches employ non-photosensitive PI, which has been proven earlier to be biocompatible. However, photosensitive polyimide (PSPI) would simplify device fabrication remarkably, but its biocompatibility has been only sparsely reported. In this study, cytotoxicity of spin-coated PSPI (HD Microsystems PI-2771) and conventional PI (HD Microsystems PI-2525) films were evaluated in vitro using BHK-21 fibroblasts according to the ISO-10993-5 standard. PSPIs were tested as cured at a temperature of 200 ◦ C (PI-2771-200) and 350 ◦ C (PI-2771-350). The PI film surfaces were characterized in terms of their roughness, energy and zeta potential which are hypothesized to affect cell–material interactions. The values of the total surface free energy (SFE), and its polar and dispersive component, were significantly (p < 0.001) greater for the PI-2525 film (SFE: 47.3 mJ/m2 ) than for the PI-2771-200 (25.6 mJ/m2 ) or PI-2771-350 films (26.2 mJ/m2 ). The curing temperature of the PI-2771 had a significant effect on the zeta potential values (p < 0.001), but not on surface energy (p = 0.091) or roughness (p = 0.717). The results from the MTS proliferation assays and live/dead staining revealed that PSPI is almost as non-cytotoxic as conventional PI and polyethylene (negative control). The morphology and spreading of BHK-21 cells were similar on all the PI materials tested. In conclusion, PSPI seems to be a promising biocompatible material, while further studies in vitro and in vivo are needed to clarify the long-term effects. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Polyimide (PI) is a high-performance polymer frequently used in the microelectronics industry as an encapsulation and stress buffer material on semiconductor chips and also as a substrate material in flexible printed circuit boards/cables [1]. PI possesses numerous desirable properties, such as excellent thermal stability and resistance to solvents, strong adhesion to metals and metal oxides and good dielectric properties [1]. Recently, PI has been investigated as a substrate/encapsulation material in numerous bio micro-electromechanical-system (bio-MEMS) applications, e.g., neural interfaces [2–10] due to its appropriate mechanical properties and biocompatibility [9,11]. It has been considered that the flexibility of PI (Young’s modulus ∼ 3 GPa) may decrease the mechanical mismatch

∗ Corresponding author. Tel.: +358 40 5572499; fax: +358 17 162585. E-mail address: sami.myllymaa@uku.fi (S. Myllymaa). 0927-7765/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2009.12.011

between rigid implant and soft neural tissues (∼3 kPa) and subsequent adverse tissue responses due to micromovement of the implant with respect to tissues, e.g., tissue damaging, inflammation reaction and scar formation [12] in comparison to the most frequently used neural implant material, silicon (∼170 GPa). PI-based neural interfaces have been developed, e.g., for intracortical multiunit neural activity recordings [2–4], cortical surface field potential recordings [5–8], and interfacing with regenerating peripheral nerves [9,10]. There are two common types of PIs for microelectronics on the market: traditional non-photosensitive PI and novel photosensitive polyimide (PSPI). Due to the incorporated photoreactive side groups in PSPI, it can be patterned directly by UV light and developer chemicals. Upon UV exposure, the photoreactive end groups undergo a free radical polymerization to form a cross-linked PI precursor. When covered with a photomask, a solubility difference between the exposed and unexposed regions results, and depending on the polarity of the PSPI used, the exposed or the unexposed

506

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

areas can then be removed by dissolving them in a developer solvent. Hence, PSPI simplifies remarkably the device fabrication because no photoresist and/or hard mask as an etch template is needed, as would be the case with standard non-photosensitive PIs. In vitro cytotoxicity tests play an important role in evaluating the biocompatibility of novel implant materials. The international standard ISO-10993-5 (biological evaluation of medical devices: tests for in vitro cytotoxicity) defines the test methods to be used for the determination of cytotoxicity in vitro. The mammalian cell line is selected according to the intended application. Cells are seeded in contact with a tested material or with an extract from a test material, and evaluations are made of the different parameters, such as cell adherence, adhesion, proliferation, morphological changes and metabolism changes. The non-cytotoxicity of PI has been demonstrated in many in vitro [9,11,13] and in vivo studies [14,15]. Stieglitz et al. [9] studied the cytotoxicity of three commercial PI grades (Pyralin PI-2611, PI2556, PI- 2566, HD Microsystems GmbH, Bad Homburg, Germany) using L-929 (mouse fibroblast) and Neuro-2A (mouse neuroblastoma) cell lines, and they reported excellent biocompatibility for PI-2611 and PI-2556 and good results for PI-2566. This last PI, since it is fluorinated, differs from the others with respect to its chemical structure. Furthermore, Lee et al. reported that fibroblasts attached, spread out and grew on the PI surfaces in a manner corresponding to the behaviour of control cells growing on the surface of polystyrene [13]. However, the previously tested PI grades are mainly nonphotosensitive ones. To the best of our knowledge, only one recent paper [16] focuses on assessing the cytotoxicity of a PSPI grade involving different chemistry. Sun et al. [16] concluded that the PSPI (Fujifilm Durimide 7020) was also noncytotoxic and that the adhesion, morphology and spreading of L-929 cells was even enhanced on PSPI compared with their activity on non-photosensitive grade (PI-2611). In our previous work, flexible microelectrode arrays that employed PSPI (PI-2771, HD Microsystems) as an encapsulating layer were developed and tested by evoked potential recordings on the surface of rat cortex [8]. For technical reasons, the curing of the PSPI was performed at a lower temperature than that recommended by the manufacturer. The array was capable of capturing a biologically meaningful signal for approximately 2 weeks, after which the responses decayed, presumably as a result of the dura mater thickening and a non-conductive scar forming around the electrode sites. The purpose of the present study was, then, to gain definite proof of the non-cytotoxicity of PSPI (PI-2771) and of its suitability for implantable biomedical use. Since the surface properties of biomaterials play an important role in cell-biomaterial interactions, the PI surfaces produced were characterized in terms of their surface roughness, energy and zeta potential. It was hypothesized that PSPI is as non-cytotoxic as traditional PI (PI-2525) and that the curing of PSPI at a lower temperature may provoke more cytotoxic effects, and hence it does not constitute an acceptable protocol in device fabrication.

2. Materials and methods 2.1. Sample fabrication Non-photosensitive PI (PI-2525) and photosensitive PI (PI2771) were purchased from HD Microsystems and used as supplied. PI-2771 is based mainly on the same backbone chemistry, i.e., polyamic acid of benzophenone tetracarboxylic dianhydride/oxydianiline/m-phenylene diamine, as PI-2525, but also contains pendant photoreactive side groups. N-methyl2-pyrrolidone (NMP) is used as a solvent in both products.

High-density polyethylene and natural latex rubber (Small Parts Inc., Miramar, FL, USA) were used as negative and positive control materials, respectively, in the cell experiments. The PI-2525 and PI-2771 films were spin-coated (5000 rpm, 60 s) under low acceleration onto 2-in., p-type 1 0 0 silicon (Si) wafers (Si-Mat Landsberg am Lech, Germany) preliminary baked for 20 min at 200 ◦ C to remove the moisture, and coated with VM-651 (HD Microsystems) to improve the adhesion between the polyimide and the Si. The PI-2771 films were then pre-baked at 100 ◦ C for 90 s on a hot plate, exposed to 365 nm UV light (Karl Suss MA45, Suss Microtec Inc., Waterbury Center, VT, USA), developed in cyclopentanone (Sigma–Aldrich Inc.) and rinsed with propylene glycol methyl ether acetate (Sigma–Aldrich Inc.). A portion of the PI-2771 coated wafers (4 of the total of 8 pieces) and all of the PI2525 coated wafers (4 pieces) were polymerized by curing them in an oven where the temperature was gradually raised to 350 ◦ C over a period of 60 min and then maintained for 60 min at that temperature. Another portion of the PI-2771 coated wafers was baked at a lower temperature (200 ◦ C). Thus, three different material groups were prepared and tested: PI-2525, PI-2771 baked at 200 ◦ C (PI-2771-200), and PI-2771 baked at 350 ◦ C (PI-2771-350). 2.2. Sample pretreatment Silicon wafers were cut into 7 mm × 7 mm pieces using a custom-made apparatus that included a diamond knife. After cutting, the pieces were sonicated for a few minutes in 7× detergent (OneMed Ltd., Vantaa, Finland) and rinsed repeatedly in ethanol and in sterile water. 2.3. Atomic force microscope (AFM) characterization of the surface The surface topography was analyzed using a PSIA XE-100 (Park Systems Corp., Suwon, Korea) atomic force microscopy at ambient temperature and humidity. AFM was performed in the non-contact mode, in which an aluminum-coated silicon cantilever (Acta-10, ST Instruments B.V., LE Grooth-Ammers, The Netherlands) was used to probe the surface across an area of 2 ␮m × 2 ␮m. The average surface roughness (Ra ) and peak-to-valley roughness (Rpv ) values were determined on six random areas per sample using instrument analysis software (XIA). 2.4. Contact angle and surface energy measurements The contact angles were determined using the sessile drop method at 22 ◦ C and 45% relative humidity with the aid of a custom-made device that included an optical microscope with a digital camera. To access the total surface free energy (SFE) and its polar/dispersive component, the contact angles for one polar (water) and one non-polar liquid (diiodomethane) were measured within 5 s of placing a drop (15 ␮l) on the surface of a sample. The drop image was stored and an image analysis software GIMP (www.gimp.org) was used to determine the left and right margin contact angles of five sessile drops in order to calculate the average contact angle. The dispersive SD and polar SP components were estimated using the Owens–Wendt model [17]: (1 + cos )L = 2((SD LD )

1/2

+ (SP LP )

1/2

)

(1)

where () is the measured (averaged) contact angle value, the superscript D labels the dispersive component, and P labels the polar component of the surface tension, while the subscripts S and L stand for solid and liquid, respectively.  L , LD and LP stand for the total SFE and its dispersive and polar components, respectively [18]. The total SFE ( S ) is the sum of its dispersive and polar components.

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

2.5. Zeta potential measurements Zeta potential measurements were performed in order to estimate the existing interface charges of PI films in a liquid. The zeta potential is generated when an electrolyte is circulated through the measuring cell and is forced under pressure to flow directly through a small gap formed by two sample surfaces. A comparative movement of the charges in the electrochemical double layer occurs and gives rise to the zeta potential. The streaming current measurements were performed using an electrokinetic analyzer (SurPASS, Anton Paar GmbH, Austria) equipped with an adjustable gap cell and two Ag/AgCl electrodes placed at the both sides of the sample. A gap of approximately 100 ␮m was adjusted between the samples. The measurements were performed using 0.001 M KCl as an electrolyte solution at a fixed pH of 7.4 ± 0.2. The zeta potential () was obtained from streaming current measurements according to the Helmholtz–Smoluchowski equation [19]: =

dI  L × × ε × ε0 A dp

(2)

where dI/dp is the slope of the streaming current vs. pressure,  is the electrolyte viscosity, ε0 is the vacuum permittivity, ε is the dielectric constant of the electrolyte, L is the length of the streaming channel, and A is the cross-section of the streaming channel. 2.6. Cell culture Baby hamster kidney fibroblasts (BHK 21, clone 13, HPA Culture Collections, Salisbury, United Kingdom) were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Euroclone, Pero, Italy) supplemented with 10% fetal calf serum (PAA, Linz, Austria), 50 ␮g/ml ascorbic acid (Sigma), 2 mM l-glutamine (PAA), 20 IU/ml penicillin (Euroclone), and 200 ␮g/ml streptomycin sulfate (Euroclone). The cells were seeded on the surface of the samples (7 mm × 7 mm) by adding a medium which contained 5.0 × 104 cells/ml in each sample cultured in 24-well cell culture plates (Greiner Bio-One, Frickenhausen, Germany). The cell cultures were incubated at 37 ◦ C gassed with 5% CO2 for 24 h. 2.7. MTS assay The MTS-based cell proliferation assay (CellTiter 96® Aqueous One Solution Reagent, Promega, USA) was used as a colorimetric method to assess the cytotoxic effects of different polyimide and control materials. This assay is based on the reduction of MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)2-(4-sulfophenyl)-2H-tetrazolium) in the presence of phenazine methosulfate (PMS) to a water-soluble formazan product in actively functioning mitochondria. The quantity of formazan formed was determined by measuring the absorbance at 490 nm, which is directly proportional to the number of viable cells. After a 24-h cell culture period, samples were transferred into unused wells and a fresh medium, after which the MTS reagent (200 ␮l) was added. After incubation for 1 h, the media were removed, the wells were washed with phosphate buffer solution and the absorbances were measured at 490 nm with an ELISA reader. The relative cell numbers were determined by normalizing the optical density values to the highest measured values. At least three samples of each material were used, and the testing was repeated twice. 2.8. Cell viability assay The viability of the fibroblasts cultivated in the samples was evaluated using a combination of 2 fluorescent probes. The specimens were incubated for 5 min in a solution containing a 60 ␮M concentration of cell-impermeable DNA-binding dye, propidium

507

Table 1 Average surface roughness (Ra ), peak-to-valley roughness (Rpv ) and contact angles values of tested polyimide surfaces. Material

PI-2525 PI-2771-200 PI-2771-350

Contact angle,  (◦ )

Roughness (nm) Ra

Rpv

 water

 diiodomethane

0.89 ± 0.13# 0.43 ± 0.15 0.39 ± 0.04

3.96 ± 0.98# 2.03 ± 0.53 1.73 ± 0.32

69.5 ± 1.8# 95.1 ± 1.3 93.2 ± 0.8

40.0 ± 1.3# 69.2 ± 1.3 69.0 ± 1.1

Values are the mean ± SD. # Significant difference (p < 0.001) compared to other PI materials.

iodide (Sigma), and a 10 ␮M concentration of cell-permeable fluorescein diacetate (Fluka, Buchs, Switzerland) in PBS. After having been washed with the PBS, the samples were viewed with a confocal scanner (PerkinElmer Life Sciences, Wallac-LSR, Oxford, UK) on a Nikon Eclipse TE300 microscope, using the wavelengths 488/10 nm (excitation) and 525/50 nm (emission) for the fluorescein, and 568/10 nm (excitation) and 607/45 nm (emission) for the propidium iodide. The density of dead (red-staining) cells (nuclei/mm2 ) and the surface area of live (green-staining) cells were counted using Image J software. Randomly taken images from at least eight parallel samples per material group were analyzed. The use of different units in dead cell (nuclei/mm2 ) and live cell (␮m2 ) estimations was necessary because only nuclei of dead cells are stained red whereas the whole cytoplasm of live cell is stained by the green color. Moreover, the numbers of live cells was technically impossible to estimate because of so high viable cell densities that they cannot any longer be separated from each others, and image analysis programs like Image J are unable to calculate them reliably as separate particles. Additionally, the mean surface area covered by the viable cells was determined for each sample group. 2.9. Scanning electron microscopic analysis of cultured cells After 24-h cultivation, the cells were fixed with 2.5% (w/v) glutaraldehyde (Sigma) in sodium cacodylate buffer (pH 7.4) and dehydrated with ethanol gradient and hexamethyldisilazane (Sigma). The BHK-21 cells on the samples were examined with a scanning electron microscope Philips XL30 ESEM-TMP (Fei Company, Eindhoven, Netherlands) at an accelerating voltage of 8–15 kV after being coated with a thin layer of gold using a Sputter Coater E 5100 (Polaron Equipment Ltd., England). 2.10. Statistical analysis The results are expressed as a mean ± standard deviation or as a mean ± standard error of the mean. One-way ANOVA (SPSS 16.0 software) variance analysis, followed by Tukey Post Hoc Tests, was used to determine the statistical significance of the observed differences between the materials. p < 0.05 was considered as significant. 3. Results 3.1. Surface characterization The average surface roughness (Ra ) and peak-to-valley roughness (Rpv ) values, determined by AFM, and also the contact angles, obtained by the sessile drop method on the different PI materials, are presented in Table 1. The roughness values obtained are all very small, i.e., all are below 4 nm. However, the values of Ra and Rpv for the PI2525 surface are significantly higher than their counterparts for the PI-2771-200 and PI-2771-350 (p < 0.001 for all). Somewhat surprisingly, the curing protocol chosen (200 ◦ C vs. 350 ◦ C) for the PI-2771 did not have a statistically significant effect on the Ra (p = 0.830) or on Rpv (p = 0.717).

508

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

Fig. 1. Total surface free energy (SFE) and its dispersive (DISP) and polar components for PI surfaces calculated using the Owens–Wendt model [17]. Error bars indicate standard deviations. # Significant difference (p < 0.001) from other PI materials.

The contact angles for water and for dioodomethane on both PI-2771 film types were significantly higher than for the PI-2525 film (p < 0.001 for all, Table 1). The total surface free energy and also the polar and dispersive component values, estimated by using the Owens–Wendt model, were all significantly lower for the PI-2771200 and PI-2771-350 films than for the PI-2525 films (p < 0.001 for all), demonstrating the more hydrophobic nature of the PI-2771 (Fig. 1). In contrast, the effect of the chosen curing temperature for the PI-2771 on SFE (p = 0.206), on the dispersive (p = 0.932) or the polar component (p = 0.091) was not statistically significant (Fig. 1). The zeta potential values measured at a fixed pH of 7.4 for the PI-2525, PI-2771-200 and PI-2771-350 were −59.9 ± 0.3 mV, −66.2 ± 1.6 mV and −54.2 ± 0.1 mV (mean + SD), respectively. All of the surfaces were negatively charged, and the differences between the surfaces were statistically significant (p < 0.001 for all). 3.2. Cell number analysis (MTS) The MTS colorimetric assay used in this study relates to the number of viable cells attached on the surfaces 24 h after the cellseeding (Fig. 2). The results were normalized relative to the highest value, i.e., the value for the PE control. The relative cell numbers in the different PI groups were at a level of 62–70% of the PE controls, with no significant differences between them. 3.3. Live-dead analysis All of the images of the live/dead-stained cells in the different PI groups appeared to be very similar and hence, of these, only PI-2525 is presented here (Fig. 3A). All of the PI types tested were highly cytocompatible, in contrast to latex rubber

Fig. 2. The relative number of BHK-21 cells at 24 h from seeding on polyethylene (PE, negative control), latex rubber (positive control), conventional polyimide (PI-2525) and two differently cured photosensitive PI (PI-2771-200, PI-2771-350) surfaces. The data was obtained by MTS assay and normalized to the highest optical density value. The error bars indicate standard deviations.

(Fig. 3B), in which viable cells were scarcely found and the dead cells were most probably detached and floating in the medium. Quantitative analysis of the proportions of the live and dead cells produced results that were consistent with the MTS assessment. The surface area of live cells (mean ± SEM) was greatest on the PI-2525 (71,900 ± 8500 ␮m2 , n = 13), followed by the PE (70,400 ± 2900 ␮m2 , n = 10), PI-2771-200 (67,300 ± 13,200 ␮m2 , n = 8), PI-2771-350 (63,000 ± 5800 ␮m2 , n = 8), and latex rubber (1400 ± 320 ␮m2 , n = 8). The only statistically significant difference (p < 0.001) was that between latex rubber and all of the other sample groups. The dead cell densities (mean ± SEM) on PI-2525, PE, PI-2771-200, latex rubber and PI-2771-350 were all in the same range with respective values of 13.5 ± 5.5, 13.9 ± 5.4, 18.8 ± 8.2, 24.9 ± 8.6 and 30.8 ± 8.8 cells/mm2 , without any significant difference between the groups. 3.4. Surface area covered by the cells The surface area covered by the BHK-21 fibroblasts at 24 h was determined by confocal scanning microscope images of the samples with live/dead-stained cells. The rank order for the amount of surface area covered by cells was PI-2525, PE, PI-2771-200, PI-2771-350 and latex rubber (Fig. 4). There were no statistically significant differences between the polyimide and PE groups. Only latex rubber, as a cytotoxic material, differed significantly (p < 0.001) from the other materials.

Fig. 3. Fluorescence live/dead staining of BHK-21 cells cultured on the surfaces of (A) PI (PI-2525) and (B) latex rubber (positive control) after 24 h. Viable cells: green, dead cells: red. Scale bar = 200 ␮m. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

509

faces. The BHK-21 cells cultivated on each PI surface showed similar morphologies in term of their size and spreading after 24 h (Fig. 5). The cells exhibited a normal fibroblastic morphology, they adhered well with filamentous extensions (filopodia), and they had started to form clusters.

4. Discussion

Fig. 4. Surface area covered by BHK-21 cells at 24 h. Polyethylene (PE) and latex rubber were used as negative and positive control materials, respectively. The error bars indicate standard deviations.

3.5. Morphological observations Scanning electron microscopic imaging was used to monitor the cell morphology, attachment and spreading on the different PI sur-

Substrate/encapsulation material underlies the vast majority of the tissue contact in the neural electrodes and prostheses and thus has to fulfill high requirements in order to permit the stable, long-term functioning of implants. An optimal insulating material should be biocompatible and biostable and have good dielectric properties. Polyimide possesses several desirable characteristics, such as mechanical flexibility, chemical stability, strong adhesion to metals and proper dielectric properties [1]. A special feature of PSPI – photopatternability – simplifies the electrode fabrication process in comparison to conventional PI by eliminating the need for complex multilevel processes (lithography, etching) to define the outer shape of the neural implants and to expose the electrode sites through an encapsulation layer [2,8,20]. These features are beneficial for implant manufacturing, permitting the process flow at lower cost and with an improved yield. Due to the differ-

Fig. 5. SEM images of BHK-21 cells after 24 h culture period grown on PI-2525 (A, D and G), PI-2771-200 (B, E and H) and PI-2771-350 (C, F and I) samples at three different magnifications.

510

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511

ent chemistry and imidization kinetics of its precursors, a degree of uncertainty is related to the biocompatibility of PSPI. Conventional PI has been proven to be biocompatible in many previous studies [9,11,13–15]. In the present study, the non-cytotoxicity of PSPI (PI-2771, HD Microsystems GmbH) was confirmed by using in vitro tests according to the guidelines stated in the ISO-10993-5 standard. The PSPI films did not adversely affect the BHK-21 fibroblast cell function in vitro, and hence it seems to be a promising candidate material for implantable biomedical use, although the possible long-term effects in living tissue environment also need to be clarified. It is well known that not only the implant material chemistry but also surface properties such as roughness/topography [21,22], energy [23,24] and charge [25,26] have an effect on the protein adsorption and cellular responses (e.g., cell attachment, adhesion, spreading, proliferation and differentiation) of biomaterial. Fibroblasts prefer smooth surfaces [27–29]. To minimize the roughness differences between the samples, we produced all the PI coatings using the same spin-coating program with a maximum rotation speed as high as 5000 rpm. The roughness values measured (Table 1) corresponded to the typical values for spin-coated PI films [30,31]. The slightly flattened surface morphology of PI-2771 may be due to the lower viscosity of PI-2771 (20 ± 5 Poise [32]) in comparison to PI-2525 (60 ± 10 Poise [33]). However, because all surfaces were extremely smooth with Ra values of less than 1 nm, the differences observed in cell response between PI materials in the present study may not be explained by differences in sample roughness values. Increased wettability with higher surface energy has been shown mainly to enhance the adhesion and spreading of fibroblasts [34,35]. In contrast, Lee et al. [36] reported that fibroblasts adhere, spread, and grow better on moderate hydrophilic (i.e. with a water contact angle of about 55◦ ) surfaces than on more hydrophobic or hydrophilic surfaces. The contact angles (water and dioodomethane) on PI-2525 were approximately 25–30◦ lower than on PI-2771 films, i.e. more beneficial for cell attachment. In general, the chemical and physical factors have an effect on the contact angle [30]. The physical factors are related to changes in the surface morphology, such as surface roughness. However, the differences in the surface roughness parameters on the different PI films revealed in this study were so small that we could conclude that the chemical factor mainly contributed to changes in the contact angle values. PI-2525 showed more hydrophilic behaviour than PI-2771 films, which may explain why the PI-2525 showed slightly enhanced cell adhesion and proliferation properties in comparison to the more hydrophobic PI-2771 films. Evidently, the PI-2771 contact angle could be reduced by using a plasma treatment or proper surface coating, as we have shown earlier for polypropylene [24]. The electric charge on a biomaterial surface is considered to be one of the main factors affecting cell–biomaterial interactions [25,26]. When a solid surface is in contact with a liquid, surface charges accumulate and an electrochemical double layer is established. In the vicinity of a solid–liquid interface the charge carriers are fixed (the so-called immobile Stern layer), whereas they are mobile in the liquid phase (mobile layer) at a greater distance. The potential at the interface between the immobile and mobile layers is known as the zeta potential, which can be measured either by electrophoresis or by streaming current/streaming potential methods. We used the streaming current method, which permits calculation of the correct zeta potential and takes into account all of the conductivity effects, such as the surface conductance, that contribute to the zeta potential. The means of the zeta potential values measured were clearly negative for all of the PI films at a pH of 7.4. The lower the zeta potential at a fixed pH, the more effectively the surface attracts positively charged particles and repels negatively charged entities such as anions or charged proteins. The

PI-2771-200 had the lowest value (−66.2 mV), while the PI-2771350 was the least negatively charged (−54.2 mV). Interestingly, the curing temperature selected had a remarkable influence on the zeta potential but not on the surface contact angle values. Hence, investigations of the relationship between the zeta potential of a biomaterial surface and the protein adsorption/cellular response would be useful to enhance understanding of the mechanisms of the biological integration of materials with tissues, and also for further development of implants and biomedical devices. Unfortunately, when implanted in the central nervous system, all artificial materials have an inconvenient tendency to induce glial scar tissue formation, which can cause serious impairment in implant performance due to enhanced cell proliferation and migration capacity of the activated reactive astrocytes, a decreased local density of neurons, and the formation of an encapsulation layer that increases the electrode impedance and lowers the signal amplitudes [12]. Undoubtedly, the tissue response of implanted material depends on the material chemistry and its surface properties, which in turn are dependent on the process parameters. According to its manufacturer, the recommended final cure temperature of PI-2771 is 350 ◦ C for 1 h. In our previous microelectrode array studies [8,37], we used lower curing temperatures (200–300 ◦ C) to avoid the mixing of titanium and platinum metallization layers in the device and subsequent oxidation of the electrode surface and a significant rise in the electrode impedance. In our view, the solvents and photoinitiators that remained in the final device were the result of an excessively low cure temperature and they might cause adverse growth of fibrous tissue [8]. Sun et al. [16] reported that fibroblasts cultured on mildly cured (200 ◦ C) PSPI showed poor adhesion, limited spreading and round morphology in contrast to cells cultured on a fully cured (400 ◦ C) PSPI attach, while they were spread very well on the surface with numerous filopodia. In the present study, this kind of phenomenon was not observed. The cells were similarly attached and spread out with several filopodia, regardless of the curing protocol or PI grade. However, possible adverse effects of low curing temperature may arise in long-term exposure in tissues. Although in vitro tests were used that are well-standardised, sensitive, relatively cheap and rapid to perform, and correlate well with short-term implant performance, they do not provide information on the systemic response when implanted inside the body. Hence, further investigations are needed to clarify the in vivo response of PSPI and to ensure that there will be no adverse, longterm effects typical of polymers and that the material is suitable for chronic implantable use. Furthermore, PSPI surface characteristics could be tailored and optimized for example by using modern plasma and deposition techniques. Acknowledgements This study was supported by the National Graduate School of Musculoskeletal Diseases and Biomaterials, the Joint Research Program in Materials Science of Kuopio and Joensuu Universities (project no. 5733), the Academy of Finland (project no. 128117), ˜ Pons and and the Ulla Tuominen Foundation. Montserrat Espanol Maria-Pau Ginebra from the Department of Materials Science and Metallurgy, Technical University of Catalonia, Barcelona, Spain, are gratefully acknowledged for acting as hosts to Katja and Sami Myllymaa during their visit to learn more about zeta potential measurements. Dr. Jukka Häyrinen (Dept. of Biochemistry, University of Kuopio) was acknowledged for his contribution in biochemical aspects. The staffs at the BioMater Centre and at the Microsensor Laboratory are also thanked for their technical assistance. References [1] M.K. Ghosh, K.L. Mittal, Polyimides: Fundamentals and Applications, CRC Press, Boca Raton, FL, USA, 1996.

S. Myllymaa et al. / Colloids and Surfaces B: Biointerfaces 76 (2010) 505–511 [2] P.J. Rousche, D.S. Pellinen, D.P. Pivin, J.C. Williams, R.J. Vetter, D.R. Kipke, IEEE Trans. Biomed. Eng. 48 (2001) 361. [3] K.C. Cheung, P. Renaud, H. Tanila, K. Djupsund, Biosens. Bioelectron. 22 (2007) 1783. [4] A. Mercanzini, K. Cheung, D.L. Buhl, M. Boers, A. Maillard, P. Colin, J.-C. Bensadoun, A. Bertsch, P. Renaud, Sens. Actuators A 143 (2008) 90. [5] A.L. Owens, T.J. Denison, H. Versnel, M. Rebbert, M. Peckerar, S.A. Shamma, J. Neurosci. Methods 58 (1995) 209. [6] B.A. Hollenberg, C.D. Richards, R. Richards, D.F. Bahr, D.M. Rector, J. Neurosci. Methods 153 (2006) 147. [7] H. Takahashi, T. Ejiri, M. Nakao, N. Nakamura, K. Kaga, T. Hervé, IEEE Trans. Biomed. Eng. 50 (2003) 510. [8] S. Myllymaa, K. Myllymaa, H. Korhonen, J. Töyräs, J.E. Jääskeläinen, K. Djupsund, H. Tanila, R. Lappalainen, Biosens. Bioelectron. 24 (2009) 3067. [9] T. Stieglitz, H. Beutel, M. Schuettler, J.-U. Meyer, Biomed. Microdev. 2 (2000) 283. [10] T. Stieglitz, Sens. Actuators A 90 (2001) 203. [11] R.R. Richardson, J.A. Miller, W.M. Reichert, Biomaterials 14 (1993) 627. [12] V.S. Polikov, P.A. Tresco, W.M. Reichert, J. Neurosci. Methods 148 (2005) 1. [13] K. Lee, A. Singh, J. He, S. Massia, B. Kim, G. Raupp, Sens. Actuators B 102 (2004) 67. [14] N. Lago, K. Yoshida, K.P. Koch, X. Navarro, IEEE Trans. Biomed. Eng. 54 (2007) 281. [15] T.G. Yuen, W.F. Agnew, L.A. Bullara, Biomaterials 8 (1987) 138. [16] Y. Sun, S.P. Lacour, R.A. Brooks, N. Rushton, J. Fawcett, R.E. Cameron, J. Biomed. Mater. Res. A 90 (2009) 648. [17] D.K. Owens, R.C. Wendt, J. Appl. Polym. Sci. 13 (1969) 1741. [18] A.L. Oliveira, P.B. Malafaya, R.L. Reis, Biomaterials 24 (2003) 2575. [19] J. Lyklema, Fundamentals of Interface and Colloid Science, vol. II. Solid–Liquid Interface, Academic Press, London, United Kingdom, 1995. [20] A.J. Spence, K.B. Neeves, D. Murphy, S. Sponberg, B.R. Land, R.R. Hoy, M.S. Isaacson, J. Neurosci. Methods 159 (2007) 116.

511

[21] R.G. Flemming, C.J. Murphy, G.A. Abrams, S.L. Goodman, P.F. Nealey, Biomaterials 20 (1999) 573. [22] A. Curtis, C. Wilkinson, Biomaterials 18 (1997) 1573. [23] N.J. Hallab, K.J. Bundy, K. O’Connor, R.L. Moses, J.J. Jacobs, Tissue Eng. 7 (2001) 55. [24] K. Myllymaa, S. Myllymaa, H. Korhonen, M.J. Lammi, H. Saarenpää, M. Suvanto, T. Pakkanen, V. Tiitu, R. Lappalainen, J. Mater. Sci. Mater. Med. 20 (2009) 2337. [25] M.T. Khorasani, S. Moemenbellah, H. Mirzadeh, B. Sadatnia, Colloid Surf. B: Biointerfaces 51 (2006) 112. [26] K. Cai, M. Frant, J. Bossert, G. Hildebrand, K. Liefeith, K.D. Jandt, Colloid Surf. B: Biointerfaces 50 (2006) 1. [27] T.P. Kunzler, T. Drobek, M. Schuler, N.D. Spencer, Biomaterials 28 (2007) 2175. [28] L. Ponsonnet, K. Reybier, N. Jaffrezic, V. Comte, C. Lagneau, M. Lissac, C. Martelet, Mater. Sci. Eng. C 23 (2003) 551. [29] C. Wirth, V. Comte, C. Lagneau, P. Exbrayat, M. Lissac, N. Jaffrezic-Renault, L. Ponsonnet, Mater. Sci. Eng. C 25 (2005) 51. [30] M. Kanno, H. Kawakami, S. Nagaoka, S. Kubota, J. Biomed. Mater. Res. 60 (2002) 53. [31] W. Liu, Y. Lan, Y. Chen, W. Lo, C. Lin, Surf. Coat. Technol. 201 (2007) 6530. [32] Technical information of PI2771 photosensitive polyimide, process guidelines, HD Microsystems GmbH, Bad Homburg, Germany. [33] Product information of Pyralin PI2525, PI2555, PI2556, HD Microsystems GmbH, Bad Homburg, Germany. [34] J. Wei, M. Yoshinari, S. Takemoto, M. Hattori, E. Kawada, B. Liu, Y. Oda, J. Biomed. Mater. Res. B: Appl. Biomater. 81 (2007) 66. [35] G. Altankov, F. Grinnell, T. Groth, J. Biomed. Mater. Res. 30 (1996) 385. [36] J.H. Lee, G. Khang, J.W. Lee, H.B. Lee, J. Colloid Interface Sci. 205 (1998) 323. [37] S. Myllymaa, K. Myllymaa, H. Korhonen, I. Gureviciene, K. Djupsund, H. Tanila, R. Lappalainen, Proc. 30th IEEE EMBS Annual Int. Conf., Vancouver, Canada, August 20–24, 2008, p. 3200.