Langmuir 2007, 23, 11693-11704
11693
Issues of Ligand Accessibility and Mobility in Initial Cell Attachment Dorota Thid,† Marta Bally,‡,§ Karin Holm,† Salvatore Chessari,§ Samuele Tosatti,§ Marcus Textor,§ and Julie Gold*,† Department of Applied Physics, Chalmers UniVersity of Technology, 412 96, Go¨teborg, Sweden, Laboratory of Biosensors and Bioelectronics, Institute for Biomedical Engineering, Department of Information Technology and Electrical Engineering, ETH Zurich, 8092 Zurich, Switzerland, and Laboratory for Surface Science and Technology, Department of Materials, ETH Zurich, 8093 Zurich, Switzerland ReceiVed April 20, 2007. In Final Form: August 13, 2007 The influence of lateral ligand mobility on cell attachment and receptor clustering has previously been explored for membrane-anchored molecules involved in cell-cell adhesion. In this study, we considered instead a cell binding motif from the extracellular matrix. Even though the lateral mobility of extracellular matrix ligands in membranes does not occur in vivo, we believe it is of interest for cell engineering in vitro. As is the case for cell-cell adhesion molecules, lateral mobility of extracellular matrix ligands could influence cell attachment and, subsequently, cell behavior in cell culture. In this paper, the accessibility and functionality of extracellular matrix ligands presented at surfaces were evaluated for the conditions of laterally mobile versus non-mobile ligands by studying ligand-antibody binding events and early cell attachment as a function of ligand concentration. We compare the initial attachment of rat-derived adult hippocampal progenitor (AHP) cells on laterally mobile, supported phospholipid bilayer membranes to non-mobile, poly-L-lysine-grafted-poly(ethylene glycol) (PLL-g-PEG) polymer films functionalized with a range of laminin-derived IKVAV-containing peptide densities. To this end, synthesis of a new PLL-g-PEG/PEG-IKVAV polymer is described. The characterization of available IKVAV peptides on both surface presentations schemes was explored by studying the mass uptake of anti-IKVAV antibodies using a combination of the surface-sensitive techniques quartz crystal microbalance with dissipation monitoring, surface plasmon resonance spectroscopy, and optical waveguide lightmode spectroscopy. IKVAV-containing peptides presented on laterally mobile, supported phospholipid bilayers and non-mobile PLL-g-PEG were recognized by the anti-IKVAV antibody in a dose-dependent manner, indicating that the amount of available IKVAV ligands increases proportionally with ligand density over the concentrations tested. Attachment of AHP cells to IKVAV-functionalized PLL-g-PEG and supported phospholipid bilayers followed a sigmoidal dependence on peptide concentration, with a critical concentration of approximately 3 pmol/cm2 IKVAV ligands required to support initial AHP cell attachment for both surface modifications. There appeared to be little influence of IKVAV peptide mobility on the initial attachment of AHP cells. Although the spread in the cell attachment data was larger for the PLL-g-PEG surface modification, this was reduced when observed after 24 h, indicating that the cells might need longer times to establish attachment strengths equivalent to those observed on peptide-functionalized supported lipid bilayers. The present study is a step toward understanding the influence of extracellular-matrix-derived ligand mobility on cell fate. Further analysis should focus on the systematic tuning of lateral ligand diffusion, as well as a comparison between the response of non-spreading cells (i.e., AHPs), versus spreading cells (i.e., fibroblasts).
Introduction Cell fate in vivo and in vitro is strongly influenced by the extracellular environment, where cues are provided to cells through their cell-surface receptors. In addition to mechanical stimuli, examples of crucial biomolecular signaling systems include paracrine and autocrine signaling molecules (such as growth factors) and cell adhesion molecules, including both cellcell and cell-extracellular matrix adhesions. In order to study such systems in more detail and under controlled environments, biomolecular cues are often immobilized on surfaces in vitro with controlled coupling chemistry, orientation, and density, and are thereby presented to cells from the underlying substrate.1-20 An interesting surface modification approach is the use of * Corresponding author. Address: Dept. of Applied Physics, Chalmers University of Technology, 412 96 Go¨teborg, Sweden. Tel: +46 31 772 1000. Fax: +46 31 772 3134. E-mail:
[email protected]. † Chalmers University of Technology. ‡ Department of Information Technology and Electrical Engineering, ETH Zurich. § Department of Materials, ETH Zurich. (1) Arnold, M.; Cavalcanti-Adam, E. A.; Glass, R.; Blummel, J.; Eck, W.; Kantlehner, M.; Kessler, H.; Spatz, J. P. ChemPhysChem 2004, 5, 383. (2) Biesalski, M. A.; Knaebel, A.; Tu, R.; Tirrell, M. Biomaterials 2006, 27, 1259.
supported phospholipid bilayers, which are unique in the sense of conserved lateral mobility of both the lipids and ligands in fluidic phase systems. Diffusion coefficients of lipid molecules in plain and functionalized supported lipid membranes in liquid phase have been measured in the range of ∼1-10 µm2/s.4,12,17 This is approximately 10 times faster compared to the movement of lipids carrying ligands4 and roughly 100 times faster than the lateral movement of receptors imbedded in a cell membrane.20 However, the actual diffusion constant for a given ligand or receptor will ultimately be a function of the size, charge, and hydrophobicity of the molecules. Using supported membranes functionalized with growth factors12 or cell-cell adhesion (3) Cavalcanti-Adam, E. A.; Micoulet, A.; Blummel, J.; Auernheimer, J.; Kessler, H.; Spatz, J. P. Eur. J. Cell Biol. 2006, 85, 219. (4) Chan, P. Y.; Lawrence, M. B.; Dustin, M. L.; Ferguson, L. M.; Golan, D. E.; Springer, T. A. J. Cell Biol. 1991, 115, 245. (5) Dori, Y.; Bianco-Peled, H.; Satija, S. K.; Fields, G. B.; McCarthy, J. B.; Tirrell, M. J. Biomed. Mater. Res., Part A 2000, 50, 75. (6) Dustin, M. L.; Ferguson, L. M.; Chan, P. Y.; Springer, T. A.; Golan, D. E. J. Cell Biol. 1996, 132, 465. (7) Houseman, B. T.; Gawalt, E. S.; Mrksich, M. Langmuir 2003, 19, 1522. (8) Jensen, T. W.; Hu, B. H.; Delatore, S. M.; Garcia, A. S.; Messersmith, P. B.; Miller, W. M. J. Am. Chem. Soc. 2004, 126, 15223. (9) Kato, M.; Mrksich, M. Biochemistry 2004, 43, 2699. (10) Marchi-Artzner, V.; Lorz, B.; Hellerer, U.; Kantlehner, M.; Kessler, H.; Sackmann, E. Chem.sEur. J. 2001, 7, 1095.
10.1021/la701159u CCC: $37.00 © 2007 American Chemical Society Published on Web 10/05/2007
11694 Langmuir, Vol. 23, No. 23, 2007
molecules,6,20,21 it was shown that ligands in supported membranes can be rearranged by the cell, resulting in cell-receptor clustering,12,20 strengthened cell attachment,4,22 and facilitated cell spreading12 compared to stationary ligands. Supported membranes have also been functionalized with extracellular matrix peptide sequences,2,5,8,10,13,14,16,17,23 where RGD has been the most commonly studied extracellular matrix peptide. In cell studies using membranes functionalized with such extracellular cell adhesion motifs, it has been shown that the orientation,8,13,14 conformation,8,14 accessibility,2,5,23 and surface density2,8 of the ligands strongly influence cell behavior, such as attachment and spreading. However, the influence of the lateral mobility of the extracellular matrix ligands has not yet been addressed. IKVAV is a peptide sequence derived from the extracellular matrix protein laminin, and has been reported to promote the attachment, spreading, migration, and differentiation of neural cells.16,17,24-31 The IKVAV sequence has attracted more and more interest in recent years; however, not much is known about its receptor or any associated signal transduction pathways. It is believed to interact with an amyloid precursor protein in neural systems32 and with the prostaglandin E2 receptor in murine mammary adenocarcinoma cells.33 Studies addressing IKVAVreceptor interactions have shown that the interaction is not divalent cation dependent, as is the case for RGD-integrin receptor binding.25 Evidence to date points to the fact that the IKVAV receptor is not an integrin-type receptor. Even though the receptor and mechanism for the interaction is not explicitly known, ligand density and spatial distribution are most likely important factors in receptor mediated binding and signaling events. We have previously reported interesting behavior of adult hippocampal progenitor (AHP) cells grown on IKVAV peptidefunctionalized, surface supported, phospholipid bilayers (Figure 1A).17 Initial cell attachment was dependent on the concentration of IKVAV peptides presented on the bilayers, with a threshold (11) Mrksich, M. Chem. Soc. ReV. 2000, 29, 267. (12) Nam, J. M.; Nair, P. M.; Neve, R. M.; Gray, J. W.; Groves, J. T. ChemBioChem 2006, 7, 436. (13) Ochsenhirt, S. E.; Kokkoli, E.; McCarthy, J. B.; Tirrell, M. Biomaterials 2006, 27, 3863. (14) Pakalns, T.; Haverstick, K. L.; Fields, G. B.; McCarthy, J. B.; Mooradian, D. L.; Tirrell, M. Biomaterials 1999, 20, 2265. (15) Schuler, M.; Owen, G. R.; Hamilton, D. W.; De Wilde, M.; Textor, M.; Brunette, D. M.; Tosatti, S. G. P. Biomaterials 2006, 27, 4003. (16) Svedhem, S.; Dahlborg, D.; Ekeroth, J.; Kelly, J.; Ho¨o¨k, F.; Gold, J. Langmuir 2003, 19, 6730. (17) Thid, D.; Holm, K.; Ekeroth, J.; Kassemo, B.; Gold, J. J. Biomed. Mater. Res., Part A [Online early access] DOI: 10.1002/jbm.a.31358. Published Online: July 23, 2007. http://www3.interscience.wiley.com/cgi-bin/fulltext/114294813/ HTMLSTART. (18) Tosatti, S.; Schwartz, Z.; Campbell, C.; Cochran, D. L.; VandeVondele, S.; Hubbell, J. A.; Denzer, A.; Simpson, J.; Wieland, M.; Lohmann, C. H.; Textor, M.; Boyan, B. D. J. Biomed. Mater. Res., Part A 2004, 68, 458. (19) VandeVondele, S.; Vo¨ro¨s, J.; Hubbell, J. A. Biotechnol. Bioeng. 2003, 82, 784. (20) Zhu, D. M.; Dustin, M. L.; Cairo, C. W.; Golan, D. E. Biophys. J. 2007, 92, 1022. (21) Perez, T. D.; Nelson, W. J.; Boxer, S. G.; Kam, L. Langmuir 2005, 21, 11963. (22) To¨zeren, A.; Sung, K. L. P.; Sung, L. A.; Dustin, M. L.; Chan, P. Y.; Springer, T. A.; Chien, S. J. Cell Biol. 1992, 116, 997. (23) Stroumpoulis, D.; Zhang, H.; Rubalcava, L.; Gliem, J.; Tirrell, M. Langmuir 2007, 23, 3849. (24) Heller, D. A.; Garga, V.; Kelleher, K. J.; Lee, T. C.; Mahbubani, S.; Sigworth, L. A.; Lee, T. R.; Rea, M. A. Biomaterials 2005, 26, 883. (25) Kasai, S.; Ohga, Y.; Mochizuki, M.; Nishi, N.; Kadoya, Y.; Nomizu, M. Biopolymers 2004, 76, 27. (26) Massia, S. P.; Holecko, M. M.; Ehteshami, G. R. J. Biomed. Mater. Res., Part A 2004, 68, 177. (27) Powell, S. K.; Kleinman, H. K. Int. J. Biochem. Cell Biol. 1997, 29, 401. (28) Powell, S. K.; Rao, J.; Rogue, E.; Nomizu, M.; Kuratomi, Y.; Yamada, Y.; Kleinman, H. K. J. Neurosci. Res. 2000, 61, 302. (29) Ranieri, J. P.; Bellamkonda, R.; Bekos, E. J.; Vargo, T. G.; Gardella, J. A.; Aebischer, P. J. Biomed. Mater. Res. 1995, 29, 779. (30) Silva, G. A.; Czeisler, C.; Niece, K. L.; Beniash, E.; Harrington, D. A.; Kessler, J. A.; Stupp, S. I. Science 2004, 303, 1352.
Thid et al.
Figure 1. Schematic illustrations of the two surface modifications used in the present study and their characterization by antibody binding. (A) From left: vesicles doped with 0-5% maleimidoterminated lipids were added in solution to a SiO2 surface, which resulted in the formation of a supported phospholipid bilayer, which was subsequently functionalized with IKVAV-containing peptides in situ. Peptide function and accessibility were tested by monitoring the amount of antibody binding. (B) From left: PLL-g-PEG/PEGIKVAV from solution adsorbed onto TiO2. Peptide function and availability was tested by monitoring the amount of antibody binding.
concentration of 1% being required for strong cell attachment. The amount of cells on IKVAV-functionalized supported bilayers was comparable to the standard polyornithine/laminin-coated reference substrates up to 8 days in culture; however, the growth morphology was drastically different. As expected, cells proliferated largely in a monolayer fashion on polyornithine/laminin coatings.34 In contrast, on functionalized bilayers, cells formed networks of three-dimensional colonies. One hypothesis put forward in that work is that this cellular behavior was triggered mainly by ligand mobility, as peptides presented on a supported membrane are laterally mobile but are considered stationary in an adsorbed protein layer.21 Through receptor-ligand binding, cells could conceivably recruit laterally mobile IKVAV ligands and deplete the surrounding bilayer of peptide. With time, newly derived cells, via cell division, would experience a phospholipid bilayer with few or no anchorage points and rather attach to cells already adherent at the surface. In this way, cell clustering could develop already at early time points. However, a comparison between supported phospholipid bilayers and polyornithine/ laminin coatings cannot only consider the ligand mobility factor, since supported bilayers exhibit only one ligand type, while adsorbed laminin presents a variety of adhesive peptide sequences,27,28 the distribution and orientation of which cannot be controlled. (31) Tashiro, K.; Sephel, G. C.; Weeks, B.; Sasaki, M.; Martin, G. R.; Kleinman, H. K.; Yamada, Y. J. Biol. Chem. 1989, 264, 16174. (32) Kibbey, M. C.; Jucker, M.; Weeks, B. S.; Neve, R. L.; Vannostrand, W. E.; Kleinman, H. K. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 10150. (33) Zhang, S. Z.; Fulton, A. M. J. Cell. Physiol. 1991, 149, 208. (34) Gage, F. H.; Coates, P. W.; Palmer, T. D.; Kuhn, H. G.; Fisher, L. J.; Suhonen, J. O.; Peterson, D. A.; Suhr, S. T.; Ray, J. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 11879.
Ligand Accessibility/Mobility in Cell Attachment
As an alternative to polyornithine/laminin-coated surfaces for laterally non-mobile ligands, a variation of the well-studied polyL-lysine-grafted-poly(ethylene glycol) (PLL-g-PEG)-based surface modification (Figure 1B) was explored in this study. As for the case for supported bilayers, PLL-g-PEG-coated surfaces are resistant to nonspecific protein adsorption35-39 and cell binding,19,40 and can be functionalized with peptides to support specific cell attachment15-19,41,42 without loss of protein resistance.16,17,39 In addition, ligand density can be varied in a controlled way for both PLL-g-PEG and lipid bilayer systems, and its influence on initial cell attachment has been previously observed.15,17 PLLg-PEG has previously been functionalized with the RGD sequence15,18,19,39,42 and successfully used to promote RGDmediated cell attachment.15,18,19 In the present study, we (a) derive and characterize a new PLL-g-PEG/PEG-IKVAV polymer as a means to present nonmobile IKVAV peptides in controlled concentrations to cells, and (b) investigate any influence of mobile versus non-mobile IKVAV peptides on early AHP cell attachment by using supported bilayers and PLL-g-PEG polymers modified with a range of IKVAV peptide densities. The characterization of accessible, as well as functional, IKVAV peptides on both surface modifications was explored by studying the binding of anti-IKVAV antibodies. Antibody binding was measured using a combination of the surface-sensitive techniques quartz crystal microbalance with dissipation monitoring (QCM-D), surface plasmon resonance spectroscopy (SPR), and optical waveguide lightmode spectroscopy (OWLS). While SPR and OWLS are optical methods that sense changes in refractive index at the surface of a gold film or waveguiding layer, respectively, QCM-D is an acoustical method that senses changes in the vibrational frequency and damping coefficient of an oscillating quartz crystal. Both optical and acoustical methods measure the uptake of molecular mass at the sensor surface; however, the QCM-D technique detects both molecular mass and the mass of associated or trapped water within the adlayer film.43-45 The detection of IKVAV peptides with anti-IKVAV antibodies indicates the availability of the ligands, offers valuable quantitative and qualitative characterization of the surface modifications, and provides a suitable model for ligand-cell receptor interactions. Materials and Methods Unless otherwise stated, chemicals were obtained from commercial sources and used without further purification. Water was always deionized (resistivity > 18 MΩ cm) and purified (Milli-Q plus, Millipore, France, or Milli-Q A10, Millipore, Switzerland). Phosphate buffered saline (PBS) (Sigma) was prepared from tablets (10 mM phosphate, 137 mM NaCl, 2.7 mM KCl, pH 7.4) and autoclaved when necessary. 10 mM N-(2-hydroxyethyl)piperazine-N′-ethane(35) Glasma¨star, K.; Larsson, C.; Ho¨o¨k, F.; Kasemo, B. J. Colloid Interface Sci. 2002, 246, 40. (36) Groves, J. T.; Mahal, L. K.; Bertozzi, C. R. Langmuir 2001, 17, 5129. (37) Kenausis, G. L.; Voros, J.; Elbert, D. L.; Huang, N. P.; Hofer, R.; RuizTaylor, L.; Textor, M.; Hubbell, J. A.; Spencer, N. D. J. Phys. Chem. B 2000, 104, 3298. (38) Pasche, S.; Textor, M.; Meagher, L.; Spencer, N. D.; Griesser, H. J. Langmuir 2005, 21, 6508. (39) Tosatti, S.; De Paul, S. M.; Askendal, A.; VandeVondele, S.; Hubbell, J. A.; Tengvall, P.; Textor, M. Biomaterials 2003, 24, 4949. (40) Andersson, A. S.; Glasma¨star, K.; Sutherland, D.; Lidberg, U.; Kasemo, B. J. Biomed. Mater. Res., Part A 2003, 64, 622. (41) Harris, L. G.; Tosatti, S.; Wieland, M.; Textor, M.; Richards, R. G. Biomaterials 2004, 25, 4135. (42) Maddikeri, R. R.; Tosatti, S.; Schuler, M.; Chessari, S.; Textor, M.; Richards, R. G.; Harris, L. J. Biomed. Mater. Res., Part A, in press. (43) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397. (44) Reimhult, E.; Larsson, C.; Kasemo, B.; Hook, F. Anal. Chem. 2004, 76, 7211. (45) Su, X. D.; Wu, Y. J.; Knoll, W. Biosens. Bioelectron. 2005, 21, 719.
Langmuir, Vol. 23, No. 23, 2007 11695 sulfonic acid (HEPES) buffer was prepared by dissolving 4-(2hydroxyethyl)piperazine-1-ethane-sulfonic acid (Fluka, Switzerland) with the addition of NaCl to 150 mM and a pH adjustment to 7.4 with NaOH. Buffers were filtered through a 0.2 µm pore filter prior to use. The synthetic peptide CSRARKQAASIKVAVSADR (Sigma, Sweden; C6171) was dissolved in citric acid buffer (5 mM, pH ) 4.0, 10 mM ethylenediaminetetraacetic acid (EDTA)) at 1 mg/mL under N2, frozen in working aliquots, and never repeatedly frozen and thawed. The rabbit anti-IKVAV antibody serum (no. 175, raised against CSRARKQAASIKVAVSADR) was a gift (Hynda K. Kleinman, NIDCR, NIH, Bethesda, MD).46 The anti-IKVAV serum was frozen in aliquots. Stock solutions of polyornithine (Sigma, Sweden, P3655) at 10 mg/mL in autoclaved water and laminin (Sigma, Sweden, L2020) at 1 mg/mL in sterile PBS were stored at -20 °C. 1-Palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC) (Avanti Polar Lipids, Alabaster, USA, 850457) was purchased. 1-palmitoyl-2-oleoyl-sn-glycero-N-[N’-(3-maleimidopropionyl)-8amino-3,6-dioxaoktanoyl]phosphoethanolamine (maleimido lipid) was previously synthesized.17 Lipids were stored at -20 °C as powders or dissolved in chloroform. Synthesis of PLL-g-PEG/PEG-IKVAV. The general architecture of the polymers used was based on a PLL backbone of approximately 120 L-lysine units (average value in view of polydispersity of the polymer), a PEG side chain of approximately 47 ethylene glycol units (PEG MW ∼ 2 kDa), and a grafting ratio g, expressed as the number of lysine monomers per PEG side chain, of 4.5. The functionalized PLL-g-PEG polymer was assembled using the following peptide sequence: N-acetyl-DGCRGYGIKVAVSADRamide. Amino acids were purchased from Novabiochem, Switzerland. The peptide synthesis was performed using standard Fmoc technique. The high-performance liquid chromatography-purified samples were analyzed and identified by electrospray ionization mass spectrometry. The synthesis of the PLL-g-PEG/PEG-IKVAV was performed as follows (Figure 2): The cysteine-containing peptide thiol was reacted with 1,8-diazabicyclo-[5.4.0]-undec-7-ene for 2 h in dried dimethyl sulfoxide, and subsequently converted into the corresponding thiolate. The reaction was confirmed by 1H NMR. After 2 h, 0.33 equiv (with respect to the peptide) of vinyl sulfone-PEG(3.4)-N-hydroxysuccinimidyl (NHS) were added, and the reaction mixture was stirred for 2 h under nitrogen. The reaction was specific toward the vinyl sulfone function (confirmed by 1H NMR), and the intermediate peptide-PEG-NHS was further reacted with PLL (MW 15-30 kDa from Sigma, USA) to obtain the PLL-g-PEG(3.4)-IKVAV adduct. This was subsequently backfilled with methoxy-PEGsuccinimidyl propionic acid (2 kDa) to achieve the desired grafting ratio of 4.5. For the resulting PLL-g-PEG/PEG-IKVAV polymer, 9.9% of the total PEG chains were PEG-IKVAV, as determined by 1H NMR. The polymer was purified by dialysis (MWCO ) 14 kDa in deionized water, 2 days), and the samples were subsequently lyophilized. PLL-g-PEG-Based Surface Modification. Non-functionalized PLL-g-PEG was purchased from SurfaceSolutions GmbH (Switzerland) and had a grafting ratio g, expressed as the number of lysine monomers per PEG side chain, of 3.3. Both the IKVAV-functionalized and non-functionalized PLL-g-PEG polymers were dissolved in HEPES buffer at 0.25 mg/mL and sterile filtered (Ministart, Sartorious, Germany). If not used immediately, the solutions were divided into aliquots and stored at -20 °C. Surfaces were coated by incubation with 0.25 mg/mL PLL-g-PEG or PLL-g-PEG/PEGIKVAV for at least 30 min and rinsed with HEPES buffer. The maximum IKVAV surface density was calculated from the polymer mass adsorbed on TiO2 as determined by OWLS, and the grafting ratios were determined by NMR. The peptide surface density was varied by using different ratios of PLL-g-PEG to PLL-g-PEG/PEGIKVAV. It was assumed that there is no preferential adsorption of one of the polymer species as observed elsewhere.47 Subsequent antibody recognition was performed by a 60 min incubation with (46) Chalazonitis, A.; Tennyson, V. M.; Kibbey, M. C.; Rothman, T. P.; Gershon, M. D. J. Neurobiol. 1997, 33, 118. (47) Huang, N. P.; Voros, J.; De Paul, S. M.; Textor, M.; Spencer, N. D. Langmuir 2002, 18, 220.
11696 Langmuir, Vol. 23, No. 23, 2007
Thid et al.
Figure 2. Synthesis scheme of PLL-g-PEG/PEG-IKVAV 1% anti-IKVAV antibody-containing serum.16 For control experiments, 100% fetal calf serum (Fisher Scientific, Sweden, 178146044) lacking anti-IKVAV antibodies was used. Lipid-Based Surface Modification. POPC was dissolved in chloroform to 100 mg/mL. For vesicle preparation, the POPC solution and, where applicable, maleimido-lipids (up to 5 mol %) were mixed to a total of 1-5 mg of lipid. A lipid film was formed on the wall of a glass flask by evaporation of the solvent, dried under N2, and hydrated in 1 mL PBS. The solution was extruded 11 times through a 100 nm filter and 11 times through a 30 nm filter (polycarbonate filter membranes, Avanti Polar Lipids, Alabaster, AL) resulting in vesicles with an average diameter of 80 nm (dynamic light scattering, BI-90 Particle Sizer, Brookhaven Instruments Corporation). Peptidefunctionalized supported phospholipid bilayers were created by surface-induced adsorption and fusion of vesicles doped with maleimido-terminated lipids, followed by exposure to cysteineterminated, IKVAV-containing peptides until stable response.16,17,43 Subsequent antibody recognition was done by a 60 min incubation with 1% anti-IKVAV antibody containing serum.16 PBS was used for all dilutions. For control experiments, 1% fetal calf serum lacking anti-IKVAV antibodies was used. QCM-D. All surface modifications were followed and quantified with the QCD-D technique, which provides real-time information on changes in mass (as shifts in resonance frequency, ∆f) and viscoelastic properties (as shifts in energy dissipation, ∆D) in surfaceadsorbed films at the surface of a quartz crystal sensor.48 For a given (48) Rodahl, M.; Ho¨o¨k, F.; Krozer, A.; Brzezinski, P.; Kasemo, B. ReV. Sci. Instrum. 1995, 66, 3924.
added mass, ∆m, there is a proportional change in f, which increases linearly with the overtone number, n ()1,3,...): ∆mSauerbrey ) -
CQCM ∆f n
(1)
where CQCM is the mass-sensitivity constant (17.7 ng/(cm2 Hz) for the 5 MHz crystals used in the present study). QCM-D senses immobilized macromolecules together with associated water, which is coupled as an additional mass via direct hydration, viscous drag, and/or entrapment in cavities in the adsorbed film. QCM-D sensors were 5 MHz, AT-cut quartz crystals (Q-Sense AB, Go¨teborg, Sweden) sputtered with 50 nm of SiO2 or 6 nm of TiO2 (reactive magnetron sputtering at PSI, Switzerland). Prior to use, they were rinsed with Milli-Q water, dried with N2, and exposed to UV/ozone for 30-60 min. Experiments were performed using a Q-Sense D300 system (Q-Sense AB, Go¨teborg, Sweden) in batch mode at 22 °C, with the liquid volume of the chamber at ∼80 µL and a sensed crystal area of ∼20 mm2. The final concentrations were peptides 10 µg/mL (in PBS), vesicles 200 µg/mL (in PBS), polyornithine 50 µg/mL (in water), laminin 5 µg/mL (in PBS), varying ratios of PLL-g-PEG and PLL-g-PEG/PEG-IKVAV 0.25 mg/mL (in HEPES), and 1% anti-IKVAV antibody serum. All QCM-D data were acquired at the third to seventh overtones. Frequency data are presented at the fundamental frequency (∆fn/n). Dissipation does not scale with the harmonic. Where applicable, the average ( SEM is reported. Modeling was performed with Q-tools 2 (Q-Sense AB, Go¨teborg, Sweden). A Voigt-based, one-layer model with a fixed density of
Ligand Accessibility/Mobility in Cell Attachment
Langmuir, Vol. 23, No. 23, 2007 11697
1000 kg/m3 was applied to two overtones, resulting in the best fit. Output thickness was used for conversion to modeled mass only for cases where elastic shear modulus and shear viscosity were in agreement with the literature.49,50 OWLS. PLL-g-PEG-based surface modifications were followed in situ by OWLS. The adsorbed mass could be calculated from the change in refractive index and thickness upon molecule adsorption from solution according to51 mOWLS ) tfilm
nfilm - nmedium dn/dc
(2)
where tfilm is the thickness of the adsorbed layer, nfilm and nmedium are the refractive indexes of the adlayer and medium, respectively, and dn/dc is the refractive index increment, which has previously been calibrated to 0.139 cm3/g for PLL-g-PEG polymers and 0.182 cm3/g for proteins.38,52 OWLS waveguides were purchased from Microvacuum (Hungary) and coated by reactive magnetron sputtering with a 6 nm TiO2 layer at PSI (Switzerland). Prior to use, the waveguides were cleaned by sonication in 0.1 M HCl (10 min), 2-propanol (5 min), and water (5 min), followed by extensive rinsing under water flow, blow-drying with N2, and oxygen-plasma cleaning (2 min) (Harrick plasma cleaner/sterilizer PDC-32G instrument (Ossining, NY)). Experiments were performed with OWLS 110 (MicroVacuum, Ltd., Budapest, Hungary) at 25 °C, in batch mode. SPR. Gold-coated SPR chips (Biacore AB, Uppsala, Sweden) were cleaned in Milli-Q water, ammonia, and hydrogen peroxide (5:1:1) for 10 min at 70 °C, rinsed with Milli-Q water, dried with N2, and covered with 1 nm of Ti and 20 nm of SiO2 by electron beam evaporation (AVAC HVC600). Coated chips were cleaned and handled the same way as QCM-D sensors. SPR experiments were performed using a Biacore 2000 instrument (Biacore AB, Uppsala, Sweden) at 22 °C. Vesicles (10 min, 10 µL/min), peptides (30 min, 10 µL/min), and antibodies (60 min, 5 µL/min) were added at the same concentrations as in QCM-D with PBS as the running buffer. However, to facilitate bilayer formation, vesicles were added in PBS containing 2 mM MgCl2, with a subsequent rinse with 5 mM EDTA in PBS. This treatment did not affect peptide or antibody binding behavior, as confirmed with QCM-D (data not shown). Four measurements were performed simultaneously: three on maleimidodoped bilayers and one on a plain POPC bilayer, where the measurement on the POPC bilayer was used as a reference. The change in relative units (∆RU) for peptide and antibody binding obtained from SPR were recalculated to mass (∆mSPR in ng/cm2) as described previously.50,53 In short, ∆mSPR )
CSPR ∆RU CSPR ∆RU ) β e-2d/δ
(3)
where the proportionality constant, CSPR, has been calibrated to 0.066 ng/cm2 for protein adsorption on flat gold surfaces.54 β is an estimated correction factor for lower sensitivity on the SPR sensor due to the SiO2 with a thickness d of 20 nm evaporated on top of the gold film.53 Further, δ is the decay length,55 which, for the present experimental setup, is ∼230 nm in SiO2. Consequently, β ∼ 0.84. Anti-IKVAV Antibody Concentration. Quantification of antiIKVAV antibody binding to IKVAV ligands with SPR (bilayers), OWLS (PLL-g-PEG), and QCM-D (bilayers and PLL-g-PEG) was performed using two antibody concentrations. The same concentration was used in QCM-D and OWLS, and the concentration was ∼2.3 times lower in SPR; QCM-D data are presented in Figure 6. The (49) Heuberger, M.; Drobek, T.; Vo¨ro¨s, J. Langmuir 2004, 20, 9445. (50) Larsson, C.; Rodahl, M.; Ho¨o¨k, F. Anal. Chem. 2003, 75, 5080. (51) Defeijter, J. A.; Benjamins, J.; Veer, F. A. Biopolymers 1978, 17, 1759. (52) Vo¨ro¨s, J.; Ramsden, J. J.; Csucs, G.; Szendro, I.; De Paul, S. M.; Textor, M.; Spencer, N. D. Biomaterials 2002, 23, 3699. (53) Jung, L. S.; Campbell, C. T.; Chinowsky, T. M.; Mar, M. N.; Yee, S. S. Langmuir 1998, 14, 5636. (54) Lo¨fås, S.; Malmqvist, M.; Ronnberg, I.; Stenberg, E.; Liedberg, B.; Lundstro¨m, I. Sens. Actuators, B: Chem. 1991, 5, 79. (55) Liedberg, B.; Lundstro¨m, I.; Stenberg, E. Sens. Actuators, B: Chem. 1993, 11, 63.
absolute values of antibody concentrations in the serum solution were not known. However, the ratio of ∼2.3 could be deducted from QCM-D measurements, where initial slopes (1 min) for antibody binding to IKVAV-functionalized bilayers were compared for the two antibody concentrations. The initial slope of analyte binding is linearly dependent on its concentration in solution, as recently shown with QCM-D.56 Cell Culture. Silicon wafers with a thermally grown 0.4 µm silicon oxide cut into 8 × 10 mm2 pieces with a diamond saw (referred to as SiO2), were used as substrates for supported bilayers. The surfaces were cleaned in accordance with the QCM-D protocol and placed in 8-well chamber slides (Nunc International, Naperville, IL, 154534). Bilayer-based surface modification was performed in a total volume of 250-300 µL using the same concentrations and incubation times as in the QCM-D measurements. After peptide incubation, PBS in the wells was exchanged with Dulbecco’s modified Eagle’s medium (DMEM)/F12. Care was taken not to dewet the surfaces at any time after bilayer formation. TiO2 was used for the PLL-g-PEG-based surface modifications. Substrates were diced into 10 × 10 mm2 pieces, sonicated twice for 10 min in toluene, and blow-dried with N2 to remove glue residues from the backside. Just before the surface modification, the samples were sonicated in 2-propanol for 15 min, blow-dried with N2, and placed in an oxygen-plasma cleaner (3 min in a Harrick plasma cleaner/ sterilizer PDC-32G, Ossining, NY, or 5 min at 250 W in a TePla 3PC MW plasma ship, Assla, Germany). Surface modification with polymers was done in accordance with the QCM-D protocol; however, for rinsing, samples were soaked for 3 min in HEPES and water, rinsed under water flow, and blow-dried with N2. Coated samples were placed in 24-well plates with a well area of 1.9 cm2 (Nunc International). The polyornithine/laminin surface modification was made in advance by 12-24 h adsorption onto sterile (10 min in 70% ethanol,) and clean (60 min in UV/ozone) SiO2 wafers of polyornithine followed by a water rinse, a 12-24 h adsorption of laminin, and a PBS rinse. Coated samples were kept in -20 °C in PBS and thawed just prior to cell experiments. All solutions, except for peptide stock, were sterile. Isolation of rat AHP cells has previously been described.57 For proliferation, cells were grown in polyornithine-coated T75 flasks in DMEM/F12 (1:1) (Invitrogen, Sweden, 21331020) supplemented with 1% N2 (Invitrogen, Sweden, 17502048), 2 mM L-glutamine (Invitrogen, Sweden, 25030024), and 20 ng/mL basic fibroblast growth factor (PeproTech, U.K., 100-18b). Passages 13-16 were used for experiments. AHP cells were detached from T75 cell culture flasks using trypsin 0.25% EDTA (Invitrogen, Sweden, 25200056) and suspended in warm DMEM/F12. For each surface modification, 500 cells/mm2 were seeded on 4-6 samples. Samples were incubated for 1 h at 37 °C in a humidified atmosphere of 95% air and 5% CO2. For 24 h experiments, media were then supplemented with 1% N2 and 2 mM L-glutamine. After incubation, cell suspension was exchanged to warm PBS. The attached cells were fixed in glutaraldehyde, and the cell nuclei were stained with 4′,6-diamidino2-phenylindole dihydrochloride (DAPI). Samples were never allowed to dry prior to mounting. Twelve pictures (0.15 mm2) were automatically acquired from each sample using an Olympus BX61 fluorescence microscope (Olympus Optical AB, Stockholm, Sweden), representing ∼0.2% of the total substrate area. Counting of attached cells was performed manually from DAPI images, taking all nuclei into consideration. The average number of cells for each surface modification was calculated from sample means. Cell data are presented as the average ( standard error of the mean (SEM) as compared to the polyornithine/laminin reference substrate from two to three independent experiments. The presented range in data for the polyornithine/laminin is 100% ( SEM on the basis of the largest experimental spread observed for this particular surface modification at each time point. A one-sided, equal variance T-test was used for statistical analysis, where p < 0.05 was considered significant. (56) Pfeiffer, I.; Hook, F. Anal. Chem. 2006, 78, 7493. (57) Palmer, T. D.; Takahashi, J.; Gage, F. H. Mol. Cell. Neurosci. 1997, 8, 389.
11698 Langmuir, Vol. 23, No. 23, 2007
Thid et al.
Table 1. Quantification of PLL-g-PEG Surface Modification and IKVAV Ligand Density
a
IKVAVfunctionalized PEG chains (mol %)
OWLS measured polymer mass (ng/cm2)
IKVAV density from OWLS and NMR (pmol/cm2)
ligand-ligand in hexagonal pattern (nm)
QCM-D measured polymer mass (ng/cm2)
9.9% 6.9% 5.3% 3.5% 0%
149 ( 15 154 ( 9 149 ( 11 139 ( 4 142 ( 14
5.12 ( 0.52 3.47 ( 0.24 2.52 ( 0.17 1.60 ( 0.05 0
6 7 9 11
624 ( 39 557 ( 93 568a 551 ( 24 571 ( 19
One data point only. Table 2. Quantification of IKVAV Ligand Density on Supported Phospholipid Bilayers maleimide lipids (mol %)
theora IKVAV (pmol/cm2)
SPR measured peptide mass (ng/cm2)
SPR measured peptide density (pmol/cm2)
ligand-ligand in hexagonal patternb (nm)
QCM-D measured peptide mass (ng/cm2)
5% 3% 1% 0%
13.6 8.2 2.7 0
29.5 ( 4.5 20.4 ( 3.0 5.8 ( 1.8
14.6 ( 2.2 10.1 ( 1.5 2.9 ( 0.92
4 5 8
217 ( 36 147 ( 16 71 ( 10
a Calculated assuming A 2 71 lipid ) 61 Å in a bilayer, equal distribution of maleimido groups between membrane leaflets, that only maleimido groups in the top leaflet are available for peptide binding, that there is no flip-flop of maleimido lipids between bilayer leaflets during the time frame of peptide functionalization,63 a 1:1 relation between maleimido groups and peptides, and a peptide weight of 2017 g/mol. b Based on theoretical IKVAV concentration.
Results Hydration of IKVAV-Functionalized Bilayers and PLLg-PEG. The results of peptide density quantification on bilayers and PLL-g-PEG by various techniques are summarized in Tables 1 and 2. The high hydration levels of these adsorbed adlayers will result in an underestimation of adsorbed QCM-D mass for both polymer- and lipid-based surface modifications, which is a typical feature of highly viscoelastic films.45,50,58 With such hydrated adlayers, modeling of QCM-D data is required in order to distill out effects in the measured values due to the viscoelasticity of the films. Comparison of the modeled mass (Voigt model, taking viscoelastic properties of the film into account, data not shown) with the measured QCM-D mass presented in Tables 1 and 2 (eq 1, assuming no viscoelastic losses) showed that the mass of the adsorbed PLL-g-PEG was underestimated by ∼20%, independent of IKVAV functionalization. Similarly, the IKVAV peptide binding to bilayers doped with more than ∼2% of the maleimido lipids was underestimated by ∼20%. For peptide functionalization of bilayers with less than ∼2%, the effect was even more pronounced, with the modeled QCM-D mass being ∼40% larger than the measured QCM-D mass (data not shown). The amount of coupled water could be determined by comparison of the modeled hydrated mass from QCM-D and the “dry” biomolecular mass from the optical methods, OWLS and SPR (Tables 1 and 2). The water content for the PLL-g-PEGbased surface coatings was independent of the degree of IKVAV functionalization (data not shown). In contrast, for the bilayer system, hydration varied with the amount of IKVAV peptide (data not shown). This is readily observed in the plot of peptide mass versus maleimido doping (Figure 3), where the slope for mQCM-D changes at ∼2% maleimido doping, while mSPR increases linearly with maleimido doping. The resulting hydrations of the films were ∼79% for PLL-g-PEG, and, for the bilayers, ∼87% for >2% peptide functionalization and ∼94% for 3 nm because of 10 amino acids preceding the IKVAV motif and the hydrophilic spacer for the maleimido group. It has previously been shown that majority of ligands spaced from the surface by 1.1-3.2 nm could be recognized by cells.64 The PLL-g-PEG polymer in the present study has previously been characterized to be an ∼8 nm thick,38,49 highly hydrated monolayer with PEG chains, in brush conformation, spaced by ∼1-2 nm.37,38 PLL-g-PEG has previously been successfully functionalized with RGD peptides and successfully adsorbed to metal oxide surfaces through electrostatic interactions.15,18,19,39 However, the charges on the grafted RGD-containing peptides and on the current IKVAV peptide are different. This may lead to a slightly different and less well-ordered architecture. The deposition of PLL-g-PEG with pre-grafted IKVAV was found to be at least 5.12 ( 0.52 pmol/cm2. This was the highest IKVAV content obtainable, which still maintained good resistance against nonspecific protein adsorption. The hydration remained constant with IKVAV functionalization since the EG chains, rather than the peptide alone, couple the majority of the water.38,49 The IKVAV sequence can extend ∼3 nm from the surrounding PEG layer. Approximately half of the distance is accounted for by an additional 32 EG moieties on PEG-peptide chains compared to the “background” PEG.37 The other half is due to the five amino acid spacer between the coupling cysteine and the active IKVAV sequence.5,64 Thus, just as in the case of bilayer-presented ligands, the PLL-g-PEG-tethered IKVAVs should also be accessible to antibodies and cells.64 The average spacing between ligands is ∼6-11 nm, assuming a hexagonal arrangement (Table 1); however, a random distribution is expected. In contrast to the supported lipid bilayer system, the possibility of peptides being embedded in the polymer brush structure cannot be excluded, as has previously been hypothesized for RGD as well as biotin in PLL-g-PEG/PEG-ligand systems.15,47 Antibody Association to Ligands. Serum containing antiIKVAV antibodies was used for the detection of available and functional peptide ligands. However, ligand quantification and direct comparison of the bilayer and polymer systems was not straightforward. The antibody-to-peptide association was different for peptides used for the bilayer and PLL-g-PEG functionalizations, even when they were both presented on a supported lipid bilayer (Figure 6). The nature of the antibodies and the valence of interactions are not explicitly known. However, detection of anti-IKVAV antibody using fluorescently labeled anti-IgG secondary antibodies has been reported, indicating that the IKVAV antibody might be of IgG type, likely interacting in a bivalent fashion with the IKVAV peptides.32 Both QCM-D and SPR/OWLS showed more antibody binding with increased ligand concentration on both the bilayers and the polymer (Figures 7 and 8). QCM-D showed that there was no rearrangement of the bilayer, peptide, or the antibody layer during antibody binding (Figure 4C). In contrast, stiffening was observed for the PLLg-PEG/PEG-IKVAV (Figure 5B) upon antibody association, where dissipation decreased during mass increase. Admittedly, such behavior of the QCM-D response might be triggered by resonance in a thick and highly viscoelastic film.65 However, thicker and more dissipative films than the PLL-g-PEG/PEGIKVAV have been successfully studied with QCM-D.65 Therefore the ∆D-∆f behavior is believed to originate from a true physical stiffening, which can be caused by antibodies penetrating into the PLL-g-PEG film and/or underlying film compression and consequent loss of water.38,47 If the antibody-peptide recognition and binding is an appropriate model of cell receptor-ligand (65) Grane´li, A.; Edvardsson, M.; Ho¨o¨k, F. ChemPhysChem 2004, 5, 729.
Ligand Accessibility/Mobility in Cell Attachment
binding, viewed from the perspective of cell attachment, the viscoelastic properties of bilayers would be conserved during the attachment process, while the polymer would release water or the cell receptors would penetrate into the film to find ligands in the polymer structure. Generally, cells are assumed to be able to interact with ligands within the top 5-10 nm of a polymer,66,67 which is comparable to the thickness of the PLL-g-PEG polymer film. Optical surface characterization methods showed that the amounts of antibody binding and ligand density were directly proportional in both the bilayer and the PLL-g-PEG cases. If an ∼150 kDa IgG molecule for the antibody is assumed, only a fraction of the ligands were recognized by the antibodies for both surface modifications. The low association ratio of antibodyto-peptide binding can be caused by an insufficient amount of antibodies in the solution and/or by low availability of peptides for antibody binding. For the PLL-g-PEG, the ligands might very well be hidden in the polymer structure, which has been hypothesized before.15,47 For the bilayer system, where the possibility of hidden ligands in the bilayer is not likely,62 the low peptide availability would rather be due to the size of the antibodies being in the same range as the average distances between peptides (Tables 1 and 2), and this argument holds true for both surface modifications. Still, for qualitative analysis, the large antibodies are considered to be a good model for cell receptors, which are generally also large. For example, integrin diameter in a membrane is ∼10 nm.68 However, such a model is limited, since cell receptors are confined to the cell membrane, while antibodies in solution have a higher degree of freedom.20 In the present study no rigorous conclusions on the quantity of available ligands can be drawn because of different interactions of antibodies with the peptides used for the two surface modifications, and noncomparable kinetics of antibody interaction with the peptides used on supported membranes versus the PLLg-PEG polymer films. Proper quantification of the accessibility and functionality of immobilized ligands requires a more welldefined probe than the non-purified polyclonal antibodies used in this study. However, trends in antibody binding for each surface modification on its own can be evaluated with certainty. We have observed that the IKVAV-containing peptides on both surfaces are recognized by the anti-IKVAV antibody in a linear, dose-dependent manner, thus indicating that the amount of available and functional IKVAV ligands increases proportionally with ligand density over the ranges tested. Cell Attachment. The fast lateral diffusion of IKVAV ligands presented in the current bilayer system might facilitate cell attachment because of the possibility for ligand recruitment and accumulation.4,6,12,21,22 On the other hand, the mobile ligands might be interpreted by the cells as a highly flexible substrate, which typically is not preferred by attachment-dependent cells when given a choice of substrate stiffness.69 Ligands presented on the PLL-g-PEG are relatively stationary compared to the bilayer ligands, as they can only move within a limited radius of gyration. Despite the difference in peptide mobility on these two surface modification schemes, the initial (1 h) cell attachment to IKVAV ligands on supported bilayers and PLL-g-PEG was comparable (Figure 9A). Thus, neither the accumulation possibility on the bilayer, nor the “rigidity” on the PLL-g-PEG seemed to enhance initial attachment. (66) Hersel, U.; Dahmen, C.; Kessler, H. Biomaterials 2003, 24, 4385. (67) Massia, S. P.; Hubbell, J. A. J. Cell Biol. 1991, 114, 1089. (68) Xiong, J. P.; Stehle, T.; Diefenbach, B.; Zhang, R. G.; Dunker, R.; Scott, D. L.; Joachimiak, A.; Goodman, S. L.; Arnaout, M. A. Science 2001, 294, 339. (69) Lo, C. M.; Wang, H. B.; Dembo, M.; Wang, Y. L. Biophys. J. 2000, 79, 144.
Langmuir, Vol. 23, No. 23, 2007 11703
For both surface modifications, a threshold ligand concentration of >3 pmol/cm2 (∼8 nm distance) was required for efficient cell attachment at 1 h (Figure 9A). The explanation behind the threshold ligand density could be that a minimum number of ligands are required at the initial cell-substrate contact to enable the adhesive interaction. In fact, this was hypothesized in a recent study for cell-cell adhesions, where various densities of laterally mobile cell-cell adhesive molecules were presented to cells on supported membranes.20 It was demonstrated that a critical coverage of >0.003 pmol/cm2 (∼240 nm distance) was required to render cell attachment for that particular ligand-cell system. In addition, the provided ligand density also determined the steady-state number of attachment points, demonstrating that, even for laterally mobile ligands, cells can only accumulate a limited amount of attachment points. Other parameters, such as binding strength and receptor/cytoskeletal rearrangements, can also influence the efficiency of cell attachment. In fact, longer incubation time on PLL-g-PEG/PEG-IKVAV yielded efficient cell attachment for ligand densities where 1 h was not sufficient (Figure 9A). This indicates that, with time, on PLL-g-PEG/PEGIKVAV, cells bind more ligands as the probability of a receptorligand interaction increases and/or that possibly hidden ligands find their way to the polymer surface.15,47 Alternatively, cell receptor or cell cytoskeletal restructuring occurs, rendering stronger attachment at 24 h. If in fact some of the IKVAV ligands were hidden in the PLL-g-PEG structure, the IKVAV surface density was overestimated in all 1 h experiments. Consequently, it cannot be excluded that the laterally non-mobile ligands can efficiently bind cells at lower threshold densities than concluded. Cell attachment with efficiency equivalent to the reference substrate could be achieved at 1 h on supported bilayers presenting ∼8 pmol of IKVAV ligands per square centimeter (∼5 nm distance) (Figure 9A). The spacings of IKVAV ligands required for efficient cell attachment seem to be an order of magnitude smaller than that for the RGD-integrin mechanism, where a cell universal spacing of ∼60 nm has recently been reported for single non-mobile ligands.1,3 Interestingly, not only distances but also clustering of RGD ligands has been shown to be of importance, where the critical cluster-cluster distance for efficient cell interaction is highly dependent on the cluster size.70 For individually presented RGD-containing peptides (cluster size of 1) on PEG-tethered polymers, the threshold ligand spacing observed for focal adhesion and stress fiber formation in fibroblasts was less than 6 nm. Differences in ligand composition (and thus ligand-receptor affinity) and substrate rigidity were quoted as possible explanations for this discrepancy. The abovementioned RGD studies were performed on immobile RGD ligands. For RGD ligands presented on a supported lipid monolayer, efficient cell attachment was observed at ∼1 pmol/ cm2 (∼15 nm distance); however, this number varied depending on, for example, the cell type (spreading endothelial cells or non-spreading hematopoietic progenitor cells) and the affinity of the ligand to the receptor.8 The mobility of the lipids in the monolayer was not described. Cell attachment was also recently investigated on polymerized, supported RGD amphiphile films (i.e., of no or limited lateral lipid mobility), where a minimum of ∼5.5 pmol/cm2 peptide concentration was required to support fibroblast cell attachment and spreading.2 This high peptide concentration was attributed to limited ligand accessibility by the integrin receptors due to the tight binding of the ligand to the lipid layer and its close proximity to the surface. Unfortunately (70) Maheshwari, G.; Brown, G.; Lauffenburger, D. A.; Wells, A.; Griffith, L. G. J. Cell Sci. 2000, 113, 1677. (71) Rawicz, W.; Olbrich, K. C.; McIntosh, T.; Needham, D.; Evans, E. Biophys. J. 2000, 79, 328.
11704 Langmuir, Vol. 23, No. 23, 2007
a comparison between polymerized and non-polymerized lipid layers was not performed. In summary, there is a large spread in reported critical ligand density required for efficient cell attachment, where the exact values are highly dependent on the investigated cell-ligand duo and the exact ligand presenting system and cell type.66 What we can conclude from our own data is that a change in IKVAV ligand spacing by as little as ∼3 nm (from 8 nm at 3 pmol/cm2 to 5 nm at 8 pmol/cm2, average spacing assuming uniform homogeneous distribution), for both laterally mobile and non-mobile presentation, was enough to render a non-cell adhesive substrate adhesive to AHP cells, a non-spreading cell type. Supported lipid membranes have also been used to study the attachment of cells to receptors involved in cell-cell adhesion.6,20,21 A systematic variation of concentration of such ligands pinpointed the critical ligand density to be ∼0.007 pmol/cm2 (∼170 nm distance), thus it is generally much lower than that for the extracellular matrix peptides.20 Lateral mobility of membrane-tethered ligands involved in cell-cell interactions has been shown to enhance the efficiency of the ligand to support cell adhesion,4,12,22 but only if a sufficient amount of ligands was provided.20 The ability of a membrane-bound ligand to diffuse laterally enhances cell adhesion by allowing the accumulation of ligands in the area of cell contact, and by increasing the rate of ligand-receptor binding. It was reasonable to believe that the same holds true for the case of cell-extracellular matrix ligand interactions. We have observed a trend of increasing cell adhesion efficiency as a function of ligand density for an extracellular matrix ligand, the IKVAV peptide, where the initial attachment of neural progenitor cells was dependent on ligand density and did not seem to be facilitated by ligand mobility. However, to address the influence of ligand mobility on cell attachment more rigorously, supported bilayers in fluid phase should preferably be compared with a surface modification for which there is no risk of ligands hiding in the structure, such as in situ-modified self-assembled monolayers7,9,11 or supported membranes with lower mobility.12 Lower lipid mobility can be achieved by using lipid systems having higher gel transition temperatures or lower diffusion coefficients compared to the PC-based lipids used in this study. Alternatively, the use of polymerizable lipid systems would allow the possibility of removing the lateral mobility of lipids, and therefore also the mobility of coupled peptides, within the same chemically defined model. Considering the amount of data and understanding already known for the RGD-integrin receptor interaction, perhaps the RGD peptide would be a better model system to study the influence of extracellular matrix peptide mobility on cell attachment. However, for the particular model system used in the present study, we are most interested in
Thid et al.
addressing the cause of the observed AHP cluster formation when grown in IKVAV-functionalized lipid bilayers. To address this issue, the use of a polymerizable lipid bilayer system, in both the rigid and fluidic states, would be most advantageous.
Conclusions The accessibility and functionality of extracellular matrix ligands presented at surfaces were evaluated for the conditions of laterally mobile versus non-mobile ligands by studying antibody binding and early cell attachment as a function of ligand concentration. For the presentation of non-mobile ligands, a new peptide-functionalized PLL-g-PEG polymer was successfully synthesized with an IKVAV-containing peptide sequence at a maximum of 5.12 pmol IKVAV ligands per square centimeter. IKVAV-containing peptides presented on laterally mobile, supported phospholipid bilayers and non-mobile PLL-g-PEG surfaces were recognized by an anti-IKVAV antibody in a dosedependent manner, thus indicating that the amount of available IKVAV ligands increases proportionally with ligand density over the concentrations tested. In comparison, the attachment of AHP cells to IKVAV-functionalized PLL-g-PEG and supported phospholipid bilayers followed a sigmoidal dependence on peptide concentration, with a critical concentration of approximately 3 pmol/cm2 IKVAV ligands required to support initial AHP cell attachment for both surface modifications. There appeared to be little influence of IKVAV peptide mobility on the number of AHP cells attaching after 1 h in culture. In order to address the influence of ligand mobility on cell attachment more rigorously, it would be beneficial to compare supported bilayers in fluid phase with a chemically and structurally more similar surface modification scheme, such as in situ peptide-modified selfassembled monolayers or supported membranes with reduced or no lateral mobility. Acknowledgment. We would like to acknowledge Dr. Sofia Svedhem, Chalmers University of Technology, and Prof. Fredrik Ho¨o¨k, Lund University, for helpful discussions. We would also like to thank Prof. Hynda Klainmann at NIH for the generous donation of the anti-IKVAV antibody serum and the late Prof. Peter S Eriksson at Go¨teborg University for kindly providing the AHP cells. The EC STREP-NANOCUES project (FP6-NMP2002-3.4.1.2-1) and Chalmers Bioscience Initiative are acknowledged for financial support. Supporting Information Available: Epifluorscence and differential interference contrast images of AHP cells grown on IKVAVpeptide-functionalized supported phospholipid bilayers and PLL-g-PEG surfaces. This information is available free of charge via the Internet at http: //pubs.acs.org. LA701159U
