Tumor Necrosis Factor (TNF)-Functionalized Nanostructured Particles

Publication Date (Web): October 27, 2005 .... Dual function of cysteine rich domain (CRD) 1 of TNF receptor type 1: Conformational stabilization of CR...
0 downloads 0 Views 514KB Size
Download PDF
Bioconjugate Chem. 2005, 16, 1459−1467

1459

Tumor Necrosis Factor (TNF)-Functionalized Nanostructured Particles for the Stimulation of Membrane TNF-Specific Cell Responses Susanne Bryde,⊥ Ingo Grunwald,|,⊥ Angela Hammer,† Anja Krippner-Heidenreich,† Thomas Schiestel,† Herwig Brunner,†,§ Gu¨nter E. M. Tovar,†,§ Klaus Pfizenmaier,† and Peter Scheurich*,† Institute of Cell Biology and Immunology, University of Stuttgart, Allmandring 31, D-70569 Stuttgart, Germany, Fraunhofer Institute for Interfacial Engineering & Biotechnology, Nobelstrasse 12, D-70569 Stuttgart, Germany, and Institute for Interfacial Engineering, University of Stuttgart, Nobelstrasse 12, D-70569 Stuttgart, Germany. Received June 24, 2005; Revised Manuscript Received September 22, 2005

Most members of the tumor necrosis factor (TNF) ligand family occur in both a membrane-bound and a soluble form, which can possess differential bioactivities. The aim of this work was the construction of a synthetic-biological hybrid system consisting of chemically nanostructured core-shell particles with a diameter of 100 nm, 1 µm, or 10 µm and the cytokine TNF to obtain a tool that mimics the bioactivity of naturally occurring membrane-bound TNF. Synthetic core-shell nanoparticles consisting of an inorganic silica core and an ultrathin organic shell bearing a maleimide group at the shell surface which allowed for a covalent and site-directed coupling of CysHisTNF mutants were prepared. The TNF mutants were modified at the N-terminus by PCR cloning by introducing a His-Tag for purification and a free cysteine group for reaction with the particle-attached maleimide group. The resulting nanostructured hybrid particles initiated strong TNF receptor type 2 specific responses, otherwise only seen for the membrane-bound form of TNF, but not the soluble cytokine, thus clearly demonstrating new and membrane TNF-like properties of the bioconjugated soluble TNF.

INTRODUCTION

Nano- to micrometer-sized particles made of inorganic and organic materials have been used in the past for different biological applications. Micrometer-sized particles for example were used to bind biological targets from biological samples, as has been published for cell sorting (Bildirici and Rickwood, 2001; Kruger et al., 2000) or protein enrichment and purification (Hurst et al., 1999). Furthermore, particles were functionalized with peptides for promoting cellular uptake of the bioconjugated particles (Becker et al., 2004), as a DNA carrier (Maruyama et al., 1997), for drug delivery (Lockman et al., 2002), or for mammalian cell transformation (Sandhu et al., 2002). In addition, fluorescent-labeled particles decorated with affinity ligands such as antibodies were successfully used for cell imaging (Willard, 2003; Gao et al., 2005). Particles exposing streptavidin on their surface to allow coating of the particles with biotin-conjugated proteins were shown to represent a valuable tool for spatially restricted receptor-selective stimulation of cells (Verveer et al., 2000). * To whom correspondence should be addressed. Phone: +49 711 685 6987; Fax: +49 711 685 7484; E-mail: peter.scheurich@ izi.uni-stuttgart.de. † Institute of Cell Biology and Immunology, University of Stuttgart. ‡ Fraunhofer Institute for Interfacial Engineering & Biotechnology. § Institute for Interfacial Engineering, University of Stuttgart. | Present address: Fraunhofer Institute for Manufacturing and Advanced Materials, Bonding Technology and Surfaces Department, Wiener Str. 12, D-28359 Bremen, Germany. ⊥ These authors contributed equally to this work.

In multicellular organisms, receptor ligands are often presented at the surface of cells. Such cell membranebound ligands bind highly specific to their corresponding interaction partners, based on molecular recognition processes. This type of intercellular communication is essential to sustain homeostasis in all higher organisms. The inflammatory cytokine tumor necrosis factor (TNF, also termed TNFR) is the representative member of a large homologous protein ligand family, mirrored by a respective family of cell surface membrane receptor proteins (Locksley et al., 2001; Aggarwal, 2003). TNF is crucially involved in inflammatory responses to elicit effective innate immune responses but is also a key player in many pathological situations such as septic shock, rheumatoid arthritis, and neurodegenerative diseases, e.g. multiple sclerosis (Grell and Scheurich, 1999; Wajant et al., 2003). In addition, TNF is capable of inducing apoptosis and is a promising candidate for new strategies in cancer therapy under conditions where systemic side effects can be avoided, as demonstrated in isolated limb perfusion (Eggermont and ten Hagen, 2001). Most studies with TNF have been performed with the soluble cytokine (sTNF). However, typical for the members of the TNF ligand family, TNF is a genuine type 2 transmembrane protein (memTNF) and initially expressed at the cell membrane. From this membranebound protein sTNF is subsequently derived from proteolytic processing by the metalloproteinase TACE (tumor necrosis factor alpha-converting enzyme) (Black et al., 1997). Both memTNF and sTNF bind to two different cell membrane receptors, termed TNFR1 (CD120a) and TNFR2 (CD120b) (Locksley et al., 2001). Whereas TNFR1 is ubiquitously expressed in all tissues, TNFR2 is highly

10.1021/bc0501810 CCC: $30.25 © 2005 American Chemical Society Published on Web 10/27/2005

1460 Bioconjugate Chem., Vol. 16, No. 6, 2005

Bryde et al.

Scheme 1. TNF-r Signaling Capabilitiesa

a TNF is a transmembrane protein (membrane or memTNF) and initially expressed at the cell surface. From this membranebound form the soluble TNF (sTNF) is proteolytically derived by action of the metalloproteinase TACE (tumor necrosis factor alphaconverting enzyme). Both memTNF and sTNF bind to two different cell membrane receptors, termed TNF receptor (TNFR) 1 and 2. Membrane TNF strongly activates both receptors, whereas sTNF is capable to only stimulate TNFR1 but not TNFR2. Functionalized particles with a recombinant TNF protein bind to and strongly activate both receptors, thus mimicking the action of memTNF.

regulated in expression and mainly found in immune cells and endothelial cells but also in neuronal tissue. The two TNF forms possess differential capabilities for the induction of larger signaling complexes. Whereas sTNF binding enforces receptor trimerization only, cellassociated memTNF might subsequently cause the formation of larger complexes by induction of capping of the membrane-bound ligand molecules and the receptors at the respective interacting cellular sites. In light of these data, it is very likely that the physiological role of TNFR2 is largely underestimated, simply because sTNF has been used in by far the most studies. Accordingly, there is a need for a memTNFadequate stimulus that can be easily applied in experimental systems in vitro but also in animal model systems in vivo. To overcome the lack of an appropriate memTNFlike stimulating reagent, we have here constructed silicabased particles covalently coupled with mutated functional TNF derivatives (here also termed as muteins), forming bioactive homotrimers at the particle surface (Scheme 1). These novel particles are able to initiate memTNF-resembling cellular responses. EXPERIMENTAL PROCEDURES

Particle Synthesis and Characterization. Nanoand microparticles of three different mean particle sizes (100 nm, 1 µm, and 10 µm) and consisting of silica were used as core material for organic modification. The smaller sized silica particles (100 nm and 1 µm) were synthesized following the method of Stober (Stober et al., 1968) as described elsewhere (Schiestel et al., 2004). Briefly, to obtain particles with a mean particle diameter of 100 nm, a mixture of 1.37 mL ammonia (25 wt. %, SAF, Taufkirchen, Germany), 2 mL of H2O, and 500 mg of tetraethoxysilane (TEOS; ABCR, Karlsruhe, Germany) in 40 mL of ethanol (HPLC-grade, SAF, Taufkirchen, Germany) were stirred for 24 h at room temperature (RT). Afterward they were separated by centrifugation

(13000 rpm, 15 min) and suspended in 10 mL of ultrapure water, then put into dialysis until conductance was below 1.3 µS/cm. Particle sizes and ζ-potentials were routinely monitored by photocorrelation spectroscopy and microelectrophoresis using a Zetasizer 3000 HSA (Malvern Instruments, Herrenberg, Germany). ζ-Potentials of the 100 nm and 1 µm particles were determined at pH 4.7 in acetate buffer. The ζ-potential of 10 µm particles was not determined by microelectrophoresis due to instability toward sedimentation during the measurements. To generate particles with a mean diameter of 1 µm, a mixture of 1.44 g of ammonia, 1.04 g of TEOS, and 10 mL of ethanol were stirred for 3.5 h at RT. The resulting particles were washed with 100% ethanol and suspended in 10 mL of 100% ethanol prior to any further modification. The 10 µm particles were purchased from Kisker (Steinfurt, Germany) and washed in the particular solvent before use. Preparation of Fluorescent Shell. An ultrathin fluorescent shell for confocal microscopy studies was covalently attached to the 100 nm, 1 µm, and 10 µm particles according to the following method. N-Hydroxysuccinimide ester Alexa dye 546 or 568 (MoBiTec, Go¨ttingen, Germany) (500 µg) was incubated for 3 h at RT with 111 µg of (3-aminopropyl)triethoxysilane (ABCR, Karlsruhe, Germany) in 64 µL of ethanol. An amount of 50 mg of silica particles in 1.6 mL of 100% ethanol, 3.3 µL (3.05 mg) of TEOS, and 167 µL of ammonia were added, and the mixture was stirred for 20 h at RT. To avoid photobleaching, all steps involving fluorescence markers were performed in the dark. The resulting particles were washed five times with 100% ethanol and once with acetate-buffered solution (pH 4.7) and suspended in acetate buffer. Particle Surface Modification. An amount of 100 mg of silica particles (in acetate buffer) was stirred with 100 µL of ammonia, 6.8 mg of TEOS, and 20.4 mg of APS for 24 h at room temperature. The resulting particles

Bioconjugate Chem., Vol. 16, No. 6, 2005 1461

Covalently Immobilized Tumor Necrosis Factor

were washed intensively and suspended in water. An amount of 1 mg of APS-modified particles was suspended in PP72 (183 µL of 10 mM phosphate buffer pH 7.2), 817 µL of a 3.06 mM sulfosuccinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (SMCC; Perbio Science, Bonn, Germany) solution in PP72 was added, and the reaction mixture was shaken for 1 h. All steps were performed under sterile conditions. The resulting particles were collected by centrifugation at 4 °C, washed with 1 mL of ice-cold PP72, suspended in 1 mL of icecold PP72, and then immediately used for protein coupling reactions. Construction and Purification of CysHisTNF and Derived Muteins. Expression constructs optimized for bacterial codon usage of a histidine-tagged soluble TNF (HisTNF) was originally obtained from Philippe Holliger, Cambridge, UK. The CysHisTNF expression construct, pQE5 CysHisTNF, was produced by site-directed mutagenesis introducing the coding sequence of a cysteine residue as well as a factor Xa cleavage site (IEGR) and a His tag. This generated an N-terminal sequence tag (MGELIEGRCAGGSGHHHHHHGSDGAS) fused to serine 79 of human TNF. Receptor specific TNF muteins (CysHisTNF143N/145R and CysHisTNF32W/86T) were also generated by site-directed mutagenesis using pQe5CysHisTNF as template. All constructs generated by PCR were verified by sequencing. For the production of CysHisTNF, E.coli XL1 blue cells transformed with CysHisTNF expression constructs were grown overnight. The following day the cultures were diluted 1:10, grown to a density of OD600 nm 0.5-0.7, and then induced by 1 mM IPTG for 6 h at 30 °C. Cells were washed in phosphate-buffered saline (PBS) and centrifuged, and the pellet was frozen at -20 °C. The cell pellet was resuspended in PBS (3 mL PBS/g pellet), cells were lysed by sonification on ice, and the cell debris was pelleted at 12000 g for 15 min at 4 °C. CysHisTNF was purified using a HiTrap column (1 mL; Pharmacia) following the instructions of the manufacturer. Briefly, the column was connected to a peristaltic pump with a flow rate of 1 mL/min and equilibrated with 10 mL of buffer 1 (20 mM NaH2PO4 pH 7.4, 0.5 mM NaCl, 10 mM imidazole), and then the supernatant was loaded and the column was washed with 10 mL of buffer 2 (20 mM NaH2PO4 pH 7.4, 0.5 mM NaCl, 30 mM imidazole) followed by elution of CysHisTNF using 7 mL of buffer 3 (20 mM NaH2PO4 pH 7.4, 0.5 mM NaCl, 500 mM imidazol). Fractions of 500 µL were collected and analyzed for their CysHisTNF content by Coomassie gel analysis as described in Grunwald et al. (2003). Positive fractions were pooled and dialyzed against PBS at 4 °C. Protein concentration was determined by Bradford method (BioRad, Germany) and the protein stored at 4 °C. TNF-Coupling to Core-Shell Particles. Similar to wildtype TNF, CysHisTNF exists in its bioactive form as a noncovalently linked homotrimer, with each monomer carrying a free Cysteine residue near its N-terminus, all three located close together. Accordingly, CysHisTNF tends to form disulfide bridges, one type linking two of its monomers and a second type bridging two CysHisTNF homotrimers. To reduce these cysteine bridges, CysHisTNF or its derivatives were pretreated with 10 mM dithiothreitol (DTT) overnight at 4 °C. DTT was removed by dialysis for 3 h against PBS, and the solution was sterilized by filtration. DTT-treated CysHisTNF showed the same bioactivity as the native protein, indicating that the reduction of the N-terminal cysteines does not influence the disulfide bridges within the TNF monomers (data not shown).

Nanostructured particles with a diameter of 100 nm, 1 µm, or 10 µm and bearing a sulfoSMCC-shell (1 mg beads/mL) were incubated with 25 µg (1 µm, 100 nm beads) of CysHisTNF in PBS or 50 µg (10 µm beads) of CysHisTNF in PBS for 1 h at RT in vials which were constantly being rolled. Control beads were prepared by addition of 200 mM β-mercaptoethanol to 1 mg of beads in PBS. Labeled beads were centrifuged at 14000 g for 10 min at 4 °C. The supernatant with surplus CysHisTNF was discarded, and the beads were washed twice (centrifugation step between the washings) with sterile cell culture medium to remove unligated CysHisTNF. The final preparation was resuspended in 200 µL of culture medium and stored at 4 °C, protected from light. Bioactivity of the particles was determined in a cytotoxicity assay as described. Cell Lines. HeLa cells stably transfected with human TNFR2 (HeLa80) or immortalized mouse fibroblasts (MF) from TNFR1/TNFR2 double knockout mice stably transfected with TNFR1-Fas or TNFR2-Fas expression plasmids (MF TNFR1-Fas, MF TNFR2-Fas) have been described elsewhere (Weiss et al., 1997; KrippnerHeidenreich et al., 2002). The human rhabdomyosarcoma cell line KYM-1 was supplied by M. Sekiguchi (University of Tokyo, Tokyo, Japan), and the human HeLa cell line was obtained from the American Type Culture Collection. HeLa cells, KYM-1 cells, and the immortalized mouse fibroblasts MF TNFR1-Fas and MF TNFR2-Fas were grown in RPMI 1640 medium supplemented with 5% (v/ v) heat-inactivated fetal calf serum and 2 mM Lglutamine. The expression of the receptor fusion proteins TNFR1-Fas and TNFR2-Fas in fibroblasts was controlled routinely by FACS analysis (Krippner-Heidenreich et al., 2002). Transient Transfection and Live Imaging by Confocal Laser Scanning Microscopy. Cells were harvested and transiently transfected with 6 µg of pFADD-EGFP or pTNFR2-EGFP expression plasmids by electroporation. An amount of 106 cells per 800 µL of culture medium was electroporated at 250 V and 1800 µF in a 0.4 cm cuvette (Peqbio Easyject Plus, Peqlab). An amount of 3 × 105 cells was seeded immediately after electroporation into 3.5 cm coated glass bottom dishes (Mattek Corp., Ashland, MA) and cultured for 24 h before stimulation. Cells were then treated with CysHisTNFlabeled particles of different sizes. Images were taken at the indicated time points using a TCS SL confocal laser scanning microscope (Leica). Cytotoxicity Assays. Mouse fibroblasts or KYM-1 cells (1 × 104 cells/well) were grown in 96-well plates overnight. Cells were then treated as indicated and cultivated overnight. The next day cells were washed three times with PBS followed by crystal violet staining (20% methanol, 0.5% crystal violet) for 15 min. The wells were washed with H2O and air-dried. The dye was resolved with methanol for 15 min, and optical density at 550 nm was determined with an ELISA plate reader (SPECTRAmax 340PC, Molecular Devices Corp., Sunnyvale, CA). RESULTS

Functionality of Organically Activated CoreShell Particles. Core-shell nanoparticles and microparticles were prepared consisting of an inorganic core and an organic shell bearing both fluorescent markers and functional surface groups suitable for the coupling of TNF muteins. The inorganic core of the particles consisted of silica prepared via sol-gel chemistry (Scheme

1462 Bioconjugate Chem., Vol. 16, No. 6, 2005

Bryde et al.

Scheme 2. Synthesis, Modification and Assembling of the TNF-Functionalized Particlesa

a Core particles are generated in a reaction mixture containing tetraethoxysilane, ammonia, water, and ethanol. Subsequently the silica particles are surface-modified using APS, ammonia, and TEOS (A). The particles are then surface activated using the heterobifunctional cross-linker SMCC (B) which is coupled to the amino groups of the modified particles. The maleimide group of the cross-linker then reacts with the free sulfhydryl groups of the TNF molecule, generating stable thioether bonds. The schemes are not drawn at the same scale (calculated sizes are: SMCC ∼ 0.13 nm, TNF ∼5.5 × 4.4 nm and nanoparticle ∼ 100 nm) with the exception of the right part of the functionalized particle in B.

2A) and showed a negative ζ-potential of -(13.9 ( 1.6) mV for the 100 nm particles and -(13.7 ( 3.3) mV for the 1 µm particles. The ζ-potential was shifted toward highly positive values during the preparation of the organic shell consisting of APS (Scheme 2B) and fluorescent markers and resulting values of (40.1 ( 4.2) mV for the 100 nm particles and (43.2 ( 6.5) for the 1 µm particles. The colloidal stability of the core-shell particles was high enough that in this case the ζ-potential of the 10 µm particles was detectable by microelectrophoresis, rendering a value of ∼52.3 mV. The organic shell was further modified by the bifunctional organic cross-linker SMCC, bearing one chemical group for reaction with the particles’ amino surface and a maleimido group as second functional group for formation of a shell surface suitable for later protein coupling reactions to the shell surface. The shell modification by SMCC led to a slightly negative ζ-potential with -(3.4 ( 6.0) mV for the 100 nm particles. Functionality of TNF-Coupled Nanoparticles. For covalent binding on the activated silica particles, CysHisTNF was used, a TNF mutein containing a His tag for purification purposes and a cysteine residue near its N-terminus (Scheme 2B). Under nonreducing conditions CysHisTNF rapidly forms covalently linked dimers (calculated Mr 39 × 103; Figure 1A, lane 1). To obtain reduced CysHisTNF, the TNF mutein was treated with DTT overnight and subsequently dialyzed against PBS (Figure 1A, lane 2). This protein shows a slightly higher apparent molecular weight as compared to soluble human TNF, in accordance with the added CysHis tag (Figure 1A, lane 3). Mouse fibroblasts derived from TNFR1 and TNFR2 double knockout animals stably expressing TNF receptor/ Fas chimeras rapidly develop apoptosis when adequately stimulated. The TNFR1-Fas chimera positive cells respond to both sTNF and memTNF, whereas the TNFR2Fas chimera only responds to the membrane-bound form of the ligand (Krippner-Heidenreich et al., 2002). TNFR2Fas positive cells were used for an initial characterization of the bioactivity of TNF-coated particles. Soluble CysHisTNF was inactive on these cells over a wide dose range and induced only at high concentrations a weak cytotoxic effect (Figure 1B), presumably caused by the presence of a minor proportion of higher molecular weight aggregates, e.g. hexamers comprised of three covalently linked dimers (data not shown). In the presence of the TNFR2-specific antibody 80M2, however, known to de-

velop a membrane TNF-like activity when combined with soluble TNF (Grell et al., 1993; Grell et al., 1995), a strong cytotoxic activity toward TNFR2-Fas positive cells was observed (Figure 1B). These data demonstrate that CysHisTNF bioactivity resembles wildtype sTNF, i.e., the introduction of the N-terminal tags did not significantly affect sTNF’s function. When coupled to beads, the parental TNFR1/TNFR2 knockout fibroblasts were fully unresponsive to treatment with CysHisTNF positive particles, similar to unconjugated control beads (Figure 1C). However, when TNFR2-Fas-expressing mouse fibroblasts were used, CysHisTNF beads induced a strong and dose-dependent cytotoxic effect, whereas control beads were ineffective (Figure 1D). These results demonstrate a memTNF-like bioactivity of CysHisTNF positive beads. Receptor Selectivity of Beads Coated with CysHisTNF Mutants. Various mutants of human TNF have been produced that exert a pronounced selectivity for one of the two TNF receptors (Van, X et al., 1993; Loetscher et al., 1993). A high selectivity for TNFR1 and TNFR2, respectively, can be obtained by exchange of only two amino acids, leading to the TNF mutants TNF32W/ 86T (TNFR1 selective) and TNF143N/145R (TNFR2 selective), that have been characterized quite extensively in the past (Beyaert and Fiers, 1994). CysHis-variants of these two TNF muteins were constructed, produced, and coupled to activated beads. CysHisTNF32W/86T shows a bioactivity on TNFR1-Fas chimeric receptors comparable to wild-type TNF but is inactive on TNFR2. CysHisTNF143N/145R possesses a mild memTNF-like bioactivity on TNFR2-Fas molecules while inactive on TNFR1-Fas (Krippner-Heidenreich et al., 2002). As expected, a marked receptor selectivity could be observed when applied to cells expressing TNFR1-Fas or TNFR2-Fas receptor chimera. Similar to control beads, CysHisTNF143N/145R positive beads induced no cytotoxic response in TNFR1-Fas-expressing mouse fibroblasts, whereas CysHisTNF32W/86T positive beads were strongly cytotoxic (Figure 2A). Vice versa, the TNFR2Fas chimera expressing cells were responsive to treatment with CysHisTNF143N/145R positive beads only (Figure 2B). Stability of CysHisTNF-Labeled Beads. To test for the stability of generated TNF-beads, CysHisTNFcoupled particles of 10 µm diameter were produced washed and their cytotoxic activity was quantified using

Covalently Immobilized Tumor Necrosis Factor

Bioconjugate Chem., Vol. 16, No. 6, 2005 1463

Figure 1. Characterization and bioactivity of soluble CysHisTNF and CysHisTNF coupled beads. CysHisTNF was incubated overnight with 10 mM DTT to reduce covalently linked dimers and dialyzed for 3 h against PBS. A. An amount of 5 µg of protein per lane was analyzed using a 15% nonreducing SDS-PAGE. Lane 1: CysHisTNF before DTT treatment, lane 2: CysHisTNF treated with DTT, lane 3: wild-type soluble TNF. The arrows at the left indicate dimeric CysHisTNF, monomeric CysHisTNF, and wild-type sTNF monomers, respectively. Numbers at the right indicate molecular weight markers. B, C, and D. Cell viability was determined in a standard cytotoxicity assay using crystal violet staining. Results are expressed as mean values of duplicates, representative results from three independent experiments are shown. B. TNFR2-Fas expressing mouse fibroblasts were stimulated with soluble CysHisTNF alone (0) or after preincubation with the mAb 80M2 (9; 1 µg/mL). C and D. DTT-treated CysHisTNF was coupled onto activated 1 µm beads. TNF-coated particles (9) and uncoated beads (0) as a negative control were used to stimulate TNFR1/TNFR2 double knockout mouse fibroblasts (C) or TNFR2-Fas expressing cells (D).

Figure 2. Receptor selectivity of beads coated with CysHisTNF mutants. The TNF derived muteins CysHisTNF32W/86T (0) and CysHisTNF143N/145R (9) were pretreated with DTT to reduce covalently linked dimers, dialyzed, and coupled onto activated beads. Bioactivity and receptor-selectivity were determined using TNFR1-Fas (A) and TNFR2-Fas (B) chimera expressing mouse fibroblasts in comparison to TNF negative control beads (b). Cell viability was quantified using crystal violet staining. Mean values of duplicates are shown.

the rhabdomyosarcoma cell line KYM-1 (Figure 3A). An aliquot was stored at a concentration of 5 µg particels/ µL in fetal calf serum containing cell culture medium for five weeks at 4 °C. Thereafter, the beads were washed once to remove unbound CysHisTNF, and their bioactivity was tested in a cytotoxicity assay (Figure 3B). Whereas control beads induced no significant cytotoxic effect in both experiments, the CysHisTNF positive beads induced a strong cytotoxic effect with half-maximum efficiency at a concentration of about 0.5 µg/well, indicating no significant loss of bioactivity during the storage period of five weeks (Figures 3, A and B).

Specific Bioactivity of CysHisTNF-Labeled Beads. The LD50 value of CysHisTNF coated beads of 1 µm size was found to be in the order of 0.2 µg beads/well (1.0 µg/ mL) in a standard cytotoxicity assay using TNFR2-Fas positive mouse fibroblasts (Figure 1D). From these values it can be calculated that about 2 × 105 particles induced half-maximum apoptosis in a single well containing 20000 cells, or 10 beads per cell (1 mg of beads with 1 µm size equals to 9.55 × 108 particles). This method to assess the bioactivity of the beads might be inadequate, however, since the particles might come into contact with their target cells only statistically. More reliable informa-

1464 Bioconjugate Chem., Vol. 16, No. 6, 2005

Bryde et al.

Figure 3. The bioactivity of TNF-coated particles remains stable upon storage at 4 °C. CysHisTNF-coupled beads (9) were stored at a concentration of 5 µg particles/µL for five weeks in FCS-containing culture medium. Subsequently they were washed, resuspended in fresh culture medium, and tested on KYM-1 cells in a standard cytotoxicity assay in comparison to control beads negative for TNF (0). A. Freshly prepared particles. B. Beads after storage for 5 weeks at 4 °C in RPMI/FCS 5%. Results are expressed as the mean values of duplicates.

Figure 4. Induction of apoptosis in TNFR2-Fas positive mouse fibroblasts.Alexa-Fluor-568 stained beads of 10 µm diameter had been coated with CysHisTNF. 106 TNFR2-Fas mouse fibroblasts were transiently transfected with human pEGFP-FADD and grown overnight in glass bottom culture dishes. An amount of 100 µg of beads was added to the cells, and a particular cell was chosen for further observation (t ) 0) using a confocal laser scanning microscope. Images were taken every 10 s over a time period of 40 min (A-I). The white arrows indicate FADD recruitment. Note that the development of the apoptotic phenotype takes place within only about 2 min after a lag time of about 29 min (F-I).

tion regarding the potency of TNF-coupled beads could be obtained from data at the single cell level. Large CysHisTNF-coated particles of 10 µm diameter, prelabeled with Alexa-Fluor-568-dye, were used for live cell imaging experiments using confocal microscopy. These beads were added to TNFR2-Fas positive mouse fibroblasts overexpressing human FADD (Fas-associated death

domain protein)-EGFP. The cells were then observed regarding their contact with one or more beads. In Figure 4, a single TNFR2-Fas-expressing mouse fibroblast is shown to undergo apoptosis after stimulation by three 10 µm CysHisTNF beads. As a receptor-proximal event, FADD-EGFP recruitment into the death-inducing signaling complex (DISC) was evident from induced membrane

Covalently Immobilized Tumor Necrosis Factor

colocalization with beads (yellow color; see white arrows). Cell shrinking, the first optically visible sign of cell death, was observed after 29 min. Blebbing and fragmentation of the cell into smaller vesicles followed rapidly within the next 2 min (Figure 4). To perform statistical analyses, we used particles of 1 µm diameter. TNFR2-Fas positive mouse fibroblasts expressing human FADD-EGFP were treated with 25 µg of red fluorescent beads that had been coupled with CysHisTNF. After an incubation time of 5 min, single cells were identified with beads in contact and observed for the next 30 min. Twenty-seven cells from nine different sample preparations were observed to develop apoptosis. These cells had been in contact, in average, with six beads (6.3 ( 3.2, n ) 27; data not shown). Approximately 600 cells from six different dishes were subsequently counted to determine the total percentage of apoptotic cells (66%). Accordingly, at a beads dose close to the LD50 value, the apoptotic cells had been in contact, on average, with six beads. This value is very similar to that determined in dose-response assays applying serial dilution of beads, where the observed LD50 value of 0.2 µg/well corresponds to 10 beads/cell (Figure 1D). Within a size ranging from 100 nm to 10 µm in diameter, the beads had been estimated to bind on average 4.5 µg of CysHisTNF per mg of carrier material (3.2-5.5 µg/mg beads; n ) 11; data not shown). These results indicate some limiting factor in the coupling procedure independent of the surface size of the beads, which varies by a factor of hundred within the used size range. Accordingly, a single particle of 1 µm diameter has bound about 50000 molecules of TNF (trimers) at the surface, a number which is in good agreement with the specific binding capacity calculated earlier (Schiestel et al., 2004). We estimated from microscopical analyses that one-third of the particle surface would be in direct contact with the cell, i.e., capable to engage receptors. From these estimations, it follows that, with an average of six beads contacting the cell surface at a LD50 dose, about 100000 TNF molecules are available for signaling. However, the receptor number available at the contact site of the cell is much lower. We assume that the 45000 TNFR2-Fas molecules expressed per cell are homogeneously distributed at the cell surface, as indicated from confocal microscopy data (Krippner-Heidenreich et al., 2002). The cellular contact site with the six beads covers about 5.6 µm2, or about 0.6% of the cell surface (with an estimated available cell surface of about 900 µm2), i.e., only about 270 receptor molecules would be directly targeted. Although the number of engaged receptors might increase somewhat due to lateral diffusion of more molecules into the contact site, these data indicate that particle-mediated activation of less than 1000 receptors is sufficient to kill a cell. These estimations indicate a very high memTNF-like potential of the carrier coupled CysHisTNF. DISCUSSION

Most of the members of the TNF-ligand superfamily exist in two bioactive forms: a membrane-bound form and its soluble counterpart, processed from the membranebound TNF by metalloproteases and/or produced by alternative splicing (Black et al., 1997; Grell and Scheurich, 1999). Both soluble and membrane-bound TNF do bind and stimulate TNFR1. Although TNFR2 also binds sTNF and memTNF, only the latter is capable to strongly activate this receptor (Grell et al., 1995). Therefore, the investigation of TNFR2 signaling pathways has long been

Bioconjugate Chem., Vol. 16, No. 6, 2005 1465

hampered by the lack of adequate memTNF-mimicking tools. In this work, we have produced nano- and microparticles with directedly and covalently coupled TNF derivatives at their surface (Scheme 1). As core particles, silica-based beads have been used that can be easily activated for covalent coupling (Schiestel et al., 2004). Silica particles of various sizes were coated with CysHisTNF, exposing a free SH-group in their N-terminal cysteine residue to generate a memTNF equivalent. In average, about 5 µg of CysHisTNF could be coupled per mg of beads, independent of its size ranging between 0.1 µm and 10 µm. This indicates a decreasing density of the coupled TNF molecules with decreasing particle size. In any case the resulting particles contained a much higher density of CysHisTNF molecules on their surface as compared to a typical receptor density of a TNF responsive cell (approximately 16000 molecules of CysHisTNF per µm2 on a 1 µm particle versus an about 100 fold lower cellular receptor density). Beside covalent coupling, some adsorption of CysHisTNF was also observed in our experiments, although the particles had been intensively washed to remove unligated CysHisTNF. This was estimated to be 10% or less of the covalently coupled material, as calculated from control experiments with wild-type TNF (data not shown). CysHisTNF-coated beads could be stored at 4 °C for several weeks without significant loss in bioactivity, although small amounts of bioactive, soluble TNF was detectable in supernatants of pelleted particles. This minor leakiness of the conjugated particles can be explained by a slow release of passively adsorbed material and/or by reversible dissociation of TNF trimers that are covalently linked to the particle surface by only one or two of its monomers, rather than all three monomers of the homotrimer (Scheme 1, Figure 3, and data not shown). The oriented presentation of CysHisTNF at the surface of the beads, with its N-terminus forming a covalent bond with sulfoSMCC, permits the formation of biologically active trimers possessing a bioactivity comparable to the membrane-expressed ligand. This could be demonstrated by the effective induction of apoptosis in cells expressing wild-type TNF receptors (Figure 3) as well as TNFR2Fas chimera (Figure 4), the latter being responsive to membrane-bound TNF only but fully resistant to soluble TNF (Krippner-Heidenreich et al., 2002). Induction of apoptosis was preceded by recruitment of FADD to the membrane-bound receptor chimeras (Figure 4), further supporting the formation of a functional death inducing signaling complex. Similarly, in cells expressing wild-type TNFR2 molecules, stimulation of the cells with CysHisTNF positive beads induced receptor clustering, recruitment of TRAF2-EGFP, and translocation of p65/ relA into the nucleus (data not shown). Cells from various human tissues have been shown to typically express between 300 and a few thousand endogenous TNF receptors (Scheurich et al., 1986; Kull, Jr. et al., 1985). Accordingly, the receptor density at the cell surface is rather low, provided that TNF receptors are distributed homogeneously, which is supported for TNFR2 by confocal microscopy analyses of ectopically expressed TNFR2-EGFP and of TNFR2-Fas chimeric receptors (Krippner-Heidenreich et al., 2002). In the case of the TNFR2-Fas chimera we could estimate the efficiency of the beads in comparison to the most efficient tool for TNFR2 stimulation so far available, the combination of sTNF with the TNFR2 specific mAb 80M2 (Grell et al., 1993; Grell et al., 1995). A comparable cytotoxic activity (50 and 66% of maximum apoptosis) could be

1466 Bioconjugate Chem., Vol. 16, No. 6, 2005

obtained with about 6500 sTNF molecules per cell in the presence of the mAb 80M2 or 6 CysHisTNF positive 1 µm beads. These beads present a large number of TNF molecules (about 100000 per particle) but are likely to interact with only a few hundred cellular receptor molecules. Accordingly, particle-coupled TNF proved to be superior in bioactivity when compared to sTNF plus 80M2 by approximately 1 order of magnitude. Human TNF-derived muteins with high receptor selectivity have been described (Van, X et al., 1993; Loetscher et al., 1993). We were questioning whether receptor selectivity would be preserved after coupling of the respective receptor-selective CysHisTNF mutants to the beads or whether subtype specificity would be overcome by a positive cooperativity of the surface-bound ligands in terms of an increased avidity. The results shown in Figure 2, however, clearly show that receptor selectivity remains preserved, and therefore TNFR2selective, strongly activating agonists can be produced on the basis of nano/microparticles covalently coupled with the TNFR2-selective mutein CysHisTNF143N/145R. In summary, we have generated TNF-functionalized particles capable to efficiently stimulate TNFR2, thus mimicking the action of membrane-bound TNF. Moreover, these tools can be used for selective and spatially restricted activation of either of the two TNF receptors and allows live imaging of receptor proximal events and apoptosis events. Future areas of application for similar but biodegradable (nano)particles might include the in vivo usage of particles carrying more than a single effector/targeting protein for therapeutic treatment. ACKNOWLEDGMENT

We thank I.-W. von Broen (Knoll AG, Ludwigshafen, Germany) for recombinant TNF, D. Ma¨nnel (University of Regensburg, Germany) for murine fibroblasts, M. Lenardo (NIH, Bethesda) for pFADD-EGFP, H. Wajant (University of Wu¨rzburg, Germany) for pTRAF2-EGFP, and A. Strasser (The Walter and Eliza Hall Institute, Melbourne, Australia) for pEF PGKpuro. We thank Eva Behrle, Nadja Reinhardt, Alexander Ganser, Gudrun Zimmermann, and Syliva Willi for technical assistance. This work was supported by the Bundesministerium fu¨r Bildung und Forschung (BMBF), grant 0312003B, and in part by Deutsche Forschungsgemeinschaft, Sonderforschungsbereich 495, project A4. LITERATURE CITED (1) Aggarwal, B. B. (2003) Signaling pathways of the TNF superfamily: a double-edged sword. Nat. Rev. Immunol. 3, 745-756. (2) Becker, M. L., Remsen, E. E., Pan, D., and Wooley, K. L. (2004) Peptide-derivatized shell-cross-linked nanoparticles. 1. Synthesis and characterization. Bioconjugate Chem. 15, 699-709. (3) Beyaert, R., and Fiers, W. (1994) Molecular mechanisms of tumor necrosis factor-induced cytotoxicity. What we do understand and what we do not. FEBS Lett. 340, 9-16. (4) Bildirici, L., and Rickwood, D. (2001) An investigation into the suitability of silica beads for cell separations based on density perturbation. J. Immunol. Methods 252, 57-62. (5) Black, R. A., Rauch, C. T., Kozlosky, C. J., Peschon, J. J., Slack, J. L., Wolfson, M. F., Castner, B. J., Stocking, K. L., Reddy, P., Srinivasan, S., and et al. (1997) A metalloproteinase disintegrin that releases tumour-necrosis factor- alpha from cells. Nature 385, 729-733. (6) Eggermont, A. M., and ten Hagen, T. L. (2001) Isolated limb perfusion for extremity soft-tissue sarcomas, in-transit metastases, and other unresectable tumors: credits, debits, and future perspectives. Curr. Oncol. Rep. 3, 359-367.

Bryde et al. (7) Gao, X. H., Yang, L. L., Petros, J. A., Marshal, F. F., Simons, J. W., and Nie, S. M. (2005) In vivo molecular and cellular imaging with quantum dots. Curr. Opin. Biotechnol. 16, 6372. (8) Grell, M., and Scheurich, P. (1999) Tumor Necrosis Factors. In Nature Encyclopedia of Life Sciences, Nature Publishing Group, London. (9) Grell, M., Douni, E., Wajant, H., Lohden, M., Clauss, M., Maxeiner, B., Georgopoulos, S., Lesslauer, W., Kollias, G., Pfizenmaier, K., and Scheurich, P. (1995) The transmembrane form of tumor necrosis factor is the prime activating ligand of the 80 kDa tumor necrosis factor receptor. Cell 83, 793802. (10) Grell, M., Scheurich, P., Meager, A., and Pfizenmaier, K. (1993) TR60 and TR80 tumor necrosis factor (TNF)-receptors can independently mediate cytolysis. Lymphokine Cytokine Res. 12, 143-148. (11) Grunwald, I., Rupprecht, I., Schuster, G., and Kloppstech, K. (2003) Identification of guttation fluid proteins: the presence of pathogenesis-related proteins in noninfected barley plants. Physiol. Plant. 119, 192-202. (12) Hurst, G. B., Buchanan, M. V., Foote, L. J., and Kennel, S. J. (1999) Analysis for TNF-alpha using solid-phase affinity capture with radiolabel and MALDI-MS detection. Anal. Chem. 71, 4727-4733. (13) Krippner-Heidenreich, A., Tubing, F., Bryde, S., Willi, S., Zimmermann, G., and Scheurich, P. (2002) Control of receptor-induced signaling complex formation by the kinetics of ligand/receptor interaction. J. Biol. Chem. 277, 44155-44163. (14) Kruger, W., Datta, C., Badbaran, A., Togel, F., Gutensohn, K., Carrero, I., Kroger, N., Janicke, F., and Zander, A. R. (2000) Immunomagnetic tumor cell selectionsimplications for the detection of disseminated cancer cells. Transfusion 40, 1489-1493. (15) Kull, F. C., Jr., Jacobs, S., and Cuatrecasas, P. (1985) Cellular receptor for 125I-labeled tumor necrosis factor: specific binding, affinity labeling, and relationship to sensitivity. Proc. Natl. Acad. Sci. U.S.A. 82, 5756-5760. (16) Lockman, P. R., Mumper, R. J., Khan, M. A., and Allen, D. D. (2002) Nanoparticle technology for drug delivery across the blood-brain barrier. Drug Dev. Ind. Pharm. 28, 1-13. (17) Locksley, R. M., Killeen, N., and Lenardo, M. J. (2001) The TNF and TNF receptor superfamilies: integrating mammalian biology. Cell 104, 487-501. (18) Loetscher, H., Stueber, D., Banner, D., Mackay, F., and Lesslauer, W. (1993) Human tumor necrosis factor alpha (TNF alpha) mutants with exclusive specificity for the 55kDa or 75-kDa TNF receptors. J. Biol. Chem. 268, 2635026357. (19) Maruyama, A., Ishihara, T., Kim, J. S., Kim, S. W., and Akaike, T. (1997) Nanoparticle DNA carrier with poly(Llysine) grafted polysaccharide copolymer and poly(D,L-lactic acid) 162. Bioconjugate Chem. 8, 735-742. (20) Sandhu, K. K., McIntosh, C. M., Simard, J. M., Smith, S. W., and Rotello, V. M. (2002) Gold nanoparticle-mediated transfection of mammalian cells. Bioconjugate Chem. 13, 3-6. (21) Scheurich, P., Ucer, U., Kronke, M., and Pfizenmaier, K. (1986) Quantification and characterization of high-affinity membrane receptors for tumor necrosis factor on human leukemic cell lines. Int. J. Cancer 38, 127-133. (22) Schiestel, T., Brunner, H., and Tovar, G. E. (2004) Controlled surface functionalization of silica nanospheres by covalent conjugation reactions and preparation of highdensity streptavidin nanoparticles. J. Nanosci. Nanotechnol. 4, 504-511. (23) Stober, W., Fink, A., and Bohn, E. (1968) Controlled Growth of Monodisperse Silica Spheres in Micron Size Range. J. Colloid Interface Sci. 26, 62-72. (24) Van, O., X., Vandenabeele, P., Everaerdt, B., Loetscher, H., Gentz, R., Brockhaus, M., Lesslauer, W., Tavernier, J., Brouckaert, P., and Fiers, W. (1993) Human TNF mutants with selective activity on the p55 receptor. Nature 361, 266269. (25) Verveer, P. J., Wouters, F. S., Reynolds, A. R., and Bastiaens, P. I. (2000) Quantitative imaging of lateral ErbB1

Bioconjugate Chem., Vol. 16, No. 6, 2005 1467

Covalently Immobilized Tumor Necrosis Factor receptor signal propagation in the plasma membrane. Science 290, 1567-1570. (26) Wajant, H., Pfizenmaier, K., and Scheurich, P. (2003) Tumor necrosis factor signaling. Cell Death Differ. 10, 4565. (27) Weiss, T., Grell, M., Hessabi, B., Bourteele, S., Muller, G., Scheurich, P., and Wajant, H. (1997) Enhancement of TNF

receptor p60-mediated cytotoxicity by TNF receptor p80: requirement of the TNF receptor-associated factor-2 binding site. J. Immunol. 158, 2398-2404. (28) Willard, D. M. (2003) Nanoparticles in bioanalytics. Anal. Bioanal. Chem. 376, 284-286.

BC0501810