| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 26, 27410-27421, June 25, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
¶¶¶¶¶¶¶¶¶||
From the
responsif GmbH and november AG, 91056 Erlangen, Germany
Received for publication, December 12, 2003 , and in revised form, April 20, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Other encapsulation strategies, like the liposome technology, have limited applicability for the delivery of low molecular weight substances such as methotrexate (MTX),1 because of leakage of enclosed but not covalently bound molecules (5, 6). For gene delivery, viral and non-viral vectors are used almost exclusively as vehicles where DNA is either not specifically encapsulated or covalently bound to the surface of the vector (7). The major techniques for loading viral or non-viral vectors with DNA fragments or low molecular weight substances are assembly in the presence of high concentrations, (8–10), application of osmotic shock (11, 12), or sonication (13).
The major advantage of the enclosure of macromolecules, especially proteins, is the protection against enzymatic cleavage. Recently, an anchoring technique to achieve a directed localization to the inward-oriented side of the capsoid-forming subunits prior to assembly has been described. Schmidt et al. (14) employed WW domains, fused to the inward-oriented side of VP1, as small modules interacting specifically with prolinerich regions of the molecules to be encapsulated.
In this study, we established a new method to efficiently encapsulate proteins and low molecular weight substances by exploiting the ability of VP2 to bind tightly into the inward-facing VP1 pentamer cavity (15). We used a conserved stretch of 47 amino acids of VP2 for the attachment of model substances like green fluorescent protein (GFP) and MTX, allowing efficient encapsulation of both substances in VP1 capsoids. Because of the open issue of the hydrophobic interplay between VP1 and VP2 molecules, we analyzed two VP2 derivates for their capacity to encapsulate attached ligands (GFP or MTX). It should be noted that both VP2 derivates have the same affinity for binding to VP1-pentamers.
Because of the fixed ratio of one VP2 binding site per VP1 pentamer, the amount of encapsulated molecules was well reproducible, and their concentrations could thus be standardized, which was essential for defined delivery of MTX in cell culture experiments. GFP fluorescence allows easy monitoring of the encapsulation efficiency during capsoid production, as well as of the cellular uptake. MTX is a well known antifolate used in tumor therapy, which is usually taken up into cells by the reduced folate carrier and converted to the intracellularly persistent MTX polyglutamates. MTX and MTX polyglutamates inhibit the folate-dependent enzymes dihydrofolate reductase and thymidylate synthase, thus abrogating DNA synthesis and cell division and resulting in cell death (16). A major mechanism of resistance to MTX (17) is a transport deficiency of the reduced folate carrier. We wanted to find out whether MTX, when endocytosed inside a macromolecular carrier-like capsoid and delivered to a different cell compartment, exhibits toxicity in the same way as the unconjugated drug and is able to overcome MTX transport resistance of cells.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-D-thiogalactoside (IPTG) was obtained from peqlab (Erlangen, Germany), and kanamycin sulfate was obtained from Calbiochem-Novabiochem. Amino acid derivatives were purchased from Calbiochem-Novabiochem AG, methotrexate, 2-mercaptosulfonic acid, ethanethiol, O-benzotriazol-1-yl-N,N,N',N'-tetramethyluronium tetrafluoroborat (TBTU), 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimid (EDCI), piperidin, diisopropylethylamin (DIPEA), Me2SO, and N,N-dimethylformamide (DMF) from Sigma-Aldrich. DNase, water, and acetonitrile were obtained from Merck. Restriction endonucleases were obtained from New England Biolabs or from MBI Fermentas, and oligonucleotide primers were synthesized by ThermoHybaid. Cloning and Vector Construction—The complete mouse polyoma virus VP1 cDNA, obtained as an EcoRI-SdaI fragment from expression vector pHB17/6 (11), was subcloned into the corresponding restriction sites of Litmus28 (New England Biolabs). After the addition of an NdeI site comprising the start codon of VP1 and removal of a second endogenous NdeI site by site-directed mutagenesis with a modified QuikChangeTM protocol (Stratagene), the VP1 cDNA was subcloned as an NdeI-BamHI fragment into the expression vector pSG1 (a pBR322 derivative containing an IPTG-inducible tac promoter), resulting in the final vector pSG1/VP1 used for expression of VP1. For construction of the expression vector pTXB1/GFP-VP2A1, the GFP-cDNA was amplified by PCR from pEGFP-N1 (BD Clontech) with primers generating a 5'-NdeI site comprising the start codon and 3'-SapI and EcoRI sites, and cloned into NdeI and EcoRI sites of pTXB1 (New England Biolabs). The resulting construct, termed pTXB1/C-GFP, was used for the fusion of the sequence coding for VP2(B) between the C terminus of GFP and the N terminus of the Mxe intein-chitin-binding domain (CBD) affinity tag by PCR-amplification of VP2(B) from vector pHB1 (11). The primers used for PCR amplification of VP2(B) generated inversely arranged terminal SapI sites, allowing directional cloning of the product into the corresponding restriction sites of the vector. The VP2(A) variant was generated by site-directed mutagenesis of VP2(B) as described above. The expression vector pTXB1/VP2A8 was obtained by PCR-amplification of the sequence coding for VP2(A), with primers generating 5'-NdeI and 3'-SapI sites, and subsequent cloning into the corresponding restriction sites of vector pTXB1.
Preparation of GFP-VP2(A)—GFP-VP2(A) was expressed as an N-terminal fusion with the Mxe intein/CBD from expression vector pTXB1/GFP-VP2A1 in E. coli BL21(DE3) cells grown at 37 °C in LB medium containing 100 µg/ml ampicillin. The expression of GFP-VP2(A) was induced at an A600 of 0.6–1.0 by the addition of IPTG to a final concentration of 1 mM, and cultivation was continued for 5 h. After harvest, the cells were resuspended in 200 ml of buffer MxeL (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 1 mM EDTA) including 10 µl of benzonase (25 units/µl; Merck) and broken by five passages through a Gaulin LAB 1000 homogenizer at 800 bar. Cell debris was removed by centrifugation at 70,000 x g and 4 °C for 1 h; the clear supernatant was supplied with 10 ml of chitin beads equilibrated in MxeL and incubated for 2 h at 4 °C with constant agitation. After a 10-min centrifugation at 3000 x g, the slurry was transferred to a chromatography column and washed overnight with buffer MxeW1 (20 mM Tris-HCl, pH 8.0, 0.5 M guanidinium-HCl, 1 M NaCl, 0.1% Triton X-100, 1 mM EDTA) at 4 °C and a constant flow rate of 0.2 ml/min. After washing the column with 50 ml MxeW2 (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA), the gel bed was flushed with MxeC buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 50 mM DTT) and incubated at 37 °C for 3 h to ensure complete cleavage of the Mxe intein-CBD moiety. The free GFP-VP2(A) fusion protein was eluted with MxeC buffer, the peak fractions were pooled and concentrated with Centriprep-10 centrifugal concentrators (Millipore). Finally, the protein solution was incubated for another 3 h at 25 °C to allow the complete formation of the GFP chromophore. Homogeneity of the protein was analyzed by SDS-PAGE and mass spectrometry, and the concentration was determined by UV-visible spectroscopy.
Purification of Recombinant Polyoma VP1—E. coli BL21 cells were grown at 37 °C in LB medium containing 70 µg/ml kanamycin. Expression of VP1 was induced by the addition of 1 mM IPTG. After continued cultivation for 5 h, the cells were harvested, resuspended in P-A buffer (20 mM Tris/HCl, 2 mM EDTA, 6 mM DTT, 5% glycerol, pH 8.5) and lysed at 800 bar using a Gaulin LAB 1000 homogenizer. Cell debris was removed by centrifugation at 15,000 x g for 2 h and 4 °C, and VP1 was captured from the supernatant by an anion-exchange column (Poros, Applied Biosystems) previously equilibrated with P-A buffer. After elution of VP1 from the column by a linear gradient of NaCl from 5 mM to 1 M, the protein was dialyzed against QS-A buffer (20 mM ethanol amine, 2 mM EDTA, 6 mM DTT, 5% glycerol, pH 10.5) overnight. The VP1-solution was applied to a Q-Sepharose column (Amersham Biosciences) and eluted again by a linear gradient of NaCl to 1 M. Afterward, the protein was dialyzed against L1 buffer (50 mM sodium phosphate, 2 mM EDTA, 6 mM DTT, 150 mM NaCl, 5% glycerol, pH 7.0). The final step of the VP1 purification was a size-exclusion chromatography (SEC) by which aggregates of VP1 were removed. Homogeneity of the protein was checked by SDS-PAGE, photon correlation spectroscopy (PCS) and mass spectrometry. Protein concentration was determined by UV-visible spectroscopy.
Expression and Purification of VP2-SEt—VP2-SEt was expressed as an N-terminal fusion with the Mxe intein/CBD from expression vector pTXB1/VP2A8 in E. coli BL21(DE3) as described for GFP-VP2(A). Cells were harvested and the pellets were stored at -80 °C. After thawing the cell pellets (which were obtained from 12 liters of culture volume) on ice, cells were resuspended in 125 ml of buffer A (20 mM HEPES, 0.5 M NaCl, 1 mM EDTA, 10 mg/ml DNase, pH 6.8) and broken by five passages through a Gaulin LAB 1000 homogenizer at 800 bar. The resulting suspension was centrifuged at 39,000 x g for 45 min, the clarified supernatant was loaded onto a chitin column, washed with 10 column volumes of buffer B (20 mM HEPES, 0.5 M NaCl, 1 mM EDTA, 0.1% Triton X-100, pH 6.8) and cleaved with buffer C (2% (v/v) ethanethiol, 50 mM sodium phosphate, 200 mM sodium chloride, 1 mM EDTA, pH 6.8) overnight at room temperature. After elution according to the instruction manual, the protein was isolated by preparative high performance liquid chromatography (HPLC) (Agilent SB C18 column; for solvent system, see above). In an HPLC-MS reanalysis of the product fraction, VP2-SEt was found to be >98% pure.
Synthesis of Dipeptide Linker—MTX-Boc-Cys(Trt)-Lys(Fmoc)-OMe was synthesized according to a standard TBTU-peptide coupling protocol. The product could be isolated quite pure from the reaction mixture by precipitation with water, filtration, and lyophilization. Electrospray mass spectrometry (ESI-MS) calculated m/z for C49H53N3O7S [M+Na]+ = 851.0; the found m/z = 850.5. Deprotection of the lysine side chain was accomplished by treatment of 50 mg Boc-Cys(Trt)-Lys(Fmoc)-OMe with 100 µl of 20% piperidin in DMF at room temperature for 20 min. The product was purified by solid phase extraction on an RP-C18-cartridge with H2O/acetonitrile and then lyophilized. After removal of the solvents, the raw product Boc-Cys(Trt)-Lys-OMe was obtained in quantitative yield as a colorless amorphous powder. ESI-MS calculated m/z for C34H43N3O5S [M+H]+ = 606.8; the found m/z = 806.2. A 1.5 molar excess of MTX (23.3 mg) was pre-activated with 8.5 mg of EDCI-HCl and 15.3 µl of DIPEA in 400 µl of DMF for 5 min and then added to a solution of 18.0 mg Boc-Cys(Trt)-Lys-OMe in 400 µl of DMF. After 30 min of incubation at room temperature, the reaction was quenched by the addition of 10 ml of H2O. The product was centrifuged (20,000 x g for 1 min), purified by preparative reverse-phase (RP)-HPLC and lyophilized to give Boc-Cys(Trt)-Lys(MTX)-OMe as a yellow amorphous powder. The yield was 28.4 mg (95%); ESI-MS calculated m/z for C54H63N11O9S [M+H]+ = 1043.2; found m/z = 1043.3. Both protecting groups of the cysteine were removed by treatment with TFA:EtSH:H2O: TIPS (50:10:4:2) for 5 h at room temperature. The deprotected MTX-dipeptide conjugate was precipitated with ice-cold diethylether. After centrifugation, the raw linker-MTX conjugate was isolated as an orange amorphous solid. ESI-MS calculated m/z for C30H41N11O7S [M+H]+ = 700.8; found m/z = 700.3 Da.
Synthesis of VP2-Cys-Lys(MTX)-OMe—15.3 mg of VP2-SEt and a 3-fold molar excess (6.9 mg) of Cys-Lys(MTX)-OMe were dissolved in 2.7 ml of 400 mM phosphate buffer, containing 10 mM EDTA and 10% (v/v) Me2SO. One equivalent of tris-carboxyethyl phosphine was added to establish reductive conditions, and 2-mercaptosulfonic acid (0.1 equivalents) was added as a catalyst. The reaction was monitored by analytical HPLC and showed a quantitative derivatization of VP2-SEt after 18 h. After final reduction by the addition of 2 equivalents of DTT at room temperature for 15 min, the solvents were removed in vacuum, and the raw product was isolated by preparative HPLC (Agilent SB C18 column; for solvent system, see above). The HPLC-MS analysis showed a purity of >98% for the obtained VP2-Cys-Lys(MTX)-conjugate.
Formation of Capsoids from VP1 Pentamers and VP2 Anchor Constructs Containing GFP or MTX—For the assembly of capsoids, DTT was removed from the VP1 pentamer solution with a desalting column (HiPrep 26/10, Amersham Biosciences; column volume = 53 ml, flow rate = 2 ml/min) equilibrated with KB1 buffer (50 mM sodium phosphate, 150 mM NaCl, 2 mM EDTA, 5% glycerol, pH 6.8). Structural integrity and concentration of the desalted pentamers were determined by PCS and UV-visible spectroscopy. VP1 pentamers were incubated with a 7-fold molar excess of GFP-VP2(A) or with a 6-fold molar excess of VP2(B)-MTX for 1 h at room temperature, respectively. Afterward, assembly of the pentamers was initiated by adding KB2 buffer (10 mM Tris-HCl, 150 mM NaCl, 5% glycerol, 3 M (NH4)2SO4, pH 8.0) to a final ammonium sulfate concentration of 250 mM. After incubation for 30 min at 25 °C, the formed capsoids were oxidized by the addition of KB3 (400 mM GSSG in KB2) to a final GSSG concentration of 7.2 mM. Afterward, the protein was dialyzed overnight at 25 °C against phosphate-buffered saline (PBS) (Biochrom KG) containing 0.7 mM CaCl2. The excess of non-assembled VP2-anchor construct was removed by SEC using a Superdex 200HR 10/30 column (Amersham Biosciences; flow rate = 0.5 ml). A precipitation with polyethylene glycol (10% PEG 3000, Sigma-Aldrich) was performed to concentrate the capsoids for cell-culture experiments.
PCS—Particle sizing was performed at 25 °C using a high performance particle sizer (ALV-NIBS). The particle size was calculated by the autocorrelation function of the ALV sizer software.
Immunoprecipitation—3.1 µM VP1 pentamers were co-incubated for 4 h at 25 °C with a 10-fold molar excess of GFP-VP2(A) or GFP, respectively. VP1 pentamers, GFP, and GFP-VP2(A) were also incubated and served as controls. Protein-A-Sepharose (Amersham Biosciences) was equilibrated with KB1 buffer and monoclonal antibodies (mab) directed against a conformational determinant of VP1-specific pentamers and appropriate capsoids (mab17 was kindly provided by Dr. M. Pawlita, Heidelberg, Germany) or GFP were coupled for 1 h at 25 °C. After washing the Sepharose two times with IP wash buffer (PBS, 0.1% Triton X-100, pH 7.4), the above described protein samples were added and incubated for 1 h at 25 °C. Unbound protein was removed by washing three times with IP wash buffer. Subsequently, the samples were analyzed by reducing SDS-10% PAGE and Western-blotting.
Fluorescence Spectroscopy—To investigate the interaction between VP1 pentamers and the GFP-VP2(A) anchor construct, GFP fluorescence polarization measurements were carried out. GFP-VP2(A) was incubated with an increasing amount of VP1-pentamers for 2 h at 25 °C. GFP fluorescence polarization was measured at 510 nm (excitation = 488 nm) for 10 min in a thermostatically controlled cell using a Fluoromax-3 spectrofluorimeter with autopolarizer (Jobin Yvon, Bensheim, Germany). The average signal was plotted against the VP1 pentamer concentration [M] and the dissociation constant (Kd) calculated according to Equations 1, 2, 3,
 | (Eq. 1) |
 | (Eq. 2) |
 | (Eq. 3) |
High Performance Liquid Chromatography (HPLC)-ESI-MS—HPLC analyses were performed on an Agilent Technologies 1100 MSD Series system equipped with diode array detector and coupled single-quadrupole MS. HPLC solvents consisted of 1% acetonitrile in H2O containing 0.1% trifluoroacetic acid (solvent A) and 90% acetonitrile in H2O containing 0.1% trifluoroacetic acid (solvent B). A Polymerlabs PLRP-S (300 Å, 4.6 x 150 mm) column was used at a flow rate of 0.8 ml/min, and the column temperature was maintained at 25 °C. The gradient started with a solvent composition of 80% solvent A and 20% solvent B, followed by a linear gradient in 20 min to 0% solvent A and 100% solvent B, after which the column was re-equilibrated.
Cell Cultivation—Swiss mouse fibroblasts (Swiss 3T3) were grown in Dulbecco's modified Eagle's medium supplemented with 2 mM glutamine (Pan Biotech) and 10% fetal bovine serum (Invitrogen). The human T cell leukemia line CCRF-CEM was cultivated in RPMI medium 1640 supplemented with 2.5% human serum, 4 mM glutamine, 25 mM HEPES buffer (all obtained from Pan Biotech), and 200 µg/ml human serum albumin (Pharma Dessau GmbH). The MTX transport-resistant CCRF-CEM/MTX cells (18) were cultivated in the same medium with additional 1 µM MTX. For experiments, the cells were grown in medium without MTX for seven passages. All cell lines tested negative for mycoplasma contamination and grown at 37 °C in a 5% CO2-air humidified incubator.
Fluorescence Immunocytochemistry with Swiss 3T3 Cells—Swiss 3T3 cells were grown on coverslips for 2 days. For the attachment of GFP-loaded capsoids or VP2-GFP(A) anchor to the cell surface, cells were incubated in Dulbecco's modified Eagle's medium containing 10 µg/ml GFP-loaded capsoids (
4 x 106 particles per cell) or 10 µg/ml VP2-GFP(A) anchor, respectively. After 1 h of incubation at 0 °C, some samples were immediately prepared for immunocytochemistry (attachment, 0 h) by washing the cells in ice-cold medium for the removal of unbound GFP-loaded capsoids or VP2-GFP(A) anchor. For the uptake of GFP-loaded capsoids or VP2-GFP(A) anchor after attachment, some samples were washed in pre-warmed Dulbecco's modified Eagle's medium, transferred to 37 °C, and incubated for 0.5, 1, and 3 h, respectively. After the indicated time points, all samples were prepared for visualization of VP1 by immunofluorescence. The following steps were performed at room temperature. Cells were fixed with 1% paraformaldehyde (Sigma) for 30 min, rinsed twice with PBS, and incubated with 80 mM glycine (Sigma) in PBS (pH 8) for 10 min to block free aldehyde functions. After washing twice with PBS, cells were permeabilized with 0.1% Triton X-100 (Sigma) in PBS for 4 min for intracellular staining of VP1, or they remained unpermeabilized for VP1 staining on the cell surface by incubating with PBS only. All samples were rinsed twice with PBS. Blocking of unspecific binding was performed with 1% bovine serum albumin in PBS (albumin fraction V, Roth, Germany) for 15 min, followed by incubation with a mouse anti-VP1 mab13 (undiluted hybridoma supernatant) for 1 h, washing twice with PBS, and labeling with a Cy3-rabbit-anti-mouse polyclonal antibody (Dianova) diluted 1:300 in 1% bovine serum albumin in PBS. The anti-VP1 mab13 (kindly provided by Dr. M. Pawlita) recognizes a conformational determinant of VP1 pentamers as well as VP1 capsoids. As shown by SDS-PAGE and subsequent Western-blotting analysis (data not shown), denatured VP1 protein could not be detected with anti-VP1 mab13. It should be noted that anti-VP1 mab13 and mab17 could recognize similar epitopes of particulate VP1-structures.2 Samples were rinsed twice with PBS, and cell nuclei were stained for 5 min with 0.2 µg/ml 4',6-diamidino-2-phenylindole (DAPI) in PBS. After washing with PBS and H2O, coverslips were mounted on microscope slides with anti-fading solution (6 g of glycerol, 2.4 g of MowiolTM; Calbiochem) and 0.2% 1,4-diazabicyclo-[2.2.2]octane in 0.2 M Tris-HCl, pH 8.5. Two control conditions were examined: incubation of cells with VP2-GFP(A) anchor and immunofluorescence staining with omission of primary antibody as negative control. None of these conditions resulted in cell staining. GFP fluorescence and the immunofluorescence signal of VP1/Cy3 were observed with a Zeiss Axioplan 2 imaging fluorescence microscope (Zeiss, Göttingen, Germany) and fluorescence filters F11–013 (DAPI), F41–054 (GFP) and F41–007 (Cy3) (AHF Analysentechnik). Pictures were taken with a digital camera (Diagnostic Instruments model 2.3.0), followed by digital image processing with Metamorph software, version 4.6 (Visitron Systems).
Dose-response Assays with CCRF-CEM and CCRF-CEM/MTX Cells—Exponentially growing CCRF-CEM cells were seeded in 96-well plates with 1.5 x 104 cells in 200 µl of RPMI per well containing different concentrations of MTX-loaded capsoids, MTX (USP grade; Sigma), capsoids (control), VP2(B)-MTX anchor (control), or medium only (untreated control). After 24, 48, and 72 h, a CalceinTM viability assay was performed. 100 µl of medium per well were aspirated, and 100 µl of a 20 µM solution of CalceinTM (Molecular Probes Europe) in Dulbecco's phosphate buffered saline (Biochrom KG) were added to a final CalceinTM concentration of 10 µM. Cells were incubated for 1 h at 37 °C, and fluorescence was measured at 485 nm/528 nm (excitation/emission) with an FLx800T BIE-microplate reader (Bio-Tek Instruments GmbH, Bad Friedrichshall, Germany). CCRF-CEM/MTX cells were grown in medium without MTX for seven passages and seeded in 96-well plates with 1.5 x 104 cells in 200 µl of RPMI medium 1640 per well containing different concentrations of MTX-loaded capsoids, MTX, capsoids (control), VP2(B)-MTX anchor (control), or medium only (untreated control). After 48 h of incubation, a CalceinTM viability assay was performed as described above.
Transmission Electron Microscopy (TEM) of Capsoids—Capsoids, GFP-loaded capsoids, or MTX-loaded capsoids were attached to formvar carbon-coated grids (300 mesh; Plano GmbH) and contrasted by the single-droplet negative-staining technique modified after Harris (19). Briefly, grids were wetted for 10 s on a 20-µl droplet of an aqueous solution of 0.1% poly-L-lysine hydrobromide (Sigma), washed on a droplet of 20-µl H2O, and transferred to a droplet of 20-µl PBS containing 100 µg/ml capsoids, GFP-loaded capsoids, or MTX-loaded capsoids. After 2 min of attachment, grids were washed twice with 20 µl of H2O and negative stained for 15 s in 20 µl of an aqueous solution of 5% uranyl acetate (Plano GmbH). After drying for 5 min, samples were ready for immediate TEM study on a Zeiss LEO 906E (LEO Elektronenmikroskopie GmbH) operating at 80 kV.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
VP2 Anchor Variants—Because of the hydrophobic character of the anchor molecule, we chose different VP2 variants for the expression of the GFP-VP2(A) fusion protein and covalent linkage of MTX to the VP2(B) anchor, to obtain soluble proteins capable of interacting with VP1 pentamers. Variant VP2(A) is based on the VP2 sequence found in the mouse polyoma virus Py-2a strain (22). Variant VP2(B) was derived from polyoma virus strain 3. Both anchor molecules differ only in the PGGA or QVVS motifs, corresponding to positions 276–279 of VP2 (Fig. 1). The x-ray structure of VP1 interacting with VP2 was performed with the VP2-derivate containing the QVVS motif (22). In contrast, our original VP2-derivate with PGGA residues showed non-conservative amino acid alterations at the same positions (276–279). Therefore, and due to the open issue of the hydrophobic interplay between VP1 and VP2 molecules, we analyzed both VP2-derivates for their capacity to encapsulate attached ligands (GFP or MTX). It should be noted that both VP2-derivates have the same affinity in binding to VP1-pentamers.
|
Expression and Purification of VP2(B) Anchor and Conjugation to MTX—A multi-step synthesis strategy to conjugate MTX covalently to the VP2(B) anchor was developed. To prevent interference of the capsoid assembly by VP2(B)-MTX, MTX was coupled to the C terminus of the VP2(B) anchor molecule, which is positioned at the sterically free accessible site of the VP1 pentamer (22). This conjugation was realized by expressed protein ligation (23). The required VP2(B)-ethylthioester was generated from the recombinant fusion protein VP2(B)-intein-CBD by induction of the intein cleavage with ethanethiol. It was sufficiently soluble and stable to hydrolysis, which enabled isolation by preparative HPLC. The same procedure was performed with the VP2(A) anchor molecule, but interestingly, the resulting VP2(A)-ethylthioester was only poorly soluble and could not be isolated. The functionalization of the dicarbonic acid MTX with the free amino acid cysteine affords a diamino linker unit, for which lysine was chosen (Fig. 1). Selective deprotection of the 
-amino group in Boc-Cys(Trt)-Lys(Fmoc)-OMe gave Boc-Cys(Trt)-Lys-OMe. Because the application of standard peptide-coupling protocols leads to the formation of multiple undesirable side products, it was necessary to develop a special protocol for the subsequent coupling of MTX to the free -amino side chain group. Variation of reaction parameters and coupling agents (O-benzotriazol-1-yl-N,N,N',N'-tetramethyluronium hexafluorophosphate, O-benzotriazol-1-yl-N,N,N',N'-tetramethyluronium tetrafluoroborat, N-hydroxysuccinimide active ester, 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimid, 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimid-hydroxybenzotriazole) revealed that selective mono-functionalization of MTX is achieved by the following parameters: (i) MTX in 1.5 molar excess; (ii) EDCI as coupling reagent; (iii) DIPEA in a limited amount (2 equivalents); and (iv) quenching of the reaction after 30 min by the addition of H2O. Consequent application of these parameters reduced the formation of side products to a minimum. Unmodified or bismodified MTX, as well as other common side products, were not detectable, and the desired product was isolated by preparative HPLC with a purity of >98%. Finally, the N/S-protecting groups of Boc-Cys(Trt)-Lys(MTX)-OMe were cleaved in one step with TFA:EtSH:H2O:TIPS (50:10:4:2). Diethylether precipitation of the reaction mixture gave the raw product, which was used directly for the recombinant protein ligation. For the ligation reaction, VP2(B)-SEt was incubated with 3 equivalents of Cys-Lys(MTX)-OMe and a catalytic amount of 2-mercaptosulfonic acid in 10% DMF. After 16 h, the conjugation of VP2(B) was quantitative, and VP2(B)-Cys-Lys(MTX)-OMe (VP2(B)-MTX) was isolated by preparative HPLC with a purity of >98%.
Interaction of the VP2 Anchor Constructs with the VP1 Pentamer—The outer shell of the polyomavirus capsid is composed of 360 VP1 monomers, arranged in 72 pentamers on an icosahedral lattice (1). VP2 binds to the hydrophobic pocket of the VP1 pentamer on the inward-facing side (15). To evaluate whether the C-terminal fragment of VP2 we used as an anchor molecule is able to interact with a VP1 pentamer, even when a protein-like GFP is fused to the N terminus, we performed gel filtration experiments. Free GFP-VP2(A) was clearly separated from the VP1 pentamer fraction. After incubation of GFP-VP2(A) with an equimolar amount of VP1 pentamer for 2 h, both proteins co-eluted from the column (data not shown). To assess the specificity of the interaction between the VP1 pentamer and the VP2 anchor, we performed immunoprecipitation experiments. VP1 was co-incubated either with GFP-VP2(A) or GFP as a control, followed by immunoprecipitation with monoclonal antibodies against VP1 or GFP, respectively. SDS-PAGE and Western-blot analysis of the precipitated samples showed that GFP-VP2(A) and VP1 co-precipitated in both cases, revealing a tight interaction between the VP1 pentamer and the anchor (Fig. 2). In contrast, no co-precipitation was observed after co-incubation of GFP and VP1, indicating that the attachment of GFP-VP2(A) to VP1 pentamers is highly specific and occurs by means of the C-terminal VP2 anchor. A corresponding fusion protein of GFP and the VP2(B) anchor gave similar results in analogous experiments (data not shown).
|
|
Encapsulation of VP2 Anchor Constructs—To achieve the encapsulation of VP2 anchor-bound molecules into VP1 capsoids, a robust method allowing both the attachment of the VP2 anchor to the pentamer and the subsequent capsoid assembly within a short time frame was developed. For capsoid formation, DTT was removed from the pentamer solution to enable the formation of intra-pentameric disulfide bonds, which stabilize assembled VP1 capsoids (14, 20, 21, 24). In contrast to the assembly method described by Stehle et al. (20, 25), we induced the assembly process by the addition of high salt buffer to the pentamer solution. As shown in Fig. 4A, this approach leads to capsoid formation in a concentration-dependent way. At the lowest pentamer concentration examined (0.9 µM), the pentamers associated with a half-time of 170 s, and the entire process was completed after 5 min. Increase of the protein concentration resulted in a faster assembly reaction. The course of assembly was also followed by PCS measurement, which showed a particle size of 8–9 nm for VP1 pentamers at the start of the reaction (Fig. 4B) and a particle diameter of 35 nm for the resulting capsoids (Fig. 4C).
|
|
Immunoprecipitation experiments confirmed that, apart from encapsulated GFP-VP2(A), additional molecules either occurred freely or unspecifically bound to the surface of the VP1 capsoids (data not shown). The gel filtration experiments suggested a rather weak interaction that was easy to disrupt during the chromatography process. After re-chromatography or extended dialysis of the capsoid fraction, no more free anchor was detectable by immunoprecipitation (data not shown).
Quantification of the encapsulated MTX by UV/visible spectroscopy and gel filtration gave an average amount of 462 anchor molecules per VP1 capsoid (data not shown), whereas free VP2(B)-MTX was not observed. In comparison, unconjugated MTX could not be encapsulated during VP1 assembly because of the small molecule size and the cavity-containing morphology of the final capsoids (data not shown).
Electron microscopy of the loaded capsoids revealed a regular morphology, showing no significant differences to empty particles (Fig. 6, A–C). Nevertheless, the capsoid GFP-VP2(A) showed minor particle formation in comparison to empty virus-like particles, and those that have encapsulated MTX ligands suggested that the lower co-assembling efficiency within a given time-frame may depend upon the charge distribution and the size of appropriate ligands (Fig. 6C, GFP versus MTX). Taken together, we have shown that the VP2-derived anchor molecule allows highly efficient encapsulation of GFP or MTX in recombinantly produced VP1 capsoids.
|
|
|
Methotrexate Delivery into Cells via Capsoids—MTX encapsulated in mouse polyoma capsoids was delivered into MTX-sensitive CCRF-CEM cells. Two independently performed dose response assays are shown (Fig. 9, A and B). Free MTX in a concentration range of 1–40 µM was cytotoxic on these cells in a time- but not a concentration-dependent manner as shown in Fig. 10A. Concentration dependence was only observed between 0.02 and 1 µM MTX. The median survival after 48 h of treatment with 0.02 µM MTX was 47.4 and 23.2% for 1 µM MTX. MTX concentrations of 10, 20, and 40 µM did not increase effectiveness and led to a median survival after 48 h of 19.8, 19.9., and 18.4%, respectively. Differentiation between growth arrest and cytotoxicity was achieved by microscopic observation, because the CalceinTM assay is a measure for cell viability only. Cell populations treated with 0.02 µM MTX showed a clearly reduced density but only a few more damaged or dead cells, as compared with the untreated control cells over the whole time period. Cells treated with cytotoxic MTX-concentrations above 1 µM appeared predominantly damaged or dead.
|
|
Empty capsoids with a concentration of 1.8 mg/ml, corresponding to the capsoid concentration of MTX-loaded capsoids containing 40 µM MTX, did not exhibit cytotoxic effects: no influence was seen on cell survival after 24 h, a low impairment resulting in median survival of 80.1% after 48 h and a recovery to 99.3% survival after 72 h. The VP2(B)-MTX conjugate with 117 µM MTX (total protein content 0.7 mg/ml) had no influence on the cells after 24 and 48 h, and the median survival after 72 h was only slightly reduced to 90.1%. The ineffectiveness of this high MTX concentration is most probably because of the lack of entry of the drug into the cell because of its conjugation to VP2(B).
Methotrexate delivered via capsoids also circumvented the MTX transport resistance of CCRF-CEM/MTX cells (Fig. 10). When these cells were treated with 0.02 or 1 µM free MTX, the median decrease of survival after 48 h of incubation was only 14.2 or 22.5%, respectively, and thus comparable with the slight growth impairment of 13.1% caused by empty capsoids. In contrast, both 0.02 and 1 µM MTX were cytotoxic in the parental MTX-sensitive cell line CCRF-CEM, causing a survival decrease of 59.0 and 83.1%, respectively. When MTX transport-resistant cells were treated with 1 µM encapsulated MTX, MTX became cytotoxic, causing a 50.4% decrease in survival, which was comparable with the reduction of 67.2% observed with MTX-sensitive cells.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-helix and the VP1 pentamer. Consequently, we decided to employ this fragment of VP2 as core of the anchor molecule. The residues N-terminal of Val-269, which is located at the top of the cavity formed by the VP1 pentamer, could not be resolved in the crystal structure, and the authors concluded that this part of VP2 might be somewhat flexible. We estimated that the VP2-mediated attachment of GFP to the inner surface of the VP1 pentamer would require a distance of at least 20 Å between Val-269 and the GFP termini (located adjacent at one narrow side of the protein) to avoid steric hindrance. This notion favored the fusion of GFP to the C-terminus of the VP2 anchor, as the corresponding Tyr-296 is located at the base of the pentamer. Because the C-terminal part of the resolved VP2 structure approaches the termini of VP1, which are very likely engaged in pentamer-pentamer interactions (20, 21), the fusion of a 27-kDa protein at this site could possibly interfere with capsoid assembly. Therefore, a GFP-fusion to the N terminus of the anchor molecule was considered as the better alternative, but obviously required an additional polypeptide stretch between Val-269 and GFP as a (preferably flexible) spacer. Based upon the average pitch of an -helix (presumed as the worst case for the spacer) of 5.4 Å per winding, we calculated that the inclusion of the 18 adjacent residues N-terminal of Val-269 in the anchor molecule would provide a spacing of at least 27 Å, thus preventing any steric hindrance with capsoid assembly. Despite a highly hydrophobic stretch present in the VP2 anchor, the resulting fusion protein GFP-VP2(A) was solubly expressed, showing that solubility is mainly determined by the larger protein domain. Furthermore, binding of the VP2 anchor to the VP1 pentamer was not hindered by GFP, indicating that the crucial hydrophobic patch of VP2 is still accessible. The strength of the hydrophobic interaction between the anchor and the pentamer is thermodynamically comparable with that between proline-rich sequence motifs and the WW domain derived from the mouse formin-binding protein 11 (FBP11). Schmidt et al. (26) inserted this WW domain into an inward-facing loop of VP1 to bind and encapsulate GFP containing a Pro-Pro-Leu-Pro ligand. The interaction between the GFP-VP2(A) molecule and VP1, which was not observed between free GFP and VP1, proved the physicochemical polarity of the GFP-VP2(A) fusion protein and the specificity of its binding to VP1. The orderly binding of the anchor moiety of GFP-VP2(A) to the inward-oriented cavity of VP1 pentamers was further confirmed by the virtually unchanged assembly into regularly formed capsoids.
For the coupling of the VP2 anchor with a low molecular weight substance such as MTX, another approach was made. Native chemical ligation (NCL) is a well established method for covalent and regioselective binding of various substances to recombinantly expressed proteins (28). This reaction affords a free cysteine residue in the molecule to be coupled. Therefore, the new dipeptide linker Boc-Cys(Trt)-Lys-OMe was used, allowing a universal and rapid modification of carboxylic acids for NCL.
MTX has frequently been coupled to macromolecular carriers such as antibodies or synthetic polymers, using carbodiimides and N-hydroxysuccinimide (29). This active ester method, however, does not only lead to the desired MTX -or 
-isomers, but also to bis-derivatives, MTX oligomers, and several by-products, e.g. N-acylisourea derivatives, because of rearrangement reactions (30). The bis-derivatives especially have to be avoided, because bis(VP2-Cys-Lys-OMe)-MTX would be able to crosslink VP1, presumably causing problems in capsoid assembly. Formation of the described by-products was avoided by modifying a recently published coupling protocol of Riebeseel et al. (31).
In the case of loading substances to the interior of VP1 capsoids, an incubation and binding step precedes capsoid assembly. The thermodynamic and kinetic stability of the binding of the anchor to the VP1 pentamer is a prerequisite for directed encapsulation of proteins and low molecular weight substances and, in the end, for an efficient loading of the capsoids. Aside, the sterical properties of the molecule to be encapsulated directly influence the encapsulation capacity. The interior of the capsoids has a volume of about 7150 nm3, which is enough space for a theoretical loading of 408 molecules of GFP. From this point of view, an amount of encapsulation near 100% should be achievable for the GFP-anchor molecule, assuming that a single capsoid is composed of 72 VP1 pentamers and one anchor binds to one pentamer.
In consideration of the determined dissociation constant of the VP1 pentamer anchor complex, a complete loading of the VP1 capsoid should be achievable employing a 10-fold molar excess of the anchor. However, under these conditions, capsoid formation was not observed, which is presumably an indication for sterical influences and/or an alteration of the VP1 pentamer-VP1 pentamer interaction via an allosteric mechanism. Finally, 64 GFP molecules were enclosed in VP1 capsoids, which is lower than achieved using the VP1-WW system (14), but does not require molecular engineering of VP1 and, therefore, leaves the delivery and cell entry properties of VP1 untouched. Besides loading capsoids with recombinant proteins, the VP2 anchor is distinguished by its capacity to enable the encapsulation of low molecular weight substances, e.g. MTX. Interestingly, in this case, the loading rate of the VP1 capsoids is significantly higher than in the case of the GFP anchor. 462 MTX anchor molecules could be enclosed into the virus-like particles, indicating that besides specific interactions, unspecific hydrophobic interactions must occur because of the increased hydrophobicity of the anchor. Unconjugated MTX was not found to be encapsulated, showing that the conjugation with the VP2 anchor is crucial for the hydrophobic interaction. Because hydrophobicity is found to be the driving force of the anchor-VP1 interaction and because VP1 shows a distinct polarity (charged on the surface, hydrophobic on the inward-facing site; Ref. 22), efficient and directed encapsulation of proteins and low molecular weight substances is ensured.
Beyond the mere delivery of a macromolecule like GFP, we wanted to show the effectiveness of an encapsulated cytostatic drug in the case of intracellular degradation of drug-loaded capsoids after uptake by the cell and release of the drug into the cytoplasm. For this purpose, we chose MTX, because it is an inhibitor of cytoplasmic folate-dependent enzymes. Furthermore, an MTX transport-deficient cell line is available (32) to demonstrate the circumvention of resistance by an alternative uptake route for MTX, as was already shown for an MTX-albumin conjugate with the same cells (33). MTX-albumin was taken up via endocytosis, degraded lysosomally, and the liberated MTX caused cytotoxicity in the MTX transport-resistant cells, which are usually unaffected by free MTX at the same concentration.
We also tried to gain information about the intracellular fate of GFP- and MTX-loaded capsoids. An approach similar to our model was developed by Guenther et al. (34), who produced VP1 virus-like particles containing a WW domain fused to the N terminus of VP1 for ligand binding. The VP1-WW virus-like particles containing GFP were efficiently delivered into NIH 3T3 cells, and both capsoids and GFP were found in endocytic vesicles. But only weak co-localization with lysosomes was found, making lysosomal degradation unlikely, although at least a partial release of GFP from the capsoids occurred (34).
A study by Richterova at al. (32) showed the entry of VP1 pseudo-capsoids into NIH 3T6 mouse cells in smooth nonclathrin-coated monopinocytic vesicles in the proximity of larger, caveola-like invaginations. They observed intracellular transport of viral particles in larger endosomes, created by fusions of monopinocytic vesicles with caveola-derived vesicles, in which viral particles might have been partially disassembled. They also found an accumulation of VP1 around the nuclear membrane, but no significant VP1 signal in the cell nucleus. They suggested that only a few intact virions entered the cell nucleus, whereas the majority became disassembled and subsequently degraded in the cytoplasm. We assume that this also applies to most of the internalized GFP- and MTX-loaded capsoids in our experiments.
The release of GFP from the capsoids was shown by the increasing separation of the GFP and VP1/Cy3 fluorescence signals into different vesicular structures 3 h after uptake, suggesting a kind of sorting mechanism. Endosomal localization of GFP-containing capsoids could affect anti-VP1 antibody binding because of acidification of the endosomes. Thus, the observed phenomenon of augmented separation of the GFP and VP1/Cy3 fluorescence signals could be also attributed to an impaired intracellular staining procedure. However, we can neither confirm nor exclude that functional GFP, or at least the intact chromatophore of the molecule, is released from vesicles. Although we noticed a diffuse distribution of GFP in the cytoplasm at later time points, these fluorescence signals were hardly distinguishable from the GFP-loaded capsoids eventually remaining on the cell surface. Neither GFP nor VP1/Cy3 fluorescence was found in the nuclei. Delivery of encapsulated molecules into the cytoplasm, however, was clearly demonstrated by the cytotoxic effect of MTX-loaded capsoids. The toxicity was time- and concentration-dependent as expected for an endocytotic-uptake mechanism with subsequent endosomal degradation and sufficient release of MTX into the cytoplasm.
The high loading rate of the capsoids and the presence of VP2 anchor molecules may account for the slight differences in intracellular processing, possibly through subtle changes in the particle surface conformation and/or a contribution of VP2 to viral entry (22). An et al. (35) showed that recombinant virus-like particles assembled from different structural polyomavirus proteins, i.e. VP1, VP1/2, VP1/3, or VP1/2/3, were equally internalized into 3T6 mouse cells and localized in the cytoplasm. The decision about the intracellular fate of the particles might be taken during and/or after internalization, when, depending upon capsoid composition and structure, different intracellular pathways are addressed or initiated. It seems likely that there is no exclusiveness for one particular pathway, but maybe a shift of preferences occurs.
From an immunological point of view, we present here a device which is very suitable for the in vivo delivery of full-length proteins like tumor antigens, because heterologous polyoma capsoids displaying a CD8 T cell epitope were already able to protect inbred mice from lethal challenge with melanoma cells that express this relevant protein antigen (36). For this reason, these capsoid-based immunotherapeutics may offer new opportunities for the treatment of cancer patients.
|
FOOTNOTES |
|---|
¶ These authors contributed equally to this work.
|| To whom correspondence should be addressed: responsif GmbH, Schallershofer Str. 84, D-91056 Erlangen, Germany. Tel.: 49-9131-75088-440; Fax: 49-9131-75088-439; E-mail: hess{at}responsif.de.
1 The abbreviations used are: MTX, methotrexate; GFP, green fluorescent protein; DMF, N,N-dimethylformamide; CBD, chitin-binding domain; DTT, dithiothreitol; PCS, photon correlation spectroscopy; HPLC, high performance liquid chromatography; RP, reverse-phase; PBS, phosphate-buffered saline; SEC, size-exclusion chromatography; mab, monoclonal antibodies; WW, domain of the formin-binding protein II.
2 M. Pawlita, personal communication.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
, 194 µg/ml (0.9 µM VP1 pentamer); 
, 387 µg/ml (1.8 µM pentamer); and 
, 775 µg/ml (3.6 µM pentamer). The assembly process was followed by light-scattering at 25 °C. PCS measurements done at the same temperature at the beginning (B) correspond to VP1 pentamers; measurements done at the end of the capsoid assembly process (C) correspond to VP1 capsoids.
Triton) or with permeabilization by Triton X-100 (D, surface-0h + Triton). Uptake of GFP-loaded capsoids was performed for 1 h at 37 °C, followed by immunofluorescence labeling of VP1 by Cy3 (VP1-Cy3) and staining of cell nuclei with DAPI as described under "Experimental Procedures." In unpermeabilized cells (B and C), VP1-Cy3-fluorescence is restricted to the cell surface (insets b and c), whereas GFP is found in intracellular vesicular structures (B and C). Intracellular GFP and VP1-Cy3-fluorescence occurred in permeabilized cells (E and F). VP1-Cy3-fluorescence (insets e and f) was co-localized with GFP in intracellular vesicular structures.
capsoids (0.9 mg/ml);