ARTICLE pubs.acs.org/Langmuir
Preparation of Metallochelating Microbubbles and Study on Their Site-Specific Interaction with rGFP-HisTag as a Model Protein Robert Lukac,†,^ Zuzana Kauerova,†,^ Josef Masek,† Eliska Bartheldyova,† Pavel Kulich,† Stepan Koudelka,† Zina Korvasova,† Jana Plockova,† Frantisek Papousek,‡ Frantisek Kolar,‡ Roland Schmidt,§ and Jaroslav Turanek*,† †
Department of Pharmacology, Toxicology and Immunotherapy, Veterinary Research Institute, Brno, Czech Republic Institute of Physiology, Academy of Sciences of the Czech Republic, Prague, Czech Republic § Hitachi High-Technologies Europe GmbH, Krefeld, Germany ‡
ABSTRACT: The histidine-metallochelating lipid complex is one of the smallest high affinity binding units used as tools for rapid noncovalent binding of histidine tagged molecules, especially recombinant proteins. The advantage of metallochelating complex over protein-ligand complexes (e.g., streptavidinebiotin, glutathiontransferase-glutathion) consists in its very low immunogenicity, if any. This concept for the construction of surface-modified metallochelating microbubbles was proved with recombinant green fluorescent protein (rGFP) containing 6Histag. This protein is easy to be detected by various fluorescence techniques as flow cytometry and confocal microscopy. Microbubbles (MB) composed of DPPC with various contents of metallochelating lipid DOGS-NTA-Ni were prepared by intensive shaking of the liposome suspension under the atmosphere of sulfur hexafluoride. For this purpose, the instrument 3M ESPE CapMix was used. Various techniques (static light scattering, flow cytometry, and optical microscopy) were compared and used for the measurements of the size distribution of MB. All three methods demonstrated that the prepared MB were homogeneous in their size, and the mean diameter of the MB in various batches was within the range of 2.1-2.8 μm (the size range of 1-10 μm). The presence of large MB (8-10 μm) was marginal. Counting of MB revealed that the average amount of MB prepared of 10 mg of phospholipid equaled approximately 109 MB/mL. Lyophilized MB were prepared with saccharose as a cryoprotectant. These MB were shown to be stable both in vitro (the estimated half-live of the MB in bovine serum at 37 C was 3-7 min) and in vivo (mouse). The stability of the MB was affected by molar content of DOGSNTA-Ni. DPPC-based metallochelating MB provided a clear and very contrast image of the ventricular cavity soon after the injection. Site selective and stable binding of rGFP-HisTag (as a model of His-tagged protein) onto the surface of metallochelating MB was demonstrated by confocal microscopy.
’ INTRODUCTION Lipid-based carrier systems represent drug vehicles composed of physiological lipids such as phospholipids, cholesterol, cholesterolesters, and triglycerides as well as synthetic auxiliary lipids that enable to provide the lipid particles, especially their surface, with a special function. Self-assembling lipid carriers offer many advantages, e.g., controlled release, selective interaction with target cells, and simple manufacture. Lipid-coated gas-filled microbubbles (MB) represent a new class of drug delivery systems with both diagnostic and therapeutic application.1 Considering drug delivery and targeting, the main advantage of microbubble application is a reduction of undesired side effects such as toxicity.2 Furthermore, the diagnostic application of ultrasound imaging using MB has become highly popular because ultrasound is a noninvasive and relatively low-cost tool. It uses portable, real-time imaging equipment and also avoids hazardous ionizing radiation.2,3 r 2011 American Chemical Society
Considering MB themselves, they are small (typically 1-8 μm in diameter) microspheres filled with high-molecular-weight gases like perfluorocarbons or sulfur hexafluoride, which results in decreased solubility and prolonged lifespan of the MB within the circulation.2 Until now, MB have been characterized by optical microscopy,4 static light scattering,6 and flow cytometry.4,5 Following the reported studies, we applied all the mentioned methods, and moreover, we attempted to preserve the MB using lyophilization, as described in a previous liposome study.7 Recently, targeting ligands that have been widely used in cardiovascular system and tumor diagnosis and therapy (reviewed by Liu et al.3) were attached to the surface of the MB. Received: November 24, 2010 Revised: January 21, 2011 Published: March 18, 2011 4829
dx.doi.org/10.1021/la104677b | Langmuir 2011, 27, 4829–4837
Langmuir Reversible interactions based on biotin-avidin/streptavidin pairs were implemented to provide the surface of the MB with the capability of attaching certain targeting ligands, e.g., monoclonal antibodies. The histidine-metallochelating lipid complex is one of the smallest high-affinity binding units available as tools for rapid noncovalent binding of histidine tagged molecules, especially recombinant proteins. The advantage of metallochelating complex over protein-ligand complexes (e.g., streptavidine-biotin, glutathiontransferase-glutathion) consists in its very low immunogenicity, if any. There have been only a few reports concerning the metallochelation of recombinant His-tagged proteins to nano- and microparticles, mainly liposomes.8-10 The preparation of such structures typically requires an insertion of a metallochelation lipid into the outer leaflet membrane of the liposome bilayer that interacts selectively in the presence of metal ions (such as nickel, zinc, or cobalt ions) with a 4-6 amino acid residue of histidine tail (His-tag) expressed at the N- or C-terminus of a recombinant protein of interest as a targeting ligand or therapeutic agent. The reversible character of the bond, fast binding kinetics at room temperature, and high affinity of metallochelation are very useful features for the construction of various self-assembled reversible supramolecular structures. The principle of metallochelating bond could be applied also to MB, and according to our knowledge, this is the first time when the metallochelating bond is implemented in the construction of proteoMB. To prove this concept, experiments were carried out using a mixture of DPPC and nickel-chelating lipid DOGS-NTA-Ni as metallochelating lipid in order to formulate proteoMB with proteins attached by metallochelation bond. Histagged recombinant green fluorescent protein (rGFP-HisTag) was used as a model protein for following structural studies. We developed a method for the preparation of metallochelating MB as a novel platform for the construction of targeted MB or other microbubble-based supramolecular structures. These MB were studied by TEM, SEM, light and confocal microscopy, flow cytometry, and static light scattering. Biocompatibility and ultrasound contrast imaging of these MB were tested in mice using the ultrasound imaging system.
’ MATERIALS AND METHODS Microbubble Preparation and Characterization. Chemicals. Lipids. 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-dioleyl-sn-glycero-3-phosphoethanolamine-N-(carboxyfluorescein) (CF-PE), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(liss-amine Rhodamine B sulfonyl) (LisR-PE), and metallochelating lipid 1,2dioleoyl-sn-glycero-3-{[N-(5-amino-1-carboxypentyl)iminodiacetic acid]succinyl} (nickel salt) (DOGS-NTA-Ni) were purchased from Avanti Polar Lipids (Birmingham, AL). Recombinant HisTag green fluorescent protein (rGFP) was obtained from Apronex (Prague, Czech Republic). MB (Vevo MicroMarker, batch QA907) commercially available from BRACCO Res. SA (Geneva, Switzerland) were used as reference for the in vivo heart imaging by ultrasound. Preparation of Liposomes. Liposomes were prepared by the lipid film hydration method. In brief, liposomes were composed of DPPC and/or appropriate amounts of auxiliary lipids (fluorescence-labeled lipids and metallochelating lipid). Lipids were dissolved in chloroform and mixed properly. The mixture was subsequently transferred into a round-bottomed flask, and the solvent was removed via rotary evaporation at 37 C. The lipid layer was hydrated by adding PBS solution up to
ARTICLE
Figure 1. Equipment for preparation of MB: (A) capsule mixing device (3M ESPE CapMix) and the small bomb with sulfur hexafluoride (SF6) linked with three-way valve and syringe for filling vials. (B) Sealed vial with dispersion of MB under a SF6 atmosphere. the final lipid concentration of 10 mg/mL at the temperature of 55 C in water bath. Afterward, the DPPC liposomes were rapidly frozen in liquid nitrogen and thawed in 55 C water bath five times. The resulting liposome suspension was digressively extruded through 400 and 200 nm polycarbonate membrane filters (Whatman; Brentfort, UK) at 55 C using an Avanti Mini-Extruder (Avanti Polar Lipids). Thereafter, the suspension was measured by Zeta Sizer NanoZS (Malvern) to check that the size of the produced liposomes is within the range of 192-205 nm. Preparation of MB. The liposome suspension (0.75 mL) was transferred into a hermetic vial (the volume of 1.5 mL). The vial was filled with sulfur hexafluoride gas (Messer Schweiz, AG) and mixed intensively for 30 s using 3M ESPE CapMix (capsule mixing device) (3M ESPEM, AG). The apparatus for filling the vials and the mixing device is shown in Figure 1. Fluorescence-labeled MB were prepared from a lipid mixture containing DPPC and appropriate fluorescence lipid at the concentration of 0-5 mol %. Residual lipid structures (liposomes and aggregates of collapsed MB shells) were separated from MB by low-speed centrifugation, removing the upper MB containing layer (without foam) and resuspending MB in appropriate buffer (PBS). The procedure was repeated three times. MB with bound rGFP were prepared by addition of rGFP solution (6 μg; 0.2 μmol) to 100 μL of metallochelating microbubble suspension (content of DOGS-NTA-Ni metallochelating lipid was 0.4 μmol). After 1 min of incubation at 25 C, 100 μL of the rGFP-MB was washed twice by PBS to remove residues of ruptured MB and non-microbubble lipids. This procedure allows obtaining clear background in confocal microscopy. Efficacy of Liposome-to-MB Conversion. The lipid concentration in liposomes and MB was measured by UV/vis spectrophotometer Uvicon XL (Bio-Tec Instrument) using Stewart’s method.11 After conversion of liposomes to MB, the vials were centrifuged at low speed to separate the MB from aqueous phase that contains residual non-microbubble structures. The samples (20 μL) were taken up from the sealed vial by a syringe. The needle was positioned near to bottom to get the sample from the aqueous phase below the layer containing the MB. This aqueous phase contains residual liposomes and nonbubble structures. The conversion rate was calculated according to equation Econv ð%Þ ¼ ðCtl - Crl Þ=Ctl 100% where Econv is the efficiency of liposomes-to-MB conversion, Ctl the content of total lipid, and Crl the content of residual nonbubble lipid. The theoretical number of monodispersed MB prepared from one mol of DPPC is expressed by equation NB ¼ NA al =4πr 2 4830
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir where NB is the number of MB prepared from 1 mol of lipid, NA the Avogadro number (6.02 1023 molecules/mol), al the effective area of the hydrophilic group for particular lipid (0.65 nm2 for DPPC), and 4πr2 the surface of a microbubble (r = microbubble radius). Provided that a monodisperse microbubble system is considered (diameter of 2.8 μm; r = 1.4 μm), the number of MB can theoretically be 1.6 1016 per 1 mol of saturated phosphatidyl choline (e.g., DPPC or DSPC). The theoretical number of MB (prepared of 10 mg of DPPC (MW 734 Da)) is about 2.16 1011. MB Characterization by Light, Epifluorescent, and Confocal Laser Scanning Microscopy. The size and concentration of the MB were determined by optical microscopy (Nikon Eclipse TE200). The microbubble samples were taken directly from the vial, diluted 100 times in PBS, and observed at room temperature. Images were captured in both bright-field and fluorescence modes using LUCIA software (Laboratory Imaging, Czech Republic). The objective Nikon LWD 20 was used to capture the images of the MB. To obtain absolute counts of MBs, the Burker counting chamber was employed, and the captured pictures of the MB were evaluated for the size distribution by LUCIA software. A true confocal scanner, Leica TCS SP2 microscope, was used to determine the structure of the MB labeled by headgroup fluorescence labeled lipids (CF-PE, LisR-PE) as well as the MB with bound rGFP. MB Characterization by TEM and SEM. The size and structure of the MB as well as the liposomes were determined by a Philips Morgagni transmission electron microscope (EM Philips 208 S, MORGAGNI software, FEI, CZ). All samples of MB were negatively stained by 2% (w/w) ammonium molybdate (pH 6.8). The sample of lyophilized MB was fixed to the SEM stub by Plano silver DAG. Images of uncoated samples were taken with the upper secondary electron detector of the Hitachi cold field emission scanning electron microscope SU8000 providing a resolution of 1.4 nm at 1 kV. In order to avoid charging artifacts, an accelerating voltage of 1.5 kV and a probe current of ∼15 pA was selected. Additionally, all images were recorded in the charge suppression or fast integration scan mode. Static Light Scattering (SLS). A HORIBA LA 950 laser diffraction particle size distribution analyzer was used to determine the microbubble size distribution. At first, the method was optimized by Megabead NIST Traceable Particle Size Standards (Polysciences, Inc., Warrington, PA) from 1, 3, 6, 10 to 15 μm. The 1.0% suspension of polystyrene microspheres in water was diluted by degassed and filtered PBS and measured. Microbubble sample (5-30 μL, 1-10 mg phospholipid/mL) was injected into cuvette-type fraction cell (filled with 10 mL of PBS) equipped with magnetic stirrer to prevent nonhomogeneous distribution owing to flotation of MB. All samples were measured immediately after the application into the cuvette and were analyzed both for number- and volume-weighted size distribution. Flow Cytometry. In this method, a FACSCalibur cell analyzer (Beckton-Dickinson, Franklin Lakes, NJ) was used to characterize microbubble fluorescence intensity (FL), light scattering profiles (FSC and SSC), and microbubble concentration. Voltage and gain settings for FSC, SSC, and FL were adjusted to delineate the microbubble populations from instrument and sample noise. Megabead NIST Traceable Particle Size Standards (Polysciences, Inc., Warrington, PA) from 1, 3, 6, 10 to 15 μm were used for the size distribution analyses. BD Trucount tubes (BD Biosciences) containing a known number of fluorescent polystyrene beads were used according to the instructions of the manufacturer. Absolute counts of MB were calculated by determining the ratio of beads to microbubble population and then multiplying this ratio by the number of beads in the tube. Subsequent data analysis was done using CellQuest Pro (Becton-Dickinson, Franklin Lakes, NJ). Counting of Microbubbles by Cell Counter. BC-2800 VET (Mindray, China) was used to count the microbubbles and to analyze
ARTICLE
their size. The microbubbles were washed two times by PBS and measured with dilution factor of 5. Lyophilization. Before lyophilization, saccharose powder (molar ratio DPPC:saccharose = 1:5) was added to 500 μL of microbubble suspension (total lipid concentration of 10 mg/mL) composed of DPPC and various concentrations (0-5 mol %) of DOGS-NTA-Ni. The vials were frozen at -80 C in a freezer and then lyophilized using the Lyovac GT2 instrument (Finn-Aqua, Finland). First, the samples were placed into the drying chamber precooled to -45 C. The lyophilization procedure was running for 24 h at 8 Pa. After this period of time, the second drying step was applied at 25 C for 12 h under 20 Pa. When the lyophilization process was finished, the gastight sealed samples were evacuated and refilled with sulfur hexafluoride by flushing with gas injected through septum. Residual water was below 1% as assayed by Karl Fisher titration. The lyophilized MB were reconstituted by addition of 500 μL of degassed and filtered water, and the hydrated preparation was gently mixed by rotation of the vial by hand. Stability of MB in Serum. The MB used for stability tests in serum were prepared by the standard way at the lipid concentration of 2 mg/mL. For reference, 50 μL of either liposome suspension used for microbubble generation or a microbubble sample was added to 3 mL of serum. The suspensions were placed into the sealed polycarbonate fluorescence plastic cuvette (10 mm, 4 mL) and tempered to 37 C. Quantification of the MB in the samples was carried out by light microscopy. The samples were taken every 5 min (interval 0-20 min), and photographs of three visual fields were acquired (at least 100 MB per field) and evaluated by LUCIA software (Laboratory Imaging, Czech Repblic). In vivo Ultrasound Imaging of Mouse Heart. The lyophilized MB were reconstituted by 0.5 mL of sterile PBS and left 10 min to hydrate, and 50 μL of the sample was injected via the cannulated right jugular vein to adult mice anesthetised by 2% isofluran (Aerrane; Baxter). For a comparison, commercial MB by Bracco were used as a standard. The preparation was diluted with PBS according to the manufacturer’s instructions. The imaging of heart was performed using ultrasonic system Vevo 770 (VisualSonic; Toronto, Canada) equipped with 25 MHz probe RMV-707-B working in the contrast mode (frame frequency 110 Hz). The left ventricle was observed in long axis view before, during, and after the injection of MB. After the experiment, the anesthetized animals were killed by cervical dislocation. Software Used. GraphPad Prism 5 software was used for statistical analyses and preparation of graphs.
’ RESULTS AND DISCUSSION Characterization of MB by Microscopy Techniques. Liposome-to-microbubble conversion of DPPC liposomes (unlabeled or labeled by fluorescence lipids) with or without DOGS-NTA-Ni was proved by light and fluorescence microscopy. Optical microscopy enabled a direct visual inspection of the yielded MB, which appeared as dark spheres with bright centers (Figure 2A,B). The analysis of the bright-field images using LUCIA software revealed that the size distribution of freshly produced MB in various batches ranged from 1 to 10 μm with the mean diameter in the range of 2-3 μm. The MB concentration determined by Burker cell counting showed that the average amount of MB (prepared of 1 mg of phospholipid) equaled approximately to (1-3) 108 MB/mL. In fluorescence mode of microscopy, MB appeared as bright rings with dark centers, clearly showing narrow lipid shell labeled by CF-PE. The preparation was free of lipid aggregates or collapsed bubbles (Figure 2B), as determined by fluorescence microscopy. A detailed structure of the MB was obtained by TEM 4831
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir
Figure 2. Pictures of MB obtained by various imaging techniques: (A) Light microscopy of DPPC/1% DOGS-NTA-Ni MB with carboxyfluorescein-PE (1%). (B) The same MB as in (A) observed by epifluorescence microscopy. (C) TEM of MB. During desiccation of the sample in the vacuum interior of electron microscope, the MB are pressed to each other, which results in the deformed polygonal shape. (D) TEM detailed picture of MB. Some liposome-like structures persist in the microbubble preparation and could be found attached to their surface (white arrow). The opening in the microbubble results probably from microexplosion during evacuation in TEM tube (black arrow). (E) SEM picture of lyophilized microbbubles. Intact MB (black arrows), crushed MB shell (white arrow), hollow inside of MB (white asterisk). Inset: details of the surface of lyophilized MB. (F) Residual liposome embedded in matrix (white arrow). Distance between two black line segments is 70 nm.
(Figure 2C,D). The high-resolution TEM pictures showed that residual liposome-like structures persist in microbubble preparation and could be found also attached to the surface of some MB. This fact is not taken into account by many researchers in the field, and this data is not shown in the literature. The TEM also revealed an opening in a microbubble owing to microexplosion
ARTICLE
during evacuation in TEM tube. During desiccation of the sample in the electron microscope, the MB are pressed to each other, which results in the deformed polygonal shape (Figure 2C,D). Characterization of Size Distribution of MB by Flow Cytometry, Optical Microscopy, and Static Light Scattering. Characterization by Flow Cytometry. Flow cytometry proved itself to be a very useful method for a precise characterization of the size distribution of MB as well as for a precise and rapid counting of the MB. The data were in a good agreement with those obtained by optical microscopy (Table 1). Moreover, a precise resolution of the multimodal distribution of MB was achieved by flow cytometry even if SLS was not able to resolve it (compare Figure 3B with Figure 3D). The minute amount of large MB (around 10 μm) was also well recognized. The limit size of the large MB was clearly around 10 μm as confirmed by Megabead NIST Traceable Particle Size Standards ranging from 1 to 15 μm. This data was confirmed by light microscopy. Optical Microscopy. Figure 3C shows a size distribution profile of MB as evaluated by optical microscopy. The multimodal distribution is recognized, however, not so precisely as in the case of flow cytometry. In optical microscopy, the number of evaluated bubbles is smaller than in flow cytometry; therefore, the multimodal distribution is not so precisely pronounced. On the other hand, direct examination and counting of real MB as objects is a great advantage of this method. Static Light Scattering. As the main advantage over the other SLS instruments, HORIBA LA 950 enables to measure in a relatively small cuvette (10 mL) with magnetic stirring to prevent changes in the distribution of MB in laser beam area owing to flotation. Microbubble size distribution by number as determined by HORIBA LA 950 ranged approximately from 1 to 18 μm with the mean around 4 ( 1.8 μm (Figure 3D). These data were not in a perfect correlation with the light microscopy and flow cytometry data (Table 1). In principle, the SLS technique tends to overestimate larger bubbles. We have found a tendency to overestimate larger MB; hence, the mean of the size distribution has shown slightly larger values compared with those determined by light microscopy or flow cytometry. Moreover, this method is very sensitive to the presence of residual liposomes and lipid aggregates formed from burst bubbles. Therefore, some samples were not analyzed correctly, and the obtained data reflected not only the presence of MB. Counting by Cell Counter. The number of microbubbles assayed by this method was in a good agreement with both FACS and microscopy techniques. Size distribution data are comparable with those obtained by static light scattering (Table 1). In conclusion, even if the evaluation of light microscopy data was more tedious in comparison to flow cytometry and SLS, we chose this method as the most precise because direct visual examination eliminates errors due to counting the nonbubble structures, which are inevitably present in the sample. Preparation of Metallochelating MB. The effect of DOGSNTA-Ni on the preparation and stability of DPPC MB was studied in order to optimize the lipid composition. DOGSNTA-Ni did not affect the conversion of liposomes into MB up to the concentration of 4 mol % in the lipid mixture (Figure 4). This concentration is high enough to provide a full coverage of the surface with His-tagged recombinant protein. 4832
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir
ARTICLE
Table 1. Comparison of Microscopy, Static Light Scattering, Flow Cytometry, and Cell Counter with Respect to Determination of MB Size Distribution and Concentrationa method of measurement
a
concn of MB [MB/mL] 108
av size of MB [μm]
minimal size [μm]
maximal size [μm]
light microscopy
2.53
2.17
0.93
9.89
static light scattering flow cytometry
DNP 1.78
4.09 2.04
2.1 0.64
16.92 10.86
cell counter
2.04
3.86
1.68
7.26
MB = microbubbles, MB = composed of DPPC (1 mg/mL), DNP = determination is not possible, and ND = not determined.
Figure 3. Comparison of size distribution analyses performed by flow cytometry, optical microscopy, and static light scattering. MB were prepared of DPPC/1% DOGS-NTA-Ni (lipid concentration of 10 mg/mL): (A) Positions of standards in forward scattering plot (flow cytometry analysis). (B) Size distribution profile of DPPC/1% DOGS-NTA-Ni MB (flow cytometry analysis). (C) Histogram of DPPC/1% DOGS-NTA-Ni MB obtained by optical microscopy. (D) Graph of size distribution measured by static light scattering (HORIBA LA 950 SLS instrument).
Generally, 1 mol % of metallochelating lipid is sufficient for this purpose.10 No significant effect on the size distribution was found by light microscopy, SLS, and flow cytometry (results not shown). Efficacy of Liposomes-to-MB Conversion. The efficacy of the liposomes-to-MB conversion process was found to be around 6.5 ( 4% (based on the measurement of the lipid concentration by Stewart’s method) for all tested compositions with the exception of DPPC/5% DOGS-NTA-Ni liposomes that exhibited a lower stability and the conversion was around 0.2-0.5%. This could be owing to the limited miscibility of DPPC (saturated lipid) with DOGS-NTA-Ni (unsaturated lipid), which results in a decreased stability of such a monolayer. This data based on the
measurement of the content of phospholipids in microbubble and nonmicrobubble structures is in a good agreement (within the order of magnitude) with the theoretical calculations showing that the theoretical number of monodisperse 2.8-μm MB (prepared of 10 mg of DPPC) is about 2.16 1011 MB. The number of MB in our preparations was within the range of (1-3) 109 (10 mg of phospholipid). The precision of the calculation is, of course, affected by the fact that a real system is not monodisperse; hence, the theoretical number of the monodisperse MB does not fit the theoretical number for a real systems with a various degree of polydispersity. Moreover, some portion of non-microbubble lipid is presented also in microbubble fraction owing to simple physical association with the MB (see 4833
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir
ARTICLE
Table 2. Effect of Lyophilization on Number and Size Distribution of DPPC and DPPC/DOGS-NTA-Ni MBa MB status
concn
av size
(before/after
of MB
of MB
size
size
composition lyophilization)
[MB/mL]
[μm]
[μm]
[μm]
DPPC
before
3.575 108 2.17 ( 0.85 0.93
6.89
after
1.23 108
2.54 ( 1.44 1.19
7.93
before
3.25 108
2.02 ( 0.68 1.04
5.13
after
1.175 108 2.11 ( 0.89 1.19
6.55
lipid
1% DOGS-
minimal maximal
NTA-Ni a
Figure 4. Influence of DOGS-NTA-Ni on the effectiveness of formation of DPPC MB. DPPC MB were prepared with addition of DOGSNTA-Ni (0-5 mol %), and the effectiveness of liposome-to-microbubble conversion process was expressed as the number of MB per 10 mg of lipid. The number of MB was evaluated by optical microscopy. /: p g 0.95.
TEM micrographs, Figure 2D). This fact affects the calculation of the number of MB based on the microbubble and non-microbubble lipid assay by Stewart’s method. Lyophilization of MB. Lyophilization is a preferable method for the preparation of long-term storage form of MB. We tested saccharose as an acceptable cryoprotective compound often used for lyophilization of liposomes.12 Saccharose powder (molar ratio Lipid (DPPC/0-4 mol % DOGS-NTA-Ni):saccharose = 1:5) was found to be effective to preserve the MB during lyophilization and ensuing reconstitution by hydration. Table 2 shows the data for DPPC and DPPC-1 mol % DOGS-NTA-Ni. About 35% of MB were preserved after the lyophilization and rehydration process. No significant differences were found within the group composed of DPPC/0-4 mol % DOGS-NTA-Ni lipids. The mean size of the reconstituted MB was similar to that of the same sample prior to lyophilization. The size limit of the large MB was shifted by about 1 μm toward the larger ones. This effect could have been caused by a fusion of small amount of MB during rehydration. The picture of MB embedded in saccharose matrix after the lyophilization process is shown in Figure 2E. The microbubbles appeared as spherical hollow objects that are a little bit wrinkled owing to lyophilization process. Some of them were crashed by manipulation with the lyophilizate during preparation of the sample; therefore, the very thin fragile lipid shell as well as the inside of MB is well visible. Also, residual liposomes which were not converted into MB are clearly visible embedded in the matrix around the MB (Figure 2F). In Vitro Stability of DPPC/DOGS-NTA-Ni Metallochelating MB in Serum. The stability of MB in 100% bovine serum at 37 C was tested by light microscopy. MB were obviously present more than 20 min, and the estimated half-live of the MB in bovine serum was more than 20 min for DPPC and DPPC/ DOGS-NTA-Ni 1 mol % and about 5-8 min for DPPC/DOGSNTA-Ni 3 and 5 mol %. The stability of the MB was affected by a content of DOGS-NTA-Ni. With an increasing concentration of DOGS-NTA-Ni, the stability slightly decreased (Figure 5). During the incubation with serum, no significant effect on the mean size and the size distribution was observed (Table 3). The in vitro stability of MB in concentrated serum is a prerequisite for the in vivo stability within blood circulation system. The destabilizing effect of DOGS-NTA-Ni at the concentration higher than 4-5 mol % is indicative of possible phase separation of DPPC and DOGS-NTA-Ni lipids. It results in inhomogeneities in lipid monolayer, which destabilizes the MB shell. It is also
Lipid concentration was 1 mg/mL.
Figure 5. In vitro stability of DPPC/DOGS-NTA-Ni metallochelating MB in serum. DPPC MB with 0, 1, 3, and 5 mol % of DOGSNTA-Ni were incubated in bovine serum (100%) at 37 C, and the decrease of their number was quantified by optical microscopy. The data were normalized and plotted as percentage of MB number versus time (s).
reasonable to suppose that high surface concentration of metallochelating function in DOGS-NTA-Ni phase-separated islands results in nonspecific interaction with serum proteins destabilizing MB. These aspects will be subjected to further research with the aim to improve in vivo stability of metallochelating MB. In vivo Visualization and Stability of MB. Figure 6 shows typical images of the left ventricle in a long axis view before and after the intravenous injection of DPPC-based MB and commercial MB (Bracco) in anesthetized mice. In both cases, no adverse effect on breathing and heart function was observed. DPPC-based MB provided a clear and very contrast image of the ventricular cavity soon after the injection. In comparison with the MB by Bracco, which are stabilized by poly(ethylene glycol), pure DPPC-based MB exhibited a lower stability in the circulation (the contrast disappeared after 1-2 min, while the MB by Bracco yielded the contrast picture for up to 10 min). For instance, Levovist and SonoVue have a half-life of 78 s and 5 min, respectively. An adequate in vivo bubble half-life is a major requirement for imaging as well as ultrasound facilitated drug targeting. In fact, an unpredictable release profile would lead to a lack of control of the drug release, regardless of a passive or active mechanism of MB targeting. As mentioned above, the bubble size is also a critical parameter that must be controlled and kept in the range of 1-7 μm, preferably around 3 μm. The MB larger than 6-8 μm would be trapped in the lung capillaries, and the bubbles smaller than 1 μm would be less stable and less ultrasound responsive.13 Our DPPC-based MB fulfill these requirements. 4834
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir
ARTICLE
Table 3. Size Distribution of DPPC/DOGS-NTA-Ni MB during Incubation in Bovine Serum at 37 C and Stability of MB Size (Measured in 0, 5, 10, 15, and 20 min) lipid composition
av size [μm]
minimal size [μm]
maximal size [μm]
prepared MB (DPPC)
2.9 ( 1.7
1.1
11.6
0 min 5 min
3.0 ( 1.6 2.9 ( 1.6
0.9 1.2
9.5 11.0
10 min
2.7 ( 2.0
0.9
9.2
15 min
2.6 ( 1.4
0.9
10.4
20 min
2.6 ( 1.5
1.0
8.0
prepared MB (1% DOGS-NTA-Ni)
2.7 ( 1.3
1.3
9.8
0 min
2.7 ( 1.6
1.0
7.5
5 min
2.9 ( 1.6
1.0
8.7
10 min 15 min
2.7 ( 2.0 2.6 ( 1.4
1.0 0.8
6.2 5.3 7.9
20 min
2.6 ( 1.5
1.1
prepared MB3% DOGS-NTA-Ni
2.3 ( 0.8
1.0
5.9
0 min
2.2 ( 1.1
1.0
8.7
5 min
2.1 ( 0.7
1.0
4.5
10 min
2.1 ( 0.8
1.0
4.3
15 min
2.0 ( 1.2
1.0
4.7
20 min prepared MB5% DOGS-NTA-Ni
2.2 ( 0.8 2.4 ( 1.0
1.3 0.7
4.2 7.3
0 min
2.5 ( 1.0
1.0
5.2
5 min
2.3 ( 0.9
1.0
5.0
10 min
2.5 ( 1.0
1.2
5.9
15 min
2.3 ( 0.8
1.2
4.5
20 min
2.5 ( 1.1
1.2
5.1
Figure 6. Ultrasound contrast images of the mouse heart in a long axis view. (A) Without MB. (B) After the intravenous administration of DPPC/1% DOGS-NTA-Ni MB. (C) After the intravenous administration of commercial MB by Bracco. c: left ventricle cavity; a: left ventricle anterior wall; p: left ventricle posterior wall.
Binding of rGFP-HisTag onto the Surface of Metallochelating MB. The rGFP-HisTag molecule bound onto the surface of metallochelating liposomes was directly detected by TEM (Figure 7A,B). Fluorescence of rGFP was used to prove binding of this protein onto the MB (Figure 7C). The binding did not affect the native conformation of rGFP as demonstrated by the preserved fluorescence characteristics of the protein. Homogeneous population of highly fluorescent rGFP MB was detected by flow cytometry (Figure 7G). Nonspecific binding of rGFP on both DPPC liposomes as well as MB lacking DOGS-NTA-Ni was observed neither by flow
cytometry nor by confocal microscopy. Confocal microscopy confirmed colocalization of bound rGFP-HisTag (green fluorescence) with the lipids forming microbubble monolayer. The monolayer was marked with LisR-PE (red fluorescence) (Figure 7D-F). Confocal microscopy revealed that a small fraction of MB had also attached liposomes or residual lipid structures of collapsed MB on their surface (as seen in Figure 7D-F). This is in a good agreement with the electron microscopy data (Figure 2D). Affinity metallochelating binding is based on complexes of bound metal ions (preferably nickel but cobalt, manganese, or 4835
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir
Figure 7. Structure of metallochelating liposomes and MB with bound rGFP-HisTtag. (A) TEM micrograph of the structure of plain liposomes. (B) TEM micrograph of the structure of rGFP-His-tag proteoliposomes (the arrow indicates rGFP-His-tag located on the liposome surface). (C) Confocal microscopic picture of rGFP-HisTagmetallochelating MB revealing the localization of functional rGFP on their surface (green fluorescence). (D) Detailed confocal microscopic picture of rGFP (green fluorescence) localized on the surface of a MB. (E) Detailed confocal microscopic picture of LisR-PE (red fluorescence) localized on the surface of a MB (white arrow indicates the lipid membrane aggregate (residuum of a ruptured MB) adsorbed on the surface of a MB; yellow arrow indicates a cluster of residual liposomes attached to the surface of a MB). (F) Merged picture demonstrating colocalization of rGFP-His-tag with MB shell labeled with LisR-PE. (G) Flow cytometry analysis of rGFP-His-tag DOGS-NTA-Ni MB (green fluorescence) and DOGS-NTA-Ni MB (nonfluorescent).
zinc could be used also), to which the polyhistidine-tag binds with micromolar affinity. The main advantage of this metallochelating binding is the site-specific immobilizations of proteins. This feature is highly demanded especially in the field of molecular recognition comprising biosensors, oriented surfaces, and targeted drug carriers. Investigations with single-molecule experiments using scanning force microscopy revealed that His-tags form various types of complexes, which significantly differ in stability and energy profile along their dissociation pathway.14,15 With respect to a potential application of metallochelating bond to the construction of targeted MB, the question of in vitro and especially in vivo stability is of great importance. This problem could be divided into two fields: first, the stability of the MB themselves and, second, the effect of the components present in biological fluids (e.g., proteins and ions) on the stability of the metallochelating bond. It is beyond the scope of this study to address thoroughly this particular question. However, the in vivo data indicated that the stability of DPPC-based MB in circulation is lower than that found for the MB stabilized in lyophilizate by poly(ethylene glycol) as cryoprotectant (SonoVue MB). With the sterically
ARTICLE
stabilized MB (e.g., by application of poly(ethylene glycol)phosphatidylcholine as the component modifying the surface of microbubbles), the immobilization of His tag protein on such a sterically protected surface could be greatly hampered by the inaccessibility of the His-tag caused by steric hindrance as described for the interaction of His-tag proteins with metallochelating affinity chromatography media.16 In this case, the metallochelating function (e.g., NTA) should be positioned at the end of PEG chain to be easily accessible to His-tag. It implies a design and synthesis of new metallochelating lipids fulfilling this specific demand and, hence, being more suitable for the construction of metallochelating MB. The lipid part of the molecule should be optimized to provide miscibility with the lipids used for the formation of the microbubble shell. The role of the spacer with metallochelating function is to provide high affinity binding of His-tag. Compared with the strong interactions reported for the complexes of avidin and biotin (KD = 10-14 M), the affinity of the hexahistidine peptide to Ni-NTA is more than a factor of 105 lower (KD = 10-9 M). A limited accessibility of the tag caused by a steric hindrance or electrostatic interactions are possible explanations for the decrease in binding affinity for the Ni2þ/6His-tagged protein complex (KD ≈ 10-6 M).17 Because there are no data in the literature dealing with metallochelating MB, we have to estimate the possible stability of the metallochelating bond according to a few papers dealing with metallochelating liposomes. The study by Ruger et al. shows that the singlechain Fv DOGS-NTA-Ni liposomes are unstable in human plasma and that most of single-chain Fv fragments are released from the liposomal surface.18 However, our experiments dealing with rOspC His-tag/DOGS-NTA-Ni proteoliposomes demonstrated their relative stability under the same conditions (Turanek, unpublished results). Similarly, Ni-NTA3-DTDA liposomes with single-chain Fv fragments (antiCD11c) bound onto the liposomal surface were able to target dendritic cells in vitro as well as in vivo.19,20 The application of the three-functional chelating lipid Ni-NTA3-DTDA probably endows the metallochelating bond with a higher in vivo stability. Generally, metal ions, physicochemical character of the metallochelating lipids, and their surface density on the particles belong to the factors that could be optimized to get a required in vivo stability.
’ CONCLUSION At present, the histidine-metallochelating lipid complex is one of the smallest high affinity binding units used as tools for rapid noncovalent binding of histidine tagged molecules, especially recombinant proteins. The advantage of metallochelating complex over protein-ligand complexes (e.g., streptavidinebiotin, glutathion transferase-glutathion) consists in its sitespecific binding and very low immunogenicity, if any. We proved usefulness of this concept for the construction of surface modified MB on the model of metallochelating MB with bound rGFP, which is easy to be detected by various techniques as flow cytometry and confocal microscopy. In vivo application of DPPC-based metallochelating MB to mice demonstrated biocompatibility and good contrast imaging comparable with that of the commercial product SonoVue (Bracco). Stabilization of the metallochelating MB preparation by lyophilization is feasible, and the in vivo stability is within the range of phospholipid-based MB (Levovist or SonoVue). An application of monolayer stabilizers like PEG-40 stearate, Pluronic F68, or Polysorbate 80 could further improve the in vitro and 4836
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
Langmuir in vivo stability. The main improvement is expected to be obtained by application of novel metallochelating lipids that are designed to be incorporated into the microbubble monolayer. To design such metallochelating lipids, several essential aspects must be taken into account: stabilization effect of the lipid part on the monolayer, high affinity of the metallochelating function linked to the accessory lipid part, and a suitable spacer (e.g., various PEG spacers) for effective binding of larger molecules onto sterically stabilized MB. The design and synthesis of these biocompatible metallochelating lipids is in progress.
ARTICLE
(17) Nieba, L.; Nieba-Axmann, S. E.; Persson, A.; H€am€al€ainen, M.; Edebratt, F.; Hansson, A.; Lidholm, J.; Magnusson, K.; Karlsson, A. F.; Pl€uckthun, A. Anal. Biochem. 1997, 252, 217–228. (18) R€uger, R.; M€uller, D.; Fahr, A.; Kontermann, R. E. J. Drug Target. 2006, 14, 576–582. (19) van Broekhoven, C. L.; Parish, C. R.; Demangel, C.; Britton, W. J.; Altin, J. G. Cancer Res. 2004, 64, 4357–4365. (20) Knecht, S.; Ricklin, D.; Eberle, A. N.; Ernst, B. J. Mol. Recognit. 2009, 22, 270–279.
’ AUTHOR INFORMATION Corresponding Author
*Phone þ 420 5 3333 1311; Fax þ 420 5 4121 1229; e-mail
[email protected]. Author Contributions ^
Contributions of the first two authors are equal.
’ ACKNOWLEDGMENT This work was supported by following grants: Grant Agency of the Academy of Sciences of the Czech Republic GAAV KAN200520703 to J.T., F.P., and F.K.; Grant Agency of the Academy of Sciences of the Czech Republic GAAV KAN200100801 to J.T.; Grant Agency of the Czech Republic GACR P304/10/1951 to J.T.; and Grant No. MZE 0002716202 to J.T. The authors thank Asst. Professor Milan Raska, Ph.D., for the schema of metallochelating microbubble in the table of contents graphic. ’ REFERENCES (1) Rawat, M.; Singh, D.; Saraf, S.; SARAF, S. Yakugaku Zasshi. 2008, 128, 269–280. (2) Hernot, S.; Klibanov, A. L. Adv. Drug Delivery Rev. 2008, 60, 1153–1166. (3) Liu, Y.; Miyoshi, H.; Nakamura, M. J. Controlled Release 2006, 114, 89–99. (4) Feshitan, J. A.; Chen, C. C.; Kwan, J. J.; Borden, M. A. J. Colloid Interface Sci. 2009, 329, 316–324. (5) Nicholson, J. K.; Stein, D.; Mui, T.; Mack, R.; Hubbard, M.; Denny, T. Clin. Diagn. Lab. Immunol. 1997, 4, 309–313. (6) Rapoport, N.; Gao, Z.; Kennedy, A. J. Natl. Cancer Inst. 2007, 99, 1095–1106. (7) Siow, L. F.; Rades, T.; Lim, M. H. Cryobiology. 2007, 55, 210– 221. (8) Malliaros, J.; Quinn, C.; Arnold, F. H.; Pearse, M. J.; Drane, D. P.; Stewart, T. L.; Macfarlan, R. I. Vaccine 2004, 22, 3968–3975. (9) Chikh, G. G.; Li, W. M.; Schutze-Redelmeier, M. P.; Meunier, J. C.; Bally, M. B. Biochim. Biophys. Acta 2002, 1567, 204–212. (10) Masek, J.; Bartheldyova, E.; Korvasova, Z.; Skrabalova, M.; Koudelka, S.; Kulich, P.; Kratochvílova, I.; Miller, A. D.; Ledvina, M.; Raska, M.; Turanek, J. Anal. Biochem. 2011, 408, 95–104. (11) Stewart, J. C. M. Anal. Biochem. 1980, 104, 10–14. (12) Koudelka, S.; Masek, J.; Neuzil, J.; Turanek, J. J. Pharm. Sci. 2010, 99, 2434–2443. (13) Cavalieri, F.; Zhou, M.; Ashokkumar, M. Curr. Top. Med. Chem. 2010, 10, 1198–1210. (14) Schmitt, L.; Ludwig, M.; Gaub, H. E.; Tampe, R. Biophys. J. 2000, 78, 3275–3285. (15) Conti, M.; Falini, G.; Samori, B. Angew. Chem., Int. Ed. 2000, 39, 215–218. (16) Miras, R.; Cuillel, M.; Catty, P.; Guillain, F.; Mintz, E. Protein Expr. Purif. 2001, 22, 299–306. 4837
dx.doi.org/10.1021/la104677b |Langmuir 2011, 27, 4829–4837
