Low-Density Lipoprotein Nanoparticles as Magnetic Resonance Imaging Contrast Agents

Abstract

Low-density lipoproteins (LDLs) are a naturally occurring endogenous nanoplatform in mammalian systems. These nanoparticles (22 nm) specifically transport cholesterol to cells expressing the LDL receptor (LDLR). Several tumors overexpress LDLRs presumably to provide cholesterol to sustain a high rate of membrane synthesis. Amphiphilic gadolinium (Gd)-diethylenetria-minepentaacetic acid chelates have been incorporated into the LDL to produce a novel LDLR-targeted magnetic resonance imaging (MRI) contrast agent. The number of Gd chelates per LDL particle ranged between 150 and 496 Gd(III). In vitro studies demonstrated that Gd-labeled LDL retained a similar diameter and surface charge as the native LDL particle. In addition, Gd-labeled LDL retained selective cellular binding and uptake through LDLR-mediated endocytosis. Finally, Gd-labeled LDLs exhibited significant contrast enhancement 24 hours after administration in nude mice with human hepatoblastoma G₂ xenografts. Thus, Gd-labeled LDL demonstrates potential use as a targeted MRI contrast agent for in vivo tumor detection.

Introduction

Nonspecific paramagnetic contrast agents, such as gadolinium-diethylenetriaminepentaacetic acid (Gd-DTPA) and other Gd chelates, have greatly improved the diagnostic capabilities of magnetic resonance imaging (MRI) [1–4]. These agents augment intrinsic tissue contrast, thereby enhancing soft tissue discrimination as well as providing important physiological information about perfusion, vascular permeability, and extracellular volume [5,6]. The recent development of targeted paramagnetic contrast agents promises to greatly expand the diagnostic specificity of these agents [7–9]. By targeting specific cell surface epitopes, these agents will enable imaging of specific cell populations in vivo. Although MRI provides high spatial resolution, it suffers from low inherent sensitivity [10]. In addition, the limited number of cell surface epitopes suitable for targeting impedes their detection by targeted MRI contrast agents. The use of nanoplatforms (nanoscale structures typically smaller than 100 nm) to simultaneously deliver multiple Gd chelates may overcome this limitation. Most of the existing nanoplatforms consist of synthetic nanostructures, such as dendrimers [11], silica-coated micelles [12], polymeric and ceramic nanoparticles [13,14], perfluorocarbon emulsions [15], and crosslinked liposomes [16]. In addition to attaching diagnostic agents, specific moieties, typically small peptides, peptidometics, or antibodies, have been conjugated to the platform to impart targeting capabilities [17–20]. Although many of these particles have shown good targeting with high payloads of Gd(III), their large size confined them to the vascular compartment [18,21]. The utility of these synthetic nanoparticles is often limited by biocompatibility, biodegradability, and toxicity problems.

Low-density lipoproteins (LDLs) are naturally occurring nanostructures that serve as the main transport vehicles for cholesterol in mammalian systems [22]. LDL particles contain a phospholipid monolayer encapsulating a hydrophobic core. Spanning the phospholipid monolayer is the apolipoprotein B-100 (ApoB-100), a very large amphipathic protein containing one or more clusters of cationic amino acids that target LDL receptors (LDLRs) on cells. After binding to the LDLR, LDL is internalized by endocytosis. Within lysosomes, LDL particles are degraded into amino acids, fatty acids, and free cholesterol, whereas the receptor is recycled to the cell surface. Given LDLR recycle time is only 10 minutes and its lifetime is 24 hours, this receptor pathway efficiently transports many LDL particles to LDLR-expressing cells. In addition to a number of normal tissues such as the liver, adrenal glands, and ovaries, which use the LDLR system, several tumors overexpress the LDLR [23–30]. Several investigators have used LDLs to selectively deliver diagnostic and/or therapeutic agents to LDLR-overexpressing tumors [31]. For diagnostic purposes most of these efforts have focused on attaching radionuclides to ApoB-100 [32,33] or intercalating radiolabeled amphiphilic conjugates into the LDL phospholipid monolayer [34,35]. Alternatively, near-infrared fluorescent probes have been attached to LDLs for tumor detection [36,37]. Although all of these studies were performed ex vivo or in cell culture, they have generated promising results that have shown selective uptake of labeled LDL by cells/tissues expressing LDLR. To date, few studies have gone on to assess the utility of these probes for in vivo tumor detection. Ponty et al. [38] used ^99mTc-LDL to visualize B16-melanoma xenographs in mice.

Scintigraphic images demonstrated accumulation of the labeled probe in the upper abdomen and in the region of the tumor xenograft; however, the poor spatial resolution of these images impaired anatomical demarcation of organ or tumor boundaries. To overcome these problems, members of our group have attempted to use paramagnetic labeled LDL for tumor detection with MRI [39]. This prototype contrast agent (PTIR267) contained both a fluorophore and a Gd(III) moiety along with two palmitate hydrocarbon chains, which facilitated intercalation into the LDL phospholipid monolayer. Fifty molecules of PTIR267 were attached to each LDL, and subsequent IV administration of this agent into a B16 tumor-bearing mouse demonstrated accumulation of labeled LDL within the liver and tumor. Although this initial finding clearly demonstrated the feasibility and potential of the LDL platform, a number of problems limit its utility. The current formulation of PTIR267 (fluorophore, Gd-DTPA moiety, and bis-palmitate chains) produces a bulky and sterically hindered molecule; hence, only a limited number (upper limit of 50) of this contrast agent can be intercalated into the LDL phospholipid monolayer. This becomes a critical issue for MR molecular imaging as high paramagnetic payloads are required to overcome the low inherent sensitivity of MRI. One method of delivering more Gd(III) to the target site would be to simply increase the amount of administered PTIR267; this approach, however, could elicit anaphylactic reactions and toxicity due to the high concentrations of protein and fluorescent dye. Clearly, alternate approaches are necessary to further optimize the diagnostic utility of paramagnetically labeled LDL. In the present study, a more versatile and less bulky amphiphilic Gd chelate was synthesized. Due to the simple structure of this agent, it was possible to incorporate much higher payloads (150–500) of the probe into LDL. Preparation of paramagnetic labeled LDL was performed in two steps: first the amphiphilic chelate was incorporated into LDL, and second chelation with Gd(III) was performed to produce the novel LDLR-targeted MRI contrast agent. Here we evaluate the characteristics and binding properties of this Gd(III)-labeled LDL and assess the diagnostic utility of this agent for detecting tumors overexpressing LDLR in vivo.

Materials and Methods

Preparation and Characterization of Gd-Labeled LDL Nanoparticle Preparation of DTPA-bis(stearylamide)

DTPA-bis(stearylamide) (DTPA-SA) was prepared by the method of Jasanada et al. [34]. Briefly, stearylamine (0.54 g; 2 mmol) in chloroform (40 ml) was slowly added to the bisanhydride of DTPA (0.40g; 1.1 mmol) in DMF (50 ml). After 2 hours of stirring at 40°C, the solution was cooled at 4°C for 2 hours. The white precipitate was filtered, washed with acetone, and dried overnight at 80°C. The precipitate was then crystallized in boiling ethanol (800 ml). After 24 hours at room temperature, the small crystals were collected by filtration and washed with water (800 ml, 80°C for 3 hours) and chloroform (800 ml, reflux for 3 hours) to eliminate unreacted DTPA and SA.

Low-Density Lipoprotein

LDL was isolated from fresh plasma of healthy donors by sequential ultracentrifugation as described previously by Lund-Katz et al. [40].

Incorporation of DTPA-SA into LDL

A 3.0 mM solution of DTPA-SA was prepared by dissolving the crystals in aqueous ammonia solution (NH₄OH/NH₄Cl, pH 9, 0.15 M) with vigorous stirring and gradual heating to approximately 40°C. Once the crystals had dissolved, the clear solution was diluted to a concentration of 1.5 mM with Tris-buffered saline. The DTPA-SA and LDL were mixed at a molar ratio of 200:1. First, an aliquot of the DTPA-SA solution was diluted with Tris-buffered saline (to 0.35mg/ml) and adjusted to pH value 8.0 with HCl (1 M). The DTPA-SA solution was added dropwise to the LDL solution. The final concentration of LDL was 0.4 mg/ml. After stirring under N₂ for 1 hour at room temperature, the sample was filtered through a 0.22-µm membrane filter and dialyzed against Tris-buffered saline overnight (16 hours) at 4°C. Any DTPA-SA that precipitated during dialysis was removed by membrane filtration (0.22 µm). The amount of DTPA-SA incorporation into LDL was determined by UV spectrometry using the zinc dithizone assay [34]. LDL particles labeled with higher payloads of DTPA-SA (up to molar ratio of 500:1) were also performed by the methods described above.

LDL Determination

The molar concentration of LDL was determined by total protein analysis using a commercial Lowry protein assay kit (Sigma-Aldrich, St. Louis, MO) on the assumption that ApoB-100 (550 kDa) was the only protein present with one protein per LDL particle.

Gadolinium Labeling

Gd citrate solution was prepared by adding GdCl₃ (17.5 µmol) in HCl to a solution of sodium citrate (87.5 µmol). Excess citrate was used to avoid formation of insoluble hydroxides. The pH of the Gd citrate solution was adjusted to 7.4 with HCl/NaOH. Gd(III) labeling of DTPA-SA-LDL was performed by slowly adding Gd citrate to a solution of DTPA-SA-LDL at a metal/chelator ratio of 1:1. After incubation for 1 hour at room temperature under N₂ with gentle stirring, the product was filtered, and a dilute solution of tropolone (10 mmol/L) was added in excess (15x) to eliminate any nonspecific binding of Gd(III) to LDL. The solution was gently stirred for 1 hour at 4°C and the final product was filtered. The solution was concentrated down to a volume of 500 µl with a 15-ml Centricon centrifugal filter unit (Millipore, Billerica, MA).

Light Scattering

The particle size distributions of native and modified LDL were measured by light-scattering photon correlation spectroscopy (Zetasizer 3000HS, Malvern Instruments, Malvern, UK) utilizing a 10-mW He-Ne laser operating at 633 nm and a detector angle of 90°. The data were modeled assuming spherical particles undergoing Brownian motion.

Electron Microscopy Studies

Transmission electron microscopy with a JEOL JEM 1010 electron microscope equipped with a charge-coupled device camera (Hamamatsu, Middlesex, NJ) operating at 80 kV using AMT 12-HR software (Advanced Microscopy Technique, Danvers, MA) was used to determine the morphology and the size of the aqueous dispersion of nanoparticles. Five microliters of lipoprotein nanoparticle suspension was placed on carbon-coated 200-mesh copper grids and allowed to stand for 5 minutes. Excess sample was removed with lens paper, and 2% saturated aqueous uranyl acetate was applied to the grid in five consecutive drops within 20 seconds. The stain was then drained off with filter paper, and the grid was air dried before digital images were taken. All electron microscopy supplies were purchased from Electron Microscopy Sciences (Fort Washington, PA).

Agarose Electrophoresis

The electrophoretic properties of LDL particles were examined by 0.5% argarose gel electrophoresis. Samples in Tris-saline buffer (6 µg protein in 4 µl) were applied to gel wells and allowed to penetrate into gel for 5 minutes before the electric field was applied. A Bio-Rad model 702 power supply was used to apply a voltage of 100 ± 2 V across a gel distance of 5.5 cm. Electrophoresis was continued for 30 minutes at 25 ± 2°C in barbital buffer (pH 8.6, 0.05 ionic strength). After electrophoresis, the gels were fixed in a solution of ethanol/acetic acid/water 60:10:30 (vol/vol/vol), oven-dried (80°C for 1 hour), and then stained (5 minutes) with a 0.15% Coomassie Blue R250 solution. Gels were destained in a solution of methanol/acetic acid/water 35:25: 40 (vol/vol/vol) for about 10 minutes or until the background adjacent to the LDL bands was clear and the stain intensity of the bands was uniform.

Determination of Longitudinal Relaxivities

The proton longitudinal relaxation rates (1/T₁) of gadodiamide and Gd-labeled LDL (Gd-DTPA-SA-LDL) were measured at 21 and 40°C on an mq60 Minispec NMR Analyzer (Bruker, Inc., The Woodlands, TX) at a field strength of 60 MHz. The contrast agents were prepared in saline, and T₁ was measured with an inversion recovery pulse sequence using 10 to 20 inversion delay values. The T₁ relaxivities were calculated from the slope of a plot of the longitudinal relaxation rates versus Gd³⁺ concentrations (millimolar).

Assessment of LDLR-Mediated Uptake of Gd-DTPA-SA-LDL

To determine the specificity of Gd-DTPA-SA-LDL uptake by LDLR, cell culture experiments were performed with MRI and optical imaging techniques.

Cell culture

Human hepatoblastoma cancer cell line (HepG₂) (overexpress LDLR) and ldlA7 [LDLR-negative Chinese hamster ovary (CHO) cells] [41] were obtained from Drs. Theo van Berkel (University of Leiden, Netherlands) and Monty Krieger (Massachusetts Institute of Technology, Cambridge, MA), respectively. HepG₂ cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml penicillin-streptomycin.

IdlA7 were cultured in DMEM/F-12 medium (Ham's nutrient mixture) supplemented with 100 U/ml penicillin-streptomycin, 2 mM L-glutamine and 5% fetal bovine serum (FBS). Cells were grown at 37°C in an atmosphere of 5% CO₂ in a humidified incubator.

Magnetic Resonance Imaging

Three and a half million cells each of HepG₂ and ldlA7 grown in T25 Corning flasks (Sigma-Aldrich) were incubated with 0.44 mM Gd-DTPA-SA-LDL for 5 hours at 37°C. An additional flask of HepG₂ was incubated in the presence of Gd-DTPA-SA-LDL and a six-fold concentration of native LDL. After this incubation period the cells were washed three times with PBS, trypsinized, and centrifuged (1000 rpm for 5 minutes) to pellet the cells. The total volume of the sample (pellet and fresh PBS) was approximately 150 µl. Cells not exposed to the contrast agent served as untreated controls. MRI evaluations were performed on the cell samples using the 4.7-T Varian magnet (Palo Alto, CA). Sagittal T₁-weighted MR images were acquired with a spin-echo sequence (TR/TE=500/15 ms, matrix = 256 x 128 and FOV = 6 x 3 cm, slice thickness = 1 mm, signal average =4).

Preparation of Gd-DTPA-SA tetra-t-Butyl Silica Phthalocyanine Bisoleate-Reconstituted LDL

The near-infrared dye tetra-t-butyl silica phthalocyanine bisoleate (SiPc-BOA) was reconstituted into the core of LDL by a minor modification of the method of Krieger et al. [42]. Briefly, 1.9 mg of dialyzed LDL was lyophilized with 25 mg starch in Siliclad-treated glass tubes. Then the LDL was extracted three times with 5 ml of heptane at —10°C. After aspiration of the last heptane extract, 6 mg of SiPc-BOA was added in 200 µl of benzene. After 90 minutes at 4°C, benzene and any residual heptane were removed under a stream of N₂ in an ice-salt bath. After about 45 to 60 minutes, completely dried reconstituted LDL was solubilized in 10 mM Tricine (pH 8.2) at 4°C for 24 hours. Starch was removed from the solution by low-speed centrifugation, and the specimen was further clarified by at least one or two additional centrifugations. SiPc-BOA-reconstituted LDL (abbreviated as r-SiPc-BOA-LDL) was stored under N₂ at 4°C. The Gd-DTPA-SA was then intercalated into the phospholipid monolayer of r-SiPc-BOA-LDL (as described above).

Confocal Microscopy

Laser scanning confocal microscopy studies were performed on HepG₂ cells to determine whether Gd-DTPA-SA(r-SiPc-BOA)-LDL was internalized through LDLRs. In brief, HepG₂ cells grown on four-well Lab-Tek chamber slides (Naperville, IL) were washed with preincubation medium containing the indicated amounts of Gd-DTPA-SA(r-SiPc-BOA)-LDL/r-SiPc-BOA-LDL and/or unlabeled LDL. After 4 hours' incubation at 37°C, the cells were washed with ice-cold PBS, fixed with 3% formaldehyde in PBS, and mounted and sealed for confocal microscopic analysis. Confocal microscopy was performed with a Leica TCS SPII laser scanning confocal microscope (Heidelberg, Germany). The filter settings were 633 nm for excitation and 638 to 800 nm for emission.

Experimental Animals and Induction of Ectopic Tumors

The following protocol was approved by the University of Pennsylvania Animal Ethics Committee. Adult female nude mice (26–29 g) were allowed free access to food and water throughout the study. HepG₂ cells (10 x 10⁶; overexpress LDLR) were inoculated subcutaneously into the left flanks of nude mice. Six weeks after tumor inoculation, mice were designated as either control or treated animals.

In Vivo MRI Imaging

Contrast agents Gd-DTPA-SA-LDL (170:1) or gadodiamide (Omniscan) (Amersham Health Inc, Princeton, NJ) were administered to mice through tail vein injection at a dose of 0.04 mmol/kg of body weight. Magnetic resonance imaging was performed before and 5 and 24 hours postcontrast injection.

Images were obtained with a 4.7-T horizontal-bore I NOVA spectrometer (Varian) equipped with a 12-cm-diameter gradient set having a maximum strength of 25 G/cm. Mice were anesthetized with a 1% isoflurane air mixture delivered through a nose cone. The core body temperature was monitored with a rectal probe connected to a small-animal monitoring system (SA Inc., Stony Brook, NY) that maintained the core temperature at 37 ± 0.1 °C by adjusting a stream of heated air in the bore of the magnet. Mice were placed prone on a fiberglass platform within a 50-mm-diameter birdcage volume coil operating at 200.8 MHz. Coronal and transverse T₁-weighted images were acquired through the abdomen (liver) and hind limb (tumor) of the mouse with a spin-echo sequence (TR/TE = 500/15 milliseconds, matrix = 256 x 128, and FOV = 4x2 cm, slice thickness = 1 mm, signal average = 4).

Imaging Data Analysis

Image contrast enhancement measurements were performed to quantify the amount of contrast agent in specific tissues. Contrast enhancement values were calculated by relating the pixel intensity values, I, of the target tissue (liver or tumor) to an unaffected tissue (skeletal muscle) according to the following equation:

$${\text{\% Contrast enhancement = (RI}}_{\text{post}}-{\text{RI}}_{\text{pre}})/{\text{RI}}_{\text{pre}}\times 100$$

where RI_post is the relative intensity (I_liver/I_muscle or I_tumor/I_muscle) after infusion, and RI_pre is the relative intensity (I_liver/I_muscle or I_tumor/I_muscle) before infusion.

Statistical Evaluation

The results were expressed as mean ± standard error. Analysis of variance (ANOVA) with Tukey's multiple comparison post hoc testing was used for evaluation of differences between groups. Differences with a P value less than .05 were deemed significant.

Results

Preparation of Gd-DTPA-SA-LDL

Following the procedure of Jasanada et al., DTPA-SA was prepared with a 48% yield (0.45g). The purity of the product was confirmed by TLC and matrix-assisted laser desorption ionization-time of flight mass spectrometry. A dominant molecular ion peak at 894.5 (m/z) verified the product (DTPA-bis(stearylamide), MW = 896.4 g/mol). A stock solution of 3.0 mM DTPA-SA was added to a solution of LDL yielding a final DTPA-SA/LDL molar ratio of 200:1. Based on zinc dithizone assay, 70% to 90% intercalation efficiency was achieved, resulting in approximately 150 to 180 DTPA-SA molecules per LDL particle. Labeling ratios as high as 496:1 were subsequently achieved, but the in vivo studies were performed with chelate to protein ratios of ∼170:1. The structure of Gd-DTPA-SA-LDL is shown schematically in Figure 1.

Figure 1.

Structure of Gd-DTPA-SA-LDL.

Size Measurements

Light scattering Figure 2 depicts the particle size distribution of native LDL and Gd-DTPA-SA-LDL analogs at probe to protein ratios of 154:1, 334:1, and 496:1. One can see that the size distribution curves of the Gd-DTPA-SA-LDL analogs are closely superimposed on that of native LDL. Mean particle diameters were 27.7, 26.3, 27.2, and 30.7 nm for native LDL, Gd-DTPA-SA-LDL 154:1, 334:1, and 496:1, respectively. The size distribution of LDL particles with up to 334 Gd-DTPA-SA chelates was similar to that of native LDL but at higher payloads (496) there was a slight right sift of the particle size distribution.

Figure 2.

Particle size distributions of native LDL and Gd-DTPA-SA-LDL analogs with various Gd-DTPA-SA/LDL ratios.

Transmission electron microscopy Negative staining electron microscopy of native LDL and Gd-DTPA-SA-LDL revealed uniform spherical particles of comparable size (Figure 3). In Table 1, measurements of particle size diameter revealed that Gd-labeled LDL at payloads of 230:1 and 496:1 (25.0 ± 4.0 and 25.7 ± 2.8 nm, respectively) were larger than native LDL (21.2 ±2.5 nm) (P < .0001).

Figure 3.

Electron microscopy of native LDL (A) and Gd-DTPA-SA-LDL (230:1 and 496:1) (B and C, respectively).

Table 1.

Diameter Measurements of Native and Gd-DTPA-SA-LDL.

	Diameter (nm)	N
Native LDL	21.2 ± 2.5	31
Gd-DTPA-SA-LDL (230:1)	25.0 ± 4.0*	30
Gd-DTPA-SA-LDL (496:1)	25.7 ± 2.8*	30

^*P< .0001 versus native LDL.

Agarose Gel Electrophoresis

Agarose gel electrophoresis was performed on Gd-DTPA-SA-LDL particles with payloads of 245:1 and 450:1 (Figure 4). Native and acetylated LDL served as controls. Gd-DTPA-SA-LDL (245:1) displayed an identical migratory pattern as native LDL, whereas Gd-DTPA-SA-LDL (450:1) migrated slightly further than native LDL. As expected, acetylated LDL showed a marked increase in its electrophoretic mobility. These findings indicate that intercalating up to 245 Gd-DTPA-SA moieties into the LDL phospholipid monolayer does not alter the valence or surface charge density of LDL. At higher payloads (450 Gd-DTPA-SA) the LDL particles exhibit a slight increase in electrophoretic mobility.

Figure 4.

Agarose gel electrophoresis of native and modified LDL species. 1. Acetylated LDL; 2. Gd-DTPA-SA-LDL (245:1); 3. Gd-DTPA-SA-LDL (450:1); 4. Native LDL.

Relaxivity Measurements

The r₁ relaxivity for Omniscan and Gd-DTPA-SA-LDL at various temperatures are listed in Table 2. The relaxivity of Omniscan was slightly higher at 21 °C (4.2 mM^-1 s^-1) compared to that at 40°C (3.5 mM^-1 s^-1). These values are consistent with previous temperature-dependent r_r measurements of Omniscan. The relaxivity per Gd(III) ion of Gd-DTPA-SA-LDL did not show a temperature-dependent effect (7.9 ± 0.22 and 8.1 ± 0.19 mM^-1 s^-1 at 21 and 40°C, respectively). Given these values and the fact that these LDL particle have approximately 180 amphiphilic Gd chelates, the r₁ relaxivity per LDL molecule is estimated to be 1440 mM^-1 s^-1.

Table 2.

T₁ Relaxivities of Omniscan and Gd-DTPA-SA-LDL.

Parameter	Contrast Agent
	Omniscan	Gd-DTPA-SA-LDL
Gd/molecule	1	180
Relaxivity/Gd, temperature
21 °C	4.2	7.9 ± 0.22
40°C	3.5	8.1 ± 0.19
Relaxivity/molecule, temperature
21°C	4.2	1440*
40°C	3.5	1440*

Relaxivity measurements were performed at 60 MHz. T₁ relaxivity of contrast agents are expressed in millimoles per liter per second. Data for Gd-DTPA-SA-LDL is expressed as mean ± standard error (n = 3 and 4 for measurements at 21 and 40°C, respectively).

^*Estimated T₁ relaxivity of Gd-labeled LDL molecule with 180 amphiphilic Gd chelates.

Assessment of LDLR-Mediated Uptake of Gd-DTPA-SA-LDL

In vitro MRI experiments Figure 5 shows representative T_1- weighted sagittal images of HepG₂ and ldlA7 cells after 5 hours' incubation with Gd-DTPA-SA-LDL (0.44 mM). In these images, the cell pellets can be distinguished from the overlaying PBS. The cell pellet of HepG₂ treated with Gd-labeled LDL showed higher signal intensity (23% higher) compared with untreated HepG₂ cells. Conversely, the low signal intensity detected in treated ldlA7 did not differ from untreated ldlA7 controls. When the corresponding Gd-DTPA-SA-LDL-treated groups were compared, marked signal enhancement (34% higher) is seen in HepG₂ over ldlA7. Finally, for HepG₂ cells treated with Gd-labeled LDL and a six-fold excess of native LDL, no signal enhancement was seen in the cell pellet. This indicates that native LDL effectively inhibited the uptake Gd-DTPA-SA-LDL in HepG₂ cells.

Figure 5.

T₁-weighted saggital spin-echo images of HepG₂ and ldlA7 cells after a 5-hour incubation with Gd-DTPA-SA-LDL. Cells not incubated with Gd-DTPA-SA-LDL served as untreated controls. Treated cells were incubated with 0.44 mM Gd-DTPA-SA-LDL. The inhibition study was conducted with a six-fold excess of native LDL.

Confocal microscopy experiments During the reconstitution process 338 molecules of SiPc-BOA were incorporated into the core of LDL. Thereafter, 160 molecules of Gd-DTPA-SA were intercalated into the phospholipids monolayer of the SiPc-BOA-LDL particle.

Figure 6 shows the confocal fluorescent images of HepG₂ cells incubated with the diagnostic probes under various conditions. Figure 6, A and Á displays images of cells alone (no nanoparticle), which demonstrate the absence of significant background fluorescence in the cells. When the cells were incubated with 240 g/ml Gd-DTPA-SA(r-SiPc-BOA)-LDL at 37°C for 4 hours, marked fluorescence was observed within the cells' cytosol (Figure 6, B and ), indicating active uptake and internalization. Similarly, in the presence of LDL reconstituted with 240 g/ml of SiPc-BOA, intense intracellular fluorescence was also observed (Figure 6, D and ). Finally, competition experiments were performed in which HepG₂ cells were incubated with 240 g/ml Gd-DTPA-SA(r-SiPc-BOA)-LDL and with a 25-fold excess of unlabeled native LDL. All intracellular fluorescence was inhibited under these conditions (Figure 6, C and Ć).

Figure 6.

Confocal fluorescent images of HepG₂ cells incubated with Gd-DTPA-SA(r-SiPc-BOA)-LDL probes (A,B,C,D) at 37°C for 4 hours as well as the corresponding bright field images (Á, Ć, ). A and Á, cell alone control; B and , cell + 240 µg/ml Gd-DTPA-SA(r-SiPc-BOA)-LDL; C and Ć, cell + 240 µg/ml probe + 25-fold excess of native LDL; D and , cell + 240 µg/ml r-SiPc-BOA-LDL

In vivo MRI experiments Representative T₁-weighted axial images through the abdomen and hind limbs from a tumor-bearing mouse after IV administration of Gd-DTPA-SA-LDL (170:1) (0.04 mmol/kg) are displayed in Figure 7 and coronal images are shown in Figure 8. In precontrast images, there was little intrinsic signal contrast between the liver parenchyma/dorsal thoracic muscle and tumor/leg muscle. At 5 hours postcontrast administration, marked signal enhancement was apparent in the liver, whereas minimal contrast enhancement was exhibited by the tumor. By 24 hours, signal enhancement within the tumor increased dramatically and clear demarcation of the boundary of the tumor was evident. These levels of tissue contrast were sustained through 36 hours.

Figure 7.

T₁-weighted axial spin-echo images through the abdomen and lower flank of a nude mouse with a subcutaneously implanted HepG₂ tumor. Images are from the mouse before administration of Gd-DTPA-SA-LDL (precontrast) and at various times after the IV administration of MR contrast agent (5, 24, and 36 hours). Long arrow indicates tumor; short arrowhead indicates liver parenchyma.

Figure 8.

T₁-weighted coronal spin-echo images of a nude mouse with a subcutaneously implanted HepG₂ tumor. The sequence of images is the same as that listed in Figure 5.

Control experiments were also performed in which the commercial Gd-DTPA-diamide chelate, Omniscan/gadodiamide, was administered to a tumor-bearing mouse. Figure 9 shows the MR images before and 5 and 24 hours after IV injection of gadodiamide (0.04 mmol/kg body weight). Five hours after administration of gadodiamide minimal signal enhancement was detected in the liver, as most of the contrast agent was expected to be cleared. At this time, no enhancement was evident in the tumor. By 24 hours, the signal in the liver returned to the precontrast levels and the tumor remained unenhanced. At 5 hours, accumulation of gadodiamide within the mouse's urinary bladder was evident. Given that gadodiamide is rapidly cleared from the body within 3 to 5 hours by renal elimination, these findings are expected.

Figure 9.

T₁-weighted axial spin-echo images through the abdomen and lower flank of a nude mouse with a subcutaneously implanted HepG₂ tumor. The images are from the mouse before administration of gadodiamide (precontrast) and at 5 and 24 hours after the IV administration of MR contrast agent. Long arrow indicates tumor; short arrowhead indicates liver parenchyma.

Table 3 presents the calculated mean contrast enhancement values from the in vivo MRI experiments. At 5 hours, significantly higher contrast enhancement was detected in the liver after Gd-DTPA-SA-LDL administration compared to Omniscan treatment (55.40 ± 6.20% and 6.28 ± 5.38%, respectively). Higher enhancement values were also observed in tumors with Gd-DTPA-SA-LDL versus Omniscan (9.40 ± 3.12% vs 0.63 ± 0.63%, respectively). At 24 hours, enhancement of the liver decreased to about half its 5-hour value (25.60 ± 4.50%), but it was still significantly higher than that of Omniscan controls (3.33 ± 3.33%). Concurrently, tumor contrast enhancement rose markedly in animals given Gd-DTPA-SA-LDL (25.60 ± 2.50%) and was significantly greater than the enhancement of Omniscan-treated controls (0.57 ± 0.57%).

Table 3.

Percent Contrast Enhancement after Administration of Omniscan or Gd-DTPA-SA-LDL (0.04 mmol/kg).

Tissue	5 hours		24 hours
	Omniscan (n = 3)	Gd-LDL (n = 5)	Omniscan (n = 3)	Gd-LDL (n = 5)
Liver	6.28 ± 5.38	55.40 ± 6.20*	3.33 ± 3.33	25.60 ± 4.50*
Tumor (HepG2)	0.63 ± 0.63	9.40 ± 3.12	0.57 ± 0.57	25.60 ± 2.50*

Percent enhancement is relative to precontrast levels in respective tissues. Values are means ± standard error (n = 5). Gd-DTPA-SA-LDL payload 170:1.

^*P < .05 versus respective control (Omniscan) group.

Discussion

The capability of actively targeted contrast agents to serve as biological markers and surrogate endpoints plays a central role in molecular imaging. To date, a variety of conjugated ligands, which include antibodies, peptides, and peptidomimetics, have been used to home synthetic nanoparticles to specific biological targets [11–13,15,16]. The LDL nanoparticle has a “built-in” targeting component. ApoB-100 is a surface transmembrane glycoprotein that directs LDL to cells expressing LDLR. Through electrostatic interactions between cationic amino acid clusters on ApoB-100 and LDLR, a series of events are activated to initiate receptor-mediated endocytosis and the delivery of the LDL cargo to the cell. Whereas cholesterol and triglycerides are the natural payloads for the LDL system, in the present study we have modified LDL by intercalating an amphiphilic Gd-DTPA chelate into the LDL phospholipid monolayer thereby providing a means for selective delivery of Gd(III) to LDLR-expressing tissues.

After surface modification of LDL, retention of avidity for LDLR is essential if selective delivery of paramagnetic contrast agents is to be observed. Unlike nuclear medicine where only a few molecules of contrast agent are needed for detection [10], a much greater number of paramagnetic probes must be conjugated onto the nanoplatform due to the lower sensitivity of MRI. This is a particularly important issue, as extensive surface modification of LDL can alter the affinity of this lipoprotein for LDLR [43,44]. The first series of experiments were performed to examine the structure and functional properties of the Gd-labeled LDL nanoparticle. Those methods that measured the physical dimensions of the particles (light-scattering photon correlation spectroscopy and transmission electron microscopy) provided slightly different findings. Light-scattering measurements revealed that the mean particle size and distribution of Gd-labeled LDL was similar to native LDL. Meanwhile, electron microscopy indicated that the diameter of Gd-labeled LDL was larger than that of unlabeled LDL. Additional experiments that examined the surface charge and hydrodynamic properties of the Gd-LDL nanoparticles (agarose gel electrophoresis) showed that intercalation of amphiphilic Gd-DTPA chelates (up to 245 molecules) into the LDL phospholipid monolayer did not alter the valence or surface charge density of LDL. This is not surprising because the Gd-DTPA-SA group is neutral and because Apo-B100 probably contains most of the charged groups in the particle. Not only were the physical properties of Gd-labeled LDL unaltered relative to its native counterpart, but both MRI and confocal microscopy cell experiments demonstrated that Gd-DTPA-SA-LDL retained its affinity for LDLR. The absence of particle uptake by ldlA7 cells (LDLR-deficient cells) and its clear inhibition by native LDL in HepG₂ cells provide strong evidence that Gd-labeled LDL selectively binds to and is taken up through LDLR. Hence, binding of the Gd-chelating group did not distort or interfere with the receptor-binding moieties on the protein component of the particle, as it should not if the chelates intercalated into the phospholipid monolayer as they were designed to do.

The remaining issue was determination if the uptake of Gd(III) was sufficient for selective enhancement/visualization of LDLR-expressing tissues in vivo. Figures 6 and 7 show that LDLR-containing tissues (liver and HepG₂ tumors) displayed significant MRI contrast enhancement after the IV administration of Gd-labeled LDL, albeit with different kinetics. Hepatic enhancement was seen early (5 hours) and tumor contrast became evident 24 hours postinjection, whereas liver enhancement diminished over the later time interval to about half its initial value. The differences in the “time to enhancement” between the two tissues can be explained by the differences in tissue mass and blood flow. The liver, a vital organ in mammals that occupies much of the upper abdomen, receives 25% of the cardiac output [45]. Thus, a large portion of the injected dose of Gd-labeled LDL will be readily accessible to the LDLR on the surface of hepatocytes. As such, high concentrations of Gd-labeled LDL will be sequestered within the liver within a short period. Conversely, the HepG₂ tumor xenograft accounted for only approximately 1% of the animal's body weight and is supplied by sporadic and tortuous blood vessels. Under these conditions, a longer time would be required for sufficient concentrations of Gd-labeled LDL to reach and be taken up by the tumor for in vivo detection. By 24 hours, significant accumulation of Gd-labeled LDL is seen in the tumor, whereas in the liver, active clearance of the amphiphilic Gd complex (through biliary elimination) is under way such that the percent contrast enhancement between the two tissues are approximately equal at this point.

Both the in vitro and in vivo findings of the present study demonstrate that LDL is an appealing nanoplatform for MR molecular imaging. A similar approach was undertaken by Frias et al. [46], who performed a fine study utilizing paramagnetic HDL-like particles to image atherosclerotic plaques in apoE knockout mice. Close examination of the LDL system reveal intriguing aspects of this particulate imaging platform. First, amphiphilic paramagnetic chelates (DTPA) can be incorporated into the phospholipid monolayer of LDL without interfering with targeting of the nanoparticle. The orientation of these chelates on LDL exposes the Gd(III) ions to the surrounding aqueous environment where they can interact with local water molecules. In the present study, up to 180 of these chelates were intercalated into one LDL particle. Typically, incorporation of Gd(III) chelates onto such macromolecules drastically increases the relaxivity of the individual Gd(III) ions [47–49]. This occurs as the Gd(III) chelate adopts the slower rotational correlation time of the macro-molecule [50]. Interestingly, the relaxivity of the amphiphilic Gd(III) chelate on LDL only showed a marginal increase over the small-molecule Gd(III) chelate, Omniscan. In addition, temperature-dependent changes in the relaxivity of the amphiphilic Gd(III) chelates on LDL were also not observed. This result would suggest that conflicting factors are involved that limit the relaxivity of these particles. These factors may include 1) high internal flexibility of the amphiphilic Gd(III) chelate on LDL and/or 2) reduced access of water to the Gd(III) center. Given that the two long sterylamide fatty acyl chains anchor the Gd-DTPA into the LDL phospholipid monolayer, it is unlikely that the Gd(III) chelate will experience a high degree of rotational freedom. However, water exchange at the Gd(III) centers near the phospholipids surface of LDL may be diminished. Unlike other liposomal/microemulsion/micelle particulate MRI contrast agents, LDL contains the large ApoB-100 at its surface. At a molecular weight of 550 kDa, ApoB-100 is one of the largest monomeric proteins known [51]. Containing both hydrophilic and hydrophobic domains, ApoB-100 spans the phospholipids monolayer and occupies between 40% and 60% of the surface area of LDL [51]. Due to the close proximity of the Gd(III) chelate to the hydrophilic portions of ApoB-100 that extend into the aqueous surface of LDL, electrostatic interactions between these components may occur, limiting the access of water to and from the Gd(III) center. Consequently, such interaction can significantly reduce the rate of water exchange. Similar “quenching” effects have previously been reported for other Gd(III) complexes bound to macromolecules [52–54]. Although apoA-1 is much smaller than ApoB-100, such interactions may also explain the moderate relaxivities reported for HDL-like paramagnetic nanoparticles [46]. Despite this shortcoming, the relaxivity of the entire Gd-DTPA-SA-LDL molecule is still quite high (estimated 1440 mM^-1 s^-1). The higher relaxivity of the Gd-LDL complex should provide signal enhancement at much lower local concentrations than traditional low molecular weight paramagnetic chelates.

Another interesting aspect of the LDLR system is its receptor-mediated uptake mechanism. Receptor-mediated endocytosis is a highly specific/high-capacity process that can concentrate many molecules of LDL inside the cell within a relatively short time [22]. Once internalized, LDLR dissociates from LDL and is recycled back to the cell surface where it is available to interact with many more Gd-LDL over its lifetime [22]. With a recycle time of approximately 10 minutes and a lifetime of about 24 hours, one can assume that each receptor mediates the transport of approximately 144 LDL particles into LDLR-expressing cells each day. Assuming approximately 1000 receptors per cell, this provides an extremely efficient system for delivery of its natural cargo of cholesterol (∼1500 cholesterol esters plus additional free cholesterol) and fatty acids. This system offers an attractive means for targeted delivery of contrast agents, as this process of receptor recycling provides an ample mechanism for signal amplification. However, one must be aware that downregulation of LDLR can alter the uptake kinetics of Gd-DTPA-SA-LDL. The cell surface expression of LDLR is regulated by cellular cholesterol concentrations [55], thus, on first pass extraction of exogenous LDL intracellular concentrations of cholesterol may rise, causing a negative feedback “retraction” of LDLR. Consequently, the uptake kinetics of these administered particles would be slower than that described above. One way of overcoming this limitation would be to replace/reduce the cholesterol-ester core with another lipophilic biomolecule, thereby avoiding the cholesterol induced LDLR down regulation.

The issue of nonspecific binding of LDL particles is also realized. Indeed, the intercalation method provides a versatile means of introducing contrast agents onto LDL; however, in this location amphiphilic molecules have a tendency to undergo nonspecific exchange with the outer phospholipids membrane of cells. To avoid this, it is essential to anchor the surface-bound chelate with a long, hydrophobic group that may penetrate into the underlying lipid core of the particle. Hook-shaped anchors consisting, for example, of cis-linoleic acid as opposed to linear stearic acid are preferable. Nonspecific transfer to cell surface bilayers would dilute the labeling of LDL and reduce the specificity of LDLR targeting. Secondly, LDLR is not the only surface receptor with affinity for native LDL. The receptors LBS (lipoprotein binding site), SR-B1/CLA-1 (scavenger receptor class B type I/CD36- and LIMPII-analogous-1), and CD36 have all been shown to bind LDL and participate in the selective-uptake LDL-derived cholesterol [56–59]. Because the majority of these observations were made in cell culture, it remains uncertain how much these receptors contribute to the catabolism of LDL in vivo. Nonspecific uptake of Gd-LDL by the reticulum endothelium system in the liver may also interfere with targeted delivery of this probe. Pegylation of the particle may render it more “stealthy” and mitigate this problem. Clearly, further studies are required to investigate these non-LDLR uptake processes.

Despite these caveats the present study demonstrates that the current formulation of Gd-labeled LDL is a simple, natural, nonimmunogenic effectively targeted contrast agent that enables one to image the LDLR in vivo. Several clinical applications for Gd-labeled LDL can be proposed; Gd-LDL may be used in the localization and detection of tumors that overexpress LDLR. Nononcologic applications could include detection for early atherosclerotic disease [60], monitoring the efficacy of gene therapy for familial hypercholesterolemia [61] as well as providing an alternative means for contrast-enhanced imaging of the adrenal glands [62].

The opportunity for multimodality MR/optical imaging approaches was also demonstrated in this study. By introducing MRI and optical probes as separate components of the LDL platform, it offers the flexibility to adjust the concentrations of the individual agents such that variable payloads of Gd(III) and optical probes can be incorporated into LDL. This approach provides optimal diagnostic capabilities of these imaging modalities—the high resolution, deep penetration, and excellent soft tissue contrast of MRI and the high sensitivity, economy, and portability of optical fluorescence imaging. In addition, SiPc-BOA is a potent photosenitizer [63] suitable for photodynamic therapy.

The potential of incorporating traditional therapeutics into LDL has long been recognized and exploited [64,56,66]. Hydrophobic drugs such as doxorubicin, taxol, and vindesine have been incorporated into LDL by simple incubation or reconstitution methods [64–67]. This approach is not limited only to hydrophobic drugs; lipophilic derivatives of drugs of moderate hydrophilicity can also be prepared and incorporated into LDL [68,69]. Several studies demonstrated that LDL can serve as an effective carrier of anticancer drugs that is capable of significantly enhancing cytotoxicity in various tumor cells [67,70–73]. Introducing both imaging agents and therapeutics into LDL offers the possibility of having a truly multifunctional nanoplatform capable of detection and treatment by a single intervention.

Recently, our laboratory has shown that by conjugating homing ligands onto the surface of ApoB-100, LDL can be redirected to alternate cell surface receptors and epitopes [74]. This was demonstrated by conjugating folic acid to the lysine residues of ApoB-100; subsequent confocal studies demonstrated uptake by folate receptor-mediated pathways, whereas LDLR uptake was blocked. To date, a diverse set of disease-homing molecules exist that could potentially be used to functionalize LDL as a multifunctional targeting platform [17–20,75–77]. The design of such a versatile biocompatible nanoplatform would not only greatly expand the possible application of LDL nanoparticles in diagnostic imaging but also in the field of targeted therapeutics.

Abbreviations

ApoB

apolipoprotein B

DTPA-SA

diethylenetriaminepentaacetic acid-bis(stearylamide)

Gd

gadolinium

Gd-DTPA-SA-LDL

Gd-labeled low-density lipoprotein

HepG₂

human hepatoblastoma G₂

LDL

low-density lipoprotein

LDLR

low-density lipoprotein receptor

SiPc-BOA

tetra-t-butyl silica phthalocyanine bisoleate