Elucidating the Structural Organization of a Novel Low-Density Lipoprotein Nanoparticle Reconstituted with Docosahexaenoic Acid

Abstract

Low-density lipoprotein nanoparticles reconstituted with unesterified docosahexaenoic acid (LDL-DHA) is promising nanomedicine with enhanced physicochemical stability and selective anticancer cytotoxic activity. The unique functionality of LDL-DHA ultimately relates to the structure of this nanoparticle. To date, however, little is known about the structural organization of this nanoparticle. In this study chemical, spectroscopic and electron microscopy analyses were undertaken to elucidate the structural and molecular organization of LDL-DHA nanoparticles. Unesterified DHA preferentially incorporates into the outer surface layer of LDL, where in this orientation the anionic carboxyl end of DHA is exposed to the LDL surface and imparts an electronegative charge to the nanoparticles surface. This negative surface charge promotes the monodisperse and homogeneous distribution of LDL-DHA nanoparticles in solution. Further structural analyses with cryo-electron microscopy revealed that the LDL-DHA nanostructure consist of a phospholipid bilayer surrounding an aqueous core, which is distinctly different from the phospholipid monolayer/apolar core organization of plasma LDL. Lastly, apolipoprotein B-100 remains strongly associated with this complex and maintains a discrete size and shape of the LDL-DHA nanoparticles similar to plasma LDL. This preliminary structural assessment of LDL-DHA now affords the opportunity to understand the important structure-function relationships of this novel nanoparticle.

1. Introduction

The low density lipoprotein (LDL) is a natural macromolecular assembly responsible for the transport of cholesterol and other lipids in the plasma (Goldstein and Brown, 1976). It exists as a quasisperical particle that is approximately 21 nm in diameter. Overall the LDL particle is organized into two major domains: namely a central apolar core of cholesteryl esters and minor amounts of triglycerides surrounded by an amphipathic shell consisting of a phospholipid monolayer, free unesterified cholesterol and a single molecule of apolipoprotein B-100 (apoB-100) (Hevonoja et al., 2000; Prassl and Laggner, 2009). Consisting of 4536 amino acid residues, apoB-100 is one of the largest monomeric proteins known (Knott et al., 1986; Yang et al., 1986). The apoB-100 serves not only to maintain the structural integrity of this assembly, but it also acts as a protein ligand that is recognized by cell surface LDL receptor (LDLR) and thus mediates the removal of LDL from the circulation by LDLR expressing tissues (mainly liver, reproductive organs, adrenal glands) (Goldstein and Brown, 1976). LDL particles and the LDLR delivery system provides an efficient and highly regulated means of transporting cholesterol and maintaining cholesterol homeostasis in the body. In many ways this system models many of the coveted attributes modern scientist strive to mimic in their nanoparticle design. First, LDL maintain a defined nanoscale size (18–25 nm) (Prassl and Laggner, 2009), these carriers have a high cargo carrying capacity (>1600 lipid molecules) (Shen et al., 1977), their cargo delivery is mediated by specific receptor recognition, lastly, being an endogenous assembly, LDL can escape recognition as a foreign entity by the immune system and evades absorption by the reticuloendothelial system (Gotto et al., 1986). Furthermore, many cancers overexpress LDLR and avidly take up plasma LDL to a greater degree over surrounding normal tissues (Chen and Hughes-Fulford, 2001; Maletinska et al., 2000; Peterson et al., 1985; Tatidis et al., 2002). For this reason many scientists have attempted to exploit the LDL system to preferentially deliver diagnostic and therapeutic agents to cancer (Allijn et al., 2013; Corbin et al., 2006; Jin et al., 2012; Kader and Pater, 2002; Masquelier et al., 2006; Masquelier et al., 2000b). In recent years, our group has engineered a novel LDL based nanoparticle for cancer treatment (Reynolds et al., 2014). Unlike traditional approaches that have attempted to formulate conventional chemotherapeutics in LDL (yielding complexes with poor stability) (Masquelier et al., 2000b), we reconstituted LDL with the natural omega-3 fatty acid, docosahexaenoic acid (DHA), referred to as LDL-DHA (Reynolds et al., 2014). Reformulating LDL with a natural lipid cargo produced stable LDL particles with high payloads of DHA. The LDL-DHA closely resembled plasma LDL in terms of overall morphology and size, however the LDL-DHA unexpectedly possesses enhanced physical stability and greater oxidative resistance compared to plasma LDL (Reynolds et al., 2014). Even when challenged with increasing temperatures the LDL-DHA nanoparticles displayed significantly higher thermal stability over its native counterpart. Furthermore, this nanoparticle also shows pronounced therapeutic promise as a novel anticancer agent. LDL-DHA nanoparticles were shown to be selectively cytotoxic towards cancer cells and are able to kill cancer cells at doses that do not harm normal cells [18, (Moss et al.). This selectivity is also evident in vivo, as a single locoregional treatment of LDL-DHA nanoparticles was able to induce pronounced necrosis (>80%) and retard the growth (3 fold) of orthotopic liver tumors in the rat (Wen et al., 2016). In contrast to this cancer selective cytotoxic activity, studies from our group have also shown that therapeutic doses of LDL-DHA can produce beneficial and protective molecules in normal tissues (Mulik et al., 2016). The LDL-DHA nanoparticles function unlike any other lipoprotein based particle described to date. Presently little is known about the structural organization of the LDL-DHA nanoparticles. Fundamental structure/function based studies maybe be able to provide further understanding and insight into the unique physical and bioactive properties of LDL-DHA. Many of the LDL-drug/contrast agent assemblies reported in the literature have not been extensively characterized in this manner. The goal of the present study is to conduct an in depth investigation into the chemical and molecular organization of lipids within LDL-DHA nanoparticles. From our findings we propose a unique molecular model for the LDL-DHA nanoparticle.

2. Materials and methods

Preparation of reconstituted LDL nanoparticles

Human LDL was isolated from apheresis plasma of patients with familial hypercholesterolemia using sequential density gradient ultracentrifugation as described previously by Lund-Katz et al. (Lund-Katz et al., 1998). The core replacement method was used to incorporate unesterified DHA, triolein (TO) or DHA methylester (DHAMeEster) (NuCheck Prep. INC, Elysian, MN) into the LDL particle (Krieger et al., 1979a). Briefly, LDL was freeze dried in the presence of starch and was subject to several rounds of organic extraction to remove nonpolar lipids from LDL. Thereafter, DHA/TO/DHAMeEster (solubilized in heptane) was added to the LDL residue and incubated at 4°C for 90 min. Heptane was then removed by nitrogen gas and the sample was re-suspended in 10 mM tricine buffer. Finally the sample was clarified by low-speed centrifugation, and stored under N₂ atmosphere at 4°C.

LDL-uniformly ¹³C labeled DHA ([U-¹³C]DHA) and LDL-([U-¹³C]DHAMeEster) nanoparticles were also prepared for the following experiments. For these nanoparticles [U-¹³C]DHA and [U-¹³C]DHAMeEster (kindly provided by Dr. Anthony Windust, National Research Council, Canada) were incorporated into LDL utilizing the reconstitution methods described above.

Characterization of LDL Nanoparticles

Composition

The composition of the LDL nanoparticles was assayed for phospholipids (ferrothiocyanate reagent method)(Stewart, 1980), free fatty acids (reverse phase-HPLC)(Mehta et al., 1998) and protein (Bradford protein assay kit-Sigma). Molar concentrations of each component in the LDL particle was determined based on the assumption that one copy of the apoB-100 protein is present per LDL particle.

Mean particle size and zeta potential

The size distribution and zeta potential of LDL nanoparticles was measured in Tricine buffer at pH 8.4 using a Zetasizer Nano (ZEN3500, Malvern Instruments, UK). All measurements were performed at 25°C temperature in triplicates. Values are expressed as mean ± S.E. from three replicate samples.

Agarose electrophoresis

The electrophoretic properties of LDL particles were examined by 1.0% agarose gel electrophoresis. Samples (10 μg) were pre-stained with Sudan Black stain (5:1 v/v) at 37°C for 30 min and finally centrifuged at 2000 rpm for 5 min. Pre-stained samples were applied to gel wells and allowed to penetrate into the gel for 5 min before the electric field was applied. Electrophoresis was performed at a voltage of 100V at 25 °C in barbital buffer (pH 8.6, 0.05 ionic strength) for one hour. The migration of the LDL particles was visualized by Sudan Black staining.

¹H and ¹³C NMR Spectroscopy

A seventy-five microliter aliquot of the concentrated stock of LDL nanoparticles was added to 0.215 mL of deuterated water (D₂O) (99.9%; Cambridge Isotope Laboratories, Inc., Tewksbury, MA). The sample was then brought to a total volume of 0.300 mL with the addition of 10μL of a 30mM solution of trimethylsilyl propionate (TSP) (Sigma Chemical Co., St. Louis, MO). The sample was adequately mixed and transferred to a 3-mm NMR tube for ¹H NMR spectroscopy.

In addition, aqueous solutions of DHA were prepared for NMR by solubilizing 1 mg of DHA in 0.250 mL of alkaline D₂O pH 8.5(via titration with 1M potassium hydroxide solution). TSP (10μl) was also added to the DHA solution and samples were prepared for NMR spectroscopy as described above.

The ¹H-NMR spectra of LDL nanoparticles or DHA solutions were obtained using Agilent VNMRS Direct Drive Console using 3mm broadband NMR probe (Agilent Technologies) at a frequency of 600 MHz with the following parameters: pulse 45°; sweep width, 9615 Hz; relaxation delay, 0.5 s; acquisition time, 4 s; with 256 scans at 25°C. Chemical shifts were referenced internal standard TSP as an internal standard. The ¹H-decoupled ¹³C-NMR spectra were obtained at a frequency of 150 MHz; 45°pulse length; 29069 Hz sweep width, 0.05s relaxation delay, 2.5 s acquisition time; 25546 number of scans,; at 25°C with proton decoupling.

For the proton NMR experiments which employed Eu³⁺ as a chemical shift and broadening reagent, small volumes of EuCl₃ (Strem Chemicals, Newbury, MA) containing D₂O (25μM) were added to the LDL nanoparticle samples. The sample was vortexed gently during the addition of the paramagnetic ions in order to obtain rapid mixing. Samples were allowed to sit for 15 mins before performing ¹H NMR spectroscopy.

Annular and Bulk Fluidity Measurements

The fluidity (viscosity/lipid packing) of lipids adjacent to ApoB100 protein (annular fluidity) and bulk fluidity in the LDL nanoparticles was measured using fluorometry methods. Measurements of resonance energy transfer from tryptophan of the LDL particles to pyrene and pyrene monomer-excimer formation was used to determine the annular and bulk fluidity respectively (Avdulov et al., 1997; Hashimoto et al., 2001; Sibmooh et al., 2004). Briefly, 10 mM PBS (pH 7.4) was added to the LDL samples to make a final volume of 2 mL containing 50 μg/ml protein. Samples were incubated at 37°C for 30 minutes in the dark and then tryptophan fluorescence at 278 nm and 334 nm for excitation and emission was measure respectively. Then, 2 μL of 1 mM pyrene (dissolved in dimethylformamide) was added to the samples and incubated for 1 minute at 37°C with gentle stirring. For measurement of annular fluidity, samples were excited at 278 nm, and the emission fluorescence intensity at 373 nm (pyrene monomer) and 480 nm (pyrene excimer) was measured. Taking into account that the Forster radius (the energy transfer-limiting distance) for the tryptophan—pyrene donor—acceptor pair is 3 nm (Tahara et al., 1992), only pyrene located in the annular lipid (close to proteins) will be excited, thus the fluidity of the annular lipids is considered proportional to the ratio of fluorescence intensity of pyrene excimer and pyrene monomer (F_e/F_m). For the fluidity of total or bulk lipids samples were excited at 334 nm, and the emission fluorescence intensities were recorded at 373 nm (pyrene monomer) and 480 nm (pyrene excimer). The F_e/F_m ratio under these setting was taken as the bulk fluidity (Avdulov et al., 1997; Hashimoto et al., 2001; Sibmooh et al., 2004).

Cryo-Electron Microscopy

LDL nanoparticle samples were prepared for cryo-electron microscopy (cryo-EM) by loading 2.5 μL of sample onto a carbon coated glow-discharged copper grid (Quantifoil grids R2/2, Electron Microscopy Science, Hatfield, PA). After a 60 second incubation the gird was mounted inside the Mark III Vitrobot chamber (FEI USA, Hillsboro, Oregon, USA), which was maintained at 4°C with greater than 90% humidity. The grid was blotted for 7 seconds using a standard Vitrobot Filter paper (Ø55/20 mm, Grade 595, Ted Pella, Inc, Redding, CA), immediately plunged into liquid ethane and stored in liquid nitrogen until EM examination. For cryoEM imaging, the grids were transferred into a JEOL JEM 2200FS Transmission Electron Microscope (TEM; JEOL USA, Inc., Peabody, MA) operating at 200 kV. A 30 eV energy filter was used for zero-loss imaging. Images were exposed with an electron dose of ~20 electrons per Ų and at defocus levels varying from −1.5 to −2.5 μm. Lastly, images were digitized directly with a Gatan UltraScan 1000XP CCD Camera at 25K magnification (Gatan, Inc., Pleasanton, CA).

DPH/TMADPH Quenching Experiment with 5,6-Carboxylfluorescein

Characterization of the core and surface properties of the LDL nanoparticles was investigated using the respective core and surface loading fluorescent probes 1,6-diphenyl-1,3,5-hexatriene (DPH) and 1-(4-(trimethylamino)phenyl)-6-phenylhexa-1,3,5-triene (TMA-DPH) (Ben-Yashar and Barenholz, 1991). Briefly, stock solutions of 0.005 M DPH in tetrahydrofuran (0.5 μL) and 0.025 M TMADPH in 2:1 tetrahydrofuran:methanol (0.5 μL) were added into the respective LDL nanoparticle samples. Next 2.5 μl of a 0.01 M methanolic solution of 5,6-carboxyfluorescein (5,6-CF) was added to each sample. All the samples were diluted to 500 μL with deionized water. The samples were gently vortexed and incubated at 37°C for approximately 50 minutes to allow the until fluorescence intensity of the fluorophores to plateau. Finally, the fluorescence spectrum of each sample was read between 400 nm to 500 nm using fluorescence spectrometer (Hitchi F-7000 Fluorescence Spectrometer, Hitachi, CA, USA) and absorbance maxima was noted. Percent fluorescence quenching of DPH and TMADPH by 5,6-CF was calculated using the following equation:

$$\%\text{fluorescence}\text{quenching}=(1\text{-F}/{\mathrm{F}}_0)\times 100$$

Where F is the fluorescence intensity of the fluorophore in the presence of 5,6-CF and F₀ is the fluorescence intensity of the fluorophore in the absence of 5,6-CF.

3. Results

Preparation and physicochemical characterization of LDL nanoparticles

The physicochemical properties of plasma LDL, LDL-DHA and LDL-DHAMeEster nanoparticles are presented in Table 1. Overall the reconstituted nanoparticles were similar to plasma LDL in regards to size and phospholipid content. In contrast, the LDL nanoparticles differed greatly in term of their cargo; plasma LDL transports a mixture of unesterified cholesterol, cholesterol esters and triglycerides, LDL-DHA and LDL-DHAMeEster are each uniformly loaded unesterified DHA and DHAMeEster, respectively. The LDL-DHA nanoparticles are able to incorporate > 1100 DHA molecules per particle, this is in keeping with our previous preparations (Reynolds et al., 2014). Similarly, close to 1900 DHA MeEsters are loaded into the LDL-DHAMeEster nanoparticle. Both of these findings are in keeping with the carrying capacity of plasma LDL which is typically known to transport >1600 lipid molecules (Shen et al., 1977). In terms of surface properties both plasma LDL and LDL-DHAMeEster had slightly negative zeta potentials around −8.6 and −13 mV respectively (Figure 1A). Conversely LDL-DHA displayed a more prominent electronegative zeta potential at −22 mV (Figure 1A). These findings were also validated by the increased electromobility of LDL-DHA compared to plasma LDL and LDL-DHAMeEster in agarose gel electrophoresis experiments (Figure 1B).

Table 1.

Composition and Physicochemical Properties of LDL Nanoparticles.

	LDL	LDL-DHA	LDL-DHAMeEster
ApoB-100 ¶	1	1	1
Phospholipid¶	681.4 ± 70.8	507.4 ± 71.0	451.8 ± 28.6
Cholesterol ¶	2958 ± 378*	ND	ND
Lipid Cargo ¶	**	1112.6 ± 23.9	1896.4 ± 428.4
Diameter (nm)	17.8 ± 0. 5	22.4 ± 0.71	21.4 ± 0.5
Surface Charge (mV)	-8.6 ± 1.3	-22.3 ± 2.7	-12.9 ± 0.8

Values are expressed as number of molecules per LDL particle.

^*Total cholesterol includes cholesteryl esters and free cholesterol. Literature values indicate that LDL typically carries between 1300–1600 cholestryl esters and 500–600 free cholesterol molecules.

^**LDL also carries about 170 triglyceride molecules. DHA typically makes up only 1% of the total fatty acid composition of LDL

Figure 1. Surface Charge Determination of LDL particles.

A. Zeta potential measurement using ZetaSizer Nano. ** indicates LDL-DHA is significantly different from plasma LDL and LDL-DHAMeEster at p<0.005 and p<0.001 respectively. B. Agarose Gel Electrophresis also showed significantly higher negative surface charge for LDL-DHA compared to plasma LDL and LDL-DHAMeEster based on electrophoretic migration of the particles.

¹H and ¹³C NMR Spectroscopy

Specific lipid moieties of plasma LDL were identified in the ¹H-NMR spectrum (Figure 2A)), these include multiple well resolved resonances arising from: C-18 of cholesterol (0.7 ppm), terminal methyl (0.9 ppm), methylene (1.3 ppm), allylic (2.0 ppm) and bis-allylic (2.8 ppm) protons from fatty acyl chains, and the prominent trimethylamine of phospholipid choline head group (3.2 ppm). Compared to plasma LDL, LDL-DHA nanoparticles displayed a unique ¹H NMR profile (Figure 2B). First, peaks arising from cholesterol are absent from the LDL-DHA ¹H NMR spectrum, due to the complete removal of neutral lipids from the LDL structure during the reconstitution process. Secondly, the incorporation of DHA in LDL produces characteristic resonances that arise from the distinct chemical structure of this omega-3 fatty acid. A prominent omega methyl peak resonates at 0.95 ppm slightly downfield from the regular terminal methyl peak. DHA is also a unique omega-3 fatty acid in having its C2 and C3 methylene groups interposed between a carboxyl group at C1 and a double bond at C4. As a result of this unique electromagnetic environment, DHA gives rise to distinct signals for the C2,3 methylene protons in the downfield region at 2.3–2.4 ppm. Lastly, as a result of the many alternating double bonds present in DHA, the bis-allylic signal in the LDL-DHA spectrum was significantly larger than that from native LDL. The unique resonances from the omega methyl and C2,3 methylenes of DHA in LDL-DHA were subsequently confirmed with the ¹H NMR spectra of: (i) organic extracts of LDL-DHA nanoparticles; and (ii) free DHA in CDCl₃ (Figures 2C,D). One distinct difference between the latter spectra and that from intact LDL-DHA nanoparticles (Figure 2B), is the sharp singlet arising from the C2,3 methylene of DHA in organic solvent versus the broad composite signal of the same protons seen in intact LDL-DHA suspended in Tris-buffer/D₂O. These differences in the NMR resonance profile of the C2,3 methylene of DHA suggest the influence of solvent effects, to this end the ¹H NMR spectrum of free DHA solubilized in alkaline D₂O was acquired (Figure 2E). Within D₂O the C2,3 methylene of DHA displayed as a well resolved quartet and triplet like pattern respectively. This NMR pattern was similar to that seen in the intact LDL-DHA nanoparticle spectrum, although within the nanoparticle the DHA C2,3 methylene signals were broad and poorly resolved.

Figure 2. Proton NMR of LDL Nanoparticles and DHA in Solution.

The ¹H-NMR spectra were obtained with 600 MHz Agilent NMR spectrometer using 3mm broadband NMR probe (Agilent Technologies). Chemical shifts were referenced internal standard Trimethylsilyl propionate (TSP). (A) Plasma LDL; (B) LDL-DHA; (C) LDL-DHA organic extract; (D) DHA in CDCl₃; (E) DHA in alkaline D₂O.

Further structural NMR analysis was performed with [U-¹³C]DHA, the LDL-([U-¹³C]DHA) and LDL-([U-¹³C]DHAMeEster) nanoparticles preparations (Figure 3). Carbon-13 NMR spectroscopy provided great detail on each of the carbons present in the DHA hydrocarbon chain. From the [U-¹³C]DHA ¹³C spectrum resonances for: the carbonyl-C1 (178 ppm); C20 (132 ppm); olefin carbons (C4,5,7,8,10,11,13,14,16,17,19 at 127–130 ppm); C2 ( 33.9 ppm); bisallylic methylene carbons (C6, C9, C12, C15, C18 at 25–27 ppm); C3 (22.8 ppm); C21 (20.9 ppm) and C22 (14.6 ppm) were all clearly resolved in CDCl₃ (Figure 3A). The ¹³C spectra from [U-¹³C]DHAMeEster in CDCl₃ (Figure 3D) and LDL-([U-¹³C]DHAMeEster) (Figure 3E) in Tris-buffer/D₂O were also well resolved. The ¹³C spectrum from LDL-([U-¹³C]DHA) (Figure 3B) differed from the previous spectra in that while resonances from intermediate to terminal carbons (C4-C22) were well resolved, those belonging to C1-C3 at 178, 33.9 and 22.8 ppm respectively were of low intensity and poorly resolved. Suspecting that solvent interactions may have caused these effects the ¹³C spectrum of [U-¹³C]DHA was also acquired in alkaline D₂O (Figure 3C). [U-¹³C]DHA in D₂O displayed broad poorly resolved resonances for C1-C3, similar to LDL-([U-¹³C]DHA). The intermediate and terminal carbons (C4-C22) were much better resolved than C1-C3, however not to the same degree as was seen in CDCl₃. Furthermore, addition broad low intensity resonances were present at 33, 28, 23 ppm and just upfield from C22 at 13.5 ppm.

Figure 3. ¹³C NMR Investigation of LDL Nanoparticles.

The ¹H-decoupled ¹³C-NMR spectra were obtained using Agilent VNMRS Direct Drive Console with 600 MHz Agilent NMR spectrometer using 3mm broadband NMR probe (Agilent Technologies). (A) DHA in CDCl₃; (B) DHA in alkaline D₂O; (C) LDL-DHA; (D) DHAMeEster in CDCl₃; (E) LDL-DHAMeEster. Arrows indicate the C1, 2 and 3 resonances of DHA.

Paramagnetic ion lanthanides (e.g. Europium⁺³) have traditionally been used to study the structure and orientation of lipid membranes. Here we performed ¹H-NMR experiments with LDL nanoparticles in the presence and absence of europium⁺³. Figure 4A,B shows that in the presence of europium⁺³ a noticeable change is observed in the LDL-DHA spectrum. While theω-CH₃ resonance remained unchanged, the characteristic DHA C2,3 peak at 2.4 ppm preferentially underwent pronounced broadening. This finding suggests that this portion of the DHA molecules was available for binding to the aqueous lanthanide ions. Identical experiments performed with LDL-DHAMeEster showed that all the peaks in the ¹H NMR spectrum, including that of DHAMeEster C2,3 protons remained unchanged in the presence of europium⁺³ (Figure 4C,D). This indicates that no segments of the DHAMeEster was available to interact with europium⁺³.

Figure 4. Structural Elucidation: Effect of Lanthanide Shift Reagent on LDL Nanoparticles.

The 1H-NMR spectra of LDL nanoparticles with 600 MHz Agilent NMR spectrometer using 3mm broadband NMR probe (Agilent Technologies). Lanthanide shift effects were induced by the addition of EuCl₃ (25 μM). (A) LDL-DHA; (B) LDL-DHA + EuCl₃; (C) LDL-DHAMeEster; (D) LDL-DHAMeEster + EuCl₃.

Membrane Fluidity Measurements

The membrane surface fluidity of the LDL nanoparticles was determined by fluorescence energy transfer from tryptophan to pyrene (Figure 5) (Avdulov et al., 1997; Hashimoto et al., 2001; Sibmooh et al., 2004). The incorporation of DHA into the LDL particle increased both annular and global fluidity. The effect of DHA was greater on the annular fluidity (~8x) than on the global fluidity (~3x). Conversely, the incorporation of DHAMeEster into LDL did not cause any significant changes the either of the membrane fluidity parameters. Collectively, these results suggest that DHA partitions more towards the surface, thus effecting its membrane fluidity, while DHAMeEster partitions greater into the core of the LDL structure.

Figure 5. Determination of Membrane Fluidity in LDL nanoparticles: FRET Experiment.

Based on resonance energy transfer from tryptophan of lipoproteins to pyrene and pyrene monomer-excimer formation, the annular and bulk fluidity was determined. For measurement of annular fluidity, samples were excited at 278 nm, and the emission fluorescence intensity at 373 nm (pyrene monomer) and 480 nm (pyrene excimer) was measured. For bulk fluidity measurement the samples were excited at 334 nm, and the emission fluorescence intensity at 373 nm (pyrene monomer) and 480 nm (pyrene excimer) was recorded. The respective Fe/Fm ratio was taken as the measure of annular/bulk fluidity. Native LDL, LDL-DHA and LDL-DHAMeEster were compared with each other. (*) represent a significant difference from the corresponding groups at p<0.005.

Cryo-Electron Microscopy

LDL nanoparticle morphology and ultrastructure was analyzed via cryo-EM (Figure 6). CryoEM micrographs of native LDL embedded in vitreous ice presented as ellipsoid structures with internal striations (2.5±0.24nm apart). This striated projection depicts the liquid-crystalline core of LDL in which the cholesteryl esters are arranged in a stacked pattern. The phase transition temperature for cholesteryl esters is high (31°C), thus as these samples are prepared at room temperature their core lipids adopt a highly ordered state. The cryoEM images of LDL-TO and LDL-DHAMeEster nanoparticles both displayed heterogeneous sized quasi-spherical structures with high density cores devoid of striations. The core lipids of LDL-TO and LDL-DHAMeEster are believe to exist in a isotropic liquid state due to their low phase transition temperatures, as a result the lipids in these nanoparticles exist in a random unordered state and project as dense structures. CryoEM images of LDL-DHA were markedly different from plasma LDL, LDL-TO and LDL-DHAMeEsther. While these nanoparticles retained a quasi-spherical morphology with a densely demarcated periphery and a low density center, at higher magnification the densely stained outline of the nanoparticle could be seen to consist of two distinct layers separated by 4.24±0.07nm. Collectively, the micrographs and measurements indicate that LDL-DHA is organized as a phospholipid bilayer nanostructure.

Figure 6. Cryo-Electron Microscopy of LDL nanoparticles.

The images were taken using JEOL JEM 2200FS Transmission Electron Microscope with a Gatan UltraScan 1000XP CCD Camera at 25K magnification. (A) Plasma LDL; (B) LDL-TO; (C) LDL-DHAMeEster; (D) LDL-DHA. Scale bar equal to 50 nm.

Fluorescence quenching with 5,6-CF

Resonance energy transfer (RET) experiments were also performed to further interrogate the structure of the LDL nanoparticles. In brief, we used the fluorescent probes DPH, which partitions between surface monolayer and the core of lipoproteins, and TMADPH, the charged derivative of DPH which anchors at the monolayer surface. Upon the addition of the water soluble 5,6-CF acceptor to the solution of probe labeled LDL nanoparticles fluorescence quenching of the fluorophores will occur as a function of the distance between the donor (DPH or TMADPH) and the acceptor (5,6-CF). From these experiments it was found that TMADPH was quenched to a significantly greater degree than DPH in plasma LDL and LDL-DHAMeEster. Conversely, for the LDL-DHA nanoparticles quenching was observed equally for DPH and TMADPH (Figure 7). These findings indicate that in plasma LDL and LDL-DHAMeEster DPH and TMADPH partition to different compartments in the LDL particles, whereas in LDL-DHA these fluorophores localize into the same compartment.

Figure 7. Fluorescence Quenching Experiments in LDL nanoparticles.

All the samples were read between 400 nm to 500 nm using fluorescence spectrometer and absorbance maxima was noted. Percent fluorescence quenching of DPH and TMADPH by 5,6-CF was calculated for each sample relative to control samples. (*), (**) represent a significant difference from the corresponding groups at p<0.005 and p<0.001 respectively.

4. DISCUSSION

Over the last six decades various biophysical methods involving microscopy and spectroscopic techniques have been employed to elucidate the structure of plasma LDL particles (Atkinson et al., 1977; Gulik-Krzywicki et al., 1979; Laggner et al., 1977; Lund-Katz et al., 1988; Spin and Atkinson, 1995). These studies have led to our current understanding of LDL as a 22nm quasi-spherical/ellipsoidal micellar assembly consisting of a single molecule of apoB-100 encompassing an organized network of approximately 3000 lipid molecules. ApoB-100 is large monomeric protein (4536 amino acid residues) that wraps around the surface of the LDL particle to help stabilize the structure of the protein-lipid assembly (Prassl and Laggner, 2009; Segrest et al., 2001). This is achieved through a series of flexible chains that connect the multiple globular domains of apoB-100, thus enabling the protein to surround the LDL particle in a belt-like manner and adopt different conformations with changes in particle size (Prassl and Laggner, 2012; Segrest et al., 2001). The noncovalent assembly of apoB-100 protein and lipids within LDL allows for fluid and dynamic interactions within the particle enabling it to respond to altered thermodynamic and kinetic conditions. ApoB-100 also contains receptor-binding domains, enriched in positively charged residues, which play a key role in recognition of the LDL receptor and its subsequent cellular uptake through receptor mediated endocytosis (Segrest et al., 2001). The lipids within LDL are arranged in three radially diffuse structural layers (Hevonoja et al., 2000). The outer surface layer of the particle consists mainly of the apoB-100 wrapping interspersed with phospholipid head groups arranged in a tangential manner. Next, the interfacial layer contains the radially situated fatty acid chains of the surface phospholipids and unesterified cholesterol. Lastly, within the neutral core is a random arrangement of cholesterol esters and triglycerides. The structural organization of LDL enables the transport of large quantities of insoluble lipids throughout the vasculature system, this in turn facilitates the circulation and homeostasis of cholesterol and fat throughout the body.

The apolar core of LDL has traditionally been viewed as a physiological reservoir for minor lipophilic constituents such as vitamins and antioxidants (Esterbauer et al., 1992). As such, many groups have tried to reconstitute the core of LDL with exogenous lipophilic agents for diagnostic or therapeutic purposes (Krieger et al., 1979b; Marotta et al., 2011; Masquelier et al., 2000a; Radwan and Alanazi, 2014). The core (an estimated 2,141 nm³) provides a confined compartment within the LDL particle capable of ferrying over 1600 lipid molecules (Teerlink et al., 2004). In the present study the reconstitution, core replacement method was also used to incorporate unesterified DHA into the LDL particle. Unlike the neutral lipids that typically reside in the core, unesterified DHA is an amphiphilic molecule that is partially ionized at physiological pH (pK_a=8.5) (Namani et al., 2007).

Hence, the reconstituted unesterified DHA is should preferentially partition into the outer surface and interfacial layer of the LDL particle (Ben-Yashar and Barenholz, 1991; Spooner et al., 1990). Several lines of evidence in the current study supports this hypothesis. The DHA molecules are believed to be orientated such that the carboxyl head groups are positioned at the surface next to the phospholipid head groups. Meanwhile, the long hydrocarbon tail of DHA will orient within the interfacial layer alongside the fatty acid tails of the phospholipids. In this position the carboxyl group of DHA will be exposed to its surrounding aqueous environment. Given the pH of Tricine buffer (pH 8.5) and the pKa of DHA, it is anticipated that half of the DHA molecules will be ionized in the anionic state. These anionic DHA molecules would impart an overall net negative charge to the LDL surface (Aggerbeck et al., 1976; Jayaraman et al., 2007). Both agarose gel electrophoresis and zeta potential measurements confirm this hypothesis as the LDL-DHA nanoparticles were found to possess a significantly greater electronegative surface charge compared to plasma LDL. This increased surface charge is believed to increase the physicochemical stability of LDL-DHA over native LDL and other control LDL nanoparticle counterparts (Jayaraman et al., 2007).

Next, membrane fluidity measurements also provided supporting evidence for structural organization of unesterified DHA in LDL. Fluorescence energy transfer experiments detected increases not only in overall global fluidity of the LDL nanoparticle, but also within the lipid domains immediately adjacent to apoB-100. This increase in surface fluidity for the LDL-DHA nanoparticles relative to native LDL and LDL-DHA methylester provides further evidence for the partitioning of DHA into the LDL nanoparticle surface. The rigidity of the cis-double bonds introduces a 30° bend to the acyl chain of DHA when fully extended (Stubbs and Smith, 1984). As a result, DHA molecules do not pack tightly with the other fatty acyl chains and this increases the average surface or cross sectional area per fatty acyl chain in the lipoprotein phospholipid layer (Stubbs and Smith, 1984). Disruption in the fatty acyl chain packing will alter the physical properties and behavior of the phospholipid layer and allow greater mobility of the lipid constituents within the phospholipid domains. In the present study the annular fluidity was twice that of the global index, this indicates that DHA preferentially associated in lipid domains in close proximity with apoB-100. From studies in cell membranes DHA is known to associate with membrane proteins and improve their function (Mitchell et al., 2003; Stillwell and Wassall, 2003). Our previous studies have demonstrated that DHA helps maintain the integrity of apoB-100 structure (Reynolds et al., 2014), this may be achieved by the relatively long, extensible fatty acyl chains of DHA which may aid in solvating the hydrophobic surfaces of apoB-100 (Palczewski et al., 2000). More studies are warranted to investigate DHA’s effects on the stability and activity apoB-100 in the lipoprotein structure.

The high resolution NMR spectroscopic findings provided further convincing data for the structural organization of DHA within the LDL nanoplatform. Within in nonpolar solvents, like chloroform, the DHA molecules are dispersed and have unrestricted molecular motion, as a result their ¹H and ¹³C NMR resonances are sharp and well defined. Conversely, in alkaline D₂O solution, DHA molecules self-assemble into micelles/vesicles structures where the anionic carboxyl head groups form the outer surface of the colloid and their hydrophobic tails are enclosed from the surrounding aqueous environment (Namani et al., 2007). Under these conditions the resolution of the ¹H and ¹³C NMR signals from DHA generally decrease from the terminal methyl to the carboxyl head group. Within the micelle/vesicle the carboxyl end of the DHA are closely assembled in the surface layer so molecular motion is restricted. Conversely, the terminal methyl end of DHA is less restricted and has much greater degrees of freedom. These findings are consistent with the spin-lattice and spin-spin relaxation measurements reported by Kobayashi et al. (Kobayashi et al., 2004) and Soubias and Gawrisch (Soubias and Gawrisch, 2007), where proton and carbon T1 and T2 values decreased from the terminal methyl to the carboxyl end of the DHA chain. The NMR profile of the DHA molecules in the LDL nanoparticle more closely reflected the features of DHA in aqueous rather than nonpolar solvents. The carbon-13 and proton NMR signals from the carboxyl end of DHA in the LDL nanoparticle were broad and of low intensity indicting short T1 and T2 values from their rigid orientation in the LDL lipid surface. Conversely, signals from the terminal methyl end of DHA were well resolved and of higher intensity suggesting a higher mobility among the phospholipid acyl chains in the interfacial layer. The presence of the large apoB-100 within the phospholipid layer of LDL is also expected to alter the chain dynamics of DHA. Indeed, longer correlation times and increased R₂ rates were reported for the carboxyl segment of DHA when closely associated with large membrane proteins like rhodopsin (Soubias and Gawrisch, 2007). Similar phenomenon are likely responsible for the broad and low intensity ¹³C NMR signals arising from the carboxyl terminal carbons of DHA in LDL relative to micelle/vesicle DHA in water. Further molecular characterization can also be achieved with ¹³C NMR method as each carbon atom along the DHA chain can be specifically described. The broad and low signal from C1-C3 of DHA in LDL is indicative of the restrictive environment these carbons encounter in the LDL lipid surface, however, from C4–C22 the resonances are noticeable sharper and of greater intensity. From this observation it can be concluded that the C4-C22 segment of DHA, penetrates deeper into interfacial layers of LDL such that they experience less restriction and greater molecular motion.

In a final series of experiments to validate the surface partitioning of DHA in the LDL nanoparticle, the lanthanide shift reagent, EuCl₃, was shown to significantly disrupt the proton signals from carboxyl segment of DHA (C2, C3). The terminal methyl protons were affected to a lesser degree with a slight increase in line broadening. EuCl₃ had no measurable effect on LDL-DHA methylester confirming the core disposition of DHAmethylester. Collectively, the NMR experiments provided convincing data for the structural organization of DHA in reconstituted LDL nanoparticles.

The experiments in this study also demonstrated that the structural organization of LDL-DHA nanoparticles more closely resembled a phospholipid bilayer structure, like nascent HDL, than the conventional phospholipid monolayer/apolar core structure of mature lipoproteins. The fluorophores DPH and TMADPH partition into the respective core and surface layers of LDL and LDL-DHAmethylester, confirming the shell-core structure of these particles (Ben-Yashar and Barenholz, 1991). In the LDL-DHA nanoparticles DPH and TMADPH similarly partition into the nanoparticle surface (equivalent quenching by 5,6-carboxy fluorescein) suggesting the absence of an apolar core. These findings were further confirmed by the ultrastructural analysis provided by cryo-EM. The plasma LDL particles displayed an ellipsoid morphology with numerous core striations, while the reconstituted LDL-TO and LDL-DHAmethylester nanoparticles were composed of heterogeneous populations of spherical particles devoid of core striations. The absence of the ordered striations in these reconstituted particles reflects the state of their TO and DHAmethylester cores, which have significantly lower transition temperatures than plasma LDL and thus exist in an isotropic liquid state (Sherman et al., 2003). The LDL-DHA nanoparticles were quasi-spherical in morphology and were more uniform in size and shape than the other reconstituted particles. The cryo-EM images clearly displayed the bilayer structure of LDL-DHA nanoparticles. The circumference of each particles was prescribed by a high density staining region which was accompanied by a second equally dense inner ring (these are believed to represent the inner and outer phospholipid bilayers). The distance between these two ring structures was measured to be approximately 4.2 nm, which is the typical thickness of a phospholipid bilayer (Atkinson and Small, 1986). The observed bilayer structure of LDL-DHA nanoparticles is consistent with the structural organization of other apolipoprotein-phospholipid complexes (nascent HDL) that lack neutral lipid components. Our proposed schematic structure of the LDL-DHA nanoparticle is presented in Figure 8. These findings indicate that the structural organization of lipids in LDL-DHA differs and is distinct from plasma LDL, hence, LDL-DHA can be considered a unique nanostructure from othe lipoprotein particles. Formative studies by Walsh and Atkinson demonstrated that recombination of ApoB with phospholipids also formed bilayer structures approximately 21nm in diameter (Walsh and Atkinson, 1983; Walsh and Atkinson, 1986). Watt and Reynolds also prepared similar apoB-phospholipid structures, their studies further demonstrated that the conformational integrity of apoB was similar to that observed in native LDL (Watt and Reynolds, 1981). In a similar manner, the inclusion of nonesterified DHA into the LDL nanoparticle formulation did not appear to disrupt the protein-lipid interactions or structural organization of apoB-100 (Reynolds et al., 2014). Native LDL and the LDL nanoparticles all maintain a high surface curvature and a diameter of approximately 20nm. For each of these structures the outer phospholipid layer opposes a lipid phase. Within the bilayer structures, the surface monolayer interfaces with the fatty acyl chains of the inner phospholipid layer, while in native LDL, LDL-TO and LDL-DHAmethylester the surface monolayer interfaces with the neutral lipid core. The particles organization in each of these cases enable the hydrophobic portion of apoB100 to be stabilized within a lipid milieu such that integrity of the protein conformation is maintained.

Figure 8. Structure of LDL-DHA Nanoparticle.

Schematic diagram of LDL-DHA nanoparticle with the proposed surface localization of DHA.

The composition data of LDL-DHA also conveys additional information regarding surface topology and structural organization of this nanoparticle. Given that apoB100 retains its conformation similar to that in plasma LDL, apoB-100 is expected to occupy a large portion of the LDL-DHA surface layer (Segrest et al., 2001). The remaining surface layer is composed of phospholipid and unesterified DHA at a molar ratio of 1:2–3. The bilayer lipid surfaces are anticipated to be fluid and highly dynamic as a result many conformers (angle-iron, helical and coiled) DHA can adopt in the lipid environments (Gawrisch et al., 2003) as well as the flip-flop interchanges of lipids across the bilayer (Rothman and Dawidowicz, 1975). Finally, it is important to note that the bilayer structure of LDL-DHA is composed of approximately 500 phospholipids and 1100 DHA molecules, in contrast the surface monolayer of plasma LDL contains around 680 phospholipids and 400 unesterified cholesterol molecules (Hevonoja et al., 2000). Intuitively, the bilayer LDL-DHA nanoparticles should have approximately twice the lipids as plasma LDL if they are to maintain the same diameter. While both particles are of similar size the lipid stoichiometries don’t match, the versatility of apoB-100 likely accounts for this difference. ApoB-100 is a highly flexible molecules that is able to adjust for the structural and compositional changes that transpire in the circulation as VLDL particles (80–200 nm in diameter) are metabolized and transformed to the 22 nm LDL particles (Hevonoja et al., 2000). In a similar manner, the lipid-associating domains of apoB-100 likely adjusts for the lipid content of LDL-DHA enabling the nanoparticle to retain a discrete size and structure.

Conclusion

In conclusion, our studies have demonstrated that the reconstitution of plasma LDL with unesterified DHA produces a unique nanoparticle with distinct molecular and structural organization. Unesterified DHA preferentially incorporates into the outer surface layer of LDL, where the carboxyl end of the PUFA is interspersed among the phospholipid head groups and the methyl terminal regions align radially with the phospholipid fatty acyl chains in the interfacial layer. In this orientation the anionic carboxyl end of DHA is exposed to the LDL surface and impart an electronegative charge to the nanoparticles surface. Further structural analyses with cryoTEM revealed that the LDL-DHA is arranges as a phospholipid bilayer nanostructure with an aqueous core, as opposed to the typical phospholipid monolayer/apolar core structure of plasma LDL. These findings indicate that LDL-DHA has a completely distinct lipid assembly from plasma LDL. Despite these differences, apoB-100 retains strong associations with the phospholipid/DHA assembly of LDL-DHA and stringently maintains the overall size and morphology of this nanoparticle similar to plasma LDL. Further studies are now being explored that utilize this current knowledge on LDL-DHA structure to better understand its unique physicochemical properties and its selective cytotoxic actions against cancer cells.

Highlights.

• LDL nanoparticles can incorporate >1100 docosahexaenoic acid (DHA) molecules per particle.

• Spectroscopic NMR analyses indicate that reconstituted DHA partitions into the outer surface/interfacial layer of the LDL particle.

• Incorporation of DHA into LDL increases the particle’s membrane fluidity.

• Cryo-TEM and fluorescence quenching experiments reveal that LDL-DHA is organized as a phospholipid bilayer structure.

• LDL-DHA is a novel and unique lipoprotein-based nanostructure.