Collagen-specific peptide conjugated high density lipoprotein nanoparticles as MR contrast agents to evaluate compositional changes in murine atherosclerotic plaques under regression conditions

Abstract

Objective

In this study, we developed magnetic resonance (MR) contrast agents based on high-density lipoprotein (HDL) nanoparticles to non-invasively visualize intraplaque macrophages and collagen content in mouse atherosclerotic plaques.

Background

Macrophages and collagen are important intraplaque components, playing central roles in plaque progression/regression. In a Reversa mouse model, plaque regression with compositional changes (from high macrophage, low collagen to low macrophage, high collagen) can be induced.

Methods

We labeled HDL nanoparticles with amphiphilic gadolinium chelates to enable target-specific imaging of intraplaque macrophages. To render HDL nanoparticles specific for the extracellular matrix, labeled HDL nanoparticles were functionalized with collagen-specific EP3533 peptides (EP3533-HDL) via poly(ethylene glycol) (PEG) spacers embedded in the HDL lipid layers. The association of nanoparticles with collagen was examined in vitro by optical methods. The in vivo MR efficacy of these nanoparticles was evaluated in a Reversa mouse model of atherosclerosis regression. Ex vivo confocal microscopy was applied to corroborate the in vivo findings and to evaluate the fate of the different HDL nanoparticles.

Results

All nanoparticles had similar sizes (10 ± 2 nm) and longitudinal relaxivity r₁ (9 ± 1 s⁻¹mM⁻¹). EP3533-HDL showed strong association with collagen in vitro. After 28 days of plaque regression in Reversa mice, EP3533-HDL showed significantly increased (p<0.05) in vivo MR signal in aortic vessel walls (normalized enhancement ratio, NER_w=85±25%; change of contrast-noise-ratio, ΔCNR_w=17±5) compared with HDL (NER_w=−7±23%; ΔCNR_w=−2±4) and non-specific control EP3612-HDL (NER_w=4±24%; ΔCNR_w=1±6) at 24 h post-injection. Ex vivo confocal images revealed the colocalization of EP3533-HDL with collagen. Immunohistostaining analysis confirmed the changes of collagen and macrophage contents in the aortic vessel walls after regression.

Conclusions

we have shown that the HDL nanoparticle platform can be modified to monitor in vivo plaque compositional changes in a regression environment, which will facilitate understanding plaque regression and the search for therapeutic interventions.

Atherosclerotic plaques are the result of a chronic inflammatory response in arterial vessel walls to the accumulated lipoproteins that contain apolipoprotein B (1). The rupture of a plaque may cause many severe health risks (2–4). Active inflammation with extensive macrophage infiltration into vessel walls, leading to plaques with a thin fibrous cap and a large lipid necrotic core are considered to be the major indicators, among several other factors, for high-risk plaques prone to rupture (5–11).

One of the therapeutic intervention goals for high-risk patients is to stabilize atherosclerotic lesions by inducing regression. In clinical studies, the stabilization of plaques has been found to correlate with lower intraplaque macrophages and higher interstitial collagen (12). Preclinical animal studies have further facilitated the understanding of plaque regression (12–15). Recently, the Fisher and Young groups developed the “Reversa” mouse model that possesses four homozygous alleles, LDLR^−/−ApoB^100/100Mttp^fl/flMx1-Cre^+/+ (16). When microsomal triglyceride transfer protein (Mttp) is conditionally ablated after plaque formation, low-density lipoprotein (LDL) production falls, and the hyperlipidemic plasma profile abates, with a consequent regression of disease as indicated by a decrease of macrophages and increase of collagen in plaques (17). Although these changes in plaque component levels during regression have been confirmed by ex vivo histological examinations, non-invasive in vivo visualization methods to detect these biological events by molecular imaging are highly desired to allow temporal evaluation of the stabilization of atherosclerosis.

Magnetic resonance (MR) imaging is one of the most powerful techniques to non-invasively visualize plaque composition and biological activity at submillimeter spatial resolution (18, 19). However, contrast agents with high payload are often essential for molecular MR imaging due to the low concentration of biomarkers and inherently low sensitivity of MR (20, 21). We have previously developed high-density lipoprotein (HDLs) based MR contrast agents to evaluate intraplaque macrophage content via their natural affinity for plaque macrophages (22–26). In this study, we report the use of HDL based MR contrast agents to visualize two key plaque components: macrophages and collagen during plaque regression conditions in Reversa mice. The HDL platform was used because of its endogenous nature, small size (~10 nm) allowing penetration into atherosclerotic plaques, and the possibility to include a higher contrast payload than small molecule platforms. In addition, since plaque component changes in the duration of this study are more evident than size changes (13, 16, 17) we were able to focus on changes in composition.

Methods

Materials

The collagen-specific peptide EP-3533 with the sequence GKWH[CTTKFPHHYC]LYBip-CONH₂, and the non-specific peptide EP-3612 containing non-natural D-Cys as shown in Figure 1 were synthesized by Peptide International. 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rhodamine-PE) and Gadolinium diethylenetriaminepentaacetate-bis(stearylamide) (Gd-DTPA-BSA) were used as fluorescence label and MR contrast agents, respectively. The preparation of HDL nanoparticles followed our previous methods (22–26), but we included 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy(polyethylene glycol)-2000] ammonium salt (DSPE-PEG-COOH) to allow attached of the peptides. Preparation of EP3533-HDL and EP3612-HDL nanoparticles: The conjugation of EP-3533 peptides to HDL nanoparticles is schematically illustrated in Figure 1. First, the HDL nanoparticles were transferred into HEPES buffer (pH=5.0) and reacted with 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) and N-Hydroxysuccinimide (NHS). After incubation at room temperature for 2 h, the reaction solution was thoroughly washed. Then 1 mg/ml EP-3533 peptides in dimethyl sulfoxide was added into the solution with a 1:1.1 DSPE-PEG-COOH:EP-3533 molar ratio. This preparation was kept at room temperature for 4 h, stored at 4 °C overnight, and then purified and buffer exchanged to PBS. The final nanoparticles were named EP3533-HDL.

Figure 1.

Illustration of (A) method to functionalize HDL nanoparticles with collagen-specific EP3553 peptides (named as EP3533-HDL), and (B) animal experiment design.

Non-specific control nanoparticles (EP3612-HDL) were prepared by the same procedure using EP-3612 peptides.

In vitro binding assay

Black 96-well plates were incubated with different proteins overnight at 4 °C, and then washed with PBS. After blocking with BSA for 3 h at 37 °C, they were washed again with PBS. Then nanoparticles was added to each well and incubated 3 h at 37 °C. After cooling to room temperature, the total fluorescence I₀ of nanoparticles was recorded (Excitation: 540 ± 25 nm; Emission: 590 ± 35 nm). Then the nanoparticle solutions were aspirated. The plates were washed three times with PBS, after which the binding fluorescence I_bind was recorded. The average fluorescence reading of empty wells I_empty was subtracted from the raw fluorescence data. In order to correct the loss of fluorescence labeling of Rhodamine-PE from nanoparticles during peptide conjugation, the binding of nanoparticles was normalized as Normalized Binding = (I_bind − Iempty)/(I₀ − Iempty)*[Gd].

Animal use

The Institutional Animal Care and Use Committee of Mount Sinai School of Medicine approved the animal protocols and procedures. At 4-weeks of age, Reversa mice were weaned and then fed on Western diet containing a 0.15% cholesterol and 21% fat for 16 weeks. Mice were then divided into “regression” and control groups as shown in Figure 1. The regression group was fed on a chow diet and Cre was induced by i.p. injection of polyinosinic-polycytidylic ribonucleic acid (pIpC) (250 μg every other day for a total of 4 injections). Day 0 was defined as the 9th day after 1st pIpC injection. The mice were scanned by MR imaging at Day 0 (n=5) and Day 28 (n=5) and then sacrificed for histological staining. The control group was fed the Western diet for the duration of the experiment and injected with saline.

MR imaging

The animals underwent in vivo MR imaging of the abdominal aorta using a 9.4T, 89mm bore magnet system. High resolution T₁-weighted images were generated using a black blood spin echo (SE) sequence (TR/TE=800ms/8.6ms, FOV=3.0×3.0cm, matrix size=256×256, 22 contiguous 500 μm-thick axial slices, number of averages=16, total scan time=54 min). After pre-injection imaging the animals were administered a nanoparticle contrast agent (50 μmol of Gd/kg) via the tail vein. Post-injection imaging with the same parameters was performed on each animal at 24 h after injection. The slices of the post-injection scan were matched to the pre-injection scan by using unique vertebral and para-spinous muscular anatomy as landmarks.

Immunohistochemistry

Frozen abdominal aorta sections were stained using as primary antibodies either rat anti-mouse CD68 or rabbit anti-mouse collagen I. The secondary antibody was Alexa Fluor 647 conjugated goat anti-rat Ab or Alexa Fluor 647 conjugated goat anti-rabbit Ab corresponding to the primary antibody. Confocal imaging was performed using a Leica SP5DM microscope.

Frozen sections were stained by Sirius red to detect interstitial collagen using birefringency illumination with polarized light.

For immunohistochemical CD68 staining, the frozen sections were stained using a rat anti-mouse CD68 primary antibody. The sections were visualized using Vector Red solution for positive areas and counter stained blue with Vector Hematoxylin QS.

Image analysis

For in vivo MR images, the signal-to-noise ratio (SNR) of the region of interest (ROI) is defined by SNR_ROI= I_ROI/I_noise where I_ROI is the intensity of either aortic vessel wall (SNR_w) or surrounding muscle (SNR_m); I_noise is the standard deviation (S.D.) outside the animal. The contrast-to-noise ratio (CNR_w) to muscle is defined as CNR_w=SNR_w-SNR_m. The difference of CNR_w between pre- and post-injection is defined as ΔCNR_w=(CNR_w)_post-(CNR_w)_pre. The normalized enhancement ratio of aortic wall (NER_w) to muscle is defined as NER_w=[(SNR_w/SNR_m)_post-(SNR_w/SNR_m)_pre]/(SNR_w/SNR_m)_pre ×100%. For ex vivo microscopy images, the percentages of positive area for macrophage and collagen contents were analyzed in ImageJ as described in Supplemental Information.

Statistics

The detailed statistical methods are described in Supplemental Information. A value of p<0.05 was considered statistically significant.

Results

EP3533-HDL nanoparticles showed higher association with collagen in vitro than controls

The conjugation of EP3533 or EP3612 peptides to HDL via DSPE-PEG-COOH linkers did not significantly change the nanoparticle size (10 ± 2 nm) or longitudinal relaxivity (9 ± 1 s⁻¹mM⁻¹). Figure 2A showed the unmodified HDL had negligible and lowest in vitro association with all the tested proteins among all three types of nanoparticles. EP3533-HDL (Figure 2B), however, had significantly higher levels of binding to collagen in comparison with HDL and nonspecific EP3612-HDL and a low level of association to other extracellular matrix components, such as heparin, fibronectin, chondroitin sulfate A, and chondroitin sulfate B. The non-specific EP3612-HDL (Figure 2C) had much lower levels of association to all the tested collagen and other ECM components in comparison with EP3533-HDL, but slightly higher than unmodified HDL.

Figure 2.

Normalized in vitro binding of (A) HDL, (B) EP3533-HDL, and (C) EP3612-HDL nanoparticles to different proteins.

EP3533-HDL enhanced in vivo MR signal of aortic walls of regressing atherosclerotic plaques in Reversa mice

Representative in vivo MR images of atherosclerotic plaques of Reversa mice in the regression group are shown in Figure 3A. At Day 0, the baseline of the regression, HDL caused enhancement of aortic vessel walls at 24 h post injection. However, EP3533-HDL did not cause significant enhancement. Non-specific EP3612-HDL injection also induced signal enhancement of MR images of aortic vessel walls. The NER_w for HDL, EP3533-HDL, and EP3612-HDL contrast agents were 91 ± 39%, 16 ± 45%, and 67 ± 37% (mean ± S.D.), respectively, at 24 h post-injection (Figure 3B). At Day 28 of regression, the signal enhancements or NER_w induced by the injection of HDL and EP3612-HDL were substantially reduced to −7 ± 23% and 4 ± 24%, respectively, at 24 h post-injection. In comparison, the enhancement caused by EP3533-HDL increased significantly to 82 ± 25% at 24 h post-injection (Figure 3B). The corresponding ΔCNR_w values are shown in Figure 3C, which showed same trends as NER_w. In the control group (i.e. continuing hyperlipidemia), the NER_w and the ΔCNR_w of all three nanoparticle contrast agents remained similar at Day 28, with values in comparison with those at Day 0 baseline (Figure 4).

Figure 3.

(A) Typical MR images, (B) NER_w and (C) ΔCNR_w of abdominal atherosclerotic plaques for pre- and 24 h post-injection of HDL, EP3533-HDL, and EP3612-HDL at Day 0 (white bars) and Day 28 (blue bars) of Reversa mice in the regression group. The arrows point to the aortas. The error bars represent mean ± S.D. The stars (*) indicate statistical significance at p<0.05. (n=5 mice × 5 slices/mice=25)

Figure 4.

(A) Typical MR images, (B) NER_w and (C) ΔCNR_w of abdominal atherosclerotic plaques for pre- and 24 h post-injection of HDL, EP3533-HDL, and EP3612-HDL at Day 0 and Day 28 of Reversa mice in the control group. Labels as in Figure 3. (n=5 mice × 5 slices/mice=25)

EP3533-HDL nanoparticles showed colocalization with collagen type I in aortic vessel walls of Reversa mice at Day 28 in the regression group

After in vivo MR imaging, the colocalization of nanoparticles was analyzed on aortic sections at 24 hrs post injection of contrast agents. The HDL nanoparticles were found to localize mainly with macrophages at Day 28 (Figure 5A), in agreement with prior results (22–25), but not with collagen type I (Figure 5B). At Day 28, EP3533-HDL was found to localize with collagen type I (Figure 5D) with negligible association with CD68 signal under confocal imaging (Figure 5C). Most of the non-specific EP3612-HDL was not found to localize with CD68+ macrophages or collagen type I at Day 28, though some of these nanoparticles were seen at the same positions of CD68 signals in confocal images (Figure 5E–F). Confocal microscopy of all three types of nanoparticles with CD68 macrophages and collagen type I staining at Day 0 is available in the Supplemental Information.

Figure 5.

Confocal microscopy of plaques at Day 28 from aortic vessel walls of Reversa mice in the regression group with CD68 (green) or Collagen type I (Col I, green) staining. The nuclei were stained with DAPI (blue) and nanoparticles were labeled with Rhodamine-PE (red). The colocalization of green and red signal appeared as yellow/orange colors. Scale bar: 25μm.

The content of CD68+ cells decreased at Day 28 in aortic plaques of Reversa mice in the regression group

The area percentage of CD68+ macrophages/foam cells (red) was 10.9 ± 4.6% in the plaques at Day 0 (Figure 6A/D). At Day 28 of regression, the CD68+ cells decreased significantly to 2.6 ± 2.0% (Figure 6B/D). However, the content of CD68+ area increased to 15.5 ± 5.4% at Day 28 in control group (Figure 6C/D).

Figure 6.

Immunohistochemical staining of CD68+ macrophages (Red) in the atherosclerotic plaques of aorta at (A) Day 0, and Day 28 in (B) regression and (C) control group. Nuclei in blue. (D) The percentage of CD68+ area in the plaques. The images were obtained at 40×. “P” denotes the plaque position. Labels as in Figure 3. (n = 5)

The collagen content increased at Day 28 in aortic plaques of Reversa mice in the regression group

As shown in Figure 7A, the plaques at aortic vessel walls at Day 0 showed little collagen (from analysis of polarized light images) corresponding to a collagen positive area of 2.9 ± 2.1% (Figure 7D). At Day 28, the collagen content increased significantly to 22.1 ± 11.6% in the regression group (Figure 7B/D). However, the collagen content maintained a similar level (2.0 ± 0.7% of plaques) at Day 28 in the control group (Figure 7C/D). The red stained areas were observed under bright field but not always as fibrils under polarized light in Figure 7C. This suggests that not all the collagen was organized into fibrils but rather degraded by matrix metalloproteinases (MMPs) expressed by macrophages/foam cells.

Figure 7.

Sirus red staining of atherosclerotic plaques in aortic vessel walls at (A) Day 0, Day 28 in (B) regression and (C) control group. The bright field images indicate the position of plaques. The collagen containing area is bright under polarized light. (D) The percentage of collagen positive area in the plaques. Italic “A”, “M”, “L”, and “P” denote adventitia, media, lumen, and plaque positions, respectively. Labels as in Figure 3. (n = 5)

The percentage of macrophage or collagen positive area was pooled together from both the regression and control groups to establish the correlations with in vivo MR enhancement (NER_w in the aortic walls). The correlation coefficients were 0.85 (R²= 0.71) for CD68+ macrophages with in vivo NER_w from HDL injection (Figure 8A) and 0.77 (R²=0.60) for collagen with in vivo NER_w from EP3533-HDL injection (Figure 8B).

Figure 8.

Correlations of (A) CD68+ and (B) collagen positive areas with NER_w obtained after HDL and EP3533-HDL injections, respectively.

Discussion

In the present study we demonstrated that HDL based nanoparticles can be used as MR contrast agents for non-invasive in vivo imaging of atherosclerotic plaque regression by targeting collagen after conjugation with collagen-specific EP3533 peptides. The interaction between HDL and collagen is improved by the EP3533 peptides (Figure 2), which have been developed to specifically target collagen (27). It is likely that EP3533-HDL may bind to collagen in the media and adventitia as well as in the plaque in vivo. Unfortunately, the resolution required to distinguish among the media, adventitia and plaque in mice arteries is not currently possible, although the contrast changes we observed are most likely due to changes in collagen in the plaque, not other parts of the arterial wall.

Macrophages and collagen play important roles in the vulnerability of atherosclerotic plaques. Macrophage content has been used to evaluate atherosclerotic plaque burden in animal models and in humans (26, 30–34). Fisher and colleagues reported a detailed study of biological characteristics of Reversa mice (17), which was confirmed again in this study (Figure 6 and 7). In advanced plaques, collagen fibrils are degraded by MMPs, resulting in many small fragments, which were not observed under polarized light microscopy or to associate with EP3533-HDL. The MR imaging and histology results together also likely imply that EP3533-HDL is targeting collagen fibrils as opposed to degraded collagen. It should be noted that rather than providing an absolute amount of collagen, EP3533-HDL indicates a relative collagen level in plaques by MR imaging.

The different trends of macrophage and collagen contents during plaque regression were found to be correlated with the in vivo MR signals of aortic vessel walls using HDL and EP3533-HDL as contrast agents (Figure 8). The correlations indicated the MR signals from aortic vessel walls are most likely reflecting the macrophage and collagen contents. Therefore collagen-specific EP3533-HDL nanoparticles can be used to visualize the collagen content non-invasively in vivo and thus to monitor the characteristics that are taken to indicate the stabilization of human atherosclerotic plaques after therapeutic intervention.

HDL nanoparticles were found to be mainly colocalized with CD68+ macrophages at both baseline (Day 0) (Figure S1 in Supplemental Information) and Day 28 of regression (Figure 5A). These results are consistent with our previous findings in several other mouse models (22–25). We also observed that not all HDL nanoparticles are associated with macrophages, which may be due to the retention of HDL by proteoglycans, especially biglycan and perlecan (36, 37) from the association of ApoAI in the lipid pools (38–40) and around necrotic cores (41) in atherosclerotic plaques. Interestingly, at baseline (Day 0), EP3533-HDL was also found to be associated with CD68+ macrophages and had a low association with collagen type I (Figure S1). This could be due to several reasons. First, activated high levels of macrophages in plaques at Day 0 could take up EP3533-HDL through non-specific endocytosis / phagocytosis. Second, conjugation of EP3533 cannot fully block HDL function, leading to off-target macrophage association. Third, the very low content of collagen results in little interaction with EP3533-HDL. However, at Day 28 under regression conditions, EP3533-HDL was found to be associated with collagen type I rich areas inside plaque due to strong binding to collagen (Figure 2A) and the readily accessible collagen contents inside plaque (Figure 7B/D), but not to be with CD68+ macrophages due to low macrophage content (Figure 6B/D) and the low possibility of access to nanoparticles. Little accumulation of EP3533-HDL was observed in the adventitia because the intravenously injected nanoparticles remained primarily inside lumen and minimally reached adventitia, which typically has little vasa vasorum compared to human plaques.

In conclusion, we demonstrate that HDL based nanoparticles are a versatile platform to non-invasively image atherosclerosis in vivo by MR to visualize not only macrophages, but also collagen after conjugation to HDL of collagen-specific peptides, which provides a rerouting strategy. MR signal enhancements of atherosclerotic plaques in aortic vessel walls after HDL and EP3533-HDL injections in Reversa mice were correlated with macrophage and collagen contents, respectively. By combination of imaging macrophage and collagen contents, the stabilization process of atherosclerotic plaques can be monitored and evaluated in vivo under regression conditions.

Abbreviations

HDL

High-density lipoprotein

LDL

Low-density lipoprotein

MMP

Matrix metalloproteinase

ECM

Extracellular matrix

MR

Magnetic resonance

Rhodamine-PE

1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl)

Gd-DTPA-BSA

Gadolinium diethylenetriaminepentaacetate-bis(stearylamide)

CNR

Contrast-to-noise ratio

NER

Normalized enhancement ratio

pIpC

Polyinosinic-polycytidylic ribonucleic acid