Imaging Metastasis Using an Integrin-Targeting Chain-Shaped Nanoparticle

Abstract

While the enhanced permeability and retention effect may promote the preferential accumulation of nanoparticles into well-vascularized primary tumors, it is ineffective in the case of metastases hidden within a large population of normal cells. Due to their small size, high dispersion to organs, and low vascularization, metastatic tumors are less accessible to targeted nanoparticles. To tackle these challenges, we designed a nanoparticle for vascular targeting based on an α_vβ₃ integrin-targeted nanochain particle composed of four iron oxide nanospheres chemically linked in a linear assembly. The chain-shaped nanoparticles enabled enhanced ‘sensing’ of the tumor-associated remodeling of the vascular bed offering increased likelihood of specific recognition of metastatic tumors. Compared to spherical nanoparticles, the chain-shaped nanoparticles resulted in superior targeting of α_vβ₃ integrin due to geometrically enhanced multivalent docking. We performed multimodal in vivo imaging (Fluorescence Molecular Tomography and Magnetic Resonance Imaging) in a non-invasive and quantitative manner, which showed that the nanoparticles targeted metastases in the liver and lungs with high specificity in a highly aggressive breast tumor model in mice.

While conventional small molecule therapeutics are distributed within cancer and healthy tissues in a non-specific manner, nanoparticle delivery systems have been developed to exploit a feature of the tumor microenvironment, the so-called ‘Enhanced Permeability and Retention’ (EPR) effect.¹ Due to the tumor’s leaky vasculature, the intratumoral accumulation of nanoparticles is high relative to normal tissues.^2–5 While the EPR strategy may be effective in large, well-vascularized tumors, it is very ineffective in metastatic disease, which presents small clusters of malignant cells within variable tissue types.⁶ Notably, the 5-year survival rate of breast cancer patients sharply decreases from 98% in cases with localized primary lesions to 23% in cases of distant metastases.⁷ The vast majority of cancer deaths are due to metastatic disease, hence it is essential to design nanotechnology-based agents that can successfully target metastatic tumors.

Targeting a metastatic lesion within a large population of normal cells presents a unique challenge. Metastases present biobarriers due to their smaller size, higher dispersion to organs, and lower vascularization than primary tumors, making them less accessible to molecular or nanoparticle agents. However, metastatic lesions can upregulate specific cell-surface molecules and secreted factors that differ from the rest of its host organ. If the appropriate chemical specificity is selected, targeted nanoparticles could provide a unique opportunity to home imaging and therapeutic compounds to metastases. Several conceptual and experimental theories of metastatic cancer have been recently explored that may be exploited for nanoparticle targeting.^{8, 9} For example, β₃ integrins have shown to play a key role in migration, invasion and metastasis.^10–16 It has been shown that the metastatic site transitions from selectin-dependent tumor cell rolling on the endothelium to firm attachment that is mediated primarily by α_vβ₃ integrins.^{10, 12} Thus, α_vβ₃ integrin is a very attractive target.¹⁷ While various studies have used integrin-binding peptides to target nanoparticles to tumors,^16–24 the preclinical development of these systems has focused predominantly on tumors at the primary site. Taking under consideration the microenvironment of metastatic tumors, a targeted nanoparticle capable of ‘sensing’ the metastasis-associated remodeling of vascular bed may offer an increased likelihood of specific recognition of tumors. In this work, we report the design of a nanoparticle targeting α_vβ₃ integrin for non-invasive imaging of metastasis (Figure 1a). Specifically, we selected the cyclic tripeptide arginine–glycine–aspartic acid (cRGD) peptide as the ligand for vascular targeting of the nanoparticle to metastases.^{19, 23, 24} We exploited the nanochain technology to fabricate a chain-shaped nanoparticle composed of four iron oxide (IO) nanospheres chemically linked into a linear assembly (Figure 1b).^{25, 26} The high aspect ratio and flexibility of the nanoparticle substantially increased its chances to successfully seek metastatic lesions due to geometrically enhanced multivalent docking²⁷ to the vasculature of metastases. Due to this high avidity and increased magnetic relaxivity of the nanochain particles²⁶, we were able to detect metastatic lesions in an aggressive breast tumor model in mice using small animal fluorescence molecular tomography (FMT) and magnetic resonance imaging (MRI). This approach should also be useful for selective targeting of therapeutic agents to metastases.

Figure 1.

Characterization of the RGD-NC nanoparticles. (a) Illustration of the models for the successful delivery of RGD-NC nanoparticles to metastasis via vascular targeting. (b) Diagram of the RGD-NC nanoparticle and its constituent components. (c) TEM image of RGD-NC nanoparticles predominantly composed of four IO spheres. (d) Size distribution of the parent IO nanospheres and RGD-NC nanoparticles obtained by DLS measurements.

RESULTS

Fabrication and characterization of the integrin-targeted nanoparticle

Fabrication of the integrin-targeted nanoparticle (termed RGD-NC) was based on the nanochain technology,²⁶ which is a two-step approach using solid-phase chemistry. In the first step, amine-functionalized IO nanospheres were attached on a solid support via a crosslinker containing a disulfide bridge. Liberation of the nanosphere using thiolytic cleavage created thiols on the portion of the particle’s surface that interacted with the solid support resulting in a particle with two faces, one displaying only amines and the other only thiols. Therefore, we were able to topologically control the conversion of amines on the surface of the IO nanospheres into thiols, resulting in a particle with asymmetric surface chemistry. In the second step, employing solid-phase chemistry and step-by-step addition of particles, the two unique faces on the same IO nanosphere served as fittings to assemble them into linear nanochains (Figure 1b). The nanochains were analyzed via visual inspection of multiple TEM images. As shown in Figure 1c, the nanochains were synthesized in a highly controlled manner. Most of the nanochains are linear and consist of 4 IO spheres with the overall geometrical dimensions of the particle being about 100 × 20 nm (length x width). To evaluate the robustness of the nanochain synthesis, the number of IO nanospheres per nanochain was measured in TEM images. While 6% of the total particles in the suspension were the parent (unbound) IO spheres, the majority of the particles (72%) comprised of nanochains with 4 IO spheres (12 and 10% were nanochains with 3 or 5 IO spheres, respectively). As shown in Figure 1d, the hydrodynamic size of the particle and its constituent IO spheres, as measured by dynamic light scattering (DLS), verified the TEM images. It should be noted that DLS measured the effective hydrodynamic diameter based on the diffusion of the particles. Since the hydrodynamic diameter measured by DLS does not correspond to the geometrical size of non-spherical particles, we relied on visual analysis of TEM images to measure the exact dimensions of the nanochain. Detailed characterization of the nanochain particles is reported in a previous publication.²⁶

The cyclo (Arg-Gly-Asp-D-Phe-Cys) or c(RGDfC) was conjugated onto the distal end of the PEG-NH₂ on the particle’s surface. While Figure 1c shows the nanochain after modification with the RGD peptide, a TEM image of the nanochain prior to modification is shown in Figure S1 in Supporting Information. In addition to conjugation of the peptide, the nanochain particles were labeled with an NIR fluorophore (VivoTag 680) to be detectable by fluorescence imaging. To evaluate the effect of the geometry on the magnetization, we compared the r2 relaxivity of the RGD-NC particle to that of its parent IO nanospheres by measuring the transverse (R2) relaxation rates at 1.4 Tesla. The r2 value of the RGD-NC particle was 121 s⁻¹ mM⁻¹, which was 2.1-fold higher than that of its constituent IO spheres. Detection of metastasis via receptor-mediated targeting depends on the generation of signal from each nanoparticle. Thus, we calculated the T2 relaxivity on a per nanoparticle basis, which was 8.4 times higher for the RGD-NC particle compared to its constituent IO nanospheres.

Targeting of the RGD-NC nanoparticles to integrin-expressing endothelial cells was evaluated in vitro under static and flow conditions. Bovine aortic endothelial cells (BAEC) were treated with TNF-α to induce expression of α_vβ₃ integrins²⁸ and then incubated with an excess of the RGD-NC nanoparticles for different periods of time. As shown in Figure S2a in Supporting Information, the time course of the nanoparticle uptake by the cells showed that the binding of the nanoparticles occurs rapidly during the first 30 min of incubation. In a similar manner, we evaluated the cellular uptake by 4T1 cells indicating that the integrin-targeting RGD-NC nanoparticles were also able to target the cancer cells (Figure S2b). This is significant, because the metastatic 4T1 cells colonize the endothelium as we show later in the histological evaluation.

Successful vascular targeting requires that a nanoparticle can escape the blood flow and drift towards the blood vessel walls (e.g. high margination), followed by strong attachment to the targeting site offsetting the blood flow forces that tend to detach the particle (e.g. high avidity). Since both margination and avidity of nanoparticles in circulation strongly depends on the geometry of the nanoparticle,^{27, 29–32} we measured the margination rates and avidity of the nanochains in microvasculature constructs under flow conditions using our previously established in vitro method.³² The experiments were conducted in a microfluidic flow network setup (Figure S2c in Supporting Information), because channel dimensions and infusion rates can be accurately controlled removing the complexity of in vivo studies. Firstly, TEM images of the nanochain suspension in cell culture media were obtained, after the nanoparticles were flowed in the microchannel for 20 min at 50 μL/min, indicating that the particles maintain their structural integrity under flow conditions (Figure S2c). To separate margination from targeting avidity, we initially evaluated the margination of non-targeted nanochains. To avoid undesirable specific binding events, the channel was coated with fibronectin, which captures marginating particles in a broad non-specific manner.³² At a flow rate of 50 μL/min, which is in the range of expected blood flow in tumor microcirculation,³³ the nanochain exhibited 2.3-fold higher margination than the IO sphere (Figure S2d). Targeting avidity of the RGD-NC particle was also assessed under flow using the microfluidic device coated with TNF-α-treated BAEC cells. RGD-NC nanoparticles and RGD-targeted nanospheres displayed a biphasic behavior comprising of an initial rapid attachment phase followed by a slower attachment rate (Figure S2e). Importantly, after 5 and 20 min, the RGD-NC nanoparticles achieved 9.5 and 2.9-fold higher attachment compared to their spherical counterparts. We should note that the ligand density on the surface of nanospheres or nanochains was the same being about 25 RGD peptides per sphere. Thus, the total number of RGD peptides on a nanochain was about 100.

Targeting the primary tumor

We used the orthotopic 4T1 mammary tumor model in mice to assess the potential utility of RGD-NC nanoparticles for detection of primary tumors and metastases. The 4T1 cell line is one of the few breast cancer models that efficiently metastasizes to sites and organs similar to that observed in the human disease.^{34, 35} Previous studies³⁴ have shown that growth of cells at the primary site displays a biphasic behavior: 1) the primary tumor rapidly grows in the first 2 weeks after inoculation of tumor cells in the mammary fat pad; 2) the tumor shrinks in the next 2 weeks due to infiltration of leukocytes and extensive necrosis; 3) during the 5th week, the tumor grows again with metastases occurring primarily in the liver, spleen and lungs.

Initially, we compared the RGD-NC particles to RGD-targeted IO nanospheres (RGD-NS) and their non-targeted variants in their ability to target the primary tumor (early-stage: week 2 after tumor inoculation). All formulations were administered at a dose containing an equal number of particles per kg of body weight (i.e. 1.3 × 10¹⁴ particles/kg b.w. corresponding to 0.21 and 0.87 mg Fe/kg b.w. for the IO nanospheres and nanochains, respectively). Using Fluorescence Molecular Tomography (FMT) imaging, we non-invasively and quantitatively monitored the time-dependent intratumoral accumulation of the various particles (equal number of NIR fluorophores/particle). Figure 2a shows representative FMT images that were taken 30 min after systemic administration of the formulations. Due to their enhanced multivalent docking, the RGD-NC nanoparticles substantially outperformed the targeted nanospheres. As expected, the non-targeted nanochains exhibited slow accumulation into the primary tumor primarily due to the EPR effect. As shown in Figure 2b, the intratumoral accumulation of the non-targeted nanochains was about 5.9% of the injected dose at t=3 hours post-injection. We should emphasize that the nanochains and nanospheres were not fully covered with the polyethylene glycol (PEG) coating. This is evident by the blood residence time of the nanoparticles shown in Figure 2c (blood t_1/2 ~20 min; measured in the heart), since the circulation time of nanoparticles depends on the degree of PEG shielding.^{36, 37} Our objective was to detect signal from the RGD-NC particles targeting integrins on the tumor-associated vascular bed with no interference from signal in the vascular site. If the particles are bound to the tumor vasculature, the time point of maximum signal from the tumor site should coincide with low concentration of the nanoparticles in the bloodstream. Figure 2d shows the time course of the fluorescent signal in the tumor for each formulation. Indeed, the RGD-NC-injected animals displayed maximum signal in the tumor in the 30–60 min time window, while the nanoparticles were almost depleted from the bloodstream. Most importantly, at t=45 min after injection, the tumor accumulation of RGD-NC was 8-fold higher than their non-targeted variant. At that time point, vascular targeting via the RGD peptide resulted in more than 40% of the administered nanochains being localized in the primary tumor. Quantitative measures of the intratumoral deposition of each particle (i.e. area under the curve and maximum signal denoted as AUC^0→1h and signal_max, respectively) are shown in the table of Figure 2d. In general, these data are consistent with the in vitro targeting experiments showing that the RGD-NC particles achieved significantly higher vascular targeting at an earlier time point. At later time points, low levels of agent remained in circulation 24 hours after administration. Mostly the agent was found in the liver and spleen with less that 5% of the injected dose being in the heart, kidney.

Figure 2.

Evaluation of the ability of the RGD-NC nanoparticles to target primary 4T1 tumors (early-stage tumor model: week 2 after tumor inoculation). (a) FMT images show the accumulation of RGD-targeted and non-targeted IO spheres and nanochains in primary tumors at 30 min post-injection (dose: equal number of particles per kg of body weight). The nanoparticles of each formulation exhibited the same fluorescence signal per particle. (b) Quantification of the time-course of accumulation of the non-targeted nanospheres and nanochains in the tumor due to the EPR effect. (c) Time-course of the amount of nanoparticles in the heart as a measure of the blood residence time of each formulation. (d) Comparison of the intratumoral accumulation of targeted nanochains and nanospheres and their non-targeted variants in the first 1 hour after administration. It should be noted the range of x- and y-axis are different between Figures 2b and c. While the RGD-targeted IO spheres exhibited higher tumor accumulation than the non-targeted formulations, they were substantially outperformed by the RGD-NC nanoparticles (data presented as mean ± standard deviation). (e) The primary tumors of animals injected with NS, NC, RGD-NS or RGD-NC were perfused, excised, and weighted 30 min after administration. After digestion of the tissues, the iron concentration was measured using inductively coupled plasma optical emission spectroscopy (ICP-OES). Control animals were used to correct for background levels of endogenous iron. In the FMT and ICP measurements, data points marked with asterisks are statistically significant relative to all groups (n = 6 animals per formulation; * P < 0.05).

To validate the in vivo FMT-based measurements, the iron concentration of the primary tumors of animals injected with the formulations was directly measured ex vivo using ICP-OES (inductively coupled plasma optical emission spectroscopy). Animals injected with saline were used for correction of the background levels of iron in the tumor tissue. Fig. 2e shows the intratumoral iron content 30 min after injection of targeted and non-targeted nanochains and nanospheres, verifying the patterns observed with FMT imaging. While the FMT measurements slightly overestimated the intratumoral concentration of the nanoparticles compared to the ICP measurement, the only statistically different condition was the case of RGD-NS. Notably, ICP confirmed that more than 35% of the administered RGD-NC particle accumulated in the tumor within 30 min after administration.

Targeting metastases

To evaluate the ability of the RGD-NC particles to target metastasis, we used mice with a late stage 4T1 tumor (week 5 after tumor inoculation). We performed whole body angiography at 99 μm resolution using a micro-CT system (Siemens Inveon) and a liposomal imaging agent encapsulating a high cargo of an iodinated contrast agent.^38–40 Consistent with previous reports,³⁴ Figure 3a indicates that the primary tumor of this animal model at a late-stage presented a necrotic core with little internal vascularization and a vascularized periphery. The same animal was systemically injected with RGD-NC nanoparticles tagged with an NIR fluorophore and imaged with the FMT system. 3D-rendered volumes of the micro-CT angiogram and the FMT image (45 min post-injection) were co-registered using a script-based software (in MATLAB) and the interactive visualization software Amira (Figure 3b). The accumulation of the RGD-NC particles in the primary tumor site was primarily observed in the location of blood vessels, which is consistent with the expression of integrins due to tumor-associated angiogenesis. Importantly, significant fluorescence signal was detected in other organs away from the primary tumor site. Based on previously published work,⁴¹ regions of interest (ROIs) were selected in the FMT image to indicate the location of major organs (Figure 3c), showing that the RGD-NC nanoparticles accumulated in other organs besides the primary tumor (e.g. liver and lungs). To confirm the colocalization of RGD-NC particles and metastatic tumors, organs were imaged ex vivo using a CRi Maestro fluorescence imaging system, since the 4T1 cell line was engineered to stably express green fluorescence protein (GFP). We should note that the fluorescence signals from GFP and the nanoparticle’s NIR tag do not overlap. Figure 3d confirms the presence of metastatic tumors in the liver, spleen and lungs (other organs are not shown, since no signal from GFP was detected). Grossly, the metastatic tumors appeared as white nodules in bright field imaging, compared to the darker liver parenchyma. More importantly, the signals from metastatic cancer cells (green) and RGD-NC particles (red) overlaid significantly (yellow) indicating the localization of the nanoparticles in metastatic lesions.

Figure 3.

Imaging of metastases. (a) Micromorphological imaging of normal and tumor vasculature at 99 Jm resolution of a metastatic 4T1 tumor (week 5) using a Siemens Inveon micro-CT and a liposome-based iodinated contrast agent. (b) Co-registration of the micro-CT image with the FMT image of the same animal injected with the RGD-NC nanoparticles. (c) The ROIs indicate the location of the tumor and different organs as obtained from previously published work.⁴¹ (d) Using a CRi Maestro fluorescence imaging system, ex vivo imaging of organs indicates the colocalization of RGD-NC particles and 4T1 metastatic cells expressing GFP.

The efficacy of the RGD-NC nanoparticles to target metastatic tumors was quantitatively evaluated in a group of mice harboring metastatic 4T1 tumors (n=6) using FMT imaging. Figure 4a shows representative images of a normal mouse (top row) and a mouse with metastases (bottom row) imaged with the FMT system at t=30 min after injection of RGD-NC. The relatively low signal in the lungs of normal animals (n=6) suggested the presence of the agent primarily in the bloodstream. Since RGD-NC nanoparticles are primarily cleared by liver Kupffer cells (and splenic macrophages), the liver of the same animals exhibited relatively appreciable signal compared to the lungs. On the other hand, FMT imaging of mice with late stage 4T1 tumors showed significant accumulation of RGD-NC primarily in regions of the liver and lungs. Using the designated ROIs for each organ (as shown in Figure 3c), we measured the concentration of RGD-NC in locations of those organs displaying significantly enhanced signal (i.e. hot spots). In each ‘metastatic’ animal, we identified 1–3 hot spots in the liver and lungs designated as ROI-1 and ROI-2 in Figure 4a, respectively. The quantitative analysis shown in Figure 4b revealed a significant concentration of the agent in these hot spots. More importantly, these hot spots displayed a 15 and 7.2-fold increase of signal compared to the background signal in healthy liver and lungs.

Figure 4.

Evaluation of the ability of the RGD-NC nanoparticles to target metastatic 4T1 tumors (late-stage tumor model: week 5 after tumor inoculation). (a) Representative FMT images show the accumulation of the RGD-NC particles in the liver and lungs of healthy and metastasis-bearing mice at 30 min post-injection. In the animal with metastases, hot spots with a significantly elevated concentration of the particles are indicated in the liver and spleen as ROI-1 and ROI-2, respectively. (b) Quantification of the fluorescence signal obtained from the FMT images of a group of healthy mice and a group of metastatic mice 30 min after injection of RGD-NC particles (data presented as mean ± standard deviation). The signal of the hot spots in the lungs and liver of the metastatic group was compared to the average signal of these organs in the healthy group (n = 6 animals per group).

Imaging of metastasis using MRI

To detect metastases using a clinically relevant imaging modality, we performed imaging with MRI. Figure 5 shows representative coronal T2-weighted images of healthy mice (n=3) and metastatic 4T1 mice (n=2) obtained using a 9.4 Tesla MRI before and after administration of the RGD-NC nanoparticles (at a dose of 1.74 mg Fe/kg b.w.). This dose is substantially lower than typical dose of IO nanoparticles used in MR imaging studies (e.g. 10 mg Fe/kg).⁴² MR images were acquired a few minutes prior to injection of the agent and 15, 30, 45 and 60 min after injection. The scanning parameters in the pre- and post-injection images were identical. Figure 5a compares the pre-injection and 45-min post-injection images of the liver in a metastatic animal. The uptake of the agent by the macrophages in the liver generated an appreciable negative contrast. However, targeting of the RGD-NC nanoparticles to metastatic lesions achieved a significantly higher negative contrast (yellow arrows in Figure 5a) that ‘overshadowed’ the background contrast in the liver. As described in the next section, histological evaluation of the liver confirms the accumulation of the agent in metastatic tumors. Figure 5b shows a 45-min post-injection image of a healthy liver demonstrating that the uptake of the agent in the liver generated homogeneous contrast with no ‘hot’ spots. To quantitatively evaluate the ability of the RGD-NC nanoparticles to target metastasis, the absolute MR signal intensity in the metastatic lesions and the healthy liver was measured using manually drawn regions of interest (ROI). Figure 5c shows that the time-course of the signal intensity in the hot spots or the entire healthy liver (normalized to the signal of an adjacent muscle; scale: 0–1). Lower values indicate greater contrast in T2 images, and a normalized intensity value of 1 corresponds to no contrast compared to the pre-contrast image. The pre-injection values for both the normal liver and the ‘hot’ spots in the metastatic liver were fairly similar and close to 1. As expected, due to clearance of the particles by the liver, injection of RGD-NC resulted in contrast enhancement in the healthy liver with an intensity value of 0.83. However, the metastatic lesions exhibited a normalized signal intensity value of ~0.44 in the post-injection images, indicating significantly higher contrast compared to the post-injection background signal of the healthy or uninvolved regions of the metastatic liver. It is important to note that this contrast may be further improved by optimization of the MRI imaging parameters or by quantifying the T2* relaxation values directly.

Figure 5.

Representative in vivo MR images of the liver in normal and ‘metastatic’ mice using a 9.4 Tesla MRI. (a) Coronal T2-weighted images of the liver of a metastatic mouse before and 45 min after injection of the RGD-NC nanoparticles. In the 45-min post-injection image, the yellow arrows show micrometastases of about 0.5 mm in size with increased contrast enhancement. (b) Coronal T2-weighted images of the liver of a normal mouse 45 min after injection of the RGD-NC nanoparticles. (c) The time-course of the MR signal intensity in the liver ‘hot’ spots was quantitatively evaluated. The absolute MR signal intensity in the metastatic lesions and the healthy liver was measured in manually drawn ROIs. The signal intensity in the hot spots or the entire healthy liver was normalized to the signal of an adjacent muscle (scale: 0–1). Since lower values indicate greater contrast in T2 images, normalized intensity values of 0 and 1 correspond to maximum and minimum contrast, respectively, compared to the pre-contrast intensity values (data presented as mean ± standard deviation; n=3; each metastatic animal exhibited 2–4 hot spots; *P<0.05).

Histological evaluation

After the last in vivo imaging session, tissues were collected and histological analysis was performed to confirm the localization of the RGD-NC nanoparticles in metastatic tumors. We histologically examined the animals used in the in vivo imaging studies verifying the presence of RGD-NC nanoparticles in the majority of metastases. We evaluated the location of the metastatic cancer cells with respect to blood vessels and the associated expression of α_vβ₃ integrin. Using fluorescence microscopy, we verified that metastatic tumors were present in the liver and lungs of all the animals at a late stage (week 5 after tumor inoculation). Images of entire histological sections of the organs were obtained at a low magnification (5x) using the automated tiling function of the microscope. A representative image of the left lobe of liver is shown in Figure 6a displaying the presence of clusters of metastatic cells (green) dispersed in the liver parenchyma. Imaging at higher magnification showed that the metastatic cancer cells were localized primarily on the endothelial walls (Figure 6b). Furthermore, these were exactly the locations that exhibited ‘remodeling’ as indicated by the overexpression of α_vβ₃ integrins (Figure 6c). We should note that negligible integrin expression was observed in the normal parenchyma of liver (images not shown). The expression of integrins of metastatic cancer cells on the endothelium should favor vascular targeting of the RGD-NC nanoparticles to metastatic tumors. Indeed, fluorescence microscopy showed that RGD-NC particles were predominantly distributed around those same blood vessels colonized by 4T1 cells. Furthermore, the nanoparticles colocalized with the integrin expression of 4T1 cancer cells (Fig, 6d–f). In addition to fluorescence microscopy showing the colocalization of cancer cells and RGD-NC particles, bright field microscopy was performed on the same histological sections after standard hematoxylin-eosin staining (Figure 6g–h).

Figure 6.

Histological evaluation of the microdistribution of the RGD-NC nanoparticles in the liver of animal with metastatic tumors. (a) Fluorescence image of a histological section of the left lobe of the liver (5x magnification; blue: nuclear stain (DAPI); green: 4T1 cancer cells (GFP)). Images of entire histological sections of the organs were obtained using the automated tiling function of the microscope. (b–c) The location of metastatic cancer cells is shown with respect to endothelial cells and expression of αvβ3 integrin in the same histological section (10x magnification; green: 4T1 cancer cells; red: endothelial cells; blue: α_vβ₃ integrin). (d–f) The RGD-NC particles accumulated in locations of 4T1 cells that expressed avβ3 integrin (10x magnification; blue: DAPI; green: 4T1 cancer cells; yellow: α_vβ₃ integrin; red: RGD-NC). (g–h) Fluorescence and bright field microscopy was performed on histological sections stained with hematoxylin-eosin showing the colocalization of RGD-NC and cancer cells and their relative anatomical location in the liver (5x and 10x magnification in g and h; green: 4T1 cancer cells; red: RGD-NC; yellow: overlay).

We also performed histological analysis on the lungs of the same animals. Figure 7a (Supporting Information) shows that micrometastases were present in the lungs at week 5 after tumor inoculation. Similarly to the liver, the RGD-NC particles accumulated in those locations that were colonized by metastatic cancer cells (Figure 7b). Notably, Figures 7c–e indicate that the location of RGD-NC coincided with overexpression of α_vβ₃ integrins on the metastasized cancer cells.

Figure 7.

Histological evaluation of the microdistribution of the RGD-NC nanoparticles in the lungs of animal with metastatic tumors. (a–b) The colocalization of fluorescently-tagged RGD-NC particles and metastatic cancer cells is shown in the same histological section (5x magnification; green: 4T1 cancer cells; red: RGD-NC; blue: DAPI). (c–e) The location of RGD-NC particles is shown with respect to metastatic cancer cells and expression of α_vβ₃ integrin in the same histological section (40x magnification; blue: DAPI; green: 4T1 cancer cells; yellow: αvβ3 integrin red: RGD-NC).

DISCUSSION

By defining the topology of two different functional groups on the surface of the parent IO nanospheres, we were able to assemble four nanospheres in a linear orientation. Furthermore, the robustness of the nanochain synthesis^{25, 26} enabled fabricating the chain-shaped nanoparticle with a high degree of uniformity. The effect of shape^{43, 44} and clustering of iron oxide cores^{45, 46} significantly increased the T2 relaxivity per particle compared to their spherical counterparts, which is highly beneficial in imaging of vascular targets.

Another advantage of the nanochain technology is the control of the shape and overall dimensions of the nanoparticle. The in vitro flow and in vivo FMT imaging studies revealed that the oblate shape of the RGD-NC nanoparticles resulted in superior attachment of the nanochains to the target site compared to spherical nanoparticles. The shape of nanoparticles has been shown to play a critical role in their margination towards the vessel walls in microcirculation.^{30, 31, 47} In contrast to spherical particles, oblate-shaped particles are subjected to torques resulting in tumbling and rotation.^{29, 30} These complex dynamics of non-spherical particles cause translational as well as rotational motions. In a previous study,³² we showed that a rod-shaped nanoparticle displayed ~8 times higher margination rate than a comparable nanosphere. Thus, by defining the geometry of the RGD-NC nanoparticle with an aspect ratio of about 5,^{27, 48} the particles had an increased ability to sense the overexpression of specific receptors (i.e. β₃ integrins) of the tumor-associated endothelium, which was our ‘docking site’ on metastatic tumors.

In addition to the high probability of the RGD-NC particles to interact with the vascular bed of tumors, the shape of the RGD-NC particles enhanced their adhesive strength for vascular targeting. In general, while nanoparticles provide enhanced targeting avidity due to multivalent interactions, the geometry of the nanochains further enhanced the ligand-receptor interactions increasing the likelihood of specific recognition of tumors. Successful vascular targeting requires that the adhesive strength of the nanoparticle offsets the hemodynamic forces applied on the particle tending to detach the particle from the endothelium. In comparison to spherical nanoparticles, oblate-shaped nanoparticles, such as the RGD-NC nanoparticle, form a greater number of ligand-receptor occurrences per particle while the blood flow-related forces are lower.²⁷

Even with their high avidity, one would expect a relatively small number of RGD-NC particles to accumulate at sites of metastasis due to the small size, dispersion and low degree of neovascularization of metastatic lesions. While tumors larger than 100 mm³ may favor the EPR effect resulting in deposition of high amounts of nanoparticles in primary tumors, this is not the case for metastatic tumors.⁴⁹ For example, liver is one of the most common organs for metastasis of breast cancer.^{50, 51} Typically, normal liver displays a higher vascular density than liver metastases.^{52, 53} Post mortem analysis of histological liver sections from women with metastatic breast cancer revealed that metastatic tumors presented a lower number of vessels than the liver parenchyma.⁵⁰ In conjunction with the absence of EPR-driven accumulation of nanoparticles into metastases, it is apparent that successful vascular targeting becomes essential.

Thus, the biological mechanisms of metastasis have to be carefully considered to design metastasis-targeted nanomaterials. During cancer progression, tumor cells show changes in their plasticity, including the epithelial to mesenchymal transition (EMT), which is strongly associated with cell invasion and migration. Numerous factors have been identified to induce EMT including substrate binding integrins, resulting in a mesenchymal-like cell that is able to move freely by forming focal adhesions.^54–56 Following intravasation, integrin expression on the circulating cancer cells plays central role in the formation of metastases at a distal site. Extensive experimental evidence also shows that tumor cells interact with platelets supporting tumor metastasis. β₃ integrin-mediated binding to activated platelets protects tumor cells from the immune system and promotes their arrest at the endothelium, supporting the establishment of metastatic secondary lesions.^{10, 11} While initial adhesion onto endothelium involves selectins, CD44, and other glycoconjugates, the metastatic site transitions from selectin-dependent tumor cell rolling on the endothelium to firm attachment that is mediated by integrins.^{10, 12} Not surprisingly, analysis of the adhesiveness of blood circulating cancer cells revealed that cell attachment to the vasculature requires a key integrin, α_vβ₃.^13–15 Therefore, peptides targeting α_vβ₃ integrin are exceptionally capable ligands to direct the nanochains to locations of metastasis.

Besides primary and metastatic tumors, the α_vβ₃ integrin is expressed in the extracellular matrix, since this integrin mediates the adhesion of normal cells to a large number of extracellular matrix proteins, including fibronectin, fibrinogen, collagen, and laminin. However, due to its size, a nanochain cannot extravasate through normal endothelium into normal tissues. More importantly, our strategy depends on vascular targeting with the targeting site being the endothelium of metastasis-associated blood vessels. While α_vβ₃ integrin is minimally expressed on normal resting blood vessels, it is significantly upregulated in the lumen of newly formed blood vessels within primary tumors or blood vessels that have been colonized by metastatic cancer cells.^{25, 57, 58} Thus, besides α_vβ₃ integrin in blood vessels during wounds healing,⁵⁹ vascular targeting of the RGD-NC particle provides specific binding to metastatic sites.

Indeed, histological evaluation of metastatic tumors showed that clusters of metastatic cancer cells were residing on the endothelium of blood vessels, which coincided with an overexpression of integrins. More importantly, the RGD-NC particles colocalized in exactly these endothelial walls associated with the metastatic cancer cells. Even though the RGD-NC particles exhibited high avidity for integrin-expressing cells, it is very challenging to discriminate the vascular targeting-based signal from the intravascular signal of the agent circulating in the bloodstream of a tumor vessel associated with a metastatic tumor.³⁹ While prolonged blood circulation of nanoparticles may be desirable in therapeutic scenarios, detection of metastasis will be ‘masked’, since the signal from high levels of circulating nanochains in the bloodstream will ‘interfere’ with the signal from successfully targeted nanochains on the tumor-associated vascular bed. The ideal scenario of a clinically relevant imaging protocol for detection of metastases would require a few hours using an imaging agent with suitable time windows for its targeting and clearance from the bloodstream. Such imaging agent should sufficiently circulate to effectively target metastatic lesions in a short time frame (i.e. within hours after administration), at which time point its concentration in the blood is low, resulting in the greatest contrast enhancement. Thus, the blood circulation (Figure 2c) and tumor targeting profiles of the RGD-NC nanoparticles (Figure 2d) satisfy the design criterion for a nanoparticle-based imaging agent.

In particular, imaging metastasis in the liver with targeted nanoparticles presents challenges due to the fact that nanoparticles are cleared by Kupffer cells in the liver, which creates background signal and potential false positive diagnoses. However, the distribution of cleared nanoparticles is relatively homogeneous in the liver resulting in homogeneous contrast enhancement without hot spots. Furthermore, the dose of the RGD-NC agent used in this study is relatively low resulting in low and homogeneous contrast enhancement in the liver due to clearance of the agent. Notably, the metastatic sites appeared as hot spots with significantly higher contrast enhancement than the background signal from the rest of the liver tissue. However, further studies are required to accurately measure the diagnostic sensitivity, specificity and positive predictive value of this technology.

To date, imaging applications of nanoparticles have focused predominantly on targeting tumors at the primary site (e.g. integrins, folate receptor, EGFR, HER2 etc).^{16–24, 60–65} Even though attention of the field has recently shifted towards metastasis, there are still very few examples of diagnostics for metastases,⁶ including contrast agents for radioimaging of molecular markers (e.g. targeting α_vβ₃ integrins^{66, 67} or CXCR-4⁶⁸), visualization aids for surgery, and HER2-targeting IO nanoparticles.⁶⁹ While the detection accuracy of CT, MRI and fluorodeoxyglucose positron emission tomography (FDG-PET) for relatively large metastatic tumors (>1 cm in size) has significantly improved in the recent past,^70–72 current imaging methods rarely detect the early spread of metastatic tumor cells,⁷³ which prohibits early and most likely effective interventions. Treatment decisions of metastatic disease (e.g. surgical resection, radiation, thermal and radio-frequency ablation) heavily depends on imaging and diagnostic tools to detect the precise location and extent of disease spread.⁷⁴ Micrometastases (0.2–2 mm in size) can rarely be detected via changes in perfusion using radionuclide or Doppler perfusion techniques.^{75, 76} In this work, we show that nanoparticles offer unique opportunities to deal with the complexity of metastatic cancer by exploiting properties that appear at the nano-scale.

CONCLUSION

In conclusion, we demonstrated that the geometry of a chain-shaped nanoparticle promoted targeting of metastatic tumors due to multivalent docking onto integrins of the vascular bed of metastasis. Using multimodal in vivo imaging (i.e. FMT and MRI), we were able to image metastatic tumors in the liver and lungs in a highly aggressive breast tumor model. As the biological mechanisms of metastasis continue to unravel, we expect that more surface markers of metastatic lesions will be identified that can be employed by nanoparticle delivery systems for targeting metastatic disease.

METHODS

Synthesis and characterization of the nanochain particles

The nanochains were synthesized following our previously published method.²⁶ Briefly, solid-phase chemistry was used to partially modify the surface functionality of IO nanospheres. CLEAR resin (Peptides International Inc, Louisville, KY) functionalized with amines was modified with a homobifunctional cleavable cross-linker reactive towards amines (DTSSP). Amine-functionalized IO nanospheres were introduced, allowed to bind to the solid support and then cleaved off using a reducing agent (TCEP). The same type of resin was used and the modified spheres with surface asymmetry were introduced in a step-by-step manner. After recovering the chain via a reducing agent, the suspension was further cleaned using dialysis. The nanoparticles were characterized in terms of their size (DLS), structure (TEM), and magnetic relaxivity (Bruker minispec relaxometer). The cyclo (Arg-Gly-Asp-D-Phe-Cys) or c(RGDfC) was conjugated onto PEG(3400) via maleimide chemistry. In addition to conjugation of the cRGDe peptide, the nanochain particles were tagged with an NIR fluorophore (Vivotag 680) to be detectable by FMT imaging or fluorescence spectroscopy or microscopy. Details of the synthesis and characterization of the nanoparticles are shown in the Supporting Information.

Mouse tumor model

All animal procedures were conducted under a protocol approved by the CWRU IACUC. We used an orthotopic 4T1 breast tumor model in mice.⁷⁷ The 4T1 cell line was engineered to stable express green fluorescent protein (GFP) to allow tracking and quantification of the cells in vivo and histologically. Briefly, we inoculated 0.5 × 10⁶ 4T1 cells orthotopically in a no. 9 mammary fat pad of female BALB/c mice that was surgically exposed while the mice were anesthetized. The animals were used in the in vivo studies at week 2 (only primary tumor) or week 5 (primary and metastatic tumors). Based on our prior experience, we chose these time points, since they represent different stages of angiogenesis, necrosis, invasion, and metastasis and are informative and relevant to the human disease.

Fluorescence Molecular Tomography

We performed fluorescence imaging on the 4T1 mammary model in mice (at week 2 or 5) using the FMT 2500 Quantitative Tomography In Vivo Imaging System (Perkin Elmer). Phantoms for each nanoparticle formulation were used to calibrate the FMT to take quantitative deposition measurements. We then intravenously injected each of the four formulations at a dose of 1.3 × 10¹⁴ particles per kg b.w. The animals were imaged before and after IV injection of the formulations at multiple time points (15, 30, 45 min and 3, 6, 24 h). After the last imaging session, tumor and organs (kidneys, lungs, brain, liver, spleen and intestine) were retrieved and weighed. To verify the findings of the in vivo imaging and confirm the presence of metastases in the organs, we imaged the organs ex vivo using the FMT and a CRi Maestro fluorescence imaging system. The organs were then processed for histological analysis.

Angiography using contrast-enhanced micro-CT imaging

Contrast-enhanced angiography was performed using a Siemens Inveon micro-CT system (isotropic 99 Jm resolution, 80 kVp, 500 μA) and a long-circulating liposomal imaging agent encapsulating a high cargo of an iodinated contrast agent.³⁹ Following IV injection of the agent at a dose of 2.6 g iodine/kg b.w., the animals were imaged with the micro-CT system. Subsequently the animals were IV injected with RGD-NC followed by imaging with the FMT system. The two images were co-registered using a semi-automatic 3D segmentation-based registration approach that was implemented in script-based software (in MATLAB) and interactive visualization software Amira (Visage Imaging Inc).⁷⁸ Fiducial markers placed around the tumor mass were visible in both the micro-CT and the FMT images. Using a region growing algorithm with seed points defined by the user, we segmented the fiducial markers from both volumes.⁷⁹ Then, we used Amira’s AffineTransform tool to register the floating volume landmarks to the reference volume landmarks.

MRI imaging

MRI images were acquired on a 9.4 T Bruker MRI system. A volume coil (3.5-cm inner diameter) was employed. High resolution images were obtained before and 15, 30, 45, 60 and 120 min after IV injection of the RGD-NC nanoparticles (at a dose of 7.5 mg Fe/kg b.w.) using a T2-weighted RARE sequence with the following parameters: TR/TE=1000/45 msec, matrix=128 × 128, FOV=5 × 5 cm, and 1 average. This resulted in an in-plane spatial resolution of 390 Jm and a slice thickness of 1 mm.

Histological evaluation

After the last imaging acquisition with FMT or MRI, tissues were collected from the mice for histological studies. The animals were anesthetized with an IP injection of ketamine/xylazine and transcardially perfused with heparinized PBS followed by 4% paraformaldehyde in PBS. Tumors and organs were explanted and post-fixed overnight in 4% paraformaldehyde in PBS. The tissues were soaked in 30% sucrose (w/v) in PBS at 4 °C for cryosectioning. Serial sections of 12 μm thickness were collected using a cryostat (Leica CM 300). To visualize the tumor microvasculature, the tissue slices were immunohistochemically stained for the endothelial antigen CD31 (BD Biosciences, Pharmingen). The tissues were also stained with the nuclear stain DAPI. Standard hematoxylin-eosin staining was also performed. The tissue sections were imaged at 5, 10 or 40x on the Zeiss Axio Observer Z1 motorized FL inverted microscope. To obtain an image of the entire tissue section, a montage of each section was made using the automated tiling function of the microscope.

Statistical analysis

Means were determined for each variable in this study and the resulting values from each experiment were subjected to one-way analysis of variance with post hoc Bonferroni test. A P value of less than 0.01 was used to confirm significant differences. Normality of each data set was confirmed using the Anderson-Darling test.