Engineered biomimetic nanoparticle for dual targeting of the cancer stem-like cell population in sonic hedgehog medulloblastoma

Microfluidic Synthesis and Characterization of eHNPs.

With our microfluidic technology enabling a simple and facile single-step reconstitution of eHNPs with high homogeneity and reproducibility (SI Appendix, Fig. S1) (27, 28), we are able to not only produce the nanoparticles with high uniformity, but also decorate an antibody of the murine SHH MB cancer stem-like cell target, stage-specific embryonic antigen-1 (SSEA-1⁺) or CD15 (SI Appendix, Fig. S2), on the surface of eHNP-A1 while incorporating LDE-225 into the nanoparticle core (Fig. 2A). Four different types of eHNPs were synthesized to evaluate the effect of each component (SI Appendix, Table S1). The synthesized eHNP-A1 exhibited a homogeneous discoidal shape of 13 nm in size (Fig. 2 B and C), like natural pre-beta HDL (29). Incorporation of CD15 into eHNP-A1 slightly increased the size to 15 nm (Fig. 2B), while maintaining the discoidal shape (Fig. 2C). Loading of LDE225, a hydrophobic drug used to treat MB by inhibiting hedgehog signaling (30), into the core of eHNP-A1 led to a large spherical particle with a size up to 25 nm (Fig. 2 B and C). The nanoparticle shell formed by apolipoprotein A1 and lipid resulted in a greater structural stability than lipid micelles while keeping LDE225 in the hydrophobic core. The target nanoparticle of eHNP-A1-CD15-LDE225 exhibited a similar size distribution with a slightly larger peak of 28 nm with a spherical shape. Our composition analysis showed that the high amounts of LDE225 in eHNP-A1-LDE225 and eHNP-A1-CD15-LDE225 were 7.2% (wt/wt) and 5.7% (wt/wt), respectively (Table 1 and SI Appendix, Fig. S3), owing to the structural stability provided by the scaffold protein apolipoprotein A1.

eHNP as a Delivery Carrier Capable of Crossing the Intact BBB to SHH MB.

We first confirmed that eHNP-A1 can cross the BBB and get into the brain parenchyma with the accumulation of eHNP-A1 in the brain observed for 24 h after i.v. injection (Fig. 3A and SI Appendix, Fig. S4). Before we tested the ability of our target nanoparticle with multiple components to cross the BBB and deliver drugs to the targeted sites in the brain, we first investigated whether the BBB is intact and maintains the barrier function predicted in an in vivo brain tumor model of a SmoA1 MB tumor-bearing mouse. The SmoA1 mice were crossed with Math1-GFP to generate homozygous SmoA1^+/+:Math1-GFP^+/+ strain. These mice spontaneously develop tumors and were used between 4 and 6 mo of age (31). Tumor-bearing mice were identified with symptoms of enlarged posterior fossa, head tilt, ataxia, and hunched posture. After confirming the tumor growth, i.v. injections of Hoechst dye and Alexa-594 conjugated lectin in SmoA1-GFP tumor-bearing mice resulted in the localization of the Hoechst dye/Lectin-594 (red) in the areas near the vasculature, and its absence from the bulk of the tumor (green) (Fig. 3 B and C), in contrast to a widespread distribution in liver tissue sections from the same tumor-bearing mice (Fig. 3D). This result confirmed that the SmoA1 mouse MB model has an intact BBB, as was previously shown by others in a different murine MB model (32). We then confirmed the ability of eHNP-A1 to cross the intact BBB and accumulate in the MB cells of the SmoA1-GFP-MB-bearing mouse model upon i.v. injection (Fig. 3 E, Upper).

eHNPs Demonstrate Receptor-Facilitated Delivery into SHH MB.

Like natural HDL, our eHNP-A1 may interact with several receptors, including SR-B1 on brain endothelial cells, enabling the release of loaded cargos into the brain parenchyma through receptor-mediated transcytosis (Fig. 1). We evaluated the cytosolic internalization of eHNPs facilitated by several receptors on SHH MB cells using confocal microscopy (Fig. 4A and SI Appendix, Fig. S5). Anti-CD15 incorporation into eHNP-A1s (eHNP-A1-CD15 and eHNP-A1-CD15-LDE225) enhanced cellular internalization up to four times greater than that of eHNP-A1, owing to the “dual targeting” receptor-mediated uptake by CD15 and SR-B1. In addition, cytosolic internalization of eHNP-A1-CD15-LDE225 was significantly reduced upon pretreatment with block lipid transporter-1 (BLT-1), a known inhibitor of SB-B1, and excess free anti-CD15, further demonstrating that dual SR-B1- and CD15-mediated cellular uptake led to the efficient cellular delivery of the eHNPs (Fig. 4B and SI Appendix, Fig. S6).

We then probed the targeting efficiency of the eHNPs in the brain tumor microenvironment, especially to the CD15-positive cancer stem-like cells. We standardized the culture conditions for the growth of SmoA1 organotypic tumor slice cultures ex vivo to analyze the effect of antitumor agents in the tumor microenvironment (33). We evaluated the targeting effect of eHNP-A1-CD15-LDE225 and its derivatives on ex vivo organotypic slice cultures (Fig. 4C). The slice cultures tagged with anti-CD15 (eHNP-A1-CD15) exhibited more localization in the perivascular niche areas of tumor tissues when compared to anti-CD15-free eHNPs (eHNP-A1 and eHNP-A1-LDE225). Subsequently, we i.v. administered eHNP-A1-CD15 to the SmoA1-GFP tumor-bearing mice and examined the cryosectioned tissues under a confocal microscope in a time-dependent manner (SI Appendix, Fig. S7). In order to visualize the preferential localization of eHNPs targeting the cancer stem cells within the tumor microenvironment, we first carried out CD15 immunofluorescence staining for cancer stem cells in MB tissue slices and confirmed the higher expression of CD15 in the perivascular areas (Fig. 4 D, Upper), as previously reported by other groups that endothelial cells interact closely with self-renewing brain tumor cells and secrete factors that maintain these cells in a stem cell-like state (Fig. 4D). Both CD15 and eHNP-A1-CD15 colocalize to the perivascular space (Fig. 4 D and E, Lower), whereas other non-anti-CD15-conjugated eHNPs are more diffusely distributed (Figs. 3E and 4C), indicating the targeting effect directed to the stem cell population. Some of the heterogeneous distribution in these images are in part due to the SR-B1-mediated uptake on all MB cells. This result is in agreement with the localization of CD15 antigen in the cancer stem-like cell niche of the brain tumor tissue (Fig. 4D) in the murine medulloblastoma, whereas eHNP-A1-CD15 showed a nontargeted distribution in the liver tissue sections (Fig. 4 E, Upper). Taken together, these results confirmed the ability of the eHNPs to traverse the BBB and target cancer stem-like cell populations in the SmoA1 mouse model of medulloblastoma through dual targeting of cells that are both CD15-positive and demonstrate high expression of SR-B1.

eHNP-A1-CD15-LDE225 Markedly Enhances LDE225 Inhibition of SHH MB.

With the successful BBB penetration and subsequent efficient targeting of SHH MB cells, we next investigated the therapeutic efficacy of eHNP-A1-CD15-LDE225 and its derivatives. To confirm the effect of each component in the proposed nanoparticles, we treated SHH MB cells (DAOY and PZp53) with free LDE225, eHNP-A1, eHNP-A1-CD15, eHNP-A1-LDE225, and eHNP-A1-CD15-LDE225, and monitored the cell viability (Fig. 5A and SI Appendix, Fig. S8). Compared to LDE225 treatment, which exhibited dose-dependent inhibition of cell viability (IC50 ∼2 μM), eHNP-A1-LDE225 treatment dramatically enhanced the therapeutic efficacy (IC50 ∼70 nM). Furthermore, the addition of targeting capability using anti-CD15 (eHNP-A1-CD15-LDE225) remarkably led to 250-fold reduction of the IC50 value (∼8 nM). More interestingly, eHNP-A1 and eHNP-A1-CD15 without drug loading also exhibited marked therapeutic effects; the cell viability following treatment with eHNP-A1 and eHNP-A1-CD15 decreased to 63%, equivalent to a treatment concentration corresponding to 10 μM LDE225, indicating that the eHNP-A1 alone has a therapeutic potential for the treatment of SHH MB, which was also supported by a recent study (10).

The enhanced therapeutic effect of eHNP-A1-CD15-LDE225 shown in SHH MB cells was further investigated using a specific inhibitor of the HDL receptor SR-B1, BLT-1, which mediates bidirectional flow of cholesterol in and out of cells, as well as free anti-CD15 (Fig. 5 B–D and SI Appendix, Fig. S9). Pretreatment with anti-CD15 affected the viability for cells treated with eHNP-A1-CD15-LDE225, but not for cells treated with free LDE225 and eHNP-A1, indicating that CD15 on SHH MB facilitates the uptake of eHNP-A1-CD15-LDE225 to SHH MB cells and contributes to the therapeutic effect. Importantly, pretreatment of DAOY cells with BLT-1 significantly further reduced the efficacy of eHNP-A1 and eHNP-A1-CD15-LDE225, suggesting that SR-B1 not only promotes the interaction of eHNPs with SHH-MB cells, but may also mediate eHNP-induced dysregulation of cholesterol homeostasis within the DAOY cells, which in turn further inhibits SHH signaling pathway. In contrast, following treatment with free LDE225, BLT-1 pretreatment showed only a minimal effect on cell viability, possibly because the blockade of lipid transport by BLT-1 may also affect the Smo pathway which is highly dependent on lipid metabolism. Moreover, we confirmed from a drug release test that acidic pH contributed to a higher drug release (SI Appendix, Fig. S10), indicating that the release of LDE225 partly occurs inside the endosome. The blocking of endosomal acidification using bafilomycin A1 (Baf A1) significantly lowered the therapeutic effect of eHNP-A1-CD15-LDE225, whereas there was an insignificant effect of Baf A1 on free LDE225. This result further demonstrates that the drug releases within the cell, more specifically inside the endosome during the endosomal acidification (SI Appendix, Fig. S11). We also found that the transport mechanism of LDE225 into the cell is collaborative: endocytosis and consequent release upon acidification, and SR-B1-mediated transport bypassing endocytic pathway (SI Appendix, Fig. S12).

Based on the striking inhibition of cell viability observed following treatment with eHNPs in the absence of SR-B1 blockade, we hypothesized that eHNPs enable not only the efficient delivery of LDE225 into the cells, but also enhance the therapeutic effect through mediating the transport of cholesterol out of cells, which is supported by studies demonstrating the critical role of cholesterol in SHH signaling in SHH subtype cells (6–8). To prove this hypothesis, we evaluated the cholesterol efflux in DAOY cells after treatment with the eHNPs (Fig. 5E). DAOY cells treated with eHNP-A1, eHNP-A1-CD15, and eHNP-A1-CD15-LDE225 exhibited much higher cholesterol efflux than nontreated and free LDE225-treated cells. Interestingly, eHNP-A1-CD15-LDE225 showed less cholesterol efflux than that of eHNP-A1 and eHNP-A1-CD15, probably due to the limited room of the nanoparticle core that was loaded with LDE225, despite promoting significantly higher cholesterol transport out of SHH-MB cells compared to nontreated or free LDE225-treated cells.

To ascertain the capability of these eHNPs to cause tumor cell death ex vivo, we first maintained the organotypic tumor slice cultures in the presence of eHNPs and then stained for the cell death marker cleaved-caspase 3 (CC3). A higher CC3 staining (blue) in slice cultures treated with eHNP-A1s containing LDE225 was observed than in non-LDE225-treated ones (SI Appendix, Figs. S13 and S14). To further demonstrate the ability of the nanoparticles to deliver drugs in the murine brain tumors in vivo, we i.v. injected eHNP-A1-CD15-LDE225 and its derivatives to MB-bearing mice and monitored the therapeutic efficacy. We employed two different mouse models of SHH MB, which are the SmoA1-GFP model and Patched (PTC) knockout model. Long-term magnetic resonance imaging (MRI) scans of the tumor growth in both SHH MB models treated by the eHNP-A1-CD15-LDE225-treated group exhibited the well-maintained cerebellum structure and minimal tumor growth, whereas the saline-treated group showed a disrupted cerebellum structure, a blunted edge of lobes, and an outstanding tumor growth (Fig. 5F and SI Appendix, Fig. S15). The volume of SHH MB (PTC) was reduced by 72% in the eHNP-A1-CD15-LDE225-treated group, whereas the saline-treated group showed a 4.3-fold increase in the tumor volume (Fig. 5G), resulting in a markedly lower tumor growth rate and significantly longer survival in the eHNP-A1-CD15-LDE225-treated group than those of the control group (Fig. 5H and SI Appendix, Fig. S16). Similarly, in the other SHH MB-bearing SmoA1-GFP model, tumor growth was significantly inhibited in the eHNP-A1-CD15-LDE225-treated group than that of the saline-treated group, and the eHNP-A1-CD15-LDE225-injected group exhibited significantly enhanced survival rate than the control group, further demonstrating the efficient treatment of SHH MB (SI Appendix, Fig. S17). In order to prove the biological activity of eHNP-A1-CD15-LDE225 and its derivatives, we treated samples to SmoA1-GFP mice and subsequently sectioned and immunostained the tissues for CC3 (Fig. 5 I and J). For eHNP-A1-LDE225 and eHNP-A1-CD15-LDE225, CC3 expressions were high in the brain tumor regions (GFP-positive), demonstrating the effective targeted delivery of LDE225 to the cancer cells in vivo. Interestingly, in a good correlation with in vitro studies, eHNP-A1 and eHNP-A1-CD15 also resulted in a greater degree of CC3 activation than negative control, whereas the degree was much less than that of LDE225-loaded nanoparticles, revealing that the effective cholesterol efflux by eHNPs led to more effective therapeutic outcome in SHH MBs reducing the drug dosage. Moreover, additional apoptosis markers including terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and cleaved poly (ADP ribose) polymerase (PARP) confirmed our findings with well-correlated images and quantitative results (Fig. 5J and SI Appendix, Figs. S18 and S19). Moreover, we identified spinal cord samples with GFP-positive regions to locate metastatic sites of SmoA1 tumors as reported previously (34). Interestingly, we also found the targeted delivery of the nanoparticle payload to all metastatic sites, compared to a minimal level in the liver tissue from the same group of SmoA1-treated mice (SI Appendix, Fig. S20). Targeted in vivo delivery of the LDE225 to the primary and metastatic tumors was further supported by negligible LDE225 release from eHNPs at neutral pH or a serum-containing environment (SI Appendix, Fig. S11). Taken all together, our findings clearly demonstrated that our multifunctional HDL-mimetic nanoparticles can serve as an efficient and potent nanomedicine by facilitating cellular uptake and by exploiting cholesterol transport, a critical factor in SHH MB.

Preparation of Nanoparticles.

A three-inlet microfluidic device that creates propagating microvortices was used for a single-step, continuous, and self-assembly-based synthesis of eHNP-A1-CD15-LDE225. Briefly, an organic phase comprising 2.75 mg/mL of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC, Avanti Polar Lipids), 0.3 mg/mL of 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine (DiD), and 0.2 mg/mL of LDE225 in absolute ethanol was introduced into the center inlet of the microfluidic device at a flow rate of 0.8 mL/min. An aqueous phase consisting of 0.2 mg/mL of apolipoprotein A1 (Millipore Sigma) and 0.05 mg/mL of anti-CD15 (BD Biosciences) in phosphate buffered saline (PBS) (pH = 7.4) was introduced into two outer inlets at a rate of 2.2 mL/min. Each of the organic and aqueous phases was injected using a programmable syringe pump (Harvard Apparatus) with the desired flow rates. After running the syringe pump, the mixture solution from outlet was collected and transferred to a centrifugal membrane filter (EMD Millipore) with a molecular weight cutoff (MWCO) of 50,000, and purified three times in PBS at a speed of 3,900 rpm for 20 min. Synthesized eHNP-A1-CD15-LDE225 was stored in PBS at 4 °C for further usage.

Characterization of Nanoparticles.

The hydrodynamic volumes of samples were obtained by a Zetasizer Nano (Malvern Instrument). Size and morphology of samples were taken from a transmission electron microscope (TEM) (Hitachi 7700) at 120 kV coupled with the Digital Micrograph camera and software suite from Gatan. The amount of whole protein in a nanoparticle was quantified by bicinchoninic (BCA) assay according to the manufacturer’s protocol, and the contents of anti-CD15 was quantified by measuring the fluorescence using the excitation/emission wavelengths of 496/578 nm. Drug loading contents were obtained by measuring the absorbance at 325 nm and subtracting the absorbance with the same amount of apolipoprotein A1 in the nanoparticle.

Cell Culture and Nanoparticle Treatment In Vitro.

DAOY human MB cells were obtained from the ATCC. PZp53 cells were provided by Matthew P. Scott, School of Medicine, Stanford University, Stanford, CA. DAOY and PZp53 cells are Shh activated and TP53 mutated. DAOY cells were seeded at 8 × 10³ cell/well in 96-well plates overnight and seven different concentrations of LDE225 (0.00001, 0.0001, 0.001, 0.01, 0.1, 1, and 10 µM) were added in 200 μL total culture medium to each well for 48 h. For testing nanoparticles on DAOY and PZp53 cells, four different groups of newly synthesized nanoparticles (eHNP-A1, eHNP-A1-CD15, eHNP-A1-LDE225, and eHNP-A1-CD15-LDE225) were added to the media at the same concentrations (corresponding to 0.00001, 0.0001, 0.001, 0.01, 0.1, 1, and 10 μM LDE225). Cell viability was analyzed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich).

Cell Viability Assay.

DAOY cells were seeded on 96-well plates at an initial density of 8 × 10³ cell/well and incubated overnight. Before the treatment of samples, cells were preincubated with 200 μM of block lipid transporter-1 (BLT-1) or 10 μM of free anti-CD15 for 30 min. Cells were then washed and replaced with total culture medium containing (or corresponding to) 0.00001, 0.0001, 0.001, 0.01, 0.1, 1, and 10 μM LDE225 and incubated for 48 h. Cell viability was measured using the aforementioned MTT assay.

Mouse Tumor Model.

All mouse model experiments were carried out in compliance with the Emory University Institutional Animal Care and Use Committee (IACUC) guidelines using Emory IACUC approved protocols. For the ex vivo and in vivo experiments, SmoA1 Transgenic (Jax 008831) and Math-Cre-ER-Ptch flox/flox were used. For the Math-Cre-ER-Ptch flox/flox mice, the pregnant female mice were administered tamoxifen (T-5648, Sigma) by oral gavage using 24-G gavaging needles (Fine Science Tools) after embryonic day 17 (E17.0) to induce MB formation. Tamoxifen was prepared as a 20 mg/mL solution in corn oil (Sigma) and tamoxifen doses were 4.0 mg/200 μL for treatment of pregnant females. After the pups were weaned and reached 5 wk old, the development of tumor was confirmed by MRI and mice were enrolled in the study.

Tumor Slice Cultures and Nanoparticle Treatment Ex Vivo.

SmoA1 mice were anesthetized using CO₂ inhalation, the GFP-positive tumor tissue was removed, and tissue was embedded in 4% agarose in 1× Hanks' Balanced Salt Solution (HBSS) supplemented with 6 g/L dextrose. Tumors were sectioned horizontally into 300-μm-thick slices with a vibratome (Leica Vibratome VT1200S). The slices were cultured on Transwell inserts (1-μm pore size; Falcon) in neurobasal media supplemented with B27, pyruvate, and glutamine. After 24 h in culture, eHNP-A1-CD15-LDE225 (40 ng/mL) was added to the media, and then the slices were cultured for 24 h more. Subsequently, the slices were fixed in 4% paraformaldehyde and immunostained with antibodies against cleaved caspase 3 (Cell Signaling 9661). Secondary antibodies (Alexa Fluor 405 Thermo Fisher Scientific A-11012) and DAPI (Thermo Fisher Scientific D1306) were applied and confocal images were obtained using an Olympus FV1000 laser confocal microscope and captured by Olympus Fluoview Software (Integrated Cellular Imaging Core, Emory University).

In Vivo Tumor Tissue Analysis.

The mouse studies were all approved by the animal care committee at Emory University. SmoA1-GFP mice (n = 4 for each group; saline, eHNP-A1, eHNP-A1-CD15, eHNP-A1-LDE225, and eHNP-A1-CD15-LDE225) were monitored for brain tumor symptoms based on head tilt, ataxia, and hunched posture. The needle catheters were prepared in-house by cutting 30-G needles and then inserting the cut end into 15 cm of polyethylene 10 tubing with a blunt 30-G Luer-lock hub at the end for attaching a syringe with either saline or nanoparticles dissolved in PBS or Hoechst dyes. The conscious mice were placed in a restrainer, and then the needle catheter was inserted into a lateral tail vein. Placement of the needle inside the vein was confirmed by infusing a small volume (100 to 150 μL). The mice were killed at 72 h postinjection and perfused with 4% paraformaldehyde. The perfused tissues from brain, spinal cord, liver, and tumors were harvested and immersed in 20% sucrose solution on 0.01 M PBS for 24 to 48 h. The tissues were then placed in optical cutting temperature (OCT) compound within a peel-away disposable embedding mold and transferred to a −20 °C freezer for 24 h. The frozen blocks were kept on metal grids that fit onto the cryostat (Thermo Scientific, Shandon AS620) and cryosectioned for 10-µm-thick slices. The slices were mounted on Permafrost microscope slides and immunostained before viewing under the confocal microscope. For dye perfusion analysis, Hoechst dye and Alexa-594 conjugated lectin were injected through the tail-vein in SmoA1-GFP tumor-bearing mice and the mice were killed after 72 h. The brain, liver, spinal cord, and tumor samples were collected after perfusion with 4% paraformaldehyde and cryosectioned before observing the sections under a confocal microscope. The slices were mounted and monitored as previously described.

In Vivo Tumor Growth and Survival Monitoring.

For long-term monitoring of tumor growth and survival test, two different SHH MB-bearing mouse models were used: 1) SmoA1-GFP mice treated with saline (n = 7) and eHNP-A1-CD15-LDE225 (n = 6) and 2) PTC mice treated with saline (n = 4) and eHNP-A1-CD15-LDE225 (n = 4). The MRI scans were obtained to determine the tumor volume with a 9.4 T Bio-spec scanner (Bruker). For MR imaging, mice were anesthetized with 1.5 to ∼2% isoflurane during the image acquisition. Respiration, body temperature, and electrocardiogram (ECG) were monitored during the imaging. T2-weighted images were acquired using a rapid acquisition with refocused echoes (RARE) sequence in sagittal direction (time to repetition [TR] = 4,000 ms, time to echo [TE] = 42 ms, RARE factor = 8, field of view [FOV] = 20 × 20 mm², matrix = 116 × 116, thickness = 0.75 mm, 15 slices, and average FOV obtained per image = 12). The tumor volume was qualified as the products of slice-to-slice separation and the sum of areas from the manually drawn tumor regions of interest (ROIs) on images.