Rationally Designed 2‑in‑1 Nanoparticles Can Overcome Adaptive Resistance in Cancer Aaron Goldman,†,‡,§,∥,▲ Ashish Kulkarni,†,‡,§,▲ Mohammad Kohandel,⊥ Prithvi Pandey,# Poornima Rao,§ Siva Kumar Natarajan,† Venkata Sabbisetti,† and Shiladitya Sengupta*,†,‡,§,¶ †
Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, United States Harvard−MIT Division of Health Sciences and Technology, Cambridge, Massachusetts 02139, United States § Division of Biomedical Engineering, Department of Medicine, Brigham and Women’s Hospital, Boston, Massachusetts 02115, United States ∥ Harvard Digestive Diseases Center, Boston, Massachusetts 02115, United States ⊥ Department of Applied Mathematics, University of Waterloo, Waterloo, ON N2L 3G1, Canada # India Innovation Research Center, Invictus Oncology, New Delhi 92, India ¶ Dana Farber Cancer Institute, Boston, Massachusetts 02115, United States ‡
S Supporting Information *
ABSTRACT: The development of resistance is the major cause of mortality in cancer. Combination chemotherapy is used clinically to reduce the probability of evolution of resistance. A similar trend toward the use of combinations of drugs is also emerging in the application of cancer nanomedicine. However, should a combination of two drugs be delivered from a single nanoparticle or should they be delivered in two different nanoparticles for maximal efficacy? We explored these questions in the context of adaptive resistance, which emerges as a phenotypic response of cancer cells to chemotherapy. We studied the phenotypic dynamics of breast cancer cells under cytotoxic chemotherapeutic stress and analyzed the data using a phenomenological mathematical model. We demonstrate that cancer cells can develop adaptive resistance by entering into a predetermined transitional trajectory that leads to phenocopies of inherently chemoresistant cancer cells. Disrupting this deterministic program requires a unique combination of inhibitors and cytotoxic agents. Using two such combinations, we demonstrate that a 2-in-1 nanomedicine can induce greater antitumor efficacy by ensuring that the origins of adaptive resistance are terminated by deterministic spatially constrained delivery of both drugs to the target cells. In contrast, a combination of free-form drugs or two nanoparticles, each carrying a single payload, is less effective, arising from a stochastic distribution to cells. These findings suggest that 2-in-1 nanomedicines could emerge as an important strategy for targeting adaptive resistance, resulting in increased antitumor efficacy. KEYWORDS: drug resistance, tumor heterogeneity, nanotechnology, drug delivery, targeted therapy
T
expressions and are conserved across drug refractory bacteria and human cancer cells.4−7 Developing novel strategies to overcome the development of such “adaptive” resistance is critical in the search for a cure for cancer. It is increasingly becoming evident that adaptive resistance needs to be targeted using combinations of drugs in spatiotemporally optimized regimens.8 For example, in a recent study, we demonstrated that breast cancer cells can mount an
he failure of chemotherapy to completely eliminate all cancer cells invariably results in relapse, which is one of the major causes of mortality. One of the primary reasons cancer therapies often fail is the emergence of drug resistance.1 The classical dogma around resistance development relies on Darwinian principles, where cells with the right mutation are selected under chemotherapy pressure, resulting in relapse.2 However, newer models to describe the development of resistance have recently emerged, which implicate nonmutational dynamics leading to favorable “drug-tolerant” states.3 Such reversible adaptations to drug-induced stress can arise from dynamic proteomic behavior or stochastic gene © 2016 American Chemical Society
Received: January 14, 2016 Accepted: June 3, 2016 Published: June 3, 2016 5823
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
www.acsnano.org
Article
ACS Nano
Figure 1. Taxanes and taxane nanoparticles induce an adaptive resistance phenotype in cancer cells. (A) Schema shows the steps in generating drug-tolerant cells. Breast cancer cells are exposed to high concentration of taxanes or taxane nanoparticles (X), and the cells that survive the initial onslaught are defined as drug (X)-tolerant cells. The DTCs primarily comprise a transient population of cells characterized by a CD44HiCD24Hi population (red) that arises from a CD44LoCD24Lo population as a result of phenotypic transition. (B) Fluorescence-activated cell sorting (FACS) plot reveals an increase in the CD44HiCD24Hi population (top right quadrant) compared to parent cells in the DTCs following treatment with taxanes (cabazitaxel, paclitaxel, and docetaxel) or taxane-based nanoparticles (STX-TC, Abraxane). The fraction of cancer stem-like cells (CD44HiCD24Lo) remains constant between parent and DTCs. (C) Cell size analysis of MDA-MB-231 and MDA-MB468 breast cancer cells and the respective DTCs generated by docetaxel was determined by flow cytometry based on the forward scatter (FSC:A) and side scatter (SSC:A) dimensions. Polygonal gate indicates the size of 20% largest parent cells. Note the systematized change in population density within this gate of the DTC from the respective cell line. (D) Cell viability analysis of MDA-MB-231 cells treated with docetaxel or doxorubicin at the indicated dose. Prior to drug treatment, cells were sorted into large and small fractions using threedimensional scatter parameters (FSC:A vs SSC:A); N > 5 biological replicates, **p < 0.01. (E) Representative histograms of CD44 expression from MDA-MB-231 and MDA-MB-468 parent cells and DTCs (top panels) and large and small parent cells (bottom panels) determined by FACS. Discrimination of large and small parent cells was performed using flow cytometry gating of the largest and smallest 20% of cells based on FSC:A and SSC:A. (F) DNA content by copy number (N) of MDA-MB-231 parent and DTC (left panel) or determined from cohorts of small and large cells of the parent population MDA-MB-231 (right panel). (G) Schematic shows experimental protocol used to derive the computational model in panel H. MDA-MB-231 cells were separated into cohorts and treated with docetaxel (DTX) for 0, 2, 4, or 8 h. The initial 0 h gate (p1) was established as the largest 20% of the parent population. This gate was used to assess changes in the population of DNA content, CD44 expression, and size over time in the absence of cell death or proliferation. A separate population of DTC was also assessed within this same p1 gate to compare population changes, which incorporate cell death parameters. (H) Quantitative analysis of the computational model based on empirical data. The model consists of four cellular compartments: large DNAHi/CD44Hi (X), small DNAHi/ CD44Hi (Y), small DNAHi/CD44Lo (Z), and small DNALo/CD44Lo (W). The model has seven transition rates between different compartments and four parameters describing the death rates for each (N → 0) of the compartments.
adaptive tolerance to taxanes via switching to a stem-like CD44HiCD24Hi copy. In this phenotypic state, it activates a Srclike family kinase (SFK) pathway, which confers a survival advantage. A sequential targeting of cancer cells with a taxane and dasatinib, where the latter inhibits SFK, exerted a greater antitumor effect.9 Similarly, in another study, the activation of phosphatidylinositol 3 kinase (PI3K) was implicated in conferring resistance to a platinum cytotoxic-based nanoparticle, and post-treatment but not pretreatment with a PI3K inhibitor was found to result in synergy.10 Indeed, as cancer nanomedicines become more commonly used in the clinic,11 similar mechanisms of adaptive resistance can limit their potential, and there will be an emerging need to use drug combinations for maximal antitumor efficacy.12−16 However, should such a combination of two drugs be delivered from a
single nanoparticle, or would the delivery of the same two drugs in two different nanoparticles achieve greater anticancer efficacy? We explored these themes in the current study in the context of adaptive resistance in cancer. We show that the physical systems governing adaptive resistance to chemotherapy in individual cells are deterministic, and that cell-state conversions under chemotherapy pressure leading to adaptive resistance are derived in a predetermined cell-state trajectory toward a “preferred phenotype” that copies cancer stem-like cells. We observed that this transient preferred phenotype was characterized by an activation of the PI3K and the SFK signaling pathways, which could confer adaptive resistance to taxanes. This presented us with two case studies to explore how optimal disruption of this deterministic program can be achieved using nanoparticles that target the PI3K and 5824
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
those of the subset of chemosensitive small cells (Figure 1E). However, fluorescence-activated cell sorting (FACS) analysis revealed that this enrichment in the DTCs for CD44 is not sizedependent (Figure S2A), which confirmed that the DTCs consisted of cell populations that had phenotypically transitioned to a chemotherapy-tolerant state following drug exposure. In addition to cell surface biomarkers, DNA content and polyploidization are known to associate with chemoresistant potential.24−28 Interestingly, the DTCs were found to exhibit a G2/M, polyploidy-heavy profile in contrast to parent cells, as determined using distribution analysis of propidium iodide (PI) incorporation (Figure 1F, left panel, and Figure S2B). Interestingly, a 4N and polyploidy state were also features of large parent cells versus the subset of small cells (Figure 1F, right panel). Mathematically Modeling the Phenotypic Transitions under Therapeutic Stress. As we observed that the DTCs phenocopy three features that are associated with inherently chemoresistant cancer cells, that is, (1) increase in cell size, (2) high DNA content, and (3) high CD44 expression, we next tested how these three features are interconnected in the acquisition of adaptive resistance. We developed a mathematical model to obtain cell-state transition probabilities of individual cells using empirical data. First, we designed a series of experiments to monitor the kinetic changes in cell size, CD44 expression, and DNA content by single-cell level analysis through FACS. As shown in Figure 1G, we began with a single ̈ cells and created a “gate” population of drug treatment naive around the largest 20% based on forward and side scatter dimensions from flow cytometry (the inherently chemoresistant population), which we define as gate “p1”. We then monitored the dynamic fluctuation within p1 following acute treatment with docetaxel over 2, 4, and 8 h or in a final population of DTCs. Consistent with our analyses above, cells in p1 exhibited concordantly increased CD44 expression and DNA content over baseline (Figure 1G). We then analyzed the FACS results to estimate the kinetics of CD44 expression and DNA content for the remaining 80% population (for simplicity, we call them “small” cells). Based on these experiments, we developed a model consisting of four cellular compartments: large DNAHi/CD44Hi (X), small DNAHi/CD44Hi (Y), small DNAHi/CD44Lo (Z), and small DNALo/CD44Lo (W) (Figure 1H). A main experimental observation was that the cells do not proliferate or die during the course of treatment (0−8 h), and therefore, any phenotypic alterations are a direct result of plasticity rather than clonal selection. Beyond 8 h, however, we observed loss of cell viability in some populations. Thus, our model consisted of four parameters for the death rates (for 8− 72 h) and seven parameters to describe possible transition rates between the compartments. The model was able to fit well to the experimental data (Figure S2C,D). In addition, the model captured two distinct features of the biological system. First, the death rate for large DNAHi/CD44Hi cells was an order of magnitude smaller than the death rates for other populations, validating our earlier experimental data that these cells are highly resistant to chemotherapy. Second, the model predicted that there was a distinct trajectory from small DNALo/CD44Lo cells toward the large DNAHi/CD44Hi cells; that is, the transition rates were significantly higher in this trajectory versus the other possible trajectories. PI3K/AKT-Driven Mechanism of DNA Synthesis in Adaptive Resistance. Having established a correlation between CD44 expression and DNA synthesis during adaptive
SFK pathways. We engineered nanoparticles with 2-in-1 drugs and compared their efficacy with two individual nanoparticles. We determined that 2-in-1 nanomedicines are critical to ensure co-delivery of the targeted agents to disrupt the origins of adaptive resistance. Indeed, the use of 2-in-1 nanomedicines provided antitumor efficacy superior to that of individual drugloaded nanoparticles. These findings give insights in overcoming chemotherapy-induced adaptive resistance and provide a rational strategy to increase antitumor efficacy of combination nanomedicines.
RESULTS AND DISCUSSION Adaptive resistance to chemotherapy is an emerging concept in cancer biology. In the following sections, we first demonstrate the biological underpinning of the emergence of adaptive resistance to taxane chemotherapy. We show that taxane-based nanoparticles are also limited by the same mechanisms of adaptive resistance. We then discuss two case studies, where we explore the potential for using 2-in-1 nanoparticles for overcoming this adaptive resistance and elucidate the mechanism why a 2-in-1 nanoparticle exerts an effect greater than that of a combination of two single payload nanoparticles. Characterizing a Model of Adaptive Resistance to Chemotherapy. We used a well-established model to study adaptive resistance.9 Breast cancer cells were treated with a high dose (20 × IC50 value)17 of taxane, a commonly used chemotherapy agent. The population of cells that were found to survive were selected using substrate reattachment and defined hereafter as drug-tolerant cells (DTC) (Figure 1A). Consistent with our previous study with docetaxel, the surviving cell population in response to all the taxanes (paclitaxel, cabazitaxel, and docetaxel) was found to comprise inherently resistant classical cancer stem cells (CSC) defined by a CD44HiCD24Lo status as well as a transient CD44HiCD4Hi population (Figure 1B).9 Interestingly, two nanoparticle-based taxanes, Suprataxel (STX), a supramolecular taxane that we have engineered,18 and Abraxane, an albumin-based paclitaxel nanoparticle,19 induced a similar phenotypic transition (Figure 1B). Indeed, the biomarker glycoprotein CD44 is known to associate with drug-resistant breast cancer cells.9,20,21 Having established that taxane nanoparticles can induce the phenotypic transition to a CD44HiCD24Hi cell state that is associated with adaptive resistance, we further characterized the DTCs. We observed that a striking feature of the DTCs is an increase in cell size, determined by quantifying forward and side scatter dimensions using flow cytometry (Figure 1C). The increased dimension made these cells comparable to the intrinsically large cells that formed ∼20% of the parent population (Supporting Information Figure S1A). A control experiment, where the cancer cells were exposed to a hypotonic solution, revealed that cell swelling was not responsible for this shift in the size (Figure S1B,C). To resolve if large cells from ̈ populations are inherently chemoresistant, we drug naive isolated and enriched small and large subpopulations using three-dimensional light scattering parameters in flow cytometry cell sorting (Figure S1D), which we validated by measuring substrate-attached cell size (Figure S1D). These cells were subsequently exposed to either docetaxel or doxorubicin, an anthracycline chemotherapy used to treat breast cancer.22,23 We observed that, similar to the CD44Hi cells, the large cells did resist the cytotoxic effects of chemotherapy compared to their smaller counterparts (Figure 1D). Interestingly, both the DTCs and the large parent cells exhibited CD44 levels higher than 5825
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
Figure 2. Co-delivery of taxanes and AKT inhibitors overcomes acquired therapy resistance. (A,B) Representative Western blot analysis of AKT family proteins and associated cortex kinases (Ezrin-Radixin-Moesin) in MDA-MB-231 parent cells and DTC (A) and cells that were either unsorted or sorted based on the high expression of CD44 (B). (C) Representative Western blot of active (phosphorylated) AKT from MDA-MB-231 cells selected by escalating dose of chemotherapy (refer to schematic Figure 1A). 10DT = 10 nM docetaxel, 50DT = 50 nM docetaxel, and 100DT = 100 nM docetaxel. (D) DNA content analysis by flow cytometry (PI staining). The use of small-molecule inhibitors directed against AKT (PI103), P21-driven DNA synthesis (Sterig), or small interfering RNA (siRNA) targeting AKT was employed prior to drug exposure (DTX). N > 8, **p < 0.01. (E) Representative flow cytometry shows cell size (FSC:A vs SSC:A) in MDA-MB-231 DTC treated for 48 h with AKT inhibitor PI103 or vehicle control. (F,G) Representative Western blot shows AKT phosphorylation in 4T1 breast cancer cells treated with STX-NP for indicated intervals (F) or for 24 h in the presence of the AKT inhibitor PI103 (G). (H−J) Combination therapy index (CI) plots (as per the method of Chou and Talalay) showing synergistic effect of STX nanoparticles in combination with PI103 in breast 4T1 and ovarian cancer 4306 cell lines. CI < 1 indicates synergy; CI = 1 indicates additivity; CI > 1 indicates antagonism. (K) Schematic illustrates the necessity for taxanes and AKT inhibitors to reach the same cell simultaneously. The spatiotemporal control of drugs ensures a co-delivery of cytotoxic agent and the mechanism of resistance that arises within the same cell.
increase in cell size (Figure 2E), suggesting that PI3K inhibitors could be used to overcome chemotherapy-induced adaptive resistance. We next validated whether the above AKT signaling is conserved in the case of treatment with a taxane nanoparticle. Indeed, as shown in Figure 2F, we observed a time-dependent increase in the phosphorylated AKT levels following treatment with the STX nanoparticles. Additionally, treatment with PI103 inhibited this increase. A Chou−Talalay analysis34 of the dose− response curves (Figure S3A), where 4T1 breast cancer cells or 4306 ovarian cancer cells were treated with a fixed concentration of STX-NP and increasing concentrations of PI103, revealed a synergistic effect with the combination, as evident by calculating the combination index (Figure 2H−J), suggesting that the inhibition of PI3K indeed overcomes adaptive resistance to the taxane nanoparticle. Case Study 1: A 1 + 1 Nanoparticle Combination vs a 2-in-1 Nanoparticle for Targeting Taxane-Induced AKTMediated Adaptive Resistance. Establishing a deeper understanding of the mechanisms underlying adaptive resistance to taxane nanoparticle-based chemotherapy provided a unique platform to test our original question of whether the chemotherapy and inhibitors of downstream signaling pathways driving adaptive resistance should be administered from a single nanoparticle or would a combination of two single payload nanoparticles be equally effective. We reasoned that a combination of the taxane chemotherapeutic (driver of adaptive resistance) and the AKT inhibitor (blocker of adaptive resistance) must be targeted to the same cell to achieve an optimal cancer cell kill effect, that is, overcome the ability of
resistance to taxane, we examined the cortex kinases that are downstream and upstream of CD44 expression and DNA synthesis, respectively, as therapeutically tractable targets. The PI3K/AKT family kinases have been previously implicated in the cell cycle,29 and its overactivity is known to drive survival and to be predictive of therapy failure.30 We first validated whether this pathway is implicated in the adaptive resistance in DTCs. Indeed, Western blot revealed that the DTCs had higher levels of activated (phosphorylated) PI3K/AKT family proteins and Ezrin/Radixin/Moesin, which have been reported to form cortex complexes with CD44 and AKT31 (Figure 2A). ̈ The inherently chemoresistant “large” cells from drug naive parent populations that exhibited a CD44Hi phenotype also demonstrated a similarly phosphorylated AKT and activated mTOR downstream of AKT (Figure 2B). However, it should be noted that a dose escalation study with docetaxel revealed that the active AKT signaling cortex in the DTCs was induced by the drug exposure (Figure 2C). To determine whether this activation of AKT mediates the drug-induced DNA synthesis and associated morphological alterations, we employed smallmolecule pharmaceutical inhibitors, such as sterigmatocystin that targets p21CIP/WAF1-driven DNA synthesis32,33 and PI103 that targets upstream PI3K, or used siRNA for AKT expression knockdown. Each of these approaches effectively inhibited the therapy-induced DNA accumulation, as determined by PI accumulation over the course of acute (2−8 h) treatment (Figure 2D). Finally, we tested the hypothesis that AKT is also coupled to DTC cell size. Based on forward and side scatter dimensions from flow cytometry, we observed that treating the DTC subset with PI103 attenuated the chemotherapy-induced 5826
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
Figure 3. Computer simulations of STX and SPI103 interactions with the lipid bilayer and supramolecular assembly of chimeric nanoparticles. (A) Snapshots of molecular dynamics simulations of 10 mol % of SPI103 + 10 mol % of STX in the SOPC bilayer after 10 ns of initial equilibrium and after 300 ns. (B) Distribution of the tilt angle measured at 500 ns in STX, SPI103, and STX + SPI103 bilayer. (C) Lipid tail order parameters for carbon atoms of the saturated lipid tail; 1 denotes the second carbon atom after the headgroup. (D) Probability of finding the water molecules in the hydrophobic region of the bilayer at different concentration of PTX and STX. (E) Representative images showing physical stability of free [PTX + PI103] nanoparticles and chimeric SNPs at room temperature, 24 h after the synthesis of nanoparticles. (F) High-resolution cryo-transmission electron microscopy image of chimeric NPs (scale bar = 100 nm). (G) Graph shows the size distribution of chimeric NPs as measured by dynamic light scattering. (H) Physical stability of chimeric NPs in phosphate-buffered saline (PBS) at 4 °C as measured using dynamic light scattering. (I) In vitro release of STX and SPI103 from nanoparticles in PBS, pH 7.4, or 4T1 cell lysate. Data represents the mean ± SE (n = 3).
SPI103 or STX alone, as indicated by significantly reduced ripple formation. Furthermore, to quantify the ordering of lipid tails, we calculated the deuterium order parameter (−SCD) for the SOPC lipid saturated tail from the last 5 ns of the trajectory,36 which revealed that the combination of 10 mol % of SPI103 and 10 mol % of STX resulted in a lipid tail ordering that overlapped with that of the pure SOPC bilayer unlike that of 20 mol % of SPI103 or STX alone (Figure 3C). The formation of ripples and reduced lipid tail order can result in perforations in the bilayer, forming a water/ion channel through the bilayer.37 Indeed, the predictions revealed that 20 mol % of both SPI103- and the dual-drug-containing systems exhibited a lower probability of water entering the bilayer compared with the 20 mol % of the STX-only system (Figure 3D). Taken together, these simulations suggested that a STX + SPI103 (10 mol % each)-containing stable nanoscale system was indeed achievable. This was also optimal as the drugs exhibited IC50 values that were comparable. We engineered the STX + SPI103 dual payload nanoparticle, which we term a “chimeric” nanoparticle (i.e., self-assembly of two distinct active molecules), using a thin film hydration followed by an extrusion approach. Interestingly, a plain liposome with 10 mol % each of paclitaxel and PI103 collapsed, resulting in the drugs precipitating out, while the chimeric nanoparticle was stable (Figure 3E). Cryo-transmission electron microscopy (TEM) and dynamic light scattering confirmed the formation of a supramolecular nanostructure of 121 ± 36 diameter (Figure 3F,G), where the ζ-potential and size were consistent over a long period at 4 °C (Figure 3H). The loading efficiency of each drug in the chimeric nanoparticle was ∼85 ± 4%. Release kinetic studies revealed that the release of the active drug moieties are accelerated in the presence of cell lysate, reaching saturation at 50 h (Figure 3I).
cancer cells to mount an adaptive resistance and survive (see schematic in Figure 2K). Therefore, we designed a 2-in-1 nanoparticle that contained both drugs. In a recent study, we demonstrated that an analogue of PI103, which we term SPI103, could facilitate supramolecular assembly into a stable nanoparticle that resulted in a sustained inhibition of the PI3K signaling pathway in vitro and in vivo.35 Additionally, encapsulation of taxanes in lipid-based nanoparticles has been a challenge, which has limited clinical translation of such structures. We therefore engineered a taxane analogue by conjugating cholesterol to the C-2′ position of paclitaxel (PTX) via a flexible and acid-labile linker. In a separate study, we observed that such a taxane analogue, which we termed Suprataxel (STX), self-assembled into stable supramolecular lipidic nanoparticles with high drug loading and exerted the desired antitumor efficacy in vitro and in vivo.18 We rationalized that STX and SPI103 could be used to engineer the 2-in-1 nanoparticles. To test whether a mixture of SPI103 and STX could indeed form a stable supramolecular assembly into a 2-in-1 nanoparticle, we used an all atomistic molecular dynamics simulation using an SOPC co-lipid as an excipient to form a bilayer (Figure 3A). The stability of the system was measured as a function of ripple formation, which was calculated from the tilt angle between the vector joining the center of mass of lipid tails and phosphorus atom on the headgroup of the SOPC molecule with the z-axis (bilayer normal). In a separate study, we demonstrated that the theoretical predictions of the ability of STX or SPI103 to form stable supramolecular nanoscale bilayers at high 20 mol % concentrations are indeed translated experimentally.10 As shown in Figure 3B, the combination of 10 mol % of SPI103 and 10 mol % of STX was predicted to costabilize the bilayer to a greater degree than 20 mol % of 5827
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
Figure 4. Delivery of synergistic combination of drugs using a single nanoparticle (chimeric NPs) shows enhanced efficacy compared to individual nanoparticles. Graphs show cell viability of (A) 4T1 and (B) MDA-MB-231 breast cancer cell lines after 72 h incubation with increasing concentrations of STX-NPs + SPI103-NPs and chimeric NPs. (C) In vivo efficacy of chimeric NPs in 4T1 breast cancer model (n = 4). Mice were treated with three doses of PBS (control), STX-NP (2 mg/kg) + SPI103-NP (1.4 mg/kg), and chimeric NPs (2 mg/kg of STX and 1.4 mg/kg of SPI103). Error bars denote mean ± SEM; **p < 0.01. Top panel shows representative images of excised tumors. (D) Representative epifluorescence images of tumor sections from different treated groups were labeled for apoptosis using TUNEL (red) and counterstained with DAPI (blue). (E) Graph shows the quantification of apoptosis from the labeled tumor sections as a percentage of TUNEL +ve cells as a function of total nuclei. Statistical analysis was performed with ANOVA. Error bars denote mean ± SEM; *p < 0.05, **p < 0.01. (F) IHC of tumor tissues indicates expression of p-AKT in STX-NP + SPI103-NP and chimeric NP groups. (G) Graph shows significant inhibition of p-AKT expression in the chimeric NP group compared to STX-NP + SPI103-NP group. (H) In vitro internalization assay using fluorescent nanoparticles shows that at lower concentration individual nanoparticles have different spatial distribution in cells, whereas chimeric nanoparticles are internalizing both green and red dye at similar locations. (I) Representative flow cytometry plots of 4T1 cells show internalization of individual fluorescent nanoparticles vs chimeric nanoparticles. (J) Representative images of tumor sections from green fluorescent NP + red fluorescent NP group and chimeric (green + red) fluorescent NP group.
We next tested the chimeric nanoparticles in vitro and in vivo. As shown in Figure 4A,B, cell viability studies revealed that the chimeric nanoparticles were more effective than the combination of STX-NP and SPI103-NP. The chimeric nanoparticles exhibited an IC50 value of 22.3 ± 5.91 and 30.16 ± 6.12 nM in 4T1 and MDA-MB-231 breast cancer cells, respectively, compared to 77 ± 5.9 and 98.5 ± 3.69 nM achieved with the combination of STX-NP and SPI103-NP. We next compared a combination of a STX-NP and SPI103-NP versus a chimeric nanoparticle in vivo. 4T1 tumor-bearing mice were treated with four cycles of either a combination of STX-NP and SPI103-NP or a chimeric nanoparticle at submaximal doses to dissect any distinctions between the two treatments. As shown in Figure 4C, treatment with the chimeric nanoparticle resulted in a greater tumor growth inhibition compared to that with the combination. This was validated by TUNEL staining, which revealed a significantly higher degree of apoptosis in the chimeric nanoparticle-treated tumors compared to that with the combination treatment (Figure 4D,E). As our in vitro results indicated that the STX nanoparticles induce PI3K signaling as an adaptive mechanism in cancer cells, we immunohistologically (IHC) stained the tumor sections for phospho-AKT. As shown in Figure 4F,G, tumors treated with the combination of STX-NP and SPI103-NP presented with a higher degree of phospho-AKT signaling, which was associated with viable regions of the tumor. In contrast, the phospho-AKT levels were significantly lower in the chimeric nanoparticle-treated tumors, which were characterized by a larger fraction of nonviable areas. Together with the in vitro observations, the increased cell kill indicated that the chimeric nanoparticle exerts a greater anticancer effect by overcoming the adaptive resistance
response that the cancer cells mount in response to cytotoxic chemotherapy. Stochastic versus Deterministic Delivery. To further test the hypothesis that the chimeric nanoparticle indeed ensures a deterministic delivery of both the cytotoxic and the targeted therapeutic to a cell versus a stochastic distribution in the case of a combination of two nanoparticles, we engineered nanoparticles where the building blocks were tagged with either a green (fluorescein) or red (HiLyte Fluor) fluorescent tracer. Cancer cells were incubated with either a combination of red and green nanoparticles or a chimeric nanoparticle that contained both tracers. Interestingly, confocal microscopy revealed that while a large fraction of cells internalized both nanoparticles following combination treatment, there were many cells that took up either red or green nanoparticles alone (Figure 4H). This was further validated by FACS, which revealed that while 48% of the cancer cells internalized both nanoparticles, 52% of the cells had taken up either a red or a green nanoparticle and not both. In contrast, 98% of the cells exhibited the presence of both tracers when incubated with the chimeric nanoparticle (Figure 4I). We next studied whether a similar result is observed in vivo. Administering the tracertagged nanoparticles in tumor-bearing animals revealed that the combination treatment indeed results in a stochastic distribution within the tumor, with regions that predominantly received either the green- or the red-tracer-tagged nanoparticles and areas that received both nanoparticles, consistent with the in vitro observations. A uniform cellular distribution of both tracers was observed in the chimeric nanoparticle-treated group (Figure 4J). These results support the hypothesis that the increased efficacy of the 2-in-1 “chimeric” nanoparticles could arise from the controlled delivery of both therapeutic agents 5828
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
Figure 5. Design of multifunctional nanoparticles overcomes drug distribution limitations. (A) Signaling schematic shows the distinct pathways perturbed by either dasatinib (CD44-driven SFK signaling) or PI103 (DNA synthesis) that are active in DTC to promote survival. (B,C) Cell viability analyses of MDA-MB-231 or MDA-MB-468 cells treated with combinations of SFK inhibitors (dasatinib), AKT inhibitor (PI103), or DNA synthesis inhibitor, sterigmatocystin (sterig) (C), N > 4. (D) Illustration shows limitations of free-form drug administration characterized by uneven cell distribution of therapy. Development of a single-drug-loaded vehicle would ensure delivery of both kinase inhibitors simultaneously. (E) Release kinetics of each drug in different pH solution. (F) ζ-Potential and dynamic light scattering graphs demonstrate stability of DPNP. (G) Cryo-TEM representative image of chimeric NPs. (H) Representative Western blot shows functional efficacy of chimeric NPs to silence signaling pathways in the AKT family of proteins. (I) Cell viability curves of MDA-MB-231 DTC treated with single-drug-loaded nanoparticles (dasatinib or PI103), two-nanoparticle treatment, or the 2-in-1 chimeric NPs. Note that free delivery of two nanoparticles is less effective than the chimeric nanoparticle, indicated by shifts in the viability curve. (J) Schematic illustrates the intratumoral injection model used throughout the study to evaluate drug tolerance from in vivo tumor tissue of a sygeneic murine carcinoma model (4T1). By variable diffusion of drug through three dimensions, the model enables dissection of drug-induced effects in viable tissue. (K) Representative confocal microscopy of drug-tolerant features in tumor tissue following treatment with DTX or dastainib-PI103 chimeric NP (DPNP). Scale bar = 22 μm. (L) Quantification scatter plots indicate mean fluorescent intensity from N > 25 individual fields; *p < 0.05, ***p < 0.001. (M) Histogram shows final tumor volumes of mice treated with indicated sequential and combination regimens (day 13).
dasatinib (a SFK inhibitor) to test whether the combination is indeed synergistic against the DTCs. While AKT inhibition alone was ineffective at low concentrations, we determined that the combination of the two inhibitors in a 1:1 ratio worked in synergy to ablate the persisting DTC, as evidenced by shifts in cell viability curves compared to either drug alone (Figure 5B). To validate that the inhibition of AKT-driven DNA synthesis was the mechanism through which synergy was achieved, we combined dasatinib with sterigmatocystin and observed a similar synergistic outcome (Figure 5C). That dual inhibition of the discrete signaling pathways led to synergism suggested that AKT and SFK signaling are drivers of adaptive resistance. We therefore next tested the hypothesis that targeting these distinct pathways with a 2-in-1 nanoparticle that ensures the delivery of both SFK and AKT inhibitors to the same cell (based on our previous observation) could exert greater efficacy than a combination of the two drugs delivered separately (schema in Figure 5D). We used SPI103 as the PI3K
uniformly to all cells; that is the targeted therapeutic is present in the same cell to block the chemotherapy-induced adaptive resistance. In contrast, in the case of combination therapy using two separate nanoparticles, a subset of cells that receive only the chemotherapeutic nanoparticle, but not the targeted therapeutic, can mount an adaptive resistance response, resulting in tumor growth and relapse. Case Study 2: Targeting Downstream Signaling of CD44 and AKT Pathway To Overcome Adaptive Resistance. We used a second case study to validate our observation. Based on our mathematical model, the induction of DNA synthesis and CD44 expression were requisite events for a cell to mount an adaptive resistance. In a recent study, we had observed that the SFK signaling pathway lies downstream of CD44.9 We therefore tested whether the simultaneous inhibition of CD44-SFK scaffolding and AKT-driven DNA synthesis could mount a synergistic outcome (Figure 5A). As a proof of principle, we first utilized low doses of PI103 and 5829
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
resistance is crucial in the design of combinatorial nanomedicines that can overcome this resistant state. Additionally, we demonstrate that a synergistic outcome is predicated by the deterministic delivery of both therapeutic agents to the same cell in the case of the combinatorial “chimeric” nanoparticle rather than a stochastic distribution of the two agents to different cells when delivered separately. Combination chemotherapy is routinely used in the clinics for maximizing the antitumor outcomes, where the drugs are classically combined at the maximum dose intensity that can be tolerated by the patient. Indeed, an increasing number of studies are now focusing on the use of combinations of nanomedicines for cancer, although the emphasis has been on creating synergy between the two agents via ratiometric dosing, which can be enabled through optimal nanoparticle design.12,16 For example, the combination of cisplatin and gemcitabine or doxorubicin and camptothecin in polymeric nanoparticles or cytarabine and daunorubicin in liposomes was used for treating different types of cancer.39−41 However, the loading of more than one active agent in a nanoparticle at levels that can be clinically translated is a challenge. We addressed this challenge by using an all atomistic molecular dynamics simulation, which allowed us to model the interactions between the two active agents and the lipid bilayer, predicting the stability of the supramolecular nanostructure using three distinct parameters: ripple formation, lipid tail ordering, and the probability of finding water molecules in the bilayer. In a separate study, we observed that the all atomistic simulation can predict the impact of minor changes in the chemical conformation of the payload on stability as well as the contribution of excipients such as PEGylated co-lipids. Indeed, the observed stability of the chimeric nanoparticle, and the drug loading efficiency achieved, suggests that an all atomistic simulation-augmented engineering of nanoparticles can emerge as an attractive approach in overcoming some of the above challenges. Interestingly, previous studies with combination nanotherapeutics have focused on synergy resulting from targeting either distinct tumor compartments such as the stroma (neovascularization) along with the parenchyma15 or different intracellular targets such as the DNA (with doxorubicin) and topoisomerase 1 (using camptothecin).40 Similarly, in a recent study, the co-delivery of paclitaxel and gemcitabine was found to be synergistic, where paclitaxel induced the activation of hemeoxygenase 1 that suppressed cytidine deaminase, which can metabolically inactivate gemcitabine.42 This study exploits combination nanotherapy to overcome chemotherapy resistance. Adaptive resistance to chemotherapy is an emerging paradigm in cancer biology that is increasingly being unraveled. For example, Gupta and colleagues recently described that the equilibrium established in heterogeneous populations is governed by probabilistic gene expression, which can influence the responses to therapy.43 Kreso et al. recently described the repopulation dynamics of clones from colorectal cancer patients, which demonstrated that phenotypic variability exists within genetically identical subsets.3 Such studies emphasize how aggregate populations influence tumor progression and the outcome of therapies through phenotypic behavior. We have described an additional paradigm in which dynamics of individual cells leading to adaptive resistance can be explained by a more definitive physical system. This model incorporates phenotypic plasticity, which influences the dynamic behavior of individual cells surviving treatment. Indeed, in the current
inhibitor, which formed stable 2-in-1 supramolecular nanoparticles when co-packed with dasatinib, as seen with cryoTEM (Figure 5E). The 2-in-1 “chimeric” nanoparticle had a hydrodynamic diameter of 201 ± 33.7 and a ζ-potential of 18 ± 1.7 mV, which was similar to the single payload nanoparticles. The size and ζ-potential remained consistent over a 10 day period (Figure 5F), and release kinetics studies revealed a pHand time-dependent release, with both agents being released at similar rates (Figure 5G). The faster release observed with cell lysates in the case of the chimeric nanoparticle in case study 1 and at a lower pH observed here is reflective of the cellular state, where cancer cells, and especially the endolysosomal compartment, have been reported to exhibit an acidic environment. Western blot analysis of cells treated with the chimeric nanoparticles revealed that the chimeric nanoparticles inhibited both AKT and SFK pathways compared with dasatinib alone (Figure 5H). Next, we performed in vitro cell viability studies that revealed the chimeric nanoparticles resulted in a potent synergistic outcome, as evidenced by a shift in the concentration−response curve compared with either of the drugs alone or a combination of two nanoparticles carrying the individual drugs (Figure 5I and Supporting Information Figure S3D). To test the functional activity of the AKT- and SFKinhibiting chimeric nanoparticles in vivo, we first designed a syngeneic murine carcinoma experiment in which docetaxel chemotherapy was injected intratumorally (i.t.). This model enabled chemotherapeutic exposure with drug diffusion through the 3-D tumor, inducing an adaptive response in the remaining viable tissue (Figure 5J). This was confirmed using hematoxylin and eosin staining of the tumor sections following i.t. injection of docetaxel that revealed an enhanced cell size and nuclear DNA content of chemotherapy-surviving populations ̈ cells seen in the compared with the drug treatment naive vehicle control group (Figure S4A). Additionally, we confirmed the expression of adaptive resistance markers CD44 and CD24 and co-expression of active AKT in these same surviving regions of the tumor (Figure S4B,C). Using this in vivo model, we administered the 2-in-1 nanoparticle 24 h prior to the i.t. injection of docetaxel. The administration of the 2-in-1 chimeric nanoparticle successfully suppressed the augmented activation of AKT and HCK (downstream of SFK) signals as confirmed using immunofluorescence imaging and FACS analysis (Figure 5K,L). Additionally, we noted that depleted signals of CD24 and AKT phosphorylation occurred in the same regions of DTX-treated tumor tissue (Figure 5K), which indicated that the 2-in-1 chimeric nanoparticle suppressed the acquisition of a resistant phenotype in a uniform manner. Consistent with these mechanistic observations, the treatment with the chimeric nanoparticles significantly enhanced the antitumor outcome (Figure 5M).
CONCLUSION Adaptive resistance, that is, the acquisition of a phenotypic state that allows the cells to tolerate the initial onslaught of chemotherapy, has emerged as an actively studied phenomenon underlying chemotherapy failure. It is distinct from the selection of a genetically resistant clone in that the phenotype is transient; that is, the cancer cells become resensitized to the chemotherapy after a “drug holiday”.38 Here, we report that adaptive resistance could also limit the efficacy of a taxanebased nanomedicine. We demonstrate that an understanding of the signaling pathways underlying the development of adaptive 5830
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano
Synthesis of Nanoparticles. For Case Study 1, L-α-phosphatidylcholine (6.4 mg), STX (2.3 mg), SPI103 (1.34 mg) and DSPE-PEG (13.7mg); for case study 2, L-α-Phosphatidylcholine (22.7 mg), SPI103 (2 mg), dasatinib (1.19 mg), and DSPE-PEG (40.98 mg) were dissolved in 1.0 mL of DCM and 0.5 mL of methanol. Solvent was evaporated to form a thin and uniform lipid−drug film using a rotary evaporator. The lipid−drug film was then hydrated with 2.0 mL of H2O for 1 h at 60 °C. The hydrated nanoparticles were light yellow to white with a slight viscous texture. It was passed through a Sephadex G-25 column and extruded at 65 °C to obtain sub-200 nm particles. To measure the loading of drug in the nanoparticle, a standard curve of SPI103 and dasatinib in dimethylformamide (DMF) was generated by measuring absorbance at 270 and 339 nm, respectively, using UV− vis spectrophotometry (Shimadzu 2450). A known concentration of nanoparticle was dissolved in DMF, and the absorbance value at 270 and 339 nm was used to calculate the loading from a standard curve. The mean particle size of the nanoparticles was measured by the dynamic light scattering method using a Zetasizer Nano ZS90 (Malvern, UK). Ten microliters of a nanoparticle solution was diluted to 1 mL using DI water, and three sets of 10 measurements each were performed at a 90° scattering angle to get the average particle size. The ζ-potential was measured using a Zetasizer ZS90 with the nanoparticles diluted in water for measurement according to the manufacturer’s manual. Release Kinetics. Drug-loaded nanoparticles (0.5 mg drug/mL, 5 mL) were suspended in pH 5.5 and pH 7.4 buffer and sealed in a dialysis tube (MWCO = 3500 Da, Spectrum Lab). The dialysis tube was suspended in 1 L of buffer with the same pH as in the dialysis tube with gentle stirring to simulate the infinite sink tank condition. A 100 μL portion of the aliquot was collected from the incubation medium at predetermined time intervals and replaced with an equal volume of PBS, and the released drug was quantified by UV−vis spectrophotometry and plotted as cumulative drug release. Cryo-Transmission Electron Microscopy. The sample was preserved in vitrified ice supported by holey carbon films on 400 mesh copper grids. The sample was prepared by applying 3 μL of sample suspension to a cleaned grid, blotting away with filter paper, and immediately proceeding with vitrification in liquid ethane. Grids were stored under liquid nitrogen until transferred to the electron microscope for imaging. Electron microscopy was performed using an FEI Tecnai cryo-bio 200 kV FEG TEM, operating at 120 keV equipped with 2 Gatan Sirius CCD cameras one 2K*2K and one 4K*4K pixel. Vitreous ice grids were transferred into the electron microscope using a cryostage that maintains the grids at a temperature below −170 °C. Images of the grid were acquired at multiple scales to assess the overall distribution of the specimen. After potentially suitable target areas for imaging at lower magnification were identified, high-magnification images were acquired at nominal magnification of 52 000× (0.21 nm/pixel) and 21 000× (0.50 nm/pixel). Images were acquired at a nominal underfocus of −5 μm (21 000×) and −4 μm (52 000×) at electron doses of ∼10−15 e/A°2. Cell Culture and Gene Knockdown with siRNA. MDA-MB-231 (ATCC) cells were cultured in DMEM containing 10% fetal bovine serum (FBS), and MDA-MB-468 (ATCC) and 4T1 mammary carcinoma cells (ATCC) were cultured in RPMI containing 10% FBS (Invitrogen, Carlsbad, CA) at 37C and 5% CO2. During treatments with chemotherapeutics, cells were grown to semiconfluence and treated with indicated concentrations of chemotherapy in serum-containing medium for indicated time points. For siRNA gene knockdown, cells were plated at a concentration of 5 × 104 cells/ mL. Prevalidated silencer select siRNA targeting (sense sequences) AKT (s659) or scrambled control was purchased from Ambion (Invitrogen, Grand Island, NY) and was transfected using lipofectamine 2000 (Invitrogen, Carlsbad CA) following the manufacturer’s protocol. Generating Drug-Tolerant Cancer Cells. Cancer cells were plated at a density of 0.5−1 × 105 cells/mL and allowed to adhere for 24−48 h. When cells reached ∼70% confluency, they were treated with cytotoxic drugs at indicated concentrations for 4−48 h and utilized for subsequent assays. Following washes with PBS, adherent
study, we observed an increase in CD44 and CD24 expression in response to chemotherapy, including treatment with cancer nanomedicines, consistent with our previous observation where chemotherapy-induced CD44 and CD24 expression correlated with a colocalization and activation of the SFK signaling pathway.9 The intrinsic stem-like cell population, defined by CD44HiCD24Lo status, was similar between parent and the DTCs, indicating that the putative cancer stem cells were also chemotolerant to the taxanes and taxane nanoparticles. Interestingly, the mathematical model demonstrated that a definitive trajectory of (1) DNA synthesis, (2) CD44 expression, and (3) cell size must be acquired within an individual cell from the distinct subpopulations of CD44Lo, low DNA content, and small cells to phenocopy inherently chemoresistant cells. Thus, targeting the phenotypic transition, for example, inhibiting the PI3K-AKT pathway that is implicated in DNA synthesis together with the SFK pathway that modulates the signals downstream of CD44HiCD24Hi status, could indeed result in a synergistic outcome in overcoming adaptive resistance to taxanes. It should be noted that the intratumoral injection of the taxane, in case study 2, creates a maximal tumor inhibition, which could mask any synergy when combined with either a dasatinib or the PI3K inhibitor. The enhanced synergy with the chimeric nanoparticle in such a setting highlights the advantage of the combinatorial 2-in-1 nanoparticle. However, the combination does not fully ablate the tumor, which highlights the possibility of additional pathways that can also drive the trajectories between different phenotypic cell states. For example, we observed that the ERM scaffolding proteins, implicated in the AKT signaling, are also phosphorylated in the DTCs. Future studies are necessary to dissect the global mechanisms underlying adaptive resistance, which can create additional targets for developing combinatorial nanomedicines. In conclusion, our results indicate that combinatorial nanomedicines may be the future in the search for a cure for cancer. Uncapping the true potential of current nanomedicines will require the overcoming of adaptive resistance, and hence, the design of next-generation nanomedicines needs to be inspired by the mechanistic understanding underlying chemotherapy-induced phenotypic state changes underlying adaptive resistance. Furthermore, we show that a combination of active agents that can target pathways implicated in adaptive resistance, delivered from the same nanoparticle (i.e., a 2-in-1 “chimeric” nanoparticle), exerts the maximal antitumor efficacy compared with that of a simple combination of two therapeutic agents delivered using two distinct nanoparticles, thus addressing a fundamental question in the design of combinatorial nanomedicines for cancer chemotherapy.
METHODS Materials. Unless noted otherwise, all reagents and drugs were purchased from Sigma-Aldrich (St. Louis, MO). Cabazitaxel and dasatinib were purchased from LC Laboratories (Woburn, MA). PI103 was purchased from Selleckchem. All chemical reagents were of analytical grade and used as supplied without further purification unless indicated. Dichloromethane (DCM), anhydrous DCM, methanol, L -α-phosphatidylcholine, and Sephadex G-25 were purchased from Sigma-Aldrich. 1,2-Distearoyl-sn-glycero-3-phosphoethanolamine-N-[amino(polythylene glycol)2000] (DSPE-PEG) and the mini hand-held extruder kit (including 0.2 μm Whatman nucleopore track-etch membrane, Whatman filter supports, and 1.0 mL Hamiltonian syringes) were bought from Avanti Polar Lipids Inc. 5831
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano cells were trypsinized and replated at a density of 1.5−2 × 105 cells/ mL and cultured in serum-containing medium. After 24 h incubation, floating cells were removed and remaining cells were washed with 1× PBS and considered as chemotherapy-tolerant cells. Cytotoxicity and Cell Viability Assays. Cells were treated with the drugs or nanoparticles for 48 h. Following incubation, cells were washed with PBS and recovered in serum and phenol red-free DMEM and subsequently treated with MTS reagent using the manufacturer’s protocol (Promega, Madison, WI). Validation of cytotoxicity was performed by using trypan blue exclusion. FACS and DNA Content Analyses. Cells were cultured as indicated and fixed with 4% paraformaldehyde in PBS for 30 min at room temperature and blocked in 10% goat serum (v/v), and 0.05% saponin was used to permeabilize cells when necessary. Following PBS washes, cells were incubated with CD24-PE and CD44-APC (BD Biosciences, San Jose, CA) for 60 min at room temperature or overnight at 4 °C and analyzed by FACS (Accuri Cyomteters Inc., Ann Arbor, MI). Single stain controls were used to set gating parameters and any compensations. All FACS results were analyzed by FlowJo software following a rigorous doublet discrimination based on FSC:A versus width as well as FSC:A versus height (Tree Star Inc., Ashland, OR). Analyses were also performed through Accuri cFlow plus software to obtain and confirm mean fluorescent intensity (GNU.org). Cell cycle analysis was performed as follows: cells were treated as described above or in figure legends, and following two washes with 1× PBS, cells were trypsinized and collected. Permeabilization was achieved by incubation in 70% ethanol overnight at 4 °C. Cells were then washed two times with 1× PBS and incubated with RNAase A for 15 min at 37 °C followed by PI solution for 30 min at 4 °C (Genscript USA Inc., Piscataway, NJ). Cells were read at 594/535 excitation/ emission by flow cytometry (BD Accuri C6, Ann Arbor, MI). All results were analyzed by FlowJo flow cytometric analysis and cell cycle analysis software following a rigorous doublet discrimination based on FSC:A versus width as well as FSC:A versus height. Cell sorting was performed on live cells. Briefly, for CD44/CD24 isolation fractions, cells were incubated with fluorescent antibody for 1 h at room temperature in PBS. Following washes, cells were sorted by FACS (BD FACS Aria IIU Special Order, Ann Arbor, MI). For isolation of fractions based on size, live cells were processed based on FSC:A and SSC:A fractions and then further discriminated for size based on FSC:A versus FSC:W and FSC:H. These parameters were used to isolate small and large cells. In Vivo Experiments. All in vivo experiments were performed in compliance with IACUC protocol approved through Harvard Medical School and in accordance with institutional guidelines, supervised onsite by veterinary staff. 4T1 mouse mammary carcinoma cells (106 cells) were injected into the flanks of 5−6 week old female balb/c mice. Docetaxel was dissolved in pure ethanol at a concentration of 50 mg/mL mixed 1:1 with polysorbate 80 (Tween 80) and brought to a final working concentration with 5% glucose in PBS. Once tumors became palpable (∼100 mm3), DTX or vehicle treatments were administered at 100 μL volumes either intratumorally or intravenously (i.v.). Dasatinib was dissolved in DMSO to a working concentration and delivered as 50 μL injections intraperitoneally on indicated days. SPI103 or chimeric nanoparticles were injected i.v. Tumor volumes were quantified using digital calipers (Starlett, Athol, MA) by a third party unaware of treatment conditions. Immunohistochemistry. Tumor tissues were fixed in 4% buffered formalin and embedded in paraffin or frozen by liquid nitrogen in OCT compound. Prior to immunohistochemical staining of target proteins, 4 μm thick tissue sections mounted in poly-L-lysine-coated glass slides were deparaffinized and rehydrated. Heat-induced antigen retrieval was achieved using citrate buffer (pH 7.8). The sections were soaked in antigen unmasking solution (Vector Burlingame, CA) for 10 min followed by retrieval using a microwave for 25 min. Endogenous hydrogen peroxidase was blocked by incubating the sections with 3% H2O2 (Merck) for 15 min and washed in running tap water for 3 min followed by a wash in 1× TBS for 7 min. After initial blocking of the slides in 10% normal goat serum (Vector Laboratory) for 1 h at room temperature, tissue sections were incubated with primary antibodies
for an additional 1 h at room temperature. The following primary antibodies were used: anti-human CD44 (Clone IM7), anti-human CD24 (ML5). Secondary antibody (signal stain boost IHC detection reagent HRP, Cell Signaling Technology) was added to the sections and incubated for 45 min at room temperature and washed four times in 1× PBS for 3 min each. Appropriate isotype-matched IgG controls were included for each secondary antibody. Chromogenic development was done by exposure of tissues to the DAB substrate (DAB peroxidase substrate kit, Vector Laboratories). Images of immunostained sections were visualized by a NIKON Eclipse TI-U microscope with a 20× ELDW, 10 or 40× Plan-Apo objective lens (Nikon, Melville, NY), and images were acquired. IHC performed from frozen sections was fixed with 10% formalin and permeabilized with 0.05% saponin or fixed and permeabilized with ice-cold methanol. Frozen section IHC was visualized by confocal microscopy as described below. Immunohistochemical images shown in figures chosen as representative are derived as examples determined by an experienced pathologist in each case to reflect overall alterations in tissue staining/architecture of the respective experiments performed. Confocal and Immunofluorescence Microscopy. Cells were generated as described above and plated in four chamber glass slides (BD Biosciences, San Jose, CA) at a concentration of 100 000 cells/ mL. Following treatments, cells were washed in PBS and fixed in 4% paraformaldehyde for 30 min. Permeabilization, when necessary, was achieved with 10% (v/v) goat serum (Vector Laboratories, Burlingame, CA) and 0.05% saponin (w/v) in PBS for 90 min. Blocking was performed in 10% (v/v) goat serum in PBS. The cells were labeled with the indicated primary antibodies CD44 (clone IM7 from eBioScience) conjugated to FITC (AnaSpec, Freemont, CA) at 1:100, CD24 (BD Biosciences, San Jose, CA) conjugated to Fluor488 (Anaspec, Freemont, CA), pHCK-Y410 (Abcam, Cambridge, MA), and pAKTS473 (Cell Signaling, Cambridge, MA) at 1:250 and masked with DAPI-containing hard-set mounting medium (Vector Laboratories, Burlingame, CA). Bright-field and fluorescent images were obtained using three channels on a NIKON Eclipse TI-U microscope with a 20× ELDW, 10 or 40× Plan-Apo objective lens (Nikon, Melville, NY). NIS Elements Viewer version 3.22 (Nikon, Melville, NY) software was used to capture the images to file. Confocal microscopy was performed with an inverted Nikon confocal microscope (TE2000) with Auto DeVlur deconvolution software and fitted with three laser detection (Nikon, Melville, NY). Gains were set manually based on negative control stains (secondary antibody only) and were left unaltered between treatment groups of similar experiments. When representative images are shown in figures, these are derived from experiments performed in at least biological triplicate on independent occasions. In general, images were obtained from more than 100 cells per conditions and chosen to represent the overall alterations in each experimental group. Quantifications were performed by visualizing at least 25 individual cells within a single frame; Adobe CS5 was employed to quantify fluorescence of a randomly assigned field and recorded in an unbiased fashion to obtain an arbitrary fluorescence intensity. Immunoblotting. Laemmli sample buffer was prepared as a 5× solution containing β-mercaptoethanol as a reducing agent. Immunoprecipitation was performed using both classic and direct IP kits purchased from Pierce following the manufacturer’s protocols (Thermo Fisher Inc., Rockford, IL). Briefly, cell lysates were prepared using IP/lysis buffer (Thermo Fisher Inc., Rockford, IL) in the presence of 2× HALT protease/phosphatase inhibitor cocktail (Thermo Fisher Inc., Rockford, IL). Protein samples were resolved by SDS-PAGE and transferred to PVDF membranes prior to incubation at 4 °C with indicated primary antibodies (β-actin, pAKT, pMTOR, pERM, pEzrin), and total protein antibodies for each target were purchased from Cell Signaling (Cambridge, MA). Cav-1 was purchased from BD Biosciences (San Jose, CA). PVDF membranes with primary antibody were incubated at room temperature with HRP-conjugated secondary antibodies (BD, Ann Arbor, MI) and resolved by chemiluminescence using the G-Box and Syngene software (Syngene, Cambridge, UK). When possible, blots were stripped (Thermo Fisher, Rockford, IL) and reprobed with a second 5832
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano primary antibody. Western blotting images chosen as representative depictions in the figures demonstrate equivalent results taken from biological replicates (N > 3). P53 phosphorylation blots were obtained using the proteome profiler array (R&D Biosciences, Minneapolis, MN). Mathematical Modeling. We assume that all large cells have high CD44 expression, and small/large cells with high expression of CD44 also have high DNA content. Thus, our model consists of four cellular compartments describing the population of (1) large cells with high DNA/CD44 content, (2) small cells with high DNA/CD44 content, (3) small cells with high DNA content, and (4) small cells with low DNA/CD44 content, denoted by X(t), Y(t), Z(t), and W(t), respectively. The model is summarized in Figure 2F. The model consists of 11 parameters, four of which describe the net proliferation rates for each of the compartments, and the remaining seven parameters describe the possible transition rates between the compartments. All the rates have the unit (1/hours). Based on our experimental data, cells start to die after about 8 h, thus all death rates are set to zero for t < 8 h. In addition, the cells do not change their phenotype after removing the chemotherapy, thus all transition rates are set to zero after 48 h. Moreover, we assume that cells do not proliferate during the observation time (0−72 h). Let dk be the death rate of cell compartment k, and rij be the rate of transfer of cells from compartment i to compartment j, we have
ACKNOWLEDGMENTS We thank the Dana Farber Cancer Institute Flow Cytometry Core Facility for their expertise, consulting, and assistance with cell sorting experiments. The authors also thank Dr. D. Goldman, MD, for consultation of histology and IHC, and Dr. Mehran Kardar, PhD, for his helpful discussion during experimental designs and computational modeling. S.S. is supported by a DoD BCRP Collaborative Innovator Grant (W81XWH-09-1-0700), a NIH RO1 (1R01CA135242-01A2), a DoD Breakthrough Award (BC132168), an American Lung Association Innovation Award (LCD-259932-N), and an Indo−US Joint Center Grant from IUSSTF. A.G. is supported by an American Cancer Society Postdoctoral Fellowship (122854-PF-12-226-01-CDD). Some images in schematics were obtained from servier.com medical image bank. REFERENCES (1) Holohan, C.; Van Schaeybroeck, S.; Longley, D. B.; Johnston, P. G. Cancer Drug Resistance: an Evolving Paradigm. Nat. Rev. Cancer 2013, 13, 714−26. (2) Cairns, J. Mutation Selection and the Natural History of Cancer. Nature 1975, 255, 197−200. (3) Kreso, A.; O’Brien, C. A.; van Galen, P.; Gan, O. I.; Notta, F.; Brown, A. M.; Ng, K.; Ma, J.; Wienholds, E.; Dunant, C.; Pollett, A.; Gallinger, S.; McPherson, J.; Mullighan, C. G.; Shibata, D.; Dick, J. E. Variable Clonal Repopulation Dynamics Influence Chemotherapy Response in Colorectal Cancer. Science 2013, 339, 543−8. (4) Cohen, A. A.; Geva-Zatorsky, N.; Eden, E.; Frenkel-Morgenstern, M.; Issaeva, I.; Sigal, A.; Milo, R.; Cohen-Saidon, C.; Liron, Y.; Kam, Z.; Cohen, L.; Danon, T.; Perzov, N.; Alon, U. Dynamic Proteomics of Individual Cancer Cells in Response to a Drug. Science 2008, 322, 1511−6. (5) Dawson, C. C.; Intapa, C.; Jabra-Rizk, M. A. ″Persisters″: Survival at the Cellular Level. PLoS Pathog. 2011, 7, e1002121. (6) Barclay, M. L.; Begg, E. J.; Chambers, S. T. Adaptive Resistance Following Single Doses of Gentamicin in a Dynamic in vitro Model. Antimicrob. Agents Chemother. 1992, 36, 1951−7. (7) Purvis, J. E.; Lahav, G. Encoding and Decoding Cellular Information Through Signaling Dynamics. Cell 2013, 152, 945−56. (8) Kummar, S.; Chen, H. X.; Wright, J.; Holbeck, S.; Millin, M. D.; Tomaszewski, J.; Zweibel, J.; Collins, J.; Doroshow, J. H. Utilizing Targeted Cancer Therapeutic Agents in Combination: Novel Approaches and Urgent Requirements. Nat. Rev. Drug Discovery 2010, 9, 843−56. (9) Goldman, A.; Majumder, B.; Dhawan, A.; Ravi, S.; Goldman, D.; Kohandel, M.; Majumder, P. K.; Sengupta, S. Temporally Sequenced Anticancer Drugs Overcome Adaptive Resistance by Targeting a Vulnerable Chemotherapy-induced Phenotypic Transition. Nat. Commun. 2015, 6, 6139. (10) Pandey, A.; Kulkarni, A.; Roy, B.; Goldman, A.; Sarangi, S.; Sengupta, P.; Phipps, C.; Kopparam, J.; Oh, M.; Basu, S.; Kohandel, M.; Sengupta, S. Sequential Application of a Cytotoxic Nanoparticle and a PI3K Inhibitor Enhances Antitumor Efficacy. Cancer Res. 2014, 74, 675−85. (11) Davis, M. E.; Chen, Z. G.; Shin, D. M. Nanoparticle Therapeutics: An Emerging Treatment Modality for Cancer. Nat. Rev. Drug Discovery 2008, 7, 771−82. (12) Ma, L.; Kohli, M.; Smith, A. Nanoparticles for Combination Drug Therapy. ACS Nano 2013, 7, 9518−25. (13) Zhang, H.; Wang, G.; Yang, H. Drug Delivery Systems for Differential Release in Combination Therapy. Expert Opin. Drug Delivery 2011, 8, 171−90. (14) Guo, S.; Lin, C. M.; Xu, Z.; Miao, L.; Wang, Y.; Huang, L. CoDelivery of Cisplatin and Rapamycin for Enhanced Anticancer Therapy Through Synergistic Effects and Microenvironment Modulation. ACS Nano 2014, 8, 4996−5009.
dX = rYXY + rWXW + rZXZ − dX X dt
dY = rZY Z + rWY W − (rYZ + rYX )Y − dY Y dt dZ = rYZY + rWZW − (rZY + rZX )Z − dZZ dt
dW = − (rWX + rWY + rWZ)W − dW W dt The transition parameters were obtained by fitting the model equations to the experimental data for cell populations at 0, 2, 4, 8, and 72 h. Statistics. Statistical analysis was carried out with Prism software (Graphpad, LaJolla, CA). Experimental data are expressed as mean ± SEM and analyzed using ANOVA followed by Bonferroni post-test or Student’s t test. Combination index was based on a median-effect equation derived from the mass action law principle known as the Chou−Talalay method,34 where two drugs are additive when CI = 1, synergistic when CI < 1, or antagonistic at CI > 1.
ASSOCIATED CONTENT S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.6b00320. Figures S1−S4 (PDF)
AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Author Contributions ▲
A.G. and A.K. contributed equally to this work.
Notes
The authors declare the following competing financial interest(s): SS is the co-founder of Cerulean Pharmaceuticals and serves on the SAB. SS is the co-founder and director of Invictus Oncology. SS holds equity in both Cerulean Pharma and Invictus Oncology. 5833
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834
Article
ACS Nano (15) Sengupta, S.; Eavarone, D.; Capila, I.; Zhao, G.; Watson, N.; Kiziltepe, T.; Sasisekharan, R. Temporal Targeting of Tumour Cells and Neovasculature With a Nanoscale Delivery System. Nature 2005, 436, 568−72. (16) Morton, S. W.; Lee, M. J.; Deng, Z. J.; Dreaden, E. C.; Siouve, E.; Shopsowitz, K. E.; Shah, N. J.; Yaffe, M. B.; Hammond, P. T. A Nanoparticle-Based Combination Chemotherapy Delivery System for Enhanced Tumor Killing by Dynamic Rewiring of Signaling Pathways. Sci. Signaling 2014, 7, ra44. (17) Fillmore, C. M.; Kuperwasser, C. Human Breast Cancer Cell Lines Contain Stem-like Cells That Self-renew, Give Rise to Phenotypically Diverse Progeny and Survive Chemotherapy. Breast Cancer Res.: BCR 2008, 10, R25. (18) Kulkarni, A. A.; Rao, P.; Goldman, A. J.; Sengupta, S. Computationally-Inspired Engineering of Supramolecular Taxane Nanoparticles. Cancer Res. 2014, 74 (19 Suppl), 5419. (19) Miele, E.; Spinelli, G. P.; Miele, E.; Tomao, F.; Tomao, S. Albumin-Bound Formulation of Paclitaxel (Abraxane ABI-007) in the Treatment of Breast Cancer. Int. J. Nanomed. 2009, 4, 99−105. (20) Zapperi, S.; La Porta, C. A. Do Cancer Cells Undergo Phenotypic Switching? The Case for Imperfect Cancer Stem Cell Markers. Sci. Rep. 2012, 2, 441. (21) Watters, J. W.; Kraja, A.; Meucci, M. A.; Province, M. A.; McLeod, H. L. Genome-Wide Discovery of Loci Influencing Chemotherapy Cytotoxicity. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 11809−14. (22) Fauzee, N. J. Taxanes: Promising Anti-Cancer Drugs. Asian Pac. J. Cancer Prev. 2011, 12, 837−51. (23) Palmieri, C.; Krell, J.; James, C. R.; Harper-Wynne, C.; Misra, V.; Cleator, S.; Miles, D. Rechallenging With Anthracyclines and Taxanes in Metastatic Breast Cancer. Nat. Rev. Clin. Oncol. 2010, 7, 561−74. (24) Roberts, J. R.; Allison, D. C.; Donehower, R. C.; Rowinsky, E. K. Development of Polyploidization in Taxol-Resistant Human Leukemia Cells in vitro. Cancer Res. 1990, 50, 710−6. (25) Al-Hajj, M.; Wicha, M. S.; Benito-Hernandez, A.; Morrison, S. J.; Clarke, M. F. Prospective Identification of Tumorigenic Breast Cancer Cells. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 3983−8. (26) Machado, H. L.; Kittrell, F. S.; Edwards, D.; White, A. N.; Atkinson, R. L.; Rosen, J. M.; Medina, D.; Lewis, M. T. Separation by Cell Size Enriches for Mammary Stem Cell Repopulation Activity. Stem Cells Transl. Med. 2013, 2, 199−203. (27) Zhang, S.; Mercado-Uribe, I.; Xing, Z.; Sun, B.; Kuang, J.; Liu, J. Generation of Cancer Stem-Like Cells Through the Formation of Polyploid Giant Cancer Cells. Oncogene 2014, 33, 116−28. (28) Harper, L. J.; Costea, D. E.; Gammon, L.; Fazil, B.; Biddle, A.; Mackenzie, I. C. Normal and Malignant Epithelial Cells With StemLike Properties Have an Extended G2 Cell Cycle Phase That is Associated with Apoptotic Resistance. BMC Cancer 2010, 10, 166. (29) Chang, F.; Lee, J. T.; Navolanic, P. M.; Steelman, L. S.; Shelton, J. G.; Blalock, W. L.; Franklin, R. A.; McCubrey, J. A. Involvement of PI3K/Akt Pathway in Cell Cycle Progression, Apoptosis, and Neoplastic Transformation: A Target for Cancer Chemotherapy. Leukemia 2003, 17, 590−603. (30) Clark, A. S.; West, K.; Streicher, S.; Dennis, P. A. Constitutive and Inducible Akt Activity Promotes Resistance to Chemotherapy, Trastuzumab, or Tamoxifen in Breast Cancer Cells. Mol. Cancer Ther. 2002, 1, 707−17. (31) Gautreau, A.; Poullet, P.; Louvard, D.; Arpin, M. Erzin, a Plasma Membrane-Microfilament Linker, Signals Cell Survival Through the Phosphatidylinositol 3-kinase/Akt Pathway. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 7300−5. (32) Essigmann, J. M.; Barker, L. J.; Fowler, K. W.; Francisco, M. A.; Reinhold, V. N.; Wogan, G. N. Sterigmatocystin-DNA Interactions: Identification of a Major Adduct Formed After Metabolic Activation in vitro. Proc. Natl. Acad. Sci. U. S. A. 1979, 76, 179−83. (33) Xie, T. X.; Misumi, J.; Aoki, K.; Zhao, W. Y.; Liu, S. Y. Absence of p53-Mediated G1 Arrest With Induction of MDM2 in Sterigmatocystin-Treated Cells. Int. J. Oncol. 2000, 17, 737−42.
(34) Chou, T. C. Drug Combination Studies and Their Synergy Quantification Using the Chou-Talalay Method. Cancer Res. 2010, 70, 440−6. (35) Kulkarni, A. A.; Roy, B.; Rao, P. S.; Wyant, G. A.; Mahmoud, A.; Ramachandran, M.; Sengupta, P.; Goldman, A.; Kotamraju, V. R.; Basu, S.; Mashelkar, R. A.; Ruoslahti, E.; Dinulescu, D. M.; Sengupta, S. Supramolecular Nanoparticles That Target Phosphoinositide-3Kinase Overcome Insulin Resistance and Exert Pronounced Antitumor Efficacy. Cancer Res. 2013, 73, 6987−97. (36) Vermeer, L. S.; de Groot, B. L.; Reat, V.; Milon, A.; Czaplicki, J. Acyl Chain Order Parameter Profiles in Phospholipid Bilayers: Computation From Molecular Dynamics Simulations and Comparison with 2H NMR Experiments. Eur. Biophys. J. 2007, 36, 919−31. (37) Choudhury, C. K.; Kumar, A.; Roy, S. Characterization of Conformation and Interaction of Gene Delivery Vector Polyethylenimine with Phospholipid Bilayer at Different Protonation State. Biomacromolecules 2013, 14, 3759−68. (38) Kuczynski, E. A.; Sargent, D. J.; Grothey, A.; Kerbel, R. S. Drug Rechallenge and Treatment Beyond Progression–Implications for Drug Resistance. Nat. Rev. Clin. Oncol. 2013, 10, 571−87. (39) Miao, L.; Guo, S.; Zhang, J.; Kim, W. Y.; Huang, L. Nanoparticles with Precise Ratiometric Co-Loading and Co-Delivery of Gemcitabine Monophosphate and Cisplatin for Treatment of Bladder Cancer. Adv. Funct. Mater. 2014, 24, 6601−6611. (40) Liao, L.; Liu, J.; Dreaden, E. C.; Morton, S. W.; Shopsowitz, K. E.; Hammond, P. T.; Johnson, J. A. A Convergent Synthetic Platform for Single-Nanoparticle Combination Cancer Therapy: Ratiometric Loading and Controlled Release of Cisplatin, Doxorubicin, and Camptothecin. J. Am. Chem. Soc. 2014, 136, 5896−9. (41) Tardi, P.; Johnstone, S.; Harasym, N.; Xie, S.; Harasym, T.; Zisman, N.; Harvie, P.; Bermudes, D.; Mayer, L. In vivo Maintenance of Synergistic Cytarabine:Daunorubicin Ratios Greatly Enhances Therapeutic Efficacy. Leuk. Res. 2009, 33, 129−39. (42) Meng, H.; Wang, M.; Liu, H.; Liu, X.; Situ, A.; Wu, B.; Ji, Z.; Chang, C. H.; Nel, A. E. Use of a lipid-Coated Mesoporous Silica Nanoparticle Platform for Synergistic Gemcitabine and Paclitaxel Delivery to Human Pancreatic Cancer in Mice. ACS Nano 2015, 9, 3540−57. (43) Gupta, P. B.; Fillmore, C. M.; Jiang, G.; Shapira, S. D.; Tao, K.; Kuperwasser, C.; Lander, E. S. Stochastic State Transitions Give Rise to Phenotypic Equilibrium in Populations of Cancer Cells. Cell 2011, 146, 633−44.
5834
DOI: 10.1021/acsnano.6b00320 ACS Nano 2016, 10, 5823−5834