Anti-Invasive Adjuvant Therapy with Imipramine Blue Enhances Chemotherapeutic Efficacy Against Glioma

See allHide authors and affiliations

Science Translational Medicine  28 Mar 2012:
Vol. 4, Issue 127, pp. 127ra36
DOI: 10.1126/scitranslmed.3003016


The invasive nature of glioblastoma (GBM) represents a major clinical challenge contributing to poor outcomes. Invasion of GBM into healthy tissue restricts chemotherapeutic access and complicates surgical resection. Here, we test the hypothesis that an effective anti-invasive agent can “contain” GBM and increase the efficacy of chemotherapy. We report a new anti-invasive small molecule, Imipramine Blue (IB), which inhibits invasion of glioma in vitro when tested against several models. IB inhibits NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase–mediated reactive oxygen species generation and alters expression of actin regulatory elements. In vivo, liposomal IB (nano-IB) halts invasion of glioma, leading to a more compact tumor in an aggressively invasive RT2 syngeneic astrocytoma rodent model. When nano-IB therapy was followed by liposomal doxorubicin (nano-DXR) chemotherapy, the combination therapy prolonged survival compared to nano-IB or nano-DXR alone. Our data demonstrate that nano-IB–mediated containment of diffuse glioma enhanced the efficacy of nano-DXR chemotherapy, demonstrating the promise of an anti-invasive compound as an adjuvant treatment for glioma.


Brain tumors present a clinical challenge with poor prognosis after treatment (1). The most recent advance in clinical treatment of glioblastoma (GBM) incorporated chemotherapy (temozolomide) with radiotherapy (2); although improved clinical outcomes were noted with this current standard of care, death still occurs within 2 years of diagnosis (3). The efficacy of therapy is limited by the invasiveness of glioma into healthy brain tissue, making it a “moving target” (4). Additionally, radiotherapy and chemotherapy have been shown to increase the invasive potential of GBM in vivo (5, 6). In animal models of glioma, including RT2 (7), 9L (8), and C6 (9), the use of temozolomide and radiotherapy has shown some enhancement in survival times, although no complete response has been demonstrated, like in humans. Therefore, a new treatment strategy is warranted: one that incorporates an anti-invasive drug with a strong cytotoxic agent.

Compounds that halt brain tumor invasion along with growth and proliferation have been investigated previously (10, 11) without much success, and their dosing to tumors remains challenging because of poor pharmacokinetics and adverse side effects (12). We have demonstrated previously that ~120-nm PEGylated liposomal nanocarriers can effectively localize to 9L and U87MG glioma models in vivo (13, 14). Compared to other nanocarrier systems, liposomal nanocarriers offer advantages for drug delivery including Food and Drug Administration (FDA)–approved status with anticancer compounds, such as doxorubicin (DXR), and well-studied pharmacokinetics and biodistribution in humans and animals (15). DXR encapsulated in liposomes has shown enhanced survival in humans with GBM compared to the standard of care while limiting negative side effects of chemotherapy (16, 17).

Here, we tested the hypothesis that limiting GBM invasion enhances the efficacy of chemotherapy with DXR, as depicted in Fig. 1A. This figure further shows that with normal therapy, invasion occurs, which leads to tumor recurrence; however, with integration of Imipramine Blue (IB), we propose that invasion can be preemptively limited before traditional antitumor therapy. Recent reports suggest that formation of invadopodia, a feature of cancer cell invasion, is NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase–mediated through phosphorylation of tyrosine kinase substrate (Tks) proteins, leading to alteration of the actin cytoskeleton (18). Triphenylmethanes have been shown to be potent inhibitors of NADPH oxidases (19, 20). Here, we characterize the anti-invasive potential of a novel triphenylmethane dye, IB. We demonstrate that IB encapsulated in liposomal nanocarriers (nano-IB) effectively inhibits tumor invasion in vivo using an aggressive rodent model of high-grade astrocytoma (RT2). Sequential delivery of nano-IB and liposomal DXR (nano-DXR) to RT2 resulted in long-term survival. Localized pretreatment with an anti-invasive agent coupled with a potent chemotherapeutic could represent a shift in the treatment paradigm for GBM—one that could positively affect the clinical outcomes in patients plagued with this disease.

Fig. 1

Preventing cancer cell invasion may enhance the benefits of traditional antitumor therapy. (A) Schematic comparing the traditional antitumor treatment (radiation, chemotherapy, or resection) to pretreatment with IB. In traditional therapy (upper panel), a tumor develops in the cortex, and antitumor therapy is immediately delivered but fails to eradicate all the tumor cells due to invasion and resistance. This results in recurrence after treatment and distant invasion. Pretreatment with the anti-invasive agent nano-IB (lower panel) allows for containment of the tumor as it grows. This treatment is then followed by aggressive DXR chemotherapy (nano-DXR), which can more effectively eradicate the tumor and prevent recurrence from invasion. (B) Structure of IB (molecular weight = 588.34 g/mol). Et, ethyl (C2H5).


Synthesis of IB

IB (Fig. 1B) was synthesized via a single-step process (fig. S1A) to yield a compound with a molecular weight of 588 g/mol. This compound was crystallized to a dark blue/brown powder that yielded a deep blue–colored solution when dissolved at neutral pH. The structure of IB was confirmed by nuclear magnetic resonance spectroscopy (fig. S1B). IB absorbs light at 610 nm, as determined through absorbance spectrophotometry (fig. S1B).

IB inhibition of glioma invasion in vitro

Glioma cell lines seeded on basement membrane extract in porous tissue culture inserts were exposed to IB, and total cell invasion was determined (as percent of total cells). Anti-invasive doses (0.01 to 20 μM) remained below cytotoxic levels of IB to RT2 glioma and primary astrocytes (fig. S2). IB at 5 μM inhibited the invasion of both human (U87MG) and rat (RT2, C6, and 9L) GBM cell lines at 24 hours (Fig. 2A). Furthermore, the ability of IB to attenuate invasion was dose-dependent, revealing an EC50 (median effective concentration) in RT2 rat GBM cells of 10 nM at 24 hours (Fig. 2B). RT2 cell spreading was inhibited and could be imaged in real time (video S1). Rat astrocyte migration remained unaffected at doses up to 10 μM in 72 hours (Fig. 2C), indicating a specificity for inhibition of cancer cell motility.

Fig. 2

IB inhibits invasion in vitro of glioma cell lines. (A) Invasion of RT2 rat astrocytoma (n = 6), C6 rat astrocytoma (n = 3), U87MG human GBM (n = 3), and 9L rat gliosarcoma (n = 3) in vitro through basement membrane–coated tissue culture inserts (5 μM IB). Data are shown as invaded cells/total cells (invasion %). *P < 0.05 by t test. (B) Dose-dependent IB-mediated inhibition of RT2 cell invasion (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001 compared to 0 μM IB by ANOVA and t test. (C) Primary rat astrocyte migration in tissue culture insert at IB doses ranging from 0 to 10 μM (n = 3). All data are shown as mean ± SEM, and n indicates an independent experiment with three replicates per condition.

To determine the applicability of the drug to humans, we embedded patient-derived neurospheres of GBM (N08-32 and N08-1002) (>100-μm diameter) in collagen matrices with or without IB and imaged at 7 days. Staining for F-actin (rhodamine-phalloidin) was used to identify cell outgrowth and showed that treated neurospheres had fewer spikes indicative of reduced invasion into the gel (Fig. 3A). These images were coupled with invasion assays with single primary cells through three-dimensional (3D) collagen matrix (Fig. 3B), demonstrating IB-mediated inhibition of invasion at a concentration of 10 μM, slightly higher than that seen in 2D cell culture, likely owing to absorption of IB by the matrix.

Fig. 3

IB inhibits invasion and growth of primary GBM neurospheres. (A) Established neurospheres from human GBM (two patients) were embedded in collagen matrix (n = 3 independent trials, one representative condition is shown) and treated with (untreated, left) or without 10 μM IB (right). Neurospheres were fixed after 7 days and stained with rhodamine-phalloidin to elucidate cell morphology. These are representative images of two neurosphere cultures N08-1002 and N08-32. Scale bar, 100 μm. (B) Invasion of dissociated neurosphere cells through collagen matrix in porous tissue culture inserts as the percentage of invasive cells of total cells seeded into the matrix. Data are means ± SEM (n = 3 independent experiments). **P < 0.01, t test.

Liposomal nanocarrier encapsulation of IB and in vivo pharmacokinetics

IB (in cremophor/ethanol/saline) was administered intravenously to adult male Fischer rats. This dose yielded a circulation half-life (t1/2) of 11 min in rats without tumor (Fig. 4A, inset). To circumvent the short t1/2, we encapsulated IB in ~160-nm PEGylated liposomes. IB was loaded consistently into liposomes with a molar drug/lipid ratio of 0.07. Systemic injection of nano-IB (3 mg/kg) into adult male rats significantly increased the circulation time of IB to a t1/2 of 18 hours (Fig. 4A) (comparison of fits F test, P < 0.001). Nano-IB (5 μM IB) and free IB yielded significant inhibition in invasion assays compared to control (saline liposomes) (Fig. 4B).

Fig. 4

Encapsulation of IB in liposomal nanocarriers prolongs circulation and targets the tumor in vivo. (A) Circulation time of free IB and liposomal IB (nano-IB) (3 mg/kg) in vivo in adult male rats (n = 4 per group), as assessed through orbital plasma collection. Data are means ± SEM, with one-phase decay curve fits [R2 = 0.93 (free) and 0.88 (nano-IB)] showing a t1/2 = 11 min (free) and 18 hours (nano-IB). Inset is free IB pharmacokinetics over the course of 1 hour. (B) RT2 cell invasion (% invaded cells/seeded cells) through tissue culture inserts coated with basement membrane extract in the presence of free IB (5 μM), nano-IB (5 μM IB), or blank liposomes (Control). Data are analyzed by one-way ANOVA with post hoc t tests. ***P < 0.001. (C) T1-weighted MR images at tumor inoculation site (dotted circle) showing Gd-DTPA–loaded nano-IB in tumor on day 4 after RT2 tumor inoculation (precontrast), then on days 6, 8, and 14 in live rats. White arrows indicate areas of enhanced T1 signal. (D) T2-weighted MR images at the same location showing brain morphology with RT2 tumor at days 4, 6, 8, and 14 after inoculation. Dotted circles indicate location of tumor inoculation. Scale bars, 1 mm.

Solid tumors display an enhanced permeability and retention (EPR) effect that allows increased tumor uptake of nanocarriers (21). The EPR effect was confirmed by intravenous injection of liposomal nanocarriers loaded with IB and the magnetic resonance imaging (MRI) contrast agent gadolinium–diethylenetriamine pentaacetic acid (Gd-DTPA) into rats 4 days after RT2 tumor inoculation. Visualization with T1-weighted MRI in RT2 tumor–bearing rats showed liposome accumulation at the tumor site after 48 hours (day 6) and subsequent diffusion into (day 8) and clearance from (day 14) the brain (Fig. 4C). Companion T2-weighted images are shown in Fig. 4D.

Organs in RT2-bearing rats receiving nano-IB or blank liposomes (n = 4 per group) were collected, and toxicity was assessed through hematoxylin and eosin (H&E) staining. Tissue sections from the spleen, liver, lungs, kidneys, and heart were examined by an anatomic pathologist for signs of acute or chronic toxicity. Tissue sections showed no evidence of acute, chronic, or granulomatous inflammation (fig. S3). There was no evidence of tissue injury, such as frank necrosis, edema, fibrosis, hemorrhage, congestion, or acute cellular toxicity. In the kidney, there was no acute tubular necrosis, and the liver showed no evidence of centrilobular necrosis. The tissue further showed no signs of enlargement, atrophy, or gross evidence of vascular congestion. All results indicate no toxicity associated with intravenously administered nano-IB.

Nano-IB inhibition of RT2 glioma invasion in vivo

Although IB was effective at inhibiting invasion of U87MG, C6, and 9L glioma cell lines in vitro, the enhanced green fluorescent protein (eGFP)–expressing RT2 rat astrocytoma intracranial model was used for in vivo studies. The RT2 model shows attributes similar to human GBM including invasion along white matter tracts and subpial linings, as well as perineuronal and perivascular satellitosis (fig. S4). The RT2 model also has reproducible time to euthanasia after implantation of 17 ± 1.9 days and physiological immune-tumor response (22).

To assess invasion in vivo, we delivered nano-IB (3 mg/kg) on days 4 and 7 after inoculation of rats with RT2 glioma. Animals were euthanized 11 days after inoculation for tumor analysis. The nano-IB–treated tumors were compacted compared to control (saline liposome–treated) tumors (Fig. 5A). Additionally, in the nano-IB–treated animals, tumor margins (defined to be the point at which there was >50% staining for neurofilament beyond the glioma) showed fewer glioma cells crossing into healthy brain (Fig. 5B). Tumor density—measured as 4′,6-diamidino-2-phenylindole (DAPI)–stained nuclei per coronal section area of tumors—increased significantly upon treatment (Fig. 5C). Correspondingly, tumor margins showed significantly decreased glioma invasion into healthy brain after treatment with nano-IB (Fig. 5D), counting the number of cells beyond the border. H&E staining of the tumor borders showed invasion similar to that seen using immunofluorescence staining (Fig. 5, B and E). Nano-IB treatment alone led to a small increase in median survival time (11.4%) when animals were treated on days 4 and 7 after inoculation (fig. S5). Because IB is only an invasion-inhibiting agent, coupling with a more powerful antitumor chemotherapeutic agent was necessary to prolong survival more effectively.

Fig. 5

In vivo delivery of nano-IB yields decreased glioma invasion. All treatments were two injections of nano-IB (3 mg/kg) (treated) or saline liposomes (untreated) on days 4 and 7 and analysis on day 11 after tumor inoculation. (A) Untreated and treated RT2 tumors in coronal slices stained for nuclei (white) (representative images from n = 4 per group). Scale bar, 1 mm. (B) Representative images of untreated (n = 4) and treated (n = 4) tumors used to quantify cellular invasion beyond the border (blue dotted line) by counting the number of glioma cells (green) per area of healthy tissue (neurofilament, red). Scale bar, 100 μm. (C) Quantification of tumor density as the number of cancer cell nuclei per tumor area in coronal slices. n = 4 per group with three sections through each tumor as shown in (A). ***P < 0.001, t test. (D) Quantification of tumor cells beyond borders as shown in (B). n = 4 per group on three sections through each tumor. ***P < 0.001, t test. (E) H&E-stained tumor borders in the same untreated and treated tumors as shown in (B). Scale bar, 50 μm.

Combination therapy with DXR

To determine IB’s efficacy as an adjuvant anti-invasive therapy, we combined nano-IB sequentially with liposomal DXR (nano-DXR) treatment. DXR has been delivered in liposomal formulations to minimize toxicity, showing benefits to survival in brain tumors in both animals and humans and a pharmacokinetic profile similar to that of our nano-IB (15, 17, 23). Treatment of RT2 glioma cells in vitro was performed at a range of DXR concentrations (50 nM to 200 μM) and indicated no additive effect (fig. S6).

Adult male Fischer rats were inoculated with RT2 glioma, and MRI was performed 7 days later to confirm tumor presence. Four groups were used in this study: control (saline liposomes), IB (nano-IB), DXR (nano-DXR), and IB+DXR (nano-IB followed by nano-DXR). Figure 6A shows the liposome treatment schedule. All groups received two injections of liposomes on days 4 and 7 after inoculation to account for increased circulation time due to multiple injections (24). Lipid load was the same for all groups (55 μmol/kg) to prevent variability in circulation time and accumulation at tumor attributed to decreased reticuloendothelial clearance.

Fig. 6

Sequential delivery of nano-IB and liposomal DXR (nano-DXR) results in increased survival. (A) Groups for survival study with treatment regimens as days after tumor inoculation: Control (saline liposomes; n = 6), IB (nano-IB, 3 mg/kg, on day 7; n = 6), DXR (nano-DXR on day 7; n = 6), IB+DXR (nano-IB, 3 mg/kg on day 4 followed by nano-DXR, 7 mg/kg on day 7; n = 5). (B) Kaplan-Meier survival plots of treatment groups. Mantel-Cox test comparing IB+DXR to DXR alone. **P < 0.01. Both of these groups were significantly different from control and IB groups (P < 0.001). (C) T2-weighted MR images from DXR and IB+DXR groups at the injection site on day 7 after inoculation (day 1 of treatment) and subsequently throughout the study until day 180. Fluid buildup (F), differentially dense tissue (T), and dark masses (M) are indicated on the MRI at day 180. Scale bar, 1 mm. (D) MR images were analyzed for growth by outlining the tumor on 1-mm slices through tumor-bearing rat brains and calculating the volume (mm3). Data points represent the means ± SEM, and the number of animals per group varies with each time point due to death of animals as shown in (B). (E) H&E-stained sections of tumors from DXR and IB+DXR groups at day 200 at the tumor site. Hemosiderin-laden macrophages are indicated by arrowheads. Scale bar, 50 μm.

We observed 100% survival in the treatment group that received nano-IB followed by nano-DXR (IB+DXR) for a period of 200 days, at which point the study was concluded (Fig. 6B). This effect was significant compared to all other treatment groups. Only 33% of the animals treated with nano-DXR alone (DXR) were surviving at day 200, with a median survival time of 44.5 days. Animals treated with saline liposomes (Control) and nano-IB alone (IB) had a median survival time of 19 and 18 days, respectively (not statistically significant). MRI was performed throughout the study to compare treatment response (Fig. 6D). Volume measurements were performed on MR images and yielded growth curves showing reduction in tumor in IB+DXR animals compared to the other treatment groups (Fig. 6C). Regrowth in the DXR group at day 180 was indicated on MRI by increased fluid in the brain and differentially dense tissue, which was not seen in the IB+DXR group (Fig. 6C).

To reveal tumor regrowth, we injected surviving animals with Gd-DTPA and imaged them with MR at 180 days (fig. S7, A and B). There was contrast enhancement seen in the DXR group (n = 2), but not in the IB+DXR group (n = 5) (fig. S7, C and D), indicating blood-brain barrier breach, often seen in patients as tumors grow (25). To further confirm features on MRI, we killed surviving rats at 200 days and processed the brains for H&E staining (Fig. 6E). Histological analysis of brain tissue indicated the presence of hemosiderin-laden macrophages at the original tumor site, creating the darkened mass seen on MRI. Macrophages were confirmed by immunohistochemistry (fig. S8). Furthermore, nano-IB + nano-DXR–treated brains showed no presence of tumor cells, whereas brains treated with nano-DXR alone showed abnormal tissue in the contralateral hemisphere and the presence of glioma cells (Fig. 6E and fig. S8).

Mechanism of IB action on glioma

RT2 cells treated in vitro with free IB showed reduction in actin fiber formation after 24 hours (Fig. 7A). Further studies on mechanism of action were conducted in vivo with the RT2 animal model. For all studies, rats were treated with nano-IB or blank liposomes on days 4 and 7 after tumor inoculation. Tumors were removed for analysis on day 11. A Rat 230 2.0 Affymetrix microarray revealed that the nano-IB–treated tumors had about 45 genes with a twofold or greater change in expression compared to untreated tumors (table S1).

Fig. 7

Mechanism of action of IB determined in vivo and in vitro. (A) RT2 glioma cells were plated on laminin-coated glass slides and treated with ethanol (untreated) or 5 μM IB. After 24 hours, cells were stained for F-actin (red, phalloidin) and nuclei (blue, DAPI). Scale bar, 10 μm. (B) IPA of microarray data from in vivo tumor samples (n = 3 per group), treated with nano-IB (3 mg/kg) on days 4 and 7. Tumor was harvested on day 11. Red, up-regulation; green, down-regulation. Greater than twofold change in response to treatment is represented. Dotted lines show indirect relationships; solid lines indicate direct protein-protein interactions. (C) Actin regulatory elements were analyzed with RT-PCR on pooled mRNA (n = 3 per treatment group, nano-IB or saline liposomes) compared to control housekeeping gene HPRT. (D) Western blots of pooled protein samples from in vivo tumors (n = 3 per group, nano-IB or saline liposomes). (E) Integrated density analysis of Western blots in (D), normalized to GAPDH. (F) H2O2 generation in Nox4-overexpressing COS cells with increasing dose of free IB after 1 hour in vitro. Data are means ± SEM. *P < 0.05 compared to 0 μM IB, t test. (G) Representative images of Src-3T3 cells treated with 6 μM IB or ethanol carrier. Cells were stained for F-actin with Alexa Fluor 568–phalloidin to view rosettes. Scale bar, 50 μm. Dose-dependent response of rosette formation (bottom) in Src-3T3 cells to free IB. Percent of cells with complete rosettes was calculated for each condition. Data are averages ± SEM analyzed by ANOVA with post hoc t tests. *P < 0.05; **P < 0.01. (H) Proposed mechanism of action of IB’s inhibition of invasion of cancer cells. Dotted lines are indirect effects and solid lines indicate direct effects. The Nox4 complex also contains the subunit p22 not specifically investigated in our studies.

Using Ingenuity Pathway Analysis (IPA), we determined relationships among genes showing more than twofold changes (Fig. 7B). One subset of genes implicated in actin polymerization was identified, including those that encode profilin-1 (pfn1, −2.342-fold change), scinderin (scin, +2.891-fold change), α-actin (acta1, −2.379-fold change), calgranulin (S100A8, −2.101-fold change), and RhoGDP dissociation inhibitor α (arhgdia, +3.205). Using reverse transcription–polymerase chain reaction (RT-PCR) on pooled mRNA samples from RT2 tumors IB-treated or untreated in vivo, we confirmed these changes quantitatively (Fig. 7C). Western blotting on pooled protein samples harvested from tumors in rats confirmed the expression level changes seen at the transcriptional level, with the exception of only a minimal decrease in profilin-1 (Fig. 7, D and E).

Upstream regulators involved in IB mechanism of action

Because compounds similar to IB have been implicated in inhibition of NADPH oxidases, we examined the ability of IB to inhibit NADPH oxidase 4 (Nox4). Free IB was applied at several concentrations to human Nox4–overexpressing COS cells in vitro as a model system of human Nox4 activity. Administration of IB led to a decrease in the release of hydrogen peroxide (H2O2), which is indicative of Nox4 inhibition (Fig. 7F). Further, Nox4 has been shown to be involved in invadopodia formation via Src kinase activation through Tks5, which leads to increased invasion in cancer cells (18). In activated Src-transformed 3T3 fibroblasts—cells that have high numbers of invadopodia and high levels of invasion—IB inhibited invadopodia formation, as indicated by disassembly of actin rosettes (Fig. 7G). Along with this, IB specifically reduced phosphorylation of the Src kinase substrate Tks5, known to be necessary for Src-activated invadopodia formation, while leaving phosphorylation levels of cortactin control protein unchanged (fig. S9). These data suggest IB’s involvement in the inhibition of the Nox4-Src axis, leading to alterations in actin regulation and invasion inhibition. The mechanism of action as determined through microarray, Western blot analysis, and in vitro signaling studies is summarized in Fig. 7H.


Tumor cell invasion, manifested by both local and distal recurrence, is a major cause of death from GBM. Current therapies for GBM reduce tumor mass but do not prevent invasion; contrarily, such therapies may actually enhance invasion by inducing hypoxia and downstream migratory pathways (4, 6). Here, we demonstrated that the small-molecule IB inhibited invasion of rat and human glioma cells in vitro. Liposomally encapsulated IB localized to tumors in an aggressive in vivo model of GBM, inhibited diffuse spread, and enhanced efficacy when combined with chemotherapy—an effect seen in vivo but not in vitro. We showed that IB acts via a putative Nox4/Src-dependent mechanism to disturb actin fiber formation. Through development of a delivery system with a new compound, nano-IB, we could better elucidate the mechanism and ability of IB to inhibit invasion in vivo to develop a therapeutic that will translate to humans.

Diffuse tumors limit safe chemo- and radiotherapeutic dosing, whereas resection of more contained tumors has a better postsurgical prognosis and longer survival times (26). Invasion can lead to resistance to therapy. Lefranc et al. proposed that invasion of GBM is tied to chemotherapeutic efficacy (27), noting several pathways involved in invasion, including nuclear factor κB (NF-κB) and phosphatidylinositol 3-kinase (PI3K) both potentially affected by IB according to pathway analysis. On the basis of these hypotheses, they have demonstrated enhanced activity of temozolomide with cimetidine (a fucose expression inhibitor) and 4-(N-benzylpiperidin-4-yl)-4-iodobenzamide (a σ1 agonist) in rodents (28, 29). Both the U373 glioma and the 9L gliosarcoma models in these studies harbored mutant p53. Here, we extended these findings to a GBM cell line (RT2) that has wild-type p53 (30). Mutations in p53 are present in less than 10% of primary GBM patients but are common in secondary GBM (31). Therefore, exploration of anti-invasive compounds affecting alternative mutation pathways has the potential to positively affect a new group of patients.

The ability of IB to enhance chemotherapy with DXR in vivo is significant. Free DXR has high levels of cardiotoxicity associated with its use clinically (32). However, in its liposomal formulation, DXR shows long-term stabilization in patients with malignant recurrent glioma (16, 17, 23). Animals receiving a single sequential treatment of nano-IB and nano-DXR survived for more than 6 months with no observable tumor mass compared to nano-DXR–treated animals. Other studies have shown comparable survival in mice bearing U87MG or U251 human glioma and rats bearing 9L gliosarcoma with multiple or prolonged treatments with temozolomide, but not in a single round of treatment (7, 33). We posit that the combination of IB with any antitumor treatment will enhance the outcomes in patients owing to the primary ability of IB to prevent invasion. Such therapy would be useful upon diagnosis before antitumor treatment begins, to create a more treatable tumor (as we saw in our in vivo studies), to prevent an invasive phenotype after treatment, and to potentially sensitize the tumor to chemotherapy through reactive oxygen species (ROS)–dependent down-regulation of NF-κB (34). MRI and histology revealed an absence of tumor in rats treated with both nano-IB and nano-DXR, whereas tumors treated with nano-DXR alone showed contrast enhancement. This enhancement is characteristic of blood-brain barrier breach that occurs during tumor regrowth in patients (25). This result indicated tumor regrowth, which was further confirmed by histology. These data—combined with IB’s efficacy against several glioma lines in vitro, including stem cell–rich neurospheres (35)—is suggestive of its potential for affecting patients with GBM.

The ability of IB to affect patients may be better understood by understanding the putative mechanism of action. IB acts upstream to actin polymerization, likely through the Nox4-Src axis. In vivo, IB causes transcription-level alterations of commonly mutated genes (scin, acta1, and arhgdia) in cancers that lead to decreased cytoskeletal stability and heightened motility (36). Altered expression of actin-related genes is observed in many human cancers, including glioma (36). α-Actin, scinderin, and calgranulin are downstream targets of RhoGTPases, which are inhibited by RhoGDP dissociation inhibitor α (RhoGDIα) (37). Decreased RhoGDIα expression has been correlated with increased malignancy in human brain tumors (38). Scinderin, a protein that severs and caps actin filaments (39), was the most highly up-regulated in this study, suggesting its role in directly decreasing the RT2 cell invasiveness in vivo. Pathway analysis further revealed that IB may modulate other signaling networks upstream of these actin regulatory elements, including major players in cancer progression and metastasis: NF-κB, PI3K, and protein kinase C (PKC).

Triphenylmethane dyes, such as IB, also inhibit ROS production by the NADPH oxidases and inactivate Src kinase (18, 19). ROS and Src activation have been implicated in human cancer cell migration (40). In human gliomas, Src kinase activity (41), Nox4 activity (42), and NF-κB activation (43) have been correlated with grade and progression of disease. Therefore, IB’s ability to inhibit these molecules has clear therapeutic potential.

The ability of nano-IB to inhibit invasion in an aggressively invasive tumor and, when combined with nano-DXR, to significantly enhance survival from a median of ~19 to 200 days in a rat model is unprecedented. Current clinical trials with more aggressive chemotherapeutics, such as DXR (16, 17), indicate that adjuvant therapy with nano-IB may be beneficial. With patients dying within 2 years (1, 3), the treatment of GBM may benefit from a paradigm shift in this direction. However, it is important to recognize that human gliomas can be diverse, and although our in vitro data suggest that IB inhibits invasion of several glioma lines, its broad ability to inhibit several tumor types has yet to be demonstrated in vivo. This said, it is also possible that the ability of IB to halt gliomas has implications to treat metastatic tumors that are not gliomas; this early promise warrants further exploration. The path the translation to human therapy of IB involves testing the generalizability of IB’s effects in human glioma in vivo and transgenic animal models, followed by scale-up and manufacturaing of nano-IB formulations for downstream approval (U.S. FDA) and clinical testing.

Materials and Methods

Cell culture and in vitro assays

RT2 and eGFP-RT2 rat glioma cell lines were a gift from H. Fillmore (Virginia Commonwealth University). U87MG human glioma and C6 rat glioma cell lines were purchased from the American Type Culture Collection. The 9L rat gliosarcoma cell line was a donation from the Neurosurgery Tissue Bank (University of California, San Francisco). All cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Mediatech) supplemented with fetal bovine serum (FBS) (Gemini Bio Products), and eGFP-RT2 cells were maintained with G418 (Gemini) for selection. Primary astrocytes were obtained from rat pups and used below passage 5. Invasion assays were performed with Basement Membrane Extract (0.5 mg/ml) (Trevigen)–coated (glioma) or bare (astrocytes) 8-μm pore size tissue culture inserts (Millipore) over 24 hours (glioma) or 48 hours (astrocytes) up an FBS gradient. Cells migrating through were identified by staining with DAPI (Invitrogen), counting five representative fields per insert, and reported as total cells invaded/total cells seeded × 100 (%) for each insert. Corresponding cell counts per field are given in the Supplementary Materials for the individual cell lines (table S2).

Primary neurospheres were derived from two patients (N08-1002 and N08-32) and maintained in cell culture conditions as described (35). Neurospheres (>100 μm) were embedded in 1.8% Type I Collagen (BD Biosciences) with either 10 μM IB or ethanol and fixed with 4% paraformaldehyde (Sigma) after 48 hours, stained with rhodamine-phalloidin (Invitrogen), and imaged. Invasion assays were conducted using tissue culture inserts (above) with 100 μl of collagen gel and dissociated neurospheres (single cells). Cells migrated up an FBS gradient for 48 hours and analyzed for percent invasion.

Liposome synthesis

IB was synthesized as described in the Supplementary Methods. Nano-IB was made by dissolving 10 mole percent (mol %) cholesterol (Sigma), 85 mol % distearoylphophatidylcholine (DSPC, Avanti Polar Lipids), 5 mol % PEG 2000 [poly(ethylene glycol), molecular weight 2000]–distearoylphosphatidylethanolamine (DSPE), 0.01 mol % DSPE-rhodamine (Genzyme), and IB (0.9 mg/ml) in ethanol at 70°C and hydrating with phosphate-buffered saline (PBS). The solution was extruded to ~160 nm as assessed by dynamic light scattering. Sepharose (GE Healthcare) column separation was performed and solution was diafiltrated to a final concentration of 1.6 ± 0.08 mg of IB per milliliter in 140 mM lipids, as assessed by absorbance at 610 nm (IB), and fluorescence at 485/610 excitation/emission (lipid). DXR nanocarriers (nano-DXR) were made according to (14), with 65:30:5 mol % DSPC/cholesterol/DSPE-PEG and extruded to ~160 nm.

Pharmacokinetic studies

All animal studies were performed in accordance with the Georgia Institute of Technology Institutional Animal Care and Use Committee (IACUC) using male Fischer 344 rats (Harlan). Free IB in saline/cremophor EL (Sigma)/ethanol [5:5:90 (v/v)] (n = 4) or nano-IB (3 mg/kg) (n = 4) was injected into the tail vein. Blood was collected orbitally at time points (free: 0, 0.25, 0.5, 0.75, 1, 2, and 3 hours; nano-IB: 0, 0.5, 1, 8, 24, and 56 hours). Plasma was collected and either measured directly (for free IB) or first lysed with 1% Triton X-100 (Sigma) solution (for nano-IB) before measurement at 610 nm using a plate reader (Bio-Tek).

Invasion and survival studies

eGFP-RT2 cells [250,000 in 10 μl of Leibovitz Media (Mediatech)] were injected 2 mm lateral and anterior to λ at a depth of 3 mm. On days 4 and 7 after inoculation, nano-IB (3 mg/kg) (nano-IB–treated, n = 4) or saline liposomes (untreated, n = 4) were delivered via the tail vein. On day 11, animals were sacrificed via intracardial perfusion with saline for histology. Brains were sectioned with a Leica cryostat at 16 μm.

For survival studies, at 4 days after implantation, rats received either nano-IB (3 mg/kg) or saline liposomes via tail vein injection. On day 7, MRI was performed followed by tail vein injection of nano-DXR (7 mg/kg), saline liposomes, or nano-IB (3 mg/kg). Animals displaying criteria for euthanasia in accordance with Georgia Institute of Technology IACUC guidelines were sacrificed by carbon dioxide inhalation. The brain was removed to confirm tumor presence by histology.

Immunohistochemistry and image analysis

Sections were stained with anti–neurofilament 160 antibody (Sigma) followed by secondary Alexa Fluor 594, Alexa Fluor 488–conjugated anti-GFP antibody, and DAPI (Invitrogen). Brain slices were imaged using wide-field microscopy with Neurolucida (MBF Biosciences). Images for margin quantification were obtained with a Zeiss LSM 510 confocal microscope. ImageJ (National Institutes of Health) was used for tumor density calculations. Invasion beyond borders was assessed on three sections through each tumor at four locations on the border. Border definition was determined by two independent parties on separate occasions on the basis of red-green spatial definition, and cells beyond border were counted (confirmed with GFP and DAPI).

Microarray analysis and quantitative RT-PCR

Rats were inoculated with RT2, treated with nano-IB on days 4 and 7 (3 mg/kg per injection), and perfused with ribonuclease-free saline 11 days after inoculation (n = 3 per group, treated and untreated). Tumors were removed and frozen. RNA was harvested with the Qiagen RNeasy kit (Millipore). Total RNA was quantified with an Agilent Bioanalyzer and converted to complementary DNA (cDNA) with SuperScript Choice System (Invitrogen). The microarray and data analysis was performed as described (44) with the Affymetrix Rat 230 2.0 gene array, with changes greater than twofold being significant. Data were analyzed with IPA (Ingenuity Systems). For RT-PCR, RNA was pooled per condition and converted to cDNA. Quantitative RT-PCR was performed with SYBR Green reagents and the StepOne Plus Analyzer (Applied Biosystems). The following primer pairs were used (Integrated DNA Technologies): acta1, 5′-GGCAAGCCAGCTTTCACC-3′ (forward) and 5′-CCTCAGGACGACAATCGACC-3′ (reverse); pfn1, 5′-GGAAGACCTTCGTTAGCATTACG-3′ (forward) and 5′-CGTCTTGCAGCAGTGAGTCC-3′ (reverse); S100A8, 5′-GCAATTAACTTCGAAGAGTTCC-3′ (forward) and 5′-CGACTTTATTCTGTAGACATATCC-3′ (reverse); arhgdia, 5′-CTACATGGTCGGGAGCTATGG-3′ (forward) and 5′-GTTGTAACTGCCCCGAGCAA-3′ (reverse); scin, 5′-TTCTCCTACCGCCTGCACTT-3′ (forward) and 5′-CGCCCAAATAGTCGTCCATCT-3′ (reverse); and HPRT (housekeeping), 5′-TGTTTGTGTCATCAGCGAAAGTG-3′ (forward) and 5′-CTGCTAGTTCTTTACTGGCCACATC-3′ (reverse). Comparison of fold change was determined by the ΔΔCT method.

Western blotting

Protein lysates were harvested from treated and untreated tumors (n = 3 per group) and pooled. Total protein was measured with BCA assay (Bio-Rad). Equivalent total protein amounts were run on a NuView (NuSep) gel followed by transfer to polyvinylidene difluoride membrane and incubation with rabbit anti-acta1, rabbit anti-scinderin, rabbit anti–profilin-1 (Abcam), goat anti-S100A8, or rabbit anti-arhgdia (Santa Cruz Biotechnology). Secondary incubation was performed with horseradish peroxidase (HRP)–tagged anti-goat (Santa Cruz) or HRP-tagged anti-rabbit (Invitrogen). Membranes were reprobed with rabbit anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (Abcam) for normalization of signal. Integrated density measurements were performed with ImageJ.

Nox4 activity assay

Homovanillic acid (HVA) assays were performed with Nox4-overexpressing COS cells, as described previously (45). Briefly, cells were treated with IB for 1 hour and then incubated with HVA solution [100 μM HVA, HRP (4 U/ml) in PBS] at 37°C for 1 hour. HVA stop buffer [0.1 M glycine, 0.1 M NaOH, 25 mM EDTA in PBS (pH 7.4)] was then added, and fluorescence was read on a luminescence spectrometer (Perkin Elmer) (excitation/emission, 312/450 nm).

Magnetic resonance imaging

Animals were anesthetized and placed in a 7-T MRI (Bruker Instruments) equipped with a 30-mm head coil. A T2-weighted image was taken through the head with the following parameters: repetition time (TR) = 2.0 s, echo time (TE) = 48 ms, field of view (FOV) = 40 mm × 40 mm with a 256 × 256 matrix, slice thickness = 1 mm, number of slices = 20, two averages per phase-encode step requiring a total acquisition time of about 6 min per rat. For T1-weighted MRI, animals were treated with either Gd-loaded nano-IB (for accumulation, 3 mg/kg) or free Gd-DTPA (Omniscan, GE Healthcare) for survival study (final concentration of Gd was 0.15 mmol/kg). T1 images were taken according to the following parameters: TR = 2.0 s, TE = 48 ms, FOV = 40 mm × 40 mm with a 256 × 256 matrix, slice thickness = 1 mm, number of slices = 20, two averages per phase-encode step requiring a total acquisition time of about 6 min per rat.

Statistical analysis

For in vitro assays, a one-way analysis of variance (ANOVA) was performed followed by Tukey post t tests for individual pair comparisons; n indicates an independent experiment (with three replicates per condition). t tests are independent sample t tests with Bonferroni adjustment of α when multiple comparisons are made to a single sample (that is, control comparisons). For in vivo invasion studies, results were compared by independent sample t test of averages for each animal (given n). For in vivo pharmacokinetic studies, points were fitted with a single-phase exponential decay curve and compared using nonlinear three-point regression (comparison of fits analysis). For the survival study, we used Mantel-Cox analysis to compare survival curves. In general, graphs depict mean ± SE with relevant P values. P < 0.05 was considered significant, and where applicable, P < 0.01 and P < 0.001 are denoted. Graphs and analyses were completed with GraphPad Prism software.

Supplementary Materials


Fig. S1. Imipramine Blue is a novel anti-invasive small molecule.

Fig. S2. IB is not cytotoxic to RT2 cells or astrocytes at effective anti-invasive doses.

Fig. S3. Nano-IB shows no toxicity to vital organs.

Fig. S4. Histological features of RT2 rat astrocytoma.

Fig. S5. Nano-IB treatment enhances survival in RT2 tumor–bearing rats.

Fig. S6. IB combined with doxorubicin does not enhance cytotoxicity in vitro.

Fig. S7. DXR-treated tumors show MRI contrast enhancement.

Fig. S8. Macrophage staining in long-term surviving animals.

Fig. S9. IB inhibits Tks5 phosphorylation.

Table S1. Fold change in gene expression from in vivo RT2 tumor microarrays.

Table S2. Cell counts from invasion assays by cell type.

Video S1. IB inhibits RT2 glioma cell migration in vivo.

References and Notes

  1. Acknowledgments: We thank P. Santangelo, B. Pai, K. McNeeley, A. Ortiz, C. Tucker-Burden, S. Alkindi, S. Nemani, M. Ahmad, E. de Hitta, and C. Calderwood. Funding: Supported by Georgia Cancer Coalition, Ian’s Friends Foundation (to R.V.B.), the Wallace H. Coulter Translational Research Grant Program (to R.V.B. and J.L.A.), and the NSF Graduate Research Fellowship (to J.M.M.). Author contributions: J.M.M. and R.V.B. wrote the paper. J.M.M. completed all experiments but those specifically listed here. L.F., M.Y.B., and J.L.A. synthesized IB. U.G.K. completed Nox4 studies. S.A.C. and B.D. completed Src and Tks5 studies. L.K. and J.M.M. analyzed microarray data. S.A.R. and J.M.M. completed neurosphere studies. D.J.B. served as pathologist and assisted with the manuscript. R.V.B. and J.L.A. provided funding. Competing interests: J.M.M., R.V.B., and J.L.A. hold a patent on nano-IB (PCT/US2010/031914). J.L.A. has a pending patent on IB (application 20090176745).
View Abstract

Navigate This Article