Gallium Maltolate Disrupts Tumor Iron Metabolism and Retards the Growth of Glioblastoma by Inhibiting Mitochondrial Function and Ribonucleotide Reductase

Glioblastoma is a primary brain tumor with a dire prognosis. Despite treatment, the median survival of patients with this disease is 14.6 months; few patients survive beyond 2 years from diagnosis (1). The need to develop new therapies for this malignancy is obvious. In this regard, drugs directed at disrupting pathways involved in tumor growth are emerging (2). Recent evidence indicates that iron metabolism and iron-dependent tumor growth are promising targets for cancer treatment as tumor cells have an increased demand for iron to support ribonucleotide reductase activity and mitochondrial function (3). Moreover, iron is required for the activity of certain cyclins and for signaling through the mTOR and WNT pathways (4). A change in the balance of proteins that regulate the cellular intake, storage, and export of iron [transferrin receptor (TfR1), ferritin, and ferroportin, and hepcidin, respectively] in tumors leads to an expanded intracellular iron pool to support iron-dependent malignant cell growth (5). It has been shown that iron homeostasis and iron transport are altered in brain tumors relative to nonmalignant cells (6, 7). Iron, as Tf-Fe, enters the brain by TfRs present on the luminal surface of brain microvascular endothelial cells (BMEC) of the blood–brain barrier (BBB; ref. 8).

Gallium nitrate, a simple metal salt that targets iron metabolism, has clinical antineoplastic activity in bladder cancer and lymphoma (9). It shares certain chemical properties with iron that enables its binding to transferrin (Tf), the transport protein for iron in the circulation (10). Gallium enters cells via TfR1-mediated endocytosis and blocks TfR1-mediated uptake of Tf iron by cells (11). Within the cell, gallium disrupts iron-dependent tumor growth and induces cell death (11).

The antineoplastic activity of gallium nitrate has prompted the development of newer gallium compounds with complex ligand structures. These agents hold the promise of greater clinical efficacy and fewer side-effects (9). For example, gallium maltolate [(tris-hydroxy-2-methyl-4H-pyran-4-onato)gallium] (GaM) displays greater cytotoxicity than gallium nitrate in lymphoma cell lines and inhibits the growth of lymphoma cells that are resistant to the cytotoxicity of gallium nitrate (12). GaM also inhibits the growth of human T-cell lymphoma xenografts in nude mice confirming its antitumor activity both in vitro and in vivo (13).

Glioblastomas have a high requirement for iron (14). As iron is taken up by cells through TfR1-mediated endocytosis and TfRs are highly expressed on the surface of glioblastoma cells (15), targeting cellular iron metabolism represents an attractive interventional strategy in glioblastoma therapy. Considering that gallium uses the transport and cellular uptake systems for iron, we hypothesized that GaM can cross the BBB via TfRs on BMECs and target TfR-bearing glioblastoma cells. In this study, we show that GaM inhibits glioblastoma cell growth in vitro and in vivo. GaM's mechanisms of action include inhibition of cellular iron uptake, disruption of mitochondrial function, and inhibition of RRM2 activity in glioblastoma cells. These findings open the door for further development of gallium-based compounds for glioblastoma treatment.

GaM was provided by Titan Pharmaceuticals. Mouse anti-human TfR antibody (anti-CD71) and rabbit anti-rat TfR antibody were from Biogenex Laboratories, and ABBIOTEC, respectively. Antibodies to RRM2, H- and L-ferritin, and TfR1 were purchased from Santa Cruz Biotechnology Inc. Human Tf, 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), oligomycin, carbonilcyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), and antimycin A were obtained from Sigma Chemical Company. Alzet mini-pumps were obtained from Durect Corporation. ¹²⁵I-Na and ⁵⁵FeCl₃ were purchased from Perkin Elmer and ¹²⁵I-Tf and ⁵⁵Fe-Tf were prepared as previously described (16).

Tissue culture media and supplements were purchased from Life Technologies, unless stated otherwise. All cell lines used were validated at their point of origin. Human glioblastoma U-87 MG and D54 cell lines were obtained from ATCC and courtesy of Dr. D. Bigner (Duke University Medical Center, Durham, NC), respectively. The U-87 MG cells were grown in MEM with Earle's salts fortified with 10% FBS, and supplemented with 1% sodium pyruvate and 0.1% gentamicin. D54 cells were grown in improved MEM with Zn option, fortified with 10% FBS, and supplemented with 0.1 % gentamicin. Human BMECs were a generous gift from Dr. Daniel Kosman (University of Buffalo, Buffalo, NY) and have been described previously (17). The glioblastoma stem cells (GSC) were developed from human glioblastoma and were authenticated, as described previously (18). Experiments were conducted with GSC lines designated GSC-22, GSC-33, and GSC-44. These cells were maintained as neurospheres in a serum-free stem cell culture medium (18, 19).

The interaction of Tf with GaM was examined by UV–Vis spectroscopy, as described by Harris and Pecoraro (10). Individual absorbance spectra of human Tf (12.5 μmol/L) and GaM (25 μmol/L) in water were obtained using a Varioskan Flash Spectral Scanning Multimode Reader (Thermo Fisher Scientific Inc.). GaM was added incrementally to the Tf-containing cuvette to achieve final Ga concentrations of 25, 50, and 100 μmol/L, respectively. Absorbance spectra were obtained at room temperature for 10 minutes after the addition of each concentration of GaM to the cuvette.

²⁵I-Tf–specific binding to intact cells was measured as reported previously (20). For Tf internalization kinetics, cells were first incubated at 4°C to allow for ¹²⁵I-Tf binding to cell surface TfR1 and then washed to remove unbound ¹²⁵I-Tf. Warm medium (37°C) was added to the cells to initiate TfR1 cycling and the fraction of acid-resistant, cell-associated ¹²⁵I-Tf (representing internalized ¹²⁵I-Tf) was measured at different time points.

The effect of GaM on cell proliferation was measured by MTT cytotoxicity assay in 96-well microwell plates using an ELX 808 Ultra Microplate Autoreader (Biotech Instruments; ref. 12).

The effect of GaM on cellular bioenergetic function was assessed by measuring the oxygen consumption rate (OCR, a measure of oxidative phosphorylation) in intact cells using a Seahorse 96XF Analyzer (Agilent Technologies), according to the manufacturer's directions. Seahorse XF analyzer methodology is reviewed by Dranka and colleagues (21). D54 cells were plated in fresh medium in a 96-well plate (10⁴ cells per well) and incubated without additives at 37°C in a CO₂ incubator. After 24 hours of incubation, various concentrations of GaM were added to the wells and the incubation continued for an additional 24 hours. The plate was transferred to an XF Analyzer and basal OCR was measured at three timepoints. This was followed by sequential additions of oligomycin (1μg/mL), FCCP (1μmol/L), and antimycin A (10 μmol/L).

X-band electron paramagnetic resonance spectroscopy (EPR) spectra of U-87 MG cells incubated without or with 100 μmol/L GaM for 24 hours were obtained at 110 K with a Bruker EMX spectrometer located at the Nation Biomedical EPR Center at the Medical College of Wisconsin (Milwaukee, WI). EPR spectra were collected as described previously by us (22).

TfR1, H-ferritin, L-ferritin, and RRM2 expression in normal brain and glioblastoma, and TfR1 expression in human brain microvascular endothelial cells were examined by IHC staining of tissue samples from surgically resected glioblastoma tumors from patients or from animal experiments. Dead cells in tumor xenografts in GaM-treated animals were identified according to the typical nuclear morphologic changes: pyknosis (nuclear condensation), karyorhexis (nuclear fragmentation), and karyolysis (complete dissolution of the nuclear fragments). Percentage of dead cells was calculated out of a total of 1,000 cells (viable and dead) counted. Human tissue was obtained from the Brain and Spinal Cord Tissue Bank of the Medical College of Wisconsin (Milwaukee, WI). IHC staining with specific primary antibodies was performed on a Dako Autostainer Plus Instrument using the Dako EnVision FLEX High pH Detection Kit protocol. Stained slides were visualized using a Nikon Eclipse 80i microscope equipped with a MicroPublisher 3.3 RTV color video camera (Q Imaging). The images were captured using NIS elements imaging software (Version 7.0, Nikon Instruments, Inc.).

Glioblastoma cells were analyzed for the expression of H-ferritin, L-ferritin, and RRM2 proteins by Western blotting using standard protocols. Protein bands on membranes were identified by primary antibodies followed by horseradish peroxidase–labeled secondary antibody. Membranes were developed in Enhanced Chemiluminescence Western blotting detection solution (Amersham Life Science) and exposed to BioMax film for autoradiography.

The antineoplastic activity of GaM in vivo was examined in an intracranial U-87 MG xenograft rat model described previously by us (23). All protocols were approved by the Institutional Animal Care and Use Committee at the Medical College of Wisconsin (Milwaukee, WI).

Male athymic rats weighing approximately 250 g were anesthetized with an intraperitoneal injection of ketamine (60 mg/kg), acepromazine (0.9 mg/kg), and xylazine (6 mg/kg). Once appropriate anesthetic depth was ascertained, the head was immobilized in a stereotactic device. A 2.5-cm skin incision was made along the midline over the bregma and a 1-mm burr hole was drilled in the skull 1 mm anterior and 2 mm lateral to the bregma on the right side (24). Using a 10-μL gas-tight syringe (Hamilton Company), 2 × 10⁵ U-87 MG cells were implanted into the right frontal lobe at a depth of 3 mm relative to the dural surface. The cells were continuously injected over 5 minutes, after which the needle was left stationary for 5 minutes and then slowly withdrawn over an additional 5 minutes. Afterwards, the skin was closed using 3M Vetbond Tissue Adhesive (3 M Animal Care Products).

Eight days following implantation of tumor cells, and after baseline MRI studies were performed, rats were anesthetized with 2% isoflurane. Using aseptic techniques, a small incision was made over the animal's neck and sharp dissection carried out down to the level of the jugular vein. The vein was skeletonized and each pump's drug release catheter was inserted and secured within the vessel lumen. A subcutaneous pocket was dissected free to hold the implanted Alzet minipump reservoir (Durect Corporation) and then the skin wound closed with sutures.

In clinical trials of gallium nitrate, the greatest antineoplastic activity and least toxicity were seen when the drug was administered by continuous intravenous infusion over 5–7 consecutive days (25). On the basis of this treatment schedule, GaM solution or saline control was administered intravenously through the subcutaneously implanted Alzet minipump to provide a steady delivery of GaM. A dose of 50 mg/kg/day was chosen because this was shown to be well tolerated without significant toxicity in rats (26).

Tumor response after 10 days of GaM therapy was assessed by MRI. The relative change in tumor size on MRI from initiation to completion of GaM treatment measured using RECIST V1.1 criteria, as used in the clinic (27). The change in cerebral blood volume (CBV) was measured as described previously (23).

Analysis was conducted on the change in tumor size and CBV (day 18–day 8) between the control group and gallium-treated group of animals with a level of significance as P = 0.05 (Mann–Whitney test).

The effect of GaM on iron uptake by U87 and D54 cells was measured using ⁵⁵FeTf, as described previously (28).

GaM is composed of three maltolate ligands bound to a central gallium atom in a propeller-like arrangement (Fig. 1A; ref. 29). At equimolar gallium concentrations, GaM proved to be more cytotoxic to D54 glioblastoma cells than gallium nitrate (Fig. 1B). This suggests that, in contrast to gallium nitrate, lower concentrations of gallium as GaM are needed to inhibit tumor growth. Accordingly, GaM would be expected to display less toxicity to normal cells than gallium nitrate.

Early studies in animals confirmed that ⁶⁷Ga citrate injected intravenously was bound entirely to Tf in the circulation (30) and that ⁶⁷Ga uptake by cells occurred primarily through cell surface TfR1-mediated uptake of Tf-Ga (31). As the binding of gallium to Tf alters the UV-Vis spectra for Tf (10), we examined whether GaM interacted with Tf by measuring changes in the individual UV-Vis spectra of Tf and GaM, and Tf mixed with GaM. As shown in Fig. 1C, a progressive spectral shift was seen when GaM was added to Tf at different ligand to metal ratios, indicating an interaction between GaM and Tf. Whereas the major shift in the spectral peak suggestive of GaM–Tf interaction was seen at ≥280 nm, an increase in the absorbance of Tf at 240–250 nm was also noted when it was incubated with GaM. The latter may be due to the binding of gallium (independent of GaM) to Tf, as the formation of Tf-Ga is known to produce an increase in the absorbance of Tf at 242 nm (10). Hence, it is possible that in solution a small amount of gallium may dissociate from GaM to bind Tf.

As Tf-Ga is preferentially taken up by TfR-bearing cells (11), TfR1 expression on these cells and on BMECs was assessed by ¹²⁵I-Tf-TfR binding assay. As shown in Fig. 1D, both D54 and U-87 MG cells expressed higher levels of specific Tf binding relative to BMECs, which displayed 63%–65% lower Tf binding. These results suggest that TfRs on glioblastoma cells can be preferentially targeted by Tf-Ga.

To mimic the treatment schedule of gallium nitrate used in clinical trials where the drug is administered continuously for 5–7 days (25), GaM was incubated with human glioblastoma cells for 5 days and the effect on cell proliferation examined. As shown in the dose–response curves in Fig. 1E, GaM inhibited the growth of glioblastoma cells in a dose-dependent manner. In contrast, it was not cytotoxic to human BMECs, thus illustrating a significant differential in GaM's cytotoxicity toward glioblastoma when compared with normal cells. Maltol alone did not inhibit cell proliferation (Fig. 1E). GaM-induced cell death was confirmed by direct visualization of cells (Fig. 1F). D54 cells incubated with 50 and 100 μmol/L GaM for 48 and 96 hours displayed morphologic changes of nuclear condensation and cellular fragmentation consistent with cell death (Fig. 1F, b and c, e and f) compared with control cells that grew normally (Fig. 1F, a and d). In other experiments, the GaM-containing culture medium was removed from the wells after 5 days of incubation and replaced with fresh medium lacking GaM. Incubation was then continued for an additional 5 days and cell growth assessed. Under these conditions, GaM-treated cells failed to regain their ability to proliferate. These latter results strongly suggest that the growth-inhibitory action of GaM is not reversible.

The IC₅₀ concentrations for GaM glioblastoma cells in vitro are relevant to gallium levels attainable in vivo. Tf has two metal-binding sites per molecule and its concentration in the blood is 25.25–45 μmol/L. Under physiologic conditions, approximately one-third of Tf in the circulation is bound by iron thus leaving two-thirds of Tf available to bind Ga. Hence, gallium blood levels of 34–60 μmol/L are possible if these remaining Tf metal-binding sites are occupied by gallium.

Several proteins of the citric acid cycle and the mitochondrial electron transport chain contain iron–sulfur clusters that are essential for their function. A decrease in cellular iron can therefore result in a loss of mitochondrial function (32, 33). Since we have previously shown that gallium compounds can inhibit cellular iron uptake (28, 34), we examined the effect of GaM on cellular respiration as a measure of mitochondrial function. In these experiments, intact cells were analyzed after they had been incubated with GaM concentrations that were not cytotoxic to cells over 24 hours. As shown in Fig. 2A, D54 cells incubated with 50–100 μmol/L GaM for 24 hours did not display evidence of cell death. However, as shown in Fig. 2B, these cells displayed a progressive decrease in mitochondrial OCR and a loss of mitochondrial reserve capacity after 24 hours of incubation with increasing concentrations of GaM. Cells incubated with 25 μmol/L GaM, displayed a loss of reserve capacity (also known as the spare respiratory capacity) without a decrease in OCR below baseline, thereby indicating that even low, noncytotoxic concentrations of GaM affected the “fitness” of glioblastoma cells. At higher concentrations, GaM produced a decrease in OCR and reserve capacity. These results suggest that GaM decreases mitochondrial function as an early event before a decrease in cell proliferation or induction of cell death can be detected.

Ribonucleotide reductase (RR) catalyzes the synthesis of deoxyribonucleotides, a rate-limiting step in DNA synthesis (35). As illustrated in Fig. 3A, RR consists of two heterodimeric subunits M1 and M2, which are under the control of different genes (35). RRM2 expression increases as cells enter S-phase (36). RRM2 contains a binuclear iron center and an EPR-detectable tyrosyl free radical, both of which are essential for its activity (Fig. 3A; ref. 37). Cellular iron deprivation or blockade of iron incorporation into RRM2 inhibits RR enzymatic activity (38). Previously, we showed that gallium nitrate and Tf-Ga inhibit RRM2 activity in leukemia cells by: (i) induction of cellular iron deprivation which limits iron availability to RRM2 (22, 39) and (ii) direct action on enzymatic function which is independent of iron (40). To examine whether GaM interfered with RRM2 in glioblastoma cells, we measured the activity of the RRM2 tyrosyl radical by EPR spectroscopy in intact U-87 MG cells incubated with GaM for 24 hours. As shown in Fig. 3B, the EPR signal in GaM-treated cells was reduced by approximately 65% relative to control cells. To determine whether this was due to a reduction in RRM2 protein, RRM2 expression in U-87 MG cells was analyzed by Western blotting after incubation with GaM for 24 hours. In contrast to the reduction in the RRM2 EPR signal, we found that there was a 1.2- to 1.4-fold increase in RRM2 protein with 25 and 50 μmol/L GaM (Fig. 3C and D). Although the reason for this increase is not obvious, one possible explanation is that cells attempt to compensate for the GaM-induced loss of iron-dependent RR activity by increasing RRM2 protein production. Indeed, an amplification of the RRM2 gene occurs during the development of drug resistance to hydroxyurea, an agent that blocks RR activity by action on the tyrosyl radical of RRM2 (41). Collectively, these results indicate that GaM decreases the activity of iron-dependent RRM2 without reducing the synthesis of RRM2 protein.

To determine whether the expression of iron-related proteins in glioblastoma cells in vitro are relevant to glioblastoma in vivo, we examined normal and glioblastoma-containing brain tissues from rodent brains for TfR1 and ferritin and from human specimens for RRM2. Consistent with prior reports, we confirmed that TfRs were present on BMECs (Fig. 4A) indicating that these receptors could serve as portals for GaM to traverse the BBB and enter the brain. Within the brain, TfR1 expression in glioblastoma cells was markedly increased relative to the adjacent normal brain (Fig. 4B). The H- and L-subunits of the iron storage protein ferritin were also increased in glioblastoma (Fig. 4C and D, respectively). Ferritin is composed of 24 subunits of H- and L-ferritin in proportions that differ in various cell types. L-subunit–rich ferritin exists in greater proportion in tissues (such as the liver) that store iron, while H-subunit–rich ferritin exists in greater proportion in metabolically active tissues such as the heart and malignant cells (42). As shown in Fig. 4C and D, both H- and L-ferritin levels were increased in glioblastoma cells relative to normal brain. These findings are consistent with the studies of Schonberg and colleagues, which showed that ferritin levels are elevated in glioblastoma (7).

Figure 4F shows that iron-containing RRM2 protein is highly expressed in human glioblastoma but not in normal brain (Fig. 4E). Collectively, the data in Fig. 4 strongly suggest that glioblastoma cells in vivo increase their expression of TfR1 and ferritin to acquire and store greater amounts of iron than the surrounding normal brain in order to support the activity of ribonucleotide reductase and other iron-dependent proteins necessary for tumor proliferation and viability.

Considering that GaM binds to Tf and inhibits the growth of glioblastoma cells in vitro, we hypothesized that GaM would enter the brain via TfR1 present on the BMECs of the BBB and inhibit the growth of glioblastoma. Thus, we conducted a proof-of-principle experiment to examine the activity of GaM against glioblastoma in vivo. In the studies shown in Fig. 5, GaM or saline solution (control) was administered to rats inoculated with U-87 MG glioblastoma xenografts in the brain. Treatment of animals with GaM or saline was initiated only after tumors were established in the brain. Tumor response to treatment was assessed by MR imaging and RECIST V1.1 criteria as used in the clinic (27). Figure 5A shows the pre- and posttreatment brain MRI from a representative experiment. Postcontrast T1-weighted images (T1+C) and MION CBV maps were obtained in a glioblastoma xenograft-bearing rat on days 8 and 18 for control (Fig. 5A a–d) and a GaM-treated rat (Fig. 5A e–h). Comparing the cohort of control versus GaM-treated animals, Fig. 5B shows that GaM significantly inhibited the growth of glioblastoma relative to untreated controls. A decrease in CBV in GaM-treated animals was also noted (Fig. 5C); while the difference in this parameter did not reach statistical significance, there was clearly a trend towards a reduction in CBV with gallium treatment.

Tumors were examined for proliferation markers and TfR1 expression. Consistent with the antiproliferative activity of GaM, tumors from GaM-treated animals displayed a significant reduction in mitotic figures (P = 0.043) and mitotic index (P = 0.045) compared with untreated tumors (Fig. 5D, a and b). This finding is consistent with an in vivo inhibition of cellular proliferation by GaM in the glioblastoma xenograft and it clearly supports a cytostatic effect of GaM. In addition, GaM-treated tumors displayed a higher percentage of dead cells (37%) than untreated control tumors (24%). Although not significantly different, the observed trend in this parameter suggests that GaM exerts a cytotoxic effect that extends beyond inhibition of proliferation. The conclusion is supported by Fig. 1F and by our prior studies, which show that GaM induces apoptotic cell death in lymphoma cells (12).

The expression of TfR1 in tumors from control and GaM-treated animals was measured by IHC after completion of treatment. As shown in Fig. 5E, TfR1 expression was markedly increased in GaM-treated tumors relative to control tumors. This finding is consistent with GaM-induced tumor iron-deprivation in vivo as we have previously shown that Tf-Ga blocks cellular iron uptake and upregulates TfR1 mRNA in human leukemic HL60 cells (43). However, to confirm that glioblastoma U-87 MG and D54 cells would respond to GaM in a similar fashion, we conducted additional experiments which showed that GaM inhibited the uptake of ⁵⁵Fe-Tf (Fig. 6A and B), and produced an upregulation of TfR1 in both these cell lines (Fig. 6C). These results support our interpretation that the increase in tumor TfR1 expression in GaM-treated animals is secondary to GaM-induced tumor iron deprivation.

As an extension of our studies with U-87 MG and D54 cell lines, we investigated whether three previously described patient-derived GSC lines designated GSC-22, -33, and -44 (19), expressed GaM-targeted iron proteins and were sensitive to its cytotoxicity. Figure 6D shows that all three GSC lines expressed immunoreactive H- and L-ferritin and RRM2 proteins. Ligand–receptor binding and ligand internalization assays confirmed that TfR1 present on GSCs bound ¹²⁵I-Tf (Fig. 6E) and rapidly internalized it (Fig. 6F). As shown in Fig. 6G, the proliferation of these GSCs could be inhibited by GaM in a dose-dependent manner. Collectively, the experiments studied in Fig. 6D–G indicate GSCs display highly functional TfR1 that can be targeted by GaM.

Our study is the first to show that a novel gallium compound, GaM, has antineoplastic activity against glioblastoma in vivo. We demonstrated that GaM inhibited the growth of glioblastoma cell lines in vitro and used MR imaging of the rat brain to measure the impact of GaM on the growth and vascularity of established glioblastoma brain tumor xenografts in live animals. These studies clearly show that GaM significantly retards the growth of tumors and reduces their relative blood volume over a 10-day period of treatment.

Early clinical studies with ⁶⁷Ga scans showed that ⁶⁷Ga was taken up by brain tumors (44); additional studies revealed that the cellular uptake and cytotoxicity of gallium was enhanced by Tf (28, 45). Consistent with earlier reports, we confirmed that TfRs were expressed on both BMECs and glioblastoma tumors in patients. These findings prompted us to investigate if TfRs could be exploited for the delivery of Tf-Ga to brain tumors.

The strategy of enhancing drug delivery to the brain by conjugating drugs or toxins to Tf or to anti-TfR1 antibodies to target BMEC TfRs has been reported by others (46). However, our approach is different in that we have taken advantage of the high-affinity binding of gallium to endogenous Tf in the circulation to deliver Ga to the brain. We propose that after crossing the BBB, Ga binds Tf in the brain leading to targeting of Tf-Ga- to TfR-bearing glioblastoma cells. It is known that Tf in the normal brain is present in oligodendrocytes and is produced and secreted by the choroidal plexus (47). Furthermore, glioblastoma cells may also secrete Tf as an autocrine growth factor, which may enable them to acquire iron for their growth. In vitro studies show that GSCs release Tf to culture medium (7). Thus, Tf in the tumor microenvironment intended to transport iron into TfR1-bearing glioblastoma cells could be hijacked by gallium to enhance its uptake by tumor cells. While our studies indicate that GaM penetrates the brain via TfR-mediated transport, it is possible that a variable amount of GaM may also cross the BBB independent of Tf; this is being investigated.

Whereas GaM retards the growth of glioblastoma, it needs to be determined whether gallium compounds will adversely impact on normal brain function. In this regard, it is encouraging to note that central nervous system toxicity was not reported in phase I and II clinical trials of gallium nitrate (25, 26). Also, we did not observe neurologic deficits in the rats treated with GaM. An important consideration is that normal glial cells do not express TfRs in vivo (8). Thus, gallium is likely to be taken up by TfR1-expressing glioblastoma cells but not by normal glial tissue. However, neuronal cells do express TfRs (8) and could be targeted by Tf-Ga, but whether this would lead to neuronal toxicity remains to be determined. Collectively, both preclinical and clinical studies suggest a therapeutic index for gallium in which gallium compounds could display antitumor efficacy at concentrations unlikely to affect normal brain cells.

We show that GaM inhibits iron-dependent RRM2 and mitochondrial function. An important point in considering RRM2 as a target in glioblastoma is that normal brain cells do not proliferate in vivo and thus would not be expected to express RRM2. In contrast, glioblastoma being a high-grade proliferating brain tumor would be expected to express RRM2. This was confirmed in our studies comparing the expression of immunoreactive RRM2 protein in glioblastoma versus normal brain. Hence GaM, as an inhibitor of iron-dependent RRM2 activity, would be expected to block DNA synthesis in glioblastoma cells but not in normal brain cells.

Whereas gallium-induced inhibition of RRM2 in itself would be sufficient to inhibit the proliferation of glioblastoma brain tumors, other iron-dependent mechanisms beyond RR could be targeted by gallium. Thus, we focused on GaM's action on the mitochondria. We hypothesized that iron–sulfur cluster–containing proteins of the citric acid cycle and mitochondrial electronic transport chain could be prime targets for disruption by GaM and that this could result in a block in energy production and cell death. We discovered that even at noncytotoxic concentrations, GaM produced loss of cellular reserve capacity in glioblastoma cells. This finding indicates that one of the initial mechanisms of action of GaM on the mitochondria is a loss of its spare respiratory capacity; this could decrease a cell's ability to cope with an energy demand (48). At higher concentrations, GaM further suppressed mitochondrial function as evidenced by a dose-dependent decrease cellular OCR and reserve capacity. This effect undoubtedly contributes to Ga-induced cell death.

Central to gallium's mechanisms of action is its ability to disrupt cellular iron metabolism at several levels (11). Consistent with this mechanism, we found that tumors in GaM-treated rats increased their expression of TfR1. Cellular iron deprivation produces an upregulation of TfR synthesis due to the enhanced interaction of cytoplasmic iron regulatory proteins-1 and -2 (IRPs-1,-2) with iron-regulatory elements (IRE) present on the 3′ untranslated region of the TfR mRNA (49). In support of this mechanism, we demonstrated that GaM inhibited iron uptake by U-87 MG and D54 cells in vitro and that this resulted in an increase in cellular TfR expression. Collectively, our studies strongly support tumor iron-deprivation as one of the mechanisms of action of GaM against glioblastoma in vivo.

In summary, our results show for the first time that a gallium compound that perturbs tumor iron homeostasis has potential in the therapy of glioblastoma brain tumors. Our animal studies serve as proof-of-principle that GaM can enter the rodent brain and retard the growth of glioblastoma tumors; this builds the foundation for additional research. Future studies will investigate the efficacy of gallium compounds alone and in combination with other drugs in GSC-derived orthotopic xenograft models. These studies will determine the optimum dose and treatment schedule for gallium compounds and their impact on survival. It is envisioned that such preclinical studies will lead to GaM-based clinical trials in glioblastoma.

No potential conflicts of interest were disclosed.

Conception and design: C.R. Chitambar, K.R. Pechman, J.P. Wereley, and K.M. Schmainda

Development of methodology: C.R. Chitambar, K.R. Pechman, J.P. Wereley, P.A. Clark, J.S. Kuo, W.E. Antholine, and K.M. Schmainda

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.R. Chitambar, H.S. Alhajala, K.R. Pechman, J.P. Wereley, J.S. Kuo, W.E. Antholine, and K.M. Schmainda

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.R. Chitambar, M.M. Al-Gizawiy, H.S. Alhajala, K.R. Pechman, J.P. Wereley, R. Wujek, J.S. Kuo, W. E. Antholine, and K.M. Schmainda

Writing, review, and/or revision of the manuscript: C.R. Chitambar, M.M. Al-Gizawiy, H.S. Alhajala, P.A. Clark, J.S. Kuo, and K.M. Schmainda

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.R. Chitambar, M.M. Al-Gizawiy, J.P. Wereley, and P.A. Clark

Study supervision: C.R. Chitambar

This work was supported by the Thomas A. and Lorraine M. Rosenberg Award for Translational Cancer Research from the Froedtert Hospital Foundation and the Medical College of Wisconsin Cancer Center (to C.R. Chitambar); NIH National Biomedical EPR Center Grant EB0011980 (to W. E. Antholine), and National Cancer Institute at the NIH (grant number R01 CA 082500) and the MCW Advancing Healthier Wisconsin Research Program (to K.M. Schmainda).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.