Background
Gliomas are primary central nervous system tumors that arise from astrocytes, oligodendrocytes or their precursors. Following the World Health Organization (WHO) classification, gliomas can be classified in 4 groups according to their histological characteristics and the most malignant form is glioblastoma multiforme (GBM). GBM is uniformly fatal and largely unresponsive to all available treatments. Despite intensive therapy including surgery, radiotherapy and chemotherapy, the average of survival of patients with glioblastoma usually is 15 months from the time of first diagnosis [
1,
2]. Conventional surgical excision, generally limited to the main tumor mass, does not remove the microscopic foci of neoplastic cells that invade the surrounding normal brain substance beyond the main tumor mass, and that are responsible for the inevitable tumor recurrence. Radiotherapy and chemotherapy, often associated to surgery, cannot ablate completely these tumors, since this would require unacceptably high radiation/chemotherapic doses that result in severe brain-neuron damage. There is therefore a clear need to accelerate progress in the development of new strategies for treatment of glioma.
Several therapeutic approaches for glioma are currently being investigated in animal models and patients. They include delivery of cytotoxic genes and proteins to glioma cells, suppression of angiogenesis, and immune stimulation [
3]. Concerning chemotherapy, alkylating agents such as temozolomide (TMZ) are widely used in the treatment of brain tumours [
4]. As a cytotoxic alkylating agent, TMZ is converted at physiologic pH to the short-lived active compound, monomethyl triazeno imidazole carboxamide (MTIC). The cytotoxicity of MTIC is primarily due to methylation of DNA at the O6 and N7 positions of guanine, resulting in inhibition of DNA replication. Chemotherapics have substantial side effects and limited efficacy, and this further underlies the need of innovative approaches for glioma treatment.
In this paper we describe a potential novel therapy for glioma, based on intracerebral administration of cytotoxic necrotizing factor 1 (CNF1), a bacterial protein toxin produced by specific strains of
Escherichia coli. CNF1 is a single-chain protein, consisting of a N-terminal domain involved in cell binding, a middle region mediating membrane translocation, and a C-terminal catalytic domain. The C-terminal part of CNF1 is released into the cytosol where it catalyzes the deamidation of a single glutamine residue of the Rho GTPases (RhoA, Rac1 and Cdc42). Rho GTPases are molecular switches that cycle between a GDP-bound inactive and a GTP-bound active state to control a multitude of cellular events, like actin cytoskeleton organization as well as gene transcription, cell proliferation, and survival [
5]. Rho GTPases deamidated by CNF1 are not able to hydrolyse GTP and remain in a persistent activated state [
6,
7] which is followed by partial deactivation of these regulatory proteins via degradation by the ubiquitin–proteasome pathway [
8]. The persistent activation of Rho GTPases by CNF1 causes a remarkable reorganization of the actin cytoskeleton with dramatic functional consequences. In particular, cultured proliferating cells exposed to CNF1 acquire a multinucleated phenotype (cytotoxic effect), due to stabilization of the actin network and prevention of cytodieresis despite ongoing nuclear division [
9]. On the other hand, our recent studies have demonstrated “plasticizing” effects of CNF1 in neurons. Specifically, intracerebral administration of CNF1 improves neuronal function, learning and memory [
10,
11], and these effects are associated with a enhancement of brain plasticity, exemplified by the increase in spine density in cortical neurons [
10].
In view of these striking, differential effects of CNF1 on proliferating cells and neurons, we have probed for the first time the potential antitumoral effects of this toxin in cell culture and in an animal model. Among the available glioma models, we adopted the well accepted GL261 syngeneic mouse model of high grade glioma [
12,
13] based on intracerebral injection of GL261 cells in C57/Bl6 mice. We demonstrate the cytostatic and cytotoxic effects of CNF1 on GL261 glioma cell line, and the survival-promoting effect on tumor-bearing animals. We have also provided proof-of-principle for cytotoxic effects of CNF1 in human tumor cells.
Methods
GL261 glioma cell cultures
The murine glioma GL261 cell line was a kind gift from Dr. C. Sala (CNR Neuroscience Institute, Milan). GL261 cells were grown in complete Dulbecco’s modified Eagle’s medium (DMEM) containing 10% Newborn calf serum, 4.5 g/L glucose, 2 mM glutamine, 100 UI/ml penicillin and 100 mg/ml streptomycin at 37°C in 5% CO2 with media changes three times per week.
CNF1
E. coli CNF1 was obtained from the 392 ISS strain (kindly provided by V. Falbo, Istituto Superiore di Sanità, Rome, Italy) and purified as described previously [
14]. The levels of lipopolysaccharide (LPS) in the CNF1 preparation were assessed by the Limulus Amebocyte Lysate (LAL) kinetic chromogenic assay (performed by LONZA Verviers Sprl). The LPS concentration determined by the assay (0.07 ng/ml) was much lower than that required to achieve biological effects (e.g. 1 ng/ml in macrophages, one of the most sensitive cells to LPS).
The activity of every batch of CNF1 was tested on GL261, treating the cells for 24 hours. Three parameters were considered: i) cells morphology (enlargement and flattening of cells), ii) increased size of nucleoli and iii) the ratio between mono and multinucleated cells. The activity of each CNF1 preparation was considered satisfactory if at least one of the three parameters indicated above were present in more than 60% of treated cells.
Clonogenic assay
GL261 cells were harvested by trypsinization, counted and seeded onto twenty four-well plates at a density of 300 cells/plate. To assess the effect of CNF1 on cell proliferation, GL261cells were incubated for 9 days with different concentrations of CNF1 (from 8 × 10
−11 to 3.2 × 10
−9 M). Nine days after treatment, cells were stained with crystal violet, the number of colonies containing at least 50 cells was counted [
15] and the effective half inhibitory dose (IC50) of the toxin was calculated on the basis of linear regression equation.
Wound healing-migration assay
The wound healing migration assay was performed according to Liang [
16] with minor modifications. Briefly, GL261 cells were seeded in 6-well cell plates and cultured to a confluent monolayer. A sterile pipette tip (200 μl) was used to make a straight scratch on the monolayer of cells and the wound was allowed to heal for 8, 24 and 48 hours. To evaluate CNF1 effects, GL261 cells were incubated with CNF1 for 24 hours before making the scratch, and wound healing was assessed 8, 24 and 48 hours after the injury. The cells were then fixed with methanol and stained with crystal violet. The extent of cell migration was photographed and the wound size measured using image analysis software (ImageJ).
Apoptosis-necrosis assays
Apoptotic and/or necrotic cells were determined using Annexin V- Propidium Iodide (PI) Staining Kit. The assay was performed following the manufacturer instructions (Annexin V kit, BD Pharmingen). Briefly, 300 GL261 cells were seeded on twenty four-well plates and incubated in CNF1 (18 nM) for 10 days. Annexin V and Propidium Iodide were diluted 1:200 in KREB medium (NaCl 120 mM, NaKCO3 25 mM, KCl 4.7 mM, KH2PO4 1.18 mM, MgSO4 1.18 mM, CaCl2 2.5 mM, EDTA 0.026 mM, glucose 5.5 mM). Cells were also stained with Hoechst dye (bisbenzimide, Sigma; 1:500 in PBS) and counted with fluorescence microscopy. We classified cells as positive for annexin V only (Ann V + PI-), positive for PI only (Ann V- PI+), positive for both markers (Ann V + PI+) or unlabeled.
Senescence-Associated Beta-Galactosidase (SA beta-gal) Assay
To determine cellular senescence, GL261 cells were plated in triplicate at low density (50% confluence) in 24-well plates. The cells were treated with CNF1 (1 nM) and incubated for 24, 48 or 72 hours before beta-galactosidase measure (senescence detection kit, Abcam catalog ab65351). The percentage of positively stained cells was determined after counting three random fields of cells. Representative microscopic fields were photographed under a 20× objective.
Human glioblastoma cell cultures
Human biopsies of glioblastoma multiforme (GBM) were collected from 2 subjects who underwent brain surgery for tumor removal. The study was approved by the Human Ethics Committee of the University of Pisa and Pisa Hospital. Written, informed consent was obtained from the patients according to institutional guidelines. Patient 1 had a left parietal lesion resembling a high grade glioma on a contrast enhanced MRI scan, while patient 2 had a similar lesion located in the left fronto-parietal region. A surgical gross total resection was performed and the histological examination confirmed the suspected tumor type (WHO grade IV) in both cases. After collection, primary tumor cells were isolated according to [
17]. Briefly, tissue explants were incubated with trypsin in DMEM for 10 minutes at 37°C and then added of 3 volumes of DMEM with 10% FBS. Tissue was completely disgregated by gentle pipetting. Cells suspension was then plated on tissue culture dishes and culture medium (DMEM, 10%FBS, 100 IU/ml penicillin,100 mg/ml streptomycin) was replaced every 3–5 days. After a week cells were trypsinized and splitted. At passage five [
18] cells were plated on cover slips and then treated with 18nM CNF1 for 9 days. At this time point, cells were fixed and stained to observe their morphology.
Animals and tumor induction
To induce glioma formation, C57BL/6 mice (12–14 weeks old) received a stereotaxically guided injection of 40,000 GL261 cells (20,000 cells/μl PBS solution) into the visual cortex (2 mm lateral to the midline and in correspondence with lambda) using fine glass micropipettes (tip diameter 40 μm). All experimental procedures were in conformity to the European Communities Council Directive 86/609/EEC. The animal experiments described in this manuscript have been approved by the Italian Ministry of Health, Department of Animal Health (Office n. 6), with decree #258-2012/B, dated Oct 23, 2012.
CNF1 injections and TMZ minipumps
Five days after GL261 cells injection (tumor induction), mice were divided into three groups. The first group received CNF1 injection, the second Tris–HCl buffer injection (control condition). Stereotaxic infusions of CNF1/Tris–HCl (1 μl of a 2 or 80 nM solution) were made into the primary visual cortex of adult mice under avertin anesthesia (intraperitoneal injection of 2,2,2-tribromoethanol solution; 0.1 ml/5 g body weight). Injections were performed in three sites: 1.5, 2 and 2.5 mm lateral to the midline in correspondence with lambda. CNF1/Tris–HCl injections was slowly delivered at a depth of 0.7–0.8 mm from the pial surface.
The third experimental group received Temozolomide (TMZ) micropumps implantation for a week. Minipump implantation was performed as described previously [
19,
20]. Micro-osmotic pumps (Alzet 1007D; Alza, USA; pumping rate 0.5 μl/hr) were filled with TMZ (20, 140 and 200 μM solution) and connected with polyethylene tubing to 30 G stainless steel cannulae [
20]. A small hole was made in the skull (2 mm lateral, 1 mm anterior) and the cannula was lowered into the cortex. The minipump was positioned subcutaneously under the neck and the cannula was secured to the skull with acrylic cement. Animals survival was checked daily and monitored up to 60 days after tumor implantation. At this time, mice that were still alive were deeply anesthetized and perfused with 4% paraformaldehyde. Coronal brain sections through the occipital cortex (45 μm thick) were cut with a freezing microtome, stained with Hoechst dye (1:500, Sigma) and histopatological examinations were carried out.
Statistical analysis
Statistical analysis was performed with SigmaPlot (version 11). Differences between three or more groups were evaluated with one way analysis of variance (ANOVA) followed by Holm-Sidak test.
Survival analysis was performed using Kaplan-Meier (LogRank) statistics.
Discussion
The data reported in this manuscript demonstrate the therapeutic potential of a bacterial protein toxin, CNF1, in blocking proliferation of glioma cells and prolonging the survival of glioma-bearing mice. At present, there are several evidences on the use of targeted toxins to treat cancer [
21]. However, the clinical efficacy of toxins has mainly been observed in hematological malignancies, but not in solid tumors, including GBM. The toxin used in this report, CNF1, is produced by certain pathogenic strains of
E. Coli and consists of a N-terminal binding domain that interacts with membrane receptors on target cells, a middle translocation domain to enter the cytosol and a C-terminal catalytic domain [
22,
23]. The catalytic moiety of CNF1 activates members of the Rho GTPase family (Rho, Rac and Cdc42) by conversion (deamidation) of a single glutamine residue into glutamic acid. This aminoacid change impairs GTP hydrolysis thus locking the Rho family proteins in their GTP-bound, active state [
22,
23]. Depletion of activated Rho GTPases is then accomplished via ubiquitin-mediated proteasomal degradation [
8].
Since Rho GTPases regulate the dynamics of the actin cytoskeleton, their activation by CNF1 triggers a rapid reorganization of F-actin [
24]. In particular, proliferating cells exposed to CNF1 acquire a multinucleated phenotype, due to the inhibition of cytokinesis despite ongoing nuclear division [
22,
25,
26]. We have observed CNF1-induced multinucleation in either GL261 glioma cells or early passage cell lines from primary GBM specimens. We have also shown that these multinucleated cells degenerate within about 15 days in vitro.
These data raise the still open question of the identification of the pathway(s) by which CNF1 causes cell degeneration. Experiments with β-Galactosidase, Annexin V and Propidium Iodide labelling allow us to conclude that senescence and then necrosis account for CNF1-induced GL261 cell death. This is consistent with the senescent-like phenotype (cell enlargement, flattening, increase in size of nuclei and nucleoli) assumed by cells treated with CNF1 (see Figure
4A, inset). At the moment, we cannot exclude the possibility that the autophagy pathway contributes to the CNF1-induced phenotype. A recent paper indicates the activation of the autophagy pathway following the treatment of glioma cells with pertussis toxin and TMZ [
27]. Future studies are needed to clarify the possible role of autophagy in the antitumoral action of CNF1.
Since CNF1 is derived from
E. coli, it might be argued that contamination by bacterial products (such as LPS) could contribute to the observed antineoplastic effects via the activation of immune responses in the brain [
28]. However, this hypothesis is very unlikely, as the amount of LPS is our CNF1 preparation was found to be extremely low, in a range unable to activate macrophages (see Methods).
In GBM and in experimental glioma models, the activation of Rho GTPases (such as Rac1) has been linked to increased cell invasion [
29], pointing to these molecules as key therapeutic targets. In this scenario, the dramatic effect of CNF1 on the actin cytoskeleton reorganization renders cells virtually immobile, and this might substantially enhance the anti-tumoral properties of the toxin. Indeed, we have demonstrated, in the wound migration assay, that CNF1 dramatically decreased the motility of GL261 cells, suggesting a further potential therapeutic feature of this toxin for its possible application in the treatment of glioma tumors. This aspect is important in the context of glioma treatment, because glioma cells tend to diffuse profusely into adjacent healthy tissue. It is well established that CNF1 causes assembly of F-actin in prominent stress fibers and extreme flattening of the cell body [
6,
24]. Since cells in motion need actin dynamics to attach and detach from the extracellular matrix, actin polymerization by CNF1 effectively renders cells stationary.
Furthermore, it is worth noting that the other key aspect of the action of CNF1 is on neuron plasticity and health [
10]. This cooperates with the antitumoral effect (reduced proliferation and motility) and may lead to better preservation of neuronal function in the brain areas surrounding the tumor. Experiments to address CNF1-mediated functional sparing in glioma models are currently ongoing in our laboratory.
Importantly, a key feature of CNF1 is the rapidity of its action. We found that exposure to CNF1 for 1 hr was sufficient to halt proliferation of most of the treated cells (Figure
2). Since CNF1 is an enzyme, entry of a few toxin molecules inside a cell can lead to the modification of several Rho GTPase targets, providing a dramatic amplifying effect. Thus, even a short CNF1 exposure produced nearly maximal effects on cell proliferation. This may be important for glioma therapy, as one problem of local therapies is the rapid washout of therapeutic molecules infused in the tumor area. Moreover, the effects of CNF1 appear to persist for weeks following one single administration of the toxin [
10,
11,
30], likely due to persistent catalytic activity of the toxin inside cells. This further strengthens the potential of CNF1 and could avoid the need for repeated drug administration.
The potential involvement of CNF1 in cell transformation is still controversial [
31]. Several studies in vitro and in vivo, including the present results, demonstrate the anti-proliferative and cytotoxic effect of CNF1 in cancer cell lines [
22,
25,
26]. Furthermore, cell transformation and tumor formation have never been observed after a single administration of CNF1 in the rodent brain (e.g. [
10,
11,
30]).
In order to evaluate the effectivness of CNF1, we compared its action to that of a current standard chemotherapic drug, namely TMZ. Even if TMZ in experimental animal models is typically given orally, here we have chosen to administer it via an intracranial route to allow direct comparison with CNF1, which does not cross the blood–brain barrier. We found that TMZ, administered for one week via minipumps, was effective in prolonging animal survival but had a quite narrow therapeutic range, with concentrations of 20 μM being ineffective and concentrations > 200 μM being toxic for the animals. The limited efficacy of TMZ chemotherapy is in line with current clinical and experimental experience [
4,
32] and can be attributed to both inherent and acquired tumor drug resistance. In contrast with conventional chemotherapy, CNF1 had greater efficacy and showed no obvious side effects with increasing doses. Remarkably, more than half of the animals treated with 80 nM CNF1 survived for at least 60 days following glioma cell inoculation. One important aspect of CNF1 is its long-lasting action, as one single intracerebral administration leads to Rho GTPase activation for at least 10–28 days [
10,
11,
30].
Competing interest
The authors declare that they have no competing interest.
Authors’ contributions
EV carried out all in vivo experiments, performed the statistical analysis and drafted the manuscript. AP carried out all in vitro experiments, performed the statistical analysis and drafted the manuscript. CC partecipated in the in vivo experiments. SL performed the in vitro experiments with human glioblastoma cells. EC performed the cell death assays. AF and CF provided CNF1 toxin. NB and RV provided human biopsies of glioblastoma multiforme. MCa and MCo conceived of the study, and participated in its design and coordination and drafted the manuscript. All authors read and approved the final manuscript.