Available online at www.sciencedirect.com

Free Radical Biology & Medicine 44 (2008) 63–72www.elsevier.com/locate/freeradbiomed

Original Contribution

Phenyl-tert-butylnitrone induces tumor regression and decreasesangiogenesis in a C6 rat glioma model

Sabrina Doblas a,b,c, Debbie Saunders b,c, Preeti Kshirsagar d, Quentin Pye c, Jenny Oblander e,Brian Gordon e, Stanley Kosanke f, Robert A. Floyd c,g,h, Rheal A. Towner b,c,f,⁎

a Oklahoma University Bioengineering Center, Norman, OK, USAb Small Animal MRI Facility, Oklahoma Medical Research Foundation, Oklahoma City, OK 73104, USA

c Free Radical Biology and Aging Research Program, Oklahoma Medical Research Foundation, Oklahoma City, OK 73104, USAd Advanced Imaging Research Center, University of Texas Southwestern, Dallas, TX, USA

e Comparative Medicine, Oklahoma Medical Research Foundation, Oklahoma City, OK 73104, USAf Department of Pathology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA

g Department of Biochemistry and Molecular Biology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USAh Merrick Foundation Chair of Aging Research, Oklahoma Medical Research Foundation, Oklahoma City, OK 73104, USA

Received 13 April 2007; revised 7 September 2007; accepted 10 September 2007Available online 20 September 2007

Abstract

The prognosis of patients who are diagnosed with glioblastoma multiforme is very poor, due to the difficulty of an early and accurate diagnosisand the lack of currently efficient therapeutic compounds. The efficacy of phenyl-tert-butylnitrone (PBN) as a potential anti-glioma therapeuticdrug was assessed by magnetic resonance (MR) imaging (T1/T2-weighted imaging) and MR angiography (time-of-flight imaging, in conjunctionwith a Mathematica-based program) methods by monitoring morphologic properties, growth patterns, and angiogenic behaviors of a moderatelyaggressive rat C6 glioma model. MR results from untreated rats showed the diffusive invasiveness of C6 gliomas, with some associatedangiogenesis. PBN administration as a pretreatment was found to clearly induce a decrease in growth rate and tumor regression as well aspreventing angiogenesis. This compound even had a 40% efficiency in reducing well-established tumors. MR findings rivaled those fromhistology and angiogenesis marker immunostaining evaluations. In this study we demonstrated the efficiency of PBN as a potential anti-gliomadrug and found it to inhibit tumor cell proliferation and prevent vascular alterations in early stages of glioma progression. The MR methods thatwe used also proved to be particularly suitable in following the angiogenic behavior and treatment response of a potential anti-glioma agent in a ratC6 glioma model.© 2007 Elsevier Inc. All rights reserved.

Keywords: Phenyl-tert-butyl-nitrone; Angiogenesis; Rat glioma; C6 cells; Magnetic resonance angiography; Magnetic resonance imaging; Free radicals

Glioblastoma multiforme, the most common and malignanttype of gliomas in humans, represents 20% of all primary

Abbreviations: MR, magnetic resonance; MRI, magnetic resonanceimaging; MRA, magnetic resonance angiography; PBN, phenyl-tert-butylni-trone; LPS, lipopolysaccharide; DMEM, Dulbecco's modified Eagle's medium;FBS, fetal bovine serum; TR, repetition time; TE, echo time; FLASH, fast lowangle shot; ROI, region of interest; vWF, vonWillebrand factor; VEGF, vascularendothelial growth factor; CT, computed tomography; COX2, cyclooxygenase2; NF-κB, nuclear factor κB; iNOS, inducible nitric oxide synthase.⁎ Corresponding author. Fax: +1 405 271 7254.E-mail address: Rheal-Towner@omrf.ouhsc.edu (R.A. Towner).

0891-5849/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.freeradbiomed.2007.09.006

brain tumors. In the best current treatment situation (surgicalremoval of the tumor followed by radiation and/or chemo-therapy sessions), patients diagnosed with a grade IV gliomahave an average survival time of 15 months, and only 3% ofthem will survive after 5 years [1]. This poor prognosis islargely due to the difficulty of obtaining early and accuratediagnoses and also due to the lack of efficient therapeuticdrugs. In order for clinicians to decide on an adequatetherapeutic drug for treatment, the growth pattern, morpho-logy, and vascular properties of the glioma must be accuratelyand quickly characterized. Also, new therapeutic drugs withthe ability to inhibit tumor growth or to suppress the

64 S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

formation of newly formed blood vessels, thus limiting tumorcell invasion, are always being sought. Ideally, these agentsshould have an effect on well-established and vascularizedgliomas.

Noninvasive in vivo imaging tools such as magneticresonance (MR) techniques are ideal diagnostic modalities,which help the assessment of glioma characteristics. In ourstudy, MR imaging (MRI) and MR angiography (MRA) wereused to characterize a C6 rat glioma model. Tumors weregenerated by intracerebral implantation of the C6 glioma cellline, which has been shown to yield uniform growth [2] withminimal distant metastasis [3]. Compared to other rat cell lineslike F98 and 9L, C6 cells have also been found to show a highersimilarity to human glioma cells in the expression of genesmainly involved in tumor progression [4]. Finally, C6 gliomashave been widely used as a reliable model of glioblastomamultiforme for assessment of treatment efficiency [3]. Themorphologic properties of glioma models are easily assessed byMRI, which allows characterization of their degree ofinvasiveness, size, and growth pattern. MRA can also be usedto assess neovascularization and vascular alterations, which arecharacteristic properties of tumors. The growth and spread ofgliomas within the host tissue depends upon their ability to getnutrients and remove waste products [5], and angiogenesis is atrademark of high-grade gliomas [6].

In addition to the use of MRI and MRA to study rat gliomas,we also assessed the efficiency of phenyl-tert-butylnitrone(PBN) to inhibit gliomas. PBN, which has been widely used as aspin trapping agent in electron spin resonance studies, has alsobeen shown to have interesting pharmacological properties. Forinstance, PBN has a neuroprotective effect in rat brains thatdevelop meningitis [7], has been found to protect againstlipopolysaccharide (LPS)-induced shock in mice [8] and rats[9], has anti-inflammatory properties [9], and has also beenshown to inhibit neoplasia in rats prone to liver cancer [10]. Inthis study, the effects of PBN treatment on glioma growth andangiogenesis were monitored using MRI and MRA to establishthe efficacy of PBN as a potential new anti-glioma therapeuticdrug.

Materials and methods

Cell culture and preparation

The C6 glioma cell line was purchased from the AmericanType Culture Collection and cultured in Dulbecco's modifiedEagle's medium (DMEM) supplemented with 10% fetal bovineserum (FBS) and 1% penicillin/streptomycin. A rat primaryastrocyte culture was obtained as described by Hensley et al.[11]. Cells were then maintained in a 37°C incubator with 5%CO2.

Before implantation, cells were briefly trypsinized, centri-fuged, and resuspended in DMEM supplemented with 10%FBS and 1% penicillin/streptomycin. Agarose was added so thatthe final suspension contained 1% ultralow gelling temperatureagarose (Sigma). Cells and syringes were kept at 37°C untilinjection.

Orthotopic glioma model

The animal study was conducted in compliance with theInstitutional Animal Care and Use Committee. Three-month-old male Fischer 344 rats (250 g) were purchased fromCharles River Laboratories and fed a choline-deficient diet(Dyets, Philadelphia, PA, USA), which was initiated 5 daysbefore cell implantation. The orthotopic glioma model wasobtained by modifying the method as described by Kobayashiet al. [12]. Each animal was anesthetized (2.7% isoflurane at0.7 L/min oxygen) and immobilized on a stereotaxic unit(Stoelting Co., Wood Dale, IL, USA). The skin of the headwas cleaned with iodine, and the skull was exposed afterincision. A hole was drilled (Ram Products, Inc., EastBrunswick, NJ, USA) through the skull at 2 mm anteriorand 2 mm lateral to the bregma in the right-hand side of thehead. Ten microliters of a 106 C6 cells/ml suspension wasinjected into the cortex at a depth of 3 mm from the dura at arate of 2 μl/min (WPI, Inc., Sarasota, FL, USA). A waitingtime of 2 min was implemented after injection and the drilledhole was sealed with bone wax to prevent any reflux. Thewound was sutured and covered with surgical glue. The wholeprocedure was performed under sterile conditions. Controlinjections of rat primary astrocytes were also conducted usingthe same conditions (n=3).

Animals were allowed to recover under supervision and weremonitored periodically. Each rat received an analgesic andantibiotics for 2 days after the surgery and was also checked forany infection with MRI.

PBN treatment

PBNwas synthesized and purified to 99% purity according tothe method described by Janzen and Haire [13] in our laboratory.The drug was administered in the drinking water at aconcentration of 75 kg rat/day (0.065% w/w, for an estimatedamount of water intake at a rate of 110–120 ml/kg rat/day) aspreviously reported by Nakae et al. [10]. Animals were dividedinto three groups. Two groups of rats were treated with PBNeither pre-tumor or post-tumor formation. PBN treatment wascontinuous and started 5 days before cell implantation for thefirst treatment group (n=5). The second group of rats wasadministered PBN-supplemented water when glioma volumeswere about 50 mm3 (approximately 14 days after cellimplantation) (n=5). A last group of animals was kept ascontrols and did not receive any treatment (n=5).

MRI and MRA experiments

Animals were imaged at the OMRF Small Animal MRIFacility using a 7-Tesla/30-cm horizontal bore magnet (BrukerBioSpin MRI GmbH, Karlsruhe, Germany) 7 days after cellimplantation and then every third day until day 40 or just beforethe death of the rat. Each animal was anesthetized (1.5%isoflurane at 0.7 L/min oxygen) and placed in a MR probe headfirst and in a supine position. A head surface coil was used toreceive the induced MR signal. A quadrature volume coil

Fig. 1. An example of a ROI chosen for quantification is represented (rectangle)in a primary astrocyte-injected animal, and the cell implantation site is alsodepicted (circle). 2D projections of the vasculature (in the (A) horizontal and (B)transverse planes) are shown superposed on T2-weighted images, with eachvessel labeled. The proximal (2 and 3) and distal (4, 5, and 6) parts of the medialcerebral artery (1) are the most altered during glioma progression.

65S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

(72-mm inner diameter) was used for transmission of allradiofrequency pulses (Bruker BioSpin MRI GmbH).

For tumor growth monitoring, T1- and T2-weighted multislice(24 transverse slices taken in a field of view of 3.5×3.5 cm2)interleaved images were obtained with a 1-mm thickness, aninterslice distance of 1mm, a repetition time (TR) of 2370ms andecho times (TE) of 17 and 64 ms, for a total acquisition time of20min, using a spin echomethod.MRAdatawere acquired usinga 3D fast low angle shot (FLASH) method in the coronal planewithin a volume of interest of 2.6×1.7×1.1 cm3 at an angle of 5°;relative to the horizontal plane, a TE of 2.3 ms, a TR of 25 ms,and a flip angle of 25°, for a total acquisition time of 27 min.The data were then reconstructed to provide a 512×512×256final image matrix, with a pixel resolution of 50× 70×40 μm3.

Modeling of C6 glioma growth patterns

Tumor sizes were assessed using ImageJ software (version1.32j, NIH). Tumor areas were measured from each slice andcompiled to obtain a total volume. Glioma volume data werethen fitted by a Gompertz function [14],

V tð Þ ¼ V0 expbad 1� exp �ad t½ �ð Þ

� �ð1Þ

which describes the tumor volume V(t) as a function of theinitial volume V0 and two Gompertz constants: the tumorspecific growth rate, α (day−1), and the retardation factor, β(day−1) [15].

If desired, the instantaneous doubling time Td can be furthercalculated using the equation [14]

Td ¼ �1a

ln 1� ln½2�ln½VlV ðtÞ�

� �ð2Þ

where V8 is an asymptotic limit of the volume given by [15]

Vl ¼ V0 expba

� �ð3Þ

In the case of tumors whose growth is described by theGompertz equation, a V(t) is then arbitrarily chosen (in our case,V(t)=100 mm3) to calculate Td, because the Td of these tumorsis not constant [16].

Finally, survival of each treatment group was represented asthe percentage of rats still surviving at a given time point.

Blood volume quantification

Calculations to quantitate blood vessel volumes wereconducted using Mathematica software (version 5.1; WolframResearch, IL, USA). Briefly, 3D intensity matrices obtainedfrom 3D-FLASH were taken into consideration in a chosenregion of interest (ROI) within the frontal right lobe, as showedin Fig. 1. A minimum signal intensity threshold was chosen (avalue of 14,000 in signal intensity) to cancel the maximumamount of background noise (determined in a vessel-free area ofthe brain), and this value was kept constant throughout the wholeexperiment. Pixels whose intensities were below this value were

set to zero. The number of remaining pixels (corresponding toflowing blood) was determined and was used to compute bloodvolumes by straightforward summation. Finally, projections ofthese high-intensity pixels were created along two differentdirections to provide 2D vasculature representations.

The ROI was chosen to be large enough to enclose a volumeof 425 mm3 (the maximum volume reached by our C6 gliomas)and kept at a constant size for each animal and for each data set.Because the glioma cells were implanted on the right-hand sideof the cortex, we chose to locate our ROI on the right-hand sideof the brain, starting at the rostral pole and enclosing most of themedial cerebral artery (these were the main arteries alteredduring glioma progression). In order to avoid any misplacementerror of this projected volume, MRI data were also consideredfor coregistration. More precisely, T2-weighted images wereused to provide spatial information on the coordinates of theglioma and its depth. To compare T2-weighted data and MRAdata, the sizes of the two different fields of views as well as theirpositions relative to the magnet isocenter (determined usingGeometry Editor in Paravision software (version 4.0; BrukerBioSpin MRI GmbH)) were taken into consideration.

Finally, in order to eliminate variability and to be able tocompare data sets between animals, blood volume ratios werecalculated for each time point, with the first day of assessmentbeing considered as a baseline for each animal.

66 S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

A quick estimation of the blood vessel length and diameterwas obtained using ImageJ. Briefly, the area and length of eachlabeled vessel (as shown on Fig. 1) were measured on the 2Dvasculature projections obtained using Mathematica software.The vessel diameter was calculated for each vessel by dividingits area by its length.

Histology and immunostaining assays

Animals were checked daily. As soon as they exhibited anyindication that may lead to death (lethargy, difficulty in moving,and/or decreased appetite) or at the end of the experiment (40days after cell implantation), they were anesthetized with 2.5%isoflurane at 0.7 L/min oxygen. Cardiac perfusion withphosphate-buffered saline was used to remove blood frombrain tissue for histological evaluations. Brains were extracted

Fig. 2. (A) The growth patterns of representative C6 gliomas from each treatment reg(triangles), and treated with PBN when the tumor was 50 mm3 (circles). Representa(other groups) were chosen. For the latter, gliomas that responded positively (opengrowth phases from day 10 to the death of the animal (or to the beginning of regressgrowth rates (continuous, dotted, and dot-dashed lines for respectively untreated, PBNa line have not been fitted. These growth curves depict the effects of PBN treatmerepresented as means±standard deviation. *pb0.05 by Student’s two-tailed t test bretardation factors and doubling times are also indicated. When used as a pretreatm

and fixed by immersion fixation in 10% buffered zinc formalin.For each rat, the contralateral “normal” side of the brain wasused as a control for normal brain tissue. Seven-micrometer-thick paraffin-embedded sections were stained with hematox-ylin and eosin or immunostained for the von Willebrand factor(vWF) (polyclonal rabbit anti-human; Dako) and the vascularendothelial growth factor (VEGF) (polyclonal IgG rabbit;GeneTex, Inc.).

Results

PBN effect on glioma morphology and growth pattern

Tumor growth curves depicted the normal growth patternsof the implanted C6 cells (Fig. 2A), which had an averagesurvival time of 23±3 days. These gliomas were classified to be

ime: untreated (diamonds), pretreated with PBN 5 days before cell implantationtive gliomas that closely resembled average (untreated group) or extreme casessymbols) or negatively (full symbols) to treatment are shown. For each animal,ion) were considered for Gompertz fitting and for the determination of specific-pretreated, and post-tumor-treated C6 gliomas). Data points not represented bynt on tumor regression. (B) Specific growth rates from each treatment regimeetween the C6 glioma untreated group and the PBN pretreated group. Averageent, PBN was found to dramatically reduce glioma growth rates.

Fig. 3. Representative T2-weighted images (from i to iii) and histology H&E staining (iv) of C6 gliomas from the different treatment groups: (A) untreated and PBN-treated starting (B) 5 days before or (C) 14 days after cell implantation. (A) (i-iii) Untreated C6 glioma at respectively 7, 10, and 17 days after cell implantation. (B)(i-iii) C6 glioma continuously treated with PBN starting 5 days before cell implantation at 7, 21, and 36 days. (C) (i-iii) C6 glioma treated with PBN once the tumorreached 50 mm3 in size and undergoing regression at 7, 16, and 35 days. (iv) For each case, the H&E stain (×4) of the tissue region outlined in (iii). PBN seemed toinduce total tumor regression when used as a pretreatment. Once tumors were well established, regression was only partial, with the brain tissue still appearinginjured.

Fig. 4. Percentage survival plots for the different treatment groups: C6 glioma-bearing untreated rats, PBN-pretreated animals, and PBN post-tumor-treatedrats. PBN seems to be an effective anti-glioma prophylactic drug and also seemsto have moderate effects on well-established tumors.

67S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

medium-grade tumors (grade II–III), presenting necroticcenters, some mitotic cells, and a diffusively infiltrative nature,as seen with MRI and histology (Fig. 3).

PBN treatment clearly modified tumor morphology andgrowth, resulting in partial or even complete regression ofthe glioma (Figs. 2 and 3). Administration of PBN 5 daysbefore cell implantation was found to be the most effective inretarding tumor growth (specific growth rate 0.05±0.007day−1; Td=7.4 days), compared to the posttumor treatment(specific growth rate 0.14±0.068 day−1; Td=3.3 days) (Fig.2B). Pretreated gliomas reached a maximum volume of only60 mm3 (Fig. 2A). We observed by MRI that the tumorsremained small in volume and completely regressed by 40days, with the brain tissue being nearly intact. This observationwas confirmed by histology (Fig. 3). Survival in this treatmentgroup reached 80% (Fig. 4) as four of five rats were still aliveat the end of the experiment (40 days after cell implantation).One rat died at the end of a MRI session (at 16 days after cellimplantation) from complications due to anesthesia. Bycomparison, none of the untreated C6 glioma-bearing ratssurvived beyond 19–26 days.

There was an overall 40% survival when PBN was used onwell-established tumors, as two of the five rats from thistreatment group survived up to 40 days (Fig. 4). Three of fiveanimals presented a normal tumor growth, with an averagesurvival time of 19 days, comparable to the untreated rats.Interestingly, two animals underwent partial regression of the

glioma, with an observed maximum volume of 270 mm3 (Figs.2A and 3). However, no effect was observed on the tumorspecific growth rate (Fig. 2B), which was similar to those in thecontrol group (0.14±0.028 day−1; Td=2.6 days). For these twoposttreated rats, MR images and histology indicated thatgliomas only partially recessed after 40 days, with the braintissue presenting a necrotic center (Fig. 3).

68 S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

PBN effect on glioma angiogenic behavior

Angiogenesis associated with the C6 gliomas was assessedfrom vasculature projections, quantification of blood volumes,and immunostaining for proangiogenic factors. For comparison(Fig. 5), quantification of the ROI blood volume for animalsimplanted with primary astrocytes instead of C6 cells was usedas a sham-control baseline. No change in vasculature wasobserved for these animals (final blood volume ratio 0.98±0.072, Fig. 5B).

C6 gliomas were found to induce a relative increase in bloodvolume (final blood volume ratio 1.26±0.157, Fig. 5B) in thetumor ROI that seemed to correlate with tumor growth. Thevasculature projections emphasized the preference of C6

Fig. 5. (A) Blood volume ratios of representative brains for each treatmentgroup: C6 gliomas untreated, PBN pretreated, and PBN post-tumor treated.Graphic lines are a linear representation between points. A sham-controlbaseline for blood volume ratio is provided by primary astrocyte-injected rats(“CTRL”). Blood volume ratios were calculated for each time point andexpressed relative to the first day of MRI assessment for each animal. MRAwasable to detect that C6 gliomas had a moderate angiogenic behavior, as well asmonitoring the effects of PBN on the vasculature in the early stages of gliomaprogression. (B) Blood volume ratios obtained at the last MRI time point (at 40days after cell implantation or just before the death of the rat) for primaryastrocyte-injected animals (CTRL) and for C6 gliomas with different PBNtreatment regimes (means±standard deviation). *pb0.05 by Student's two-tailed t test, between the CTRL group and the untreated C6 glioma group, aswell as between the untreated C6 glioma group and the PBN-pretreated group.

gliomas to use preexisting blood vessels for their nutrientsources, with vessels becoming longer (50±15% increase),wider (30±5% increase in diameter), and more meandering inthe tumor area (Fig. 6A). For animals that underwent PBNpretreatment, the blood volume ratios never increased beyond 1(final blood volume ratio 0.79±0.144, Fig. 5B), with the vesselsappearing to reduce in size over time (50±8% decrease indiameter) (Fig. 6B). For animals that underwent post-tumorPBN treatment, there were two distinct groups. For those ratsthat did not show any effect of PBN on tumor growth, vascularalterations comparable to those seen in untreated rats wereobserved, with blood volume ratios increasing in a similarfashion. The second group consisted of rats that presentedglioma regression, with a correlated decrease in blood volumeratio and in vessel length (20±7% decrease) (Figs. 5 and 6C).However, this was not as dramatic as observed in pretreated rats,as the general angiogenic behavior of this treatment group wasvery similar to that from untreated rats (Fig. 5). Blood vesselsalso remained at a normal thickness (Fig. 6C) rather thandecreasing as observed with the PBN-pretreatment group.

The moderate angiogenic behavior of C6 gliomas as shownby MRA data was also revealed by immunostaining assays forthe proangiogenic factor VEGF and a marker of endothelial cellproliferation, vWF (Fig. 7). C6 gliomas showed moderateincreased expression of VEGF, which was found to bepredominantly around the necrotic areas and on the border ofthe tumor, as well as an increase in vWF (Fig. 7B). Theefficiency of PBN pretreatment to prevent vascular alterations,as detected by MRA, was also confirmed by these immunos-taining assays, in which PBN-pretreated brain tissue did notexpress either VEGF or vWF and was found to be very similarto normal brain tissue (Figs. 7A and 7C). Rats that wereunsuccessfully treated posttumor had normal VEGF and vWFlevels, which resembled the untreated C6 glioma group (Figs.7B and 7E). Finally, posttreated rats that underwent tumorregression exhibited low to moderate VEGF levels and weakvWF expression (Fig. 7D).

Discussion

Without a contrast agent, MRA provided an accurate view ofthe tumor-related blood vessels in our glioma model. Measure-ment of blood vessels with diameters greater than 100 μm wassufficient to follow the vascular alterations induced by C6gliomas and establish the ability of PBN to prevent these earlyalterations (as observed with MRA) and the generation of newcapillaries (as revealed by vWF immunostaining). This methodcould be easily applied to characterize any tumor model andprovide a noninvasive and in vivo assessment of angiogenesis.Indeed, this assessment is currently done in cancer patientsusing dynamic contrast-enhanced MRI, dynamic contrast-enhanced computed tomography (CT) [17], or immunohisto-chemistry methods (microvessel density assessment developedbyWeidner et al. [18]). These MRI or CT techniques require theuse of contrast agents, and parameters such as the size of theagent, its extravasation, and the exchange constants of the agentwithin the vascular system must be identified using relatively

Fig. 6. 2D vasculature projections in the horizontal (i and iii) and in the transverse (ii and iv) planes, for early (i and ii) and late (iii and iv) time points, for eachtreatment group: C6 gliomas (A) untreated, (B) PBN pretreated, and (C) post-tumor treated. The early MRI time point was taken 7 days after cell implantation, and thelate time point corresponded to day 15 (A) or 35 (B and C). The whole ROI is represented, as well as the tumor boundaries (circled). The arrows depict the vesselsmostly affected by vascular alterations (A) or by PBN (B and C).

69S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

complex models. Furthermore, only relative blood volumes canbe obtained, whereas MRA allowed us to calculate absolutevolumes (for blood vessels with diameter greater than 100 μm)within the rat brain. Finally, MRA is noninvasive and the patientcan be examined as often as needed, which is difficult withbiopsies.

The angiogenic behavior of any tumor is closely related to itsmalignant potential, and it is widely known that dramaticallyincreased vascular proliferation is a trademark of all types ofmalignant cancers [6,19]. Precise characterization of the tumorvascular structure could provide noninvasive and in vivodiagnosis for angiogenesis and be used to monitor treatment, asthe delivery of any therapeutic compound relies upon thenetwork supplying the tumor [19].

Currently, investigators tend to use perfusion imaging toidentify tumor type and grade [20–22], as well as to followantiangiogenic treatment responses in gliomas [23,24]. Thismethod relies on the use of contrast agents and their effect ontissue relaxation times, which only indirectly characterize tissuevasculature, whereas MRA provides a direct assessment ofblood water molecules. To our knowledge, only Brubaker et al.

have used MRA to get a precise view of tumor-relatedvasculature in genetically engineered mice and were able tocompute vessel radius and density [25]. This report representsthe first in vivo assessment of absolute blood volume in a ratglioma model and correlation of the MRA results with VEGFand vWF immunostaining assays.

OurMathematica-based programwas able to quantitate bloodvolumes in carefully chosen ROIs, as well as to provide a precisedepiction of the alterations in vasculature induced by a glioma. Itwas also established that MRA results could be correlated withimmunostaining assays. For example, the increase in bloodvolume (accompanied by an increase in vessel diameter andlength) exhibited by C6 gliomas as shown in Fig. 5 wasaccompanied by increases in the level of vWF (a marker ofendothelial cell proliferation) and VEGF, both specific andrepresentative markers of tumor angiogenesis. 2D vasculatureprojections also indicated than C6 gliomas tended to preferen-tially use preexisting blood vessels for their nutrient source. Thesame method as the one used to compute blood volume could beimplemented to determine vessel length, diameter, and tortuos-ity, therefore providing further information on the tumor an-

Fig. 7. Immunostaining assays for vWF (i; original magnification×10) and VEGF (ii; original magnification×40) for (A) normal brain tissue, (B) untreated C6 glioma,and C6 glioma (C) pretreated with PBN and post-tumor treated with PBN (D, in presence, and E, in absence of tumor regression). The arrows show examples ofpositive staining.

70 S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

giogenic behavior. The method used in this paper to estimatevessel length and diameter is an estimation, as it was based on 2Dprojections and did not take into consideration the angles of thevessels and their 3D path. We are currently working on thisimprovement, by which parameters such as diameter and lengthcould be extracted from a 3D data set.

PBN is a highly efficient antioxidant that has been found tohave neuroprotective [7], anti-inflammatory [9], anti-ischemic[26], and anticarcinogenic properties [10]. These characteristicscoupled with its ability to freely cross the blood–brain barrier,being a species both lipophilic and hydrophilic, makes PBN aninteresting candidate therapeutic drug to be studied in a ratglioma model. Two different treatment regimes were imple-mented in this study. A pretreatment started 5 days before cellimplantation was used as a prophylactic approach. Anothergroup of animals was treated with PBN when the gliomas were50 mm3 in size, i.e., well established and vascularized, whichprovided us with a clinically relevant post-tumor treatmentregime.

This report is the first to show that PBN is an effectiveprophylactic agent and a possible anti-glioma agent whentumors are presented, and it is the first in vivo assessment ofthe effects of PBN on tumor growth and angiogenesis, asshown respectively by MRI and MRA. Indeed, PBN has beenfound to slow down tumor growth when administered as apretreatment, decreasing the glioma specific growth rate from0.14±0.028 day−1 (untreated C6 gliomas) to 0.05±0.007day−1 (pretreated animals) in the early growth phase, and tobe able to induce complete or partial glioma regressionaccording to the treatment regime. PBN was also able toprevent the generation of new blood vessels and decrease thelevel of expression of the proangiogenic factor VEGF in earlystages of tumor progression, as found by MRA andimmunostaining assays. The best current treatment situationis surgical removal of the tumor followed by radiation therapyand/or chemotherapy. Because PBN is most efficient in theearly stages of glioma progression, we propose that thistherapeutic compound could be administered to the patient

71S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

after surgical resection of the tumor or after radiation therapy,because it could prevent the reestablishment and growth ofsurvivor tumor cells.

It is known that a relationship exists between carcinogenesisand inflammatory mediators [27] and that prostaglandin E2

(produced from arachidonic acid by the cyclooxygenase 2(COX2) enzyme) promotes cell proliferation and invasionthrough the action of the E-prostanoid receptors [28].Antioxidants have the ability to down-regulate the nuclearfactor κB (NF-κB) [29], which regulates COX2 expression[30], and PBN has been found to disrupt COX2 activity in a ratliver carcinogenesis model [10]. It is highly probable that PBNcould inhibit tumor cell proliferation in the early stages ofglioma progression by inhibiting COX2 expression promotedby NF-κB.

It is also widely known that the generation of new bloodvessels that help the transport of nutrients and the removal ofwaste products is required for tumor growth and spread [5]and that high-grade gliomas generally present an increasedvascular network [6]. This process may be promoted byinducible nitric oxide synthase (iNOS)-generated nitric oxide(NO) [31,32]. Constitutively expressed NO has major roles invarious physiological processes, such as a neurotransmitter,blood flow regulator, or mediator of host defense mechanisms[33], whereas iNOS is responsible for high NO productioninvolved in various pathological conditions [34]. Forexample, heightened iNOS expression has often beenshown in human brain tumors [35] and is strongly correlatedwith the degree of malignancy of a carcinoma [36]. NO andVEGF also tightly cooperate to promote angiogenesis.Indeed, NO has been shown to induce VEGF expression invitro in glioblastoma and hepatocellular carcinoma cells [37],and NOS inhibition strongly prevents VEGF-induced angio-genesis [38], by decreasing VEGF-mediated endothelial cellmigration and proliferation [39], as well as vascularhyperpermeability [40]. However, PBN is able to suppressiNOS induction and by doing so NO production [8] andVEGF expression, as revealed by our immunostaining assaysshowing decreased VEGF expression with PBN treatment.Indeed, cytokines activate NF-κB, which promotes iNOSexpression [41], and PBN has been found to down-regulatecytokine expression in a LPS mouse model [42]. Antioxidantsare also known to down-regulate NF-κB [29]. PBN mayprevent angiogenesis in the early stages of tumor progressionby inhibiting cytokine-regulated and NF-κB-regulated NOproduction.

Acknowledgments

The authors acknowledge Dr. Yasvir Tesiram for his help inestablishing optimal MR parameters, Dr. Molina Mhatre andDr. Yujun Pan for their help and their advice regarding thesurgical procedures for the intracerebral implantation of gliomacells, as well as Ms. Melinda West for helping with the brainextraction procedure and the primary astrocyte culture. Thisproject was supported by National Institutes of Health GrantP20RR016478.

References

[1] Central Brain Tumor Registry of the United States. Statistical report:primary brain tumors in the United States, 1998–2002. Chicago: CBTRUS;2005.

[2] Zhang, X.; Wu, J.; Gao, D.; Fei, Z.; Qu, Y.; Jing, J. Development of a ratC6 brain tumor model. Chin. Med. J. (Engl.) 115:455–457; 2002.

[3] Grobben, B.; De Deyn, P. P.; Slegers, H. Rat C6 glioma as experimentalmodel system for the study of glioblastoma growth and invasion. CellTissue Res. 310:257–270; 2002.

[4] Sibenaller, Z. A.; Etame, A. B.; Ali, M. M.; Barua, M.; Braun, T. A.;Casavant, T. L.; Ryken, T. C. Genetic characterization of commonly usedglioma cell lines in the rat animal model system. Neurosurg. Focus 19:E1;2005.

[5] Goldbrunner, R. H.; Wagner, S.; Roosen, K.; Tonn, J. C. Models forassessment of angiogenesis in gliomas. J. Neurooncol. 50:53–62; 2000.

[6] Plate, K. H.; Breier, G.; Weich, H. A.; Risau, W. Vascular endothelialgrowth factor is a potential tumour angiogenesis factor in human gliomasin vivo. Nature 359:845–848; 1992.

[7] Endoh, H.; Kato, N.; Fujii, S.; Suzuki, Y.; Sato, S.; Kayama, T.; Kotake, Y.;Yoshimura, T. Spin trapping agent, phenyl N-tert-butylnitrone, reducesnitric oxide production in the rat brain during experimental meningitis.Free Radic. Res. 35:583–591; 2001.

[8] Miyajima, T.; Kotake, Y. Spin trapping agent, phenyl N-tert-butyl nitrone,inhibits induction of nitric oxide synthase in endotoxin-induced shock inmice. Biochem. Biophys. Res. Commun. 215:114–121; 1995.

[9] Sang, H.; Wallis, G. L.; Stewart, C. A.; Kotake, Y. Expression of cytokinesand activation of transcription factors in lipopolysaccharide-administeredrats and their inhibition by phenyl N-tert-butylnitrone (PBN). Arch.Biochem. Biophys. 363:341–348; 1999.

[10] Nakae, D.; Kotake, Y.; Kishida, H.; Hensley, K. L.; Denda, A.; Kobayashi,Y.; Kitayama, W.; Tsujiuchi, T.; Sang, H.; Stewart, C. A.; Tabatabaie, T.;Floyd, R. A.; Konishi, Y. Inhibition by phenyl N-tert-butyl nitrone of earlyphase carcinogenesis in the livers of rats fed a choline-deficient, L-aminoacid-defined diet. Cancer Res. 58:4548–4551; 1998.

[11] Hensley, K.; Abdel-Moaty, H.; Hunter, J.; Mhatre, M.; Mou, S.; Nguyen,K.; Potapova, T.; Pye, Q. N.; Qi, M.; Rice, H.; Stewart, C.; Stroukoff, K.;West, M. Primary glia expressing the G93A-SOD1 mutation present aneuroinflammatory phenotype and provide a cellular system for studies ofglial inflammation. J. Neuroinflammation 3:2; 2006.

[12] Kobayashi, N.; Allen, N.; Clendenon, N. R.; Ko, L. W. An improved ratbrain-tumor model. J. Neurosurg. 53:808–815; 1980.

[13] Janzen, E.G., Haire, D.L., Tanner, D.D. eds. Two decades of spin trapping.JAI Press, Greenwich; 1990.

[14] Rygaard, K.; Spang-Thomsen, M. Quantitation and Gompertzian analysisof tumor growth. Breast Cancer Res. Treat. 46:303–312; 1997.

[15] Demicheli, R.; Foroni, R.; Giuliani, F.; Savi, G. Analysis of free growthand post-treatment regrowth of C3H mammary carcinoma in individualmice. Cell Tissue Kinet. 19:471–476; 1986.

[16] Steel, G. G. Growth kinetics of tumors. Clarendon Press, Oxford; 1977.[17] Goh, V.; Padhani, A. R.; Rasheed, S. Functional imaging of colorectal

cancer angiogenesis. Lancet Oncol. 8:245–255; 2007.[18] Weidner, N.; Semple, J. P.; Welch, W. R.; Folkman, J. Tumor angiogenesis

and metastasis-correlation in invasive breast carcinoma. N. Engl. J. Med.324:1–8; 1991.

[19] Acker, T.; Plate, K. H. Role of hypoxia in tumor angiogenesis-molecularand cellular angiogenic crosstalk. Cell Tissue Res. 314:145–155; 2003.

[20] Choyke, P. L.; Dwyer, A. J.; Knopp, M. V. Functional tumor imaging withdynamic contrast-enhanced magnetic resonance imaging. J. Magn. Reson.Imaging 17:509–520; 2003.

[21] Provenzale, J. M.; Wang, G. R.; Brenner, T.; Petrella, J. R.; Sorensen, A. G.Comparison of permeability in high-grade and low-grade brain tumorsusing dynamic susceptibility contrast MR imaging. Am. J. Roentgenol.178:711–716; 2002.

[22] Knopp, E. A.; Cha, S.; Johnson, G.; Mazumdar, A.; Golfinos, J. G.;Zagzag, D.; Miller, D. C.; Kelly, P. J.; Kricheff, I. I. Glial neoplasms:dynamic contrast-enhanced T2

*-weighted MR imaging. Radiology 211:791–798; 1999.

72 S. Doblas et al. / Free Radical Biology & Medicine 44 (2008) 63–72

[23] Moffat, B. A.; Chen, M.; Kariaapper, M. S.; Hamstra, D. A.; Hall, D. E.;Stojanovska, J.; Johnson, T. D.; Blaivas, M.; Kumar, M.; Chenevert, T. L.;Rehemtulla, A.; Ross, B. D. Inhibition of vascular endothelial growthfactor (VEGF)-A causes a paradoxical increase in tumor blood flow andup-regulation of VEGF-D. Clin. Cancer Res. 12:1525–1532; 2006.

[24] Gossmann, A.; Helbich, T. H.; Kuriyama, N.; Ostrowitzki, S.; Roberts, T.P.; Shames, D. M.; van Bruggen, N.; Wendland, M. F.; Israel, M. A.;Brasch, R. C. Dynamic contrast-enhanced magnetic resonance imaging asa surrogate marker of tumor response to anti-angiogenic therapy in axenograft model of glioblastoma multiforme. J. Magn. Reson. Imaging 15:233–240; 2002.

[25] Brubaker, L. M.; Bullitt, E.; Yin, C.; Van Dyke, T.; Lin, W. Magneticresonance angiography visualization of abnormal tumor vasculature ingenetically engineered mice. Cancer Res. 65:8218–8223; 2005.

[26] Vergely, C.; Renard, C.; Moreau, D.; Perrin-Sarrado, C.; Roubaud, V.;Tuccio, B.; Rochette, L. Effect of two new PBN-derived phosphorylatednitrones against postischaemic ventricular dysrhythmias. Fundam. Clin.Pharmacol. 17:433–442; 2003.

[27] Hong, W. K.; Sporn, M. B. Recent advances in chemoprevention of cancer.Science 278:1073–1077; 1997.

[28] Mann, J. R.; Backlund, M. G.; Du Bois, R. N. Mechanisms of disease:inflammatory mediators and cancer prevention. Nat. Clin. Pract. Oncol.2:202–210; 2005.

[29] Schreck, R.; Meier, B.; Mannel, D. N.; Droge, W.; Baeuerle, P. A.Dithiocarbamates as potent inhibitors of nuclear factor kappa B activationin intact cells. J. Exp. Med. 175:1181–1194; 1992.

[30] Schmedtje, J. F.; Jr.; Ji, Y. S.; Liu, W. L.; Du Bois, R. N.; Runge, M. S.Hypoxia induces cyclooxygenase-2 via the NF-kappaB p65 transcriptionfactor in human vascular endothelial cells. J. Biol. Chem. 272:601–608;1997.

[31] Ambs, S.; Merriam, W. G.; Ogunfusika, M. O.; Bennett, W. P.; Ishibe, N.;Hussain, S. P.; Tzeng, E. E.; Geller, D. A.; Billiar, T. R.; Harris, C. C. p53and vascular endothelial growth factor regulate tumor growth of NOS2-expressing human carcinoma cells. Nat. Med. 4:1371–1376; 1998.

[32] Andrade, S. P.; Hart, I. R.; Piper, P. J. Inhibitors of nitric oxide synthaseselectively reduce flow in tumor-associated neovasculature. Br. J.Pharmacol. 107:1092–1095; 1992.

[33] Bredt, D. S.; Snyder, S. H. Nitric oxide: a physiologic messenger molecule.Annu. Rev. Biochem. 63:175–195; 1994.

[34] Moncada, S.; Palmer, R. M.; Higgs, E. A. Nitric oxide: physiology,pathophysiology, and pharmacology. Pharmacol. Rev. 43:109–142; 1991.

[35] Cobbs, C. S.; Brenman, J. E.; Aldape, K. D.; Bredt, D. S.; Israel, M. A.Expression of nitric oxide synthase in human central nervous systemtumors. Cancer Res. 55:727–730; 1995.

[36] Lala, P. K.; Orucevic, A. Role of nitric oxide in tumor progression: lessonsfrom experimental tumors. Cancer Metastasis Rev. 17:91–106; 1998.

[37] Chin, K.; Kurashima, Y.; Ogura, T.; Tajiri, H.; Yoshida, S.; Esumi, H.Induction of vascular endothelial growth factor by nitric oxide in humanglioblastoma and hepatocellular carcinoma cells. Oncogene 15:437–442;1997.

[38] Ziche, M.; Morbidelli, L.; Choudhuri, R.; Zhang, H. T.; Donnini, S.;Granger, H. J.; Bicknell, R. Nitric oxide synthase lies downstreamfrom vascular endothelial growth factor-induced but not basic fibroblastgrowth factor-induced angiogenesis. J. Clin. Invest. 99:2625–2634;1997.

[39] Uhlmann, S.; Friedrichs, U.; Eichler, W.; Hoffmann, S.; Wiedemann, P.Direct measurement of VEGF-induced nitric oxide production bychoroidal endothelial cells. Microvasc. Res. 62:179–189; 2001.

[40] Wu, H. M.; Huang, Q.; Yuan, Y.; Granger, H. J. VEGF induces NO-dependent hyperpermeability in coronary venules. Am. J. Physiol. 271:H2735–2739; 1996.

[41] Xie, Q. W.; Whisnant, R.; Nathan, C. Promoter of the mouse geneencoding calcium-independent nitric oxide synthase confers inducibilityby interferon gamma and bacterial lipopolysaccharide. J. Exp. Med.177:1779–1784; 1993.

[42] Pogrebniak, H. W.; Merino, M. J.; Hahn, S. M.; Mitchell, J. B.; Pass, H. I.Spin trap salvage from endotoxemia: the role of cytokine down-regulation.Surgery 112:130–139; 1992.