Subclinical Photodynamic Therapy TreatmentModifies the Brain Microenvironment andPromotes Glioma GrowthANA C. deCARVALHO,1 XUEPENG ZHANG,1 CYNTHIA ROBERTS,1 FENG JIANG,1

STEVEN N. KALKANIS,2 XIN HONG,2 MEI LU,3 AND MICHAEL CHOPP1,4*1Departments of Neurology, Henry Ford Health Sciences Center, Detroit, Michigan2Department of Neurosurgery, Henry Ford Health Sciences Center, Detroit, Michigan3Department of Biostatistics and Research Epidemiology, Henry Ford Health Sciences Center, Detroit, Michigan4Department of Physics, Oakland University, Rochester, Michigan

KEY WORDSorthotopic glioma model; photodynamic therapy; reactivegliosis; tumor growth

ABSTRACTPhotodynamic therapy (PDT) has been clinically investi-gated as an adjuvant local therapy for brain tumors. Thera-peutic interventions intended to promote tumor cell deathcan also promote changes in the tumor microenvironmentthat could favor tumor growth. We have previously shownthat PDT can activate pro-angiogenic factors in the normalrodent brain. This study seeks to further elucidate theeffects of subtherapeutic doses of Photofrin-PDT on normalbrain and to establish a mouse model for studying gliomaprogression in an environment modified by oxidative stress.Photofrin was administered to nude mice, and a defined in-tracranial area was illuminated with laser to deliver an op-tical dose equivalent to 80 J/cm2. Three and 7 days afterPDT, mice were sacrificed and brains were fixed and ana-lyzed by immunohistochemistry. PDT treatment resulted intransient increase in cell proliferation, associated with a ro-bust activation of astrocytes and microglia in the treatedregion, without causing substantial cell death. To test howthis modified environment would affect glioma growth,human glioblastoma U87 cells were implanted in the PDT-treated hemisphere or in the control brain subjected tosham surgery. Significantly larger tumors were observed af-ter 3 weeks in the PDT treated brains relative to controltreatment. Our results indicate that subclinical Photofrin-PDT locally alters the brain homeostasis without inflictingsignificant disruption to the tissue architecture, providing amodel to study the effects of the microenvironment on gli-oma growth, with implications for the optimization of theclinical use of PDT for brain tumors. VVC 2007 Wiley-Liss, Inc.

INTRODUCTION

Central nervous system tumors occur at an annualincidence of 18,500 in the US, were responsible for12,760 deaths in 2005 and are a leading cause of cancermortality for people younger than 40 years (Jemal et al.,2005). Malignant gliomas, the most frequent and aggres-sive of these cancers (Gurney and Kadan-Lottick, 2001),are refractive to standard treatments, such as cytoreduc-tive surgery, radiation, and chemotherapy, creating the

need for the development of additional treatments.Photodynamic therapy (PDT) is currently a treatmentoption or a potential adjuvant for several neoplastic andnon-neoplastic diseases (Dolmans et al., 2003), includingbrain cancer (Muller and Wilson, 2006; Stylli and Kaye,2006).

For internal malignancies, the photodynamic action isachieved by systemic administration of a photosensitizerand local exposure to light. After excitation with visiblelight, singlet oxygen, and other reactive oxygen species(ROS) are generated from tissue oxygen, leading to oxi-dative damage to cellular components and cell death(Almeida et al., 2004; Dolmans et al., 2003). Selectiveuptake of photosensitizer by tumor cells and associatedvasculature is one of the prerequisites for tumor eradica-tion with minimum toxicity for the surrounding hosttissue. Photofrin (porfimer sodium) has been the mostcommonly used photosensitizer in diverse clinical appli-cations, including brain tumor. Clinical use of PDT asan adjuvant treatment for gliomas has been investigatedfor more than 20 years, and commonly consists of preo-perative administration of Photofrin and subsequentillumination of the resection cavity by long wavelengthlaser light (Muller and Wilson, 1985). Promising clinicaloutcome as a result of high optical dose application aftermaximal resection has been recently reported (Stylliet al., 2005). Because PDT targets neoplastic cells invad-ing into the normal brain, evaluating its effects on thebrain parenchyma is essential for the optimization ofthe treatment. Results from clinical studies show selec-tive accumulation of Photofrin in tumors, in relation tothe normal brain, yet significant photosensitizer uptakeby peritumoral brain tissue has also been reported (Hillet al., 1990).

Photofrin-PDT effects on the normal brain are dose de-pendent. High therapeutic doses of Photofrin-PDT result

Grant sponsor: NIH; Grant numbers: CA043892, CA100486.

*Correspondence to: Michael Chopp, Ph.D., Department of Neurology, E&RBuilding Room 3056, Henry Ford Hospital, 2799 West Grand Boulevard, Detroit,MI 48202, USA. E-mail: chopp@neuro.hfh.edu

Received 10 October 2006; Revised 26 April 2007; Accepted 30 April 2007

DOI 10.1002/glia.20525

Published online 5 June 2007 in Wiley InterScience (www.interscience.wiley.com).

GLIA 55:1053–1060 (2007)

VVC 2007 Wiley-Liss, Inc.

in brain toxicity in clinical and animal studies (Chen etal., 1996; Lilge and Wilson, 1998; Lilge et al., 2000; Mul-ler and Wilson, 1990). High doses of PDT result in vas-cular damage, leading to compensatory growth of bloodvessels in rodent models (Jiang et al., 2004), mediatedby the hypoxia - Hif1a- vascular endothelial growth fac-tor (VEGF) pathway (Ferrario et al., 2000). In contrast,sublethal doses of PDT, such as those received by tissuessurrounding the focally treated area, have the potentialto modify the brain environment in the absence of exten-sive cell death. For example, we have previouslyobserved increased VEGF expression in the rodent brainafter very low doses of PDT (Zhang et al., 2005). Here,we report that the normal mouse brain responses to sub-therapeutic doses of Photofrin-PDT include localizedincrease in mitotic activity, activation of astrocytes andmicroglia and VEGF upregulation in the reactive astro-cytes. Notably, these alterations in the host tissue inresponse to the PDT treatment favored tumor growth inan orthotopic mouse model for glioma. Our resultsunderscore the need for optimization of PDT protocolsleading to selective killing of tumor cells while attenuat-ing effects on the host tissue environment that couldfavor tumor recurrence.

MATERIALS AND METHODSMouse Model for Intracranial

Photodynamic Therapy

Athymic nude mice were obtained from the NationalCancer Institute (Frederick, MD). All experimental pro-tocols were approved by the Institutional Animal Careand Use Committee of the Henry Ford Health SciencesCenter.

PDT treatment of mouse brain was performed as pre-viously described (Zhang et al., 2005). Briefly, Photofrin(Axcan Scandipharm) was systemically administered tonude mice at a dose of 2 mg/kg. After 24 h, mice wereanesthetized and fixed in a stereotaxic device, the skullwas exposed, a 5 mm diameter craniotomy was drilledover the right hemisphere 2.5 mm to the midline and2.0 mm anterior to the bregma. Light (635 nm) at a flu-ency rate of 260 mW/cm2 was delivered by a diode laser(University Health Network, Toronto, Canada) throughthe craniotomy, to deliver 80 J/cm2 optical dose. Controlanimals were submitted to sham surgery, with no expo-sure to light.

Experimental Glioma

U87 human glioblastoma cells (ATCC, Manassas, VA)were maintained at 37�C in a humidified, 5% carbondioxide atmosphere, in growth medium consisting ofDulbecco’s modified eagle medium (DMEM) supplemen-ted with 10% fetal bovine serum and 100 U/mL penicil-lin G sodium, 100 lg/mL streptomycin sulfate (GibcoInvitrogen Cell Culture). Cells were harvested andviable cell count was obtained by Trypan blue exclusion.

Three or 7 days after PDT treatment or sham surgery,U87 cells were implanted intracranially through thesame craniotomy used for laser exposure. The mice wereanaesthetized and fixed in a stereotaxic device, a Hamil-ton syringe was lowered to a depth of 2.5 mm beneaththe dura so as to inject 3 3 105 U87 cells in 5 lL volumeover a 5 min interval into the subcortical white matter,just ventral to the area affected by the PDT. Animalswere monitored daily and sacrificed 3 weeks post im-plantation.

Tissue Preparation, Tumor Volume Measurement,and Immunohistochemistry

The brains were removed, fixed, and embedded in par-affin. Serial 6 lm thick sections from approximately ev-ery 0.5 mm of tissue were stained with hematoxylin andeosin (H&E), or analyzed by immunohistochemistry. Forevaluation of the tumor size, the tumor area was meas-ured in each digitalized coronal section image, and thevolume was calculated using the Global Lab Image anal-ysis program (Data Translation, Marlvoro MA).

Primary antibodies used were: Ki76 (Abcam), GFAPand von Willebrand factor (vWF, Dako), vimentin andVEGF (Santa Cruz Biotechnology). Peroxidase conju-gated Isolectin B4 was used as a microglial cell marker(Sigma). For bright field images, immunodetection wasperformed using the Elite Vector Stain ABC System(Vector Laboratories) and diaminobenzidine tetrachlor-ide (DAB) chromogen (Sigma). Nuclei were counter-stained with hematoxylin. Images were captured withan Olympus BX41 microscope, equipped with an Olym-pus DP70 camera. For fluorescent detection, FITC- orCy3-conjugated secondary antibodies (Jackson Immu-noResearch Laboratories) and DAPI nuclear stain wereused and images were captured with an epifluorescenceequipped Zeiss Axiophot2 microscope.

Quantitative analysis of digitalized images wasperformed using an image analysis program (MCID,Imaging Research). For each coronal section, four highmagnification fields (140 lm2) were considered within aregion extending 1.5 mm laterally from the midline and1.5 mm ventrally from the pial surface, in the PDT-trea-ted and contra-lateral hemispheres (Fig. 1A). The cellproliferation index was calculated by counting the num-ber of Ki67 positive nuclear profiles and expressed aspercentage of total hematoxylin- labeled nuclei. VEGFand IB4 positive cells were detected by DAB stainingand counted for each field. GFAP expression level wasdetected using a FITC-conjugated secondary antibodyand the immunoreactive area was calculated andexpressed as percentage of total area for the fieldsanalyzed.

Statistical Analysis

Data are represented as the mean 6 S.E, for n 5 5–15, as indicated. Values were compared by the Student’s

1054 deCARVALHO ET AL.

GLIA DOI 10.1002/glia

Fig. 1. Sublethal Photofrin-PDT induces gliosis and cell proliferation in the mouse brain. (A) Schematic coronal section showing visual fieldsused for quantitative analysis. (B,C). Representative sections stained with H&E, 7 days after PDT. The arrows indicate the focal area illuminatedby laser. No morphological changes were present for most samples (B); minor alterations in tissue architecture was observed for other sections(dashed line, C). (D) Quantification of GFAP expression in the PDT-treated and corresponding contralateral region. (E) Pronounced increase inGFAP (green) staining in a representative section. (F) Ki67 proliferation index in the PDT-treated region and corresponding contra-lateral hemi-spheres. (G) Representative Ki67 immunolabeled section 3 days after PDT, counterstained with hematoxylin. (H,I) Double labeling of PDT-treatedarea for Ki67 (red) and GFAP (green) (H) or IB4 (green) (I); nuclei were labeled with DAPI (blue). Arrows indicate colocalization of the prolife-ration marker with the astrocytic marker (H) or with the microglial/macrophage marker (I). (J) Quantification of IB4 positive cells in the PDT-treated and corresponding contra-lateral region, 3 and 7 days after PDT. Results represent mean 6 SE for n 5 5 animals. Asterisks, statisticallysignificant difference between PDT-treated and untreated hemispheres, (*) P < 0.001 and (**) P < 0.05. Scale, 200 lm (B,C,E,G) and 50 lM (H,I).

t-test or by one-way ANOVA, followed by multi-compari-son test, with P < 0.05 considered significant.

RESULTS

The optimization of PDT as an adjuvant treatment ofinvasive intracranial tumors depends in part on a betterunderstanding of the side-effects of the treatment on thehost brain tissue surrounding the targeted neoplasticcells. The goal of this study is to evaluate the PDTmediated activation of defense mechanisms in the nor-mal mouse brain and the effect of the modified microen-vironment on glioma growth.

Sublethal PDT Dose Induces Reactive Gliosisin the Mouse Brain

To study the effect of subclinical dose of Photofrin-PDT in the normal brain, we have selected a 80 J/cm2

optical dose. Although lower doses of PDT are therapeu-tic for certain malignancies such as superficial bladdercarcinoma (Skyrme et al., 2005), have an effect in thenormal rodent brain (Zhang et al., 2005) and in experi-mental tumors (Zhang et al., 2006), higher doses (>200J/cm2) are currently used clinically to treat brain tumors(Stylli et al., 2005) and other malignancies such as basalcell carcinoma (Oseroff et al., 2006). Therefore, the PDTregimen employed in this study is below clinical dosescurrently employed in the treatment of brain tumorswhile still having the potential to affect the brain tis-sues. Ten nude mice were divided into two groups of 5and sacrificed 3 and 7 days after intracranial PDT treat-ment. Brains were removed, fixed, and coronal sectionswere analyzed by immunohistochemistry. For quantita-tive studies, four high magnification fields from digitizedimages of the PDT-treated region and correspondingcontra-lateral area were considered (Fig. 1A).

Therapeutic doses of PDT are optimized to target tu-mor cells to death by apoptosis or necrosis (Almeidaet al., 2004; Oleinick and Evans, 1998). The sub-clinicalPhotofrin-PDT dose selected for this study did not resultin significant tissue damage. No change in tissue archi-tecture was observed for 70% of the treated brains (Fig.1B), while acellular regions were observed for 30% ofthe samples, restricted to a subpial area not larger than0.15 mm2, (Fig. 1C, dashed line). At the time points ana-lyzed, 3 and 7 days after PDT, no apoptotic cells weredetected in the PDT-treated brains, as assessed by insitu terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labeling (TUNEL) assay (datanot shown). We cannot rule out that limited cell deathcould take place at different time points.

Reactive astrogliosis is a common defensive responseby the central nervous system (CNS) to diverse insults.Reactive astrocytes are commonly recognized by hyper-trophy and increased levels of the astrocyte specific in-termediate filament protein GFAP. Extensive astrogliosisin response to the Photofrin-PDT treatment, as inferred

from the significant increase in GFAP immunoreactivity,was observed in the treated hemisphere in relation tothe contra-lateral side (Figs. 1D,E). Reactive astrogliosiswas consistently restricted to the grey matter, within anarea extending 1.1 mm ventrally from the pial surfaceand 1.5 mm laterally from the midline.

The astrocytic reaction was accompanied by a signifi-cant increase in cell proliferation in the treated region,detected by Ki67 immunostaining, 3 days after the Pho-tofrin-PDT treatment (Figs. 1F,G). After one week, thenumber of proliferating cells had returned to untreatedcontra-lateral levels (Fig. 1F), indicating that cell prolif-eration was an acute response to treatment. The mitoticactivity was observed in a smaller area within theregion affected by reactive gliosis, extending between 0.5and 0.8 mm from the pial surface. Colocalization studiesreveled that few proliferating cells (less than 15%) inthe PDT-treated area coexpressed the astrocytic markerGFAP (Fig. 1H, arrows), while approximately twice asmany cells co-labeled with the microglia/macrophagemarker, isolectin B4 (IB4) (Fig. 1I, arrows). In agree-ment with the microglial mitotic activity, a significantincrease in IB4-labeled cells in the PDT-treated areawas observed 3 and 7 days after treatment (Fig. 1J).These results suggest that the increase in cell prolifera-tion observed in response to PDT-treatment can be par-tially attributed to the activation of astrocytes andmicroglia. Similar microglial and astrocytic reactionwithin 1–3 days after mild indirect insult to the CNShas been reported (Raivich et al., 1999). Additionally, weobserved that between 10 and 26% of proliferating cellswere colabeled with the endothelial cell marker vWF;however, no increase in microvascular density wasobserved (not shown), in agreement with our previouswork (Zhang et al., 2005).

PDT Treatment of Normal Mouse Brain Resultsin Upregulation of VEGF

Increased secretion of cytokines and growth factors byneoplastic and host cells in the tumor microenvironmentcan drive tumor growth. VEGF expression is upregu-lated by diverse stimuli, including hypoxia and oxidativestress (Pages and Pouyssegur, 2005), which can occur asa result of PDT. Thus, upregulation of VEGF in endothe-lial (Schmidt-Erfurth et al., 2003; Zhang et al., 2005)and neoplastic cells (Ferrario et al., 2000) in response toPDT has been reported. Photofrin-PDT treatment signif-icantly increased the number of cells expressing VEGFin the mice brain 3 and 7 days after treatment (Figs.2A,B), relative to the contra-lateral untreated hemi-sphere (Figs. 2A,C), temporally coincident with reactivegliosis in the PDT-treated brain (Figs. 1D,E). Reactiveastrocytes have been shown to be an important source ofVEGF secretion in diverse neuropathologies, includingbrain tumors (Antunes et al., 2000; Salhia et al., 2000).Accordingly, double-labeling analysis revealed thatnearly all cells expressing detectable levels of VEGFcoexpressed the astrocytic marker GFAP (Figs. 2D–F),

1056 deCARVALHO ET AL.

GLIA DOI 10.1002/glia

at 3 days after treatment. At 7 days, between 70 and93% of VEGF positive cells coexpressed GFAP, while theremaining VEGF positive cells were associated with thevasculature. Spatially, VEGF positive cells were re-stricted to a 0.12 mm deep subpial region. Therefore,the majority of reactive astrocytes located deeper intothe cortex did not coexpress detectable levels of VEGF.Secreted VEGF by other cells, however, could occur atlevels below what can be detected by immunohistochem-istry. From these results, we conclude that astrocytesare an important source of VEGF upregulation followingPDT treatment; however, in our system only a small su-perficial subregion affected by reactive gliosis containedcells expressing detectable levels of VEGF.

PDT-Mediated Changes in the NormalMice Brain Favors Glioma Growth

The results above indicate that the subtherapeuticPhotofrin-PDT treatment of normal mouse brain did notresult in substantial tissue injury, yet it induced reactivegliosis. To test how this modified environment wouldaffect glioma growth, Photofrin was administered to 35nude mice. Twenty-four hours later a craniotomy wasmade to expose a defined brain region to a diode laser,exactly as described earlier. Control mice (n 5 15) werenot exposed to laser and the remaining animals (n 5 20)received the subtherapeutic dose of 80 J/cm2. Three or 7

days later, 3 3 105 U87 human glioma cells wereimplanted through the same craniotomy. Three weekslater the animals were sacrificed, brains were removedand fixed. Sequential coronal sections were obtained andtumor volumes for each animal were calculated asdescribed in the Materials and Methods section. Signifi-cantly larger tumors were observed in the PDT-treatedbrains in relation to control (P < 0.001) (Fig. 3). Thetime interval between PDT treatment and U87 cell im-plantation, 3 or 7 days, was not a factor, since no signifi-cant difference in tumor volumes was observed betweenthe two groups (P 5 0.16) (Fig. 3A). Low dose PDTtreatment (10 J/cm2) did not result in statistically signif-icant increase in tumor volumes relative to control (notshown). The glioma cells were injected at a position ven-tral to the PDT affected cortex region, and 54 and 90%of the tumor xenografts grew dorsally into the cortex ofcontrol and PDT treated brains, respectively (Figs.3C,D). For the control treated brains, the tumors thatexpanded into the cortex were larger than the tumorsrestricted to the white matter and striatum, meanvalues of 0.69 and 0.23 mm3, respectively, however, thedifference was not statistically significant (t-test, P 50.07). Importantly, tumors growing into the cortex ofPDT-treated animals (mean 4.43 mm3) were significantlylarger (P 5 0.02) than tumors growing into the cortex ofcontrol treated brains (mean 0.69 mm3), indicating thatthe modified environment was the main factor affectingtumor growth.

Fig. 2. Reactive astrocytes contribute to the PDT-mediated increase in VEGF positive cells. (A) Quantification of VEGF-positive cells in thePDT-treated and contra-lateral hemispheres, 3 and 7 days after PDT. Results represent mean 6 SE for n 5 5 animals, (*) P < 0.05 and (**) P <0.001. Representative image showing VEGF positive cells for the PDT treated area (B, arrows) and no signal for the corresponding contra-lateralregion (C). Images showing PDT-treated region labeled for GFAP (D), VEGF (E) and merged images (F); nuclei were stained with DAPI. Whitearrows point to cells co-expressing GFAP and VEGF. Scale, 30 lm.

1057GLIOMA PROGRESSION IN PDT-TREATED MICE BRAIN

GLIA DOI 10.1002/glia

Reactive Astrocytes and Microgliain the Experimental Glioma

Because PDT treatment concomitantly induced theactivation of astrocytes and microglia in the normalbrain and resulted in increased glioma volumes, weinvestigated the extent of the presence of astrocytes andmicroglia in the experimental glioma model employedfor this study. Reactive astrocytes in the peritumoralbrain has been previously characterized in other precli-nical models (Le et al., 2003; Nagano et al., 1993; Salhiaet al., 2000). The intermediate filament protein vimen-tin, a malignant glioma marker, is expressed in severalhuman glioma cell lines, including U87 (Rutka et al.,1999), therefore a species–specific anti-human vimentinmouse monoclonal antibody was used to specifically labelthe U87 cells in the tumor xenograft in double-labelingexperiments with the astrocytic marker GFAP and themicroglial/macrophage marker IB4 (Fig. 4). Reactiveastrogliosis in the peritumoral region (Figs. 4A,B) andtumor infiltrating astrocytes (Fig. 4B) were observed forall sections obtained from animals sacrificed 3 weeks af-ter U87 implantation. The anti-GFAP antibody usedrecognizes mouse and human protein, however, U87cells express very low levels of GFAP (Mandil et al.,

2001) and no colocalization of GFAP with vimentin inU87 cells was observed. Microglia/macrophage cells werealso located to the peritumoral regions and within thetumor mass (Figs. 4C,D). The difference in reactive glio-sis at the time of U87 implantation between PDT-treatedand control animals (Fig. 1) was not observed 3 weekslater when the animals were sacrificed, as extensive tu-mor associated reactive gliosis was observed in controland PDT-treated animals. These results show that inthe orthotopic glioma model used for this study, abun-dant host astrocytes and microglia/macrophage cells arepresent within the tumor microenvironment.

DISCUSSION

Injury or even mild indirect insults to the brain canresult in acute responses mediated by the surveillanceand repair mechanisms, such as inflammation, increasedmitotic activity, activation, and attraction of astrocytesand microglia to the site of injury, and induction ofangiogenesis. All these responses can influence eachother and affect tumor development and growth. Forexample, chronic tissue inflammation has been recog-nized as a risk factor for cancer development (Coussensand Werb, 2002) and development of brain cancer aftertrauma and radiation has been reported (Moorthy andRajshekhar, 2004; Salvati et al., 2003).

In the present study we investigated the acuteresponses to PDT elicited in the normal mouse brainand the consequence of this modified microenvironmentfor experimental glioma growth. In addition to directlytargeting neoplastic cells and tumor associated-vascu-lature, PDT can affect other host cells present in thetumor environment (Dolmans et al., 2003; Gomeret al., 2006; Korbelik, 2006). PDT directly targets tumorcells by activation of a cell associated photosensitizerand consequent formation of ROS, and occasionallydecreased oxygen levels (hypoxia), resulting from oxygenconsumption in the photoreaction. Subsequent cell deathby apoptosis or necrosis takes place if the cytotoxicthreshold is achieved. Conversely, sublethal doses ofPDT can lead to the upregulation of oxidative stress andhypoxia responsive genes, many of them involved in cellsurvival. PDT can also indirectly induce changes in themicroenvironment, such as the activation of immuneand inflammatory responses (Korbelik, 2006). Here wereport a significant activation of astrocytes and micro-glia in response to Photofrin-PDT, in the absence ofsignificant tissue damage. Astrocytosis is a commonreaction to a diverse range of brain insults, includinggliomas (Daumas-Duport et al., 1987; Le et al., 2003;Nagano et al., 1993; Nagashima et al., 2002) or meta-static brain tumors (Zhang and Olsson, 1997). Similarly,significant infiltration of glioma by microphage/microgliahas been reported (Watters et al., 2005). The conse-quences of the presence of reactive astrocytes and mic-roglia in the tumor microenvironment is not firmlyestablished, but these cells have the potential to secretea variety of molecules in the tumor microenvironment

Fig. 3. PDT-mediated increase in experimental glioma volume. Micewere injected with photofrin and 24 h later were subjected to craniot-omy to expose a defined brain area to the indicated optical dose of laser.Three or 7 days later, 3 3 105 human glioma U87 cells were implantedthrough the same opening and animals were sacrificed 3 weeks later.(A) Tumor volumes expressed as mean 6 SE for the number of animalsindicated. (*) indicates significant increase in tumor volumes in relationto control (0 J/cm2), P < 0.001. Representative images of H&E stainingof tumor-bearing brain sections for animals subjected to craniotomyalone (B) and 80 J/cm2 laser (C) 7 days before U87 cell implantation.Arrows point to the tumor borders. [Color figure can be viewed in theonline issue, which is available at www.interscience.wiley.com.]

1058 deCARVALHO ET AL.

GLIA DOI 10.1002/glia

which could promote tumor growth, such as cytokines,growth factors, and ECM components (Raivich andBanati, 2004). For example, VEGF is expressed in peri-tumoral reactive astrocytes in human glioblastoma spe-cimens (Nagashima et al., 2002) and in experimentalglioma (Antunes et al., 2000). In addition to the wellestablished effect in angiogenesis and vascular perme-ability (Machein and Plate, 2004), VEGF can induceinflammation (Croll et al., 2004) and a role for VEGF inproliferation and maintenance of reactive astrocytes inresponse to brain injury has been suggested (Krum andKhaibullina, 2003). VEGF has been shown to be upregu-lated in rodent astrocytes by hypoxia (Ijichi et al., 1995),ischemia (Gao and Chopp, 2005) and ionizing radiation(Park et al., 2001). The results presented here show thatastrocytes contributed to the PDT-induced VEGF upre-gulation in the mouse brain.

Our results are in agreement with published datareporting a conspicuous presence of microglia and astro-cytes in the tumor microenvironment. Whether an envir-onment prepopulated by activated astrocytes and micro-glia directly contributes to the enhanced tumor growthdescribed here remains to be determined. Since reactivegliosis is a common response to a variety of insults tothe brain, it will be informative to assess whether pre-implantation surgical lesion and trauma would havesimilar effect on experimental glioma growth reportedhere. The use of a syngeneic glioma model, implanted in

transgenic mice defective for reactive gliosis (Okadaet al., 2006) would be useful in elucidating the role ofreactive glia in glioma growth.

Since recurrence of malignant gliomas typically occursin the proximity of the resection cavity (Wallner et al.,1989), there is a need for adjuvant therapies targetingneoplastic cells invading into the brain parenchyma.Results from a recent clinical trial have reinforced thepotential of PDT as a local adjuvant in the treatment ofbrain cancer (Stylli et al., 2005). However, the success isstill limited by less than optimal tumor specificity andpotency of the photosensitizers currently approved for clin-ical use. Our results further support the need for the de-velopment of photosensitizers with increased selectivity fortumor cells, and of combination therapies that include thepharmacological inhibition of the undesired effects of PDT.

ACKNOWLEDGMENTS

The authors thank Ms. Supata Santra for technicalassistance.

REFERENCES

Almeida RD, Manadas BJ, Carvalho AP, Duarte CB. 2004. Intracellularsignaling mechanisms in photodynamic therapy. Biochim BiophysActa 1704:59–86.

Fig. 4. Tumor associated reactive glia. (A,B) Reactive astrocytes, identified by the increased expression of GFAP (green), accumulate at the pe-riphery of the U87 tumor labeled with anti-human vimentin (red, arrows); astrocytes infiltrating the tumor are also observed (arrowheads). (C,D)IB4 (green) stained microglia/macrophage cells at the periphery (arrows) and within the U87 tumor mass (arrowhead). Scale, 100 lm (A,C), 50 lm(B), 30 lm (D).

1059GLIOMA PROGRESSION IN PDT-TREATED MICE BRAIN

GLIA DOI 10.1002/glia

Antunes L, Angioi-Duprez KS, Bracard SR, Klein-Monhoven NA, Le FaouAE, Duprez AM, Plenat FM. 2000. Analysis of tissue chimerism in nudemouse brain and abdominal xenograft models of human glioblastomamultiforme: What does it tell us about the models and about glioblastomabiology and therapy? J Histochem Cytochem 48:847–858.

Chen Q, Chopp M, Madigan L, Dereski MO, Hetzel FW. 1996. Damagethreshold of normal rat brain in photodynamic therapy. PhotochemPhotobiol 64:163–167.

Coussens LM, Werb Z. 2002. Inflammation and cancer. Nature 420:860–867.

Croll SD, Ransohoff RM, Cai N, Zhang Q, Martin FJ, Wei T, KasselmanLJ, Kintner J, Murphy AJ, Yancopoulos GD, Wiegand SJ. 2004.VEGF-mediated inflammation precedes angiogenesis in adult brain.Exp Neurol 187:388–402.

Daumas-Duport C, Scheithauer BW, Kelly PJ. 1987. A histologic andcytologic method for the spatial definition of gliomas. Mayo Clin Proc62:435–449.

Dolmans DE, Fukumura D, Jain RK. 2003. Photodynamic therapy forcancer. Nat Rev Cancer 3:380–387.

Ferrario A, von Tiehl KF, Rucker N, Schwarz MA, Gill PS, Gomer CJ.2000. Antiangiogenic treatment enhances photodynamic therapyresponsiveness in a mouse mammary carcinoma. Cancer Res60:4066–4069.

Gao Q, Li Y, Chopp M. 2005. Bone marrow stromal cells increase astro-cyte survival via upregulation of phosphoinositide 3-kinase/threonineprotein kinase and mitogen-activated protein kinase kinase/extracellularsignal-regulated kinase pathways and stimulate astrocyte trophic factorgene expression after anaerobic insult. Neuroscience 136:123–134.

Gomer CJ, Ferrario A, Luna M, Rucker N, Wong S. 2006. Photody-namic therapy: Combined modality approaches targeting the tumormicroenvironment. Lasers Surg Med 38:516–521.

Gurney JG, Kadan-Lottick N. 2001. Brain and other central nervous systemtumors: Rates, trends, and epidemiology. Curr Opin Oncol 13:160–166.

Hill JS, Kaye AH, Sawyer WH, Morstyn G, Megison PD, Stylli SS.1990. Selective uptake of hematoporphyrin derivative into humancerebral glioma. Neurosurgery 26:248–254.

Ijichi A, Sakuma S, Tofilon PJ. 1995. Hypoxia-induced vascular endothelialgrowth factor expression in normal rat astrocyte cultures. Glia 14:87–93.

Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, FeuerEJ, Thun MJ. 2005. Cancer statistics, 2005. CA Cancer J Clin55:10–30.

Jiang F, Zhang ZG, Katakowski M, Robin AM, Faber M, Zhang F,Chopp M. 2004. Angiogenesis induced by photodynamic therapy innormal rat brains. Photochem Photobiol 79:494–498.

Korbelik M. 2006. PDT-associated host response and its role in thetherapy outcome. Lasers Surg Med 38:500–508.

Krum JM, Khaibullina A. 2003. Inhibition of endogenous VEGFimpedes revascularization and astroglial proliferation: Roles forVEGF in brain repair. Exp Neurol 181:241–257.

Le DM, Besson A, Fogg DK, Choi KS, Waisman DM, Goodyer CG,Rewcastle B, Yong VW. 2003. Exploitation of astrocytes by gliomacells to facilitate invasiveness: A mechanism involving matrix metal-loproteinase-2 and the urokinase-type plasminogen activator-plasmincascade. J Neurosci 23:4034–4043.

Lilge L, Portnoy M, Wilson BC. 2000. Apoptosis induced in vivo byphotodynamic therapy in normal brain and intracranial tumour tis-sue. Br J Cancer 83:1110–1117.

Lilge L, Wilson BC. 1998. Photodynamic therapy of intracranial tissues:A preclinical comparative study of four different photosensitizers.J Clin Laser Med Surg 16:81–91.

Machein MR, Plate KH. 2004. Role of VEGF in developmental angiogenesisand in tumor angiogenesis in the brain. Cancer Treat Res 117:191–218.

Mandil R, Ashkenazi E, Blass M, Kronfeld I, Kazimirsky G, RosenthalG, Umansky F, Lorenzo PS, Blumberg PM, Brodie C. 2001. Proteinkinase Calpha and protein kinase Cdelta play opposite roles in theproliferation and apoptosis of glioma cells. Cancer Res 61:4612–4619.

Moorthy RK, Rajshekhar V. 2004. Development of glioblastoma multi-forme following traumatic cerebral contusion: Case report and reviewof literature. Surg Neurol 61:180–184; discussion 184.

Muller PJ, Wilson BC. 1985. Photodynamic therapy: Cavitary photoillu-mination of malignant cerebral tumours using a laser coupled inflata-ble balloon. Can J Neurol Sci 12:371–373.

Muller PJ, Wilson BC. 1990. Photodynamic therapy of malignant braintumours. Can J Neurol Sci 17:193–198.

Muller PJ, Wilson BC. 2006. Photodynamic therapy of brain tumors-Awork in progress. Lasers Surg Med 38:384–389.

Nagano N, Sasaki H, Aoyagi M, Hirakawa K. 1993. Invasion of experi-mental rat brain tumor: Early morphological changes followingmicroinjection of C6 glioma cells. Acta Neuropathol (Berl) 86:117–125.

Nagashima G, Suzuki R, Asai J, Fujimoto T. 2002. Immunohistochemi-cal analysis of reactive astrocytes around glioblastoma: An immuno-histochemical study of postmortem glioblastoma cases. Clin NeurolNeurosurg 104:125–131.

Okada S, Nakamura M, Katoh H, Miyao T, Shimazaki T, Ishii K,Yamane J, Yoshimura A, Iwamoto Y, Toyama Y, Okano H. 2006. Con-ditional ablation of Stat3 or Socs3 discloses a dual role for reactiveastrocytes after spinal cord injury. Nat Med 12:829–834.

Oleinick NL, Evans HH. 1998. The photobiology of photodynamictherapy: Cellular targets and mechanisms. Radiat Res 150 (Suppl 5):S146–S156.

Oseroff AR, Blumenson LR, Wilson BD, Mang TS, Bellnier DA, ParsonsJC, Frawley N, Cooper M, Zeitouni N, Dougherty TJ. 2006. A doseranging study of photodynamic therapy with porfimer sodium (Photo-frin) for treatment of basal cell carcinoma. Lasers Surg Med 38:417–426.

Pages G, Pouyssegur J. 2005. Transcriptional regulation of the vascularendothelial growth factor gene—A concert of activating factors. Car-diovasc Res 65:564–573.

Park JS, Qiao L, Su ZZ, Hinman D, Willoughby K, McKinstry R,Yacoub A, Duigou GJ, Young CS, Grant S, Hagan MP, Ellis E, FisherPB, Dent P. 2001. Ionizing radiation modulates vascular endothelialgrowth factor (VEGF) expression through multiple mitogen activatedprotein kinase dependent pathways. Oncogene 20:3266–3280.

Raivich G, Banati R. 2004. Brain microglia and blood-derived macro-phages: Molecular profiles and functional roles in multiple sclerosisand animal models of autoimmune demyelinating disease. Brain ResBrain Res Rev 46:261–281.

Raivich G, Bohatschek M, Kloss CU, Werner A, Jones LL, KreutzbergGW. 1999. Neuroglial activation repertoire in the injured brain:Graded response, molecular mechanisms and cues to physiologicalfunction. Brain Res Brain Res Rev 30:77–105.

Rutka JT, Ivanchuk S, Mondal S, Taylor M, Sakai K, Dirks P, Jun P,Jung S, Becker LE, Ackerley C. 1999. Co-expression of nestin andvimentin intermediate filaments in invasive human astrocytomacells. Int J Dev Neurosci 17:503–515.

Salhia B, Angelov L, Roncari L, Wu X, Shannon P, Guha A. 2000.Expression of vascular endothelial growth factor by reactive astro-cytes and associated neoangiogenesis. Brain Res 883:87–97.

Salvati M, Frati A, Russo N, Caroli E, Polli FM, Minniti G, Delfini R.2003. Radiation-induced gliomas: Report of 10 cases and review ofthe literature. Surg Neurol 60:60–67; discussion 67.

Schmidt-Erfurth U, Schlotzer-Schrehard U, Cursiefen C, Michels S,Beckendorf A, Naumann GO. 2003. Influence of photodynamic ther-apy on expression of vascular endothelial growth factor (VEGF),VEGF receptor 3, and pigment epithelium-derived factor. InvestOphthalmol Vis Sci 44:4473–4480.

Skyrme RJ, French AJ, Datta SN, Allman R, Mason MD, MatthewsPN. 2005. A phase-1 study of sequential mitomycin C, 5-aminolaevu-linic acid-mediated photodynamic therapy in recurrent superficialbladder carcinoma. BJU Int 95:1206–1210.

Stylli SS, Kaye AH. 2006. Photodynamic therapy of cerebral glioma—Areview. Part II—Clinical studies. J Clin Neurosci 13:709–717.

Stylli SS, Kaye AH, MacGregor L, Howes M, Rajendra P. 2005. Photo-dynamic therapy of high grade glioma—Long term survival. J ClinNeurosci 12:389–398.

Wallner KE, Galicich JH, Krol G, Arbit E, Malkin MG. 1989. Patternsof failure following treatment for glioblastoma multiforme and ana-plastic astrocytoma. Int J Radiat Oncol Biol Phys 16:1405–1409.

Watters JJ, Schartner JM, Badie B. 2005. Microglia function in braintumors. J Neurosci Res 81:447–455.

Zhang M, Olsson Y. 1997. Hematogenous metastases of the humanbrain–characteristics of peritumoral brain changes: A review. J Neu-rooncol 35:81–89.

Zhang X, Jiang F, Kalkanis SN, Zhang ZG, Hong X, Yang H, Chopp M.Post-acute response of 9L gliosarcoma to PhotofrinTM-mediated PDTin athymic nude mice. Lasers Med Sci, 2007; Feb 15 (Epub ahead ofprint).

Zhang X, Jiang F, Zhang ZG, Kalkanis SN, Hong X, Decarvalho AC,Chen J, Yang H, Robin AM, Chopp M. 2005. Low-dose photodynamictherapy increases endothelial cell proliferation and VEGF expressionin nude mice brain. Lasers Med Sci 20:74–79.

1060 deCARVALHO ET AL.

GLIA DOI 10.1002/glia