Viola Caretti in Brain Pathology ... - Stichting Semmy · RESULTS Development of a human E98 DIPG...

R E S E A R C H A R T I C L E bpa_468 1..11

Monitoring of Tumor Growth and Post-Irradiation Recurrencein a Diffuse Intrinsic Pontine Glioma Mouse ModelViola Caretti1–3; Ilse Zondervan2,3; Dimphna H. Meijer3,8; Sander Idema2,3; Wim Vos4; Bob Hamans6;Marianna Bugiani4; Esther Hulleman1–3; Pieter Wesseling7; W. Peter Vandertop2; David P. Noske2,3;Gertjan Kaspers1; Carla F.M. Molthoff5; Thomas Wurdinger2,3,9

1 Department of Pediatric Oncology, 2 Department of Neurosurgery, 3 Neuro-oncology Research Group, 4 Department of Pathology,5 Nuclear Medicine & PET Research, VU University Medical Center, Amsterdam.6 Department of Radiology, 7 Department of Pathology, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands.8 Department of Cancer Biology, Harvard Medical School and Dana-Farber Cancer Institute, 9 Molecular Neurogenetics Unit, Department ofNeurology, Massachusetts General Hospital and Harvard Medical School, Boston, Ma.

AbstractDiffuse intrinsic pontine glioma (DIPG) is a fatal malignancy because of its diffuse infiltra-tive growth pattern. Translational research suffers from the lack of a representative DIPGanimal model. Hence, human E98 glioma cells were stereotactically injected into the ponsof nude mice. The E98 DIPG tumors presented a strikingly similar histhopathology toautopsy material of a DIPG patient, including diffuse and perivascular growth, brainstem-and supratentorial invasiveness and leptomeningeal growth. Magnetic resonance imaging(MRI) was effectively employed to image the E98 DIPG tumor. [18F] 3′-deoxy-3′-[18F]fluorothymidine (FLT) positron emission tomography (PET) imaging was appliedto assess the subcutaneous (s.c.) E98 tumor proliferation status but no orthotopic DIPGactivity could be visualized. Next, E98 cells were cultured in vitro and engineered to expressfirefly luciferase and mCherry (E98-Fluc-mCherry). These cultured E98-Fluc-mCherrycells developed focal pontine glioma when injected into the pons directly. However, thediffuse E98 DIPG infiltrative phenotype was restored when cells were injected into the ponsimmediately after an intermediate s.c. passage. The diffuse E98-Fluc-mCherry model wassubsequently used to test escalating doses of irradiation, applying the bioluminescent Flucsignal to monitor tumor recurrence over time. Altogether, we here describe an accurateDIPG mouse model that can be of clinical relevance for testing experimental therapeuticsin vivo.

Keywords

brainstem glioma, DIPG, imaging, luciferase,pons.

Corresponding author:

Thomas Wurdinger, PhD, Neuro-oncologyResearch Group, Cancer Center Amsterdam,VU University Medical Center, Amsterdam, theNetherlands, De Boelelaan 1117, 1081 HV,Amsterdam, the Netherlands (E-mail:[email protected])

Received 20 October 2010; accepted 22November 2010.

Funding: This work was supported by theSemmy foundation and the Egbers foundation.

Conflict of interest: None declared.

doi:10.1111/j.1750-3639.2010.00468.x

INTRODUCTIONDiffuse intrinsic pontine glioma (DIPG) is the most common andmalignant form of primary brainstem tumors. DIPG is a majorcause of pediatric brain tumor-related death (9, 14), whereas focalbrainstem glioma presents a more circumscribed growth and abetter prognosis (8). DIPG generally develops in the pons region ofthe brainstem, where tumor cells diffusely invade the parenchyma,spreading throughout the brainstem and often into the cerebellum,spinal cord and to supratentorial brain regions (32). Of note, arecent study reported a high incidence of leptomeningeal dissemi-nation and the importance to monitor this tumor component inDIPG patients (26).

Palliative radiotherapy is the standard treatment but it only pro-vides temporary relief of symptoms and is mostly followed by fastrecurrence of tumor growth (9). Overall survival is about 10months from the time of diagnosis (14) and less than 10% of

patients survive longer than 2 years (29). Different factors accountfor such a dismal prognosis in DIPG. First, resective surgery is notpossible without serious morbidity because of its diffuse nature andthe anatomical complexity of the brainstem (1). Second, in the pastdecades no chemotherapeutical scheme has proved successful inconferring a survival advantage (14). Third, radiotherapy is notcurative and effective radiosensitizers have not been identified yet.Fourth, a better understanding of the biology, and subsequently thepossibility to develop and test targeted treatment strategies, hasbeen hampered by: (i) the significant reduction in biopsy sampleacquisition since the advent of high-quality magnetic resonanceimaging (MRI); (ii) the low number of autopsies performed onDIPG patients (4); and (iii) the low number of pre-clinical studiescarried out on this type of cancer.

Until now only a few studies have been published on brainstemglioma models and none of them demonstrated specific targeting ofthe pontine region of the brainstem. Additionally, the brainstem

Brain Pathology ISSN 1015-6305

1Brain Pathology (2010)

© 2010 The Authors; Brain Pathology © 2010 International Society of Neuropathology

glioma rodent models developed by using 9L, F98, or C6 ratglioma cells (16–19, 30), human glioma cell lines, ie, U87 andU251 and primary human glioma cells maintained as neurospherecultures (15, 28), failed to show the typical diffuse and infiltrativeDIPG growth. Two recently published rodent brainstem gliomamodels, developed by genetically overexpressing platelet-derivedgrowth factor (PDGF) (2, 20), presented a more invasive pheno-type, but still did not resemble the robust diffuse infiltrative growthobserved in DIPG patients.

In the present study, we developed an orthotopic DIPG mousemodel presenting diffuse and infiltrative growth by injectinghuman adult E98 glioma cells in the pontine region of the brain-stem of nude mice. Direct comparison of the E98 DIPG model withautopsy material of a DIPG patient revealed a similar histopathol-ogy. The E98 DIPG model was also used to investigate MR- andpositron emission tomography (PET) imaging modalities. In addi-tion, we stably modified the E98 DIPG cells with expression cas-settes for firefly luciferase (Fluc) and the fluorescence reportermCherry (E98-Fluc-mCherry), allowing bioluminescence imaging(BLI), while preserving the infiltrative DIPG phenotype. Finally,we monitored the response of the E98-Fluc-mCherry DIPG tumorsto different doses of irradiation (IR) via BLI.

METHODS

E98 in vivo models

Female athymic nude mice (age 6–8 weeks; Harlan, Zeist, theNetherlands) were kept under specific pathogen-free conditionsin air-filtered cages and received food and water ad libitum. Allanimal experiments were approved and performed according to theguidelines established by the VU University Ethical and ScientificCommittees on animal experiments. E98 cells were originallyobtained from a surgical glioma specimen of an adult patient. TheE98 human glioma mouse model has been previously described(6). E98 glioma cells were cultured as adherent monolayersin Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen,Carlsbad, CA, USA) supplemented with 10% fetal bovine serum(Sigma, St Louis, MO, USA), 100 U/mL penicillin and 100 mg/mLstreptomycin (Invitrogen). After two passages, the cells were trans-duced with a lentiviral vector containing firefly luciferase (Fluc)and mCherry, as described elsewhere (21, 31). The E98-Fluc-mCherry xenograft model was established by transplantation of3 ¥ 106 cells, after <10 passages in vitro. Tumor growth was moni-tored via charge-coupled device (CCD) camera imaging and bycaliper measurements. For stereotactic surgery mice were sedatedusing gas anesthesia (1.5 L 02/minute and 2.5% isoflurane). A totalof 0.5 ¥ 106 E98 glioma cells were injected stereotactically in avolume of 5 mL (5). The skull of the mouse was exposed and asmall burr hole (0.5 mm) was made using a high-speed drill at theappropriate stereotactic coordinates. E98 or E98-Fluc-mCherryglioma cells were injected at a speed of 2 mL per minute into thepons with a 26-gauge (2 mm, point style AS) Hamilton syringe(Bonaduz, GR, Switzerland), after which the needle was removedat a speed of 0.5 mm/minute. After closing the scalp, mice wereplaced on a warming pad and returned to their cages after fullrecovery. Upon discomfort and >15% weight loss, mice were sacri-ficed and brains were removed and formalin-fixed or snap frozen.

Immunohistochemistry (IHC)

Sections of 5 mm were cut from mouse brains. IHC and hematoxy-lin and eosin (H&E) staining were performed according to standardprocedures. For IHC antibodies against human vimentin (cloneV9, VUMC, Amsterdam, the Netherlands) and mouse glucosetransporter-1 (Glut-1; DakoCytomation, Glostrup, Denmark) wereused after antigen retrieval with citrate. Omission of primary anti-body was used as negative control. EnvisionHRP and Liquid DAB(DakoCytomation) were used for visualization.

MR- and PET imaging

MRI scans were obtained on a 7T/30cm horizontal-bore magnetinterfaced to a clinical console (ClinScan, Bruker BioSpin, Ettlin-gen, Baden-Württemberg, Germany). A circular polarized receive-only coil optimized for mouse brain was used in combination witha circular polarized birdcage bodycoil for transmit. A localizer wasmade in the three orthogonal directions, followed by a T2-weightedfat-suppressed image series with slice sets in the three orthogonaldirections (TSE: TE 43 ms; TR 3880 ms; echo spacing 14.2 ms;turbo factor 7, voxel size 0.1 ¥ 0.1 ¥ 0.7 mm). Subsequently, athree-dimensional T1-weighted CSF-suppressed image series(MPRAGE: TE 3.09 ms; TR 2000 ms; TI 1000 ms; voxel size0.1 ¥ 0.1 ¥ 0.5 mm) was acquired. Mice were anesthetized using1.3% isoflurane in N2O/O2 and positioned on a 37°C water bedto maintain body temperature. For analysis of the data sets,MR scanner bundled image analysis software was used (Syngo,Siemens, Erlangen, Germany).

FLT (3′-deoxy-3′-[18F]fluorothymidine) was synthesizedaccording to an earlier described method (22, 27). PET scans wereacquired under inhalation anesthesia (mixture of 5% isoflurane:95% oxygen) and performed, as previously described, using adouble lutetium oxyorthosilicate (LSO)-layer High ResolutionResearch Tomograph (HRRT; CTI, Knoxville, TN, USA) (3, 7).

In vivo BLI and IR

Mice were anesthetized as described earlier and BLI wasperformed 10 minutes after i.p. injection of 150 mL D-Luciferin(150 mg/kg body weight). Photon counts were recorded over 5minutes using a cooled CCD camera with no illumination, usingthe Xenogen IVIS Lumina System (Xenogen Corp., Alameda, CA,USA). A light image of the animal was taken in the chamber usingdim polychromatic illumination. Following data acquisition, post-processing and visualization was performed using image displayand analysis suite, as defined by the LivingImage software(Xenogen Corp.). For visualization purposes, bioluminescenceimages were fused with the corresponding white light surfaceimages in a transparent pseudocolor overlay.

For the IR study, 13 days after stereotactic surgery mice werestratified on the basis of BLI signal intensities into four groups withcomparable mean Fluc activity. Mice were anesthetized using ket-amine and xylazine and received single cranial IR using Clinac D/E(Varian Medical Systems, Palo Alto, CA, USA), 6 MV, dose rate300 MU/min. The body of the mouse was shielded using lead.Three mice per group received a dose of 2, 6 or 10 Gy, whereasthree control mice were not treated.

Development of a DIPG mouse model Caretti et al

2 Brain Pathology (2010)


RESULTS

Development of a human E98 DIPGmouse model presenting a diffuseinfiltrative phenotype

We first determined whether the pons of nude mice could be tar-geted via stereotactic injection of human E98 glioma cells. Thefollowing coordinates were selected: -0.8 mm on the y plane,-1 mm on the x plane and -5 mm on the z plane (Figure 1A–C).First, E98 cells were injected and mice were directly sacrificedonce the stereotactic injection was ended; the pons was success-fully targeted (Figure 1D).

To determine whether E98 glioma cells injected into the ponsformed tumors and to characterize the development of such tumors,three groups of nude mice underwent stereotactic surgery and weresacrificed 8 (n = 3), 16 (n = 3) and 28 (n = 3) days after injection.Brains were harvested and tissue sections were stained with H&Eand anti-human vimentin in order to identify human E98 gliomacells and to distinguish single infiltrating E98 cells from the mouseglial cells, in particular reactive astroglial cells. E98 DIPG tumorsdeveloped in all mice and consistently showed diffuse infiltrativegrowth in the pons as determined by H&E (Figure 2A–C), andhuman vimentin immunoreactivity (Figure 2E–G), which is visiblein more detail in Figure 2I–K and M–O. In the E98 DIPG model,areas of compact growth were present in the leptomeninges in allE98 DIPG mice analyzed (Figure 2F) and occasionally within theventricles (~20% of cases) (Figure 2M), in particular inside theaqueduct of Sylvius and the fourth ventricle. The E98 DIPG mousemodel showed key histological features of human glioma suchas nuclear atypia, cellular pleomorphism and increased mitoticactivity.

At day 8 after injection the tumors were already evident(Figure 2A and E). No clear cut border separated the tumor from

the surrounding parenchyma; instead, single cells invading theadjoining tissue were clearly visible (Figure 2I, arrows). CompactE98 foci were identifiable in the fourth ventricle (Figure 2M).Interestingly, tumor cells from these foci started invading the teg-mentum of the pons (Figure 2M, arrows). At day 16 the tumors hadincreased in dimension (Figure 2B and F) and two different compo-nents were identifiable, a more solid core area and an invasive front.The core area corresponded to the original tumor focus, visible atday 8 (Figure 2I), where small islands of proliferating tumor cellscontinued to form (Figure 2J, asterisk). An invasive front com-posed of groups of cells as well as single cells invading the ponstissue were also observed (Figure 2J). Moreover, compact compo-nents were visible in the subarachnoid space (Figure 2F). The non-neoplastic pontine tissue was compressed both by the intraparen-chymal and the leptomeningeal tumor mass, which seemed toinvade through permissive routes, such as the white matter tracts inthe middle cerebellar peduncule (Figure 2N). Between day 16 and28 the tumors underwent major expansion, with more than 60% ofthe pons being diffusely invaded by the E98 cells (Figure 2C andG). Even 28 days after injection the invading fronts with infiltratingsingle cells were still present, as shown in a view of the secondcerebellar lobule (Figure 2K, arrows), and in the area between theinitial tumor core and the subsequently grown tumor spreading inthe pons parenchyma (Figure 2O).

In conclusion, we were able to develop an E98 DIPG mousemodel presenting 100% take rate. Additionally, the stereotacticcoordinates selected and the use of gas anesthesia allowed us toreproducibly target the pons of nude mice, to achieve full recoveryof the mice within half an hour from tumor cell injection and toreport no surgery related mortality.

For comparison, E98 glioma cells were also injected in thesupratentorial region (Figure 2D and H) of the brain of nude mice(n = 3). Here the infiltrative component was preferentially presentin major white matter fiber tracts (Figure 2L), as previously

*

B

C

A

D

Figure 1. Stereotactic approach used totarget the pons of nude mice. A. Stereotacticcoordinates useful to identify the site fordrilling (red star) in relation to the mouse skulllandmarks. B. Mouse head mounted on thestereotactic frame during tumor cell injection.The head of the mouse is positionedhorizontal and symmetrical to the ear bars.The arrow indicates the direction of the gasanesthesia flow. C. Anatomical trajectory ofthe needle reaching the z coordinate. D.

Hematoxylin and eosin (H&E)-stained sagittalview of the mouse brain, corresponding tothe anatomical illustration in (C). E98 gliomacells (asterisk) directly after stereotacticinjection are visible in the pontine region ofthe brainstem. Size bar = 1 mm, A and C

partly adapted from (24).

Caretti et al Development of a DIPG mouse model



reported (6). The supratentorial tumor component in the hypotha-lamic area revealed less pronounced margins (Figure 2P), whereasin the E98 DIPG model the diffuse component infiltrated through-out the pontine parenchyma (Figure 2K and O).

The orthotopic E98 DIPG mouse modelresembles the histopathology of clinical DIPG

Next the histopathological characteristics of the E98 DIPG tumorswere directly compared with autopsy material from a DIPG patient.The tumor of the DIPG patient presented the typical diffuse growthin the basis (Figure 3A and B, arrowheads), as well as in the teg-mentum pontis (Figure 3E and F, arrowheads). The tumor of theDIPG patient, originally located in the pons, invaded supratentorial(data not shown) and other infratentorial structures, by means ofdirect intraparenchymal infiltration and leptomeningeal dissemina-tion. Figures 3J and 3N (arrowheads) show human DIPG cellsinvading the medulla oblongata via intraparenchymal infiltration.

Leptomeningeal and intraventricular dissemination occurredthroughout the brain, including the fourth ventricle, the cerebellumand ventral part of the medulla oblongata (Figure 3E, I and M,respectively, indicated by asterisks). A similar direct infiltration ofthe surrounding parenchyma was also observed in E98 DIPGtumors (Figure 2K and O). The same applies to the leptomeningealdissemination (Figure 2F) with secondary infiltration of the under-lying parenchyma from the compact subarachnoid component,which was observed in both the E98 model (Figure 2N) and thepatient (Figure 3M, arrows).

Furthermore, perivascular tumor dissemination was detectedthroughout the brain of the clinical DIPG case (Figure 3G), as wellas in the preclinical E98 DIPG model (Figure 3H), providing per-missive routes to invade distal brain areas (10). Interestingly,perivascular migration appeared to be a route for the tumor cellslocated in the subarachnoidal space to invade the brain parenchymaboth in the clinical (Figure 3K) and E98 (Figure 3L) DIPG tumors.Additionally, vascular proliferation was detected in the intraparen-

Day 8 Day 16 Day 28

E98 DIPG E98 supratentorial V

IME

NT

IN

Day 21

E H G F

LJI

M

K

P O PN

*

H&

E

A C DB

Figure 2. E98 diffuse intrinsic pontine glioma (DIPG) tumor develop-ment. Gross appearance of a representative E98 DIPG tumor at 8(A,E), 16 (B,F) and 28 (C,G) days after E98 glioma cells implantation.E–G. Areas enclosed by the two dotted squares are magnified in

I,M,J,N,K,O.D,H. Gross appearance of a supratentorial E98 tumor; theareas enclosed by the two dotted squares are magnified in L,P. Sizebar = 1 mm (A–H), 125 mm (I), 200 mm (J,K,L,P), 500 mm (M,O) and62.5 mm (N).




chymal and leptomeningeal (Figure 3O and P) tumor parts, both inthe human and E98 DIPG tumors where also the preexisting lep-tomeningeal vessels may have been incorporated into the subarach-noid tumor component.

Endothelial expression of Glut-1 is considered to be a marker forthe integrity of the blood–brain barrier (BBB) (13). In the humanautopsy material and in the E98 DIPG tumors the majority of bloodvessels in the intraparenchymal diffuse tumor component stainedpositive for Glut-1 (Figure 3C and D, respectively).

MR- and PET imaging of the E98 DIPGmouse model

The applicability of MRI and PET was investigated as imagingmodalities for the E98 DIPG tumors in the pons of mice. Two

mice underwent MR-scans at 28 days after stereotactic implanta-tion of E98 glioma cells in the pons, together with a healthycontrol mouse (Figure 4). The pons of the control mouse wasclearly visible (Figure 4A, arrows). MR-images of the E98 DIPG-bearing mice revealed the presence of tumors in the pons region(Figure 4B). MRI scans also showed E98 DIPG tumor invasionof the midbrain, cerebellum, medulla oblongata and thalamus(Figure 4B). The sagittal and transversal MRI views revealedenlargement of the mouse pons (Figure 4B, upper and centralpanels), resembling the human situation (Figure 4C). The coronalMRI scan (Figure 4B, mouse 1, lower panel) clearly reproducesthe H&E- and vimentin-stained tissue sections from the samemouse (Figure 2C and G).

To test whether PET could be used to monitor E98 DIPG meta-bolic activity, we first determined the uptake of the [18F]FLT tracer

B C D

F G H

J K

O

L

P N

*

*

*

*

GPID89EGPIDlacinilCB

AS

IS

PO

NT

IS

TE

GM

EN

TU

M

PO

NT

IS

ME

DU

LL

A

OB

LO

NG

ATA

C

ER

EB

EL

LU

M

A

E

I

M

* *

Figure 3. Comparative analysis of clinical and E98 diffuse intrinsicpontine glioma (DIPG) histopatology. A,E,I,J,M. Hematoxylin and eosin(H&E) staining of clinical DIPG at different anatomical locations. Theareas enclosed by the dotted squares are magnified in the central andright panels of the clinical DIPG case (B,F,G,K,N,O). (C,D) Glut1 stainingof human DIPG (C) and of E98 DIPG in the pons (D). H,L,P. Vimentin

stained images of E98 DIPG xenografts. Asterisks indicate leptom-eningeal growth. Arrows indicate blood vessel positive for Glut-1 (C,D),perivascular growth (H,K,L) and blood vessels in compact tumor compo-nents (O,P). Size bars = 250 mm (A,E,J,L,M,P), 62.5 mm (B,C,F,G,K),31.25 mm (D), 125 mm (H,N,O) and 500 mm (I).




in E98 cells in vitro. [18F]FLT tracer uptake and retention, an indi-cator of cell proliferation (25), increased proportionally to E98cells number (Supporting Information Figure S1). Next [18F]FLTPET imaging was performed in E98 tumor bearing mice to investi-gate whether the proliferative activity of E98 tumors could bemonitored in vivo. First, the tracer was administered to micebearing s.c. E98 tumors (n = 4) in both flanks. Figure 4D (leftpanel) shows a representative PET image of a mouse bearing bilat-eral s.c. E98 tumors, presenting [18F]FLT tumor uptake (arrows).[18F]FLT PET was also performed in mice 21 days after injection ofE98 cells in the pons (n = 3) or in the striatum (n = 3). Althoughpost-mortem analysis of the brains did reveal the presence of E98tumors in these locations (data not shown), the resolution (2.3–2.7 mm) of this HRRT PET scanner was not high enough to signifi-cantly measure [18F]FLT tracer activity in E98 tumors in the mousebrain (Figure 4D, right panel).

BLI of E98-Fluc-mCherry DIPG tumors

In order to establish DIPG tumors in mice that could be monitoredlongitudinally, E98-Fluc-mCherry cells were engineered. Morethan 99% of the cells expressed the red fluorescent proteinmCherry (Figure 5A, insert), and transduction efficiency remainedconstant after multiple passages. Additionally, the BLI signal cor-related directly to E98-Fluc-mCherry cell number (Figure 5A).E98-Fluc-mCherry cells were injected stereotactically into thepons of nude mice. Mice injected with E98-Fluc-mCherry cells

showed an increasing Fluc signal over time, indicating DIPG tumorengraftment and growth (Figure 5B and C). After sacrificing themice, brain slices were analyzed by fluorescence microscopy toidentify mCherry positive E98-Fluc-mCherry cells and resultswere confirmed by IHC. All mice analyzed showed mCherryexpression throughout the pons (Figure 5D). However, IHC evalu-ation revealed a more compact phenotype (Figure 5E), resemblingthat of focal pontine glioma. At best a limited invasive growthinstead of diffuse infiltration was detected, even though the dura-tion before onset of clinical signs and weight loss were similar tothe parental E98 cells. In Figures 5F and G we show the tumor thatdeveloped with the most invasive margins.

To determine whether the focal phenotype of the E98-Fluc-mCherry tumors was caused by in vitro culture prior to intracranialinjection, E98-Fluc-mCherry cells were implanted s.c. in the flankof nude mice. The Fluc activity of the E98-Fluc-mCherry s.c.tumors correlated with the caliper measurements (Figure 6A andB). After 3 weeks, E98-Fluc-mCherry cells were isolated from thes.c. tumors and transplanted directly into the pons of nude mice.Again, these mice showed an increasing Fluc signal over time(Figure 6C and E). Interestingly, after the s.c. transplantation step,the brains analyzed by IHC revealed a major diffuse and infiltrativephenotype (Figure 6F and G), and in addition the perivascular andleptomeningeal growth (Figure 6H), which was already observedin the parental E98 model. Figure 6I depicts single cells invadingthe second cerebellar lobe, similar to the parental E98 DIPGxenografts (Figure 2K). The E98-Fluc-mCherry focal and diffuse

s.c.

DIPG patient C

D

i.c.

A DIPG mouse 1 DIPG mouse 2 Control mouse B

s.c.. i.c..

Figure 4. Magnetic resonance (MR)- and positron emission tomography(PET) imaging of the E98 diffuse intrinsic pontine glioma (DIPG) mousemodel. Upper, central and lower panels represent, respectively, sagittal,transversal and coronal views of a healthy mouse (A) and of mice bearing

DIPG (B). C. MR images of a DIPG patient. All MRI scan are T2 weightedimages. D. PET images of a mouse bearing s.c. E98 tumors (left) and E98DIPG tumor (right). I.c. stands for intracerebral.




pontine tumors presented tumor growth in all injected mice and areproducible tumor phenotype. Finally, BLI allowed the identifica-tion of supratentorial brain invasion as well as the identification ofspinal cord metastases (Figure 6D).

IR treatment of nude mice bearingE98-Fluc-mCherry DIPG

Radiotherapy is the standard treatment for DIPG. Hence to provethe feasibility of the E98-Fluc-mCherry DIPG model for therapeu-tic testing, an IR dose–response experiment was performed. Thedoses used are 2, 6 and 10 Gy, as described previously (2). The BLIsignal, detected by CCD camera imaging, decreased in all irradi-ated mice, whereas the non-treated tumors continued to grow(Figure 7A and B). The 6 and 10 Gy doses showed a similar effecton DIPG tumor recurrence, ie, resumption of DIPG tumor growthapproximately 13 days after IR. The 2 Gy dose had a less strong

effect resulting in tumor recurrence 7 days after IR. Similarly,survival analysis showed that mice irradiated with 2 Gy survivedlonger than non-treated mice, but less compared with mice thatreceived a dose of 6 or 10 Gy (Figure 7C).

DISCUSSIONThe most important hallmark of DIPG is its extreme infiltrativediffuse phenotype, which is also the major obstacle to effectivetreatment. While resective surgery may constitute a treatmentoption for focal brainstem tumors because of their compact pheno-type, such treatment is not possible for DIPG. Another hallmark isthe specific anatomical location where DIPG presents its epicenter,the pontine region of the brainstem, as revealed by MRI of DIPGpatients at diagnosis. Hence, such a diffuse infiltrative growthpattern and the specific anatomical location where DIPG developsshould be represented in a robust DIPG animal model. Brainstem

A

C

B

104

105

106

107

Mouse 1

Mouse 2

Mouse 3

Mouse 4

Days after injection 5 10 15 20 25

Flu

c ac

tivi

ty (

phot

ons/

seco

nd/c

m2

)

107

E98-Fluc-mCherry

E98

101 102 103 104 105 106

102

103

104

105

106

107

Cell number

Flu

c ac

tivi

ty (

RL

U/s

econ

d) Mouse 1

Mouse 2

Mouse 3

Day 4 7 13 18 21

Mouse 4

ED

F G

Figure 5. Cultured E98-Fluc-mCherry cells develop focal pontinegliomas. A. The graph depicts the relative light units (RLUs) persecond as determined using a luminometer for different numbers ofE98 and E98-Fluc-mCherry glioma cells. Results are presented asmean � standard deviation (SD). The insert shows a fluorescencemicroscopy image of E98-Fluc-mCherry cells overlaid to the correspond-ing phase contrast image. B. Charge-coupled device (CCD) camera

images of mice bearing E98-Fluc-mCherry DIPG. Days refer to days afterstereotactic surgery. C. Quantification of the Fluc activity in (B). D. Fluo-rescence microscopy image of an E98-Fluc-mCherry DIPG showingmCherry expression. E. Vimentin stained histological section of the E98-Fluc-mCherry DIPG tumor in Mouse 2. F. A view of DIPG (Mouse 1)invasion of the midbrain. G. Higher magnification of the image depictedin (F). Size bar = 125 mm (A), 300 mm (D), 1 mm (E,F) and 250 mm (G).




animal models previously developed show a more compact pheno-type or lack the specific DIPG anatomical location in the pons.Such models represent brainstem gliomas with a more focal growthrather than the diffuse pattern characteristic of DIPG. Here, to thebest of our knowledge, we report for the first time the developmentof a DIPG specific mouse model and not of a brainstem gliomamodel. In particular, this study was constantly centered over adirect comparison with the histopathology and clinical data ofhuman DIPG with the ultimate goal to develop a mouse modelpresenting clinical relevance.

We characterized the histopathology of the E98 DIPG xenograftsat different time points during their growth, demonstrating a robustinfiltrative growth pattern present since day eight after tumor cellsinjection until mice were sacrificed. Such a detailed descriptionprovides a useful histopathological reference for future studiesemploying this model. This especially holds true as we directlycompare the E98 DIPG mouse model with autopsy material of aDIPG patient, highlighting salient common characteristics such astumor brainstem- and supratentorial invasiveness and perivascularand leptomeningeal growth. A recent study on MRI neuroaxissurveillance revealed that 56% of DIPG patients presented leptom-eningeal dissemination and a shorter overall survival than patients

with localized disease (26). The E98 DIPG model would also allowinvestigation of the leptomeningeal component.

The E98 DIPG xenografts presented here are a very accuratephenotypical model for this type of glioma, although it has to benoted that they do not necessarily represent the genotype of thisdisease. Recently two studies reported for the first time genomicprofiling of DIPG tumor samples (23, 33). Although DIPGappeared to be heterogeneous on chromosomal level, we identifieddiscrepancies in chromosomal aberrations between E98 (6) and theDIPG samples analyzed (23, 33). E98 tumors revealed chromo-somal loss of 2q, 4q, 9p and 10q, while DIPG tumors showed lossof 13q, which is in common with pediatric high-grade astrocytoma(HGA), and loss of 11p, 17p, 14q, 18p and 22q. E98 tumors pre-sented gain of 7 and 19p, while DIPG and HGA showed gains in 1qand 9q, whereas only DIPG showed gain of 10p and 17q (23, 33).Hence, a future genotypic improvement over the E98 DIPG modelpresented here may be the use of primary glioma cells derived fromDIPG patients. Although the development of suited human DIPGcell lines has not been reported yet, such cells could be injectedfollowing the stereotactic approach described here.

Another caveat of this study is the impaired immune system ofnude mice. While genetic mouse models better recapitulate the

B A

E D

C Day 5 10 17 21 28

Day 21 Day 21

G F

H I

*

7 14 21 Day

Flu

c ac

tivi

ty (

phot

ons/

seco

nd/c

m2)

103

104

105

106

107

5 10 15 20 25 30

Mouse 1 Mouse 2 Mouse 3

108

Days after injection

Fluc Tumor Vol

5 10 15 20 25

104

105

106

101

102

103

Days after transplantation F

luc

acti

vity

Tum

or v

olum

e (m

m3 )

Figure 6. Transplanted E98-Fluc-mCherry glioma cells develop diffuseintrinsic pontine glioma (DIPG) tumor. A. Charge-coupled device (CCD)camera images of a mouse bearing s.c. E98-Fluc-mCherry tumor. B.

Corresponding quantification of the s.c. Fluc activity and tumor volume.C. CCD camera images of mice bearing E98-Fluc-mCherry DIPG. Daysrefer to days after stereotactic surgery. D. CCD camera images of dropmetastasis (left) and supratentorial invasion (right). E. Quantification of

the Fluc activity in (C). F. Vimentin IHC of a representative E98-Fluc-mCherry DIPG tumor (Mouse 1). G. Higher magnification of (F), tumorcells diffusely invade the pons parenchyma (arrowheads). H. Midbrain(Mouse 1) where diffuse DIPG invasion is visible (arrowheads). I.

Cerebellum. The asterisk indicates leptomeningeal growth, the arrowperivascular growth and the arrowheads diffuse intraparenchymal inva-sion. Size bar = 1 mm (F), 556 mm (G), 250 mm (H) and 100 mm (I).




impact that inflammatory or stromal cells have in the pathogenesisof tumors, xenograft mouse models allow the use of human celllines. Although thus far other human glioma cell lines (15, 28),once injected into the brainstem of rodents, did not show an infil-trative diffuse phenotype, as shown for the E98-Fluc-mCherryglioma cells, we can not exclude that other cancer cell lines couldgive rise to such phenotype.

While characterizing the E98 DIPG xenografts and human DIPGwe found Glut-1, a BBB marker (13), to be positive both in thehuman DIPG autopsy material and in the E98 DIPG tumor, indicat-ing a functional BBB, and suggesting that tumor cells propagatearound pre-existing vessels that retain normal expression of BBBmarkers (11). These results suggest that the BBB in the E98 andpatient DIPG tumors remains intact in areas of diffuse tumorgrowth. Interestingly, DIPG tumors often show only limited con-trast enhancement on MRI, as expected by the extreme diffuse andinfiltrative nature of the DIPG cells growing in areas with an intactBBB. Furthermore, a reason for the inefficiency of systemic che-motherapy in DIPG patients may be the lack of effective drugdelivery caused by an intact BBB in the diffuse DIPG areas. Theseresults warrant further investigation of the role of the BBB inDIPG.

The applicability of MRI and PET imaging for the E98 DIPGmouse model was also investigated. Here we showed thatMR-imaging of E98 DIPG tumors is feasible, allowing DIPGtumor localization within the different anatomical structures of themouse brainstem along with DIPG invasiveness of other brainstructures. In addition, future studies could employ MRI contrastagent to further investigate the function of the BBB in E98 DIPGxenografts, also during preclinical investigation of novel treat-ments for this disease. [18F]FLT PET imaging using the HRRT canbe used to determine proliferative activity in s.c. E98 tumors, butnot in the E98 orthotopic brain tumors. For this purpose the use ofthe more sensitive micro-PET or micro-PET-computed tomogra-phy (CT) will need to be investigated (25).

Next, E98 glioma cells were genetically engineered to expressFluc and mCherry to allow BLI monitoring of tumor engraftment,growth and response to therapies. Interestingly, by injecting cul-tured E98-Fluc-mCherry cells into the pons of nude mice, focalpontine gliomas developed, whereas injection after an intermediates.c. passage gave rise to typical DIPG tumors, presenting the samehistopathology as the parental E98 DIPG tumors. These resultssuggest that specific environmental stimuli play a role in the abilityof glioma cells to invade the surrounding tissue. Previous studies

0 5 10 15 200

20

40

60

80

1000 Gy2 Gy6 Gy10 Gy

Figure 7. E98-Fluc-mCherry diffuse intrinsic pontine glioma (DIPG) response to irradiation (IR) monitored by bioluminescence imaging (BLI). A.

Quantitation of Fluc activity in treated and control groups, results are presented as mean � standard deviation (SD) B. Charge-coupled device (CCD)camera images of a representative mouse from each group. C. Survival analysis of A. Day 0 refers to the day mice were irradiated.




on primary human glioma cells demonstrated that serial s.c. pas-saging in mouse flanks allowed retention of tumor invasivenessupon subsequent implantation intracranially (12). This versatilityof the E98-Fluc-mCherry glioma cells could be used to investigatedifferences in biology and treatment strategies between focal brain-stem tumors and DIPG tumors.

As a proof of principle for the applicability of BLI monitoring infuture studies, an IR dose–response study was performed. In mostDIPG patients, IR causes temporary DIPG regression and partialrelief of symptoms. However, recurrence of the tumor always occurswithin a period of months. The challenge for translational researchin DIPG is the discovery of effective sensitizers able to potentiate orprolong the effects of radiotherapy, preventing DIPG recurrence.The E98-Fluc-mCherry DIPG model may be suited for this purposeas it responds to IR while presenting subsequent tumor recurrence.

CONCLUSIONWe have characterized a unique orthotopic DIPG mouse modelusing human adult E98 glioma cells. Although the E98 DIPGmodel may not fully represent the genotypic hallmarks of humanDIPG, this model does present the characteristic diffuse infiltrativephenotype of human DIPG, and DIPG recurrence after IR treat-ment, one of the major obstacles in the treatment of this devastatingdisease. Additionally, we have demonstrated that this model can beuseful for studies on DIPG growth and for preclinical testing ofpotential therapeutic approaches.

ACKNOWLEDGMENTSWe are grateful to Petra van der Stoop, Tonny Lagerweij, MarcoBarazas (Neuro-oncology Research Group), Paul van der Valk andInge Greeuw (Department. of Nuclear Medicine & PET Research)for skilled technical support. We thank Dannis G. van Vuurden andMarc H.A. Jansen (Department. of Pediatric Oncology) for supportwith the MR-imaging and Phil Koken (Department. of RadiationOncology) for expert technical support during the IR study, allfrom the VUMC, Amsterdam, the Netherlands. We would also liketo thank Anneke Navis and Judith W.M. Jeuken (Department. ofPathology, RUNMC, Nijmegen, the Netherlands) for precious helpwith Glut-1 staining and unraveling the genetic background of theE98 model. This work was supported by the Semmy and Egbersfoundations.

REFERENCES1. Angeles Fernández-Gil M, Palacios-Bote R, Leo-Barahona M,

Mora-Encinas JP (2010) Anatomy of the Brainstem: A Gaze Into theStem of Life. Semin Ultrasound CT MR 31:196–219.

2. Becher OJ, Hambardzumyan D, Walker TR, Helmy K, Nazarian J,Albrecht S et al (2010) Preclinical evaluation of radiation andperifosine in a genetically and histologically accurate model ofbrainstem glioma. Cancer Res 70:2548–2557.

3. Boellaard R, de Jong HWAM, Molthoff CFM, Buijs F, Lenox M, NuttR, Lammertsma AA (2003) Use of an in-field-of-view shieldto improve count rate performance of the single crystal layerhigh-resolution research tomograph PET scanner for small animalbrain scans. Phys Med Biol 48:335–342.

4. Broniscer A, Baker JN, Baker SJ, Chi SN, Geyer JR, Morris EB,Gajjar A (2010) Prospective collection of tissue samples at autopsyin children with diffuse intrinsic pontine glioma. Cancer116:4632–4637.

5. Cetin A, Komai S, Eliava M, Seeburg PH, Osten P (2006) Stereotaxicgene delivery in the rodent brain. Nat Protoc 1:3166–3173.

6. Claes A, Schuuring J, Boots-Sprenger S, Hendriks-Cornelissen S,Dekkers M, van der Kogel AJ et al (2008) Phenotypic and genotypiccharacterization of orthotopic human glioma models and its relevancefor the study of anti-glioma therapy. Brain Pathol 18:423–433.

7. Direcks WGE, Berndsen SC, Proost N, Peters GJ, Balzarini J,Spreeuwenberg MD et al (2008) [18F]FDG and [18F]FLT uptake inhuman breast cancer cells in relation to the effects of chemotherapy:an in vitro study. Br J Cancer 99:481–487.

8. Donaldson SS, Laningham F, and Fisher PG (2006) Advances towardan understanding of brainstem gliomas. J Clin Oncol 24:1266–1272.

9. Frazier JL, Lee J, Thomale UW, Noggle JC, Cohen KJ, Jallo GI(2009) Treatment of diffuse intrinsic brainstem gliomas: failedapproaches and future strategies. J Neurosurg Pediatr 3:259–269.

10. Friedl P, Gilmour D (2009) Collective cell migration inmorphogenesis, regeneration and cancer. Nat Rev Mol Cell Biol10:445–457.

11. Gambarota G, Leenders W, Maass C, Wesseling P, van der Kogel B,van Tellingen O, Heerschap A (2008) Characterisation of tumourvasculature in mouse brain by USPIO contrast-enhanced MRI. Br JCancer 98:1784–1789.

12. Giannini C, Sarkaria JN, Saito A, Uhm JH, Galanis E, Carlson BLet al (2005) Patient tumor EGFR and PDGFRA gene amplificationsretained in an invasive intracranial xenograft model of glioblastomamultiforme. Neuro Oncol 7:164–176.

13. Guo X, Geng M, and Du G (2005) Glucose transporter 1, distributionin the brain and in neural disorders: its relationship with transport ofneuroactive drugs through the blood-brain barrier. Biochem Genet43:175–187.

14. Hargrave D, Bartels U, and Bouffet E (2006) Diffuse brainstemglioma in children: critical review of clinical trials. Lancet Oncol7:241–248.

15. Hashizume R, Ozawa T, Dinca EB, Banerjee A, Prados MD, JamesCD, Gupta N (2009) A human brainstem glioma xenograft modelenabled for bioluminescence imaging. J Neurooncol 96:151–159.

16. Jallo GI, Penno M, Sukay L, Liu JY, Tyler B, Lee J et al (2005)Experimental models of brainstem tumors: development of aneonatal rat model. Childs Nerv Syst 21:399–403.

17. Jallo GI, Volkov A, Wong C, Carson BS Sr, Penno MB (2006) A novelbrainstem tumor model: functional and histopathologicalcharacterization. Childs Nerv Syst 22:1519–1525.

18. Lee J, Jallo GI, Guarnieri M, Carson BSS, Penno MB (2005) A novelbrainstem tumor model: guide screw technology with functional,radiological, and histopathological characterization. NeurosurgFocus 18:1–6.

19. Liu Q, Liu R, Kashyap M, Agarwal R, Shi X, Wang C, Yang S (2008)Brainstem glioma progression in juvenile and adult rats. J Neurosurg109:849–855.

20. Masui K, Suzuki SO, Torisu R, Goldman JE, Canoll P, Iwaki T(2010) Glial progenitors in the brainstem give rise to malignantgliomas by platelet-derived growth factor stimulation. Glia58:1050–1065.

21. Mir SE, De Witt Hamer PC, Krawczyk PM, Balaj L, Claes A, NiersJM et al (2010) In silico analysis of kinase expression identifiesWEE1 as a gatekeeper against mitotic catastrophe in glioblastoma.Cancer Cell 18:244–257.

22. Molthoff CFM, Klabbers BM, Berkhof J, Felten JT, van Gelder M,Windhorst AD et al (2007) Monitoring response to radiotherapy inhuman squamous cell cancer bearing nude mice: comparison of




2′-deoxy-2′-[18F]fluoro-D-glucose (FDG) and 3′-[18F]fluoro-3′-deoxythymidine (FLT). Mol Imaging Biol 9:340–347.

23. Paugh BS, Qu C, Jones C, Liu Z, Adamowicz-Brice M, Zhang J et al(2010) Integrated molecular genetic profiling of pediatric high-gradegliomas reveals key differences with the adult disease. J Clin Oncol28:3061–3068.

24. Paxinos G, Franklin KBJ (2001) The mouse brain in stereotaxiccoordinates. Academic Press, San Diego, CA.

25. Salskov A, Tammisetti VS, Grierson J, Vesselle H (2007) FLT:measuring tumor cell proliferation in vivo with positron emissiontomography and 3′-deoxy-3′-[18F]fluorothymidine. Semin Nucl Med37:429–439.

26. Sethi R, Allen J, Donahue B, Karajannis M, Gardner S, Wisoff J et al(2010) Prospective neuraxis MRI surveillance reveals a high risk ofleptomeningeal dissemination in diffuse intrinsic pontine glioma. JNeurooncol 99:1–7.

27. Shields AF, Grierson JR, Dohmen BM, Machulla HJ, Stayanoff JC,Lawhorn-Crews JM et al (1998) Imaging proliferation in vivo with[F-18]FLT and positron emission tomography. Nat Med 4:1334–1336.

28. Siu I, Tyler BM, Chen JX, Eberhart CG, Thomale U, Olivi A et al(2010) Establishment of a human glioblastoma stemlike brainstemrodent tumor model. J Neurosurg Pediatr 6:92–97.

29. Wolff JEA, Classen CF, Wagner S, Kortmann R, Palla SL, Pietsch Tet al (2008) Subpopulations of malignant gliomas in pediatricpatients: analysis of the HIT-GBM database. J Neurooncol87:155–164.

30. Wu Q, Tyler B, Sukay L, Rhines L, DiMeco F, Clatterbuck RE et al(2002) Experimental rodent models of brainstem tumors. Vet Pathol39:293–299.

31. Wurdinger T, Badr C, Pike L, de Kleine R, Weissleder R, BreakefieldXO et al (2008) A secreted luciferase for ex vivo monitoring of in vivoprocesses. Nat Methods 5:171–173.

32. Yoshimura J, Onda K, Tanaka R, Takahashi H (2003)Clinicopathological study of diffuse type brainstem gliomas: analysisof 40 autopsy cases. Neurol Med Chir 43:375–382.

33. Zarghooni M, Bartels U, Lee E, Buczkowicz P, Morrison A, Huang Aet al (2010) Whole-genome profiling of pediatric diffuse intrinsicpontine gliomas highlights platelet-derived growth factor receptoralpha and poly (ADP-ribose) polymerase as potential therapeutictargets. J Clin Oncol 28:1337–1344.

SUPPORTING INFORMATIONAdditional Supporting Information may be found in the onlineversion of this article:

Figure S1. FLT uptake analysis in vitro. E98 cells were plated intriplicates in 24-well plates 2 days prior to radiotracer addition.Medium was replaced with DMEM without L-glutamine, and after4 h of incubation, 1 MBq well-1 of FLT was added to each well andincubated for 60 minutes. Cells were harvested and cell-boundradioactivity was measured in a Compugamma gamma counter.

Please note: Wiley-Blackwell are not responsible for the content orfunctionality of any supporting materials supplied by the authors.Any queries (other than missing material) should be directed to thecorresponding author for the article.




Viola Caretti in Brain Pathology ... - Stichting Semmy · RESULTS Development of a human E98 DIPG...

Documents

Transcript of Viola Caretti in Brain Pathology ... - Stichting Semmy · RESULTS Development of a human E98 DIPG...