www.elsevier.com / locate / bres
Short communication
Cyclooxygenase (COX)-1 expressing macrophages / microglial cells
and COX-2 expressing astrocytes accumulate during
oligodendroglioma progression
a ,
*
a a bMartin H. Deininger
, Richard Meyermann , Katrin Trautmann , Matthias Morgalla ,
b b c a
Frank Duffner , Ernst H. Grote , Juergen Wickboldt , Hermann J. Schluesener
a
Institute of Brain Research, University of Tuebingen, Medical School, Tuebingen, Germany
b
Department of Neurosurgery, University of Tuebingen, Medical School, Tuebingen, Germany
c
Department of Neurosurgery, Asklepios Klinik Schildautal, Seesen, Germany
Accepted 29 August 2000
Abstract
Cyclooxygenases (COX, prostaglandin endoperoxide synthases, PGG / H synthases) are potent mediators of edema, impeding blood flow and immunomodulation in the pathologically altered brain. Two COX iso-enzymes have been associated with brain disease, the constitutively expressed COX-1 and the cytokine-inducible COX-2. We have used single and double labeling immunohistochemistry to analyse COX-1 and COX-2 expression in twenty-six primary WHO grade II oligodendrogliomas, sixteen primary WHO grade III anaplastic oligodendrogliomas, twenty-seven matched recurrences and ten neuropathologically unaltered brains. COX-1 immunoreactivity was predominantly observed in macrophages / microglial cells. The number of COX-1 expressing macrophages / microglial cells was significantly lower in primary oligodendrogliomas than in primary anaplastic oligodendrogliomas (P,0.0001) and in anaplastic oligodendroglioma relapses (P50.011). Patients with low COX-1 labeling scores in the primary tumors had significantly longer time to progression and overall survival (P50.0285) than those with high COX-1 labeling scores. COX-2 immunoreactivity was predominantly observed in disseminated neurons and astrocytes. In glioblastoma multiforme relapses, accumulation of COX-2 expressing astrocytes was observed surrounding areas of focal necrosis. The number of COX-2 expressing astrocytes was significantly (P50.0471) lower in primary oligodendrogliomas than in high grade oligodendroglioma relapses. These data provide convincing evidence for the differential accumulation of cyclooxygenase isoforms during oligodendroglioma progression in vivo. 2000 Elsevier Science B.V. All rights reserved.
Theme: Disorders of the nervous system
Topic: Neuro-oncology
Keywords: Oligodendroglioma; Cylcooxygenase-1; Cyclooxygenase-2; Immunocytochemistry
1. Introduction modulators of the immune system, blood perfusion, and
tumor growth. Two cyclooxygenase isoforms, COX-1 and
Cyclooxygenases (COX, prostaglandin endoperoxide COX-2, have been described that both catalyse identical
synthases, PGG / H synthases) catalyse the synthesis of the reaction products [8,16]. COX-1 is an enzyme of a
eicosanoid prostaglandin metabolites PGG2 and PGH2 molecular mass of 66 kd. COX-2 is a 70 kd homologue
which are metabolised to PGE , PGD2 2 and PGF , the2a that shares 61% sequence identity with COX-1. While
thromboxane TXA , or the functional antagonist of TXA ,2 2 constitutive COX-1 expression is associated with
physio-PGI2 (prostacyclin) [7,15]. Eicosanoids are potential logical eicosanoid production, COX-2 is
cytokine-induc-ible following proinflammatory stimuli.
In a wide variety of diseases of the brain, induction of
*Corresponding author. Tel.:149-7071-298-2283; fax:1
49-7071-294-COX-2 expression is associated with complex
846.
E-mail address: [email protected] (M.H. Deininger). pathophysiological derangements like ischemia [14],
trauma [6], HIV-induced encephalitis [1], inflammation peroxidase complex (Dakopatts) were diluted 1:400 in
[11], Alzheimer’s disease [13] and fever [2]. More im- BSA-TBS for 30 min, developed with diaminobenzidine
portantly, COX-2 expression modulates tumor growth. (Sigma, St. Louis, MO) and counter-stained with
hemax-Recent reports have revealed distinct interactions of cyclo- onylin.
oxygenases in tumor development. In human colorectal
cancer cells, COX-2 regulates angiogenesis [19], apoptosis 2.3. Double labeling experiments
[17] and cell adhesion [18]. Inhibition of cyclooxygenase
has therefore become a promising therapeutic strategy in In double labeling experiments, we first labeled a
cell-colorectal cancer [5]. In this context, it is of note that in type-specific antigen using the ABC procedure in
combina-brain neoplasia, inhibitors of eicosanoid biosynthesis sup- tion with alkaline phosphatase conjugates. Briefly, slices
press proliferation of glioblastoma cells [20]. were deparaffinised, irradiated in a microwave oven for
In order to provide a pathological basis for the in- antigen retrieval and incubated with nonspecific porcine
volvement of cyclooxygenases in oligodendrogliomas, we serum as described above. Then the differentiating
anti-have analysed the expression of COX-1 and COX-2 in 69 bodies directed against GFAP (glial fibrillary acidic
pro-oligodendroglioma tissue samples by immunohistochemis- tein, Boehringer Mannheim, Germany), HLA-DR, -DP,
try. Twenty-six primary WHO grade II oligodendrogliomas -DQ (MHC class II), CD68 (macrophages) and NSE
and sixteen primary WHO grade III anaplastic oligo- (neuron-specific enolase) (all Dakopatts, Glostrup,
Den-dendrogliomas were included. Nineteen grade II tumors mark) were added to the slices at a dilution of 1:100 in
progressed, ten were again grade II oligodendrogliomas, BSA-TBS. Visualization was achieved by adding
and nine had progressed to higher grade lesions. Eight biotinylated rabbit anti-mouse IgG diluted 1:400 in
BSA-anaplastic oligodendrogliomas progressed, five were again TBS for 30 min and alkaline phosphatase conjugated ABC
WHO grade III tumors, and three had progressed to complex diluted 1:400 in BSA-TBS for 30 min followed
glioblastoma multiforme. Double labeling experiments by Fast Blue BB visualization. Between double labeling
confirmed the nature of COX-1 and COX-2 expressing experiments, slices were irradiated in a microwave for 20
cells. min in citrate buffer [12]. Alkaline phosphatase was
completely inhibited as previously described [3]. Then COX-1 and -2 were immunolabeled as described above.
2. Materials and methods
2.4. Analyses and statistics 2.1. Oligodendroglioma patients
Staining results were evaluated at 4003 magnification
All oligodendrogliomas were resected at the Department using an eye-piece grid. COX-1 and COX-2
immuno-¨
of Neurosurgery in Tubingen or at the Department of reactivity was evaluated in ten regions of solid tumor
Neurosurgery of the Asklepios Klinik Schildautal in growth each. Positively stained cells were counted and
Seesen and have been described in detail before. Resection compared to the total number of counterstained nuclei.
was documented by the surgeons as incomplete or macro- Evaluation of the staining score was as follows: 0, no
scopically complete (Table 1). Diagnosis was performed staining; 1, up to 2% labeled cells; 2, 3% to 10% labeled
according to the WHO classification system [10]. cells; 3, 11% to 30% labeled cells; 4, 31% to 50% labeled
cells. Mean labeling scores (MLS) were calculated and
2.2. Immunohistochemistry compared using the two-tailed Mann-Whitney rank sum
test. To identify a predictive role of the immunoreactivity
Brain samples were removed immediately after death, of COX-1 with regard to survival and time-to-progression,
fixed in buffered formalin and embedded in paraffin. Five we compared tumors ranked 0 and 1 with tumors ranked 2
mm sections were deparaffinised and rehydrated. For and 3. The data were correlated with absolute survival in
antigen retrieval, the sections were immersed in 0.01 M months and with radiological or clinical
time-to-progres-citrate buffer and irradiated in a microwave oven set at 750 sion in months. Significance of effects on survival and
W, five cycles of 5 min. Endogenous peroxidase was time-to-progression was calculated using Kaplan-Meyer
blocked with 1% H O in methanol. Then the slices were2 2 survival analysis and the logrank test.
incubated with non-specific porcine serum. Monospecific
polyclonal antibodies directed against COX-1 and COX-2 2.5. Controls
(Santa Cruz, CA) were diluted in 1% BSA (bovine serum
albumin) TBS (Tris–balanced salt solution, pH 7.5, con- Single labeling immunohistochemistry controls included
taining 0.025 M Tris, 0.15 M NaCl) at a dilution of 1:400 incubation of the tissue slices with non-immune BSA /
and added to the slices for 1 h at room temperature. Then, TBS, blocking experiments with the respective peptide and
secondary antibody (biotinylated anti-goat IgG, Dakopatts, blocking experiments with the differentiated peptide (Santa
over-Table 1
a
Patients and labeling scores of COX-1 and COX-2
Primary tumors Relapses
night incubation of antibodies with the peptides was at 48C oligodendroglioma relapses (MLS51.83, SEM50.31).
according to the manufacturer’s instructions. Ten previous- Double labeling experiments revealed coexpression of
ly described neuropathologically unaltered control brains CD68 in COX-1 positive cells (Fig. 1E). 20–50% of
were included in this study [4]. COX-2 positive cells surrounding areas of necrosis
ex-pressed GFAP (Fig. 1F).
3. Results
In control brains without neuropathological alterations, 4. Discussion
COX-1 immunoreactivity was detected in few (5-10% of
all cells) macrophages / microglial cells (Table 1). COX-2 The pathophysiological role of COX-1 expression has
immunoreactivity was observed in singular neurons located been obscured due to its constitutive expression in a wide
in the neocortex and allocortex. No endothelial or glial range of organs including the brain. COX-1 expression in
cells expressed COX-2. macrophages / microglial cells has been described before
In oligodendrogliomas, COX-1 immunoreactivity was but its expression was again constitutive and unrelated to
predominantly detected in grouped accumulations of pyogenic stimuli [11]. In this context it is of note that both
macrophages / microglial cells throughout areas of solid COX-1 and COX-2 catalyse identical reaction products
tumor growth with frequent accumulations in the immedi- [16]. The numerical increase of COX-1 expressing
macro-ate vicinity of the tumor vasculature (Fig. 1A,B). No phages / microglial cells in high grade or progressing versus
significant (P50.235) increase in the number of COX-1 low grade oligodendroglioma therefore is surprising and
expressing cells (MLS51.57, SEM50.11) compared to the provides evidence for the synergistic involvement of
COX-analysed control brains (MLS51.3, SEM50.15) was 1 in the production of soluble prostaglandins in response to
observed. Significantly higher numbers of COX-1 express- proinflammatory stimuli in the brain. Constitutive COX-2
ing cells were observed in primary anaplastic oligodendro- expression has been described in subpopulations of
excitat-gliomas (Mean52.93, SEM50.18, P,0.0001), anaplastic ory neurons. Therefore it has been speculated that COX-2
oligodendroglioma relapses (Mean52.33, SEM50.33, P5 expression in neurons may play a role in postsynaptic
0.0430) and pooled anaplastic oligodendroglioma and signaling of excitatory neurons in cortex and associated
glioblastoma relapses of all examined patients (Mean5 structures [9]. Enhanced COX-2 expression in the brain
2.43, SEM50.17, P50.0011) compared to primary WHO has been associated with inflammatory disorders [11],
grade II primary oligodendrogliomas. Kaplan–Meyer sur- trauma, ischemia [6,14] and Alzheimer’s disease [13] and
vival analyses revealed that patients with low COX-1 administration of COX-2 inhibitors has been suggested to
labeling scores, median survival was significantly (P5 be a promising therapeutic strategy in these diseases.
0.0285) longer (36 months) compared to those with high Prostaglandins promote tumorigenesis by modulating
labeling scores whose median survival was 10 months. BCL-2 expression in colon cancer cells and recent reports
Single COX-2 positive cells were disseminated through- have revealed distinct interactions of cyclooxygenases in
out the tumor parenchyma of low and high grade oligo- tumor development. In human colorectal cancer cells,
dendrogliomas with morphological characteristics of neu- COX-2 regulates angiogenesis [19], apoptosis and cell
rons (Fig. 1C) and astrocytes. They accumulated in areas adhesion [17]. Inhibitors of eicosanoid biosynthesis
sup-of infiltrative tumor growth. Only singular endothelial cells press proliferation in glioblastoma cells [20]. Further
expressed COX-2. In glioblastoma multiforme relapses, we investigations revealed that COX-1 regulates angiogenesis
detected accumulation of COX-2 positive cells in the in endothelial cells [19]. Inhibition of cyclooxygenase has
immediate vicinity of necroses (Fig. 1D). In primary therefore become a promising therapeutic strategy in
oligodendrogliomas, we calculated significantly (P50.004) colorectal cancer [5]. These findings are of particular
higher numbers of COX-2 expressing cells (MLS51.0, interest, since expression and function of COX-1 and
SEM50.11) than in the examined healthy control brains, COX-2 can be selectively inhibited by glucocorticoids,
significantly (P50.0471) lower numbers than in pooled non-steroidal anti-inflammatory drugs (NSAIDs), and other
anaplastic oligodendroglioma and glioblastoma relapses of agents. The detection of COX-2 expression in astrocytes
all examined patients (MLS51.57, SEM50.2) and sig- surrounding areas of necrosis in glioblastoma multiforme
nificantly (P50.0268) lower numbers than in anaplastic relapses therefore might indicate a novel therapeutic
cDNA cloning, expression, and gene chromosomal assignment,
strategy using COX-2 inhibitors in the treatment of
FASEB J. 5 (1991) 2304–2312.
oligodendroglioma patients.
[8] T. Hla, K. Neilson, Human cyclooxygenase-2 cDNA, Proc. Natl.
These data provide convincing evidence for the differen- Acad. Sci. USA 89 (1992) 7384–7388.
tial accumulation of cyclooxygenase isoforms during [9] W.E. Kaufmann, P.F. Worley, J. Pegg, M. Bremer, P. Isakson,
COX-2, a synaptically induced enzyme, is expressed by excitatory
oligodendroglioma progression in vivo. However, further
neurons at postsynaptic sites in rat cerebral cortex, Proc. Natl. Acad.
studies are necessary to evaluate COX inhibitors in the
Sci. USA 93 (1996) 2317–2321.
treatment of oligodendroglioma patients. [10] P. Kleihues, P.C. Burger, B.W. Scheithauer, The new WHO
classifi-cation of brain tumours, Brain Pathol. 3 (1993) 255–268. [11] S. Lacroix, S. Rivest, Effect of acute systemic inflammatory
response and cytokines on the transcription of the genes encoding Acknowledgements cyclooxygenase enzymes (COX-1 and COX-2) in the rat brain, J.
Neurochem. 70 (1998) 452–466.
¨
Supported by a grant from the fortune program ([638- [12] H.Y. Lan, W. Mu, Y.Y. Ng, D.J. Nikolic-Paterson, R.C. Atkins, A
simple, reliable, and sensitive method for nonradioactive in situ
0-0) of the University of Tuebingen. We thank Thai Dung
hybridisation: use of microwave heating to improve hybridisation
Nguyen for expert technical assistance.
efficiency and preserve tissue morphology, J. Histochem. Cytochem. 44 (1996) 281–287.
[13] W.J. Lukiw, N.G. Bazan, Cyclooxygenase 2 RNA message abun-dance, stability, and hypervariability in sporadic Alzheimer neocor-References
tex, J. Neurosci. Res. 50 (1997) 937–945.
[14] C. Iadecola, C. Forster, S. Nogawa, H.B. Clark, M.E. Ross, [1] G. Bagetta, M.T. Corasaniti, A.M. Paoletti, L. Berliocchi, R. Cyclooxygenase-2 immunoreactivity in the human brain following
Nistico, A.M. Giammarioli, W. Malorni, A. Finazzi-Agro, HIV-1 cerebral ischemia, Acta. Neuropathol. Berl. 98 (1999) 9–14. gp120-induced apoptosis in the rat neocortex involves enhanced [15] B. Samuelsson, Prostaglandin endoperoxides and thromboxanes: expression of cyclo-oxygenase type 2 (COX-2), Biochem. Biophys. role in platelets and in vascular and respiratory smooth muscle, Res. Commun. 244 (1998) 819–824. Acta. Biol. Med. Ger. 35 (1976) 1055–1063.
[2] C. Cao, K. Matsumura, K. Yamagata, Y. Watanabe, Induction by [16] W. Smith, R. Garavito, D. DeWitt, Prostaglandin endoperoxide H lipopolysaccharide of cyclooxygenase-2 mRNA in rat brain; its synthases (cyclooxygenases)-1 and -2, J. Biol. Chem. 271 (1996) possible role in febrile response, Brain Res. 697 (1995) 187–196. 33157–33160.
[3] M.H. Deininger, R. Meyermann, Multiple epitope labeling by the [17] M. Tsujii, R.N. DuBois, Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperox-exclusive use of alkaline phosphatase conjugates in
immunohisto-ide synthase-2, Cell 83 (1995) 493–501. chemistry, Histochem. Cell Biol. 110 (1998) 425–430.
[18] M. Tsujii, S. Kuwano, R.N. DuBois, Cyclooxygenase-2 expression [4] M.H. Deininger, M. Weller, J. Streffer, M. Mittelbronn, R.
Meyer-in human colon cancer cells Meyer-increases metastatic potential, Proc. mann, Patterns of Cyclooxygenase-1 and -2 expression in human
Natl. Acad. Sci. USA 94 (1997) 3336–3340. gliomas in vivo, Acta Neuropathologica 98 (1999) 240–244.
[19] M. Tsujii, S. Kawano, S. Tsuji, H. Sawaoka, M. Hori, R.N. DuBois, [5] D.J. Elder, C. Paraskeva, COX-2 inhibitors for colorectal cancer,
Cyclooxygenase regulates angiogenesis induced by colon cancer Nat. Med. 4 (1998) 392–393.
cells, Cell 93 (1998) 705–716. [6] E.F. Ellis, J. Chao, M.L. Heizer, Brain kininogen following
ex-[20] D.E. Wilson, K.M. Anderson, T.M. Seed, Ultrastructural evidence perimental brain injury: evidence for a secondary event, J.
Neuro-for differentiation in a human glioblastoma cell line treated with surg. 71 (1989) 437–442.
inhibitors of eicosanoid metabolism, Neurosurgery 27 (1990) 523– [7] C.D. Funk, L.B. Funk, M.E. Kennedy, A.S. Pong, G.A. Fitzgerald,
