The expression of nitric oxide synthases in human brain tumours and peritumoral areas

Journal of Neurological Sciences 155 (1998) 196–203

The expression of nitric oxide synthases in human brain tumours and peritumoral areas a b b, a c A. Bakshi , T.C. Nag , S. Wadhwa *, A.K. Mahapatra , C. Sarkar a

Department of Neurosurgery, C.N. Centre, All India Institute of Medical Sciences, New Delhi, India b Department of Anatomy, All India Institute of Medical Sciences, New Delhi, India c Department of Pathology, All India Institute of Medical Sciences, New Delhi, India Received 5 May 1997; received in revised form 27 August 1997; accepted 3 September 1997

Abstract Nitric oxide, a potent signalling molecule produced from L-arginine by nitric oxide synthase (NOS), has been implicated in diverse pathophysiological processes. Many characteristics of malignant tumours such as increased vascular permeability, vasodilation, neovascularisation and free radical injury to the tumour and adjacent normal tissues are believed to be mediated by nitric oxide. While NOS expression has been demonstrated in brain tumours, no equivalent studies have yet been reported on the adjacent peritumoral brain region. The present study examined the pattern of NOS expression in the human tumour and peritumoral brain areas. Biopsies were obtained from eight patients (six gliomas, one each of meningioma and metastatic adenocarcinoma) from three areas: tumour, peritumoral, and apparently ‘normal’ adjacent brain tissue. Immunohistochemical staining was performed for three isoforms of NOS: brain NOS (BNOS), endothelial NOS (ENOS) and macrophage-specific NOS (MacNOS). Except for glioblastoma multiforme and metastatic adenocarcinoma, the tumour cells expressed all three NOS isoforms. In four tumours, there was a demonstrable gradient of ENOS expression falling away from the tumour. In three gliomas, many glial cells were intensely labelled with BNOS. This labelling decreased in the peritumoral tissues. In four tumours, cells (presumably lymphocytes, and CD 45 positive macrophages) were labelled intensely with MacNOS in and around the blood vessels. These results suggest that nitric oxide is produced in the tumour cells and endothelium of tumour vasculature, while occasionally glial cells may also be induced to produce it. The possible role of nitric oxide in the production of peritumoral oedema is discussed.  1998 Elsevier Science B.V. Keywords: Brain tumour; Endothelium; Nitric oxide synthases; Oedema; Peritumoral brain

1. Introduction Peritumoral brain oedema (PBO) plays an important role in the pathophysiology and clinical profiles of brain tumours. The phenomenon varies with the tumour type and is more a problem with malignant gliomas, metastatic carcinomas and meningiomas than with well-differentiated astrocytomas or oligodendrogliomas (Ito et al., 1986). PBO is thought to be associated with the disruption of the blood–brain barrier and increased vascular permeability (Brightman and Broadwell, 1976; Reichhman et al., 1986; Kimelberg, 1995). PBO is primarily vasogenic and due to *Corresponding author. Tel.: 191 11 6594875; fax: 191 11 6862663. 0022-510X / 98 / $19.00  1998 Elsevier Science B.V. All rights reserved. PII S0022-510X( 97 )00315-8

intrinsic capillary abnormalities it is likely that the permeability of tumour capillaries is further increased by immunological activity, tumour-related permeability factors and inflammatory mechanisms (Baethman et al., 1980; Del Maestro et al., 1984). The possible role of biochemical factors in the pathogenesis of vasogenic brain oedema has been the subject of considerable research. Tissue destruction, vascular thrombosis, ischemia, and immunological and inflammatory reactions are characteristic histopathological features of many brain tumours. Therefore, a wide range of compounds generated by these processes could secondarily potentiate oedema formation (Baethman et al., 1980; Del Maestro et al., 1984). Oxygen free radicals, such as

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

superoxide anions and hydroxyl radicals, and vasogenic permeability factors are a few agents that are thought to be important mediators of PBO (Chan et al., 1984; Criscuolo, 1993). Many recent experimental studies have suggested a role for nitric oxide (NO) in causing PBO, increased tumour blood flow and vascular permeability (Andrade et al., 1992; Maeda et al., 1994). Semiquantitative data have established that malignant brain neoplasms express unexpectedly high levels of nitric oxide synthase (NOS), the enzyme that converts L-arginine to NO (Moncada et al., 1991; Kiechle and Malinski, 1993), and suggest that NO production may be associated with pathophysiological processes important to these tumours (Cobbs et al., 1995). Furthermore, experimental studies have established a direct role for NO and superoxide as important mediators of injury in vasogenic brain oedema (Oury et al., 1993; Koedel et al., 1995). Because of the strong evidence for NO as an important mediator of PBO, we decided to study the expression of NOS in the peritumoral brain tissue by immunocytochemistry. Thus far, there has only been one published report which has evaluated the expression of NOS in human brain tumours (Cobbs et al., 1995). The present study was carried out on both the human tumoral and adjacent brain tissues. The differential expressions of three known isoforms of NOS (Moncada et al., 1991), viz. MacNOS (an inducible form present in macrophages), BNOS (a constitutive form of neuronal source) and ENOS (a constitutive form present in endothelial cells), were examined in all tissues.

2. Material and methods

2.1. Sample collection Craniotomy and surgical excision of the tumour was performed in all patients. Tissue samples were collected during surgery from three sites: (a) tumour brain (T 0 )– from the main bulk of the tumour tissue; (b) peritumoral brain tissue (T 1 )–from the brain tissue adjacent to the tumour and involved in oedema; and (c) apparently ‘normal brain’ (T 2 )–from brain tissue lying immediately adjacent to the peritumoral region which was excised during removal of the tumour but was not involved in either the tumour or oedema, as diagnosed by the surgeons at surgery. CT scans of all patients except cases 3 and 8 showed definite evidence of peritumoral oedema. Histopathology was performed to confirm the presence of oedema. The age of the patients ranged from 32 to 50 years. All tumours were in the supratentorial compartment.

2.2. Fixation and processing Tissue samples were collected immediately in 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (pH

197

7.4) and transported within 15 min to the laboratories where they were fixed for 2 h at 48C. After several washes in buffer, the samples were cryoprotected in 30% sucrose in buffer overnight. Sections of 25–35 mm thickness were cut with a Reichert–Jung cryocut, collected in vials containing phosphate buffer and stored at 48C.

2.3. Immunocytochemistry The sections were initially treated in 70% methanol containing 0.1% hydrogen peroxide (H 2 O 2 ) for 30 min to block endogenous peroxidase activity. They were washed thoroughly and treated in 20% horse normal serum (for those incubated in mouse monoclonal antibodies) and 10% goat normal serum (for sections incubated in rabbit polyclonal antibody) for 3 h at 48C. Following this treatment, the sections were incubated in primary antisera against MacNOS (1:50, monoclonal), ENOS (1:100, monoclonal), and BNOS (1:250, polyclonal), all from Transduction Laboratories, Kentucky, USA, for 2–3 days at 48C. The antisera were diluted in 0.01 M phosphate buffer saline containing 0.5% Triton X-100 and 5% normal sera. Sections were washed and placed in secondary antisera (1:200; horse anti mouse IgG for MacNOS and ENOS, and goat anti rabbit IgG for BNOS; Vector Laboratories, Burlingame, California) for 5–6 h at 48C. After washing, the sections were incubated in avidin–biotin peroxidase complex (Vector) for 2 h at room temperature. For visualization, the sections were treated with 0.05% diaminobenzidine tetrahydrochloride (DAB, Sigma) and 0.05% H 2 O 2 in 0.1 M acetate-imidazole buffer (pH 7.4) containing 0.3% nickel sulphate (as intensifier) for 20 s. Finally, the sections were rinsed in water, mounted on gelatin-coated slides, dehydrated in ethanol and coverslipped with DPX mountant. For control sections, primary antisera incubation was omitted and the remaining processing was as already outlined. Sections from adult rat retina and / or colon, both fixed in 4% paraformaldehyde, were additionally processed as positive controls. Adjacent sections from each tumour tissue sample were cut at 15–20 mm thickness and stained with haematoxylin and eosine for histological examination. The slides were examined by four observers having a background in immunocytochemistry and the labelling was graded as 0 when there was no detectable labelling, and 1, 2 and 3 for weak, moderate and intense labelling, respectively. In glioblastoma multiforme (case 2), a densely stained perivascular infiltrate was observed in the tumour and peritumoral tissues. To characterise the cellular identity of the infiltrate, the same tissue sections processed for MacNOS labelling were also incubated with FITC linked monoclonal CD45 antiserum (which recognizes macrophages; dilution 1:1000; Becton Dickinson, San Jose, USA) for 2 days at 48C. After washing, the sections were mounted in 80% glycerine and examined under a Leica

198

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

fluorescence microscope using a combination of filters specific for FITC fluorescence. In the case of astrocytoma (case 3), the adjacent tissue sections were incubated with antiserum to GFAP (1:200, monoclonal, Sigma, USA), a marker for astrocytes, for 2 days at 48C. Following this step, the sections were placed in horse anti mouse IgG, washed, treated with avidin– biotin peroxidase complex and reacted with DAB.

had gliomas (cases 2, 3, 4, 5, 6, 8) of varying grades, and one each showed meningioma (case 1) and metastatic adenocarcinoma (case 7). The histopathological profiles from all three sites are depicted in Table 1. It is evident from the table that the T 2 region in all samples did not show completely normal brain (as in cases 1, 3 and 5). All patients in this series, except cases 3 and 8, exhibited definite radiological evidence of significant peritumoral oedema, which was corroborated in the histological preparations. Case 8 had a large tumour cyst.

3. Results

3.2. Immunoreactivity of the tumours 3.1. Histopathology Eight patients were examined in this study of which six

Table 2 summarises the results of the immunolabelling of the tumour region with all three NOS isoforms. It is

Table 1 Histological description of the biopsy materials Case No.

Tumour tissue (T 0 )

Peritumoral tissue (T 1 )

Normal brain (T 2 )

1. Meningioma

Sheaths of round to spindle shaped cells. Many large blood vessels (Fig. 1A)

Normal brain

2. Glioblastoma multiforme, GBM

Large areas of necrosis. Few tumour cells. Marked pleomorphism (Fig. 2A) Marked increase in cellularity. No pleomorphism / mitoses (Fig. 3A)

No tumour. Large reticular spaces. Normal brain with oedema No tumour. Normal brain with reactive infiltrate

3. Astrocytoma

4. Anaplastic oligoastrocytoma with calcification

5. Gemistocytic astrocytoma

6. GBM

7. Metastatic adenocarcinoma

8. Oligoastrocytoma with cyst

Tumour cells with dense nuclei and pleomorphism. Significant endothelial proliferation (Fig. 4A) Many large gemistocytes. Smaller nongemistocytic tumour cells (Fig. 5A) Pleomorphic tumour cells. Frequent mitoses. Marked endothelial proliferation (Fig. 6A) Moderate sized cells with large nuclei, forming glandular structures. Excessive mitoses / pleomorphism Both oligodendroglial and astrocytic components. Calcification

Increased cellularity though less than T 0 – oedematous Fewer tumour cells present, infiltrating into the surrounding brain

Normal brain with few infiltrative cells Some areas involved in tumour cells. Rest normal Normal brain

Fewer gemistocytes infiltrating into the surrounding brain

No gemistocytes but increased cellularity

No tumour tissue. Normal brain

Normal brain

No tumour. Normal brain with oedema

Normal brain

No tumour. Normal brain with oedema

Normal brain

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

199

Table 2 NOS immunoreactivity of the tumour region Case No.

ENOS

BNOS 1

1

MacNOS 1

Tumour cells 2 . Endothelium 3 Necrotic areas only 0 1

3. Astrocytoma

Tumour cells 2 1 . Endothelium 3 1

4. Anaplastic oligoastrocytoma

Tumour cells (cytoplasm) 2 1

Tumour cells 2 1 . Spidery glial cells 2 1 . No neuronal staining Tumour cells 2 1 . Endothelium 2 1

5. Gemistocytic astrocytoma

Tumour cells 1 1 . Endothelium 3 1

Tumour cells 0 1 . Fuzzy glial cells 3 1

6. GBM 7. Metastatic adenocarcinoma

Tumour cells 0 1 . Endothelium 3 1 Tumour cells 0 1 . Vascular endothelium 1 1 . Perivascular cells 3 1 Tumour cells 0 1 . Endothelium 0 1 . Perivascular infiltrate 3 1

Tumour cells 1 1 . Fuzzy glial cells 3 1 Tumour cells 0 1

Tumour cells 1 1 . Perivascular macrophages 1 1 Tumour cells 1 1 . Endothelium 1 1 No staining 0 1 Same as ENOS

Tumour cells 2 1

Same as ENOS

8. Oligoastrocytoma with cyst

Tumour cells 2 . Necrotic areas only 0 1

Tumour cells 1 1 Necrotic area 0 1 . Perivascular infiltrate 3 1 Tumour cells 1 1

1. Meningioma 2. GBM

GBM, glioblastoma multiforme.

evident that most tumour cells significantly expressed ENOS and / or BNOS, whilst the expression of MacNOS was weak to absent. When present, the NOS activity was localised in both cytoplasm and nuclei of the cells, although it was more intense in the latter. The immunoreactive staining was specific since the NOS-positive cells were observed to lie adjacent to cells which showed no immunoreactivity. We found a dense labelling of the endothelium of tumour vasculature with ENOS (Fig. 1B, Fig. 6B) in four tumour types, although weak labelling was also seen with MacNOS. We also observed several smallto-moderate sized cells immunopositive for MacNOS (Fig. 2B) and / or ENOS around and inside the blood vessels in four of the eight tumours. In some areas, these had the appearance of mononuclear cells, as lymphocytes, while in other areas these were macrophages, confirmed by CD 45 immunolabelling. Another interesting finding was the dense and very specific labelling of the glial cell bodies and their processes by BNOS (Fig. 3B, Fig. 4B) in three tumours.

3.3. Immunoreactivity of peritumoral ( T1 ) and ‘ normal’ ( T2 ) brain samples Only the pertinent findings of the immunolabelling pattern of samples from areas T 1 and T 2 from each case are discussed here. Case 1. Although meningioma (Fig. 1A) is a benign extraparenchymal tumour, it may frequently be associated with extensive peritumoral oedema as seen in this case. A remarkable gradient of ENOS expression falling away from the tumour was evident. The endothelium of the vessels from areas T 0 , T 1 , and T 2 showed 3 1 , 2 1 and 1 1 reactivity, respectively (Fig. 1B–D). This indicates that there is some factor secreted by the tumour which causes its endothelium to express NOS, and this factor also affects the surrounding tissue, although to a lesser extent. Simultaneously, the glial cells of the T 1 area showed intense

staining with BNOS (Fig. 7A) which was reduced to moderate staining in the T 2 area (Fig. 7B). Case 2. This was a case of glioblastoma multiforme (GBM, Fig. 2A), the most malignant cerebral neoplasm, which was characterised by very rapid growth leading to marked endothelial proliferation and development of large areas of necrosis. Samples from the T 1 area demonstrated a densely stained perivascular infiltrate immunopositive for all three NOS isoforms. Although similar cells were seen in the T 2 areas, they were less frequent and stained less densely. Moreover, the extent of NOS expression decreased with increasing distance from the tumour. Case 3. This intrinsic brain tumour (astrocytoma, Fig. 3A) infiltrated into the brain parenchyma through fingerlike processes making the identification of the tumour– normal brain interface difficult. The T 2 area also showed infiltrative tumour cells. As in case 1, an endothelial gradient of immunoreactivity for ENOS was seen. The tumour cells, glial cells and some persisting parenchymal neurones showed moderate staining with BNOS (Fig. 3B) in contrast to very faint to no staining in the T 1 and T 2 areas. The GFAP immunolabelling on adjacent sections from the T 0 area showed many positively labelled reactive astrocytes (Fig. 7E). Case 4. This was a case of anaplastic oligoastrocytoma, a malignant mixed tumour that demonstrated a uniform 2 1 staining of the endothelium with ENOS in all three areas. No gradient was demonstrable. Tumour cells and multiple glial processes were moderately labeled by BNOS (Fig. 4B). MacNOS showed weak staining of the tumour cells. Case 5. Gemistocytic astrocytomas (Fig. 5A) are a subcategory of astrocytic neoplasms. The T 2 area of this tumour was also involved in some tumour cells as in case 3. BNOS demonstrated a remarkable gradient in the staining of the glial cells, which had a spidery appearance. There was 3 1 staining in the tumour area (Fig. 5B), 2 1 staining in T 1 and no staining in T 2 . ENOS showed uniformly 3 1 staining of the capillary endothelium in all

200

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

three areas. MacNOS did not show any significant staining in the peritumoral areas. Case 6. This was another highly malignant tumour, glioblastoma multiforme (GBM, Fig. 6A), in which the endothelium showed a staining gradient with ENOS in the three regions examined (Fig. 6B,C). Glial cells staining with BNOS also demonstrated a similar pattern. MacNOS did not label cells in any of these areas. In contrast to case 2, which was also a GBM, this case did not show any perivascular infiltrate and necrotic areas. Case 7. In this case of metastatic adenocarcinoma, the tumour cells did not express NOS. However, the endothelium of the tumour vasculature showed weak expression of ENOS, whilst the blood vessels in the peritumoral area showed moderate staining. Thus the factor stimulating the endothelium seems to have affected the brain capillaries more than the tumour vessels. Perivascular cells in the tumour area showed significant ENOS and MacNOS staining. Case 8. Oligoastrocytomas are low grade malignant tumours and this case was associated with a large cyst. The tumour cells showed 2 1 staining with BNOS while the surrounding T 1 and T 2 areas showed 3 1 staining, indicating that the neighbouring brain regions were being stimulated to produce NO. Negative control sections in all cases from the three regions did not show any labelling; one such section from the tumour region of case 3 is shown in Fig. 7C. Positive control sections from rat retina showed a BNOS-positive inner plexiform layer and amacrine cells (7D), as expected.

4. Discussion In this study, we have examined the pattern of expression of three isoforms of NOS in the normal, peritumoral and tumoral regions of the human brain, in an attempt to

201

determine if NO is an important mediator of peritumoral brain oedema (PBO). In order to identify agents as mediators of brain oedema, certain requirements have to be met (Oettinger et al., 1976). These are: (a) the factors should induce brain oedema at concentrations that occur physiologically; (b) the factors should be released or activated under conditions that lead to brain oedema; (c) the extent of oedema produced in a given situation should correlate with the amount of agent released; and (d) measures that inhibit activation or release of the factor should ameliorate brain oedema. No agents have so far been discovered which completely fulfil these requirements. Many experiments that fulfil criteria (a) and (d) have identified NO as a possible agent involved in the formation of brain oedema (Oury et al., 1993; Buttery et al., 1993; Maeda et al., 1994; Koedel et al., 1995; Orucevic and Lala, 1996). Our observations demonstrate a gradient of NOS expression from the tumoral to peritumoral oedematous brain thereby fulfilling criterion (b). Although the observations are suggestive of a correlation between NOS expression and peritumoral oedema it is difficult to make a strong argument for a cause and effect relationship. Much work has been done to elucidate the role of NO in vascular permeability, oedema formation and neovascularisation. Koedel et al. (1995) conclude from NOS inhibitorbased experiments that both NO and superoxide radicals are involved as mediators of brain oedema. Orucevic and Lala (1996) have demonstrated that L-NAME (a NOS inhibitor) reduced the severity of interleukin-2-induced capillary leakage, thus linking cytokines with NO production. NO may also act as a part of the signalling cascade for neovascularisation in tumours as demonstrated by the studies of Jenkins et al. (1995). Buttery et al. (1993) have used immunohistochemistry to demonstrate that NOS could be induced in the endothelium of the newly formed capillaries of tumours. Thus experimental evidence tends to indicate that nitric oxide may be a mediator of oedema

Fig. 1. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (3160). (A) Meningioma (H&E stained), showing many blood vessels. ENOS labelling in the tumour (B), peritumour (C) and normal tissue (D). A gradient (i.e. 3 1 , 2 1 and 1 1 , respectively, in three areas) is apparent in the labelling of endothelium and tumour cells. Fig. 2. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (3160). (A) Necrotic region of glioblastoma multiforme showing perivascular infiltrate (H&E stained). (B) MacNOS-positive 3 1 labelling of infiltrative cells in glioblastoma multiform. Fig. 3. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (3160). (A) Astrocytoma (H&E stained). (B) BNOS-positive tumour cells, glial cells and neurones (arrows) with processes can be seen. Fig. 4. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (A, 3160; B, 3400). (A) Anaplastic oligoastrocytoma with calcification (H&E stained). (B) BNOS-positive multiple glial processes (arrows). The tumour cells express 3 1 labelling. Fig. 5. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (3160). (A) Gemistocytic astrocytoma (H&E stained). Note the prominent gemistocytes (arrows). BNOS-positive glial cells (arrows), with 3 1 labelling in the tumour (B) and 2 1 labelling in the peritumoral region (C). Fig. 6. Photomicrographs showing histological profile of the tumours, stained with haematoxylin and eosine (H&E), and NOS immunoreactivity in the three brain regions studied (3160). (A) Glioblastoma multiforme (H&E stained). ENOS labelling of the endothelium (arrows) in the tumour (B, 3 1 ) and peritumour (C, 2 1 ) tissues.

202

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

Fig. 7. BNOS immunoreactivity in peritumoral (A) and normal (B) region from a meningioma case showing 3 1 and 2 1 labelling of the glial cells (arrows), respectively. (C) Negative control section from tumoral region of astrocytoma (case 3) showing absence of immunoreactivity. (D) Positive control section from rat retina, showing BNOS-positive inner plexiform layer (IPL) and amacrine cell (arrow). (E) Section from the tumoral region of astrocytoma (case 3). The GFAP-positive reactive astrocytes are indicated by arrows (A,B, 3160; C–E, 3400).

formation and that oedema formation may be decreased by NOS inhibitors. All the above conclusions were drawn from experimental studies on animals. Cobbs et al. (1995) performed the first study of NOS expression in human brain tumours. The present study demonstrates that not only do human brain tumours express NOS as shown by Cobbs et al. (1995), but also that the brain tissue in the immediate vicinity of the tumour expresses NOS. In four tumours, the endothelium of the blood vessels was stained intensely with ENOS, the gradient of which fell away from the tumour. NO, being produced by endothelial cells, may contribute to the increased permeability of the blood–brain barrier and vasodilation of the tumour vessels. It may also contribute to oedema formation not only by increasing the vascular permeability but also by inducing and maintaining vasodilation in and around the tumours. In addition, this study has also demonstrated the presence of BNOS-positive glial cells in three grades of astrocytic tumours. The intensity of staining with BNOS

decreased in areas T 1 and T 2 . GFAP labelling showed that a large number of the glial cells were reactive astrocytes. It has already been demonstrated that astrocytes produce NO in response to several stimuli including lipopolysaccharide, glutamate and epinephrine (Agullo and Garcia, 1992). Bioassay studies suggest that activated astrocytes produce NO in quantities sufficient to produce relaxation of cerebral arteries (Murphy et al., 1992; Faraci and Brian, 1994). It is likely that the NOS-immunopositive glial cells observed in this study may have contributed to increased vascular permeability and vasodilation by producing NO. Another unique observation, hitherto unreported, was the presence of densely staining MacNOS-positive perivascular infiltrate, probably lymphocytes, and macrophages, seen in four tumours. These cells were present both inside and around the blood vessels and their significance remains to be determined. Possibly, they are cells of the immune system which were stimulated by the cytokines to produce NO and which may also contribute to oedema formation.

A. Bakshi et al. / Journal of Neurological Sciences 155 (1998) 196 – 203

In conclusion, this study has shown that NOS is expressed in human brain tumours and in the surrounding brain parenchyma. Vascular endothelium is the main site of NO production, although glial cells may also be induced to produce it. The expression of both ENOS and BNOS demonstrates a gradient falling away from the tumours, in some cases. This suggests that the processes leading to NO production are centred in the tumour and not in the surrounding brain.

References Agullo, R., Garcia, A., 1992. Different receptor mediated stimulation of nitric oxide-dependent cGMP formation in neurons and astrocytes in culture. Biochem. Biophys. Res. Commun. 182, 1362–1368. Andrade, S., Hart, I., Piper, P., 1992. Inhibitors of nitric oxide synthase selectively reduce flow in tumour-associated neovasculature. Br. J. Pharmacol. 107, 1092–1095. Baethman, A., Oettinger, W., Rothenfusser, W., Kempski, O., Unterberg, A., Geiger, R., 1980. Brain edema factors: Current state with particular reference to plasma constituents and glutamate. Adv. Neurol. 28, 171–195. Brightman, M.W., Broadwell, R.D., 1976. The morphological approach to the study of normal and abnormal brain permeability. Adv. Exp. Med. Biol. 69, 41. Buttery, L.D., Springall, D.R., Andrade, S.P., Polak, J.M., 1993. Induction of nitric oxide synthase in the neo-vasculature of experimental tumours in mice. J. Pathol. 171, 311–319. Chan, P.H., Schmidley, J.W., Fishman, R.A., Longar, S.M., 1984. Brain injury, edema, and vascular permeability changes induced by oxygen derived free radicals. Neurology 34, 315–320. Cobbs, C.S., Brenman, J.E., Aldape, K.D., Bredt, D.S., Israel, M.A., 1995. Expression of nitric oxide synthase in human central nervous system tumours. Cancer Res. 55, 727–730. Criscuolo, G.R., 1993. The genesis of peritumoral vasogenic brain edema and tumour cysts: A hypothetical role for tumour derived vascular permeability factor. Yale J. Biol. Med. 66, 277–314.

203

Del Maestro, R.F., Megyesi, J.F., Farrell, C.L., 1990. Mechanisms of tumour associated edema: A review. Can. J. Neurol. Sci. 17, 177–183. Faraci, F.M., Brian, J.E., 1994. Nitric oxide and cerebral circulation. Stroke 25, 692–703. Ito, U., Reulen, H.J., Huber, P., 1986. Spatial and quantitative distribution of human peritumoral brain edema in computer tomography. Acta Neurochir. (Wien) 81, 53–60. Jenkins, D.C., Charles, I.G., Thomsen, L.L., Rhodes, P., Moncada, S., 1995. Role of nitric oxide in tumour growth. Proc. Natl. Acad. Sci. USA 92, 4392–4396. Kiechle, F.L., Malinski, T., 1993. Nitric oxide: Biochemistry, pathophysiology and detection. Am. J. Clin. Pathol. 100, 567–575. Kimelberg, H.K., 1995. Current concepts of brain edema. J. Neurosurg. 83, 1051–1059. Koedel, U., Bernalowicz, A., Paul, R., Frei, K., Pfister, H.W., 1995. Experimental pneumococcal meningitis: Cerebrovascular alterations, brain edema and meningeal inflammation are linked to the production of nitric oxide. Ann. Neurol. 37, 313–323. Maeda, H., Noguchi, Y., Sato, K., Akaike, T., 1994. Enhanced vascular permeability in solid tumour is mediated by nitric oxide and inhibited by both new nitric oxide scavenger and nitric oxide synthase inhibitor. Jpn. J. Cancer Res. 85, 331–334. Moncada, S., Palmer, R.M.J., Higgs, E.A., 1991. Nitric oxide: Physiology, pathophysiology and pharmacology. Pharmacol. Rev. 43, 109– 142. Murphy, S., Kardos, S., Moore, S.A., Orgren, K.I., Faraci, F.M., 1992. Astrocyte modulation of cerebral vessel function via eicasanoids and nitrosyl factors. Soc. Neurosci. Abstr. 18, 784. Oettinger, W., Baethman, A., Rothenfusser, W., Geiger, R., Mann, K., 1976. Tissue and Plasma Factors in Cerebral Edema. Dynamics of Brain Edema. Springer, Berlin, pp. 161–163. Orucevic, A., Lala, P.K., 1996. N-Nitro-L-arginine methyl ester, an inhibitor of nitric oxide synthase ameliorates interleukin-2 induced capillary leakage and reduces tumour growth in adenocarcinoma bearing mice. Br. J. Cancer 73, 189–196. Oury, T.D., Piantadosi, C.A., Crapo, J.D., 1993. Cold induced brain edema in mice: Involvement of extracellular superoxide dismutase and nitric oxide. J. Biol. Chem. 268, 15394–15398. Reichhman, H.R., Farrell, C.L., Del Maestro, R.F., 1986. Effects of steroids and nonsteroidal anti-inflammatory agents on vascular permeability in a rat glioma model. J. Neurosurg. 65, 233–237.