MATERIAL AND METHODS

Tumour biopsies were obtained from routine neurosurgical management. Tumours were usually preserved in their native state and processed as frozen sections, fixed in buffered formalin and glutaraldehyde for neuropathological diagnoses using routine histological methods.

Using immunohistochemistry, 95 single specimens were analysed for CRLR immunostaining and directly compared with the staining pattern of glial fibrillary acid protein (GFAP). Table 1 gives the histological WHO classification of the selected tumours.

Development, purification, and application of antibodies

The raising of rabbit antisera against the commercially synthesised human CRLR sequence conjugated with thyroglobulin and mixed with Freund’s adjuvant has been described in detail previously.¹² One antiserum MR 567 gave best results in western blotting and immunohistochemistry and was therefore affinity purified.

Immunohistochemistry

Immunohistochemistry was performed on paraffin wax embedded 1–2 μm sections, mounted on SuperFrost®Plus slides. After dewaxing and rehydration, sections were microwaved for 20 minutes to assist antigen exposure. After rapid cooling and two washes, the sections were incubated with primary antibody for 30 minutes at room temperature. The most effective dilution of the antibody in this series of staining procedures was 1/100 to 1/200. The routinely used peroxidase–antiperoxidase method,¹⁶ which involves the use of three subsequent antibodies and diaminobenzidine as chromogen, was used for visualisation of the antigens.

Evaluation

The various patterns of CRLR immunostaining were evaluated, including the relative number of CRLR immunostained tumour cells and the degree of co-staining with the astrocyte marker, GFAP. The counting device was a simple quadrangular intersectional microscopic ocular counting field. The clinical classification of the tumour tissue was determined by one of the authors (HDM), according to the rubrics and grading scheme of the most recent issue of the WHO brain tumour classification.¹⁷

RESULTS

In human gliomas, CRLR immunostaining was seen predominantly in two cell types; namely, vascular endothelial cells and astrocytic tumour cells. In addition, neuronal perikarya were also immunostained in well preserved neighbouring cortical tissue.

Vessels

Gliomas of increasing malignancy produce a wealth of blood vessels of normal and pathological morphology. Various vessel types exhibiting CRLR immunoreactivity could be distinguished. In low grade astrocytomas, larger veins stain positively for CRLR (fig 1A) and, therefore, could be distinguished from the small vacuoles that are usually formed by this tumour type. In highly aggressive gliomas (WHO III and IV), blood vessel proliferation may start with endothelial growth, which could be detected by the presence of a double layer of endothelial cells undergoing mitosis (fig 1B). Garland-like capillary and angiomatous vessel formations, typical of malignant gliomas, were clearly CRLR immunostained, whereas adjoining tumour cells were unstained or only slightly stained (fig 1C, D). Although vessels were surrounded by CRLR immunostained tumour cells, the vascular contours could be clearly distinguished from these tumour cells, because the CRLR immunostaining was evidently restricted to the endothelial layer of the vessels missing in the adjacent vessel walls.

Tumour cells

Tumour cells were stained inconsistently, but clear and distinctive CRLR immunostaining of single tumour cells, in addition to small and large clusters of tumour cells, could be seen frequently. Different patterns of CRLR immunoreactivity in tumour cells could be identified. Single immunostained cells, within a rather inconspicuous surrounding, exhibited cytoplasmic extensions and were recognisable as typically shaped astrocytes (fig 2A). They correspond to the protoplasmic subtype common in reactive and tumorous conditions. These were found to be preferentially aggregated around vessels and a small distance from the vessel (fig 2B).

Some, but not all, “stuffed” gemistocytes were intensely CRLR immunoreactive. These cells occur in small clusters, similar to other multipolar tumour cells, and possibly represent focal cellular multiplication.

A further pattern of expression consisted of diffuse staining in most of the newly growing tumour cells. This was recognisable in the form of a very “loose network” of low grade fibrillary astrocytic cells (fig 2C), in addition to a much denser bipolar arrangement of anaplastic cells, and in the variegated cellular morphology pattern of highly malignant glial tumour cells around pathological vessels (fig 2D).

Neurones

Cortical neurones, especially the perikarya, presumably entrapped in the tumour itself, also exhibited strong and constant CRLR immunoreactivity (fig 3). Most of them had an altered shape, presumably as a result of the adjacent tumour tissue acting upon them.

Cellular localisation and correlation of relevant structures

Staining in all cases was within the confines of the perinuclear cytoplasm. This held true for neuronal perikarya, endothelial cells, and astrocytes. However, astrocytes did not show a constant intensity of immunostaining. Some giant tumour cells were stained in the periphery although the protoplasmic centre was negative. Gemistocytes displayed glial fibrillary acid protein and CRLR immunoreactivity. Using pertinent aniline dyes and GFAP immunostaining, the restriction of CRLR antigen to a subpopulation of astrocytes and endothelial lining could be demonstrated.

Morphometric and statistical results

Numbers of GFAP positive cells in defined areas of gliomas of different grades were counted and compared with CRLR positive cell numbers in the randomly selected fields. There was a small decrease in the proportion of GFAP immunostained cells with ascending grade of malignancy (glioma II (mean, 24.8%; SD, 14.8%), glioma III (mean, 19.8%; SD, 10.2%), and glioblastoma (mean, 14.8%; SD, 9.5%)). No correlation was found for CRLR immunostained cells. The highest numbers of CRLR immunostained cells were found in glioma III tumours (mean, 41.8%; SD, 33.6%).

DISCUSSION

CRLR is the cognate receptor of two vasoactive peptides—adrenomedullin, which is produced by a variety of tissues and cells in the cardiovascular system, and CGRP, which is produced by perivascular nerves.^18,19 We have recently used our anti-CRLR antibody to demonstrate the widespread distribution of CRLR in peripheral tissues¹⁴ and its prevalence in the microvasculature. Here, we have demonstrated the distribution of CRLR in human gliomas of different grades, and have shown that it is expressed by different cell types within the tumour tissue. To our knowledge, these are new findings, and they suggest that CRLR plays an important role in the pathology of human gliomas. In particular, it is important to view our findings in light of the well documented evidence that adrenomedullin may play an important role in glioma tumorigenesis. The presence of CRLR immunoreactivity in the vascular endothelium and in astrocytic tumour cells is consistent with the notion that adrenomedullin may influence glioma growth by means of both direct mitogenic and indirect angiogenic effects.²⁰

It has been proposed by Folkman that tumour growth depends on new vessel growth,²¹ and recent studies²² indicate that the growth of gliomas exhibits two growth phases. Initially, normal cerebral vessels are recruited to support early tumour growth. The second phase is characterised by angiogenesis and neovascularisation, which appears to be preceded by transient apoptosis of the pre-existing vascular cells. Both adrenomedullin²³ and CGRP²⁴ have been shown to promote endothelial growth in in vitro studies of human umbilical vein cells. In addition, there is in vivo evidence of adrenomedullin acting as a mediator of physiological angiogenesis.²⁵ Furthermore, vasodilation is a key early process in angiogenesis.^26,27 Both adrenomedullin and CGRP are potent vasodilators. The high amounts of CRLR protein that we found in the endothelial cells of the various vascular conformations of human glioma suggest that CRLR plays a role in angiogenesis in gliomas. This should be further evaluated, in addition to the respective roles of adrenomedullin and/or CGRP.

“Our immunohistochemical analysis did not indicate a clear correlation between the concentration of the calcitonin receptor-like receptor and the tumour grade”

Recently, it has become increasingly evident that adrenomedullin plays an important role in tumour pathogenesis.^7,8,28–30 The expression of this endogenous CRLR ligand has been demonstrated in normal and neoplastic cells, including gliomas. Adrenomedullin acts as a glioma growth factor^20,31 via direct (possibly autocrine) actions, and by promoting vacularisation of glioma tissue. Our findings concerning the presence of CRLR protein in glioma tumour cells and vasculature are consistent with these modes of action.

Real time quantitative polymerase chain reaction has shown that the degree of expression of adrenomedullin in human gliomas correlates with the grade of tumour, with especially high expression in grade IV.²⁰ Our immunohistochemical analysis did not indicate a clear correlation between the concentration of the CRLR protein and the tumour grade. This suggests that the degree of vascularisation and tumour differentiation may not depend on concentrations of CRLR, but on those of its ligand adrenomedullin. In addition, the expression of RAMPs and their degree of expression in gliomas in relation to the tumour grade need to be considered in further studies.

Regions of hypoxia are common in solid tumours and are associated with the transcription of a battery of genes, including adrenomedullin in tumour cells.¹⁵ Adrenomedullin is thought to be a mediator of hypoxia driven angiogenesis,³² which is a major factor in tumour growth. It has been shown to protect tumour cells from hypoxia induced apoptosis, probably via activation of CRLR.³³ The antiapoptotic effects of adrenomedullin explain (at least partly) the mitogenic actions of adrenomedullin, which have been demonstrated in a variety of non-tumour and tumour cells.⁸

Our morphometric and statistical results comparing GFAP and CRLR containing neoplastic astrocytes lead to the notion that neoplastic glial cells expressing CRLR form a subset of astrocytic tumour cells mainly expressed in glioma grade III neoplastic astrocytic tumour cells.

Apparently normal cortical perikarya within the glioma tissues were also found to express CRLR protein, which is consistent with earlier findings of the presence of CGRP and CGRP binding sites within the human cerebral cortex.^34,35 Adrenomedullin has also been detected in rat cerebral cortical neurones (in pyramidal cells and small cells).³⁶

In conclusion, in our study we have demonstrated the wide distribution of CRLR in different grades of human glioma, especially in vessels, astrocytes, and neurones. Our findings are particularly important for glioma biology because CRLR may play a major role at distinct stages of angiogenesis and carcinogenic tumour progression, and may be a useful basis for further studies on this topic.