The induction of malignant tumors of the central and peripheral nervous system of the rat by pulse exposure to the alkylating N
-nitroso carcinogen ethylnitrosourea (EtNU1
The abbreviations used are: EtNU
polyacrylamide gel electrophoresis
high pressure liquid chromatography
polymerase chain reaction
) provides a model for the molecular analysis of cell type-specific carcinogenesis (
Rajewsky et al., 1977
- Rajewsky M.F.
- Augenlicht L.H.
- Biessmann H.
- Goth R.
- Hülser D.F.
- Laerum O.D.
- Lomakina L.Ya.
). Specific DNA-alkylation products formed by this carcinogen in proliferation-competent target cells may lead to mutations in genes critically associated with cell differentiation and malignant transformation. Thus, a specific mutation of the neu/erbB-2
gene, whose expression is diagnostic for EtNU-induced malignant schwannomas in the peripheral nervous system, may be traced back to a small number of mutant Schwann precursor cells with uncontrolled proliferation which first become detectable shortly after carcinogen exposure (
). In the central nervous system, target genes of similar relevance to malignant conversion by chemical carcinogens have not yet been identified.
The neuro-oncogenic effect of EtNU in the rat is strongly dependent on the developmental stage at the time of carcinogen exposure, with maximum susceptibility during the late prenatal to early postnatal period. To characterize cell subpopulations in the developing central nervous system in terms of cell lineage relationships, phenotypic differentiation, and relative risk of malignant conversion by EtNU, monoclonal antibodies (mAb) have been produced after immunization with intact fetal brain cells isolated at different stages of development (
). One of these antibodies (mAb RB13-6) recognizes a cell surface antigen expressed by a small subpopulation of neural precursor cells, but not by cells of the mature brain. In contrast, sustained expression of the RB13-6 antigen is observed in the cells of most EtNU-induced brain tumors analyzed thus far and in all tumorigenic neuroectodermal cell lines derived from such tumors or from fetal brain cells that underwent malignant transformation in culture after exposure to EtNU in vivo
cell line BT4Ca; Kindler-Röhrborn et al.
S. Blass-Kampmann, T. Bilzer, and M. F. Rajewsky, manuscript in preparation.
Expression of the RB13-6 antigen, as observed in brain tumors, could thus specify differentiating neural precursor cells particularly sensitive to malignant conversion by EtNU.
Prior to biochemical characterization and molecular cloning of the RB13-6 antigen, the rate and equilibrium constants for binding of mAb RB13-6 to the surface of intact cells were determined and the number of antigenic determinants/cell was calculated to be ∼6,000/fetal brain cell and ∼30,000/BT4Ca cell (
). Initial biochemical analyses suggested that the RB13-6 antigen was a membrane glycoprotein with an apparent molecular mass of 125 kDa (
Cleeves, V., 1991, Biochemische Charakterisierung Antikörper-definierter Zelloberflächenmoleküle Normaler und Maligner Neuraler Zellen der Ratte. Ph.D. dissertation, University of Essen, Essen, Germany.
). Here we describe the characterization of the membrane glycoprotein recognized by mAb RB13-6 and its large scale purification by immunoaffinity chromatography, followed by sequencing of endoproteinase-generated peptides, and isolation of the cDNA encoding gp130RB13-6
from a fetal rat brain cDNA library. Potential structural and functional features of gp130RB13-6
are discussed on the basis of its significant sequence homology to the human and murine plasma cell membrane glycoprotein PC-1 (
The surface antigen recognized by mAb RB13-6 on cells of the malignant neuroectodermal rat cell line BT4Ca was identified by affinity purification and cDNA cloning as a glycoprotein (gp130) related to the human and murine PC-1 proteins. PC-1 was originally described as a cell surface antigen of differentiated, antibody-secreting B cells (
), but later found to be expressed in other tissues also (
). The class II transmembrane protein PC-1 is a homodimer consisting of two disulfide-linked 115-kDa polypeptides. cDNA cloning of murine and human PC-1 had neither revealed homology to other proteins nor possible biological functions (
). Rebbe et al.
(1991) reported that nucleotide pyrophosphatase/alkaline phosphodiesterase I activity is associated with murine PC-1 which was confirmed by expression of enzymatically active PC-1 in CHO cells (
). PC-1 isolated from bovine liver can be phosphorylated at a threonine residue, and it has been suggested that it may act as a threonine-specific ectoprotein kinase (
- Oda Y.
- Kuo M.-D.
- Huang S.S.
- Huang J.S.
). Recently, evidence for the existence of a soluble form of PC-1 generated by proteolytic cleavage has been published (
The protein purified from BT4Ca cells most probably corresponds to the antigenic determinant detected by the same antibody on a subpopulation of neural precursor cells of rat brain and in EtNU-induced brain tumors of the rat (
, 1994). Direct proof of this assumption by biochemical analysis of different cell types of the immature rat brain has thus far not been possible due to the relatively weak antigen expression and the small number of antigen-positive fetal brain cells (
The EtNU-induced, malignant cell line BT4Ca was chosen for biochemical analysis and purification of gp130RB13-6
because it combines a high rate of cell proliferation with high antigen expression. Several parameters of the purification procedure had to be defined and optimized before sufficient amounts of gp130RB13-6
could be isolated, probably due to an unstable epitope recognized by the antibody, retaining its native structure only under certain conditions. The unusually high excess of affinity matrix required may be a consequence of the moderate binding affinity of the antibody (K
= 1.87 × 108
) and the low concentration of solubilized antigen.
The electrophoretic mobility of unreduced gp130RB13-6
isolated from BT4Ca cells indicates that it is not a disulfide-linked homodimer, in contrast to the related PC-1 glycoproteins which migrate in SDS-polyacrylamide gels in accordance with the molecular mass expected for a dimer (
). It will be of interest to investigate different cell types, including transformed phenotypes, for the presence of dimeric gp130RB13-6
and with respect to the monomer-dimer equilibrium. This analysis will require the production of a new anti-gp130RB13-6
antibody suitable for Western blot analysis.
The sequencing of endoproteinase-generated gp130RB13-6
peptides has facilitated the isolation of the corresponding cDNA which was aided by inclusion of an RT-PCR step to generate a homologous probe for stringent screening of a cDNA library. Assuming the first ATG in-frame to be the translation initiation site, as also concluded from the similarity to the PC-1 proteins, the molecular mass calculated from the sequence (100 kDa) is lower than the 115 kDa expected from the electrophoretic mobility of N
. This difference may be explained by additional protein modifications and/or bound nonionic detergent molecules used for solubilization, which influence SDS binding and mobility. However, it cannot be excluded and will be subject to further investigation that the isolated cDNA represents a shorter splice variant of gp130RB13-6
, although longer cDNA sequences were not detected in the fetal brain cDNA library. The existence of three phosphorylated rat liver proteins reacting with an anti-PC-1 antibody described by
invites speculations on the existence of such variants of the related rat protein. Apart from such speculation, the isolated cDNA must be regarded as coding for the isolated protein because all sequenced peptides obtained from the purified protein occur in the derived amino acid sequence.
The limited sequence homology between gp130RB13-6
and PC-1 compared to the high degree of conservation between human and murine PC-1 sequences (Fig. 6
) suggests that gp130RB13-6
may not be the rat homolog of PC-1, but rather a different member of a larger protein family. This will have to be confirmed by cloning of the human gp130RB13-6
and rat PC-1 homologs, respectively.
The sequence motifs of gp130RB13-6
and its homology to PC-1 point to several possible functions of this molecule in normal and transformed cells. 5′-Nucleotide pyrophosphatase activity, as expected from the presence of a highly conserved catalytic domain (see Fig. 7
), was shown for PC-1 (
, 1993) and could be confirmed for purified gp130RB13-6
. The biological role of a molecule with this primary function could be: (i) hydrolysis of nucleoside triphosphates to promote uptake of metabolically used nucleosides and (ii) control of the concentration of extracellular adenosine which can influence various important cellular functions (
). Ectonucleotidase activity was also found as an accompanying function of cell adhesion molecules C-CAM (
) and N-CAM (
). Ectoprotein kinase activity, as reported for PC-1, must also be considered for the RB13-6 antigen, although the in vitro
data for PC-1 (
, 1993) are only circumstantial evidence for a similar function in vivo
. The predicted intracellular domain contains a potential serine phosphorylation site for a cAMP-dependent protein kinase in a context (KKDSLKR) reminiscent of a consensus sequence of integrin α subunits (K X
GFFKR) responsible for calreticulin binding (
). In integrin α subunits, this consensus sequence is also found immediately adjacent to the transmembrane domains. Two somatomedin B-like domains (
) were detected in the cDNA-derived amino acid sequence of gp130RB13-6
. Such domains have thus far only been found in PC-1, in two proteins with unknown function (placental protein 11,
), and in the cell-substrate adhesion molecule vitronectin. The serum peptide somatomedin B is released by proteolytic cleavage from vitronectin which interacts with integrin receptors via an RGD tripeptide (
). The RGD sequence of gp130RB13-6
invites the hypothesis that it also binds to integrins, as other membrane proteins and components of the extracellular matrix containing this tripeptide (
). As speculated on the basis of data indicating the existence of soluble PC-1 (
), proteolytic cleavage of gp130RB13-6
could release a soluble protein fragment capable of blocking integrin receptors. In view of the putative structural features deduced from the presence of sequence motifs, involvement of gp130RB13-6
in cell-cell or cell-substrate adhesion seems possible and would be in accordance with an expected function during physiological brain development and its subversion by malignant transformation. Future experiments will be focused on the elucidation of the main function of gp130RB13-6
in normal and malignant neural cells. The purification and cDNA cloning of gp130RB13-6
provides the basis for such functional characterization and the search for alterations in malignant phenotypes.
© 1995 ASBMB. Currently published by Elsevier Inc; originally published by American Society for Biochemistry and Molecular Biology.