Human tumour necrosis factor: precursor structure, expression and homology to lymphotoxin

Human tumour necrosis factor has about 30% homology in its amino acid sequence with lymphotoxin, a lymphokine that has similar biological properties. Recombinant tumour necrosis factor can be obtained by expression of its complementary DNA in Escherichia coli and induces the haemorrhagic necrosis of transplanted methylcholanthrene-induced sarcomas in syngeneic mice.     

Characterization of receptors for human tumour necrosis factor and their regulation by gamma-interferon

Tumour necrosis factors, TNF-alpha and TNF-beta (previously called lymphotoxin), are the products of activated monocytes and lymphocytes, respectively, and both have recently been purified, sequenced and cloned by recombinant DNA methods, revealing 35% identity and 50% homology in the amino-acid sequence. Both proteins have been found to be specifically toxic to many tumour cells. Furthermore, it has been reported that various interferons are synergistic with TNF for anti-tumour effects in vitro, while activities attributed to the two proteins have also been shown to necrotize various tumours in vivo. We have now prepared 125I-labelled highly purified recombinant human TNF-alpha to study in detail its binding to the human cervical carcinoma cell line ME-180. Our results indicate that there is a single class of specific high-affinity receptors for TNF on this cell line which has a Kd of about 0.2 nM and an average of 2,000 receptor sites per cell. The binding of labelled TNF-alpha to these cells can be inhibited by both TNF-alpha and TNF-beta but not by gamma-interferon (IFN-gamma). However, preincubation of cells with IFN-gamma increases the total number of TNF receptors two to threefold without any significant change in the affinity constant. This is the first report that TNF-alpha and -beta share a common receptor and that the receptors can be up-regulated by interferon. Our results may explain previous observations regarding similar biological activities observed for these two cytotoxic proteins and also their synergistic action with interferons.     

Mediation of mouse natural cytotoxic activity by tumour necrosis factor

Natural cell-mediated cytotoxic activity in the mouse has been associated with two types of effector cells, the natural killer (NK) cell and the natural cytotoxic (NC) cell, which seem to differ with regard to their patterns of target selectivity, cell surface characteristics and susceptibility to regulatory factors. During studies on the mechanism of action of cytotoxic molecules, it became evident that WEHI-164, the prototype NC target cell, was highly susceptible to direct lysis by both human and mouse recombinant tumour necrosis factor (TNF). Here we show that NC, but not NK activity mediated by normal splenocytes, is abrogated by rabbit antibodies to recombinant and natural TNF, respectively. Thus, the cell-mediated activity defined as NC is due to release of TNF by normal spleen cells and does not represent a unique natural effector mechanism.     

Cloning and expression in Escherichia coli of the gene for human tumour necrosis factor

Tumour necrosis factor (TNF) was found originally in mouse serum after intravenous injection of bacterial endotoxin into mice primed with viable Mycobacterium bovis, strain Bacillus Calmette-Guerin (BCG). TNF-containing serum from mice is cytotoxic or cytostatic to a number of mouse and human transformed cell lines, but less or not toxic to normal cells in vitro. It causes necrosis of transplantable tumours in mice. TNF also occurs in serum of rat, rabbit and guinea pig. Rabbit TNF has been purified recently to give a single band on SDS-polyacrylamide gel electrophoresis (PAGE). The purified TNF had a relative molecular mass (Mr) 40,000 +/- 5,000 measured by gel filtration, and 17,000 by SDS-PAGE. Its isoelectric point is 5.0 +/- 0.3. The necrotic activity in vivo and the cytotoxicity in vitro are produced by the same substance. The gene encoding TNF has been identified in a human genomic DNA library using as a probe a cloned cDNA encoding a portion of rabbit TNF. The regions of this gene encoding an amino-acid sequence corresponding to mature TNF have been expressed in Escherichia coli and the product of this expression isolated in pure form and shown to produce necrosis of murine tumours in vivo.     

Tumour necrosis factor and lymphotoxin genes map close to H-2D in the mouse major histocompatibility complex

Tumour necrosis factor (TNF-alpha) and lymphotoxin (TNF-beta) are related proteins, secreted by macrophages and lymphocytes respectively, which play a role in destruction of tumour cells and virally infected cells (for reviews see refs 1,2). TNF-alpha is a non-glycosylated protein of relative molecular mass 17,000 (Mr 17 K), whereas TNF-beta is a glycoprotein of Mr 25 K. Both TNF-alpha and TNF-beta aggregate into multimers and act through the same receptor molecule on target cells. Genes encoding these two TNF proteins have been cloned from mouse and man and in both are closely linked, being separated by approximately 1 kilobase (kb) of DNA. In the mouse these genes are located on chromosome 17, but in man they are on the short arm of chromosome 6. This segment of chromosome 6 also contains the genes of the major histocompatibility complex (MHC), as does chromosome 17 in the mouse. To find out whether the TNF genes are located within the MHC, we used polymorphic restriction sites to analyse a panel of MHC congeneic and intra-MHC recombinant mouse strains. Initially, we mapped the TNF genes the D or Qa region in the distal half of the mouse MHC. We then studied a gene cluster encompassing part of the D and Qa regions and found the TNF genes are located 70 kb proximal to the D gene.     

Formation of a nine-subunit complex by HLA class II glycoproteins and the invariant chain

HLA class II molecules are heterodimeric transmembrane glycoproteins that bind and present processed antigenic peptides to CD4-positive T lymphocytes. Intracellularly, class II molecules associate with a third subunit termed the invariant (I) chain. Here we describe the physical characteristics of the intracellular class II alpha beta I complex. Chemical crosslinking, size exclusion chromatography and sedimentation velocity studies demonstrate that the alpha beta I complex is a nine-subunit transmembrane protein that contains three alpha beta dimers associated with an I chain trimer. The organization of class II alpha- and beta-subunits in such a multimer may have a role in the documented ability of the I chain to inhibit peptide binding to class II molecules. In addition, the formation of the nine-chain complex may induce the structural changes necessary to overcome the cytoplasmic retention signal responsible for the localization of free I chain in the endoplasmic reticulum, releasing class II-I chain complexes for transport to endosomes.     

A minimum of four human class II alpha-chain genes are encoded in the HLA region of chromosome 6

The major histocompatibility complex (MHC) in man, also called the HLA region, is located on the short arm of chromosome 6 and encodes antigens involved in immunological processes. The class II HLA antigens consist of two noncovalently associated polypeptide chains, one of molecular weight 34,000 (alpha) and the other of molecular weight 29,000 (beta). The extensive polymorphism of the beta chain(s) has allowed the genetic mapping of the corresponding beta gene(s) to the HLA-DR region. cDNA clones for the HLA-DR alpha chain have been used to map the non-polymorphic DR alpha-chain gene to chromosome 6 using mouse-human somatic cell hybrids. Similarly, the DR alpha-chain gene has been mapped to the short arm of chromosome 6 centromeric to the HLA-A, -B and -C loci by in situ hybridization experiments. We isolated a cDNA clone that is related to the DR alpha chain and encodes the class II antigen DC alpha chain. We describe here how this DC alpha clone was used to find two or three additional alpha-chain genes by cross-hybridization and how HLA-antigen loss mutants of a human lymphoblastoid cell line (LCL) were used to ascertain that these additional class II antigen alpha-chain genes are also located in the HLA region.     

Stimulation of bone resorption and inhibition of bone formation in vitro by human tumour necrosis factors

When leukocytes are exposed to mitogens or antigens in vitro, they release bone-resorbing activity into the culture supernatants which can be detected by bioassay. Like many lymphocyte-monocyte products, this activity has been difficult to purify because of its low abundance in activated leukocyte cultures and the unwieldy bioassay required to detect biological activity. Partially purified preparations of this activity inhibit bone collagen synthesis in organ cultures of fetal rat calvariae. Recent data suggest that both activated lymphocytes and monocytes release factors which could contribute to this activity. Recently, monocyte-derived tumour necrosis factor alpha (TNF-alpha) and lymphocyte-derived tumour necrosis factor beta (TNF-beta) (previously called lymphotoxin), two multifunctional cytokines which have similar cytotoxic effects on neoplastic cell lines, have been purified to homogeneity and their complementary DNAs cloned and expressed in Escherichia coli. As both of these cytokines are likely to be present in activated leukocyte supernatants, we tested purified recombinant preparations for their effects on bone resorption and bone collagen synthesis in vitro, and report here that both cytokines at 10(-7) to 10(-9) M caused osteoclastic bone resorption and inhibited bone collagen synthesis. These data suggest that at least part of the bone-resorbing activity present in activated leukocyte culture supernatants may be due to these cytokines.     

A novel HLA class II molecule (DR alpha-DQ beta) created by mismatched isotype pairing

Human major histocompatibility complex (MHC) class II molecules are heterodimeric glycoproteins composed of non-covalently associated alpha and beta chains. Only isotype-matched alpha-beta associations have been described in man; these can occur either by cis- or trans-complementation (HLA-DR, DQ, DP). Here evidence is provided for the existence of a new type of hybrid molecule (DR alpha-DQ beta) arising by mixed-isotype pairing in human B-cell lines. Class II isotype-mismatched heterodimers have been recently reported in the mouse after transfection of class II genes, and our data demonstrate that such interisotypic pairing can occur in untransfected cells. This crosspairing greatly enhances the repertoire of the class II antigens that regulate immune responses and leads us to reconsider the HLA-disease association.     

Murine cells expressing an HLA molecule are specifically lysed by HLA-restricted antiviral human T cells

Class I HLA (histocompatibility locus antigen) molecules are the targets of allospecific cytolytic T lymphocytes (CTL) in graft rejection, and constitute the restricting elements necessary for the interaction between antiviral CTL and virus-infected cells. Cells expressing only one HLA in the absence of other human molecules would provide a remarkable model for studying the function of these molecules. However, HLA+ murine cells transfected with human genes are generally not lysed by allospecific human CTL, and this is ascribed to insufficient HLA expression, lack of human beta 2-microglobulin, alteration of HLA molecules or absence of receptors for human T8 or LFA1 molecules in murine cells. Here we report, for the first time, the specific lysis of virus-infected HLA+ murine cells by HLA-restricted antiviral human CTL. Therefore, these murine cells constitute an excellent model for studying the role of HLA molecules.     

Epithelial cells expressing aberrant MHC class II determinants can present antigen to cloned human T cells

The first step in the induction of immune responses, whether humoral or cell mediated, requires the interaction between antigen-presenting cells and T lymphocytes restricted at the major histocompatibility complex (MHC). These cells invariably express MHC class II molecules (HLA-D region in man and Ia in mouse) which are recognized by T cells of the helper/inducer subset in association with antigen fragments. Interestingly, in certain pathological conditions, for example in autoimmune diseases such as thyroiditis and diabetic insulitis, class II molecules may be expressed on epithelial cells that normally do not express them. We speculated that these cells may be able to present their surface autoantigens to T cells, and that this process may be crucial to the induction and maintenance of autoimmunity. A critical test of this hypothesis would be to determine whether epithelial cells bearing MHC class II molecules (class II+ cells) can present antigen to T cells. We report here that class II+ thyroid follicular epithelial cells (thyrocytes) can indeed present viral peptide antigens to cloned human T cells.     

Recognition of pre-processed endogenous antigen by class I but not class II MHC-restricted T cells

Class I and class II MHC-restricted T lymphocytes recognize non-native forms of antigen. The presentation of antigen to these two classes of T lymphocytes can occur through distinct pathways. Several mechanisms, including differences in antigen processing in different intracellular compartments, have been proposed to account for these pathway differences. Here we describe a T-cell epitope located on the influenza virus haemaglutinin, which is recognized by both class I and class II MHC-restricted cytolytic T lymphocytes (CTL). When expressed de novo in target cells, from a synthetic minigene encoding only the epitope, this pre-processed antigenic site is recognized by class I but not class II MHC-restricted T lymphocytes, even though target cells treated with the exogenously introduced peptide can be recognized by both classes of T cells. Because endogenous expression of the pre-processed antigenic fragment results in differential presentation to class I and class II MHC-restricted CTL, differences between the two different pathways of presentation could lie not at the level of processing but at the level of targeting and/or interaction of processed antigen with MHC.     

Defective processing and presentation of exogenous antigens in mutants with normal HLA class II genes

Presentation of an exogenous protein antigen to helper (CD4+)T-lymphocytes by antigen presenting cells (APC) generally requires that the APCs degrade the native protein antigen into an immunogenic peptide, a process termed 'antigen processing', and that this peptide bind to a major histocompatibility complex (MHC) class II molecule. The complex of peptide and MHC molecule on the APC surface provides the stimulatory ligand for the alpha beta T cell receptor. The intracellular pathways and molecular mechanisms involved in the generation of the peptide-MHC complex are not well understood. Here, we describe several mutant APCs which are altered in their ability to present native exogenous protein antigens but effectively present immunogenic peptides derived from these proteins. The lesions in these mutants are not in the class II structural genes, but they affect the conformation of mature class II dimers.     

Vascular endothelial growth factor is a potential tumour angiogenesis factor in human gliomas in vivo.

Clinical and experimental studies suggest that angiogenesis is a prerequisite for solid tumour growth. Several growth factors with mitogenic or chemotactic activity for endothelial cells in vitro have been described, but it is not known whether these mediate tumour vascularization in vivo. Glioblastoma, the most common and most malignant brain tumour in humans, is distinguished from astrocytoma by the presence of necroses and vascular proliferations. Here we show that expression of an endothelial cell-specific mitogen, vascular endothelial growth factor (VEGF), is induced in astrocytoma cells but is dramatically upregulated in two apparently different subsets of glioblastoma cells. The high-affinity tyrosine kinase receptor for VEGF, flt, although not expressed in normal brain endothelium, is upregulated in tumour endothelial cells in vivo. These observations strongly support the concept that tumour angiogenesis is regulated by paracrine mechanisms and identify VEGF as a potential tumour angiogenesis factor in vivo.