Genes involved in the control of tumor progression and their possible use for gene therapy

Gene Ther Mol Biol Vol 1, 381-398. March, 1998.

Genes involved in the control of tumor progression and their possible use for gene therapy

Georgii P. Georgiev¹, Sergei L. Kiselev¹, and Evgenii M. Lukanidin^1,2

¹Institute of Gene Biology, Russian Academy of Sciences, 34/5 Vavilov Street, 117334 Moscow, Russia; and

²Danish Cancer Society, Department of Cancer Molecular Biology, Copenhagen, Denmark.

__________________________________________________________________________________________________

Correspondence: Georgii P. Georgiev, Fax: +7-095-135 41 05, E-mail: georg@biogen.msk.su

Summary

Three major groups of genes may be used for cancer gene therapy: (i) oncogenes and tumor suppressor genes; (ii) genes involved in the control of tumor progression and metastasis; and (iii) genes encoding proteins protecting the organism from tumor cells. Each group contains numerous genes, and the discovery of new important genes is an exciting prospect in cancer research. We are working on the search and characterization of the genes over- or under-expressed in metastatic comparing to non-metastatic tumors of the same origin. Two mouse systems are being used: (i) VMR-0 (non-metastatic mammary adenocarcinoma cells) - VMR-100-Liv and VMR-100-Ov cells (metastatic preferentially to the liver or ovaries, respectively); and (ii) CSML-0 - CSML-100 (mammary adenocarcinoma cells non-metastatic and metastatic to the lungs, respectively). Several different genes were found to be over-expressed in metastatic cells, but only few of them were shown to be necessary and sufficient for maintaining the metastatic phenotype using stably transfected cells and/or transgenic animals. Among them are the mts1 and c-met genes. The mts1 gene, encoding a calcium-binding protein of 101 amino acids of the S-100 family, was extensively characterized. Its expression induced a number of changes in cell functions connected with cytoskeleton features, attachment properties of the cell, mesenchyme formation and possibly tumor vascularization. As a multifunctional regulator, the mts1 gene is a promising target for gene therapy of cancer.

Other genes identified are over-expressed only in few metastatic tumors and do not seem to be connected directly with the acquisition of the metastatic phenotype. However, during the transfection experiments some interesting features emerged for these genes, raising the possibility of their exploitation in cancer gene therapy. The most interesting is the tag7 gene encoding a new cytokine, 182 amino acids long, with a far distant relation to cytokines of the TNF-Lymphotoxin family. The tag7 gene is expressed in lymphoid cells, in a limited set of other normal cells, and in few cancer cells including myelomas. The Tag7 protein is secreted to the culture medium and possesses a strong cytotoxic activity inducing apoptosis. VMR-0 cells were stably transfected with a construct containing the tag7 gene under control of the CMV promoter. The original VMR-0 tumors killed mice in one month after subcutaneous transplantation; animals displayed large necrotic foci at this stage. However, the VMR-0/tag7 cells, synthesizing very low amounts of Tag7 protein, exhibited dramatically different growth properties: they grew much slower; even after 4 months, no mice were killed by tumors arising from the transplanted cells and no necrotic foci were formed. Histological analysis of VMR-0/tag7 tumors showed a strong inhibition in mitotic rates and an enhanced rate of apoptosis compared to VMR-0 tumors. The tumors induced by transplantation of a mixture of VMR-0 and VMR-0/tag7 cells also grew much slower than VMR-0 cells alone, suggesting an activation of the immune system against tumor (tumor vaccination effect), which may be mediated through induction of CTL cells. Experiments with nude mice gave similar results. In fact at later stages of development in nude mice, VMR-0/tag7 tumors were completely eradicated. It seems that the effect of tag7 expression is complex and includes activation of an immune response as well as a direct cytotoxicity. The higher tag7 expression in culture cells is incompatible with cell survival. Experiments are in progress for further elucidating the role of Tag7 and its exploitation for the development of tumor vaccines.

I. Introduction

Three major groups of genes may be used for cancer gene therapy: (i) oncogenes and tumor suppressor genes; (ii) genes involved in the control of tumor progression and metastasis; and (iii) genes encoding proteins protecting the organism from tumor cells. Each group contains numerous genes, and the discovery of new important genes is an exciting prospect in cancer research. We are working on the search and characterization of the genes over- or under-expressed in metastatic comparing to non-metastatic tumors of the same origin. Below, we briefly summarize the general data on genes and proteins involved in the control of tumor progression (for more information see review articles in the Reference list). Thereafter, we present the data obtained in our laboratories on two genes from the second and third groups mentioned above. At least one of these genes and its protein product may be used for the gene therapy of cancer.

II. Tumor progression and tumor metastasis

The tumor is not a static formation. It is developing and changing constantly. This depends on the genetic instability of the tumor. As a result of transformation, tumor cells acquire a partial independence from regulatory signals arising from neighboring cells and grow more or less independently of these signals. Another important feature is the elimination of the cells with damaged DNA. Normally such cells cannot overcome the cell cycle checkpoints and progress through the apoptotic process leading to their death. The p53 protein plays an important role in the direction of damaged cells along the apoptotic way. Many mutations in the p53 gene, frequently occurring in tumors, lead to the loss in the ability to induce apoptosis and to down regulate cell proliferation. As a result, cells become able to survive after DNA damage, and this results in the increase in mutation rate in such cell populations. Interestingly, another protein involved in the control of apoptosis, the product of the bcl-2 gene which down regulates apoptosis, is a potential oncogene.

This, and possibly some other processes, result in accumulation of different types of mutations in tumor cells: translocations, loss of heterozygosity, point mutations and transpositions. Consequently, this leads to an accumulation of heterogeneity in tumor cell populations. If the mutated cell acquires some advantage for rapid growth or other properties useful for the cell itself, it has some good chances for survival and multiplication. Such changed cells may replace the original population of tumor cells over time. This phenomenon is known as tumor progression, that usually leads to appearance of a more malignant phenotype. Usually the same tumor contains cells with different genotypes and several clones obtained from the same tumor may differ in the level of malignancy.

It should be pointed out that in some cases, the malignant phenotype can appear just at the first stage of tumor development, as for example, in mouse mammary tumors induced by activation of the neu oncogene. However, in many cases, the malignancy develops in the course of tumor progression. One of the major features of malignant tumor cells is the ability to give metastases, i.e. new foci of tumor growth in distantly located regions of the organism.

The process of tumor progression and tumor metastasis is very complex and includes several independent steps: 1) vascularization of a primary tumor node; 2) detachment of tumor cell from the primary focus; 3) invasion into surrounding tissues including the blood vessels; 4) transfer to the new site and arrest at this site; 5) adhesion to the endothelial cells; 6) extravasation; 7) vascularization of a novel focus and its invasive growth at the new place. Each step depends on new special properties of tumor cells, such as ability to induce angiogenesis, detachment from or attachment to cell aggregates, cell invasiveness and cell motility (Fig. 1).

Each step is a complex one and is controlled by a number of genes and proteins. Therefore, one can expect a number of genes/proteins to be involved in the control of tumor metastasis. Several activated oncogenes themselves can generate a metastatic phenotype. However, in many other cases, the processes of oncogenesis and metastasis are uncoupled, and special genes are responsible for the appearance of the metastatic phenotype.

The genes for tumor progression and metastasis may be separated into two groups: effector genes and upstream regulatory genes. The protein products of effector genes directly determine the invasiveness and other features of the metastatic tumor, while proteins encoded by the genes in the second group act in an indirect way. They either control the expression of different effector genes or control some general cell function indirectly determining the features characteristic of malignant tumor cell.

III. Some main steps of tumor progression and some genes involved in their control

A. Angiogenesis.

Before vascularization, or angiogenesis, takes place the tumor can only grow up to 2 mm in diameter due to the shortage of nutrition. After vascularization has occurred, tumor cells can grow to much larger dimensions. In addition, tumor cells from vascularized tumors can penetrate into the blood vessels and this is the first step for metastasis development.

The invasion of endothelial cells into the tumor node and formation of capillary sprout are influenced by many different cytokines produced either by tumor cells or by normal inflammatory cells whenever inflammation accompanies tumor development. There are positive and negative cytokines. Among positive cytokines inducing tumor angiogenesis are fibroblast growth factors, especially bFGF, vascular endothelial growth factor (VEGF), interleukin-8 (IL-8) and many others. On the contrary, cytokines such as interferons (IFN-a and IFN-b), angiostatic cartilage-derived inhibitors and several others inhibit angiogenesis in tumor foci. The production of corresponding cytokines by tumor cells depends on the complex interactions between tumor and surrounding host cells. Sometimes, angiostatic cytokines are produced by primary tumor itself and this prevents overgrowth of metastases. In such cases, surgical removal of the primary focus may lead to induction of metastasis. The control of angiogenesis by antiangiogenic polypeptides, like angiostatin, endostatin and some other agents is an important approach for cancer therapy (see this volume).

B. Detachment of tumor cells from the original cell cluster.

The next important step in metastasis is the detachment of tumor cells from aggregates to enter the blood vessel. Cell aggregation depends on several factors, the most important being represented by cadherins, immunoglobulins, integrins and selectins.

Aggregation of the cells plays an especially important role in the progression of the cancers, or epithelial tumors, as epithelial cells originally form firm aggregates and have very little motility. They bind to each other and to the basal membrane. Adherence junctions between normal and tumor epithelial cells depend on the cadherin family of proteins. The best characterized member of the family, E-cadherin, is a ca. 120 kDa protein with a large N-terminal extracellular part containing four calcium-binding domains, a transmembrane domain and a C-terminal intracellular domain. The extracellular domains of E-cadherins located on different cells interact in a calcium-dependent way. The C-terminal domain interacts with the C-terminal region of b-catenin (plakoglobin), whose N-terminal part binds to a-catenin which interacts with the cytoskeleton. This chain of protein-protein interactions firmly holds epithelial cells at their position.

There is a strong correlation between cell behavior and E-cadherin content (Vlamincks et al. 1991). In transfection experiments, stable transfection and expression of the E-cadherin gene strongly suppresses the metastatic phenotype of different tumor cell lines. On the other hand, antibodies to E-cadherin make the tumor cells more aggressive in some cases. Mutations in the E-cadherin and b-catenin genes leading to their inactivation have been observed in some tumors and found to be associated with enhancement of metastatic phenotype. Down regulation of E-cadherin gene expression leads to a similar effect. Detachment from the cell aggregate leads to the aquisition by epithelial cells of several properties typical of mesenchymal cells: mesenchymal transformation of epithelial cells takes place. This phenomenon correlates neatly with the increase of invasiveness and appearance of metastatic phenotype in tumor cells (see also below).

C. Invasive growth.

Different groups of genes/proteins are involved in determination of invasive tumor growth. Examples include the tyrosine protein kinase-type receptors, the autocrine motility factor (AMF) and its receptor and different types of degradative enzymes.

Among the genes encoding tyrosine kinase receptors, c-met, c-neu, c-ret and c-ros were shown to be associated with the invasiveness of tumor cells. In more detail, the pair consisting of the ligand, scattering factor (SF)/hepatocyte growth factor (HGF), and of the receptor, c-Met, was studied. Their interaction leads to the induction of liver morphogenesis and at the same time to the increase of motility and invasiveness of tumor cells into the collagen matrix. The synthesis of both c-Met and SF/HGF is increased in different malignant tumors, in particular, in many cases of metastatic human breast cancer. In contrast to effector proteins, the system SF/HGF-cMet is part of the control proteins determining signal transduction resulting in the change of expression of several different genes.

AMF is a 64 kD protein that interacts with the receptor, 78 kD glycoprotein, gp78, and activates cell motility. The synthesis of the components of this ligand-receptor system is activated in parallel with progression of some tumors, for example, bladder carcinoma.

A wealth of data have been obtained on the role of different degradative enzymes and factors in tumor progression. Examples of degradative enzymes include different metalloproteinases and serine proteases, such as cathepsins, collagenases, stromelysins; plasminogen activator and its inhibitor, as well as gualuronidase and several other enzymes destroying extracellular matrix. The genes encoding these degradative enzymes are frequently activated in metastatic and, in general, in invasive cancer cells, although the correlation is not absolute. This group of enzymes may obviously play a role in degradation of extracellular matrix synergizing to the overgrowth and invasion of cancer cells, and in particular their invasion into blood vessels.

D. Attachment to endothelial cells (arrest in capillary bed) and extravasation.

The process of attachment of tumor cells to endothelial cells is induced by cytokines produced in tumor or inflammatory cells. These are IL-1, TNF, lipopolysaccharides (LPS) etc. Weak attachment is mediated by selectins. Synthesis of E-selectin is induced in endothelial cells. Its extracellular part interacts with tumor cell or leukocyte carbohydrates through an N terminal lectin-like domain. Strong attachment is mediated by interaction of integrins of tumor cells with the members of immunoglobulin superfamily located on the surface of endothelial cells. For example, a4b1 integrin usually present in melanomas and sarcomas binds to VCAM-1 (vascular cell attachment molecule), while a6b1 (present in colon carcinomas) and a6b4 (present in lung carcinomas) bind to ICAM-1 (intercellular cell attachment molecule). These interactions are responsible for a firm attachment.

Several reports have appeared indicating the special role in tumor metastasis of the CD-44 transmembrane hyaluronate receptor. Both metastatic and non-metastatic tumors contained the major variant of this protein, but only metastatic cells contained some minor variants of the protein characterized by the presence of additional domains, that were found to be responsible for intercellular interactions. The appearance of such variants was a result of alternative splicing that led to inclusion into mRNA of additional small exon(s). Stable transfection of non-metastatic cells with the construct expressing the CD-44 variant in some cases led to the enhancement of metastatic potential, although some opposite results were also obtained. Since both the attachment and detachment of cells play a role in metastasis just at different stages of the process, these controversial results may not be too surprising. Further experiments are needed before final conclusions can be reached.

Many other genes/proteins important for tumor progression have been described; the examples mentioned here are those of extensively studied. Some additional genes will be mentioned in further discussion.

Table 1. The genes with changed expression detected in the VMR-0 - VMR-100 and CSML-0 - CSML-100 pairs of tumor cells and/or tumors

Cell-line system

/ The gene or protein

Metastatic tumors comparing to non-metastatic

VMR-100 / VMR-0

tag7*

c-met

novel serine threonine protein kinase*

LAR (tyrosine protein phosphatase)

MHC class I H2-L-a antigen

Na,K-ATPase, catalytic sub-unit

Prothymosin

Cathepsin D

MMTV LTR

selectively expressed

over-expressed

selectively expressed

over-expressed

under-expressed

over-expressed

CSML-100 / CSML-0

mts1*

ly6

new Semaphorin*

selectively expressed

*Novel, previously undetected genes.

IV. Search for new genes involved in the control of tumor metastasis and the systems used.

Each of above mentioned genes plays a certain role in the acquirement by tumor cells of an invasive and metastatic phenotype. Probably, in different cases, different genes may be involved; one could expect that a number of genes controlling metastasis are still unknown. In particular, this may include upstream genes that are not directly involved in cell functioning but control other genes or the activity of proteins. An important direction for further studies is to discover such genes and to understand the role of their protein products in tumor progression and metastasis. In this respect, the most interesting are genes that either play a key role in producing a metastatic phenotype in various tumors or which can potentially be exploited for metastasis diagnostics and/or treatment.

Different approaches can be used in the search of new genes involved in the control of tumor progression. One approach is to search for genes, whose protein products are directly connected with tumorigenesis and metastasis. In this case, one can usually expect to find the “effector genes” encoding proteins directly participating in corresponding cell functions.

Another approach is a less direct, but still may give interesting results. It is the search for genes over-expressed or under-expressed in metastatic tumor cells compared to non-metastatic tumor cells of the same origin. In an ideal case, one can discover a gene belonging to a class of upstream genes with wide functions in the generation and maintenance of the metastatic phenotype. In many other cases, some “occasional” genes may be fished out, that are differentially expressed only in few types of tumor cells. Yet, in many cases, these new genes may become valuable for understanding several aspects of tumor progression and, more important, in development of the methods for gene therapy. For example, one such gene identified from our experiments, tag7, seems to belong to the group of genes encoding proteins protecting the organism from tumor cells, and its transfer to cancer patient could constitute an approach for gene therapy. Another gene discovered in our laboratory, mts1, may play an important role in metastasis as an upstream regulator gene.

Two mouse systems elaborated in the Russian Oncology Center (E. Revasova and V. Senin) have been used in our experiments. Both use the cell lines obtained from spontaneous mammary adenocarcinomas. The first is represented by a pair of VMR-0 non-metastatic cells and VMR-100 metastatic cells. Originally, VMR-0 cells were obtained and maintained as a cell line by subcutaneous transplantation. Occasionally, metastatic foci appeared, and the cells from these foci were taken for preparing cell culture. As a result of such selection, highly metastatic cell lines were obtained with preferential metastasis to the liver (VMR-100-Liv cells) or to the ovaries (VMR-100-Ov cells). Another pair are the CSML-0 and CSML-100 cells, non-metastatic and highly metastatic to the lungs, respectively, obtained in about the same way, also from spontaneous mammary adenocarcinoma. These two pairs of cell lines were further used for screening of the genes differentially expressed in metastatic cells. Two technologies were used: subtraction of cDNA libraries at the earlier stage and the mRNA display method at the later stage of the screening methodology. The mRNA display method, although giving a lot of false clones, is still much easier to apply and, ultimately, more genes of interest may be obtained with this technique.

Several different genes were found to be over-expressed or under-expressed in metastatic cells in the two pairs

described above (Table 1). One of them, the mts1 gene was found to be over-expressed in many different metastatic tumors and also shown to be necessary and sufficient for maintaining the metastatic phenotype at least in some tumor cells in experiments on stably transfected cells and on transgenic animals. Several other new genes may arise; however, additional experiments on stable transfection of cells with the corresponding gene constructs followed by phenotypic analysis, are required.

Some of the discovered genes correspond to those described before, and are mostly effector genes (see Table 1). Among the upstream genes, the c-met gene seems to be an important regulator of tumor cell invasiveness and, as a result, of metastatic behaviour. It is the receptor for the SF/HGF (see above). Interestingly, the difference in expression of c-met between VMR-100 and VMR-0 cells appeared only in vivo. The effect can be mimicked in cell culture after contact with stromal cells. The c-met gene expression is used as a test for the level of malignancy.

Finally, some of the discovered genes were over-expressed only in few types of metastatic cells, and therefore, they could not play a general role in determining the metastatic phenotype. This was confirmed in transfection experiments. Among them are the gene encoding a novel serine-threonine protein kinase and the tag7 gene, both over-expressed in VMR-100 cells. Although, transfection with the tag7 construct did not induce a metastatic phenotype, it strongly influenced the properties of tumor cells. Therefore, this gene and its protein product were studied in more detail.

V. A novel cytokine, Tag7, and its properties

A. Structure of the tag7 gene.

The tag7 gene was obtained by the mRNA display method, as a gene over-expressed in VMR-100-Liv and VMR-100-Ov tumors growing in mice after subcutaneous transplantation, as compared to tumors induced by VMR-0 cells (Kustikova et al. 1996). The full-length cDNA clone as well as a genomic clone were obtained and sequenced. The putative Tag7 protein consists of 182 aminoacids with no homology in the data-base (Fig. 2). Three exons were determined. The upstream region of the tag7 gene contains several binding sites for well known transcription factors, Ets1, NFkB, Sp-1 and MyoD. It was noticed that the arrangement of these sites was similar to that in the gene encoding Lymphotoxin-b. A careful manual comparison of the Tag7 and TNF-Lymphotoxin amino acid sequences revealed a very low level of homology in different regions (Fig. 2), suggesting Tag7 to be a far distant relative of the members of TNF-Lymphotoxin family of cytokines.

The tag7 gene is located on the seventh mouse chromosome in the A3 region, i.e. in a different chromosome than the TNF gene cluster. The 7A3 chromosomal region is interesting, as it has a genetic relation to the development of such autoimmune diseases, as lupus erythematosus.

B. Expression of the tag7 gene in normal tissues and tumors.

High expression of the tag7 gene is not a general characteristic feature of metastatic tumors (Fig. 3). For example, CSML-100 cells do not express the tag7 gene, whereas CSML-0 cells do. Most metastatic and non-metastatic tumor cells tested do not express the gene at all. Thus, it is clearly not a marker for the metastatic pheno-

Fig. 3. Northern blot hybridization of RNAs from cell culture and tumors with the tag7 probe type.

Exceptionally, the human multiple myeloma cells all express the tag7 gene at a rather high level. Strong tag7 expression in VMR-100 cells takes place only in tumors in vivo. In cells in culture, the level of tag7 mRNA is very low, suggesting that active tag7 expression depends on interactions between tumor and host (possibly stromal) cells.

Expression of the tag7 gene takes place in several tissues of normal organism (Fig. 4). The highest signal after Northern blot hybridization was obtained in the spleen and lungs. It is actively expressed in isolated lymphoid cells, circulating monocytes, thymocytes, splenocytes and resident peritoneal macrophages. The level of tag7 mRNA synthesis and Tag7 protein accumulation in cultured splenocytes is moderately enhanced by LPS induction (but not by IL-2 or PHA). The LPS effect on TNF and Lymphotoxin-a expression is much stronger and faster. Interestingly, in situ hybridization reveals the active tag7 expression in some specific cell sets in different organs, for example in Purkinje cells of cerebellum and in some neurons of hippocampus (Fig. 4). Its expression was found in the duodenal cells of 7-8 day embryos.

Fig. 5. Tag7 protein exists in soluble and cell-associated forms. Western blot analysis with affinity-purified antibodies to Tag7 protein. 1, Freshly isolated splenocytes; 2, 3, Splenocytes in culture after 0.5 h. LPS induction; 4, 5, The same after 24h induction; 6, 7, VMR-100-Liv cells after LPS stimulation. 1, 2, 4, 6, Tag7 protein from the cells; 3, 5, 7, Tag7 protein from the culture medium. 8, Recombinant Tag7 protein.

C. Properties of the Tag7 protein.

The recombinant Tag7 protein was obtained in inclusive bodies of E. coli and used for preparing polyclonal and monoclonal antibodies. Western blot analysis showed the presence of Tag7 protein both in the cells and in the culture medium (the major part) of Tag7-producing cells (Fig. 5). Thus, Tag7 is a secreted protein. Tag7 protein possesses a rather strong cytotoxic activity in respect to several cell lines, in particular, mouse L929 and human Jurkatt and MCF-7 cells (Fig. 6). Affinity-purified polyclonal or monoclonal antibodies destroy the cytolytic activity of Tag7 protein, while antibodies to TNF and Lymphotoxin-b do not. Vice versa, antibodies to Tag7 are not efficient in preventing the cytotoxic effect of TNF. The cytotoxicity of Tag7 is higher than that of the best commercial TNF preparation at the same concentration. The cytotoxicity of Tag7 protein is mediated through apoptosis as deduced from cytological analysis and the appearance of oligonucleosomal DNA repeats in the nuclei of target cells (Fig. 6).

High level of cytotoxicity of Tag7 protein creates a number of experimental problems. Synthesis of recombinant Tag7 in the periplasm of bacterial cells kills them. In transfection experiments, the cells producing high amount of Tag7 rapidly die. Therefore, only a limited amount of Tag7 protein can be obtained from the conditioned medium from the cultured cells producing small amount of Tag7. The problem of obtaining of native Tag7 from inclusive bodies has not yet been solved.

VI. Possible exploitation of the tag7 gene for antitumor gene therapy

A. Influence of tag7 expression on the growth of VMR-0 tumors.

The possibility to use the tag7 gene for cancer gene therapy was revealed from transfection experiments aimed to analyse the role of Tag7 in tumor metastasis. VMR-0

Fig. 6. Cytotoxicity of Tag7 protein. The L929 cells were incubated with Tag7 from VMR-100-Liv supernatant or with commercial TNF. In some experiments the indicated antibodies were added.

A. Cell death assayed by trypan blue staining; B. Appearance of nucleosome repeats after incubation of cells with TNF and Tag7.

Fig. 7. Influence of tag7 expression on the tumor cell growth.

A. Northern blot hybridization of RNAs from different VMR/tag7 cell lines (SX4 and SX12) with tag7 probe. The level of tag7 expression in transfected cells is much lower than in VMR-100-Liv.

B. 10⁶VMR-0 cells (), mock-transfected VMR-0/Neo (s) and tag7 transfected SX4 (l) and SX12 (D) cells were subcutaneously injected to 10, 5, 10 and 10 A/Sn mice, respectively. The mean values of tumor size were determined at different periods after injection. SX4 cells expressed tag7 at higher level, than SX12 cells. (m). Inhibition of SX4 effect by purified polyclonal antibodies to Tag7 (3 mice). (n), 10⁶VMR-0 cells were coinjected with 10⁶ SX4 cells (3 mice). *Animals of these groups died 4-5 weeks after injection.

cells were stably transfected with a construct containing the tag7 gene under control of the CMV promoter. Two stable cell lines, VMR-0/tag7, were obtained, both expressing tag7 at a low level. Yet, the level of expression in one cell line was two- three-fold higher than in another. It should be pointed out that the growth rate of VMR-0/tag7 cells in culture was the same as of control VMR-0 cells. However, the attempts to obtain the higher level of tag7 expression in VMR-0 cells led to the inhibition of growth of the cells in culture followed by their death at later stages.

The VMR-0/tag7 cells (10⁶) were subcutaneously transplanted to isogeneic mice. The untreated VMR-0 cells or cells transfected with neomycin gene alone were used as a control. The original VMR-0 tumors grow fast at the site of injection and kill mice in one month after subcutaneous transplantation. The tumors contain large necrotic foci at this stage (Fig. 7).

The VMR-0/tag7 cells have a dramatically changed growth properties. They grow much slower. Even after 4 months, no mice were killed by the tumor. No necrotic foci were formed, even at later stages, when the tumors reached a large size. Histological analysis of VMR-0/tag7 tumors recovered strong inhibition of mitotic rate and activation of apoptosis frequency comparing to VMR-0 tumors (Table 2). Growth inhibition was much stronger in the case of VMR-0/tag7 cell line producing higher amount of Tag7. The injections of antibodies to Tag7 accelerated the tumor growth at the period of their application (Fig. 7).

The tumors induced by transplantation of the mixture of VMR-0 (10⁶) and VMR-0/tag7 (10⁶) cells also grow much slower, than VMR-0 cells alone (Fig. 7), suggesting the activation of immune system against tumor cells (tumor vaccination effect), which may be realized through formation of CTL cells. This interpretation is supported by observation, that the growth of tumor cells of another line (CSML-100) is not inhibited by cotransplantation with VMR-0/tag7 cells.

Fig. 8. The scheme with a possible explanation of the inhibition of tumor growth by activation of cytolytic T lymphocytes.

APC-antigen-presenting cells. Filled circles, Tag7; empty circles, tumor antigens.

B. On the mechanism of tumor growth inhibition by tag7 expression.

It is usually accepted that tumor cells expressing some cytokines, in particular GM-CSF or IL-2, attract the antigen presenting cells (APC). Then, APC convert naive T-lymphocytes into cytolytic T-lymphocytes (CTL) recognizing the antigens present in tumor cells and attacking both primary foci and metastases. One can suggest, that VMR-0/tag7 cells possess the same properties (Fig. 8).

To check this, the experiments with nude mice were performed as they lack or have a very weak system for T lymphocyte response. The results were very similar. VMR-0/tag7 cells grow much slower, than VMR-0 cells. VMR-0 cells killed mice in 2 months, i.e. later than in the case of isogenic mice. However, the general properties of growth and tumor morphology remain unchanged (Fig. 9). The VMR-0/tag7 cells grow in nude mice even slower. The tumors increased in size at the beginning but later on they frequently decreased in size. Some waves of growth followed by degeneration could be observed. Finally either

Fig. 9. Nude mice injected with 10⁶ VMR-0 (A) or SX4 (B) cells after 2 months of tumor development.

Fig. 10. EM pictures of mitotic (M) and normal (N) cells in VMR-0 tumor and apoptotic cells (A) in SX4 tumor transplanted to nude mice.

Fig. 11. Nucleotide sequence of the mts1 gene coding sequence and amino acid sequence of Mts1 protein.

In the amino acid sequence, the calcium-binding domains are underlined.

small tumors remained or complete dissolving of the tumor took place (Fig. 9). Histological analysis again shows strong inhibition of mitotic rate and at least tenfold increase in frequency of apoptosis. Typical pictures of apoptosis are observed under electron microscopy (EM) (Fig. 10). It seems that the effect of tag7 expression is rather complex and at least includes activation of an immune response mediated by CTL cells and direct cytotoxicity of the Tag7 protein. Experiments are in progress for more precise understanding of Tag7 action and for its application to cancer treatment.

VII. The mts1 gene

A. General properties of the mts1 gene.

Another extensively studied gene is the mts1 gene. It may play an important role in the control of tumorprogression and appearance of metastases at least in some tumors. The mts1 gene has been discovered in experiments on cDNA libraries subtraction using CSML-100 and CSML-0 cell lines (see above). The mts1 gene is transcribed to a 0.55 kb mRNA, which was abundant in CSML-100 cells and absent from CSML-0 cells. This mRNA was detected in many metastatic tumor cell lines of different origin, but not in non-metastatic tumors, although several exceptions could be observed. Therefore, the gene was called as mts1, a gene encoding Metastasin 1 protein, Mts1 (Ebralidze et al. 1989).

The cDNA was sequenced and the protein structure was deduced. Mts1 was found to be a protein 101 amino acid long with two typical calcium-binding domains (Fig. 11). It belongs to the S-100 sub-family of calcium-binding proteins. The mts1 gene was described at about the same time in several other groups under different names, but without any relation to tumor metastasis. The mts1 has also been cloned from the human genome. Human Mts1 protein differs from its mouse counterpart just by 7 amino acid substitutions.

The mts1 gene is expressed in several normal tissues: embryonic fibroblasts, trophoblasts, and lymphoid cells, in particular, in T-lymphocytes and activated macrophages. At least some of these cells possess invasive properties. The level of mts1 expression can be readily modulated by different lymphokines or calcium ionophores (Grigorian et al. 1993).

No sequence rearrangement usually takes place during the change of tumor phenotype to a metastatic one, as deduced from Southern blot hybridization experiments. The only exception was observed in the VEHI-3 cell line (myelomonocytic leukaemia), where a deleted copy of IAP retrovirus-like mobile element was found to be inserted into the first intron of the mts1 gene. As a result, the transcription was started within the IAP LTR, and chimeric mRNA was synthesized, while the protein remained unchanged (Tarabykina et al. 1996). Therefore, activation of mts1 mRNA synthesis should usually result from changes in concentration of certain trans-regulatory factors. Finding of such factors responsible for mts1 activation may lead to discovery of new genes involved in the creation of metastatic phenotype, acting upstream in respect to the mts1 gene.

The mts1 gene consists of three exons and two introns. The first exon is small and does not contain translated sequences. The gene is located in the gene cluster containing several other members encoding proteins belonging to S-100 family. The distances between genes in the cluster are rather small. Examination of the mts1 upstream region up to the 3’-end of the neighbor gene of the cluster has led to the conclusion that cis-regulatory elements are not present but instead a TATA-box containing promoter. All cis-regulatory elements have been found within the first intron. Several well known positive and negative cis-regulatory element binding was determined as well as some new transcription factors have been found (Fig. 12) (Grigorian et al. 1993, Tulchinsky et al. 1996, 1997).

The first element (from the 5’-end of the intron) is the enhancer of a moderate strength possessing no homology

Fig. 12. Regulatory region of the mts1 region located in the first intron.

with known enhancers. It binds a novel transcription factor, which is present in both CSML-100 and CSML-0 cells. However, in vivo the protein binding was detected only in CSML-100 cells. Thus, the enhancer selectively works in metastatic cells in spite of the presence of activating protein in both. One can suggest the involvement of structural changes in chromatin. Actually, the test for DNA methylation showed the absence of mts1 methylation in CSML-100, while in CSML-0 cells the mts1 gene was heavily methylated. The latter may interfere with binding of activating proteins to the enhancer. The effect should be indirect as the enhancer core sequence does not contain CpG dinucleotides.

The second cis-regulatory element is represented by the sequence TGACTCG differing from the consensus AP-1 binding sequence (TGACTCA) by one base substitution. As a result of substitution, a CpG dinucleotide appears that is the subject for deoxycytidine methylation. Both methylated and non-methylated sequences interact with nuclear proteins, as followed from band shift experiments. The protein binding to methylated sequence is different from that binding to non-methylated one and is much more abundant in nuclear extracts. The consensus AP-1 binding sequence competes only with the methylated element. This suggests that only the methylated sequence binds AP-1 factor. The conclusion was proved in supershift experiments with antibodies to Jun and Fos proteins. Thus, methylation of CpG creates a novel site for AP-1 binding. In the mts1 gene, methylated AP-1 binding element plays the role of a transcription silencer of a moderate strength. This inhibition seems to play a role in vivo, as a particular CpG sequence in CSML-0 cells is not methylated, while in CSML-0 cells, it is completely methylated (Tulchinsky et al. 1996).

The third cis-regulatory element is located further downstream and is represented by the GGGGTTTTTCCAC sequence, related to NFkB-binding sites. The sequence does actually bind NFkB (p50/p50 and p50/p65), but this binding does not change mts1 transcription, at least in experiments with transient expression. On the other hand, the same sequence binds another factor, p200, of a higher molecular weight. As was shown in experiments with different constructs, the latter was responsible for activation of transcription in transient transfection assays. The p200 concentration in CSML-100 extract is about tenfold higher, than that in CSML-0 cells. In vivo footprinting showed the occupancy of the DNA sequence only in CSML-100 cells. These data suggest the role of the factor in the in vivo activation of mts1 transcription (Tulchinski et al., 1997). Closely to the previous element, a fourth element is located, which contains a microsatellite motif and is protected from nucleases in both in vivo and in vitro footprinting assays. It binds a protein interacting with microsatellite. It seems to play in vivo a role of a moderate transcription activator, as follows from stable transfection experiment (Prokhourtchuk et al., in preparation).

The fifth positive cis-regulatory element binds a novel protein which is present only in CSML-100 cells. Therefore, it may represent a metastasis-specific transcription factor. Its cloning is in progress. Finally, the sixth regulatory element is represented by the enhancer binding AP-1 protein. In CSML-100 it is a FRA-JunD

Table 3. Results of transfection experiments with the mts1 gene constructions

Cells	Phenotype	mts1 construction	Mts1 in transfected cells	Change of metastatic phenotype
CSML-100	M	Antisense	Decrease	Strong decrease
OHS (human)*	M	Ribozyme	Decrease	Strong decrease
CSML-0	NM	Sense	Absence	No change
Line 1 + DMSO	NM	Sense	Appearance	Strong increase
Rama 37 (rat)**	NM	Sense	Appearance	Strong increase
MCF-7 (human)	NM	Sense	Appearance	Increase
Transgenic mice; spontaneous adenocarcinoma	NM	Sense	Appearance	Strong increase

complex. CSML-0 cells are FRA-deficient, and this may play an important role in the transcription control.

Thus, the first intron of the mts1 gene contains a complex regulatory system, which is sensitive to methylation. Some factors present in this system may play an important role in the control of metastatic behaviour of tumors.

B. The role of mts1 expression in tumor metastasis.

The central question is whether the over-expression of the mts1 gene in tumor cell can or can not change the phenotype from non-metastatic to metastatic and vice versa, i.e. whether the presence of Mts protein is casual for metastatic behavior of the tumor cell or an occasional coincidence. Certainly, the over-expression of any cellular gene could not be expected to constitue the only factor responsible for metastasis (see above), but some genes on certain background of expression of other genes may become indispensable for that. These “key genes” can be detected in transfection experiments that allow to switch on or off the gene functioning.

Several cell systems were used in such experiments (Table 3). First, CSML-100 cells were transfected with the construct containing mts1 cDNA in antisense orientation under the control of Moloney sarcoma virus promoter/enhancer element present in its LTR. The cell lines actively expressing antisense RNA were selected and used for subcutaneous transplantation to isogeneic mice.

They had a dramatically decreased metastatic potential compared to highly metastatic CSML-100 cells. Instead of hundreds of metastatic foci expected in the lungs of mice subcutaneously injected with the original CSML-100 cells, either no foci or single metastases appeared after transplantation of the same cells but expressing mts1 antisense RNA. Thus, mts1 expression was necessary for maintaining a metastatic phenotype in CSML-100 cells (Grigorian et al. 1993).

Another technology to switch off the gene expression is to use a construct encoding ribozymes, i.e. RNAs that specifically cleave a particular RNA. The ribozyme specifically cleaving human mts1 RNA at the second exon was transfected into human osteosarcoma (OHS) cells. The control OHS cells gave metastases to bone marrow of nude rats after intracardiac injection. The stable transfectants strongly suppressed the metastatic phenotype. The Mts1 protein content in such cells was decreased (Maelandsmo et al. 1996).

The reverse experiment with CSML-0 cells stably transfected with a construct expressing sense mts1 mRNA gave negative results. However, in spite of active mts1 transcription, these cells did not contain Mts1 protein. Thus, in addition to the control of mts1 expression at the transcription level, the control at translational level is also important and some cells can not translate mts1 mRNA. Therefore, these experiments are non interpretable until the translation suppression is overcome.

Yet sense constructs were successfully tested in three other cell lines. One of them is line 1 cells, that are mouse small cell lung carcinoma cells highly metastatic to lungs upon intravenous injection. However, after dimethylsulfoxide (DMSO) treatment, they lose the ability to metastasize. DMSO treatment was also shown to strongly inhibit mts1 expression. Sense constructs were transfected to these cells and the transfectants, actively expressing exogenous mts1 gene, acquired the ability to give metastases even after DMSO treatment. The latter did not interfere with mts1 expression governed by MSV-LTR control elements (Grigorian et al. 1993).

A strong increase in metastatic potential was found in rat mammary epithelial Rama37 cells after stable transfection with the mts1-expressing constructions. The original cells are benign and do not metastasize, while transfectants gave metastases to lungs and lymph nodes (Davies et al. 1993).

Finally, experiments on the well-characterized human mammary adenocarcinoma MCF-7 cells were performed. MCF-7 cells are rather benign. They can grow after transplantation to nude mice only when supported with estrogen and only when transplantation into the mammary fat pad is performed. The growth is non-invasive and no metastases can be observed. MCF-7 cells do not contain any significant amount of Mts1 protein. Only a low level of mts1 expression could be observed in stromal cells. The expression of the exogenous mts1 gene induced by stable transfection with the construct containing the mts1 gene strongly changed the properties of the MCF-7 cell growth. First, their growth in nude mice became hormone-independent. Second, they could grow after just subcutaneous transplantation. Third, an invasive growth at the primary focus could be detected. Fourth, metastases to regional lymph nodes and small-size metastases to the lungs were observed. The tumor cells contained varying amounts of Mts1 protein (Grigorian et al. 1996). Thus, in all mentioned cases, the appearance or disappearance of Mts1 protein led to a significant modulation of metastatic phenotype in the expected direction.

Yet, a weak point in transfection experiments is the heterogeneity of the cell population used for transfection. For example, CSML-0 cells consist of three morphologically distinct cell types that are reproduced after cloning from individual cells. It can not easily be excluded that only cells with pre-existing differences in metastatic potential have been selected during transfection experiments. Some other approaches should also be used.

The most clear evidence for the casual role of the mts1 gene expression for creation of metastatic phenotype was

Fig. 13. The scheme of the experiment with transgenic animals.

(m), mammary adenocarcinomas; (), metastases; (s) presence of Mts1 protein.

obtained in experiments with transgenic animals (Ambartsumian et al. 1996) (Fig. 13). Transgenic mice were obtained with the construct containing the mts1 gene under control of the MMTV-LTR promoter/enhancer

Fig. 14. Staining of the non-transgenic (A) and transgenic tumor (B) with antibodies to Mts1 and the metastasis of transgenic tumor to the lungs (C).

Table 4. Appearance of metastases in spontaneous and transplanted tumors depending on the presence of actively expressed mts1 transgene

Type of the tumor	Mice line and generation	Number of tumors	Number of tumors with metastases
Spontaneous	Tg463 (F1-F4)	23	9 (39%)
mammary	Tg507 (F1)	5	2 (40%)
adenocarcinoma	Transgenic total	28	11 (40%)
	Non-transgenic	21	1 (5%)
Transplantation	Transgenic metast.	5	3
to nude mice	Transgenic non-met.	2	2
	Transgenic total	7	5 (70%)
	Non-transgenic	2	0 (0%)

element. Transgenic mice expressed exogenous mts1 in several tissues. The highest level of expression was found in lactating mammary glands, where the MMTV promoter is very active. The endogenous mts1 gene is not expressed in lactating mammary gland. Interestingly, the phenotype of transgenic mice was not changed compared to normal mice. Even the presence of a high amount of Mts1 protein in lactating mammary glands did not interfere with their functioning and no mammary gland tumors could be observed. Obviously, the mts1 gene is not an oncogene.

Thereafter, the transgenic mice were crossed with mice from the GRS/A strain characterized by a high incidence of mammary gland tumors appearing after several cycles of pregnancy and lactation. These tumors are non-metastatic. As transgenic mice were heterozygous, only half of the offspring carried the transgene, while another half represented the control group. The tumors appeared with the same high frequency in both groups and they were morphologically indistinguishable. The tumor growth rate also did not depend on the presence of the transgene.

However, a dramatic difference in the metastatic potential of tumors from the two groups was found. As was mentioned above, non-transgenic tumors never metastasize. Just in only one case (out of 21 tested), non-transgenic tumor gave metastases to lungs, but this probably depended on certain additional genetic changes. On the other hand, 40% of transgenic tumors were metastatic (Table 4). Then, the tumors were subcutaneously transplanted to athymic mice to determine their metastatic phenotype. Both transgenic tumors that had metastasized before and transgenic non-metastasizing tumors gave rise to lung metastases. Thus, 40-50% incidence of metastasis is an intrinsic feature of spontaneous mammary carcinomas expressing the mts1 gene. Non-transgenic tumors never metastasized after transplantation, with the above mentioned exception.

The distribution of Mts1 protein was detected with the aid of immuno-staining (Fig. 14). Mts1 was found in transgenic tumor cells at the primary focus as well as in the metastatic foci. The concentration varied in a wide range among different cells even in the same tumor. Non-transgenic tumor cells in neither case contained Mts1 protein. The stromal cells in the both types of tumors contained the same small amounts of Mts1 expressed from endogenous gene.

All mentioned experiments clearly demonstrate that in some tumors mts1 expression is necessary and sufficient for the acquisition of the metastatic phenotype. It is quite clear that mts1 can not be responsible for all metastatic phenotypes, as several metastatic tumors, in particular one which appeared among the non-transgenic tumors, do not express mts1. However, the described results show that the mts1 gene is one of the key metastatic genes. The question arises, what is a possible mechanism of the action of Mts1 protein.

C. A possible biological role of the Mts1 protein.

To answer the question, intensive studies of the protein had to be performed. For this, one needs high amounts of protein. It was obtained in a bacterial system with oligohistidine tail that allowed an easy purification of the protein on nickel columns. The purified protein was used for preparing polyclonal and monoclonal antibodies.

Western blot analysis and immunostaining of cells with these antibodies showed the Mts1 protein to be localized in the cytoplasmic fraction, like other calcium-binding proteins. Experiments on fractionation of cell extracts suggested the presence of a significant fraction of Mts1 protein in the cytoskeleton.

To understand Mts1 function, attempts to determine the targets of Mts1 protein were performed. The in vivo labeled proteins which bind Mts-1 were immunoprecipitated with antibodies to Mts1 and the proteins specifically precipitated by these antibodies were

analyzed. This approach resulted in the isolation of different components of myosin complex. On the other hand, antibodies to myosin co-precipitated Mts1. Ultracentrifugation of cell extracts on sucrose gradients also demonstrated cosedimentation of a rather significant fraction of Mts1 with a much heavier myosin complex suggesting their reversible association. After double immuno-staining with antibodies to Mts1 and myosin, one can see the exact coincidence in fluorescence distribution for both antibodies. Mts1 interacts only with the heavy chain of non-muscle myosin as followed from different overlay experiments and using antibodies specific for different types and different chains of myosin. Mts1 does not interact with the light chain and with heavy chain of smooth muscle myosin (Kriajevska et al. 1994).

To further analyze Mts1-myosin interaction, different fragments of the heavy chain of non-muscle myosin (HCNMM) were prepared using recombinant DNA and tested for interaction with Mts1 in protein overlay experiments. Only the carboxy-terminal part of HCNMM did react. Analysis of the deletions obtained in this part of HCNMM showed that the only peptide responsible for interaction with Mts1 was located between the 1908 and 1938 aminoacid residues (Fig. 15). Then the effect of such binding on protein kinase mediated phoshorylation of heavy chain of non-muscle myosin was tested. Mts1 specifically inhibited phoshorylation of serine residue no. 1917 by protein kinase C without any effect on other phosphorylation sites and without interference with casein protein kinase action. The target serine residue is located just inside the binding region for Mts1 (Fig. 15). One can suggest that at least one of Mts1-induced effects is an inhibition of this phosphorylation reaction. The latter was claimed to play a role in non-muscle myosin functioning putatively leading to changes in cell motility. This may be a possible way for changing the metastatic phenotype of tumor cells. However, for the time being, this is just a hypothesis.

Another consequence of interaction between Mts1 and carboxy-terminal part of non-muscle myosin is the solubilization of the non-muscle myosin. The simple addition of a native Mts1 to the myosin polypeptide precipitate at physiological salt concentration completely solubilized the precipitate, confirming the role of Mts1 in myosin disaggregation.

It should be pointed out that myosin is not the only target for Mts1. In particular, the method of binding to affinity column revealed another protein, p37, interacting with the Mts1 column. The p37 protein binds to Mts1 in a calcium-dependent manner. Interestingly, this binding strongly changes the interaction of Mts1 with calcium. Two calcium-binding domains in Mts1 act cooperatively and in general the affinity to calcium is increased. Possibly, p37 may play some role in the control of Mts1 functions connected with binding of calcium ions (Dukhanina et al. in press).

Another important function of Mts1 protein may be excerted via mesenchymal transformation of epithelial cells. Strutz et al (1995) found, that expression of the mts1 gene in epithelial cells might induce their mesenchymal transformation. We have found that appearance of Mts1 proteins in the cells of transgenic mice was accompanied by the loss of E-cadherin. The reverse correlation between Mts1 and E-cadherin content may be especially important as E-cadherin is one of most clear-cut “antimetastatic proteins” with well understood function (Vlamincks et al. 1991). The mechanism of Mts1 influence on E-cadherin remains unclear.

Finally, Onishchenko et al. (1997) recently observed the destabilization of blood vessels in tumors expressing mts1; this means that endothelial cells of blood vessels may constitute another target of Mts1 and this additional phenotypic effect of Mts1 may be effected through its influence on tumor vascularization and on the state of endothelial cells.

Thus, Mts1 may be a multifunctional regulator of cell functions connected with the acquisition of an invasive metastatic phenotype. The mts1 gene may be one of the critical genes for metastasis development and, therefore, a promising target for cancer gene therapy.

VIII. Conclusions

In general, this study summarizes our results on the search for new genes and proteins controlling tumor progression. Here we described explicitly two examples of such genes. One gene, (mts1), tentatively assigned to the second class (tumor progression genes), and another gene, (tag7), putatively belonging to the third class (genes for biological defense against tumors) were discovered in this research program together with many other, yet less characterized genes. The tag7 gene seems to be a promising factor to be used for gene therapy or even for the direct treatment of tumors by its product, Tag7 protein. We propose that the mts1 gene may also find a place in the constellation of target genes for cancer gene therapy.

Acknowledgements.

This work was supported by Moscow Program for Cancer Treatment, Russian Fund for Basic Researches, INTAS and PECO grants

References.

1. Review articles:

Anderson MW, Reynolds SH, You M, Maronpot RM (1992) Role of proto-oncogene activation in carcinogenesis. Environ Health Perspect 98,13-24

Folkman J (1992) Inhibition of angiogenesis. Semin. Cancer Biol. 3, 65-71.

Barbacid M (1987) ras genes. Ann. Rev. Biochem. 56, 779-827.

Georgiev GP, Kiselev SL, Lukanidin EM Tumor progression and metastasis. (1997) In “Genome structure and function” (Nicolini C, ed.) Kluwer Acad. Publ., Netherlands, pp217-237.

Gunthert U, Birchmeier W, Eds. (1966) Attempts to understand metastasis foemation. II. Regulatory factors (Current topics in Microbiol. Immunol.) Springer-Verlag, Berlin- Heidelberg, v. 213/II.

Vogelstein B, Kinzler KW (1993) The multistep nature of cancer. Trends Genet. 9, 138-141.

White E (1996) Life, death, and the pursuit of apoptosis. Genes Dev. 10, 1-15.

2. Experimental papers:

Ambartsumian NS, Grigorian MS, Larsen IF, Karlstrom O, Sidenius N, Rygaard J, Georgiev G, Lukanidin E (1996) Metastasis of mammary carcinomas in GRS/A hybrid mice transgenic for the mts1 gene. Oncogene 13, 1621-1630.

Davies BR, Davies MPA, Gibbs FEM, Barrachlough R, Philip S, Rudland PS (1993) Induction of the metastatic phenotype by transfection of a benign rat mammary epithelial cell line with the gene for p9Ka, a rat calcium-binding protein, but not with the oncogene EJ-ras-1. Oncogene 8, 999-1008.

Grigorian MS, Tulchinsky EM, Zain S, Ebralidze AK, Kramerov DA, Kriajevska MV, Georgiev GP, Lukanidin EM (1993) The mts1 gene and control of tumor metastasis. Gene 135, 229-238.

Grigorian M, Ambartsumian N, Lykkesfeldt AE, Bastholm L, Elling F, Georgiev G, Lukanidin E (1996) Effect of mts1 (S100A4) expression on the progression of human breast cancer cells. Int. J. Cancer 67, 831-841

Ebralidze A, Tulchinsky E, Grigorian M, Afanasyeva A, Senin V, Revasova E, Lukanidin E (1989) Isolation and characterization of a gene specifically expressed in different metastatic cells and whose deduced gene product has a high degree of homology to a Ca2+-binding protein family. Genes Dev. 3, 1086-1093.

Kriajevska MV, Cardenas MN, Grigorian MS, Ambartsumian NS, Georgiev GP, Lukanidin EM (1994) Non-muscle myosin heavy chain as a possible target for protein encoded by metastasis-related mts-1 gene. J. Biol. Chem. 269, 19679-19682.

Kustikova OS, Kiselev SL, Borodulina OR, Senin VM, Afanasyeva AV, Kabishev AA (1996) Cloning of the tag7 gene expressed in metastatic mouse tumors. Genetika (in Russian) 32, 621-628.

Maelandsmo GM, Hovig E, Skrede M, Kashani-Sabet M, Engebraaten O, Florenes VA, Myklebost O, Grigorian M, Lukanidin E, Skanlon KJ, Fodstad O (1996) Reversal of the in vivo metastatic phenotype of human tumor cells by an anti-CAPL (mts1) ribozyme. Cancer Res. 56, 5490-5498.

Onischenko A, Chenard MP, Lefebvre O, Bruyneel E, Rio MC (1996) Defective tumor vascularization induced by metastasin 1 expression. Invasion Metastasis 16, 160-168

Strutz F, Okada H, Lo CW, Danoff T, Carone RL, Tomaszewsky JE, Neilson G (1995) Identification and characterization of a fibroblast marker: FSP1. J. Cell Biol. 130, 393-405.

Tarabykina S, Ambartsumian N, Grigorian M, Georgiev G, Lukanidin E (1996) Activation of mts1 transcription by insertion of a retrovirus-like IAP element. Gene 168, 151-155.

Tulchinsky E, Georgiev G, Lukanidin E (1996) EM Novel AP-1 binding site created by DNA-methylation. Oncogene 12, 1737-1745.

Tulchinsky E, Prokhortchouk E, Georgiev G, Lukanidin E (1997) A kappaB-related binding site is an integral part of the mts1 gene composite enhancer element located in the first intron of the gene. J. Biol. Chem. 272, 4828-4835.

Vlemincks K, Vakaek L, Mareel M, Fiers W, VanRoy F (1991) Genetic manipulation of E-cadherin expression by epithelial tumor cells reveals an invasion suppressor role. Cell 66, 107-119.

Tumor cells	Mitotic cells	Apoptotic cells	Ratio M:A
(i) VMR-0	3-5%	<0.5%	ca. 10/1
(ii) VMR-0/tag7	<1%	5-8%	ca. 1/10
Ratio, (ii):(i)	ca. 1/5	ca. 20/1	ca. 1/100