Yuri Ovchinnikov Lecture
Human Leukemia Genes: Search for the Villains
J. D. Rowley    Haematol. Bluttransf. Vol 35

This Ovchinnikov Lecture provides an occasion to review our progress in a central area of cancer research, namely the genetic changes that occur within the cancer cell that are critically involved in the transformation of a normal to a malignant cell. To concentrate on genes to the exclusion of cell biology would be too narrow and shortsighted a perspective. Nonetheless, I am convinced that until we have isolated the genes that are centrally involved in at least some of the malignant processes in different cell types we will be unable to answer the fundamental questions about malignant transformation. More importantly, we will be unable to answer the questions with precision. I will limit my consideration to those changes that have been detected by analyzing the karyotypic pattern of human cancer cells using chromosome banding, and in particular to those found in leukemia. We are living in a golden age of the biomedical sciences. Increasingly sophisticated instruments and creative scientific strategies allow remarkably precise understanding of some aspects of cancer biology. It is clear that during the course of the last three decades, the scientific community's assessment of the role of chromosome changes in the complex process of malignant transformation has changed from considering them to be merely trivial epiphenomena to recognizing their fundamental involvement at least for some tumors. This change in attitude has occurred for at least two reasons. First, the demonstration of specific recurring chromosome rearrangements, including translocations and de letions, that were often uniquely associated with a particular type of leukemia, lymphoma, or sarcoma provided clear evidence that these rearrangements were critically involved in malignant transformation [1- 3]. About 70 recurring translocations as well as many nonrandom deletions and other structural abnormalities are listed in the chapter on structural chromosome changes in neoplasia included in Human Gene Mapping 10 [4]. The evidence for the presence of recurring chromosome abnormalities in a wide variety of human neoplasms was the result of 30 years of painstaking chromosome analysis by my cytogenetic colleagues around the world. Second, and I believe an even more powerful force acting within the general scientific community to reassess the role of karyotypic alterations, was the identification of the genes involved in some of the chromosome rearrangements and the discovery that some of these genes were the human counterparts of the viral oncogenes [5]. In a sense, each group of investigators gave the other scientific validity. The fact that oncogenes were directly involved in chromosome translocations demonstrated that both translocations and oncogenes were critically involved in human cancer. The genetic changes that occur in different types of malignant cells are quite varied, and clearly several different changes occur in the same cell as it is altered from a normal to a fully malignant cell. Cytogenetic analysis has been the key to defining at least two major categories of rearrangements, namely recurring translocations and consistent de letions. One of the first translocations, identified in 1972, was the 9; 22 translocation in chronic myeloid leukemia [6]. There are now at least 70 recurring translocations that have been detected in human malignant cells. The identification of consistent chromosome deletions has been equally important because it has provided the absolutely essential information regarding the chromosome locations of the genes that are involved in cancer. I submit that the retinoblastoma gene would not have been cloned, or at least not yet, if cytogeneticists had not identified deletions of the long arm of chromosome 13, and specifically of band 13 q 14, in patients with constitutional chromosome abnormalities who had a high incidence of retinoblastoma [7]. This is not to detract from the careful and exciting work of many scientists is actually cloning the gene, but at least they knew where to look [8]. This triumph has now been joined by the recent cloning of the DCC (deleted in colorectal carcinomas) gene on chromosome 18; the fact that a gene important in the transformation of colorectal cells was located on chromosome 18 was the result of cytogenetic analysis of colon cancer cells that revealed that loss of chromosome 18 was a recurring abnormality [9-10]. I must acknowledge that it has been a source of great disapointment to me that we have progressed so slowly in cloning most of the genes located at the breakpoints in the recurring translocations or inversions in human leukemia. This emphasizes the fact that knowing the location of the breakpoint is very helpful in selecting the genes to use as probes for these rearrangements. However, a chromosome band contains at least five million base pairs and the likelihood that the DNA probe that you "pull off the shelf' is at the breakpoint and can detect a rearrangement on Southern blot analysis is vanishingly small. The lymphoid leukemias and lymphomas are the major exceptions to this slow progress, because the immunoglobulin genes in B cell tumors and the T cell receptor genes in T cell tumors have provided the essential DNA probes to clone several dozen trans locations [1113]. Fortunately , the rapid progress being made in mapping the human genome, coupled with major advances in working effectively with large pieces of DNA, has already made important contributions to the successful mapping of some of the recurring translocations in the acute leukemias and sarcomas. The use of cosmids or yeast artificial chromosomes (Y AC) as probes to screen much larger segments of DNA for rearrangements provides anew strategy for the analysis of these chromosome abnormalities. We have used these probes in in situ hybridization with biotin labeling of the DNA and detection with a fluorescein isothiocyanate-(FITC)-tagged avidin antiavidin conjugate. The focus of our research has been the analysis of chromosome translocations involving band 11q23. This band is of great interest because it is affected in a large number of different recurring rearrangements. The translocations may occur in either acute lymphoblastic or acute myeloid leukemias, especially of the mono blastic or myelomonocytic subtype. Finally, about two-thirds of chromosome abnormalities in leukemia cells of children under 1 year of age involve 11 q 23, regardless of the morphological classification of the leukemia. We have used a series of cosmid probes as well as a yeast clone containing two Y ACs to map the 11 q 23 breakpoint in four different translocations, namely t(4;11), t(6;11), t(9;11), and t(11;19) [14]. The cosmid probes were isolated by Evans et al. [15] and they were mapped to the region 11 q 22 to 11 q 25 by Lichter et al. [16]. The yeast clone with the Y ACs was identified using polymerase chain reaction primers specific for the CD3G gene. We showed that this yeast clone contains two Y ACs of 320 and 275 kb that differ only because of a 45 kb deletion in one of the Y ACs; the deletion is in the end opposite the CD3G/CD3G complex. With the use of cosmid probes we obtained essentially the same results in all four translocations. The cosmid probes 3.16, 23.20,1.16,4.13, ZB6, and CD3D all remained on chromosome 11. The cosmid probes XH5, XB1, ZC9, PBGD, 9.4,ZA7, THY1,8.5,SRPR,XB2,ETSl 23.2, and 5.8 all were translocated to the other chromosome. Seven cosmid probes (XH5, XB1, ZC9, PBGD, 9.4, ZA 7, and THY1) were deleted in one t(9:11) patient, presumably simultaneously with the translocation. The CD3G Y ACs localized only to chromosome 11 in normal cells. However, in addition to labeling the normal chromosome 11, the Y ACs were split in cells with the four translocations; thus one portion remained on chromosome 11 and the other was translocated to the other chromosome. There was no labeling of any other chromosomes in these cells. Thus the breakpoint in these translocations, which occur in both lymphoblastic and myeloid leukemia, is within the same 320 kb region of human DNA. We have no evidence, at present, of whether the break involves the same segment in the different translocations. Thus the use of Y AC clones provides a new strategy for screening large pieces of DNA and for focusing intensive molecular analysis only on the segment that is shown cytogenetically to be of interest. Y ACs will also be of great benefit in defining the genetic boundaries of chromosome deletions. These probes also provide powerful tools for detecting these same rearrangements in interphase cells. U sing more conventional techniques, a colleague of mine, Dr. Timothy McKeithan, has cloned the translocation breakpoint found in some patients with B cell chronic lymphatic leukemia [17]. The translocation involves the immunoglobulin heavy chain locus (IGH) located at chromosome band 14q 32 and a previously unknown gene that we have called BCL3 on chromosome 19 (band 19q13). They cloned the translocation breakpoint from two of our patients as well as from several others from material provided by other laboratories. There was evidence for rearrangement adjacent to one of the IGH constant regions in each case. In four of the cases, this rearranged band has been cloned; all showed a rearrangement with sequences from chromosome 19. Three of the breakpoints on chromosome 19 were within 170 bp of each other; the fourth lay 19 kb centromeric. Overall, a region of about 35 kb surrounding these breakpoints has been cloned and mapped. A cluster of CG-containing restriction sites was found close to the cluster of breakpoints on chromosome 19. These "CpG islands" are usually associated with the 5' ends of genes. The presence of a CpG island adjacent to the cluster of t(14; 19) breakpoints was confirmed by sequencing. Probes from this region were used in Northern blot experiments, which detected a 2.1- 2.3 kb transcript in many hematopoietic cell lines. S 1 protection experiments confirmed this result and showed that transcription occurred in a direction away from the breakpoint toward the telomere. The BCL3 cDNA was cloned and sequenced [18]. A basic protein of 446 amino acids and a molecular weight of 46741 is predicted, which shows a remarkable structure. The N-terminus is highly enriched in proline (25% ), and the Cterminus in proline (23% ) and serine (28% ). Almost the entire remainder of the protein (about half) is made up of seven tandem repeats of 33-37 amino acids. Comparison with proteins in the available data bases showed significant homology to the Dro.sophila Notch protein. The homology is in the region of the repeats. Notch has six repeats with clear similarity to the repeat in BCL3. These repeats have been found in three additional proteins namely, lin-12 of Caenorhabditis elegans (six repeats), cdc10 in Schizo.saccharomyces pombe (two) and SW16 in Saccharomyces cerevisiae (two) [reviewed in 18]; the repeats are generally referred to in the literature as cdc 10 repeats. The role of this motif is not known. Total RNA from two patients with chronic lymphocytic leukemia (CLL) and the t(14; 19) one with a break on chromosome 19 close to BCL3 and one with a break more than 25 kb away -was hybridized on Northern blots to determine the level of BCL3 expression in cells containing the t(14; 19). The samples were compared with total RNA from the peripheral blood of two other patients with CLL, as well as with three cell lines derived from the prolymphocytic variant of CLL; none of the cell lines contain the t ( 4; 19). In addition, five other hematopoietic cell lines were examined. The level of message in the two CLL samples with a t(14; 19) was higher than that found in any other sample examined. By hybridization to blots containing various quantities of RNA, the two t(14; 19) samples were found to contain 5- 7 times and 10-15 times the level of message present in the CLL cell line with the greatest quantity of message [18]. The message present in the cells with the t(14; 19) was identical in size to that present in normal hematopoietic cells, as would be expected from the fact that the translocation breakpoints occur upstream of the transcription start site. The apparent normality of the message suggests that the increased message level results from increased transcription and not from an increased message stability arising from changes in the structure of the transcript itself. The known functions of the other proteins containing the cdc10 motif may offer a clue to the function of BCL3; unfortunately, however, the divergent structure and function of these proteins makes it difficult to image a common role for the motif. Notch (in Drosophila) and lin-12 (in the nematode Caenorhabditi.s) are transmembrane proteins involved in cell lineage determination. On the other hand, the two yeast proteins are not transmembrane proteins and they share functions involved in control of the cell cycle. cdc 10 is one of two genes in Schizosaccharomyces pombe known to be required for commitment to the cell cycle; this control point, in G 1, is known as "start". Much more is known about the function of the other required gene, cdc 2, encoding a protein kinase which is highly conserved among eukaryotes and is required both for start and for mitosis. Little is known about the function of the cdc 10 protein. Recently, a specific antibody to cdc10 was shown to detect a protein of similar size in mammalian cells, suggesting that, like cdc2, the protein may be conserved throughout the eukaryotes. S W 16 is one of several genes known to be required for transcription of the HO gene, which encodes the endonuclease which initiates mating type switching in Saccharomyces cerevisiae. HO is activated immediately after commitment to start, and a particular repeated sequence in the 5' flanking region of the gene has been shown to be responsible for cell cycle control of its transcription. SW 16 and SW 14 (whose sequence has not yet been reported) are the only genes known to be specific for this control element. These two genes appear to be at least partially interchangeable since neither single mutation is lethal, but double mutations are nonviable. While the function of SW 16 strongly suggests that it is a nuclear trans-activating protein, it has not been directly shown to interact with DNA or even to be a nuclear protein. If the cdc 10 motif is involved in protein-protein interactions, there may be little commonality in function between the two yeast proteins and the two invertebrate proteins. Nevertheless, there are a few plausible models in which the proteins could have related functions. For example, BCL3, cdc10, and SW16 may be peripheral membrane proteins which interact with the cytoplasmic domains of transmembrane proteins and are involved in signal transduction. According to this interpretation, the ancestor to the lin 12 and Notch genes could have resulted from the fusion of two genes in evolution -one encoding a transmembrane protein, and the other, a cdc 10-related protein. The increased levels of BCL3 message following mitogenic stimulation and the homology of the gene to cell cycle control genes suggest that abnormally large quantities of the protein present in CLLs with the t(14; 19) may lead to an increased proliferative rate in these cells. This superficially seems inconsistent with the very low mitotic rate of CLL cells. Perhaps this mitotic rate, while low, is nevertheless greater than that of normal CD5+ B lymphocytes. Alternatively, a subpopulation of CLL cells, perhaps those present in pseudofollicular growth centers in lymph nodes, may show an abnormally high rate of proliferation. One of the major reasons to concentrate on cloning the genes involved in rearrangements is that the consistent chromosome changes pinpoint the location of the genes whose functions are critical in the growth potential of that cell type. The chromosome changes that we concentrate on are present in all of the malignant cells; thus they are not random events affecting one or a few cells in the involved tissue. Moreover, they are clonal in origin and are derived from a single cell in which the intial chromosome change occurred. These changes are somatic mutations in individuals who otherwise virtually always have a normal kyrotype in their uninvolved cells. These observations provide the evidence that cancer is a genetic disease. This notion seems self-evident today, but it was not generally accepted several decades ago when many of us began working in cytogenetics. Clearly, I am using ""genetic" in a special way, referring to changes in genes within the affected cell, not in the more usual sense of a constitutional genetic disease such as hemophilia or color blindness. I will conclude with some comments regarding the longer-term potential impact of. discovering new genes via chromosome rearrangements. Once these genes are identified, many previously unknown, BCR or BCL3 for example, they become the focus of very active investigation. Scientists try to find answers regarding the function of these genes in normal cells; how are they altered by the chromosome rearrangements, and how does this relate to malignant transformation? The questions are endless. The answers will provide insights into cell biology that have very profound implications. Within the next decade or two, we should be able to define the major genetic abnormalities in many types of cancer and to identify the specific changes in the tumor cells of many patients. For most leukemias, lymphomas, and sarcomas, unique chromosome changes are often associated with a particular subtype of these neoplasms. Cloning of the genes involved in these chromosome changes will provide specific DNA markers that will have diagnostic importance. For some solid tumors, on the other hand, current evidence suggests that deletions of the same chromosome region may occur in different types of tumors, such as the deletions of 13 q in retinoblastoma, osteosarcoma (not secondary to radiation for retinoblastoma), breast cancer, and lung cancer. The deletion of the same region does not necessarily imply that the same gene is involved or that the change within the gene is identical, witness the fact that two different translocations in band 22q 11 involve different genes, namely, the lambda light chain gene in the 8; 22 translocation in Burkitt's lymphoma and the BCR gene in chronic myelocytic leukemia (CML); furthermore, the breakpoints within BCR in Ph1-positive leukemia are also somewhat variable. The multistep process of malignant transformation is complex. In the leukemias and lymphomas, we often see specific chromosome translocations combined with loss or gain of particular chromosome segments. Some combination of alterations in dominantly acting proto-oncogenes and in recessively acting tumor-suppressor genes certainly act synergistically to enhance the malignant phenotype. In the future, the precise definition of the genetic changes in the malignant cells of a patient will be used to select the most appropriate treatment for cells with these genetic defects. This treatment will be less toxic for the normal cells in the patient. Moreover, this genetic profile may allow monitoring of the patient's course and early detection of relapse. These same genetic markers may be used to detect the involvement of other tissues such as bone marrow, spleen, or lymph nodes. These changes in treatment strategies will clearly benefit the patient. Of more general scientific importance, however, will be the identification of dozens of genes, many hitherto unknown, that can be used to study the complex process of the regulation of cell growth and differentiation. This development may be the most significant result of our success in understanding the genetic changes that occur in cancer cells. 1 would like to conclude with a more personal note based on my continuing amazement at the interrelatedness of the biomedical sciences. This should be no surprise to me, but it is. Many investigators have found that a successful system for carrying out some function in primitive organisms has evolved and then this system is used repeatedly with varying modifications as the organisms become more complex. As a cytogeneticist, 1 had to learn something about the cell cycle and DNA replication, about chromosome structure, and about various agents that can alter both of these. More recently, 1 have had to become an amateur tumor virologist at least with regard to the action of viral oncogenes and their cellular counterparts, the protooncogenes. With the cloning of translocations, especially some of the recent ones, a knowledgeable cancer cytogeneticist must understand cell cycle control genes in yeast (cdc 10 and SWI4 and 6) and developmentally regulated genes in Drosophila (Notch for example) and nematodes (lin-10, glp-1). 1 have already described in some detail, the cloning of the BCL3 gene by McKeithan et al. [17, 18]. If 1 am to understand the possible roles this gene plays in B cell transform ation, 1 must understand how its homology to portions of the cdc 10 and Notch genes might provide clues as to its functions in both normal and malignant cells. As more genes involved in translocations and deletions are defined, many of us in cancer research will continue to have to "go back to school" to be able to incorporate the knowledge provided by molecular geneticists and cell biologists into our concepts of carcinogenesis. The golden age in biology and in medicine are nourishing one another as never before in history.

My colleagues, Drs. Manuel Diaz, Michelle Le Beau, and Timothy McKeithan, have been generous with their review of this manuscript. 1 acknowledge the expert secretarial assistance of Ms Felecia Stokes. The research described has been supported by the Department of Energy Contract No. DE-FGO2-86ER60408 and by United States Public Health Service Grant CA 42557.


1. Rowley ID (1989) Principles of molecular cell biology of cancer: chromosomal abnormalities. In. de Vita VT, Hellman S, Rosenberg SA (eds) Cancer. principles and practice of oncology, 3rd edn. Lippincott, Philadelphia, pp 81-97
2. Heim S, Mitelman F (1987) Cancer cytogenetics. Liss, New York
3. Mitelman F (1988) Catalog of chromo some aberrations in cancer. Liss, New York
4. Trent IM, Kaneko Y, Mitelman F (1989) Human gene mapping 10: report of the Committee on Structural Chromosome Changes in Neoplasia. Cytogenet Cell Genet 51.533-562
5. Bishop IM (1987) The molecular genetics of cancer. Science 235:305-311
6. Rowley ID (1973) A new consistent chromosomal abnormality in chronic myelogenous leukemia identified by quinicrine fluorescence and Giemsa staining. Nature 243.290-293
7. Francke U (1976) Retinoblastoma and chromosome 13. Cytogenet Cell Genet 16:131-134
8. Friend SH, Bernards R, Rogel S, et al. (1986) A human DNA segment with properties of the gene that predisposes to retinoblastoma. Nature 323.643-646
9. Fearon ER, Cho KR, Nigro JM, et al. (1990) Identification of a chromosome 18q gene that is altered in colorectal cancers. Science 247.49-56 10. Mulleris M, Salmon RJ, Zafrani B, et al. (1985) Consistent deficiencies of chromosome 18 and of the short arm of chromosome 17 in eleven cases of human large bowel cancer: a possible recessive determinism. Ann Genet (Paris) 28:206-213
11. Leder P, Battey J, Lenoir G, et al. (1983) Translocations among antibody genes in human cancer. Science 222.765-771
12. Croce CM, Nowell PC (1986) Molecular genetics of human B cell neoplasia. Adv ImmunoI38.245-274
13. Rabbitts TH, Boehm T, Mengle-Gaw L (1988) Chromosomal abnormalities in lymphoid tumours; mechanisms and role In tumour pathogenesis. Trends Genet 4:300-304
14. Rowley JD, Diaz MO, Espinosa R (1990) Mapping chromosome band 11q23 in human acute leukemia. identification of 11 q 23 breakpoints with a yeast artificial chromosome. Proc Natl Acad Sci USA 87:9358-9362
15. Evans GA, Lewis K, Rothenberg BE (1989) High efficiency vectors for cosmic microcloning and genomic analysis. Gene 79:9-20
16. Lichter P, Tang C-JC, Call K, et al. (1990) High-resolution mapping of human chromosome 11 by in situ hybridization with cosmid clones. Science 247.64-69
17. McKeithan TW, Rowley JD, Shows TB, Diaz MO (1987) Cloning of the chromosome translocation breakpoint junction of the t ( 14; 19) in chronic lymphocytic leukemia. Proc Natl Acad Sci 84:9257-9260
18. Ohno H, Takimoto G, McKeithan TW (1990) The candidate proto-oncogene bcl3 is related to genes implicated in cell lineage determination and cell cycle control. Cell 60:991-997