This is a brief personal account of a career in cancer research and tumor cytogenetics, recalling the circumstances of some early findings, the period of frustration when theoretical concepts could not be tested, and the excitement of the last decade as the work has been carried to the molecular level. It is presented from the narrow perspective of one investigator, with no attempt to cover the development of the whole field, and it offers few profound perceptions, except the importance to young researchers of an environment that allows them to pursue unexpected findings, question accepted dogma, and enjoy the privilege of investigating the complexities that underlie human disease.
Introduction to Cancer
I came to medical school at Penn in 1948, knowing that I wanted to pursue a career in research, probably cancer. My chemistry professor in college, having noted my remarkable lack of manual dexterity in the laboratory, had suggested that I consider a Ph.D. in theoretical chemistry but conversations with several physician-scientists convinced me that the M.D. degree would give me more research options relevant to human disease. I found a conducive environment in the Department of Pathology with Drs. Balduin Lucké and Dale Coman and spent several summers working on cancer-related projects. After an internship, I spent slightly more than a year in a pathology residency, at the Presbyterian Hospital in Philadelphia with Dr. Philip Custer, a world authority on leukemias and lymphomas, and one of the few “lumpers” in a field largely occupied by “splitters.” His views on the systemic nature of hematological neoplasms, and the relationships among different clinical presentations, helped to shape my future thinking.
My research career really began, however, in 1954, when the “doctor draft” made me the pathologist for the U.S. Naval Radiological Defense Laboratory (N.R.D.L.) in San Francisco. Here, I performed 5,000 mouse autopsies to serve my country, but, more importantly, formed a collaboration with a little-remembered creative investigator, Leonard Cole, that lasted over the next decade. Cole, who never bothered to obtain a doctorate, was working on problems of radiation carcinogenesis and also attempts to prevent acute death from large doses of radiation. In the former category, we collaborated on a series of studies indicating the multiple factors and steps in radiation carcinogenesis that presaged my later views on clonal evolution in tumors.
Of more immediate impact was the inadvertent demonstration of xenogeneic bone marrow transplantation. When I arrived at the N.R.D.L., Cole was injecting lethally irradiated mice with ground-up rat spleens, and they were surviving long after the controls had died. Of course, they all succumbed later to what we then called “secondary disease.” Nevertheless, Cole was very excited and believed that he had discovered a means to “transfect” rat DNA into the irradiated mice and, thus, allow them to reconstitute their hematopoietic system. Having had my year of pathology training, I suggested that maybe we should look at the rat spleen preparations under the microscope, and found that they did contain intact nuclei. I then stumbled on a paper which indicated that rat granulocytes contained large amounts of alkaline phosphatase, whereas mouse granulocytes were negative. Using this as a marker, we demonstrated that our recovered mice had, in fact, been reconstituted with rat marrow cells, and a group at Oak Ridge published similar findings, using erythrocyte markers. This demonstration that bone marrow could be transplanted across a histocompatibility barrier stimulated George Santos (who joined us in San Francisco in 1956) and others around the world to spend the next 30 years grappling with the problems of graft-vs.-host reactions and ultimately making marrow transplantation a practical clinical tool — with, incidentally, chromosome studies and molecular probes derived from them being used to monitor its efficacy in some patients with leukemia. And thus, cleverly, we finally come to the real subject of this chapter, the cytogenetics of tumors.
Introduction to Chromosomes
With this background at the N.R.D.L., I returned to Philadelphia in 1956 and began some poorly defined studies of leukemia, looking at the growth and differentiation of human leukemic cells in irradiated mice and in vitro. The culture technique was one that had been developed by Edwin Osgood and utilized an extract of navy beans, phytohemagglutinin, to remove the erythrocytes from the leukemic blood — a story I will take up later. The leukemic cells were grown on small slides, and, having been partially trained in pathology, I prepared them for examination by rinsing them with tap water before staining them with Giemsa. I had no idea that I was reenacting the discovery, made earlier by T. C. Hsu and by others, and often equally serendipitously, that hypotonic treatment could be a major aid to mammalian cytogenetics.
When I looked at the slides, I was aware that there were dividing cells and that in many the mitotic spindle had been sufficiently disrupted to allow counting of the individual chromosomes. At that time, I knew nothing of cytogenetics. I was vaguely aware of very recent evidence that the human chromosome number was 46 rather than 48, but I don’t recall having read anything on the chromosomes of tumors. Subsequently, of course, I learned of the long history of speculation, going back to Von Hansemann and Boveri, concerning the importance of chromosome alterations in the pathogenesis of neoplasia; and of limited cytogenetic studies, primarily on experimental tumors, in the early 1950s.
In the meantime, I had not yet met my future collaborator, but I began looking around to see if anyone might be interested in the chromosomes of my leukemic cells. I checked first at the Wistar Institute, across the street, and they referred me to a graduate student at the Institute for Cancer Research in Fox Chase. This was David Hungerford, who was trying to do a thesis on human chromosomes with Jack Schultz, but had little material to work with. A natural collaboration resulted. I continued my investigation of leukemic cell differentiation in culture, but also prepared some slides, under the more sophisticated direction of Dave Hungerford, and sent them up to him for karyotypic analysis. Using the “squash” preparations of the time, we initiated studies on both acute and chronic myelogenous leukemia (AML and CML) as well as on some poorly defined “myeloproliferative disorders.” We found no consistent abnormality in early studies of AML, although three cases had an extra chromosome in group C, probably representing, in retrospect, the trisomy 8 that is the most common alteration in this group of diseases. The results with CML were more encouraging. Dave spotted a characteristic small chromosome in the neoplastic cells of two male patients with CML, and we published this observation with caution because the group in Edinburgh had recently reported negative findings in this disease. We also thought that the “minute” chromosome might be derived from the Y chromosome, since we had not studied any female patients.
Subsequently, with the help of Paul Moorhead, we introduced the improved “air-drying” technique, into our investigations and collected a small series of seven CML patients, including women, in which all but one had the abnormal chromosome. One of these patients is still alive 30 years later and has provided some interesting insights into lymphoid as well as myeloid biology. With these better preparations, Dave correctly determined that the abnormality was derived from the larger pair of the four G-group chromosomes. At the time, this was designated chromosome 21, but when banding techniques subsequently indicated that the trisomy in Down syndrome involved the smaller pair of G’s, they were left as number 21 for historical reasons, and the larger pair became number 22.
We could not state whether the “minute” chromosome resulted from a deletion or a translocation, realizing that the small missing piece would not be recognizable if translocated to a larger chromosome. Even the techniques for quantification of DNA available in the early 1960s did not resolve the problem, and it remained for Janet Rowley, with the banding methods of the 1970s, to demonstrate the true nature of the t(9;22) translocation.
We were readying this first series of CML cases for publication when A. N. Richards, our vice president for health affairs, asked us to present something at a National Academy of Sciences regional meeting that he was organizing at Penn. An abstract of that presentation was published in Science late in 1960, and the finding was promptly confirmed by the group in Edinburgh. They were gracious enough to offer the name “Philadelphia chromosome,” in accord with the 1960 Denver conference on nomenclature which suggested that abnormal chromosomes be named for the city of origin.
This observation of a consistent somatic genetic change in nearly all cases of a specific human neoplasm, and in all the cells of the tumor, seemed to us strong evidence that a tumor could arise from a critical “mutation” in a single cell, providing a growth advantage to the progeny of that cell as they expanded into a recognizable neoplasm. We speculated that there might be similar consistent changes in the AML cases we had studied, but that they were submicroscopic, and that the cytogenetic abnormalities we had observed in AML might represent “secondary” rather than “primary” alterations, or “primary” alterations that varied from case to case. We had no clue to the specific gene or gene product involved in CML, other than the consistently reduced alkaline phosphatase in these cells, nor any means to approach such questions at that time. This, plus the continuing failure to demonstrate consistent cytogenetic changes in other leukemias, and subsequently in solid tumors, led many initially to dismiss the Philadelphia chromosome as an isolated “epiphenomenon.”
Meanwhile, however, our cultures of leukemic cells had yielded another unexpected finding, with major implications not only for cytogenetics, but also for immunology and for general studies of growth regulation. This was the phytohemagglutinin (PHA) story. One day my technician and I traveled to the Presbyterian Hospital, where I had trained briefly as a resident, to obtain some leukemic blood. We found that the patient was in remission, but rather than waste the trip we cultured the peripheral blood leukocytes anyway. To our surprise, we found many mitotic figures. This led to the culturing of my own cells, with similar results, and the conclusion that either I had leukemia or we had found a way to get normal peripheral blood leukocytes to divide in culture.
Again, I knew nothing of early speculations such as that of Haldane in 1932 that “satisfactory mitosis [for cytogenetics] might be observed in a culture of leukocytes,” or of some success in Russia in the 1930s that was interrupted by Lysenkoism. Twenty years later, there was still considerable debate concerning the proliferative capacity and functions of lymphoid cells, and some workers felt that circulating small lymphocytes were poor, effete end-cells with no real function other than possibly serving to transport prepackaged DNA to epithelial cells in the periphery.
Our findings, however, suggested otherwise, and after eliminating one by one the variables in the culture system, it eventually became clear that the mitogen was the PHA we had been using to remove the erythrocytes. I was disappointed that the stimulant was not more physiologic, and tended to agree with one reviewer of the resultant manuscript, submitted to Cancer Research, that it was an interesting observation but of no obvious importance. We did promptly use PHA-stimulated blood cultures and squash preparations to study nonleukemic individuals, including the demonstration of a female (XX) karyotype in a young man who was a phenotypic intersex. This did not deter him from joining the army, and we did not pursue other phenotypic anomalies.
Subsequently, of course, PHA combined with improved hypotonic solutions and the air-drying technique, described in collaboration with Paul Moorhead and Bill Mellman as well as Dave Hungerford, became the standard lymphocyte culture method that remains widely used for routine constitutional chromosome studies. Early on, the chief supplier of PHA created a major crisis by marketing a nonmitogenic product just as many laboratories began to use it for cytogenetics, but they soon began to assay the product not only for its erythrocyte agglutinating ability, but also for its capacity to stimulate normal lymphocytes.
Another early confusion was the nature of the mononuclear cell in the peripheral blood that was being stimulated by PHA. Because circulating large lymphocytes and monocytes had been shown to be labeled following in vivo injections of tritiated thymidine, they seemed to be the most likely candidates for a cell capable of proliferation in vitro. This was apparently confirmed when I did the only “logical” experiment in all of these early studies — and got the wrong answer! I attempted to stimulate with PHA the most homogeneous population of small lymphocytes that I could find, the circulating lymphocytes in chronic lymphocytic leukemia (CLL). These, of course, did not proliferate in response to PHA, apparently ruling out the small lymphocyte as the dividing cell in cultures of normal peripheral blood. It was more than a decade before the differences between B-cells and T-cells were appreciated and this failure was explained (as well as early problems with the cytogenetics of CLL). Fortunately, in the interim, investigators such as Sir James Gowans demonstrated the proliferative capacity of the small lymphocyte in vivo and its central role in immune responses.
These observations, as well as the evidence that corticosteroids could inhibit PHA-initiated proliferation in vitro, promptly led to the wide use of both mitogen- and antigen-stimulated cultures for a variety of investigations in immunology, and also of the sequential steps by which a resting normal mammalian cell moved from G0 into the cell cycle. In several of our own subsequent studies, we were able to combine our experience with radiation effects, cytogenetics, and leukocyte culture. For example, we utilized unstable chromosome changes (acentrics, dicentrics, rings) induced by therapeutic radiation to identify long-lived antigen-responsive T-cells in humans; and in studies with Darcy Wilson, in the rat, used cytogenetics, including radiation-induced chromosome markers, to demonstrate lymphohematopoietic stem cells and various aspects of lymphocyte function.
Thus, the serendipitous observation of metaphases in human leukemic cell cultures spawned several major fields of inquiry, which are still active. With respect to tumor cytogenetics, however, they also ushered in a decade when the lack of additional technical approaches prevented more productive exploitation of the early observations.
Clinical Correlations and Clonal Evolution
During most of the 1960s and into the early 1970s, we and others made very little progress in our attempts to utilize chromosome studies either to understand better the fundamental nature of neoplasia or to find many practical clinical applications. The relatively crude chromosomal preparations, without banding, were a major limitation, and it also became apparent that tumor cells responded less well to the standard preparative techniques than did normal cells, often yielding metaphases with “fuzzy” and overlapping chromosomes. Furthermore, it was recognized that in PHA-stimulated bone marrow or blood cultures, normal lymphocytes might also proliferate and cause the technically poor neoplastic metaphases to be overlooked.
We did initiate some longitudinal studies of patients with the spectrum of “preleukemic” states that are now called myelodysplastic and myeloproliferative disorders. The early results suggested that a chromosomally abnormal clone in the bone marrow indicated a poor prognosis, and subsequent investigations have confirmed and extended this conclusion. The cytogenetic studies also helped demonstrate the relationship of these preleukemic disorders to the subsequent leukemias, through progression of karyotypic evolution.
Much of my time during this period, however, was occupied with being a department chairman and then director of our new cancer center, and by the time I returned actively to the laboratory, a revolution had occurred. Chromosome bonding techniques had been developed and were being applied around the world, with Janet Rowley and many others describing nonrandom chromosome changes in a variety of leukemias and lymphomas, so that the Philadelphia chromosome was finally no longer an isolated phenomenon. These banding studies also began to be extended to solid tumors, with similar but more complex results, and it was increasingly clear that nearly all neoplasms had visible somatic genetic changes importantly involved in their development. Interestingly, as early as 1964, I had observed, in lymphocyte cultures treated with mitomycin, metaphases that contained elongated chromosomes with horizontal bands or “chromomere” patterns, as well as many bizarre chromatid exchanges, but did not attempt to use this approach for chromosome identification.
One concept that I did pursue was the idea of clonal evolution within tumor cell populations. In the 1950s, several workers had examined both biologically and cytogenetically the phenomenon of tumor progression — the tendency of neoplasms to become more aggressive in their behavior and more “malignant” in their characteristics during their life history. By the 1960s, my own work in tumor cytogenetics, in human leukemias and preleukemias, and in various animal species (rat, rabbit, dog), as well as the studies of radiation carcinogenesis, had helped to crystallize my thinking, and I found over the next decade that the resultant model of tumor development, extending the earlier concepts, was very useful in my introductory lectures on neoplasia to medical students.
As finally written for a review article in Science in 1976, the clonal evolution model was hardly revolutionary, although some diehards were still resisting the idea that tumors resulted from somatic genetic change. In that review, I cited both cytogenetic and other evidence to support the concept that tumors arise from a single “mutated” cell, and that biological and clinical progression results from subsequent additional genetic alterations, giving rise to more aggressive subpopulations within the original neoplastic clone. More controversial was the further suggestion that the likelihood of such sequential changes in tumor cells was enhanced by increased genetic instability in these cells, acquired as part of the neoplastic process. Although several possible mechanisms for this increased lability were proposed, I really had very little firm evidence to offer.
We subsequently did a few speculative studies on chromosomal fragility syndromes and considered possible parallels between both clinical and “subclinical” inherited genetic instability disorders and the lability acquired as a somatic alteration in neoplastic clones, but only very recently has any real information on the mechanisms underlying this phenomenon begun to be developed.
Interestingly, the initial reaction to the clonal evolution article was generally favorable despite its rather pessimistic implications for simple answers to cancer therapy. Recently, of course, there has been increasing cytogenetic and molecular evidence on the specific steps involved in the pathogenesis of different tumors, and that brings us to the final decade of this reminiscent ramble.
Oncogenes and Oncogenesis
As we moved into the 1980s, recombinant DNA technology finally made it possible to use the clues provided by nonrandom cytogenetic alterations in human tumor cells to look for the growth regulatory genes involved and the mechanisms by which the function of these genes is altered. These explorations, involving translocations, deletions, chromosomal additions, even microscopically visible gene amplification units, have demonstrated the involvement of a few previously known stimulatory oncogenes and suggested many more that were previously unknown. Cytogenetics has also helped contribute to the recognition of the equally important family of tumor suppressor genes and the overall complexity of tumorigenesis that still remains to be unraveled.
I became involved when Dr. Carlo Croce, at the Wistar Institute, suggested that we use a combination of cytogenetic, molecular genetic, and somatic genetic techniques to explore chromosome translocations in lymphoid tumors. With Guy Tsujimoto, Jan Erikson, Beverly Emanuel, and others, as well as my long-time chief cytogeneticist, Janet Finan, this collaboration first explored the t(8;14) translocation of Burkitt’s lymphoma and helped to demonstrate that the c-myc gene is deregulated when it is brought into juxtaposition with a transcriptionally active immunoglobulin heavy chain locus on chromosome 14. The same approach was extended to other translocations and led to the identification of the bcl-2 gene in the t(14;18) translocation of many follicular lymphomas. Thus far, this is the best-characterized of many previously unknown growth regulatory genes apparently involved in the pathogenesis of these B-cell tumors, and others are being actively investigated.
Similar methods were then used for T-cell neoplasms, where we and others demonstrated that the c-myc gene or a different set of previously unknown oncogenes might be deregulated, through chromosome translocation, by being brought adjacent to a T-cell receptor gene. Sequential involvement of several such oncogenes was also shown, in association with clinical progression of both B-cell and T-cell tumors, and other genes capable of “activating” c-myc have also been identified.
From these studies with Carlo Croce and the work of other laboratories, some two dozen translocations in human lymphoid tumors have now been investigated that appear to result in the inappropriate expression of a growth regulatory gene and, thus, apparently contribute importantly to the development of a neoplastic clone. Interestingly, among all of these, only the myc gene represents a previously known “oncogene.” Also, the newly recognized putative oncogenes involved in these processes seem to function in either the B-cell or T-cell lineage, but not in both.
We have also been involved, to a limited degree, in sorting out the molecular variations in Ph-positive acute leukemias and in defining the sequential cytogenetic and molecular changes associated with tumor progression from benign nevus to malignant melanoma. All of these studies, as well as data from many other types of tumors, are indicating every day the enormous variability in the sequence of events leading to the full development of different neoplasms, and the lineage specificity of many of the genes involved. We are continuing our studies with Carlo Croce and extending the work into efforts to identify tumor suppressor genes and mechanisms for their deletion in both lymphoid and myeloid neoplasms. The work is exciting, because we finally have the tools to explore the clues provided by the chromosomal observations; it is also frustrating, because it is clear that there are no simple, easy answers.
In 1985, Mike Bishop, in one of his refreshing reviews, noted that studies of tumor chromosomes were no longer “mere amusements for the myopic microscopist,” because they did provide useful leads for the molecular geneticist to pursue. I suggested to him that we microscopists had known all along that these were important clues, but had to wait for the retrograde retrovirologists to provide us with the means to exploit them. He graciously acknowledged the interdependence, and it is clear that a happy marriage has occurred in the joint effort to unravel the genetic basis of human neoplasia.
So, this personal reminiscence of 40 years of cancer and chromosomes ends on a note of excitement and enthusiasm. The next several decades will undoubtedly be an era of profound discovery and ultimately important practical applications. Tumor cytogenetics will continue to provide important signposts for more detailed molecular dissection. I can only hope that the young people entering the field, and having the newer tools to extend these studies, will also find, as I did, an environment that allows inquiry in unexpected directions and easy links to clinical applications. These are the factors that have made, for one investigator, a remarkably satisfying career.