From fruit flies to fallout: Ed Lewis and his science


  • Howard D. Lipshitz

    Corresponding author
    1. Program in Developmental Biology, Research Institute, The Hospital for Sick Children & Department of Molecular and Medical Genetics, University of Toronto, Toronto, Ontario, Canada
    • Program in Developmental Biology, Research Institute, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada
    Search for more papers by this author

  • This article is based on a talk presented in honor of Ed Lewis at the California Institute of Technology on February 4, 2004 to coincide with the publication of “Genes, Development and Cancer: The Life and Work of Edward B. Lewis.” The present tense has been retained despite Lewis' recent death on July 21, 2004. The article was published in the Journal of Genetics (Volume 83, pages 201–218, 2004) and is reprinted with the permission of the publisher.


Edward B. (“Ed”) Lewis was born in Wilkes-Barre, Pennsylvania, on May 20, 1918, and educated at the E.L. Meyers High School in Wilkes-Barre. He grew up during the Great Depression; his father, a jeweler and watchmaker, supported Ed's educational and musical aspirations, despite the economic hardships. One of the highlights for Ed was the public library in Wilkes-Barre, which he used to visit daily. Two books and a journal, which he read there as a teenager, had a major impact on his life. The journal was Science and the first book was Bertrand Russell's The Scientific Outlook (Russell,1931), both of which I will return to later. The second book, The Biological Basis of Human Nature by H. S. Jennings, a Professor at Johns Hopkins University, and dedicated by him to Calvin Bridges (Jennings,1930), is where Ed first read about Drosophila and the power of Drosophila genetics (Lewis,2004). Shown in Figure 1 are two illustrations from the book. On the left is “sex-linked inheritance” of the white gene, which was the subject of T.H. Morgan's first paper on Drosophila genetics in 1910; on the right is the inheritance of the dominant Bar mutation, which A.H. Sturtevant did a lot of work on, and which was important in motivating Ed to his graduate work. Ed does not remember seeing the Bar figure and says that, if he did, he probably would not have understood it anyway!

Figure 1.

Drosophila genetics as illustrated in H.S. Jennings' 1930 book The Biological Basis of Human Nature, which Ed Lewis read while in high school. On the left are the parents and progeny from a cross of white females to wild-type males. Because of sex-linkage, the male progeny are white eyed whereas the female progeny have red eyes. On the right is a cross of males carrying the sex-linked dominant Bar mutation to wild-type females, which results in females with Bar eyes and males with wild-type eyes. Reprinted from Jennings (1930).

Ed was a member of the E.L. Meyers High School biology club, which at that time was chaired by Ed Novitski, who also become a Drosophila geneticist. In late 1934, Ed (Lewis) noticed an advertisement in Science for Drosophila cultures from Turtox Products in Chicago for a dollar each. I searched through 4 years worth of Science to find this advertisement but failed. Both Eds, however, assure me that it existed. After a Web search, I did manage to obtain the 1933 issue of a Turtox catalog (Turtox,1933) and sure enough, there the ad is, offering live Drosophila mutants for two dollars a culture (Fig. 2). Ed thinks that, because of the Great Depression, by 1934 the price had dropped to only 1 dollar. Novitski tells me that not being Drosophila geneticists, when he wrote the letter to obtain this culture, he asked whether for the same price they could throw all the mutants into the same vial. Professor Rifenberg at Purdue who filled the order, wrote back saying that it was clear that they were new to genetics, but for the price, he would send several vials. That was the beginning of Ed's career as a geneticist.

Figure 2.

A 1933 Turtox catalog with an ad for Drosophila stocks. The ad is very similar to the one that Ed saw as a high school student in the Osterhout Library in Wilkes-Barre, Pennsylvania, in late 1934. Lewis and Ed Novitski obtained the stocks for 1 dollar, launching both of them on the road to becoming Drosophila geneticists. [Color figure can be viewed in the online issue, which is available at].

In his autobiography, Ed emphasizes that “by allowing Novitski and me freedom to use the biology lab and its supplies to carry out our experiments on Drosophila, our biology teacher, who also was the athletic coach, could not have been more helpful in furthering our careers. There was none of the present attitude that one cannot become a scientist without having had the benefit of teachers skilled in the art of keeping their students constantly motivated” (Lewis,2004). In other words, Ed was motivated from the beginning and has stayed motivated throughout his life.

Ed is an excellent musician (Fig. 3). He was given a wooden flute by his great-uncle, Thomas Wyllie, in 1928. Later, during the Depression, his father somehow managed to come up with funds to buy Ed a silver flute. He played in the High School Orchestra and the Wilkes-Barre Symphony, beginning his tertiary education at Bucknell College on a music scholarship.

Figure 3.

Ed playing the flute in the living room of the author's home in Toronto in 1996. He began his tertiary education on a music scholarship to Bucknell College in 1936 but switched to the University of Minnesota after 1 year to focus on biostatistics and genetics. He remained an enthusiastic and accomplished flautist for the rest of his life. Reprinted from Lipshitz (2004). [Color figure can be viewed in the online issue, which is available at].

After a year at Bucknell, science triumphed and Ed went to the University of Minnesota to complete his undergraduate studies. He chose that university for two reasons: first, money was tight, and the University of Minnesota had the lowest out-of-state tuition of any major state university; second, because the Reserve Officer Training Corps (ROTC) was not compulsory. Ed's older brother, Jimmy, who went on to a distinguished career in the US diplomatic service, assisted financially, enabling Ed to complete his undergraduate degree in biostatistics, which he did in 2 years.

It was at the University of Minnesota that Ed started working on the recessive allele of Star, which Novitski had found and had sent to him (Lewis,1939). Ed continued these analyses as a graduate student at the California Institute of Technology (Caltech) as one of A.H. Sturtevant's students. During his 3-year graduate career (1939–1942), Ed was to invent the “cis–trans test” for position effects that all of us study in our introductory genetics courses (Lewis,1941,1942,1945).

The Second World War interrupted Ed's career as a geneticist. He stayed on at Caltech to complete a Master's degree in meteorology and then joined the US Army Air Corps, first in Hawaii and subsequently as a weather officer for the US 10th army in Okinawa before the invasion of Japan (Fig. 4).

Figure 4.

Lieutenant Lewis at his weather officer's desk, G2 section, US Tenth Army, Okinawa, 1945. Reprinted from Lipshitz (2004).

Ed returned to Caltech in 1946. In his autobiography he writes that, in 1943, “Millikan called me into his office and simply said: ‘This war will not last forever. When it is over we would like you to come back as an instructor’…Obviously Sturtevant had a hand in this style of making appointments that was typical of the times” (Lewis,2004). Ed has spent his entire independent career at Caltech, becoming the Thomas Hunt Morgan Professor of Biology in 1966, attaining Emeritus status in 1988, but remaining active in his lab.

The other big event of 1946 was that Pamela Harrah became Pamela Lewis. Beadle had returned to Caltech from Stanford in 1946 to chair the Biology Division. That same year Ed, as an Instructor at Caltech, had taken over the Drosophila stock collection and was looking for a stock keeper. While still at Stanford, Beadle called Pam (who had done an undergraduate degree there) into his office and said, “Hey Pam, how tall are you?” to which Pam replied, “5' 3”.” Beadle then said, “Your new Boss is 5' 4” tall, he's 28 and maybe you will like him so much, you will fall in love and decide to stay there at Caltech” (Berg and Singer,2003). A few months after meeting, Ed and Pam were married and have remained so for over 57 years (Fig. 5).

Figure 5.

Ed, Pam, and Hugh Lewis (3 months old, in the high chair) en route to the United Kingdom on the Queen Mary ocean liner in 1948. Ed had been awarded a Rockefeller Foundation Fellowship, which he chose to spend in Dr. David Catcheside's laboratory at Cambridge University. Reprinted from Lipshitz (2004).


Understanding Ed's approach to science is crucial to understanding his motivations as well as his papers. Ed chose to quote from Bertrand Russell's book The Scientific Outlook (Russell,1931)—which he had first encountered as a high school student—to begin his Nobel lecture: “The power of using abstraction is the essence of intellect and with every increase in abstraction, the intellectual triumphs of science are enhanced” (Lewis,1995). Abstraction lies at the heart of Ed's science.

Genetics is an abstract discipline; therefore, Ed was at home in genetics, and his first love has always been genetics. The two greatest abstractions in genetics are Mendel's 1866 framing of the laws of inheritance (Mendel,1866) and Sturtevant's invention, as an undergraduate at Columbia University in 1913 while working in Morgan's fly room, of linear mapping of genes based on meiotic recombination frequencies (Sturtevant,1913b).

Almost all of Ed's papers on Drosophila present his data in terms of abstract models, attempting to accomplish Russell's goal. I am not going to describe the models here, but I thought it might be interesting to look at a few of them, taken from Ed's papers published over a 40-year period (Fig. 6). It should be noted that Ed not only abstracted models but he always made them current, based on what was happening in other disciplines such as biochemistry and molecular biology.

Figure 6.

Some of Ed's abstract models. Top: A model for the control of sequential chemical reactions by pseudoallelic genes (Lewis,1951). Second from top: A model for transvection (reprinted with the permission of the University of Chicago Press from Lewis,1954). Third from top: An operon-type model applied to the bithorax pseudoalleles (Lewis,1963a). Bottom left: An additive model for BX-C gene function in specifying segmental identity (Lewis,1981; reprinted with permission of Elsevier). Bottom right: Ed's model for transvection that incorporated a substance ‘E’ (Lewis,1985), which he speculated might correspond to the recently discovered noncoding RNAs derived from the cis-regulatory regions of the BX-C (Lipshitz et al.,1987).

It is very important when reading Ed's papers to distinguish between what he calls “rules” and what he calls “models.” Each rule is Ed's description of a particular genetic phenomenon that he has discovered. The models, in contrast, are his abstractions of those phenomena. To take one example, consider the “cis-vection” rule, which derived from Ed's invention of the cis–trans test (Fig. 7, left; Lewis,1945,1951). If the bx3 and Bxl mutations are in cis, there is a very weak phenotype, but, when they are in trans, there is a strong phenotype. The abstraction attempts to explain this phenomenon using the biochemical pathway terminology of the time (Fig. 7, right; Lewis,1951).

Figure 7.

“Rules” vs. “models.” On the left is the cis-vection rule, which derived from the results of the cis–trans test (shown for two bithorax complex “pseudoalleles”). On the right is Ed's abstract model of the cisvection phenomenon. In this case, “+” and “a” would represent the function of the wild-type and bx3 alleles while “+” and “b” would represent the function of the wild-type and Bxl alleles. On the left, these are shown in cis and on the right in trans. From Lewis (1951).

What struck me as I was writing the commentary to accompany Ed's papers, was that—even 50 years after many of these rules were framed—we have very little understanding of their underlying molecular basis. We know quite a lot about the mechanisms of cis-vection but we do not yet understand “trans-vection” or “cis-overexpression.” We also have little knowledge of the molecular basis of the “colinearity” rule, for which Ed is perhaps best known: that the order of the bithorax complex (BX-C) genes in the chromosome is the same as the order, along the fly's body axis, of the segments whose development they control (Fig. 8; Lewis,1978).

Figure 8.

The colinearity rule. The order of the BX-C genetic functions on the chromosome (bottom) correspond with the order of the body segments whose identity they control (top). From Lewis (1998). [Color figure can be viewed in the online issue, which is available at].


Ed's scientific work can be divided into two parts: first his Drosophila research and, second, his work on ionizing radiation and cancer. I will first briefly review his Drosophila research. For the first 25 years of Ed's career in Drosophila, his goals were (Lewis,1955) “understanding the gene, how it functions, how it mutates and how it evolves.” Ed's later progression to studying development, which is what he is most famous for, reflects that when you do science, you never end up going where you think you will.

Ed began this work several years before DNA was shown to be the genetic material and long before the double helical structure of DNA was proposed. At the time, genes were regarded as indivisible units distributed on chromosomes much like beads on a string. It was thought that meiotic recombination occurred only between—never within—genes. Sturtevant's multiple allelism hypothesis had stated that, if two recessive mutations, a and b, give a mutant phenotype when in trans, they are alleles of the same gene (Sturtevant,1913a). Recombination, thus, would not be expected to occur between a and b. In addition, Sturtevant, in 1925, had discovered a “position effect” at the Bar locus (Fig. 1B): when the two Bar mutations are present in cis on the same chromosome the phenotype, which is a reduction in the size of the eye, is more severe than when the two mutations are present in trans (Sturtevant,1925). Finally, Bridges, in assembling the Drosophila polytene chromosome maps in 1935, had hypothesized that polytene chromosome doublets might represent tandem gene duplications (it was thought that each chromosome band represented a single gene), which “could subsequently mutate separately and diversify their effects” (Bridges,1935).

The challenge for Ed, an iconoclastic young scientist, was to determine whether it might indeed be possible to obtain recombination between members of a multiple allelic series. In particular, how would one detect recombination? Ed's invention of the cis–trans test for position effects accomplished this goal. He applied the test in his Ph.D. thesis research to the Star and asteroid pseudoalleles (Fig. 9; Lewis,1942,1945) and, later, to Stubble and stubbloid (Lewis,1951), white and apricot (Lewis,1952), and the bithorax mutant series, beginning in 1946 (Lewis,1951; for example, see Fig. 7, left). All of these experiments showed that recombination can, in fact, occur between what Ed termed “pseudoalleles” and that the Star and asteroid as well as the bithorax series of pseudoalleles map to polytene chromosome doublets (Fig. 10), as Bridges had suggested might be the case for newly evolving genes.

Figure 9.

The cis–trans position effect between the Star and asteroid pseudoalleles, which Ed studied during his Ph.D. thesis research. On the left is the cis configuration (+ +/S ast), which results in an almost wild-type eye. On the right is the trans configuration (S +/+ ast), which almost completely eliminates the eye. From Lewis (1951).

Figure 10.

Pseudoalleles map to polytene chromosome doublets. In particular, Star and asteroid map to 21E1-2 and the bithorax mutations to 89E1-2. From Lewis (1951).

This development led Ed to spell out very clearly the hypothesis that gene evolution may occur by tandem gene duplication (Lewis,1951): “We picture…a two step process; namely (1) the establishment of the duplication of that gene…and (2), the occurrence in one of the two…genes thus formed, of a ‘mutation into a new function.’ ” He goes on to describe the potential mechanisms and constraints on the process in some detail. The gene duplication and divergence hypothesis was popularized approximately 20 years later by Susumu Ohno (Ohno,1970).

In the mid 1950s, Ed discovered another strange position effect that he called trans-vection (Lewis,1954). When he introduced a chromosomal rearrangement that would disrupt somatic chromosome pairing, he found that the trans phenotype of two pseudoalleles often was distinct from that exhibited when the chromosomes were permitted to pair. A striking example is shown with Ubx and bx (Fig. 11); in this particular case, the phenotype becomes much worse when pairing is disrupted. Ed hypothesized that there must be some substance that is able to pass from one chromosome to the other, thus partially rescuing the mutant phenotype. In the case of Ubx and bx, disruption of pairing prevents passage of this substance, making the phenotype more severe. Although substantial progress has been made, we do not fully understand the molecular basis of transvection 50 years after Ed discovered the phenomenon.

Figure 11.

Transvection. When chromosomes are able to pair somatically (left), only a weak transformation of the third thoracic segment toward second occurs (the enlarged halteres can be seen). When the chromosomes are unable to pair (right), a more severe transformation occurs, particularly of the notum or body (arrow). Ed's abstract model is shown above the photographs of mutant flies. Reprinted with the permission of the University of Chicago Press from Lewis (1954).

Another major advance derived from Ed's detailed analysis of the phenotypic differences among the bithorax family of mutants (Fig. 12). Together with his genetic analyses, this led Ed, in his 1951 paper, to the idea that the bithorax family mutants control the development of particular segments and that the second thoracic segment in some sense is the developmental “ground state” (Lewis,1951). The function of the bithorax genes, thus, is to convert segments from this ground state to more posterior segmental identity (i.e., from second thoracic to third thoracic as well as abdominal identity; Fig. 13).

Figure 12.

Distinct but related developmental transformations of the third thoracic haltere toward second thoracic wing caused by mutations in what Ed later came to call the bithorax gene complex (BX-C). Right: a wild-type haltere. Center: bxd mutations result in transformation of the posterior part of the haltere toward wing (posterior is to the right). Left: In contrast bx mutations result in transformation of the anterior part of the haltere toward wing. From Lewis (1951).

Figure 13.

The second thoracic segment represents a developmental “ground state.” The role of the BX-C genes is to convert the more posterior body segments away from a second thoracic level of identity (L-ms) toward third thoracic (L-mt) or abdominal (L-ab) identity. After Lewis (1951).

In the mid-1950s Ed discovered a very bizarre pair of alleles that he called postbithorax (pbx) and Contrabithorax (Cbx; Lewis,1955). The pbx mutation converts part of the haltere (third thoracic segment) toward wing (second thoracic segment; Fig. 14, left). The Cbx mutation, in contrast, produces the opposite phenotype, converting wing toward haltere (Fig. 14, right; the striking Haltere-mimic phenotype is shown, which is similar but not identical to Cbx). Analyses of Cbx in particular led Ed, by 1963, to define “gain-of-function” and “loss-of-function” mutations: Mutations such as pbx are “in the direction of loss of function,” whereas Cbx “effects…are in the direction of…a gain in…function” (Lewis,1963a). “It is as if the Cbx mutant causes the wild-type alleles of certain other genes in the series to begin acting in the second thoracic segment instead of acting in the third thoracic or any abdominal segment.” When molecular biology elucidated the nature of the pbx and Cbx mutations 20 years later (Bender et al.,1983), it turned out that Ed's hypothesis was correct in its essence: the Cbx mutation causes the Ubx protein to be expressed inappropriately in the second thoracic segment (Cabrera et al.,1985; White and Akam,1985).

Figure 14.

The pbx loss-of-function phenotype is the reciprocal of the Cbx gain-of-function phenotype. Left: pbx mutations cause posterior haltere to transform to posterior wing. From Lewis (1963a). Right: The Haltere-mimic mutation—which falls into the Cbx class of mutations—causes wing to transform into haltere. From Lewis (1982).

The second quarter century of Ed's career in Drosophila commenced in 1963, with a change in focus from “genes, their function and evolution,” to development. He begins both his 1963 and his 1964 papers essentially laying out the genetic approach to development: “There are many experimental approaches to the problem of growth and development. We would like to discuss what might be called the genetic approach. The underlying concept is that the genetic mechanisms which are believed to control and regulate biosynthetic pathways may be applicable with relatively little modification to the control and regulation of developmental pathways” (Lewis,1964). Little did Ed know that he was to discover a series of rules about how genes control development, which could in no way have been derived from the earlier studies on biosynthetic pathways!

During the 1960s, Ed began to study genes that behave as positive or negative regulators of the bithorax complex—what we now call the trans-acting regulators (Lewis,1968). He also started to address the spatial and temporal control of development by bithorax complex genes through his analyses of genetic mosaics (Lewis, 1963,1964). Using mosaics, he was able to ask whether the bithorax complex genes confer the fate of cells autonomously or whether they encode diffusible substances, which, therefore, function non–cell-autonomously. He began to study cell-memory as well as to analyze lethal mutations in the bithorax complex using genetic mosaics. One example is Ed's analysis of the lethal, Ubx alleles, which he studied using both gynandromorphs and mitotic recombination-induced mosaics. To generate gynandromorphs, Ed inserted a small duplication carrying the wild-type alleles of the complex into an unstable ring-X chromosome (Fig. 15, upper panel). Loss of the ring chromosome during development results in patches of mutant tissue in an otherwise genetically wild-type fly. Strikingly, Ubx behaved completely cell-autonomously (Fig. 15, lower panel), consistent with this gene encoding, a nondiffusible substance that gives identity instructions to each epidermal cell in which it is expressed (Lewis,1963a,1964).

Figure 15.

Genetic mosaics enabled Ed to analyze lethal BX-C alleles (in the case illustrated, Ubx) and proved that the genes function cell-autonomously in determining the identity of the cuticular structures. Top: The BX-C was moved from the third chromosome onto an unstable ring-X chromosome. Loss of this ring-X results in patches of mutant cells in an otherwise wild-type fly. Bottom: Single genetically wild-type bristles can develop an appropriate identity in an otherwise mutant region (lower arrow), proving that Ubx functions cell-autonomously. From Lewis (1963a).

With the recently discovered “lac operon” in mind, Ed suggested that the bithorax genes “evidently…[produce] a whole set of substances that repress certain systems of cellular differentiation and, thereby, allow other systems to come into play” (Lewis,1964). Subsequently, he extended this finding to suggest that the bithorax substances would function through both activation and repression. Twenty years later, molecular analyses proved this hypothesis to be correct: the BX-C was shown to encode homeodomain-containing transcriptional regulatory proteins (Laughon and Scott,1984; Shepherd et al.,1984).

Ed's 1978 article (Lewis,1978) is “a paradigm for the genetic control of development” (Lipshitz,2004). It summarizes 30 years worth of research in approximately six pages. It is very difficult to read but it is an amazing paper that is well worth the effort.

In the 10- to 15-year publication drought that preceded his 1978 paper, Ed moved from studying bithorax in adults to examining the mutant phenotypes in embryos, using the landmarks on the embryonic cuticle to determine segmental transformations. This strategy laid the groundwork for rapid acceptance of the work of Christiane Nüsslein-Volhard and Eric Wieschaus who, starting in the late 1970s, carried out comprehensive screens for embryonic lethal mutants with cuticular pattern defects. They shared the Nobel Prize with Ed in 1995.

For me, the most remarkable part of Ed's 1978 article is his invention of what can be called “add-back genetics” (Lipshitz,2004). Standard genetics involves mutating or deleting genetic functions and inferring the wild-type function of genes from their mutant phenotypes, a strategy that Ed had applied very successfully to the bithorax complex for almost 30 years. In contrast, add-back genetics began by deleting the entire bithorax complex and then adding back, bit by bit, wild-type pieces of the complex. In this way, Ed was able to define the location and the wild-type function of genes for which he had not yet obtained mutations. He, thus, also avoided a common trap that people doing standard genetics fall into: depending on the nature of the mutations analyzed, one can incorrectly infer the function of the wild-type gene.

Ed was able to invent add-back genetics because, over the previous 30 years, he had collected a large number of chromosomal rearrangements that enabled him to delete the whole bithorax complex (BX-C) and then add it back one piece at a time (Fig. 16). Thus, starting with a deficiency for the whole BX-C in which all segments look like the second thoracic segment (Fig. 17), he was able to add back wild-type subsets of the complex and assess how each piece directs development. That approach led him to propose that there are 12 different genes in the BX-C, which turn on progressively one at a time from more anterior (fewer genes “on”) toward more posterior (more genes “on”) segments. Thus, the fate of any particular segment would be specified additively by the sum of the BX-C genes turned on in it (Fig. 18; Lewis,1978,1981).

Figure 16.

Ed's extensive collection of chromosomal rearrangements that affect the BX-C enabled him to invent “add-back” genetics. A genetic map of the BX-C is shown at the top. Below are some of the available chromosomal rearrangements whose breakpoints are diagrammed relative to the genetic map as well as to the polytene chromosome map (bottom). From Lewis (1978); reprinted with permission from Nature.

Figure 17.

a: Removal of the entire BX-C (using Df-P9) results in transformation of all thoracic and abdominal segments toward mesothoracic identity (T2, left). b: A wild-type mature embryo is shown on the right. Add-back genetics involved adding pieces of the BX-C to Df-P9 and asking which structures and segmental identities develop. Ed's use of cuticular landmarks (arrows) pioneered analyses of embryonic lethal pattern mutants. From Lewis (1978); reprinted with permission from Nature.

Figure 18.

The additive model for BX-C gene function. Ed postulated that the BX-C genes turn on sequentially from anterior to posterior along the body axis and stay on in the more posterior segments. Thus, the identity of any segment is determined by the sum of all BX-C genes active in that segment. The second thoracic segment (T2) has no active BX-C genes, whereas abdominal segment 8 (A8) has all of the genes turned on. From Lewis (1981; reprinted with permission of Elsevier).

In his 1978 article, Ed also followed up on a mutant, Polycomb, which Pam Lewis had identified in his lab in 1947. From the transformation of thoracic and abdominal segment identity toward the eighth abdominal segment (Fig. 19), Ed inferred that it must be a trans-acting negative regulator of the BX-C (Duncan and Lewis,1982; Lewis,1978). In terms of his additive model, in Polycomb mutants, there would be no repression; thus, all the BX-C genes would be on in all segments, which would, therefore, adopt the identity of the most posterior segment.

Figure 19.

Polycomb is a trans-acting negative regulator of the BX-C. Left: Wild-type embryonic cuticle. Center and right: Cuticles from Polycomb mutant embryos carrying either two (left) or four (right) copies of the BX-C. Thoracic and abdominal segments are converted toward the eighth abdominal segment in identity. From Duncan and Lewis (1982).

A giant leap forward in the mechanistic understanding of development commenced in 1978 with the famous collaboration between David Hogness, a molecular biologist at Stanford, and Ed (Fig. 20). This collaboration led to the first positional cloning of a gene—the Ubx gene in the BX-C—and the first functional genomic analyses, which correlated the DNA map, the transcripts, the mutations, and the phenotypes (Bender et al.,1983). Featured on the cover of the issue of Science in which the research was reported in July 1983, was Ed's famous four-winged fly (Fig. 21). Ed for the first time saw the mutations that he had identified and studied molecularly mapped on the DNA cloned from the bithorax complex (Fig. 22).

Figure 20.

Ed with David Hogness at the Thirteenth Biennial EMBO Workshop on Molecular and Developmental Biology of Drosophila, Kolymbari, Crete, Greece, in June 2002. Their collaboration, begun in 1978, resulted in the first positional cloning of a gene and the first functional genomic analysis. From Lipshitz (2004). [Color figure can be viewed in the online issue, which is available at].

Figure 21.

Ed's famous four-winged fly graced the July 1, 1983, cover of Science in which the first functional genomic analysis—of part of the BX-C—was reported. Reprinted with permission from Science, copyright AAAS. [Color figure can be viewed in the online issue, which is available at].

Figure 22.

With the cloning of the BX-C, Ed now saw the mutations he had studied during the previous 37 years, mapped on the DNA of the first chromosomal “walk.” Note in particular that pbx1 is a 17-kilobase deletion of DNA, which was inserted some 40 kilobases to the left to produce the Cbx1 mutation. These two mutations had arisen together after X-ray exposure and had been separated by recombination in 1955 by Ed. Reprinted with permission from Bender et al. (1983), copyright AAAS.

The molecular analyses revealed that Ed in his earlier hypothesis had been both right and wrong. Right in that the genes in the complex had indeed evolved by tandem duplication: there are three tandemly duplicated protein-coding genes in the BX-C, which are characterized by the so-called homeobox (Fig. 23; Regulski et al.,1985). Wrong in that most of the 12 “genes” that Ed had identified in the BX-C are in fact cis-regulatory regions that control the time, place, and level of expression of the homeobox-containing mRNAs (Beachy et al.,1985; O'Connor et al.,1988; Kornfeld et al.,1989). One could not have predicted this molecular complexity solely on the basis of Ed's genetic analyses; the combination of molecular with genetic methods was required.

Figure 23.

Tandem duplication gave rise to the BX-C. There are three protein-coding transcription units in the BX-C, each of which contains the evolutionarily conserved homeobox (H). All of the rest of the DNA comprises giant cis-regulatory regions that control the time, place, and amount of these transcripts. From Lewis (1998). [Color figure can be viewed in the online issue, which is available at].

The molecular analyses led, after the discovery of the homeobox, to the amazing finding in 1984 that the homeobox gene (HOX) complexes are in fact conserved from flies to mammals (McGinnis et al.,1984). Furthermore, Ed's colinearity rule—that the order of the homeotic complex genes in the chromosomes corresponds to the order of the body segments that they control along the body axis—applies all the way from flies to mammals (Duboule and Morata,1994; Duboule,1998). Thus, the HOX complex predates the divergence of the ancestors of flies and mammals over five hundred million years ago. It was this extension and generalization of four decades of Ed's genetic analyses that led directly to the realization, in 1995, that it was time to award him the Nobel Prize (Fig. 24).

Figure 24.

Ed receives the 1995 Nobel Prize in Physiology or Medicine from King Carl XVI Gustaf of Sweden in Stockholm. Eric Wieschaus and Christiane Nüsslein-Volhard, who shared the Prize with Ed, can be seen standing to Ed's left in the front row, third and fourth from the center aisle. [Color figure can be viewed in the online issue, which is available at].


H.J. Muller, in 1927, had reported in Science that X-rays induce genetic mutations in Drosophila (Muller,1927); for this work, he received the Nobel Prize in 1946. Shortly thereafter, Muller and others showed that the frequency of X-ray induced mutations in Drosophila is approximately in direct linear proportion to the dose received (Muller,1928; Hanson and Heys,1929; Oliver,1930). Importantly, over the next 20 years, it was shown that X-rays given to Drosophila in relatively small doses over a long period produce the same genetic effects as an equal dose administered all at once (Uphoff and Stern,1949).

One sentence in Muller's 1927 article highlights that he was already thinking about the possibility that the “…effect of X-rays, in occasionally producing cancer, may also be associated with their action in producing mutations” (Muller,1927). Ten years later, he expanded on this idea, hypothesizing “that the carcinomas, sarcomas, and leukemias arising after irradiation represent mutations induced by [X-rays]” (Muller,1937).

By the mid-1950s, the Cold War was at its height. Admiral Lewis Strauss, Chairman of the Atomic Energy Commission, assured Americans in March 1954 that atomic weapons tests result in “a small increase in background radiation in some localities within the continental United States…far below the levels which could be harmful in any way to human beings.” A.H. Sturtevant of Caltech, in his presidential address to the Pacific Division of the American Association for the Advancement of Science countered Strauss' statement by emphasizing that “there is, in fact, no clearly safe dosage—all high-energy radiation, even of low intensity and brief duration, must be considered as potentially dangerous to the exposed individual… I regret that an official in a position of such responsibility should have stated that there is no biological hazard from low doses of high-energy radiation” (Sturtevant,1954).

Ed's entrée into the area of ionizing radiation and cancer was motivated in part by the discovery “in a discussion of nuclear weapons testing at… [a Caltech] faculty lunch table—that some of the physicists present assumed that a threshold would be operating” (Lewis,2004). His decision in late 1955 to “see what was known” culminated in his 1957 article in Science on “Leukemia and Ionizing Radiation” (Lewis,1957).

George W. Beadle (Fig. 25), then Chair of the Biology Division at Caltech, was an advisor to the Atomic Energy Commission and requested that Ed be given access to the data on leukemia cases among Japanese atomic bomb survivors. In the report, Ed plots the distance from the hypocenter at Nagasaki and Hiroshima and the dose received in rems (Fig. 26). He then derives the first risk estimates of leukemia incidence among the survivors in relation to estimated doses of ionizing radiation: for the atomic bomb survivors, Ed's best estimate of absolute risk of leukemia was two cases per million persons per rem per year. This was by no means a high risk, but it was not, as Strauss had claimed, negligible.

Figure 25.

G.W. Beadle (left), A.H. Sturtevant (center), and Ed in the late 1950s during the period that Ed's 1957 article in Science on leukemia and ionizing radiation was creating scientific controversy and politically motivated public attacks on Ed. Courtesy of the California Institute of Technology Archives.

Figure 26.

To begin his analyses of the relationship between radiation dose and leukemia induction in Japanese atomic bomb survivors, Ed plotted the distance in meters from the hypocenter at Hiroshima and Nagasaki against the dose of ionizing radiation received in rems, the unit in use at that time. Reprinted with permission from Lewis (1957), copyright AAAS.

Ed goes on to examine data on other groups that had been exposed to low doses of ionizing radiation: arthritic patients, patients who have been irradiated for thymic enlargement, and radiologists. In each case, his calculated risk estimate was one or two cases of leukemia per million persons per rem per year. Thus, Ed was led to conclude that “a striking feature of the Japanese data…is that the incidence of leukemia in…the zone with a calculated average ‘air’ dose of 50 rem—is significantly higher than in…the ‘control’ zone… Moreover, chronic irradiation at a relatively low dose rate (perhaps 0.1 rad per day or less) appears to induce leukemia in radiologists at a rate per rad which is comparable to that observed for the Japanese survivors…This finding also fails to support the concept of a threshold dose below which leukemia will not develop” (Lewis,1957).

In the last part of the report, Ed turns to radioactive fallout, which the Atomic Energy Commission (AEC) and the National Commission on Radiation Protection (NCRP) had assumed would only induce bone cancer from the Strontium-90, which concentrates in skeletal tissue. Furthermore, they had assumed that a dose of 2,000 rads would be needed before bone cancer would be expected. Ed points out in the article that Strontium-90 could well result in an increased risk of leukemia by irradiating the bone marrow. Using his earlier estimate of one in a million per individual per rad to bone marrow per year, he calculated a 5 to10% expected increase in leukemia incidence in the United States from a maintained level of one tenth the so-called maximum permissible concentration (MPC) of Strontium-90 that had been recommended by the NCRP.

It is important to emphasize that Ed's goal in conducting research into the somatic effects of radiation was not to stop nuclear weapons tests; it was to provide policy makers with accurate scientific information upon which they could then base their decision about whether to continue with the atmospheric tests.

There were several political consequences of Ed's Science paper. The first actually predated publication of the paper because Harrison Brown, who was a geochemistry professor at Caltech, had shared Ed's draft manuscript with Albert Schweitzer, the 1952 Nobel Peace Prize Laureate. At that time, Schweitzer was drafting the “Declaration of Conscience,” which the Nobel Prize Committee released on 24 April 1957. In that declaration, Schweitzer used Ed's results to warn that radiation from accumulated radioactive elements from nuclear fallout causes “mainly serious blood diseases…since the radiation goes from the bone tissue to the bone marrow.”

In the issue of Science in which Ed's paper was published, the Editor, Graham DuShane, wrote an editorial called “Loaded Dice”: “Ed Lewis shows that there is a direct linear relationship between the dose of radiation and the occurrence of leukemia. Thanks to Lewis it is now possible to calculate—within narrow limits—how many deaths from leukemia will result in any population from any increase in fallout or other source of radiation. And for the individual, it is possible to calculate the probability of death from leukemia as a result of any particular dose of radiation…We are approaching the point at which it will be possible to make the phrase ‘calculated risk’ for radiation mean something a good deal more precise than the ‘best guess’. It is apparent that the atomic dice are loaded. The percentages are against us and we ought not to play unless we must assure other victories” (DuShane,1957).

Shortly after Ed's article appeared, he was publicly attacked on NBC's “Meet the Press” (May 26, 1957), by Admiral Strauss, who charged that “[Lewis] is not a specialist in leukemia…it is too early to say what acceptance his hypothesis will receive among scientists who are students of leukemia…his hypothesis should rather be used as an illustration of the kind and degree of risk which might conceivably be involved if a radiation dose as high as 10 roentgens were received.” Interviewer Martin Agronsky challenged Strauss: “You're saying he [Lewis] is wrong and you are right, sir?” Strauss responded: “No. No. I say I think the chances are that he is wrong and I believe the chances are that the advisory commission on biology and medicine of the Atomic Energy Commission is right” (Lipshitz,2004).

Ed was called to give testimony before the US Congressional Joint Committee on Atomic Energy a couple of weeks later. His testimony was carefully phrased: “I think that one thousandth of a Roentgen is a conservative rate per year [for fallout] that one can estimate…if we reach this level and maintain it continuously, that leads to 10 as the number of deaths from leukemia per year from fallout sources” (Special Subcommittee on Radiation,1957). Not mentioned by Ed, who was careful to avoid hyperbole, was that—using the same estimates—if the so-called “safe” dose for the public of 0.1 MPC, which is 1.5 rads, had been reached, there would have been 15,000 extra leukemia deaths per year from fallout sources. This ran counter to what Admiral Strauss had claimed. Life magazine thought the topic of fallout danger to be important enough to run an article that included a picture of Ed in front of a black board that was covered in calculations (Fig. 27).

Figure 27.

Life magazine published an article on radioactive fallout dangers on June 10, 1957, which included a photograph of Ed in his Caltech office. The caption stated: “Warning of danger was sounded by Dr. E.B. Lewis of Caltech. In an article in Science he proved that there is a direct relationship between radiation and leukemia. He predicts a 5 to 10% increase in leukemia if strontium-90 level in humans reaches a figure which the AEC still considers harmless.” Reprinted courtesy of Life magazine.

Ed revisited the effects of low doses of ionizing radiation in several publications over the next 10 to 15 years. Richard Doll, a British epidemiologist, had in 1957 shown that chronic lymphatic leukemia (CLL) is not elevated in irradiated arthritic patients. In 1963, Ed obtained a small grant from the American Cancer Society, which enabled him to obtain and then carry out a very detailed analysis of 425 death certificates for American Board certified radiologists who had died at ages 35 through 74 from 1948 to 1961. He showed that their death rate was higher than that of the average US white male. Very significantly, none of the deaths was from CLL, enabling Ed to exclude the possibility that differential diagnosis was the cause of the excess deaths (Lewis,1963b). That study was also the first to show that low doses of ionizing radiation induce multiple myeloma. Ed went on in 1970 to examine the mortality of dermatologists and neurologists—who had been using X-ray machines in their offices for several decades—and found that there was a statistically significant increase in their mortality from leukemia (Lewis,1970).

So, almost 50 years after publication of his landmark article, was Ed right? The answer is an unequivocal “yes.” Reports from the National Research Council on the Biological Effects of Ionizing Radiation have confirmed his original estimates (BEIR I,1972; BEIR V,1990). Indeed, if anything, the estimates of risk have increased (Fig. 28).

Figure 28.

History has supported Ed's conclusions regarding the effects of low doses of ionizing radiation in cancer induction. Ed concluded in 1957 that leukemia is induced at a rate of 1.0 to 2.0 cases per million persons per rad (or rem) per year. Listed are recent estimates that range from 1.0 to 3.4 cases per million persons per rad (or rem) per year. Data extracted from BEIR I (1972) and BEIR V (1990).

I end my overview of Ed's science with another of his radiation effects article, this one published in 1959 (Lewis,1959). It specifically focused on radioactive fallout from the atmospheric weapons tests in Nevada. The AEC had claimed that the fallout from those tests produced radiation doses far below background levels. Ed pointed out in his paper the special hazard to the thyroid of infants and children from the 131I that was released from those explosions. His calculations showed that the radiation dose received was in fact one or two times—not far below—background. He went on to discuss how “the level of radioiodine in cow's milk may well be the most important index of short-term environmental contamination with fission products whether from weapons tests or other sources” (Lewis,1959). In 1986, the meltdown of the Chernobyl nuclear reactor in the Ukraine released large amounts of radioiodine into the atmosphere of Northern Europe, particularly Belarus. A study 12 years later showed a significant increase in thyroid cancer among individuals who were infants or children at the time of accident and were likely to have ingested contaminated cow's milk (Astakhova, 1998).


I do not want to leave the impression that Ed is in any way a stodgy, boring individual. To the contrary, he is energetic and full of fun. One of my favorite memories is from a surprise party thrown in honor of my naturalization as a US citizen in 1991. The party was held on the third floor of Kerckhoff Biological Laboratories, outside our labs—which were across the hall from each other (Fig. 29). Ed enjoys dressing up in even more innovative costumes, which for many years he created from scratch for the annual Biology Division Halloween party. These were usually modeled after paintings by his favorite artist, René Magritte (Fig. 30).

Figure 29.

Ed at a surprise party in honor of the author's naturalization as a US citizen in May 1991. The party was held in the hallway of the third floor of Kerckhoff Biological Laboratories at Caltech. Upper left: Ed, the author (cutting the cake), Joanne Topol (then a postdoctoral fellow with the author and subsequently a research fellow with Ed), Elliot Meyerowitz (currently Chair of the Biology Division at Caltech). Lower left: the author, Susan Celniker (then a research associate with Ed and currently codirector of the Drosophila Genome Sequencing Project at the Lawrence Berkeley National Lab), Ed, Elliot Meyerowitz. Right: Ed. [Color figure can be viewed in the online issue, which is available at].

Figure 30.

Ed in two of his Halloween costumes from the late 1980s. Upper: After “Le thérapeute (The Therapist)” by René Magritte. Lower: After “Le Mouvement Perpetuel (Perpetual Motion)” by René Magritte. Reprinted from Lipshitz (2004). [Color figure can be viewed in the online issue, which is available at].

Ed and Pam are always together at conferences and on other occasions, fondly being referred to by all present almost as a single item: “edandpam.” One such occasion, documented in one of Ed's favorite photographs, is of Pam meeting Hilary Clinton at the White House in 1995, where they had been invited to celebrate the announced award of the Nobel Prize (Fig. 31).

Figure 31.

Ed watches with pleasure as Pam (left) meets Hilary Clinton in 1995 at the White House celebration in honor of the American recipients of that year's Nobel Prizes. [Color figure can be viewed in the online issue, which is available at].

Ed is a consummate biologist. In his laboratory and home, aquaria are always present, in which he breeds octopuses, clown fish, and other marine animals. The pleasure he and Pam attain as they periodically raise the offspring of their pair of desert tortoises is palpable.

Ed exhibits a rare combination of intellectual rigor and iconoclasm that is coupled with remarkable personal and scientific integrity and humility. He is kind, gracious, and generous in both his personal and his scientific life. In his science, he has continued the tradition of sharing data and materials that was begun by Morgan and his coworkers almost a century ago. On a personal level, for nearly 20 years, he has served as a wonderful colleague, mentor and friend. Thanks Ed!


Ed died of cancer on July 21, 2004. At 86, he was active in the lab until shortly before his death. Figure 32 shows the last page of his lab notebook with details of genetic cross number 53,446. Because Ed began numbering his crosses as a graduate student in 1941 he evidently averaged over 900 crosses a year during the following 6 decades. He is sorely missed as a role model, colleague, and friend to geneticists worldwide.

Figure 32.

The last page of Ed's lab notebook showing cross number 53,446. Ed's final cross ended almost seven decades in Drosophila genetics. Photograph courtesy of Elliot Meyerowitz (Caltech).