Norman Harold Horowitz, 1915–2005
Robert L. Metzenberg

Anecdotal, Historical and Critical Commentaries on Genetics

Edited by James F. Crow and William F. Dove

AS a new graduate student of Herschel Mitchell at Caltech in 1951, I soon had occasion to visit Horowitz's office. I immediately noticed a newspaper clipping taped to the door. “It is a Privilege to be Living in the Same Century as Horowitz,” gushed the title. The article was, as it turned out, about a recent concert by Vladimir Horowitz, but it prepared me for the wry Horowitz sense of humor, always delivered deadpan. Later, there was Thanksgiving dinner at his house with his wife, Pearl, his son and daughter, and his mother. Norm sat down to the piano and we all sang old chestnuts. Vladimir he was not. But full of food and drink as we were and enthralled with our own wine-enhanced voices, it was a perfect evening. I left knowing Norm Horowitz and I had begun a lifelong friendship.

My full appreciation of Norm Horowitz as a scientist developed more slowly. I had especially admired the ground-breaking article of Srb and Horowitz (1944). They had shown that a metabolic pathway, leading to arginine from precursors that included ornithine and citrulline, could be understood as a linear series of enzymatically catalyzed steps, each of which was dependent on the intactness of a single gene. To me, newly versed in chemistry and physics and almost completely innocent of any knowledge of biology, this was a very appealing model, one that seemed almost intuitively obvious. How could anyone doubt it? But many did doubt it, mostly on the basis of the notion that it was too simple to be true. More about that later. A more serious criticism of the model was raised by Max Delbrück. His argument was that the great majority of enzymes or other proteins might be indispensable for life because they were needed in themselves or because their products were not present in the commonly used “rich” media or, if present, could not be transported into the cell. Any mutation that altered two or more proteins was very likely to alter an indispensable one and hence to be unrecoverable. The sample of mutants examined would in consequence be highly biased in favor of ones that altered only one protein. Horowitz and Leupold (1951) performed an ingenious test of Delbrück's argument. It involved isolating a rather large sample of temperature-conditional mutants of Neurospora and Escherichia coli. These strains grew on minimal medium at the lower of two temperatures (“permissive”), but not at an elevated (“restrictive”) temperature, which would nevertheless support the growth of the wild-type parent. It was assumed that mutations in genes encoding dispensable and indispensable proteins would occur in about the same proportions in these temperature-conditional mutants as in temperature-independent mutants. The question, then, was, “What fraction of temperature-conditional mutants are also unable to grow at the restrictive temperature if they are offered a medium containing all the common vitamins, amino acids, and purines and pyrimidines?” The answer turned out to be: only about one-half for the Neurospora mutants and one-fourth for the E. coli. This was not a large-enough fraction to create the degree of bias necessary to support Delbrück's criticism. Delbrück himself was satisfied and apparently considered the matter closed.

Much of the rest of the biology community, however, remained skeptical until overwhelming evidence from molecular genetics and biochemistry accumulated during the 1950s and 1960s in favor of the one gene-one enzyme model, by this time modified to the “one gene-one polypeptide” or “one cistron-one polypeptide” model. See also Yanofsky (2005) and Hickman and Cairns (2003). An additional early source of skepticism was traceable to an all-but-forgotten controversy. Some brilliantly conceived but ultimately misleading experiments led to the belief, widespread at the time, that proteins were synthesized by stitching together small preformed peptides, not by processive addition of individual amino acids. In their 1951 article, Horowitz and Leupold pointed out that the peptide-as-precursor model of protein synthesis was inconsistent with the one gene-one enzyme model because failure to make any given small peptide would almost certainly lead to failure to make several different proteins. The fact that one gene-one enzyme implied processive synthesis of proteins did not go unnoticed by advocates of the peptide-as-precursor hypothesis and added to the ranks of doubters.

When I was a beginning graduate student, I understood nothing of the upheavals in biological thinking demanded by the “one gene-one enzyme” hypothesis, nor how widely this hypothesis was rejected by the geneticists of that time. Neither did I sense the stubborn courage required for Horowitz to espouse and defend this initially unpopular idea and see it through to general acceptance. Chemists, including biochemists, are, by reflex, reductive thinkers. Their gold standard is the study of individual reactions involving the purest reactants that can be obtained, determination of the standard free energy and entropy of a reaction, the mechanism of reaction, and the measurement of kinetic properties such as the energy of activation. Geneticists in the first half of the 20th century, by contrast, were far less unanimous about reductive thinking. While many of them sought to delimit the boundaries of individual genes and to understand the phenotypes generated by mutation of individual genes, many others doubted that genes, as definable entities, even existed. They rather mystically attributed the phenotypes of mutants to altered properties of entire chromosomes. Even among geneticists who believed in the existence of individual genes, there was often a vague, holistic feeling that no simple causative relationship could exist between genes and proteins.

George W. Beadle (1966), rereading the volume that issued from the storied 1951 Cold Spring Harbor meeting, recorded his impression that believers in the one gene-one enzyme model “could be counted on the fingers of one hand, with a couple of fingers left over” (Horowitz 1996, p. 3). This appears to be a bit of an exaggeration, but if he had said “counted on the fingers of both hands,” it would probably have been accurate. The foreword to the 1951 volume by Miroslav Demerec, the keynote presentation by Richard B. Goldschmidt (1951), and its postpresentation comment by J. Herbert Taylor (1951) all reflect a strong undercurrent of skepticism about the very existence of genes as we know them today. In a 1995 Perspectives, Nathaniel Comfort, a historian of science, gave the impression that in 1951, the one gene-one enzyme model was already the canonical view among geneticists and that those who challenged it were fascinating mavericks and intellectual pioneers. Horowitz (1996) forcefully rebutted this view. He believed, as did Einstein, that scientific reality exists “out there” (Horowitz 1995), and he was deeply disturbed by postmodern historians of science who, with varying degrees of certitude, continue to argue that scientific findings are cultural artifacts. Horowitz's 1996 Perspectives article on Comfort's (1995) Perspectives article exemplifies his rejection of what he saw as an attempt to rewrite the history of science.

The 1944 Srb-Horowitz article soon budded off a short theoretical article on the evolutionary origin of biochemical pathways (Horowitz 1945). His theory was based on the idea, introduced by A. I. Oparin in 1924 and elaborated further in 1938, that life had originated in what J. B. S. Haldane (1929) called “hot dilute soup.” Later, Urey and Miller, Oró and Orgel, and others, showed that such a “dilute soup” could actually arise from simple nitrogen and carbon sources that would plausibly have been present in the early terrestrial atmosphere. Horowitz envisioned the evolution of a hypothetical pathway, which I will reformulate for present purposes asMathwhere A, B, C, D, and E were all present in the prebiotic soup, but only A was a necessary building block for the emergent “ur-biote.” The Greek letters represent pathway enzymes that catalyze the reactions in modern organisms. Once the ur-biote started propagating itself, the soup would soon become depleted of A and become limiting for life. If a spontaneous mutant that had acquired a gene encoding enzyme α appeared, it would allow the still-plentiful B to become a source of A. This mutant would soon become the regnant ur-biote, until it was out-competed by a descendant strain that had also acquired a gene encoding enzyme β, etc.: hence the Horowitz model of “retro-evolution.” It is still the best model going, and yet it remains untested and, at first glance, untestable. Note, however, that it casts giant shadows forward into the present. Horowitz instinctively chose a model in which natural selection could be called upon for the driving force, however slow it might be, and rejected the idea that E “knew” it had to find its way to A. Under the Horowitz hypothesis, the temporal order in which the enzymes would come into existence would therefore be first α and then β, γ, and δ.

Yet a proponent of “intelligent design” would presumably predict that historically the enzymes would show up in the reverse order, or at best, simultaneously, because the “designer” would envision the problem as one of converting E to A. It would be hard to find a sharper contrast of predictions between the doctrine of Darwin and the doctrine of “intelligent design.” Unfortunately, there is no obvious way to attach dates to the appearance of α, β, γ, and δ. More recently, however, Adam Wilkins (1995) has proposed a clever (and possibly testable!) hypothesis in which the Horowitz model of retro-evolution could explain the origin of cascades of sex-determining genes, most notably in the nematode Caenorhabditis elegans. This sort of argument seems potentially capable of explaining the origin of other cascades as well, for example, the sequence of events in blood clotting or in signal transduction.

Given Horowitz's interest in the origin of life, it was perhaps inevitable that he would become involved in planetary science, and beginning in 1964, this became his primary activity. Initially, he was responsible for screening externally generated proposals for experiments to be given “face time” on earth orbiters. Many of these proposals involved the triumph of optimism over experience in predicting that some well-understood process, such as hydrolysis of a chromogenic ester, would proceed differently in a zero-gravity field than on the earth's surface. Horowitz once remarked to me (but not to the proposers) that it would be much cheaper to heave a Beckman spectrophotometer into an empty elevator shaft and have it radio back readings for a few seconds on the way to its final splatter.

I think that this period, in which Horowitz had to deal with scientists whose defense was always “Show me!,” was a daily exercise in patience, which was soon to serve him well. Over the next 11 years, he directed the Jet Propulsion Lab's bioscience work on the Mariner and Viking missions. He was largely responsible for the design of the experiments carried out by the Viking landers to test the possibility of current or past life on the surface of Mars. The results from incubating Martian soil with complex media, and from pyrolysis of soil followed by gas chromatography and mass spectrometry of the pyrolysis products, convinced almost all scientists that the surface of Mars is lifeless and that if living organisms ever thrived there, they have left no credible biochemical footprints in the form of organic compounds. Spectroscopic analysis of the Martian atmosphere from the Mariner and Viking orbiters and from earth-based labs confirmed that there could be no liquid water on the surface of Mars, and hence no life as we know it.

This was more of a philosophical jolt for many people than might be guessed today. Since the Enlightenment, scientists as distinguished as Helmholtz, Kelvin, and Arrhenius had believed that life was widely dispersed in the universe and would certainly be found on the other planets of our solar system. As information on the planets accumulated, it became clear that only Mars was earth-like enough to have any possibility of supporting life. As Horowitz (1986) pointed out in his book To Utopia and Back: The Search for Life in the Solar System, Mars and the presumed intelligent beings who lived there, had engaged the attention of an amateur astronomer, Percival Lowell. Soon this became an obsession with Lowell, rather than an object for scientific study. In his enthusiasm, Lowell carried with him a large body of scientific and lay opinion. Despite a brief panic in 1938 caused by people tuning in late to a radio drama in which our Earth was invaded by technologically advanced but violent Martian conquistadores, many people found it comforting to think that we were not alone in the solar system. The mythology of a Mars populated by intelligent beings, or even by bacteria, should have disappeared abruptly with the first data from the Viking landers, if not before. However, it was kept simmering for years by excited but misleading statements to the press and to the U. S. Senate, particularly by three scientists who surely knew better; it seems clear that this was a media-savvy attempt to generate additional funding for the manned space program.

In a 1987 letter to Science, Horowitz, truth teller that he was, roundly criticized this trio for ignoring the overwhelming evidence against life on Mars. This cannot have endeared him to NASA. He also made the point that whenever the manned space program, which he considered to be a hugely expensive stunt rather than true exploration, had gone head to head against scientific content in NASA budgetary skirmishes, science had always lost—a point well worth considering in the light of today's seemingly unstoppable drumbeat for manned exploration of Mars.

A central tenet of Horowitz's philosophy was his devotion to reductionist thinking. In Horowitz's 1986 book, he noted that from the time of Aristotle to the late Renaissance, very little happened in biology, nor could it happen before the idea of spontaneous generation was discredited. Until the experiments of philosophical reductionists like Francesco Redi and Lazzaro Spallanzani, educated, intelligent people still believed that rotting meat spontaneously produced fly larvae and maggots and that mice were generated de novo in flour bins plugged with dirty laundry. In a 1994 book review, highly critical of a 1993 book by Lily E. Kay, Horowitz wrote:Kay condemns the idea that life can be explained by “upward causation.” She is an antireductionist—a position sometimes adopted by those who dislike what they perceive to be the direction of modern genetics. It must be understood that antireductionism is not a scientific position, but a political one, just as Lysenkoism was before it. Like Lysenkoism, which also claimed to be scientific, it is actually antiscience. Its adherents do not perceive that science must be reductionist—that natural systems can be said to be understood only after they have been reduced to and reassembled from their components [italics in original]. On the contrary, Kay claims that a variety of nonreductive biologies exist, any of which would have served better than molecular biology as a major theme for understanding life. She says: “the abundance of rigorous quantitative antireductionist models that have developed during the second half of the twentieth century attests to the limits of the mechanistic and physicochemical approach for solving problems of biological organization.” Needless to say, neither Kay nor the cited reference describes even one such model.A decade later, we have a wealth of knowledge in the form of complete genome sequences of many organisms reduced to their component genes. Horowitz would be pleased at the rate at which new knowledge is accumulating. We are indeed studying “upward causation” with tools such as prediction of amino acid sequences of proteins and functional motifs of those proteins, microarray analysis to analyze expression patterns, and two-hybrid analysis to detect protein interactions. Inferring the organization of entire pathways is still in its infancy. While all this is occurring, “downward causation” observations and experiments will doubtless continue, as they should. As more sophisticated answers on pathways and cellular organization are developed—as is already happening—it will be a job for historians of science to sort out the relative contributions of reductionist and holistic thinking. One hopes that they will approach their task with the tough-minded honesty for which Horowitz has set the example.

Horowitz will surely be remembered for his important contributions to our understanding of genes and proteins, for his insights into evolution, and for his leadership of the bioscience activities of the Mariner and Viking space ventures. The parts of this article memorializing those contributions are, in a sense, superfluous. What is at more risk of oblivion is the Norm Horowitz as conscience for our time: the blunt but ever-civil truth teller. As the old saw goes, “Always tell the truth, but don't always be telling it.” Somehow, Norm always managed to tell the truth without ever becoming a scold. There can never be enough of such people, and his legacy must be kept alive.

After more than 5 decades, I think again of that clipping on his door. Yes, it is a privilege to have lived in the same century as Horowitz.