Abstract

John W. Drake died 02-02-2020, a mathematical palindrome, which he would have enjoyed, given his love of “word play and logic,” as stated in his obituary and echoed by his family, friends, students, and colleagues. Many aspects of Jan’s career have been reviewed previously, including his early years as a Caltech graduate student, and when he was editor-in-chief, with the devoted assistance of his wife Pam, of this journal for 15 impactful years. During his editorship, he raised the profile of GENETICS as the flagship journal of the Genetics Society of America and inspired and contributed to the creation of the Perspectives column, coedited by Jim Crow and William Dove. At the same time, Jan was building from scratch the Laboratory of Molecular Genetics on the newly established Research Triangle Park campus of the National Institute of Environmental Health Science, which he headed for 30 years. This commentary offers a unique perspective on Jan’s legacy; we showcase Jan’s 1969 benchmark discovery of antimutagenic T4 DNA polymerases and the research by three generations (and counting) of scientists whose research stems from that groundbreaking discovery. This is followed by a brief discussion of Jan’s passion: his overriding interest in analyzing mutation rates across species. Several anecdotal stories are included to bring alive one of Jan’s favorite phrases, “to think like a geneticist.” We feature Jan’s genetical approach to mutation studies, along with the biochemistry of DNA polymerase function, our area of expertise. But in the end, we acknowledge, as Jan did, that genetics, also known as in vivo biochemistry, prevails.

A select group of scientific heavyweights have had sustained influence through both vision and happenstance. Jan undoubtedly punched the ticket on “the vision thing” through his selfless devotion to making GENETICS a standout journal, and also by founding the Laboratory of Molecular Genetics at the National Institute of Environmental Health Science (NIEHS). As editor-in-chief of GENETICS, Jan faced many challenges, which included dealing with the “reviewer from hell,” Jan’s exact words to describe a reviewer who agreed to review a paper but then did not. Jan described this and many additional challenges that he and his wife Pam faced during their 15-year service to GENETICS in his autobiographical Perspective, “The Mom and Pop Editorial Shop” (Drake 1998).

Jan was recruited to the NIEHS in 1977. When its new campus in Research Triangle Park (RTP), NC, was completed (1981/1982), Jan headed up the new Laboratory of Molecular Genetics, which he staffed with several talented, independent, young scientists. Jan brought Lynn Ripley with him from the University of Illinois, where she had developed the first assay to measure transversion mutagenesis in phage T4 (Ripley and Drake 1972; Ripley 1975). And luckily for the NIEHS, Jan’s move to RTP coincided with the ready availability of two Larry Loeb trainees, Tom Kunkel and Roel Schaaper. Jan’s vision for the laboratory came to fruition with these earliest recruits, and many who came later with world-class reputations. Nancy Nossal and one of us, Myron Goodman, were asked to come for a site visit during the early days of the NIEHS and to provide in situ verbal and written feedback of Jan’s laboratory, and also of Burke Judd’s new Drosophila laboratory. Nancy and I were, quite frankly, surprised and gratified to see that the NIEHS leadership had the broad vision to support researchers who saw the value of using model organisms—fruit flies for Burke Judd and phage T4 for Jan—which Jan acknowledged as a fortuitous choice in his autobiographical Commentary (Drake 2006). We were also honored, as biochemists, to be asked to contribute to Jan’s vision of developing cutting edge genetic research at RTP. Our visit foreshadowed our long collegial interactions with Jan.

Jan served as head of the NIEHS Laboratory of Molecular Genetics for ∼30 years, stepping down in 2012. Goodman, having served two terms on the NIEHS Board of Scientific Councilors, was able to see first-hand that Jan’s program was successful in providing a molecular mechanistic component that added strength and breadth to the applications-oriented environmental role envisioned for the NIEHS. It was obvious from their yearly presentations that Jan had hired a cohort of creative young scientists who were afforded the opportunity to flourish. For the past 38 years, Tom and Roel have continued to occupy their same respective laboratories on the third floor of the Rall Building. A symposium “The Fidelity of DNA Replication: From Basic Mechanism to Disease,” was held August 29–30, 2019, to commemorate Tom Kunkel’s 70th birthday. It would have been a perfect symposium except for the unintended absence of Jan, who was not feeling well enough to attend.

While the above contributions distinguish Jan as a superb leader and administrator, he is remembered for much more. We focus the lions’ share of our comments here on what we view as Jan’s principal sustaining contribution to science: his 1969 Nature paper in which he reported the discovery of T4 DNA polymerase mutants that exhibited antimutator phenotypes (Drake et al. 1969). Jan’s paper, along with the 1965 paper by Joseph Speyer that reported T4 DNA polymerase mutants possessing mutator phenotypes (Speyer 1965), provided a critical cornerstone for the fields of DNA replication fidelity, mutation, and repair.

Bacteriophage T4 Genetic Studies Reveal DNA Polymerase Mutators and Antimutators

Jan discovered antimutator T4 DNA polymerases when he was at the University of Illinois, where he had set up a phage T4 genetics laboratory. T4 was a powerful model organism in the 1960s that continues to have impact in DNA replication studies, especially since the determination of the structure of the DNA polymerase of the closely related phage RB69 (Wang et al. 1997). Many important discoveries with phage T4 are described in two books—T4 “bibles”—published by the American Society for Microbiology. The first edition (Mathews et al. 1983) was followed by a second edition in 1994 with Jan as coeditor along with Jim Karam, Elizabeth (Betty) Kutter et al. (Karam et al. 1994). The influential Evergreen International Phage Meetings, several of which featured T4 DNA replication, have been organized by Betty Kutter since 1975.

The T4 DNA polymerase story began in 1963 when Dick Epstein, Bob Edgar, and co-workers (Epstein et al. 1963) identified T4 essential genes, including gene 43 encoding the DNA polymerase, (De Waard et al. 1965). Joe Speyer and colleagues (Speyer 1965) observed that a few gene 43 temperature-sensitive alleles (ts) appeared to encode error-prone DNA polymerases; some ts mutants, including tsL56, displayed considerable plaque variability (plaques are clearings in a lawn of bacteria that are produced by multiple rounds of phage infection that start with a single, phage-infected bacterium). Normally, T4 plaques are uniform, but some of the ts DNA polymerase variants that Speyer examined produced a wide range of plaque morphologies at the permissive temperature. This would be expected if the starting lysate contained many plaque morphology mutants and/or if different plaque morphology mutations emerged during the multiple rounds of infection and lysis required to produce a plaque. The variation in plaque morphologies suggested a high number of mutations, a mutator phenotype that Speyer interpreted to mean that the mutant DNA polymerases had reduced base selectivity (Speyer et al. 1966). However, there are many ways to make an error-prone DNA polymerase, as will be seen as this story unfolds. While mutant DNA polymerases and replicases with low replication fidelity have been indentified in many organisms, this was breaking news in 1966: T4 DNA polymerase was fingered as a major determinant of replication fidelity. This observation pushed Jan, as he told this story, to wonder if there were mutator DNA polymerases that specifically made transversion mutations. But while screening for this phenotype, Elizabeth Allen discovered something more interesting: antimutator DNA polymerases (Drake and Allen 1968). The new concept of “antimutation” was born.

Since mutations are rare and antimutator DNA polymerases produce even fewer mutations, a sensitive assay to determine mutation frequencies was needed. Jan took advantage of Seymour Benzer’s observation that mutations that inactivate the T4 rII genes prevent T4 phage from replicating in bacterial hosts that carry an integrated copy of the phage lambda genome (a lysogen), and that this restriction is exceptionally tight (Benzer 1961). It is just what was needed for a sensitive reversion assay: a single rII revertant could be detected in a population of 100 million or more phage. This assay enabled the quantitation of mutation frequencies; development of assays to quantitatively determine mutation frequencies is a feature of many of Jan’s publications.

Jan and his group determined reversion frequencies for many ts T4 DNA polymerase mutants from the Epstein and Edgar collection (Drake and Allen 1968) using rII mutants that he had isolated (Drake 1963). They observed that the mutator T4 DNA polymerases identified by Speyer increased reversion frequencies, especially for two rII mutants (rUV199 and rUV183), but that two other mutant T4 DNA polymerase strains (tsL141 and tsL42) dramatically reduced reversion frequencies of rUV199 and rUV183 by as much as ∼100-fold (Drake et al. 1969). These are the classical T4 antimutator DNA polymerases. An important accompanying observation was that the antimutator DNA polymerases reduced the reversion frequencies of only some rII mutants.

Why was there a large reduction in reversion frequency of some rII mutations and not others? This work was done well before DNA sequencing had been invented. But Jan and his research group, following in the footsteps of Ernst Freese (Freese and Freese 1967), determined the reversion pathways for several rII mutants by using base analog mutagens, chemicals that damage DNA, and ingenuity (Drake and Greening 1970). To summarize a lot of clever work, phage T4 antimutator DNA polymerases suppress reversion of the rUV199 and rUV183 mutations via AT→GC base pair transitions at some sites, but not GC→AT transitions or transversions, as was later confirmed by DNA sequencing (Reha-Krantz 1995). Some transversion pathways were even increased by antimutator DNA polymerases (Ripley and Drake 1972). Thus, only certain mutation events, at hot spots, were reduced by antimutator DNA polymerases. Jan concluded that antimutators reduced only some errors of replication and some errors of incorporation (Drake and Greening 1970). Biochemists, with Jan’s assistance, then stepped in.

Revealing the Biochemical Basis for the Mutator and Antimutator Phenotypes of T4 Polymerases

DNA polymerase biochemists at that time were stymied by the problem of exonuclease activities that copurified with DNA polymerases. At the top of the list was the “pesky” 3′→5′ exonuclease activity that was present in Escherichia coli DNA polymerase I and the T4 DNA polymerase, as recounted in Bob Lehman’s must-read Reflections article (Lehman 2003). One of us, L.J.R.-K., remembers a story told by her Ph.D. supervisor, Maurice Bessman, about his postdoc days (1955–1958) in the Kornberg laboratory. Bessman related that it was always tense when the postdocs gathered each morning to decide who would have to tell Kornberg that the presumed contaminating exonuclease activity in E. coli DNA polymerase I preparations had still not been removed. It simply did not make sense that a DNA polymerase would have an activity that degraded newly synthesized DNA. Yet, as time would tell, a 3′→5′ exonuclease is a bona fide, integral activity of both the E. coli and T4 DNA polymerases.

Maurice Bessman and Nancy Nossal carried out the first biochemical studies of antimutator T4 DNA polymerases. Jan welcomed input from biochemists, as demonstrated by a minisymposium that he organized circa 1975 at the University of Illinois, which consisted of just two talks: one by Nancy Nossal and the second by Maurice Bessman. Jan was a no-nonsense eminence grise, like Dragnet’s Sgt. Joe Friday1: “all we want are the facts, ma’am.” Thus, he advertised the talks simply as “Her Work; His Work.” Nancy described her interactions with Jan in the special 1998 GENETICS issue, created to honor Jan, which ranged from his view of her “very” biochemical work, which she thought may have not been a compliment, to Jan reading bedtime stories to her children (Nossal 1998). Both Nancy and Maurice expressed their considerable gratitude to Jan for sharing his T4 gene 43 strain collection, and for encouraging their enzymological studies of DNA polymerase antimutators (Nossal 1998; M. Bessman, personal communication).

The biochemical studies were revealing. In late 1972, Bessman’s laboratory published a benchmark2 paper (Muzyczka et al. 1972) that reported that T4 DNA polymerase 3′→5′ exonuclease activity was lower in the mutator tsL56 DNA polymerase than observed for the wild-type T4 DNA polymerase, but was higher for the antimutator tsL141 and tsL42 DNA polymerases. Furthermore, the 3′→5′ exonuclease activity preferentially removed wrong nucleotides at the primer end; in other words, the antimutator DNA polymerases were really fast at removing the wrong nucleotide and the mutators were exceptionally slow. The T4 DNA polymerase studies corroborated studies of E. coli DNA polymerase I, in which the 3′→5′ exonuclease activity was discovered to be a proofreading activity that preferentially excised wrong nucleotides (Brutlag and Kornberg 1972). The T4 DNA polymerase studies were compelling because in vitro biochemical assays were interpreted in view of Drake’s and Speyer’s in vivo studies of mutator and antimutator T4 DNA polymerases. Thus, insights into DNA polymerase function came from joining genetic and biochemical approaches. More examples of this dual approach follow.

Bessman explained the lower and higher 3′→5′ exonuclease activities of mutator and antimutator DNA polymerases in terms of a ratio between nuclease and polymerase activities, which can be illustrated as a balance, like the scales of justice (Figure 1). The antimutator (AM in the following equation) phenotype was linked to increased exonuclease (N) proofreading activity relative to its DNA polymerase (P) synthesis activity, N/PAM > N/PWT. However, the opposite was observed for the mutator (M in the following equation) phenotype, a reduced N/P ratio (N/PM < N/PWT) (Reha-Krantz 2010).

Figure 1

The balance model to explain replication fidelity by mutator and antimutator DNA polymerases. The N/P ratio is the ratio of exonuclease and polymerase activities. While exonuclease (exo) activity is the same for the wild-type (wt) T4 DNA polymerase (pol) (A), and for antimutator (B) and some mutator DNA polymerases (D), the balance is shifted because of decreases in polymerase activity for the antimutators and increased synthesizing activity for the mutators. The weakened polymerase activity tips the balance toward increased proofreading (B), but some mutator DNA polymerases have increased polymerase activities that tip the balance away from proofreading activity, which allows incorporation of wrong nucleotides and extension of the mismatched primer ends as well as error-prone replication of damaged DNA (D). The mutator phenotype is also observed if exonuclease activity is reduced by mutations that encode amino acid substitutions that prevent Mg2+ or DNA binding in the exonuclease active site (C). But most importantly, proofreading activity is regulated in wild-type DNA polymerases so that DNA replication is not unduly impeded by excessive exonuclease activity, but reasonable accuracy is achieved in reasonable time (Drake 1993; Reha-Krantz 1998; Reha-Krantz 2010).

For wild-type T4 DNA polymerase, the N/P balance greatly favors polymerase activity, which protects the newly synthesized DNA from wanton degradation (Figure 1A); 3′→5′ exonuclease activity is called into action only when polymerase activity is reduced, for example if a wrong nucleotide is inserted at the primer end that cannot be extended. This is the classic proofreading model; however, polymerase activity is not reduced only by a mismatched primer terminus, but also by skewed dNTP pools (Clayton et al. 1979; Hopkins and Goodman 1979; Mathews 2015). Low dNTP pools slow replication and trigger proofreading while high concentrations support nucleotide incorporation and suppress proofreading. The effect of raised dNTP concentrations is so great that the suppression of 3′→5′ exonuclease activity by high concentrations of the next nucleotide, the so-called next nucleotide effect, can be used to identify proofreading DNA polymerases (Clayton et al. 1979; Fersht and Knill-Jones 1981; Kunkel et al. 1981). Support for these proposals came from experiments with mutator and antimutator DNA polymerases where high dNTP concentrations suppressed the antimutator phenotype of the tsL141 DNA polymerase, while low dNTP concentrations increased proofreading by the mutator tsL56 DNA polymerase. Furthermore, dNTP concentrations are also important for DNA polymerases that lack intrinsic proofreading activity. Mismatch extension rates by these error-prone polymerases are also increased at high concentrations of the “next” dNTP (Mendelman et al. 1990).

In a series of insightful biochemical studies, Nossal explained (Gillin and Nossal 1976; Spacciapoli and Nossal 1994; Nossal 1998) that the simultaneous observation of reduced polymerase activity coupled with high exonuclease activity of the classic antimutator DNA polymerase tsL141 (aka tsCB120, A737V) is due to a defect in strand displacement synthesis. The reduced processivity of the antimutator DNA polymerase, which often results in enzyme dissociation from the replication fork, creates an increased opportunity for the polymerase to rebind the DNA primer in the exonuclease active site and to excise the terminal nucleotide, even though the primer end is correct. Many correct nucleotides are removed during DNA replication by antimutator DNA polymerases! This idling reaction, in which correct nucleotides are inserted and then removed without forward advancement along the template strand, can be created in the laboratory by providing just a single nucleotide in reactions with proofreading DNA polymerases. The costly waste of dNTPs in DNA polymerase idling is driven home by the observation that the tsL141 (A737V) DNA polymerase is temperature-sensitive not because the protein is sensitive to high temperature, but because exonucleolytic proofreading is increased to an excessive level by higher temperature (Lo and Bessman 1976). We know from the Nossal experiments discussed above that the A737V DNA polymerase has reduced processivity. Thus, increased temperature tips the balance to increased proofreading (Figure 1B) by increasing enzyme dissociation and likely by also assisting strand separation at the primer end, which is a prerequisite to binding the primer end in the exonuclease active site.

Genetic studies support the Nossal hypothesis that antimutator DNA polymerases have more opportunity to proofread. If the antimutator A737V DNA polymerase has reduced processivity in vivo, as observed in vitro, then suppressors of the excessive proofreading are predicted to restore processivity. Indeed, such suppressor mutations were selected that restored processivity (Stocki et al. 1995; Nossal 1998). Interestingly, the suppressor mutations were clustered in four regions throughout the DNA polymerase; these mutations provide the means to probe how the T4 DNA polymerase maintains processivity during DNA replication.

An important application of the reduced processivity of antimutator DNA polymerases is in the detection of DNA damage. For example, the antimutator I417V DNA polymerase (Reha-Krantz and Nonay 1994) has difficulty replicating past an ethyl phosophotriester lesion in the DNA backbone (Tsujikawa et al. 2003). Wild-type DNA polymerases in general were thought to readily replicate DNA with alkyl backbone modifications, but stalling of antimutator T4 DNA polymerases observed at subtle lesions suggests that stalling is also a feature of wild-type DNA polymerases. This proposal is supported by a recent report that the phage T7 replisome pauses frequently during DNA replication, which results in backtracking and removal of correct nucleotides (Singh et al. 2020). Thus the high turnover of nucleotides, first observed for antimutator T4 DNA polymerases nearly 50 years ago (Muzyczka et al. 1972), is now an established aspect of DNA replication.

There are many potential problematic consequences caused by DNA polymerase pausing, in addition to the removal of correct nucleotides. Notably, fork stalling in difficult-to-replicate DNA sequences followed by replisome dissociation provides an opportunity for misalignment of the unbound primer strand with ensuing production of deletions, insertions, duplications, base substitutions, and frameshift mutations (Kunkel and Soni 1988), replication errors that are increased substantially by antimutator T4 DNA polymerases (Wang and Ripley 1998), as predicted because antimutator DNA polymerases with reduced processivity are more likely than wild-type DNA polymerases to dissociate following fork stalling. Ripley also demonstrated that DNA sequences with imperfect inverted repeats (quasi-palindromes) are predisposed to frameshifts and deletion and insertion mutations, many appearing simultaneously (Ripley 1982; Papanicolaou and Ripley 1989; Wang and Ripley 1998).

These findings have medical implications since drugs that slow replication, like the human immunodeficiency virus drug azidothymidine, promote mutations that arise in imperfect inverted repeats (Seier et al. 2012). The mutations appear to arise by fork stalling at short inverted repeats, followed by dissociation and then replacement of the replisome with translesion DNA polymerases DNA polymerase ζ and Rev1, and finally error-prone replication that produces a cluster of mutations (Northam et al. 2014). Polina Shcherbakova coined the term “defective replisome-induced mutagenesis” or DRIM to label this class of mutagenic events. Incidentally, Polina prospered for many years in Jan’s NIEHS as a postdoc in Tom Kunkel’s laboratory. Mutational clusters caused by transient hypermutation during replication were an interest of Jan’s, and were discussed in impressive breadth and depth in his 2007 paper “Too Many Mutants with Multiple Mutations” (Drake 2007). Furthermore, there is a cancer connection. The latest DNA sequencing methods were used to screen for the DNA sequences of mutational hot spots in many cancers (Alexandrov et al. 2020), which are the mutational signatures of replication and repair failure. Jan would have enjoyed pouring over this mutational data, and he would have been proud that one of his colleagues at the NIEHS, Dmitry Gordenin, is a leader of these studies.

Because of many potential negative consequences of too much proofreading, the process must be regulated to avoid an unacceptable level of mutations (Reha-Krantz 1998). Thus, the level of proofreading observed for wild-type DNA polymerases is a compromise that reduces some replication errors but not so much as to generate more errors than are avoided. Jan concluded that general antimutators are improbable (Drake 1993), an important insight.

In contrast to antimutator T4 DNA polymerases, the biochemical basis of the mutator phenotype of T4 mutant DNA polymerases is easier to understand. Biochemical studies show that reduced replication fidelity is observed by mutations that result in amino acid substitutions in the exonuclease active site that directly reduce 3′→5′ exonuclease activity (Figure 1C), by amino acid changes that reduce the specificity of nucleotide incorporation (Hershfield 1973; Nossal 1998), and by amino acid changes that reduce proofreading by reducing the ability of the polymerase to switch between synthetic and exonuclease activities (Figure 1D) (Stocki et al. 1995; Reha-Krantz 2010; Reha-Krantz et al. 2014).

Studies of mutator DNA polymerases have medical applications, especially in understanding origins of cancer. For example, mice homozygous for mutations decreasing the 3′-exonuclease of DNA polymerase δ suffer a variety of cancers, principally epithelial, and die much sooner than wild-type or heterozygous mice (Goldsby et al. 2002). DNA polymerase ε proofreading appears to play an even more important role in tumorigenesis (Albertson et al. 2009). Several mutations in the exonuclease domain of human DNA polymerases ε and δ have been implicated in cancer (for example see Barbari and Shcherbakova 2017). However, mutations that confer the antimutator phenotype are also part of the cancer mutator story, since antimutator mutations arise during the evolution of mutator cells to curb strong mutator activity, as demonstrated recently in yeast (Tracy et al. 2020).

A big breakthrough in understanding how T4 DNA polymerase achieves high replication fidelity was the determination of the structure of the phage RB69 DNA polymerase, a close relative of T4 DNA polymerase (Wang et al. 1997). This was the first structure of a family B DNA polymerase to be determined, and thus provides an important model for eukaryotic DNA polymerases. Nossal analyzed several ts and am T4 DNA polymerase mutants in view of the polymerase structure in her paper entitled “A New Look at Old Mutants” (Nossal 1998). Many puzzles about locations of amino acid substitutions encoded by the ts and am gene 43 mutants were resolved with the protein structure. Yes, a picture is worth a 1000 words, and more.

Technical problems have prevented the crystallization of T4 DNA polymerases for structural studies, but the RB69 phage does not have the genetic reporters that Jan developed to measure mutation rates in phage T4. Jan’s solution, along with Anna Bebenek, was to express the RB69 DNA polymerase from a plasmid in bacteria infected with a double amber T4 DNA polymerase mutant (Bebenek et al. 2001), making the RB69 DNA polymerase the only DNA polymerase available to replicate the T4 genome. Remarkably, T4 phage replication is supported by leaky expression of the RB69 DNA polymerase. Several genetic experiments with this RB69–T4 complementation system were carried out by Jan’s research team, as well as biochemical experiments by Kadyrov et al. (Kadyrov and Drake 2003; Kadyrov and Drake 2004), and in collaboration with Jim Karam and Bill Konigsberg (Bebenek et al. 2002). Despite the information learned from structural and biochemical studies of DNA polymerases, Jan concluded that fidelities of wild-type and mutant RB69 DNA polymerases determined in vitro only partially reflected their fidelities in vivo (Bebenek et al. 2002). For example, biochemical studies of engineered RB69 DNA polymerases showed that increasing the size of the nucleotide-binding pocket reduced nucleotide insertion fidelity, but that was countered by in vivo mutation studies (Trzemecka et al. 2010) that forced reconsideration of the biochemical studies (Xia et al. 2011). Hence, there is a real and continuing need for in vivo replication fidelity experiments.

Despite >50 years of genetic and biochemical studies of antimutator and mutator DNA polymerases, conundrums remain. The following is an appeal to researchers to follow Jan’s footsteps to solve remaining puzzles about DNA polymerase function. The rewards will be the discovery of new aspects of DNA polymerase function that maintain the fidelity of DNA replication, new insights into the evolution of mutation rates, and potentially revealing genes that are responsible for sporadic and inherited human disease.

One outstanding conundrum is the specificity of antimutator T4 DNA polymerases for strikingly reducing only some AT→GC mutations, but apparently not for GC→AT transitions or for several transversion mutations. Jan liked to cite a Salts and Ronen report that conversion of ochre (TAA) to opal (TGA) triplets varied 1000-fold across the rII genes, even though the neighboring base pairs were identical (Salts and Ronen 1971). This large variation in mutability in vivo is not observed in vitro since all mismatches are proofread in biochemical assays (Sinha 1987), but with increased efficiency in AT-rich DNA sequences and decreased efficiency in GC-rich sequences (Bloom et al. 1994). Furthermore, antimutator DNA polymerases not only decrease some AT→GC mutations, but they increase a variety of other types of mutations as revealed in forward mutation assays for resistance to acriflavine (Wang and Ripley 1998), and in screens for rI mutants (Bebenek et al. 1999). Thus, there is no net reduction in mutational load for DNA polymerase antimutator strains; antimutator DNA polymerases are not generally improved polymerases since the decrease in some mutations is offset by increases in other types of mutations.

Different models have been proposed to explain the specificity of proofreading by antimutator DNA polymerases (Reha-Krantz 1995; Schaaper 1998), but irrespective of the model it is clear that antimutator T4 DNA polymerases reduce mutations at hot spot sites that escape proofreading by the wild-type DNA polymerase. Thus, an important question is: “what is the biochemical basis for ‘hot spot’ mutation sites that are replicated more accurately by antimutator DNA polymerases?” We suggested >20 years ago that a remaining challenge in replication fidelity was to determine the origins of mutational hot and cold spots (Goodman and Fygenson 1998). Earlier, Jan raised the possibility of “wrong bases in the right places” that result from misalignment of the primer strand (Drake 1991) and other “horror” stories that include DNA damage. It is important to consider that DNA polymerase proofreading is not just a spell-checking activity; enhanced opportunity to proofread also deters translesion synthesis (Goodman et al. 1993), which was demonstrated for the antimutator I417V DNA polymerase discussed above (Tsujikawa et al. 2003).

A second outstanding puzzle is to determine the function of the N-terminal domain of the T4 and RB69 DNA polymerases. While functions for their exonuclease, polymerase, finger, and palm domains are known, does the N-terminal domain do anything of importance? Yes, since strong mutator and antimutator T4 DNA polymerase mutations reside there (Reha-Krantz 1988; Stocki et al. 1995; Reha-Krantz and Wong 1996; Li et al. 2010). The classic mutator tsL56 DNA polymerase has two amino acid substitutions in the N-terminal domain, A89T and D363N (Reha-Krantz 1989), not in the exonuclease domain as might be expected for a mutant DNA polymerase with apparent reduced exonuclease activity (Muzyczka et al. 1972). Another N-terminal domain substitution, L340P, was identified in a selection for T4 mutator DNA polymerases (Reha-Krantz and Bessman 1981)3. The L340P substitution in the N-terminal domain causes a strong mutator phenotype without reducing exonuclease activity (Reha-Krantz and Bessman 1981), yet the nearby R335C substitution results in a strong antimutator phenotype (Reha-Krantz and Wong 1996). These and other mutations in the N-terminal domain and elsewhere were proposed to affect a network of protein interactions that facilitate DNA polymerase translocation and processivity, which allows the polymerase to replicate DNA under a variety of conditions without invoking exonuclease action except when needed (Li et al. 2010). But this proposal needs to be tested.

Three Fidelity Conferences Play an Instrumental Role in Adding Theory, Structure, and Kinetics to Genetics and Biochemistry

Following the geneticists and biochemists, the theorists stepped into the replication fidelity field. The considerable excitement spurred by the initial T4 polymerase mutator/antimutator studies of 1969–1972 quickly led to the development of theories, one of Jan’s favorite activities. John Hopfield (Hopfield 1974) and Jacques Ninio (Ninio 1975) proposed DNA synthesis mechanisms that didn’t employ “brute force” 3′-exonuclease proofreading, but rather the presence of internal nonequilibrium DNA polymerase transition steps in the dNTP insertion pathway that facilitated the ejection of wrong nucleotides prior to their incorporation into DNA. While insertion specificity is important, David Galas and Elbert Branscomb developed an alternative model that explicitly included 3′-exonuclease proofreading to analyze tsL56 mutator, wild-type, and tsL141 antimutator data (Galas and Branscomb 1978). Heady stuff indeed, all of it inspired by the initial studies by Speyer and Drake.

A new and expanding field needs a scientific meeting. In 1977, Ninio, Hopfield, and Chuck Kurland organized a superb meeting at a French chateau in Plasir-Grignon, accompanied by a multistar menu and gourmet wine selection. The meeting brought together a congenial mix of geneticists, biochemists, and theoreticians in an incipient field, which jump-started collaborations that melded experiment with theory to address fidelity mechanisms of DNA and protein synthesis (Hopfield et al. 1976; Clayton et al. 1979). Roel Schaaper has fond memories of the Plasir-Grignon meeting. He told Goodman that as a beginning postdoc attendee, immediately upon entering the chateau he ran across three people at a table drinking wine and debating T4 biochemical genetics. The three turned out to be Goodman, Chris Mathews, and Jan Drake. Roel’s first sighting of Jan was a harbinger of good things to come as it was Jan with whom Roel had a long and successful professional and personal relationship.

Tom Kunkel and Roel Schaaper, from Jan’s Laboratory of Molecular Genetics, organized a second meeting in early fall 1989, “The Fidelity of DNA Synthesis: Structural and Mechanistic Perspectives,” beginning just 2 days after arrival of Hurricane Hugo, a spectacular category 4 storm that hit North Carolina’s shore. The meeting was to have taken place in Beaufort, located on North Carolina’s Outer Banks, a stone’s throw from Jan and Pam Drake’s vacation home. Tom and Roel wisely relocated the meeting to the safer, albeit less exotic, setting of Raleigh, NC. Maurice Bessman’s first slide introduced to the audience the obscure, yet à propos, Beaufort Hurricane Scale, with the category 4 Hugo equivalent to a Beaufort 12. Tom and Roel convened a follow-up (storm-free) meeting in Wrightsville Beach, NC, in 1995.

Structural and kinetic studies on base selection at polymerase active sites were just then getting started, and the meeting featured an animated “discussion” between Hugette Pelletier representing the Joe Kraut and Sam Wilson groups’ studies on eukaryotic DNA polymerase β (Pelletier et al. 1994), and Tom Steitz presenting the structure of the E. coli polymerase I Klenow fragment (Beese et al. 1993). Both vigorously contested structures were of course correct, just different. Jan’s influence on the melding of the biochemical basis of DNA replication fidelity with kinetic and structural studies is reflected in two Annual Reviews of Bio‐chemistry articles, one by H. Echols and M. F. Goodman that described the period between the Plasir-Grignon meeting and 2 years beyond the Beaufort meeting (Echols and Goodman 1991), and another by Tom with Kasia Bebenek (Kunkel and Bebenek 2000) that described the remarkable growth in the fidelity field during the 5-year aftermath of the Wrightsville Beach meeting, especially with publication of the structure of the RB69 DNA polymerase, as discussed above.

“Drake’s Law” on Mutation Rates

Maurice Fox wrote an insightful Perspective about Jan’s creative and persistent interest in mutation and mutation rates (Fox 1998), but readers are encouraged to also read Jan’s own autobiographical account (Drake 2007). Jan explored the relationship between mutation frequency, which is what is typically measured, and the more fundamental mutation rate (mutation per nucleotide pair), which is estimated from mutation frequency, subject to a number of key assumptions (e.g., exponential, linear, mixed replication, and mutation rate often assumed to be the same for all replication events). Indeed, Jan devoted a book to this subject (Drake 1970). In “Twists and Turns,” Jan’s memoir published in the journal DNA Repair, the ideas discussed in his book were further refined and presented alongside an analysis of mutational loads for a variety of organisms (Drake 2012). The key take-home message from these studies can be embodied in “Drakes’s law.” Michael Lynch (Lynch 2010) explained that “Drake concluded that the mutation rate/nucleotide site/generation scales with genome size in DNA-based microbes, which further implies that the mutation rate/genome/generation is essentially constant across all microbial life.” Furthermore, the approaches Drake developed back in 1991 retain their validity two decades later. Although RNA-based organisms and organisms with more DNA follow different rules, the analyses developed by Jan nonetheless remain useful today.

Jan was also one of the first to observe hypermutability. Jan compared the link between his experimental work with RB69 that showed mutational clustering, with Jacques Ninio’s theoretical analysis of hypermutability (Ninio 1991). To quote the first sentence of Jan’s 2007 paper, “In 1991, Jacques Ninio argued that microbial populations would contain clones initiated by cells that had experienced some transitory attenuation in replication fidelity and therefore expressed a higher-than-average mutation frequency” (Drake 2007). Jan went on to say that Ninio’s proposal languished, but then he observed far too many multiple mutants with RB69 polymerase to be attributed to chance. This is another observation from Jan that has medical implications. High mutation frequencies and subpopulations of cells with mutator phenotypes help to account for the observation by I. Fidler that tumors are heterogeneous, having a variety of subpopulations with differing potential to be invasive (Fidler 1978), as well as the proposal by L. Loeb that a mutator phenotype is needed to drive the development of cancer cells (Loeb 2001).

Some Final Reflections

Our interaction with Jan began when we were members of Maurice Bessman’s laboratory at Johns Hopkins in the early 1970s. We experienced first-hand the discovery of proofreading by the T4 DNA polymerase. Thus, we have grown up scientifically following in Jan’s footsteps. We saw how Jan’s discovery of antimutator DNA polymerases led to the field of DNA replication fidelity and important offshoots that we discussed here. After a Monty Python-like metaphor, we liken Jan’s initial two short antimutator papers (Drake and Allen 1968; Drake et al. 1969) to a Rube Goldberg contraption4, in which a relatively microscopic initial event sets in motion a chain reaction culminating in a “megascopic” outcome. As biochemists, we have commented on Jan’s sustained impact on our understanding of DNA synthesis fidelity from a biochemical viewpoint, but a comprehensive view of fidelity from a genetics point-of-view was provided by Jan himself in a 2006 Genetics Perspectives article, “Chaos and Order in Spontaneous Mutation” (Drake 2006). We also recommend the thoroughly engaging view of “Young Jan,” by Matt Meselson and Frank Stahl (Meselson and Stahl 1998). We conclude with a story that Jan told Mark Johnston, the current GENETICS Editor-in-Chief: although Jan had published a paper with Jim Crow (Drake et al. 1998), he regretted never having published a paper authored by Dove, Drake, and Crow. In the absence of such an article, a bird watchers’ sighting will have to suffice (Figure 2). Jan was truly a geneticist par excellence.

Figure 2

A photo of William Dove, Jan Drake, and James Crow. William Dove is on the left, Jan Drake center, and James Crow on the right. The picture was taken at the 1992 Genetics Society of America Meeting, and was published in Bill Dove’s 2016 GENETICS Perspectives (Dove 2016).

Acknowledgments

We thank Polina Shcherbakova, Phuong Pham, and Malgorzata Jaszczur for commenting on the manuscript; Tom Kunkel and Roel Schaaper for sharing recollections; and offer our deep and abiding gratitude to Maurice J. Bessman for teaching by example, since Moishe worked at the bench each and every day in his Mergenthalar Hall laboratory at Johns Hopkins during our fledgling years. Support was provided to M.F.G. from National Institutes of Health grants R35 ES-028343 and GM-130450. The authors have no conflicts of interests.

Footnotes

Communicating editor: A. S. Wilkins

↵1 The Dragnet radio and TV show began in the early 1950s, written by and starring Jack Webb, an ushered in the iconic phrase “just the facts, ma’am.”

↵2 The use of the laudatory term “benchmark” is ascribed to Jan Drake’s inclusion of Muzyczka et al. (1972) in his Benchmarks in Genetics series (Drake and Koch 1976).

↵3 L.J.R.-K. acknowledges Jan’s encouraging words when she was a graduate student that “she was thinking like a geneticist,” a complement she cherished throughout her career, along with the support from Maurice Bessman. L.J.R.-K. took to heart Jan’s advice that in vivo studies open the doors to many insights.

↵4 For the unfortunate many born after 1960, we provide here an active internet link to see Rube Goldberg’s ingenious contraptions, all of which violate fundamental physical laws such as the conservation of energy and momentum, and assuredly others as well (https://www.artsy.net/article/artsy-editorial-cartoonist-rube-goldbergs-machines-turned-simple-tasks-epic-spectacles).

Received September 7, 2020.Accepted October 22, 2020.Copyright © 2020 by the Genetics Society of America