JOURNAL OF BACTERIOLOGY, June 2000, p. 2993–3001
11
0021-9193/00/$04.00⫹0
Copyright © 2000, American Society for Microbiology. All Rights Reserved.
MINIREVIEW
A Biochemical Mechanism for Nonrandom
Mutations and Evolution
BARBARA E. WRIGHT*
Division of Biological Sciences, The University of Montana,
Missoula, Montana
As this minireview is concerned with the importance of the
environment in directing evolution, it is appropriate to remem-
ber that Lamarck was the first to clearly articulate a consistent
theory of gradual evolution from the simplest of species to the
most complex, culminating in the origin of mankind (71). He
published his remarkable and courageous theory in 1809, the
year of Darwin’s birth. Unfortunately, Lamarck’s major con-
tributions have been overshadowed by his views on the inher-
itance of acquired characters. In fact, Darwin shared some of
these same views, and even Weismann (106), the father of
neo-Darwinism, decided late in his career that directed variation
must be invoked to understand some phenomena, as random
variation and selection alone are not a sufficient explanation
(71). This minireview will describe mechanisms of mutation
that are not random and can accelerate the process of evolu-
tion in specific directions. The existence of such mechanisms
has been predicted by mathematicians (6) who argue that, if
every mutation were really random and had to be tested
against the environment for selection or rejection, there would
not have been enough time to evolve the extremely complex
biochemical networks and regulatory mechanisms found in
organisms today. Dobzhansky (21) expressed similar views by
stating “The most serious objection to the modern theory of
evolution is that since mutations occur by ‘chance’ and are
undirected, it is difficult to see how mutation and selection can
add up to the formation of such beautifully balanced organs as,
for example, the human eye.”
The most primitive kinds of cells, called progenotes by
Woese (108), were undoubtedly very simple biochemically with
only a few central anabolic and catabolic pathways. Wächter-
häuser (103) theorizes that the earliest metabolic pathway was
a reductive citric acid cycle by which carbon fixation occurred
(64). At that point in time, some four billion years ago, how did
the additional, more complex metabolic pathways found in
even the simplest prokaryotes evolve? For that matter, how are
they evolving today? As pointed out by Oparin (79), it is in-
conceivable that a self-reproducing unit as complicated as a
nucleoprotein could suddenly arise by chance; a period of
evolution through the natural selection of organic substances
of ever-increasing degrees of complexity must intervene.
Horowitz (40) suggests a plausible scheme by which biosyn-
thetic pathways can evolve from the successive depletion and
interconversion of related metabolites in a primitive environ-
ment, as the rich supply of organic molecules is consumed by a
burgeoning population of heterotrophs. Thus, a possible sce-
* Mailing address: Division of Biological Sciences, The University of
Montana, Missoula, MT 59812. Phone: (406) 243-6676. Fax: (406)
243-4184. E-mail: bewright@selway.umt.edu.
2993
Vol. 182, No.
nario begins with the starvation of a self-replicating unit for its
precursor, metabolite A, utilized by enzyme 1 encoded by gene
1. When metabolite A is depleted, a mutation in a copy of gene
1 gives rise to gene 2 and allows enzyme 2 to use metabolite B
by converting it to metabolite A. Then metabolite B is de-
pleted, obtained from metabolite C, and so on, as an increas-
ingly complex biochemical pathway evolves. In fact, there are
Downloaded from jb.asm.org by on November 23, 2009
examples in which a similar series of events can actually be
observed in the laboratory, for example, involving enzymes
that are “borrowed” from existing pathways, via regulatory
mutations, to establish new pathways (75).
The starvation conditions that may initiate a series of events
such as those described above target the most relevant genes
for increased rates of transcription, which in turn increase rates
of mutation (111). Transcriptional activation can result from
the addition of a substrate or from the removal of a repressor
or an end product inhibitor. The latter mechanism, called
derepression, occurs in response to starvation for an essential
substrate or for an end product that represses its own synthesis
by feedback inhibition. Since evolution usually occurs in re-
sponse to stress (41), transcriptional activation via derepres-
sion is the main focus of this minireview.
EVOLUTION OF BIOCHEMICAL PATHWAYS
A number of events initiated by carbon source starvation can
facilitate the evolution of a new catabolic pathway. Under
these circumstances, cells with gene duplication and higher
enzyme levels have a selective advantage (87, 95). In some
systems, duplicated segments are specifically subject to higher
mutation rates (93), providing ideal and expendable material
for mutations representing minor modifications of existing
genes (58). These new genes can encode modified enzymes
catalyzing reactions closely related and/or complementary to
those in existence (56). An additional consequence of starva-
tion is the removal of feedback controls, resulting in the dere-
pression of genes previously inhibited by the now absent me-
tabolite. Increased rates of mutation in these derepressed
genes increase the probability of creating a new gene-enzyme
system. A number of examples exist in which derepression of a
gene has enabled an enzyme to use a new substrate. For ex-
ample, altros-galactoside can be used by ␤-galactosidase after
it is derepressed (53); other examples are ␤-glycerolphosphate
via alkaline phosphatase (100), putrescine via diamine-␣-keto-
glutarate transaminase (44), and D-mannitol via D-arabitol de-
hydrogenase (55).
An excellent example of the evolution of biochemical path-
ways involves the modification of two genes to serve the new
demands imposed by carbon source starvation (56, 112). Rib-
itol dehydrogenase, which is induced by ribitol in wild-type
2994 MINIREVIEW
TABLE 1. Evolution of a catabolic pathwaya
Generation Km (mM) [14C]xylitol uptake
Strain
time (h) (cpm)
Xylitol Ribitol
b come apparent, genetic derepression may be the only mecha-
X 4.1 0 3 104
X1c 4.1 290 3 104
X2 1.7 120 2 185
X3 0.9 130 2 685
a
Data taken from reference 112.
b
Wild-type strain with inducible ribitol dehydrogenase.
c
Mutant strain with constitutive ribitol dehydrogenase.
Enterobacter arogenes (strain X in Table 1), is unable to use
xylitol. Starvation for ribitol in the presence of xylitol results in
a mutation to strain X1, in which ribitol dehydrogenase is
constitutive and able to use xylitol, which is a poor substrate for
the enzyme but not its inducer. By repeated growth cycling on
xylitol, derivative mutants X2 and X3 are obtained with lower
Km for xylitol. The enhanced uptake of labeled xylitol in the
final mutant, X3, is due to the acquisition of a constitutively
expressed active transport system for xylitol, originating from
the modification of an inducible transport system for D-arabi-
tol. Thus, two preexistent gene-enzyme systems evolve to ini-
tiate a new catabolic pathway in response to the stress of
imminent starvation.
In the evolution of a growth rate-limiting amino acid bio-
synthetic pathway, starvation, derepression, and higher muta-
tion rates can result in a lower Km for the rate-controlling
endogenous precursor of the pathway or in the ability to use a
new and more plentiful precursor for the synthesis of that
amino acid.
SPECIFICITY OF STARVATION-INDUCED
DEREPRESSION
Starvation for any essential nutrient activates systems that
protect the vulnerable cells from environmental damage (37,
72, 91). In addition, elaborate and specific feedback mecha-
nisms are deployed that counteract the particular crisis created
by the absent nutrient (9, 13, 39, 54, 66, 77). For example,
inorganic phosphate (Pi) starvation derepresses the pho regu-
lon, including a new high-affinity Pi transport system able to
cope with lower phosphate levels, and a hydrolase able to
obtain Pi from new sources (67). Nitrogen starvation dere-
presses the ntr regulon, including glutamine synthetase, which
has a higher affinity for NH4⫹ than the constitutive glutamate
dehydrogenase (65). Starvation for leucine specifically targets
derepression of the genes in the leu operon (111). Regulation
of amino acid biosynthetic operons by attenuation is exquis-
itely sensitive (over ranges of 1,000-fold) to the need for, and
the supply of, all the amino acids (18, 49, 97). Attenuation
regulation is an impressive example of the remarkable mech-
anisms that have evolved to ensure the conservation of pre-
cious reserves and the derepression and activation of only
those systems essential for survival under particular conditions
of starvation. As seen in Table 2, each amino acid operon
encodes in its leader sequence a series of codons for the amino
acid product of that operon. This sets the stage for a very
complex and specific mechanism to monitor the precise
amount of amino acid required relative to the amount avail-
able (49, 107). If the Leu codons are replaced with Thr codons,
regulation of the leu operon by leucine is abolished (12).
Presumably, feedback mechanisms existing today evolved in
the past to prevent unnecessary and wasteful metabolic activ-
J. BACTERIOL.
ities by coordinating these activities with the presence or ab-
sence of nutrients in the environment. High mutation rates in
derepressed genes prepare cells to respond rapidly to new
challenges should the stress become more severe. As will be-
nism by which particular environmental conditions of stress
target specific regions of the genome for higher mutation rates
(hypermutation). Although this direct avenue for increasing
variability is probably not available to multicellular organisms
in which germ cells and somatic cells are separated, the dere-
pression of biosynthetic pathways is essential to increased lon-
gevity in mammals subjected to caloric restriction (54), and
amino acid limitation in rats can also induce gene expression
(9).
MECHANISMS OF MUTATIONS VERSUS
MECHANISMS OF EVOLUTION
In a scientific context, the word spontaneous is meaningless.
Every event is preceded by, and dependent upon, innumerable
known and unknown prior events and circumstances. There-
fore, the work background will be used when referring to the
many environmental conditions, cellular events, and repair
Downloaded from jb.asm.org by on November 23, 2009
processes that affect mutation rates in nature. Although back-
ground mutations, such as deamination, alkylation, and depuri-
nation, occur with low frequencies, they have characteristic,
finite activation energies under physiological conditions (31,
32, 48, 60), and the molecular mechanisms by which they occur
are well established. For example, protonation of the N-3 of
cytosine results in its deamination to uracil. This process oc-
curs 140 times more frequently in single-stranded DNA
(ssDNA) than in double-stranded DNA (dsDNA) (in which
N-3 is hydrogen bonded to N-1 of guanine), in mismatched
base pairs, and in AT-rich regions that “breathe.” These back-
ground mutations and frameshift-induced deletions and addi-
tions are DNA sequence directed in that they occur most
frequently in ssDNA and in unpaired, mispaired, and methyl-
ated bases (15, 26, 28, 31, 32, 60). Such vulnerable bases can
arise as a consequence of slippage in tandem repeats or as a
result of stem-loop DNA secondary structures that arise from
sequences containing intrastrand inverted complements. The
loops in these structures consist of unpaired bases particularly
vulnerable to mutations (other more complex structures con-
taining vulnerable bases are also important as precursors of
mutations but will not be discussed in this minireview).
In an evolutionary context, we are not concerned with the
above molecular mechanisms by which individual types of mu-
tation occur but with another kind of mechanism. Evolution
depends upon events that enhance mutation rates, thus in-
creasing the supply of variants from which the fittest are se-
lected. Therefore, the word mechanism in the present context
will refer to the circumstances affecting mutation rates. That
any DNA-destabilizing event will increase mutation rates is
TABLE 2. Leader sequences of attenuation-regulated operons
Operon Amino acid sequence of codons in leader RNAa
his..................Met-Thr-Arg-Val-Gln-Phe-Lys-His-His-His-His-His-His-
His-Pro-Asp-
phe.................Met-Lys-His-Ile-Pro-Phe-Phe-Phe-Ala-Phe-Phe-Phe-Thr-
Phe-Pro-
thr..................Met-Lys-Arg-Ile-Ser-Thr-Thr-Ile-Thr-Thr-Thr-Ile-Thr-
Ile-Thr-Thr-Gly-
leu..................Met-Ser-His-Ile-Val-Arg-Phe-Thr-Gly-Leu-Leu-Leu-Leu-
Asn-Ala-Phe-
a
Operon-specific sequences shown in bold type.
VOL. 182, 2000
axiomatic. In growing cells, such events include polymerase
and proofreading errors during DNA replication, as well as
recombination, transcription, and repair. DNA transcription
and repressor binding affect the rate of deletions in Escherichia
coli plasmids (101). The availability of ssDNA (leading- and
lagging-strand DNA templates) facilitates the slippage of tan-
dem repeats and the formation of stem-loop structures (89). In
nongrowing cells, however, the DNA-destabilizing events of
replication are probably not primary causes of mutations.
Within an hour following starvation, bacterial cells undergo
major metabolic transitions (the stringent response [13]) in
which genes required for cell division are repressed while a
number of other genes (depending upon the starvation regi-
men) are derepressed. During this transition from exponential
growth to stationary phase, events related to gene activation
parallel a sharp increase in supercoiling, suggesting that tran-
scriptional activation may drive supercoiling and the resulting
DNA secondary structures that are precursors of mutations
(discussed below). Among the known DNA-destabilizing
events, only transcription can be selectively activated (7), ei-
ther by induction or derepression. Derepression of the leu
operon in E. coli is specifically correlated with an increased
rate of leuB mRNA turnover (62; J. M. Reimers, A. Longacre,
and B. E. Wright, Conf. DNA Repair Mutag., abstr. B29, p. 80,
1999) and an increased reversion rate of the leuB mutant gene;
this mutation is located at the site of a predicted stem-loop
structure (111).
RANDOM VERSUS NONRANDOM
HYPERMUTATION
As discussed above, background mutations are sequence
directed and not random in the sense that they occur in bases
made vulnerable by virtue of their particular location within
specific DNA sequences, such as tandem repeats, or the un-
paired and mispaired bases of stem-loop structures. Dobzhan-
sky’s statement (22) enlarges upon this point: “The structure of
a gene is a distillate of its history, and the mutations that may
occur in a gene are determined by the succession of environ-
ments in which that gene and its ancestors existed since the
beginnings of life. The environment prevailing at the time
mutation takes place is only a component of the environmental
complex that determines the mutation.” The definitions of
directed and random that are appropriate in the above context
are neither relevant nor useful, however, when discussing
mechanisms of evolution. By the neo-Darwinian definition, a
mutation is random if it is unrelated to the metabolic function
of the gene and if it occurs at a rate that is undirected by
specific selective conditions of the environment. For example,
mutagenic DNA-destabilizing events associated with cell divi-
sion are random, as they are dependent upon growth rate and
selective conditions of the environment only insofar as those
conditions affect the rate of cell division. However, the focus of
this minireview concerns the consequences of environmental
stress on evolution. What are the DNA-destabilizing processes
operative in stressed, nongrowing organisms forced to mutate
before they can continue to multiply? Mechanisms must have
evolved in starving cells to stimulate metabolic changes and
mutations that facilitate adaptation to new circumstances.
With the above neo-Darwinian definition of random in
mind, an impressive array of circumstances that enhance back-
ground mutation rates in response to environmental stress may
be examined with respect to whether or not they are random
(undirected). Examples of conditions that result in undirected,
genomewide hypermutation include those caused by UV irra-
diation, reactive oxygen species, mismatch repair-deficient mu-
MINIREVIEW 2995
tator phenotypes (35, 98), horizontal gene transfer by trans-
duction with a viral particle, and mobile genetic elements that
increase mutation rates by inserting at particular regions or at
target sequences within the genome (73, 76). Such mechanisms
are undirected because, for example, a mismatch repair defi-
ciency will result in failure to repair a particular kind of lesion
regardless of whether or not it confers a selective advantage
upon its host.
In higher organisms, environmental conditions of stress do
not have direct access to the cells involved in reproduction, and
different mechanisms resulting in hypervariation have evolved.
For example, localized DNA rearrangements and shuffling
produce extensive beneficial variation (82, 96), and hypervari-
able sequences provide continual changes in the composition
of venoms produced by snakes (29) or snails (78) to overcome
resistance developed by their predators or prey. These mech-
anisms are also random. The threat of predators does not
result in hypermutation; there is no evidence that the circum-
stances selecting such hypermutable genes bear any metabolic
relationship to the mechanisms by which they originally arose.
A gene may be hypermutable because it contains a hot spot
due to a particular DNA sequence, and if a high mutation rate
is advantageous to its host, that gene will be selected during
Downloaded from jb.asm.org by on November 23, 2009
evolution. However, its hypermutability per se is undirected,
since it is unrelated to those selective conditions and to the
function of the gene. These random mechanisms resulting in
hypermutation are in essence serendipitous relationships; in
contrast, hypermutation resulting from derepression is local-
ized as a direct consequence of a specific response to environ-
mental challenge.
TWO MECHANISMS BY WHICH TRANSCRIPTION CAN
INCREASE MUTATION RATES
Transcription exposes ssDNA. The most common base sub-
stitution events in the spectra of background mutations in
E. coli and mammalian cells are G 䡠 C-to-A 䡠 T transitions. Fix
and Glickman (28) observe that 77% of these mutations orig-
inate on the nontranscribed strand in E. coli mutants unable to
repair deaminated cytosines. This suggests that the unpro-
tected single strand in the transcription “bubble” is significant-
ly more vulnerable to mutations than the transcribed strand,
which is protected as a DNA-RNA hybrid (Fig. 1A). The fre-
quency of UV-induced lesions in the lacI gene is also higher in
the nontranscribed strand than in the transcribed strand (46).
In fact, cytosines deaminate to uracils in ssDNA at more than
100 times the rate in dsDNA (31, 32, 60). The relative muta-
bility of the nontranscribed strand is also seen in a plasmid
system in which a fourfold increase in the frequency of tran-
sitions occurs selectively in the nontranscribed strand when
transcription is induced (4). Transcription may therefore be a
prerequisite for many C-to-T transition mutations, since other
mechanisms resulting in the transient generation of single-
stranded sequences, such as replication or breathing (102) do
not lead to asymmetry in the two strands. Apparently, the ob-
served strand bias cannot be explained by transcription-cou-
pled repair (36), since base mismatches are poor substrates for
this kind of repair, and the same strand bias is observed when
the host is deficient in repairing U 䡠 G and T 䡠 G mismatches
(4). Thus, transcription may be implicated as a major cause of
background transition mutations in nature.
Transcriptional activation as a mechanism for increasing
mutation rates was first proposed in 1971, by Brock (8) and
Herman and Dworkin (38). Their work demonstrates that
recA-independent lac reversion rates of frameshift and point
mutations are higher when transcription is induced by isopro-
2996 MINIREVIEW
FIG. 1. (A) Exposure of the nontranscribed strand during transcription; (B)
effect of transcription on supercoiling; (C) a typical stem-loop structure contain-
ing unpaired and mispaired bases; (D) mutation 1, a C-to-T transition in the loop.
pyl-␤-D-thiogalactopyranoside (IPTG), and that the effect is
specific. More recently, specifically induced, transcription-en-
hanced mutations have also been shown for a lys frameshift
mutation in Saccharomyces cerevisiae (16, 74). Starvation-in-
duced stringent response mutations in E. coli (62, 109–111)
and Bacillus subtilis (90) occur as a result of transcriptional
J. BACTERIOL.
activation triggered by gene derepression, not induction. In
this system, mutations arise during the transition between
growth and stationary phase and they are recA independent,
similar to the lac reversions mentioned above. This distin-
guishes them from prolonged stress-induced adaptive muta-
tions (11) and from DNA damage-induced SOS mutagenesis
(104), both of which require recA (and will not be discussed in
this minireview). It is noteworthy that the experiments de-
scribed above on the effects of artificially induced transcription
on mutation rates in growing cells are all examples of specifi-
cally directed mutations. However, none of the researchers
come to that conclusion or challenge the assumptions and
implications inherent in the experiments of Luria and Del-
bruck (63), which reinforce neo-Darwinism. This situation may
be due to the dominance of current dogma and to the assump-
tion that mechanisms operative during growth cannot also be
critical during evolution under conditions of environmental
stress. In fact, the limited evidence now available suggests that
only growing cells, or cells in transition between growth and
stationary phase, have the metabolic potential required for
specific, transcription-induced mutations in response to envi-
ronmental challenge. Thus, IPTG induction enhances lac re-
version rates in growing cells (38) but not in cells subjected to
Downloaded from jb.asm.org by on November 23, 2009
prolonged stress (17). Transcriptional activation is the mech-
anism for enhancing mutation rates both in the artificially
induced systems and in stringent response mutations (90, 111;
J. M. Reimers, A. Longacre, and B. E. Wright, Conf. DNA
Repair Mutag., 1999). However, only the latter are relevant to
evolution, since they occur naturally as a result of starvation-
induced derepression. Mutations that most benefit organisms
and accelerate evolution may occur as an immediate response
to imminent starvation, when cells still have the metabolic
resources to respond specifically to the particular conditions of
stress at hand.
Transcription drives localized supercoiling. Chromosomal
DNA from bacterial cells is negatively supercoiled. The level of
global negative supercoiling in E. coli cells is maintained within
a physiologically acceptable range by two opposing enzyme
activities: DNA gyrase, which introduces negative supercoils,
and topoisomerase I, which relaxes them. Investigations with
plasmids grown in E. coli (59, 81) demonstrate the presence of
stem-loop structures in naturally occurring supercoiled circular
DNA molecules (Fig. 1B). Analyses with single strand-specific
nuclease show that DNA molecules with high superhelical
densities are selectively cleaved, in contrast to their linearized
counterparts with which they are in dynamic equilibrium in
vivo. The sequence surrounding the area of cleavage reveals
inverted complementary sequences that hydrogen bond to be-
come the stem separated by noncomplementary bases that
become the single-stranded loop and the substrate for nuclease
cleavage. Such complex structures form preferentially in easily
denatured AT-rich stretches of DNA and occur about 10,000
times more frequently than expected by chance (30, 59), sug-
gesting their selection during evolution. Data indicate that
stem-loop-based recombination may have evolved in the early
“RNA world” (94) and that the potential to generate stem-
loops was later conserved, for example, in hypervariable snake
venom genes under strong selection to keep one step ahead of
both predators and prey (29).
A number of variables, such as temperature, anaerobiosis,
osmolarity, and nutritional shifts, affect DNA supercoiling (1,
19, 47, 85, 86). Some environmental perturbations affect plas-
mid systems and chromosomes in a similar manner, while some
apparently do not (25). Transcription both responds to and
promotes changes in supercoiling. The optimal level of super-
coiling for gene expression varies for different genes, and su-
VOL. 182, 2000
FIG. 2. Depletion of limiting levels of glucose by E. coli immediately triggers
several metabolic events. Depletion of a limiting amino acid has similar conse-
quences. Curve A represents growth (optical density). Curve B represents the
concentration of cAMP, ppGpp, ␴S, ␤-galactosidase, 30 other proteins, and
supercoils.
percoiling-induced conformational changes may be required
for structural changes in regulatory complexes and for recog-
nition by RNA polymerase (RNAP) (85). Transcription has a
profound effect on supercoiling, because RNAP distorts and
destabilizes dsDNA. As indicated in the twin-domain model
(Fig. 1B) of Liu and Wang (61), negative supercoiling is gen-
erated behind, and positive supercoiling in front of, the ad-
vancing RNAP transcription complex. Many investigations
provide evidence demonstrating that transcription drives su-
percoiling in vivo (1, 19, 20, 27, 86) and that the wave gener-
ated can be as long as 800 bp (47). Negative supercoiling
induces and stabilizes a transition from the right-handed B-
form to the left-handed Z-form of DNA (42); a chemical assay
detecting these distortions reveals that transcription-induced
supercoiling is highly localized (47, 86). During the induction
of transcription, supercoils are found inside each transcribed
region, as well as upstream and downstream of each individual
RNAP complex. Transcription from a strong promoter leads to
greater negative supercoiling than transcription from a weak
one (27). A major role of DNA topoisomerase I is now con-
sidered to be the relaxation of local negative supercoiling dur-
ing transcription, thus preventing unacceptably high levels of
supercoiling and associated R-loops that form when nascent
RNA moves behind the advancing RNAP to bond with its
original template DNA (69, 70, 105). The bulk of plasmid
DNA does not exhibit stem-loop conformations during loga-
rithmic growth. However, supercoiling may play a particularly
important role in stressed cells, in which a disruption can occur
between transcription and translation, thus promoting both
R-loop formation and supercoiling (68, 69). The inhibition of
protein synthesis by chloramphenicol, which uncouples tran-
scription and translation, induces stem-loop formation in the
overwhelming majority of DNA molecules (19).
When E. coli is grown with limiting levels of glucose (Fig. 2),
a burst in supercoiling occurs precisely at the moment of glu-
cose depletion, as the cells cease logarithmic growth and enter
stationary phase (1). This is also the moment at which a sharp
increase occurs in the concentration of cyclic AMP (cAMP)
(10), ppGpp (23, 51), ␴S (50), and about 30 new proteins,
including ␤-galactosidase (34, 37). The increase in negative
supercoiling under these circumstances can be due to the in-
crease in transcription known to occur as a result of derepres-
MINIREVIEW 2997
sion in response to starvation for any essential nutrient. The
two “alarmones,” cAMP and ppGpp, activate transcription in
derepressed genes in a number of systems. Both are required
for the synthesis of enzymes that catabolize alternative carbon
sources, such as ␤-galactosidase (83, 84, 91). The alarmone
ppGpp activates the synthesis of ␴S (33), which in turn governs
the expression of a number of stationary-phase genes involved
in the starvation-mediated resistance to osmotic, oxidative, and
heat damage (37, 72). Under circumstances in which a group of
related genes become activated, such as those dependent upon
␴S, the topological changes in DNA could provide a mechanism
by which transcriptional activation in one gene may influence
adjacent genes (47, 85). Changes in stem-loop formation and
superhelicity similar to that caused by glucose starvation (Fig.
2) are also observed immediately following amino acid starva-
tion or the inhibition of protein synthesis (19). Within 30 min
following treatment with chloramphenicol or valine (which cre-
ates an isoleucine deficit and ppGpp accumulation in E. coli),
stem-loop formation is evident. Isoleucine starvation results in
the formation of stem-loops in a much smaller number of
DNA molecules than are affected by chloramphenicol inhibi-
tion, consistent with the selective derepression of relatively few
genes in the absence of a single amino acid.
Downloaded from jb.asm.org by on November 23, 2009
In response to starvation for any essential metabolite, the
immediate problem is addressed specifically (e.g., derepression
of a higher-affinity transport system for that metabolite), cou-
pled with a general increase in stress resistance. Starvation
results in derepression, and transcription drives localized su-
percoiling; the formation of stem-loop structures at regions of
high superhelicity results in localized hypermutation (Fig. 1).
Although energetic considerations do not favor the creation of
complex structures in metabolically inactive dsDNA, transcrip-
tion clearly accelerates supercoiling and transitions to second-
ary DNA structures (1, 19, 20, 27, 47, 59, 69, 70, 81, 86).
SECONDARY DNA STRUCTURES: ARE THEY
PRECURSORS TO MUTATIONS?
Almost 40 years ago, Benzer (5) demonstrated that the mu-
tability of specific sites in the genome varied by orders of
magnitude, suggesting that these differences in mutation rate
might reflect particular characteristics of the DNA sequence
associated with hot spots. In fact, mutable sites are frequently
the consequence of their location within DNA secondary struc-
tures. Simple stem-loops arise from ssDNA sequences contain-
ing two segments that are inverted complements, usually about
10 to 15 bases long, separated by 5 to 10 noncomplementary
bases that become the loop at the end of the stem formed by
hydrogen bonding of the two complementary segments (Fig.
1C). These structures are called hairpins if the loop is very
small and cruciforms if they form opposite one another in each
DNA strand. Perfect complementarity (a palindrome) is rare;
the more common quasipalindromes or stem-loops contain
bases that are left unpaired or mispaired and therefore vulner-
able to deamination (mutations 1 and 2 in Fig. 1C), deletion
(mutations 3 and 4), replacement (mutations 5 and 6), or
complementation by the insertion of a new base to the struc-
ture (mutation 7). The stem-loop is in dynamic equilibrium
with linearized DNA, and changes such as those indicated in
Fig. 1C only become fixed as mutations in the course of further
metabolic events such as repair or replication (Fig. 1D). For
example, the entire structure depicted in Fig. 1C will be ex-
cluded and deleted by virtue of new DNA synthesis across the
base of its stem. However, if this structure returns to its linear
form prior to new DNA synthesis, the potential changes indi-
cated above can be immortalized due to synthesis templated by
2998 MINIREVIEW J. BACTERIOL.

the altered sequence. An example of this process is indicated in

Fig. 1D, in which a C in the loop is deaminated to uracil, which
codes for A, which then codes for T during DNA synthesis,
resulting in a C-to-T transition.
What is the probability that these structures actually exist in
vivo and constitute precursors of background mutations in
nature? One kind of evidence for their existence is the striking
correlation of deletion end points with tandem repeats and the
ends of potential secondary structures. The leuB, argH, and
pyrD mutations in E. coli all occur at the end of predicted
stem-loops (111). The lacI gene has frequently been used as a
model system for investigating these correlations by comparing
its nucleotide sequence (26) to those of various mutant strains.
Among the sequenced deletion mutations in lacI, 7 occur in
tandem repeats, 6 consist of deleted stem-loop structures, and
4 of these 13 mutations share both characteristics. Although
three remaining mutations fail to coincide exactly with pre-
dicted vulnerable sites, they may be explained by different
DNA secondary structural intermediates including interstrand
misalignments (26, 88, 92). Todd and Glickman (99) analyzed
102 amber and 71 ochre mutants in the lacI gene, correlating
mutation hot spots with the locations of predicted unpaired
sites, many of which are located in stem-loop structures. Mu-

Downloaded from jb.asm.org by on November 23, 2009

tational hot spots are highly localized, and 50% of the non-
sense mutations arose in a segment comprising only 6% of the
DNA sequence analyzed. Each hot spot is found to be located
at an unpaired site within potential secondary structures.
Clearly, these correlations depict causal relationships.
The mechanism of frameshift mutagenesis has also been
examined in vitro during DNA polymerization (80). In this
system, sequence misalignments result from intrastrand com-
plementary pairings between two segments within a single
newly synthesized strand, as well as from interstrand pairings
between a segment in the new strand and a complementary
sequence in the template strand. When these misaligned seg-
ments are used as templates for DNA synthesis, mutant se-
quences are produced. Polymerase pausing, or the local rate of
DNA polymerization, was also measured and correlated with
the misalignments, since pausing is sequence specific as well
(43). Pausing serves to increase the time of exposure of muta-
ble bases and is known to promote mutagenesis (2, 3). The
correlation between pausing, positions of frameshift misalign-
ments, and subsequent deletions can account for 97% of the
mutations observed. Moreover, the most common mutations
coincide precisely with the strongest pause sites, and the ter-
mini of pause sites correlate with the sequence at which the
deletions begin.
Thus, in vivo and in vitro investigations strongly implicate
the existence of DNA secondary structures as mutagenic sub-
strates and/or as structural precursors to mutagenic substrates
that give rise to mutations that are immortalized during new
DNA synthesis or repair (Fig. 1D). As discussed earlier, mu-
tations occur preferentially in ssDNA and in unpaired and
mispaired bases (28, 31, 32, 48, 60). Other kinds of evidence
also support a role for secondary structures as precursors of
mutations. Many insertion mutations (Fig. 1C) can best be
explained if the other strand of a predicted transient stem were
used as a template for DNA synthesis prior to replication. The
fact that mutations are grouped closely together much more
frequently than could occur by chance implicates a single ini-
tiating event (structure). As the researchers of the above in-
vestigations point out, other variables undoubtedly contribute
to and modify the correlation observed between DNA se-
quences, misaligned structures, polymerase pausing, and mu-
tations. Nevertheless, these investigations are but a small frac-
tion of an enormous literature providing compelling evidence
FIG. 3. An algorithm for evolution.
that the sequence-dependent secondary structures created and
stabilized by supercoiling are precursors to mutations.
CONCLUSIONS
Many scientists may share Dobzhansky’s intuitive conviction
that the marvelous intricacies of living organisms could not
have arisen by the selection of truly random mutations. This
minireview suggests that sensitive, directed feedback mecha-
nisms initiated by different kinds of stress might facilitate and
accelerate the adaptation of organisms to new environments.
The specificity in the series of events summarized by Fig. 3
resides entirely in the first step, which is meant to suggest a
pattern of derepression elicited by a corresponding pattern of
adverse conditions. Microorganisms in nature must be con-
fronted simultaneously by a complex set of problems, for ex-
ample, the threat of oxidative or osmotic damage together with
suboptimum concentrations of many essential nutrients. Tran-
scriptional activation of genes derepressed to various degrees
would expose the nontranscribed strands to mutations and
stimulate localized supercoiling. Vulnerable bases in the com-
plex DNA structures resulting from supercoiled DNA will also
contribute to localized hypermutation in the genes activated to
cope with the stresses that initiate the above series of events.
A multitude of random mechanisms result in hypermutation
under conditions of environmental stress and clearly contrib-
ute to the variability essential to evolution. However, since
most mutations are deleterious, random mechanisms that in-
crease mutation rates also result in genomewide DNA damage.
Among microorganisms, from phage to fungi, the overall mu-
tation rate per genome is remarkably constant (within 2.5-
fold), presumably reflecting an obligatory, delicate balance be-
tween the need for variation and the need to avoid general
genetic damage (24, 45, 57). Thus, mutator strains are not
selected in nature but remain at 1 to 2% of the population (35,
52); under certain adverse conditions, they flourish for short
periods but are then selected against, apparently because of
widespread deleterious effects intrinsic to genomewide hyper-
VOL. 182, 2000
mutation. In contrast, hypermutation that is the consequence
of starvation-induced derepression and transcriptional activa-
tion represents a very rapid and specific response to each
adverse circumstance. The extent to which normal background
mutations in nature are due to derepression mechanisms is
difficult to estimate, but the location of most C-to-T transitions
on the nontranscribed strand suggest that it may be significant.
Regardless, a mechanism that limits an increase in mutation
rates to genes that must mutate in order to overcome prevail-
ing conditions of stress would surely be beneficial and there-
fore selected during evolution.
The environment gave rise to life and continues to direct
evolution. Environmental conditions are constantly controlling
and fine-tuning the transcriptional machinery of the cell. Feed-
back mechanisms represent the natural interactive link be-
tween an organism and its environment. An obvious selective
advantage exists for a relationship in which particular environ-
mental changes are metabolically linked through transcription
to genetic changes that help an organism cope with new de-
mands of the environment. In nature, nutritional stress and
associated genetic derepression must be rampant. If mutation
rates can be altered by the many variables controlling specific,
stress-induced transcription, one might reasonably argue that
many mutations are to some extent directed as a result of the
unique metabolism of every organism responding to the chal-
lenges of its environment. Thus, mutations are brought within
the realm of metabolic events to become the final, irreversible
act of metabolism in the constant struggle to adapt or die.
ACKNOWLEDGMENTS
I am indebted to my colleagues in the Division of Biological Sciences
and to Lynn Ripley and Karl Drlica for suggestions and valuable
criticisms of this manuscript.
This work was supported in part by the Eppley Foundation, PHS
grant GM/OD54279, a MONTS grant from the NSF EPSCoR pro-
gram, and by the University of Montana.
REFERENCES
1. Balke, V. L., and J. D. Gralla. 1987. Changes in the linking number of
supercoiled DNA accompany growth transitions in Escherichia coli. J. Bac-
teriol. 169:4499–4506.
2. Bebenek, K., C. M. Joyce, M. P. Fitzgerald, and T. A. Kunkel. 1990. The
fidelity of DNA synthesis catalyzed by derivatives of Escherichia coli DNA
polymerase I. J. Biol. Chem. 265:13878–13887.
3. Bebenek, K., J. Abbotts, S. H. Wilson, and T. A. Kunkel. 1993. Error-prone
polymerization by HIV-1 reverse transcriptase. J. Biol. Chem. 268:10324–
10334.
4. Beletskii, A., and A. S. Bhagwat. 1996. Transcription-induced mutations:
increase in C-to-T mutations in the non-transcribed strand during transcrip-
tion in Escherichia coli. Proc. Natl. Acad. Sci. USA 93:13919–13924.
5. Benzer, S. 1961. On the topography of the genetic fine structure. Proc. Natl.
Acad. Sci. USA 47:403–415.
6. Bernhard, R. 1967. Heresy in the halls of biology: mathematicians question
Darwinism. Sci. Res. (New York) 2:59–66.
7. Bick, M. D., C. S. Lee, and C. A. Thomas. 1972. Local destabilization of
DNA during transcription. J. Mol. Biol. 71:1–9.
8. Brock, R. D. 1971. Differential mutation of the ␤-galactosidase gene of
Escherichia coli. Mutat. Res. 11:181–186.
9. Bruhat, A., C. Jousse, and P. Fafournoux. 1999. Amino acid limitation
regulates gene expression. Proc. Nutr. Soc. 58:625–632.
10. Buettner, M. J., E. Spitz, and H. V. Rickenberg. 1973. Cyclic adenosine 3⬘,
5⬘-monophosphate in Escherichia coli. J. Bacteriol. 14:1068–1073.
11. Cairns, J., and P. L. Foster. 1991. Adaptive reversion of a frameshift
mutation in Escherichia coli. Genetics 128:695–701.
12. Carter, P. W., D. L. Weiss, H. L. Weith, and J. M. Calvo. 1985. Mutations
that convert the four leucine codons of the Salmonella typhimurium leu
leader to four threonine codons. J. Bacteriol. 162:943–949.
13. Cashel, M., D. R. Gentry, V. J. Hernandez, and D. Vinella. 1996. The
stringent response, p. 1458–1496. In F. C. Neidhardt, R. Curtis III, J. L.
Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M.
Riley, M. Schaechter, and H. Umbarger. (ed.), Escherichia coli and Salmo-
nella: cellular and molecular biology, 2nd ed., vol. 1. American Society for
Microbiology, Washington, D.C.
MINIREVIEW 2999
14. Chatterji, D., N. Fujita, and A. Ishihama. 1998. The mediator for stringent
control, ppGpp, binds to the beta-subunit of Escherichia coli RNA poly-
merase. Genes Cells 3:279–287.
15. Coulondre, C., J. H. Miller, P. J. Farabaugh, and W. Gilbert. 1978. Mo-
lecular basis of base substitution hotspots in Escherichia coli. Nature 274:
775–780.
16. Datta, A., and S. Jinks-Robertson. 1995. Association of increased sponta-
neous mutation rates with high levels of transcription in yeast. Science
268:1616–1619.
17. Davis, B. D. 1989. Transcriptional bias: a non-Lamarkian mechanism for
substrate-induced mutations. Proc. Natl. Acad. Sci. USA 96:5005–5009.
18. Davis, M. G., and J. M. Calvo. 1977. Relationship between messenger
ribonucleic acid and enzyme levels specified by the leucine operon of
Escherichia coli K-12. J. Bacteriol. 131:997–1007.
19. Dayn, A., S. Malkhosyan, D. Duzhy, V. Lyamichev, Y. Panchenko, and S.
Mirkin. 1991. Formation of (dA-dT)n cruciforms in Escherichia coli cells
under different environmental conditions. J. Bacteriol. 173:2658–2664.
20. Dayn, A., S. Malkhosyan, and S. M. Mirkin. 1992. Transcriptionally driven
cruciform formation in vivo. Nucleic Acids Res. 20:5991–5997.
21. Dobzhansky, T. 1950. The genetic basis of evolution. Sci. Am. 182:32–41.
22. Dobzhansky, T. 1950. Heredity, environment, and evolution. Science 111:
161–166.
23. Donini, P., V. Santonastaso, J. Roche, and A. J. Cozzone. 1978. The rela-
tionship between guanosine tetraphosphate, polysomes and RNA synthesis
in amino acid starved Escherichia coli. Mol. Biol. Rep. 4:15–29.
24. Drake, J. W. 1991. A constant rate of spontaneous mutation in DNA-based
microbes. Proc. Natl. Acad. Sci. USA 88:7160–7164.
25. Drlica, K., R. J. Franco, and T. R. Steck. 1988. Rifampin and rpoB muta-
Downloaded from jb.asm.org by on November 23, 2009
tions can alter DNA supercoiling in Escherichia coli. J. Bacteriol. 170:4983–
4985.
26. Farabaugh, P. J., U. Schmeisser, M. Hofer, and J. H. Miller. 1978. Genetic
studies of the lac repressor. VII. On the molecular nature of spontaneous
hotspots in the lacI gene of Escherichia coli. J. Mol. Biol. 126:847–863.
27. Figueroa, N., and L. Bossi. 1988. Transcription induces gyration of the
DNA template in Escherichia coli. Proc. Natl. Acad. Sci. USA 85:9416–
9420.
28. Fix, D. F., and B. W. Glickman. 1987. Asymmetric cytosine deamination
revealed by spontaneous mutational specificity in an UngG strain of Esch-
erichia coli. Mol. Gen. Genet. 209:78–82.
29. Forsdyke, D. R. 1995. Conservation of stem-loop potential in introns of
snake venom phospholipase A2 genes: an application of FORS-D analysis.
Mol. Biol. Evol. 12:1157–1165.
30. Forsdyke, D. R. 1995. Reciprocal relationship between stem-loop potential
and substitution density in retroviral quasispecies under positive Darwinian
selection. J. Mol. Evol. 41:1022–1037.
31. Frederico, L. A., T. A. Kunkel, and B. R. Shaw. 1990. A sensitive genetic
assay for the detection of cytosine deamination: determination of rate
constants and the activation energy. Biochemistry 29:2532–2537.
32. Frederico, L. A., T. A. Kunkel, and B. A. Shaw. 1993. Cytosine deamination
in mismatched base pairs. Biochemistry 32:6523–6530.
33. Gentry, D. R., V. J. Hernandez, L. H. Nguyen, D. B. Jensen, and M. Cashel.
1993. Synthesis of the stationary-phase sigma factor ␴S is positively regu-
lated by ppGpp. J. Bacteriol. 175:7982–7989.
34. Groat, R. G., and A. Matin. 1986. Synthesis of unique proteins at the onset
of carbon starvation in Escherichia coli. J. Ind. Microbiol. 1:69–73.
35. Gross, M. D., and E. C. Siegel. 1981. Incidence of mutator strains and
coliforms in nature. Mutat. Res. 91:107–110.
36. Hanawalt, P., and I. Mellon. 1993. Stranded in an active gene. Curr. Biol.
3:67–69.
37. Hengge-Aronis, R. 1996. Regulation of gene expression during entry into
stationary phase, p. 1497–1512. In F. C. Neidhardt, R. Curtis III, J. L.
Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M.
Riley, M. Schaechter, and H. Umbarger (ed.), Escherichia coli and Salmo-
nella: cellular and molecular biology, 2nd ed., vol. 1. American Society for
Microbiology, Washington, D.C.
38. Herman, R. K., and N. B. Dworkin. 1971. Effect of gene induction on the
rate of mutagenesis by ICR-191 in Escherichia coli. J. Bacteriol. 106:543–550.
39. Hinnebusch, A. G. 1988. Mechanisms of gene regulation in the general
control of amino acid biosynthesis in Saccharomyces cerevisiae. Microbiol.
Rev. 52:248–273.
40. Horowitz, N. H. 1945. On the evolution of biochemical syntheses. Proc.
Natl. Acad. Sci. USA 31:153–157.
41. Jablonski, D., J. Sepkoski, Jr., D. J. Bottjer, and P. M. Sheehan. 1983.
Origin of species in stressed environments. Science 222:1112–1124.
42. Jaworski, A., N. P. Higgins, R. D. Wells, and W. Zacharias. 1991. Topo-
isomerase mutants and physiological conditions control supercoiling and
Z-DNA formation in vivo. J. Biol. Chem. 266:2576–2581.
43. Kaguni, L. S., and D. A. Clayton. 1982. Template-directed pausing in in
vitro DNA synthesis by DNA polymerase ␣ from Drosophila melanogaster
embryos. Proc. Natl. Acad. Sci. USA 79:983–987.
44. Kim, K.-H. 1963. Isolation and properties of a putrescine-degrading mutant
of Escherichia coli. J. Bacteriol. 86:320–323.
3000 MINIREVIEW
45. Kimura, R. 1960. Optimum mutation rates and degree of dominance as
determined by the principle of minimum genetic load. J. Genet. 57:21–34.
46. Koehler, D. R., S. S. Awadallah, and B. W. Glickman. 1991. Sites of
preferential induction of cyclobutane pyrimidine dimers in the nontran-
scribed strand of lacI correspond with the sites of UV-induced mutation in
Escherichia coli. J. Biol. Chem. 266:11766–11773.
47. Krasilnikov, A. S., A. Podtelezhnikov, A. Vologodskii, and S. M. Mirkin.
1999. Large-scale effects of transcriptional DNA supercoiling in vivo. J.
Mol. Biol. 292:1149–1160.
48. Kunkel, T. A. 1984. Mutational specificity of depurination. Proc. Natl.
Acad. Sci. USA 81:1494–1498.
49. Landick, R., C. L. Turnbough, Jr., and C. Yanofsky. 1996. Transcription
attenuation, p. 1263–1286. In F. C. Neidhardt, R. Curtis III, J. L. Ingraham,
E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M.
Schaechter, and H. Umbarger (ed.), Escherichia coli and Salmonella: cel-
lular and molecular biology, 2nd ed., vol. 1. American Society for Micro-
biology, Washington, D.C.
50. Lange, R., and R. Hengge-Aronis. 1994. The cellular concentration of the
␴S subunit of RNA polymerase in Escherichia coli is controlled at the levels
of transcription, translation, and protein stability. Genes Dev. 8:1600–1612.
51. Lazzarini, R. A., M. Cashel, and J. Gallant. 1971. On the regulation of
guanosine tetraphosphate levels in stringent and relaxed strains of Esche-
richia coli. J. Biol. Chem. 246:4381–4385.
52. Le Clerk, J. E., B. Li, W. L. Payne, and T. A. Cebula. 1996. High mutation
frequencies among Escherichia coli and Salmonella pathogens. Science 274:
1208–1211.
53. Lederberg, J. 1951. Genetic studies with bacteria, p. 263–289. In L. C. Dunn
(ed.), Genetics in the 20th century. Macmillan, New York, N.Y.
54. Lee, C.-K., R. G. Klopp, R. Weindruch, and T. A. Prolla. 1999. Gene
expression profile of aging and its retardation by caloric restriction. Science
285:1390–1393.
55. Lerner, S. A., and E. C. C. Lin. 1962. A method for isolating constitutive
mutants for carbohydrate-catabolizing enzymes. Fed. Proc. 21:268.
56. Lerner, S. A., T. T. Wu, and E. C. C. Lin. 1964. Evolution of a catabolic
pathway in bacteria. Science 146:1313–1315.
57. Levins, R. 1967. Theory of fitness in a heterogenious environment. VI. The
adaptive significance of mutation. Genetics 56:163–178.
58. Lewis, E. B. 1951. Pseudoallelism and gene evolution. Cold Spring Harbor
Symp. Quant. Biol. 16:159–174.
59. Lilley, D. M. J. 1980. The inverted repeat as a recognizable structural
feature in supercoiled molecules. Proc. Natl. Acad. Sci. USA 77:6468–6472.
60. Lindahl, T., and B. Nyberg. 1974. Heat-induced deamination of cytosine
residues in deoxyribonucleic acid. Biochemistry 13:3405–3410.
61. Liu, L. F., and J. C. Wang. 1987. Supercoiling and the DNA template
during transcription. Proc. Natl. Acad. Sci. 84:7024–7027.
62. Longacre, A. L., J. M. Reimers, and B. E. Wright. 1999. Specificity of
transcription-enhanced mutations. Ann. N. Y. Acad. Sci. 870:383–385.
63. Luria, S., and M. Delbruck. 1943. Mutations of bacteria from virus sensi-
tivity to virus resistance. Genetics 28:491–504.
64. Maden, B. E. H. 1995. No soup for starters? Autotrophy and the origins of
metabolism. Trends Biochem. Sci. 20:337–341.
65. Magasanik, B. 1996. Regulation of nitrogen utilization, p. 1344–1356. In
F. C. Neidhardt, R. Curtis III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B.
Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. Umbarger
(ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd
ed., vol. 1. American Society for Microbiology, Washington, D.C.
66. Mager, W. H., and A. J. J. De Kruijff. 1995. Stress-induced transcriptional
activation. Microbiol. Rev. 59:506–531.
67. Makino, K., M. Amemura, S.-K. Kim, A. Nakata, and H. Shinagara. 1994.
Mechanism of transcriptional activation of the phosphate regulon in Esch-
erichia coli, p. 5–12. In A. Torriani-Gorini, E. Yagil, and S. Silver (ed.),
Phosphate in microorganisms: cellular and molecular biology. ASM Press,
Washington, D.C.
68. Masse, E., P. Phoenix, and M. Drolet. 1997. DNA topoisomerases regulate
R-loop formation during transcription of the rrnB operon in Escherichia
coli. J. Biol. Chem. 272:12816–12823.
69. Masse, E., and M. Drolet. 1999. Relaxation of transcription-induced neg-
ative supercoiling is an essential function of Escherichia coli DNA topo-
isomerase I. J. Biol. Chem. 274:16654–16658.
70. Masse, E., and M. Drolet. 1999. Escherichia coli DNA topoisomerase I
inhibits R-loop formation by relaxing transcription-induced negative super-
coiling. J. Biol. Chem. 274:16659–16664.
71. Mayr, E. 1982. The growth of biological thought: Diversity, evolution, and
inheritance. Belknap Press of Harvard University Press, Cambridge, Mass.
72. McCann, M. P., J. P. Kidwell, and A. Matin. 1991. The putative ␴ factor has
a central role in development of starvation-mediated general resistance in
Escherichia coli. J. Bacteriol. 173:4188–4194.
73. McClintock, B. 1950. The origin and behavior of mutable loci in maize.
Proc. Natl. Acad. Sci. USA 36:344–355.
74. Morey, N. J., C. N. Greene, and S. Jinks-Robertson. 2000. Genetic analysis
of transcription-associated mutation in Saccharomyces cerevisiae. Genetics
154:109–120.
J. BACTERIOL.
75. Mortlock, R. P. 1984. Microorganisms as model systems for studying evo-
lution. Plenum Press, New York, N.Y.
76. Naas, T., M. Blot, W. M. Fitch, and W. Arber. 1995. Dynamics of IS-related
genetic rearrangements in resting Escherichia coli K-12. Mol. Biol. Evol.
12:198–207.
77. Ninfa, A. J. 1996. Regulation of gene transcription by external stimuli,
p. 1246–1262. In F. C. Neidhardt, R. Curtis III, J. L. Ingraham, E. C. C. Lin,
K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H.
Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular
biology, 2nd ed., vol. 1. American Society for Microbiology, Washington, D.C.
78. Olivera, B. M. 1997. Conus venom peptides, receptor and ion channel
targets and drug design: 50 million years of neuropharmacology. Mol. Biol.
Cell 8:2101–2109.
79. Oparin, A. I. 1938. The origin of life. Macmillan, New York, N.Y. [Trans-
lated by S. Margulis.]
80. Pananicolaou, C., and L. S. Ripley. 1991. An in vitro approach to identifying
specificity determinants of mutagenesis mediated by DNA misalignments.
J. Mol. Biol. 221:805–821.
81. Panayotatos, N., and R. D. Wells. 1981. Cruciform structures in supercoiled
DNA. Nature 289:446–470.
82. Prescott, D. M. 1992. The unusual organization and processing of genomic
DNA in hypotrichous ciliates. Trends Genet. 8:439–445.
83. Primakoff, P. 1981. In vivo role of the relA⫹ gene in regulation of the lac
operon. J. Bacteriol. 145:410–416.
84. Primakoff, P., and S. W. Artz. 1979. Positive control of lac operon expres-
sion in vitro by guanosine 5N-diphosphate 3N-diphosphate. Proc. Natl.
Acad. Sci. USA 76:1726–1730.
85. Pruss, G. J., and K. Drlica. 1989. DNA supercoiling and prokaryotic tran-
Downloaded from jb.asm.org by on November 23, 2009
scription. Cell 56:521–523.
86. Rahmouni, A. R., and R. D. Wells. 1992. Direct evidence for the effect of
transcription on local DNA supercoiling in vivo. J. Mol. Biol. 223:131–144.
87. Rigby, P. W. J., B. D. Burleigh, Jr., and B. S. Hartley. 1974. Gene dupli-
cation in experimental enzyme evolution. Nature 251:200–204.
88. Ripley, L. S., and B. W. Glickman. 1983. Unique self-complementarity of
palindromic sequences provides DNA structural intermediates for muta-
tion. Cold Spring Harbor Symp. Quant. Biol. 47:851–861.
89. Rosche, W. A., L. S. Ripley, and R. Sinden. 1998. Primer-template misalign-
ments during leading strand DNA synthesis account for the most frequent
spontaneous mutations in a quasipalindromic region in Escherichia coli. J.
Mol. Biol. 284:633–646.
90. Rudner, R. A. Murray, and N. Huda. 1999. Is there a link between mutation
rates and the stringent response in Bacillus subtilus? Ann. N. Y. Acad. Sci.
870:418–422.
91. Saier, M. H., Jr., T. M. Ramseier, and J. Reizer. 1996. Regulation of carbon
utilization, p. 1325–1343. In F. C. Neidhardt, R. Curtis III, J. L. Ingraham,
E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M.
Schaechter, and H. Umbarger (ed.), Escherichia coli and Salmonella: cel-
lular and molecular biology, 2nd ed., vol. 1. American Society for Micro-
biology, Washington, D.C.
92. Schaaper, R. M., B. N. Danforth, and B. L. Glickman. 1986. Mechanisms of
spontaneous mutagenesis: an analysis of the spectrum of spontaneous mu-
tation in the Escherichia coli lacI gene. J. Mol. Biol. 189:273–284.
93. Selker, E. U. 1990. Premeiotic instability of repeated sequences in Neuro-
spora crassa. Annu. Rev. Genet. 24:579–613.
94. Sobell, H. M. 1972. Molecular mechanism for genetic recombination. Proc.
Natl. Acad. Sci. USA 69:2483–2487.
95. Sonti, R. V., and J. R. Roth. 1989. Role of gene duplication in the adapta-
tion of Salmonella typhimurium to growth on limiting carbon sources. Ge-
netics 123:19–28.
96. Stemmer, W. P. 1994. Rapid evolution of a protein in vitro by DNA
shuffling. Nature 370:389–391.
97. Stephens, J. C., S. W. Artz, and B. Ames. 1975. Guanosine 5N-diphosphate
3N-diphosphate (ppGpp): positive effector for histidine operon transcrip-
tion and general signal for amino acid deficiency. Proc. Natl. Acad. Sci.
USA 72:4389–4393.
98. Taddei, F., M. Radman, J. Maynard-Smith, B. Toupance, P. H. Gouyon,
and B. Godelle. 1997. Role of mutator alleles in adaptive evolution. Nature
387:700–702.
99. Todd, P. A., and B. W. Glickman. 1982. Mutational specificity of UV light
in Escherichia coli: indications for a role of secondary structure. Proc. Natl.
Acad. Sci. USA 79:4123–4127.
100. Torriani, A., and F. Rothman. 1961. Mutants of Escherichia coli constitutive
for alkaline phosphatase. J. Bacteriol. 81:835–840.
101. Vilette, D., M. Uzest, D. Ehrlich, and B. Michel. 1992. DNA transcription
and repressor binding affect deletion formation in Escherichia coli plas-
mids. EMBO J. 11:3629–3634.
102. von Hipple, P. H., and K.-Y. Wong. 1971. Dynamic aspects of native DNA
structures: kinetics of the formaldehyde reaction with calf thymus DNA.
J. Mol. Biol. 61:587–613.
103. Wächterhäuser, G. 1990. Evolution of the first metabolic cycles. Proc. Natl.
Acad. Sci. USA 87:200–204.
104. Walker, G. C. 1995. SOS-regulated proteins in translesion DNA synthesis
VOL. 182, 2000
and mutagenesis. Trends Biochem. Sci. 20:416–420.
105. Wang, J. C. 1985. DNA topoisomerases. Annu. Rev. Biochem. 54:665–697.
106. Weismann, A. 1893. Über die Vererbung. Gustav Fischer, Jena, Germany.
107. Wessler, S. R., and J. M. Calvo. 1981. Control of the leu operon expression
in Escherichia coli by a transcription attenuation mechanism. J. Mol. Biol.
149:579–597.
108. Woese, C. R. 1987. Bacterial evolution. Microbiol. Rev. 51:221–271.
109. Wright, B. E. 1996. The effect of the stringent response on mutations in
Escherichia coli K-12. Mol. Microbiol. 19:213–219.
MINIREVIEW 3001
110. Wright, B. E., and M. F. Minnick. 1997. Reversion rates in a leuB auxo-
troph of Escherichia coli K-12 correlate with ppGpp levels during expo-
nential growth. Microbiology 143:847–854.
111. Wright, B. E., A. Longacre, and J. M. Reimers. 1999. Hypermutation in
derepressed operons of Escherichia coli K12. Proc. Natl. Acad. Sci. USA
96:5089–5094.
112. Wu, T. T., E. C. C. Lin, and S. Tanaka. 1968. Mutants of Aerobacter
aerogenes capable of utilizing xylitol as a novel carbon. J. Bacteriol. 96:447–
456.
Downloaded from jb.asm.org by on November 23, 2009