Clinical Science (1997) 93,383-390 (Printed in Great Britain)
383
1997 Bayer Lecture
Spontaneous and induced minisatellite instability in the human
genome
Alec J. JEFFREYS
Deportment of Genetics, University of leicester, University Rood, leicester LEI 7RH, U.K.
1. Minisatellites, originally developed for forensic
DNA fingerprinting and profiling, have also provided extremely informative systems for analysing
processes of tandem repeat DNA turnover in the
human genome.
2. Minisatellite instability involves distinct mutation processes in somatic and germline cells. In
sperm, tandem repeat instability most likely arises
at meiosis as a by-product of high-frequency meiotic
recombination events occurring in and near minisatellites. Instability is not an intrinsic property of
minisatellite repeat DNA, but instead appears to be
controlled by flanking recombination initiation elements.
3. Minisatellites not only show high-frequency
spontaneous mutation in the germline, but also
appear to be very sensitive to mutation induction by
ionizing radiation, both in experimentally irradiated
mice and in human populations exposed following
the Chernobyl disaster; the mechanisms of mutation
induction by radiation remain enigmatic.
INTRODUCTION
The ultimate source of all variation in the human
genome is de novo mutation in the germline, resulting in sequence alterations that can range from
simple base substitutions through deletion, duplication and transposition to turnover at tandem repeat
loci. Unfortunately, most classes of mutation occur
so rarely that spontaneous mutation rates cannot be
directly measured by family analysis, and have to be
inferred indirectly from population diversity studies
or by the ascertainment of individuals with inherited
disease arising from de novo dominant mutation.
Such studies only give population-averaged estimates of genome instability, and no information on
individual mutation rates, precluding any detailed
analysis of mutation process and the identification of
intrinsic and environmental factors that can influence mutation rate.
This situation has recently changed with the discovery that some loci in the human genome can
show extraordinarily high rates of spontaneous mutation in the germline, enabling, for the first time,
the detailed dissection of mutation process and the
first steps towards identifying factors that control
mutation rate. These unstable loci are tandem
repeat minisatellites at which germline instability
results in the alteration of repeat copy number and
therefore the length of the tandem repeat array.
Minisatellites were originally developed as highly
informative ultravariable markers for linkage analysis, but rapidly swept to prominence with the realization that they provided a powerful new method
for individual identification.
A brief history of forensic DNA typing
Human gene characterization in the early 1980s
revealed the first examples of highly variable minisatellites composed of repeat arrays, typically several
kilobases in length and with repeats 10-60 bp long,
depending on the locus. The discovery of sequence
similarities between repeat units at different loci,
the so-called minisatellite ‘core’ sequence [l], made
it possible to develop hybridization probes capable
of detecting multiple highly variable loci to produce
a Southern blot profile that was so variable between
individuals that it could be legitimately termed a
DNA fingerprint [2]. DNA fingerprinting was first
developed in 1984, and within a year was being used
as a powerful new tool to resolve paternity and
immigration disputes; even to this day, DNA fingerprinting is used in thousands of family relationship
disputes each year, particularly through Cellmark
Diagnostics, a company established to provide a
commercial outlet for DNA typing.
psented as the 1997 b y e t bcrur~at the Spring M&hg dthe Madid Rtscvch
at the
I5 May 1997.
wordr: con&,
DNA firlgerprinting DNA profilig, ionking r a d i i , minisateJli instabilii, mutation, recombination.
Awrcvhbknr: MVR-PCR minisatellite variant repeat mappigby PCR; SP-PCR, small pod PCR;SIR, dmple tandem repeat.
cdkgc
h
d on~
304
A. J. Jeffreys
DNA fingerprinting also allowed, for the first
time, the systematic isolation of large numbers of
different human minisatellites, an essential prerequisite to any studies of the biology of these loci [3, 41.
Some minisatellites so isolated proved to be extraordinarily variable, with scores or even hundreds of
alleles differing in repeat copy number. These loci
not only proved to be efficient markers for gene
mapping and loss of heterozygosity studies, but also
provided the second approach to human identification, namely single-locus DNA profiling, which
yields simple two-band hybridization profiles rather
than the complex multi-band phenotypes generated
by DNA fingerprinting [4]. DNA profiling saw its
forensic debut in 1986 with the Enderby murder
case involving two young girls who had been raped
and murdered; semen DNA profiling resulted in the
release of a young man falsely accused of the
crimes, the first individual to be exonerated by DNA
typing [4], and the successful entrapment of the true
murderer following the world’s first DNA-based
man hunt. DNA profiling is now standard practice
in forensic laboratories worldwide, and despite
extensive scientific and legal controversy, has maintained its position as a robust and highly reliable
DNA typing technology.
The next revolution in DNA typing came in 1988
with the invention of user-friendly PCR driven by
thermostable DNA polymerase [5]. Numerous PCRbased DNA typing systems have been developed,
based on assays of base-substitutional polymorphisms in the nuclear genome, on sequence variation
in mitochondrial DNA, on the amplification of minisatellites and the like. PCR also allowed repeat
copy number polymorphism to be detected at the
much shorter microsatellite or simple tandem repeat
(STR) loci comprised of 2-5 bp repeats [6]. STR
markers have had a profound impact on linkage
mapping in humans and other species. The semiautomated multiplex genotyping systems so
developed have also proved to be directly transferable to forensic DNA analysis, greatly speeding up
sample processing and increasing sensitivity, and
allowing even trace or degraded DNA to yield its
secrets. Perhaps the most spectacular case resolved
by PCR was the identification of the skeletal
remains of Tsar Nicholas and his family by bone
DNA typing, using STR markers to establish a
family grouping at the grave site plus mitochondrial
DNA to link the remains to living matrilineal
descendants of the Romanovs, including Prince
Philip [7]. STR typing is also having a major impact
on body part identification in mass disasters, for
example the Aeroflot crash in Spitsbergen in 1996
[8], and in developing DNA databases of convicted
criminals [9]. In particular, the U.K. National DNA
Database was established in April 1995 and currently holds the DNA profiles of over 150000 individuals to provide a powerful tool for identifying
re-offenders and for establishing links between
apparently unrelated crimes, for example serial rape.
Minisatellite instability
DNA fingerprint analysis of families soon
revealed the occasional appearance of new mutant
bands in children that could not be attributed to
either parent [l]. This provided the first indication
that minisatellites can show significant levels of
instability, though not so great as to interfere with
family relationship testing. Systematic pedigree
analysis of different loci by DNA profiling revealed
a remarkable spectrum of instability, with germline
mutation rates varying across minisatellites from
immeasurably low ( < lop4per gamete) up to a maximum of 6% for minisatellite CEBl [lo, 111. Mutation rates correlated with population variability in
accordance with the neutral mutation/random drift
predictions [ 101, but did not correlate with features
such as repeat homogeneity and high repeat copy
number which are strong predictors of variability,
and therefore instability, at STR arrays [12]; indeed,
the most unstable minisatellite, CEBl, is characterized by relatively low repeat copy number alleles,
fairly long repeat units (37-42 bp) and very substantial repeat sequence heterogeneity, even between
repeats in the same allele [13].
These pedigree surveys established basic features
of the mutation process, in particular a tendency to
gain or lose relatively small numbers of repeats, and
with a preferential bias to instability in the male
rather than female germline. However, they provided no information on the mutation process or the
timing of mutation during germ cell development. In
particular, studies of allele length changes cannot
distinguish between the two classic models of repeat
instability, namely replication slippage, resulting in
intra-allelic gain or loss of repeats, and recombination-based processes, such as unequal cross-over
between misaligned alleles at meiosis. Furthermore,
mutant recovery from pedigrees is tedious and gives
no information on the mutational behaviour of individual alleles. It was therefore necessary to develop
new methods both for the efficient mass isolation of
mutants and for the detailed structural characterization of alleles before and after mutation.
Detection of mutants by single molecule PCR
Mutation detection in pedigrees uses the genotype
of a mutant child to identify whether the sperm or
egg giving rise to the offspring carried a mutation.
In theory, offspring can be by-passed by direct
analysis of mutants in gametic DNA, using the
remarkable efficiency and fidelity of PCR to amplify
minisatellites at the single molecule level. Unfortunately, human oocytes are not readily obtainable,
leaving pedigrees as the only current source of
maternal mutants. In contrast, a single ejaculate of
semen can contain lo8 or more sperm, providing, in
principle, a virtually limitless source of new mutant
molecules derived from a given donor.
There are several approaches to detecting new
Minisatellite instability in the human genome
mutant alleles in single sperm. Conceptually the
simplest is to deliver individual sperm to a PCR
reaction, followed by amplification of the allele in
each sperm to determine whether it is mutant in
length. This approach is, however, cumbersome. A
more efficient method is to dilute bulk sperm DNA
into multiple PCR reactions such that each reaction
contains about 100 sperm equivalents of DNA,
amplification followed by gel electrophoretic analysis
of PCR products will reveal whether each small pool
of sperm DNA contains any mutant molecules
amongst the hundred or so progenitors. This small
pool PCR (SP-PCR) approach [14] has proved to be
very efficient and remarkably free of PCR artefacts
(Fig. l), and can be used to measure mutation rates
as low as
per cell. SP-PCR can be applied to
any source of genomic DNA, not just sperm. For
lower mutation rates, it is possible to use gel electrophoretic fractionation of restriction-digested bulk
genomic DNA to separate mutant molecules from
progenitors before mutation detection by SP-PCR
[15]. This approach is more laborious, but can yield
quantitative mutation rate data at rates as low as
iO-5-iO-6
per cell.
Single molecule PCR has been used to characterize mutation at several human minisatellites. Mutant
molecules are far more frequent in sperm that in
input molecules
A0
II
1fin
I
no.
repeats
I
E40
0
*
Fig. I. Detecting de novo mutant minisatellite molecules in sperm
DNA by SP-PCR. Sperm DNA from a donor heterozygous for 29- and
200-repeat alleles was diluted, and 0.5 or I ng of DNA was added to each
PCR reaction, corresponding to 80 or 160 amplifiable molecules of each
allele. Minisatellite molecules were amplified using primers located 5' and 3'
to the repeat array, and PCR products were analysed, by agarose gel
electrophoresis and Southern blot hybridization, to detect products derived
from the 29-repeat progenitor allele and from mutant molecules with
altered repeat copy number [14]. Most mutants involve gains or losses of
small numbers of repeats, with an excess of gain mutants. The 200-repeat
allele is too large to be analysed by SP-PCR.
385
somatic (blood) DNA, indicating that instability is
largely restricted to the germline [14, 151. Singlesperm mutation rates are in accord with paternal
rates measured in pedigrees [14, 161, and are therefore not significantly perturbed by excess mutation
in a subset of infertile sperm that do not contribute
to the next generation. All loci show a bias towards
gaining rather than losing repeats [14, 161; this gain
bias greatly accelerates the evolutionaly rate of
minisatellite expansion, and helps explain why the
most variable and unstable human loci are evolutionary newcomers in the human genome, arising
within the last few million years after the divergence
of humans and great apes [17]. A major question is
how such gain-biased loci are prevented from indefinite expansion, thus swamping the human genome.
Minisatellite mutation shows a curious dependence on allele length, with evidence .for a threshold
of instability between 10 and 15 repeats, above
which mutation occurs at a maximal rate, irrespective of allele length [14, 181. Similar thresholds have
been seen in triplet repeat expansion disorders, but
may have an entirely different mechanistic basis (see
~91).
Structural analysis of minisatellite alleles
Minisatellite alleles are rarely, if ever, sequencehomogeneous, but instead are composed of two or
more variant repeat types present along the repeat
array [l, 41. These patterns of variant repeats can be
read off an allele using variant repeat-specific PCR
primers (minisatellite variant repeat mapping by
PCR, MVR-PCR) to provide detailed internal structural information on an allele [20].
MVR-PCR has been developed for several minisatellites, to reveal that variant repeats are common
and are almost always heavily intermingled with
each other along an allele (Fig. 2), implying that the
germline mutation process must actively scramble
the order of repeat units. MVR-PCR has also
revealed extraordinary levels of allelic diversity, with
los or more different and distinguishable alleles
present at some loci in the current world population
[20]. It also enables different but closely related
groups of alleles to be identified (Fig. 2), providing
a method for tracing allele lineages and human origins [21]. Most loci also show the curious phenomenon of polarized variability [20], whereby most
differences between closely related alleles are
restricted to one end of the repeat array (Fig. 2);
this in turn implies that germline repeat turnover
does not occur at random along the array but is
instead restricted to the end showing maximal population variability.
The structural basis of minisatellite mutation
MVR-PCR has been used to characterize somatic
(blood) mutants and germline mutants recovered
a
b
C
--
A. j. jeffreys
d
e
f
F
l
9
Fig. 2. Two levels of variation in human minisatellites: array length
and internal allelic structure. Examples are shown of minisatellite MS32
alleles characterized by two-state MVR-PCR [20] to distinguish two repeat
types (white, grey). The alleles shown constitute a closely related group
identified from a set of I100 different MS32 alleles mapped to date. These
related alleles show typical interspersion of variant repeats plus polar variability, with most variation between aligned alleles restricted to the beginning
of the repeat array. These alleles also fall into two subgroups (a-c, d-g)
based on shared internal structural features, showing how allele structure
can be used to identify allele lineages and sublineages.
from pedigrees or by single sperm analysis. Somatic
and sperm instability have a completely different,
and apparently mutually exclusive, structural basis
[14, 151. Low-frequency blood mutants arise by
simple intra-allelic duplication or deletion of blocks
of repeat units located at random along the repeat
array; such events could arise by replication slippage
or by recombination events such as unequal sister
chromatid exchange [15]. In stark contrast to the
simple blood mutants, sperm mutants can be astonishingly complex and show features completely
incompatible with replication slippage or simple
recombination events [13, 141. Instead, sperm mutation is dominated by the inter-allelic transfer of
blocks of repeat units from one allele (the donor) to
the other (the recipient) in a process strongly reminiscent of gene conversion (Fig. 3). For loci showing
polar variation, these transfers are, as predicted,
almost completely restricted to the end of the allele
showing maximal variability and, further, tend to
occur in register between the alleles, implying that
donor and recipient alleles are paired over the
unstable end before repeat transfer [14]. Furthermore, both alleles in an individual mutate by the
same gain-biased process, implying that the donor
allele is not permanently altered by a mutation
event and therefore that donor information is copied before transfer to the recipient [14, 221. While
some transfers are simple, the majority are complex
(Fig. 3), with rearrangements including deletions or
duplications in the recipient allele, scrambling of
repeat units during transfer, and imperfect reduplications of donor-recipient junctions. In some cases,
these rearrangements can be so extreme as to profoundly remodel allele structure in a single, astonishingly complex, sperm mutation event [14].
Minisatellites and meiotic recombination
These complex sperm conversion events, though
recombinational in nature, are restricted to the
repeat array and do not result in true recombination
events with exchange of markers flanking the minisatellite (Fig. 3). To determine whether there is any
relationship between minisatellite conversion and
meiotic recombination, SP-PCR-type methods have
been developed which allow the selective amplification of recombinant sperm molecules from semen
donors showing multiple base substitutional heterozygosities in the DNA flanking the repeat array
(A.J. Jeffreys, unpublished work). Analysis of the
best characterized human minisatellite, MS32, has
shown that true meiotic recombination events do
arise by inter-allelic cross-over within the repeat
array, but at a much lower frequency than gene conversion (Fig. 3). Unequal and equal (in register)
cross-overs both occur at similar frequencies, to
yield simple recombinant alleles seldom showing any
evidence of the complex rearrangements characteristic of conversion. Recombination breakpoint
progenitor alleles
1
B
C
Fig. 3. Representative examples of minisatellite MS32 mutant structures detected in sperm. The sperm donor is heterozygous for allele I
(200 repeats) and allele 2 (27 repeats). Internal structures were established
by &state MVR-PCR [36], enabling four different repeat types to be identified; the central region of allele I is omitted for clarity. Alleles I and 2 also
show multiple heterozygosities in the flanking DNA (circles). (A) Examples
of inter-allelic conversion events in allele 2 without exchange of flanking
markers: these are the predominant mutation products and arise at a frequency of 0.8% per sperm. Mutant 3 shows simple polar in-phase transfer
of repeats from the donor allele I without gain or loss of recipient repeats.
Mutant 4 shows in-phase transfer plus acquisition of additional repeats
derived from the recipient allele. Mutant 5 is complex, involving imperfect
duplication of a donor-recipient junction. (B) An example of inter-allelic
unequal cross-over within the repeat array, resulting in exchange of flanking
markers and a simple recombinant array. (C) An example of equal crossover leading to a recombinant allele equal in length to progenitor allele 2.
Unequal and equal recombinant molecules occur much less frequently than
conversion products.
Minisatellite instability in the human genome
polarity (Fig. 3) and asymmetries in recombination
events point to a fundamental connection between
conversion and meiotic recombination; the unstable
end of minisatellite MS32 is thus a site both of
intense conversional activity and of elevated meiotic
recombination. Curiously, this meiotic recombination hotspot is not restricted to the repeat array, but
appears to extend out into the adjacent flanking
DNA for a distance of perhaps 500-1000 bp. Thus,
MS32 appears to be located at the boundary of a
recombination hotspot.
Flanking DNA and minisatellite mutation
Polarized instability and the flanking recombination hotspot both point to DNA flanking the
-mmmmmmmmmmmmm=
- m ~ ~ m m m m m t--conversion
387
unstable end of the repeat array as having a major
influence on repeat instability. The simplest model
(Fig. 4) is that the flanking DNA contains elements
necessary to initiate recombination and conversion;
initiation could occur via the introduction of a
double-strand break or staggered nicks into the
recipient allele, followed by pairing with the other
allele and strand invasion, to produce a recombination complex. In most instances, and for unknown
reasons, this complex is aborted to yield a conversion product; the abortion process must be complex
to give the extraordinary rearrangements that can
accompany conversion. Occasionally, the recombination complex is correctly processed to yield a true
meiotic recombinant.
This model predicts the existence of a recombina-
-
-mimmwmm~mmmm-r!aWmmmmmmmmirecombination
Fig. 4. Model for the co-initiation of minisatellite conversion and recombination, based on the double-strand break repair model for recombination in yeast
[37l. The allele destined to become the conversion recipient is activated for mutation by introduction of a double-strand break or staggered nicks near the beginning of
the array (I);this activation is controlled by a flanking initiator (blue). The break expands to a gap and the recipient allele pairs in register with the donor (2). The gap is
bridged by strand invasion from the donor allele (3). For conversion, this bridge provides a template for repair synthesis (blue arrow) (4). The donor strand is then
extruded (S), leaving the donor allele intact, and the resulting single strand gap in the recipient is repaired (blue arrow) to yield a simple conversion product. For
recombination, the strand invasion complex is isomerired (pink arrow) and the Holliday junctions resolved (purple arrows) (6).followed by repair synthesis (7) to yield a
recombinant allele. Only one of the two recombinant products presumed to arise from this process is shown.
388
A. J. Jeffreys
tion/conversion initiator, adjacent to the repeat
array, which controls tandem repeat instability. If
correct, then variation in mutation rates across different loci could reflect differences in initiator efficiency. Two lines of evidence support this model.
First, instability does not increase with array length
above the instability threshold, as predicted for a
flanking regulator that can only address the proximal part of the repeat array; the existence of a
repeat threshold of instability may indicate a minimal array size requirement for the production of a
recombination/conversion complex. Second and
more significantly, a single G - + Cbase transversion
has been discovered 48 bp upstream of the unstable
end of MS32 within the flanking recombination hotspot; C-linked alleles show a profound (up to
100-fold) reduction in conversion rate, not only
associated with, but apparently directly caused by,
this substitution [22]. The C variant suppresses
instability but does not alter the residual conversion
process [15], suggesting a block in conversion initiation rather than down-stream processing. This block
operates in cis, but does not prevent a C-linked
allele from acting as a conversion donor to an
unstable G-linked allele, suggesting that donor and
recipient alleles are functionally distinct during conversion (Fig. 4). Moreover, the C variant also suppresses meiotic unequal cross-over with a G-linked
allele, providing further evidence that conversion
and recombination are products of the same recombination initiation process. Finally, the C variant
does not affect somatic (blood) instability [15], again
indicting a fundamental difference between germinal
and somatic instability. The C variant thus provides
strong evidence that instability is not an intrinsic
property of the repeat array, but is instead conferred
upon the array by a local recombination/conversion
initiator element disrupted by the C variant.
Modelling in mice
Further analysis of flanking instability regulators
will be of considerable interest in defining basic
aspects of chromosome pairing and the initiation of
recombination in human meiosis, and would be
greatly facilitated by a more experimentally tractable
animal model. However, attempts to transfer minisatellite MS32 sperm instability into transgenic mice
have so far completely failed, although the cosmidbased constructs tested to date have shown successful transfer of somatic instability, as judged by the
frequency and spectrum of MS32 blood mutants
seen in transgenic mice [23]. The reasons for complete MS32 sperm stability in mice is unclear, but
might point to a fundamental difference in the way
that humans and mice process tandem repeat DNA
at meiosis. Preliminary analyses of endogenous
mouse minisatellites supports this possibility, with
no clear evidence as yet for loci showing high-level
germline instability plus polar variation/mutation/
conversion that frequently characterize the most
unstable human minisatellites. Further analyses are
underway to define the nature of tandem repeat
turnover processes that do operate in the mouse
germline.
Minisatellite mutation and ionizing radiation
It has long been known that ionizing radiation
induces both somatic and germinal mutation. However, the lack of efficient mutation monitoring
systems and of appropriate exposed human populations, other than the survivors of the Japanese atom
bombs, have resulted in considerable uncertainty
about the effects, if any, of environmental radiation
on germline mutation in man. Minisatellites currently provide the only loci appropriate for the efficient monitoring of germline mutation in human
populations. We have therefore investigated the
potential of minisatellites for monitoring radiationinduced germinal mutation.
Mutation and radiation in mice
Exposure of male mice to y radiation, followed by
breeding and the monitoring of mutations in offspring by multilocus DNA fingerprinting [24], has
provided strong evidence for radiation-induced
germline mutation, with a doubling dose of approximately 0.5 Gy, a dose similar to that required to
double mutation rate in classic mutation monitoring
systems that require the analysis of very large
numbers of offspring. There is uncertainty about the
relative radiosensitivity of different stages of spermatogenesis, though recent data from this laboratory suggest that pre-meiotic spermatogonia are
particularly sensitive. Most induced mutations arise
at the intrinsically most unstable loci detected in
mouse DNA fingerprints; these loci are not minisatellites but instead are highly expanded tetra- and
penta-nucleotide STR arrays [24, 251. Curiously, the
absolute level of induced mutation is far higher than
would be predicted if mutation arises directly as a
consequence of radiation-induced damage, such as
double-strand breaks, that happen to occur within
tandem repeat DNA [25]. Instead, there appears to
be an indirect induction mechanism whereby radiation is sensed (by monitoring DNA damage?) and in
some way transduced into a (meiotic?) signal that
somehow destabilizes tandem repeat loci. If correct,
this would represent a completely novel mechanism
for radiation mutagenesis. Unfortunately, the DNA
fingerprint and STR systems used to monitor mutation in mice cannot be used to investigate the
developmental timing or mechanisms of induced
mutation.
The Chernobyl disaster
The catastrophic explosion of reactor 4 at the
Chernobyl nuclear power station on 26 April 1986
led to the accidental release of some 5 x lo7 Ci of
assorted radionuclides which resulted in widespread
Minisatellite instability in the human genome
pollution, particularly in areas of the former Soviet
Union. The immediate health risk to the general
population was from the emission of 1 x lo7 Ci of
short-lived 1311, resulting in a significant thyroid dosage whose consequences are now becoming
apparent in terms of a substantial increase in the
incidence of thyroid cancer. Longer-term exposure
has arisen mainly from environmental pollution with
137Cs(half-life 30 years), which caused persistent
ground contamination as well as entering the food
chain. The biological consequences of chronic
environmental 137Cs contamination for humans,
including the effects on germline mutation, are
unknown.
Minisatellite germline mutation was therefore
studied in a cohort of 127 babies born during 1994
to parents continuously resident in polluted areas of
the Mogilev district of Belarus, some 300 km to the
north of Chernobyl. Analysis using DNA fingerprinting plus eight different single-locus minisatellite
mutation monitoring systems, chosen for their high
spontaneous-mutation rates, showed a highly significant (P <0.0001) two-fold increase in mutation rate
over most loci in the Belarus families, compared
with unexposed families from the U.K. ([26];
Y. Dubrova and A.J. Jeffreys, unpublished work).
This represents the first evidence for a systematic
genome-wide shift in mutation rate between two different human populations, but does not necessarily
implicate environmental radiation as the inducer.
However, dose-response analysis within the Belarus
families, using estimates of individual parental dose
deduced from the levels of environmental 137Cscontamination in the locale of each family, showed a
significant tendency for the highest mutation rates to
occur in the most exposed families. These data,
which should be regarded as suggestive but not definitive, are consistent with a radiation origin of
induced mutation and, if correct, provide the first
direct evidence for radiation-induced germline mutation in humans. Similar, though more limited,
studies have been carried out on the children of the
Hiroshima and Nagasaki survivors with, interestingly, no evidence for induced mutation, despite
much higher gonadal doses than those estimated for
the Belarus parents [27]. One obvious interpretation
is that there may be different biological consequences of a single acute radiation dose received on
atomic bomb detonation compared with chronic
environmental exposure.
Concluding remarks
These studies establish, for at least some modes
of genomic instability, that detailed mechanistic
studies of germline mutation are now possible in
humans. Minisatellites provide the most detailed
picture to date of processes of tandem repeat
instability, which are now fairly conclusively established to be a by-product of meiotic recombination.
389
Most minisatellite mutation is likely to be without
phenotypic consequence, though there are exceptions, such as the insulin- and c-H-ras oncogeneassociated minisatellites that act as risk factors for
diabetes and various cancers respectively [28, 291.
Instances of minisatellite array expansion, causing in
one case myoclonus epilepsy [30] and in another a
chromosomal fragile site [31], have also been identified, though the mechanistic basis of expansion
remains unclear. The most dramatic examples of
array expansions causing disease are dynamic mutations operating at trinucleotide repeat arrays in
neurological disorders such as fragile X mental
retardation and Huntington’s disease (see [19]).
While most models for triplet repeat instability
focus on replication slippage promoted by hairpin
DNA [32], unanchored Okazaki fragments [33],
DNA flaps at replication forks [34] and the like, the
detailed mechanisms remain obscure, and a recombinational component to trinucleotide array instability cannot yet be excluded. Indeed, studies of the
(CAG),, triplet array in the Machado-Joseph
disease locus have shown that germline mutation of
one allele can be influenced by its allelic partner,
implying some sort of recombinational inter-allelic
cross-talk in the mutation process [35].
Finally, evidence is now accumulating that minisatellites may be very sensitive to radiation-induced
germline mutation in both mice and humans. In
both cases, the scale of induction implies some
unexpected and mysterious mechanism whereby
radiation indirectly results in mutation at tandem
repeat loci. It is not possible to extrapolate these
findings to other, more biologically relevant, classes
of mutation in the human genome, nor to infer any
adverse health consequences to descendants of the
post-Chernobyl population. The current data do,
however, highlight our ignorance of the effects of
chronic radiation on the human germline, and raise
issues ranging from specific questions about mechanisms of radiation mutagenesis to general issues of
the biological consequences of environmental radiation.
ACKNOWLEDGMENTS
I am very grateful to numerous colleagues at
Leicester for their help, support and encouragement
over the years. The work described in this review
was supported by grants from the Medical Research
Council, Wellcome Trust, Howard Hughes Medical
Institute and Royal Society. A.J.J. is an International Research Scholar of the Howard Hughes
Medical Institute.
REFERENCES
I. Jeffreys AJ, Wilson V, Thein SL. Hypervariable‘minisatellite’ regions in human
D N A Nature (London) 1985; 3 14: 67-74.
390
A. J.Jeffreys
2. JeffreysAJ, Wilson V, Thein SL. Individual-specific 'fingerprints' of human DNA
Nature (London) 1985; 316 76-7.
3. NakamuraY, Leppert M, OConnell P, et al. Variable number of tandem repeat
(VNTR) markers for human gene mapping. Science 1987; 235: I616-22.
4. Wong Z, Wilson V, Patel I,Povey S, Jeffreys A]. Characterizationof a panel of
highly variable minisatellites cloned from human D N A Ann Hum Genet 1987;
5 I: 269-88.
5. Saiki RK, Gelfand DH, Stoffel S, et al. Primerdirected enzymatic amplificationof
DNA with a thermostable DNA polymerase. Science 1988 239: 487-9 I.
6. Weber JL, May PE. Abundant class of human DNA polymorphisms which can be
typed using the polymerase chain reaction. Am J Hum Genet 1989; 44:388-96.
7. Gill P, lvanov PL, Kimpton C, et al. Identification of the remains of the Romanov
family by DNA analysis. Nature Genet. 1994; 6 130-35.
8. Olaisen B, Stenersen M, Mevag 8. Identification by DNA analysis of the victims of
the August 1996 Spitsbergen civil aircraft disaster. Nature Genet 1997; IS:
402-5.
9. Gill P. Urquhart A, Millican E, et al. A new method of STR interpretationusing
inferential logic: development of a criminal intelligence database. Int J Legal Med
1996; 109 14-22.
10. Jeffrey AJ, Royle NJ.Wilson V, Wong Z Spontaneous mutation rates to new
length alleles at tandem-repetitive hypervariable loci in human DNA. Nature
(London) I988 332 278-8 I.
I I . Vergnaud G, Mariat D. Apiou F, Aurias A, Lathrop M, Lauthier V. The use of
synthetic tandem repeats to isolate new VNTR loci: cloning of a human
hypermutable sequence. Genomics 1991; I I: 135-44.
12. Weber JL. Informativeness of human (dC-dA)n.(dGdT). polymorphisms.
Genomics 1990 7: 524-30.
13. Buard J, Vergnaud G. Complex recombination events at the hypermutable
minisatelliteCEBl (D2S90). EMBO J 1994; 1 3 3203-10.
14. JeffreysAJ, Tamaki K. MacLeod A, Monckton DG, Neil DL, Armour JAL.
Complex gene conversion events in germline mutation at human minisatellites.
Nature Genet 1994; 6 136-45.
15. JeffreysA]. Neumann R Somatic mutation processes at a human minisatellite.
Hum Mol Genet 1997; 6 129-36.
16. May CA,JeffreysAJ, Armour JAL. Mutation rate heterogeneity and the
generation of allele diversity at the human minisatellite MS205 (D165309).Hum
Mol Genet 1996; 5: 1823-33.
17. Gray IC, JeffreysAJ. Evolutionarytransience of hypervariable minisatellites in man
and the primates. Proc R SOC London Ser B 1991; 243 241 -53.
18. Jeffrey AJ, Bois P, Buard J, et al. Spontaneous and induced minisatellite
instability. Electrophoresis 1997; 18 in the press.
19. Sutherland GR Richards RI. Simple tandem repeats and human genetic disease.
Proc Natl Acad Sci USA 1995; 9 2 3636-4 I.
20. JeffreysAJ, MacLecd A, Tamaki K, Neil DL, Monckton DG. Minisatelliterepeat
coding as a digital approach to DNA typing. Nature (London) I99 I; 354 204-9.
21. Armour JAL, Anttinen T, May CA, et al. Minisatellite diversity supports a recent
3 154-60.
African origin for modern humans. Nature Genet I996 1
22. Monckton DG, Neumann R Guram T, Fretwell N, Tamaki K, MacLeodA,
Jeffreys AJ. Minisatellite mutation rate variation associated with a flanking DNA
sequence polymorphism. Nature Genet 1994; 8 162-70.
23. Bois P, Collick A, Brown J, Jeffreys A]. Human minisatellite MS32 (DI58) displays
somatic but not germline instability in transgenic mice. Hum Mol Genet 1997; in
the press.
24. Dubrova YE, Jeffreys AJ, Malashenko AM. Mouse minisatellitemutations induced
by ionizing radiation. Nature Genet 1993; 5: 92-4.
25. Sadamoto S, Suzuki S, Kamiya K, Kominami R, Dohi K, Niwa 0. Radiation
induction of germline mutation at a hypervariable mouse minisatellite locus. lnt
J Radiat Biol 1994; 65: 549-57.
26. Dubrova Y, Nesterov VN, Krouchinsky NG, Ostapenko VA, Neumann R Neil
DL, JeffreysAJ. Human minisatellite mutation rate after the Chernobyl accident.
Nature (London) 1996; 380: 683-6.
27. Kodaira M, Satoh C, Hiyama K,Toyama K Lack of effects of atomic bomb
radiation on genetic instability of tandem-repetitive elements in human germ
cells. Am J Hum Genet 1995; 57: 1275-83.
28. Bennett ST, Lucassen AM, Gough SCL, et al. Susceptibility to human type-I
diabetes at lDDM2 is determined by tandem repeat vatiation at the insulin gene
minisatellite locus. Nature Genet 1995; 9 284-92.
29. Krontiris TG, Devlin B, Karp DD, Robert N, Risch N. An association between
the risk of cancer and mutations in the HRASl minisatellite locus. New England
J Med 1993; 329 517-23.
30. Lalioti MD, Scott HS, Buresi C, et al. Dodecamer repeat expansion in cystatin B
gene in progressive myoclonus epilepsy. Nature (London) 1997; 386 847-5 I.
3 I. Yu S, Mangelsdorf M, Hewett D, Hobson L, et al. Human chromosomal fragile
site FRA16B is an amplified AT-rich minisatellite repeat. Cell 1997; 8 367-74.
32. Mariappan SVS, Catasti P, Chen X, et al. Solution structures of the individual
single strands of the fragile X DNA triplets (GCC),.(GGC),. Nucleic Acids Res
1996; 24 784-92.
33. Richards RI, Sutherland GR Simple repeat DNA is not replicated simply. Nature
Genet 1994; 6 114-6.
34. Gordenin DA, Kunkel TA, Resnick M A Repeat expansion: all in a flap? Nature
Genet 1997; 16 116-8.
35. lgarashi S, TakiyamaY, Cancel G, et al. lntergenerationalinstability of the CAG
repeat of the gene for Machado-Joseph disease (MIDI) is affected by the
genotype of the normal chromosome: implicationsfor the molecular mechanisms
of the instability of the CAG repeat. Hum Mol Genet 1996; 5 923-32.
36. Tamaki K, Monckton DG, MacLeod A, Allen M, Jeffreys A]. Four-state MVR-PCk
increased discrimination of digital DNA typing by simultaneous analysis of two
polymorphic sites within minisatellitema
' nt repeats at DIS8. Hum Mol Genet
1993; 2 1629-32.
37. SzostakJW, Orr-Weaver TL, Rothstein RJ, Stahl FW. The double-strand-break
repair model for recombination. Cell 1983; 33 25-35.