Note to the Reader
Text file started Sept 1, 1998. This version updated July 3, 2000.
Please refer to: Williams RW (1999) A targeted screen to detect recessive
mutations that have quantitative effects. Mammalian Genome 10:734–738.
A Targeted Screen to Detect Recessive Mutations that have Quantitative
Robert W. Williams
Center for Neuroscience and the Department of Anatomy and Neurobiology,
University of Tennessee, Memphis TN 38163
Materials and Methods
Figure 1: The Mutagenesis
Genotyping G2 Females
Recessive Mutations in QTLs
Colony Costs of the Screen
Objections to a
Other Targeted Mutagenesis Tactics
Related Links on
Articles in PubMed
Abstract. Chromosome substitution strains have been used for
more than four decades to isolate, map, and characterize recessive
mutations in Drosophila. This paper describes an experimental cross
in mice that can be used to define and map induced germ-line mutations
that map to a single chromosome. The cross is a modification and extension
of a conventional three-generation recessive mutagenesis screen. It begins
with an inbred consomic strain that carries the desired target chromosome.
Consomic males are mutated and bred back to the background strain. G1
heterozygous males are also crossed to the background strain. Second
generation male and female progeny that inherited non-recombinant target
chromosomes are identified by genotyping and are bred. The third
generation progeny are genotyped for multiple markers on the target
chromosome, but the aim in this generation is to identify males and
females that are homozygous for the same short interval of the target
chromosome (5 to 25 cM). These near-congenic pairs are bred. Multiple
litters of fourth generation animals that are homozygous for known
segments of the target chromosome are phenotyped. Mutations with
relatively small and somewhat variable effects can be detected and then
mapped to a precision as high as 5–10 cM.
The genome is a robust code. Not only do
diploid organisms have two copies of most genes, but many genes also have
closely related variants, or paralogs that serve similar functions (Nadeau
and Sankoff, 1997). Even when both alleles of a given gene are mutated,
the phenotypic repercussions can be subtle. This is what Sewall Wright
(1977) had in mind when he commented that sets of related genes may be
"maintained by selection for quantitative effects, while protected from
drastic effects of inactivation or loss." It follows that an important
problem in a systematic screen for mutations is to devise protocols
capable of detecting more subtle variants. Without sufficient sensitivity,
mutant alleles in a significant fraction of genes may never be detected.
Screens will lack power. In contrast, mutations in genes that lack backup
may be detected many times. To cite an extreme example, a mutagenesis
screen in C. elegans recently produced 63 novel mutations in a single
mechanoreceptor gene (Chalfie and Au, 1989; García-Añoveros and Corey,
1997). In contrast, mutations and knockouts of numerous important genes,
for example, the retinoic acid receptors, often have surprisingly modest
effects that would elude most screens (Luo et al., 1995; Mendelsohn et
al., 1994; Zhou et al., 1998). In this paper I describe a promising
technique designed with this problem in mind. The technique relies on a
novel type of mouse called a consomic mouse and on a four-generation
breeding protocol (Fig. 1).
A consomic mouse is a fully inbred animal in which a single intact
pair of homologous chromosomes from an inbred donor strain has been
transferred into the genome of a fully inbred recipient strain (Fig. 1,
top; Hudgins et al., 1985). If chromosome 7 from donor strain A is
introgressed into recipient strain B, the resulting animals are designated
B-7A consomic mice. Several consomic strains have been generated by
Jean-Louis Guénet and Maria Rosa Arnau (PWK and SPRET chromosomes
introgressed into C57BL/6). We are using the C57BL/6-19PWK strain to begin
a mutagenesis screen of Chr 19 (RW Williams, JL Guénet, D Goldowitz, in
progress). A complete set of consomic strains in which chromosomes from
A/J have been transferred to C57BL/6J are now being generated by J Nadeau,
J Singer, and E Lander (personal communication). With this set of strains
any chromosome can become the focus of a recessive screen. Although, the
particular screen outlined here is just now being tested in mice,
chromosomal substitution lines of Drosophila with similar properties have
been used very effectively to isolate mutations with quantitative effects
on viability, sensory bristle number, behavior, and body weight (e.g.,
Lyman et al., 1996; Anholt et al., 1996; Currie et al., 1998).
Key features of this method are:
- Entire litters of homozygotes are examined, not isolated individuals
within litters. A large number of mutants can be generated in the first
round of analysis. Animals with recessive mutations can be studied at
several ages or can be sent to multiple investigators. The analysis of
entire litters increases the power of the screen.
- Mutations are screened on a relatively homogeneous genetic
background. This increases the sensitivity of the screen. Control lines
for comparative analysis are easily generated.
- Only individuals known to be non-recombinant on defined intervals of
a target chromosome are bred, and only homozygous progeny are phenotyped.
This increases the overall efficiency of phenotyping. This selectivity
relies on highly efficient genotype assays.
- Mutations are mapped to specific intervals of a target chromosome
without extensive genotyping or additional rounds of breeding and
phenotyping. Resolution of 5–20 cM is achieved in the first pass. This
increases the efficiency with which mutations are mapped and,
- Embryonic lethal alleles are readily detected and simultaneously
mapped in the form of “missing homozygotes.” Lethal alleles can be
recovered and studied by crossing heterozygotes.
Figure 1: Design of a Consomic Recessive Screen
Download a color high-resolution PDF version of this figure.
Consomic mutagenesis cross. The upper-left box represents the genome of a
fully inbred consomic male mouse (4 of 19 autosome pairs are illustrated).
Three of the consomic male's chromosomes are identical to those of the
recipient strain female (upper-right box). The exceptional chromosome
(black target chromosome) is derived from a donor strain that is not
illustrated. The original male (G0) is mutagenized with ENU and is crossed
to a recipient strain female. The small box (center-left) represent a
single gamete of the mutagenized male. Eight mutations are indicated using
small open circles, two of which are on-target, six of which are
off-target. The G1 male (A) is also crossed to a recipient strain female.
G2 progeny are genotyped and non-recombinant individuals (e.g., animal B)
are identified and bred. Note that the number of off-target mutations
carried by the G2 male (B), is half that carried by the G1 male (labeled
A). The diagonal line marked backcross dilution represents an optional
step that can be used one or more times to sequentially dilute off-target
mutations. G3 progeny are genotyped and pairs of animals that are
homozygous on common intervals of the target chromosome are mated. In this
example, the G3 breeders share a homozygous interval that extends from
approximately 15 to 40 cM. A set of litters of G4 animals that have a
comparatively uniform genome are phenotyped to identify and map recessive
mutations on the target chromosome. Note that the figure only illustrates
one of a potentially large number of such G4 test litters; each homozygous
on different parts of the target chromosome.
Methods and Materials
The key starting material for a targeted recessive screen is one or
more consomic strains of mice. These strains are produced by introgressing
entire chromosomes derived from an inbred donor into an inbred recipient
strain (Hudgins et al., 1985). Consomic lines are also referred
substitution lines or insertion lines because an exogenous pair of
homologous chromosomes is substituted for the usual pair. Several consomic
strains have been successfully generated by Jean-Louis Guénet and
colleagues (SPRET/Ei chromosomes moved onto a C57BL/6 background; JL
Guénet, personal communication). A complete set of consomic strains in
which A/J chromosomes have been transferred to a C57BL/6J background are
now being generated by Joseph Nadeau, Jonathan Singer, and Eric Lander
(1998). Consomic mice in which Chr 7 is derived from A/J have the
formal designation C57BL/6J-7A/J (B.A7 for
short). I will use the B.A7 strain for illustration in this paper, but any
consomic or chromosome substitution strain can be used.
Consomic males are mutagenized with N-ethyl-N-nitrosourea (ENU) or
another mutagen (Fig. 1, top left). Those animals that regain fertility
are mated with several females belonging to the recipient strain used to
make the consomic line (inbred females, top right). ENU is a potent
mutagen of spermatogonia. Each sperm will carry a load of mutations. Tests
of specific gene loci suggest that the right dose of ENU can mutagenize up
to one gene in a thousand (Russell et al., 1979; 1982; Rinchik et al.,
1990). Chr 7 makes up roughly 5% of the mouse genome, and single gametes
produced by a B.A7 male will harbor an average of five of the
approximately 100 mutations generated randomly across the genome. The
great majority of mutations (>90%) carried by the G1 male will be
recessives (Kacser and Burns, 1981; Lynch and Walsh, 1998, pp. 63–65).
Generating and Genotyping G2 Females
Four chromosomes are depicted in Figure 1—the target and three other
chromosomes. The G1 male is a heterozygote at all mutant loci and at all
Chr 7 loci that are polymorphic between A/J and C57BL/6J (left margin).
Like his father, he is mated to the recipient or background strain
(C57BL/6J, in this case) to generate a set of G2 progeny. A majority of
his progeny will inherit recombinant Chr 7s, but a significiant fraction
will inherit a non-recombinant A/J-type chromosome 7 that traces its
descent from a single gamete of the mutated G0 male (e.g., left-most G2
female). For short chromosomes (50 cM) subject to a single obligatory
recombination event during meiosis, two of the four chromatids will
usually be non-recombinant (Lawrie et al., 1995). However, for longer
chromosomes the fraction of non-recombinant chromatids is approximately
0.5L/50, where L is the length of the chromosome in centimorgans. For Chr
7, with a length of approximately 75 cM, the estimated fraction of
non-recombinant chromosome is 0.35, of which half will be derived from the
consomic parental strain. This equation gives estimates that are biased
upward because it fails to account for the non-independence of
recombination events within the tetrad. Interference extends over
approximately 60 Mb (Lawrie et al., 1995) and results in a more uniform
distribution of second crossover events. To sort mice with recombinant and
non-recombinant target chromosomes, all G2 mice are genotyped. The number
of G2 progeny that are genotyped will depend on the length of the target
chromosome, but an average of 30 to 60 G2 animals will typically need to
be tested. To reduce this effort a two step procedure can be employed. All
animals are initially genotyped for markers near proximal and distal ends
of the target chromosome. Mice that are heterozygotes for both markers are
subsequently genotyped at five to ten internal markers spaced at
approximately 10 cM intervals to determine whether or not they inherited
an intact A/J chromosome 7 from their G1 sire. Only those G2 males and
females that are non-recombinant across the entire target chromosome are
retained for breeding.
To generate sufficient numbers of non-recombinant G2 progeny, the G1
male is bred to five or more C57BL/6J females, generating 40 or more
offspring. The probability that any one G2 offspring carries a
nonrecombinant A/J chromosome is about 0.25 for small chromosomes (~50 cM)
and 0.125 for long chromosomes (~100 cM). In the case of Chr 7 (75 cM),
about 1 in 6 G2 animals are expected to carry an intact A/J chromosome
bearing a set of mutations. We expect 3 nonrecombinant mice in 20, and the
probability of finding at least 1 nonrecombinant of each sex in 40 is
close to 95%. It will be necessary to identify enough nonrecombinant
females to ensure that 50 or more G3 progeny can be generated quickly.
The G2 females are bred to G2 males, and are not backcrossed to the
G1 male as in a standard G3 recessive screen. The key difference between
G1 and G2 males is that G2 progeny carry half the load of extraneous
mutations (mutations on non-target chromosomes) as does the G1 male. As
described below, it is important to reduce the load of these extraneous
Genotyping and Breeding G3 progeny
Nonrecombinant G2 females are crossed to one or more G2 nonrecombinant
males to generate G3 progeny (Fig. 1). All G3s are genotyped at moderate
resolution on the target chromosome (makers spaced at 10 cM). Individuals
homozygous for segments of the target chromosome longer than 5–10 cM are
retained for phenotyping and breeding. An average of one-fourth of G3
progeny will be homozygous for a given part of the target chromosome.
G3 males and females that are homozygous for shared segments of the
target chromosome are paired. The goal is to ensure that the target
chromosome is fully represented by overlapping and nested sets of breeding
pairs. In this way the load of mutations carried on the target chromosome
is effectively split apart and mapped in one process. The larger the
number of breeding pairs with overlapping homozygous intervals, the better
the resolution and assurance with which mutations can be mapped. Assuming
that a line does not carry lethal alleles, it should be possible to screen
all of Chr 7 for recessive mutations using only a few breeding pairs, each
homozygous over matched a 30 to 40 cM interval. In subsequent cycles of
breeding, different G3 males and females sharing progressively smaller
common intervals can be paired to refine intervals that harbor mutations.
Detecting Lethal Mutations
Recessive lethal alleles can make up a large fraction of induced
mutations—up to 90% in some screens of Drosophila (Wright, 1977).
Unless visible markers are incorporated into a screen these recessive
lethals can be difficult to detect. However, in a consomic screen it is
easy to recognize and even map recessive lethal mutations. G3 progeny are
genotyped at a set of markers that span the target chromosome, and if few
or no G3 individuals are homozygous for a particular interval, then this
is good evidence that a recessive embryonic lethal mutation has been
generated and detected. Similarly, alleles that compromise prenatal and
early postnatal viability should lead to segregation distortion of
genotypes assayed at weaning. It should subsequently be possible to choose
appropriate G3 animals that are heterozygotes across these missing
intervals and breed them to confirm, recover, and characterize lethal
mutations. Assessing the frequency of missing homozygous intervals among
G3 progeny may be an effective way to estimate the efficacy of mutagen
Phenotyping G4 progeny
Matched G3 animals are bred and entire litters of G4 individuals are
phenotyped. All are homozygous across large segments of the target
chromosome. If a segment harbors a recessive mutation, then there should
be a consistent abnormality within a G4 litter. Litters of G4 progeny can
be compared to the original consomic strain and to other litters with
overlapping and non-overlapping regions of homozygosity.
G3 parents will share mutations with their G4 offspring and it may
initially be puzzling that G4 animals are generated prior to an extensive
analysis of their G3 parents. The main reason is that if G3 animals appear
normal, then by allowing the G3 pair to produce one or more G4 litters we
can get enough test class animals to search for mutations with subtle
qualitative and quantitative effects. A related reasons is that in a
recessive screen a factor that often limits progress is the number of test
class animals that can be generated. If a pair of G3 animals is fertile,
then large numbers of test class animals that are homozygous on defined
intervals can be generated efficiently. They can be sent to several
investigators for analysis. Third, if G3 animals are obviously mutants,
then mating these animals allows us to determine the impact that mutations
have on reproduction.
How isogenic are G4 test class animals? G4 animals are similar
in genetic composition to both the consomic parental strain and to the
original inbred recipient strain (Fig. 1). However, even G4 littermates
produced from interval-matched G3 parents are not isogenic. First, they
carry a variable set of segregating mutations on otherwise inbred
non-target chromosomes (Fig. 1). Second, the G4 progeny are genetically
heterogeneous in the variable length intervals that flank the homozygous
interval of the target chromosome (Fig. 1, bottom). G4 progeny from single
litters will therefore have some residual genetic variance. Variance due
to off-target mutations can be reduced significantly using an intervening
backcross to the wildtype recipient strain (see next section on
Collateral Damage). However, the genetic variance within G4 litters
should not prevent the detection of mutations. If a mutation maps to the
chosen interval of the target chromosome, then other G4 litters that share
this interval but that were produced by mating different G3 mice should
have the same aberrant phenotype. It is therefore possible to confirm that
a mutation maps to the target chromosome and simultaneously refine the
position of mutations by comparing several litters from different G3
parents. [Furthermore, by crossing G4 animals to mice carrying small
deletions in selected homozygous intervals (Schimenti and Bucan, 1998), it
should also be possible to quickly refine the location of a mutation by
If G4 animals are used to generated additional generations of progeny,
a significant number of these mutations will become fixed.
In both G2 and G3 generations animals that do not meet the genotype
criteria are never mated. This filtering can be done before weaning and
reduces the size of the colony and number of individuals that need to be
Collateral Damage: Are Mutations On-Target?
Selective breeding of G2 and G3 progeny enriches for homozygotes on
defined intervals of the target chromosome. However, there will still be
many non-target segments of the genome that harbor mutant alleles (see G3
and G4 progeny in Fig. 1). Mutations that are off-target in G2 progeny are
diluted twofold relative to mutations on the target (compare the mutation
load in the G1 male to that in G2 progeny in figure 1). For this reason,
the probability that a single G3 parent is homozygous for an off-target
mutation originally carried by the G1 male is about 1 in 8. The
probability that two randomly chosen G3 animals will both be homozygous
mutants for off-target intervals is 1 in 64. However, the non-targeted
part of the genome is on average about 20 times larger than the target.
The approximate likelihood that a mutation is off target is therefore
about 1 in 3 (1:64 times 20:1).To compute a more precise estimate would
require information on several factors, including the relative size of the
target chromosome and the homozygous target interval, the relative gene
density of the particular interval, and the relative abundance of lethal
alleles that are off-target. A practical way to assess whether a mutation
is on- or off-target is to examine litters generated from different sets
of G3 parents that share common homozygous intervals. If enough G3 mating
pairs are bred, every part of the target should be represented by two or
more G4 litters. The likelihood that two G4 litters from different parents
will have mutations associated with the same recessive allele on a
non-target chromosome is low—on the order of ~0.005. If three or more
independently derived litters have the same aberration then the mutation
is highly likely to be on-target.
One other factor adds some complexity to the analysis of G4
progeny. Numerous mutations that are off-target will be carried in a
heterozygous state by the G3 progeny. Phenotypes that appear on average in
one-quarter or one half of mice in one G4 litters, but not another, will
for the most part be caused by segregation at these mutated loci. An
example is shown on the third pair of chromosomes at the bottom of Figure
1 in which the male G3 parent is –/+ and the female G3 parent is –/– for
an off-target mutation. The source of this variability can be rapidly
assessed using the same method mentioned above; namely, to compare
phenotypes of sets of litters with matched or overlapping homozygous
intervals. However, there is an alternative method that may be useful,
depending on the sensitivity of the screen to this collateral genetic
damage. By backcrossing G2 nonrecombinant males to parental strain females
the off-target mutations can be diluted twofold per generation. This step
is shown in Figure 1 by an oblique line labeled dilution backcross.
With each generation of backcrossing to the recipient strain, the load of
residual off-target mutations is halved.
Detecting Recessive Mutations in Quantitative Trait Loci
A key issue in a mutagenesis screen is to have sufficiently sensitive
and economical assays to detect altered traits. Minimizing extraneous
genetic variance is critical. Mackay and colleagues (Anholt et al., 1996)
have used a chromosome substitution protocol and they have been able to
detect and map numerous mutations in Drosophila that affect
behavioral responses to volatile organic compounds. In a consomic screen
the final analysis of deviant phenotypes is based on a comparison between
several litters of G4 progeny and three possible control populations: (1)
the consomic strain, (2) the recipient inbred strain, and (3) non-mutagenized
G4 control lines generated in the same way as the experimental G4 litters
and homozygous for the same interval.
For phenotypes that are inherently noisy, such as many behavioral
and disease-susceptibility traits, entire litters of homozygotes can be
screened, and the litter's mean, variance, or incidence becomes the key
measure. As is true of recombinant inbred lines, the precision and
reliability of results can be assessed and improved by examining numerous
offspring from multiple litters. Unusual phenotypes in the G4 should be
found in at least several members of each litter. Ultimately, the
phenotype should also be noted in G3 parents. The G4 screen is robust in
the sense that episodic, but unfortunately common, non-heritable
abnormalities (developmental phenocopies) should not become a distraction.
For this reason exhaustive assessment of G4 animals may be warranted. For
example, quantitative histological analysis might reveal reliable
differences between G4 litters and control lines. Extensive quantitative
serological, complex expression assays, and extensive behavioral and
neurological testing might be justified.
How small an effect can one expect to detect with a consomic design?
The statistical power depends primarily on the number of G4 animals and
litters that one has the patience to phenotype. A reasonable expectation
is that for any part of the target chromosome one will have at least two
homozygous litters available for analysis (x2 coverage). The mean
phenotypes of G4 animals should be compared with those of a G4 litter that
is homozyogus on the same interval but generated from a non-mutagenized
consomic stock. The statistical power of the comparison is that of a
conventional t test (Snedecor and Cochran, 1980, p. 70). For example, with
12 animals in the mutagenized G4 group, the power to detect a mutation at
the p = 0.05 level with an effect that shifts the mean +/- 1 SD is
approximately 0.94. With 25 animals (4–5 litters) one should have a 50%
chance of detecting mutations with an effect size of only 0.4 SD. In some
cases the power will be less because the appropriate n for traits
sensitive to maternal effects will be litter number. The power of a search
for a mutation with an effect of +/- 1 SD is approximately 0.50 using four
Genotyping and Colony Costs of the Screen
Genotyping Costs. A targeted screen using consomic strains
relies on efficient genotyping. Approximately 100 to 200 G2 and G3 animals
will need to be genotyped to characterize on-target mutations carried each
G1 male. If 10 markers were typed per individual, a maximum of 2000
reactions would be required. A more typical estimate is 500 to 1000
reactions. Thanks to the high density of MIT microsatellite markers
(Dietrich et al.,
1992), simple DNA extraction procedures, and streamlined PCR protocols
(Williams and Strom,
<http://www.nervenet.org/papers/PCR.html>), the time, effort, and
expense of genotyping is now comparatively low (~$1/genotype).
Colony Costs. While the G4 design is somewhat complicated,
this does not mean that the colony size needs to be large or that the
screening will be slow. Each G1 male is initially bred to 5 to 8 inbred
females in one large cage. G2 animals can be culled shortly after they
have been genotyped and just before weaning. The G1 male is retired and a
son and several daughters that all have nonrecombinant target chromosomes
(Fig. 2, middle) are left in the cage. The G3 progeny will need to be
marked, genotyped, culled, or retained prior to weaning. At this point,
pairs of matched G3 progeny can be moved to smaller breeding cages. As few
as 3 pairs, or as many as 20 pairs, could be used to cover the entire
target chromosome. Additional heterozygote mating cages may need to be set
up to characterize intervals harboring putative recessive lethal
mutations. Approximately four months after the initial G1 matings, the G2
mating cage will have been supplanted by 3 to 15 small breeding cages of
G3 parents and their G4 offspring. Each cage will probably need to be
retained in the colony for a minimum of two months to allow enough time to
phenotype each litter. The average colony requirement associated with a
single G1 lineage is approximately 1 large cage for a duration of 4 months
followed by 10 small cages for a duration of 3 months. The frequency with
which mutations are identified will determine the cost of success—those
costs associated with characterizing, archiving, and distributing mutant
Screening for Late-Onset Mutations. It should be practical to
retain small numbers of G3 and G4 animals in the colony to study the
long-term consequences of novel recessive alleles. The original G3 mating
pairs will often be 100-days-old before the analysis of their first litter
of G4 progeny is complete. If no interesting phenotypes are initially
detected in the G4 animals, then it may nonetheless be useful to hold
several sets of G3 or G4 breeders in anticipation of possible late-onset
traits. It may often be possible to test an entire chromosome by retaining
only animals with long congenic intervals. In the absence of recessive
lethal alleles, as few as one to five pairs of mice could cover an entire
chromosome. Periodic test matings may be useful to ensure that mutations
with late lethal effects are not lost.
Targeted Screens—Disadvantages and Advantages
A targeted screen ignores mutations generated across most of the
genome—19 out of 20 mutations will be rejected in a typical consomic
screen. For this reason, a targeted approach may not be appropriate (1) if
the screening procedure is tightly focused, reliable, rapid, and
inexpensive; (2) if there is good evidence that traits of interest are
known to be controlled by small numbers of loci; and (3) if mapping
mutations is a low priority and one that can be more conveniently done
after mutations have been identified. One particular potential problem
with a targeted approach is the possibility that a narrow chromosomal
focus will be coupled with a low yield of mutations (Schimenti and Bucan,
1998). The consomic approach enriches for homozygotes and this does
increase the efficiency and sensitivity of the screen. However, even after
selectively breeding G2 and G3 individuals, some mutations that do not map
on the target chromosome will be detected among G4 progeny. In order to
retain focus on mutations that can be rapidly mapped and for which a list
of candidate genes can be identified, these off-target mutations will
generally need to be ignored. Whether this focused approach yields a
sufficiently high output of novel recessive alleles will depend in part on
the ingenuity, sensitivity, and number of tests that can be performed on
One potential problem with a targeted approach is the possibility
that a narrow focus will be coupled with a low yield of mutations (Schimenti
and Bucan, 1998). On the one hand, the consomic targeted approach enriches
for homozygotes, increasing the efficiency and sensitivity of the screen
and this method also simplifies subsequent high-resolution mapping. On the
other hand, even after genotyping G2 and G3 individuals, some mutations
that are off-target are likely to be detected. In order to retain focus on
those mutant phenotypes that can be rapidly mapped and for which a list of
candidate genes can be identified, off-target mutations will generally
need to be ignored. Whether this focused approach yields a sufficiently
high output of novel recessive alleles will depend in part on the
ingenuity, sensitivity, and number of tests that can be performed on G4
Although a consomic recessive screen targets particular chromosomes,
there is no reason why it would not be possible to systematically target
each chromosome in turn. In this respect, a targeted approach could be as
comprehensive in scope as a random whole-genome screen. One possible
advantage of the targeted approach is that it is more practical to assess
when a particular part of the genome has been saturated with mutations (Rinchik
et al., 1990). It should be possible to fine-tune the relative intensity
of mutagenesis and screening to reflect the large regional differences in
gene density. Furthermore, if the consomic target chromosome were derived
from the particular strain of mouse whose genome is soon be sequenced
(C57BL/6J), then once a set of candidate genes have been identified in a
given homozygous interval, it will be straightforward to sequence
candidates as a first step in identifying the responsible mutation.
[Because the G4 progeny are homozygous on well-defined intervals,
one may want to finetune the particular type of phenotype screen to detect
novel mutations in well mapped genes and quantitative trait loci. For
instance, if a particular interval is known to contains a family of
voltage-gated potassium channels or a QTL that specifically affects organ
weights, then selected litters might be subjected to special functional or
Two other issues associated with the G4 consomic design deserve
mention. (1) The ENU dosages appropriate for this design has not yet been
established. Given the likelihood of generating numerous recessive lethal
alleles, it may be difficult to determine a suitable dosage for a given
set of consomic strains. (2) In other targeted recessive designs it is
often possible to take advantage of visible markers and inversions to
simplify the generation and categorization of homozygous animals. This
completely avoids the somewhat onerous genotyping that is vital to a G4
consomic screen. Except in Drosophila, it is generally not possible to use
visible markers or inversions as part of a consomic design. However, to
give one example in which this will be possible, the B-7A consomic strain
will be a white albino. The interval around the tyrosinase locus could in
principle be targeted with minimal genotyping.
Using Congenic Strains?
A screen could in principle be adapted to exploit smaller target
intervals that have been isolated in congenic, rather than consomic,
strains of mice. In congenic strains the length of the donor interval is a
function of the number of backcrosses, but even after ten backcross
generations, the donor interval is typically 20 cM in length (~200/n cM,
where n is the number of generations of successive backcrosses). One
disadvantage of using congenic strains is that a significant fraction of
mutations will occur in linked intervals that flank the congenic segment.
These adjoining intervals will be larger than the congenic interval
itself. This will make mapping mutations generated in congenic strains
somewhat more unpredicable than those generated using consomic strains.
Precedents: Other Targeted Methods
Targeted screens for recessive mutations have been carried out for many
years and it is important to compare the consomic design with well known
alternatives. One of the first and best protocols to screen for recessive
mutations takes advantage of hemizygous lines that are missing all or part
of particular chromosomes (Muller, 1927; Justice et al., 1997, Schimenti
and Bucan, 1998). Muller's classic mutagenesis experiments exploited the
hemizygosity of the X chromosome of male fruit flies. But artificial
hemizygous intervals can now be generated in mice of both sexes by
deleting a short segments from one autosome. Mice from which 5 to 15 cM
has been deleted are usually viable (Justice et al., 1997; Schimenti and
Bucan, 1998). Recessive alleles that are in trans to the deletions are
unmasked and mice have phenotypes similar, if not identical, to those of
homozygote mutants. To expose a recessive mutation involves breeding G1
heterozygote carriers like those in Figure 1 to deletion stock. An average
of one in four progeny of such a cross will be a hemizygous mutant.
Rinchik and colleagues used mice from which the interval on chromosome 7
extending from tyrosinase to shaker 1 had been deleted (Rinchik et al.
1990; Rinchik, 1991). G1 females carrying mutations inherited from
ENU-mutagenized sires were crossed to males hemizygous for the deletion.
By a clever choice of coat-color mutations, they were able to restrict
their main analysis to the hemizygous G2 test class animals. Schimenti and
Bucan (1998) review a G2 screen in which recombination in a deletion
interval is suppressed by using a stock that carries a large inversion and
a coat-color mutation. This refinement simplifies the analysis by forcing
a more nearly perfect concordance between visible markers and genotype
(see their figure 3).
In order to use a G2 screen of this type, deletions need to be
generated and hemizygotes must be viable. Until recently, making deletion
stocks would have been a serious technical impediment, but Schimenti and
colleagues (You et al., 1997;
<http://lena.jax.org/~jcs/Delbank.html>) are now systematically
generating embryonic stem cells that can be used to generate mice with
deletions quickly. A set of 500 lines, each with a unique pair of
thymidine kinase tags embedded 5 cM apart, are being generated. The
interval between tags is deleted and the stem cells are then used as
starting material to generate deletion stock. By combining deletions with
visible markers and chromosomal inversions that suppress
recombination—genetic tricks first perfected in Drosophila—it is
possible to have a nearly perfect concordance between visible markers and
genotypes (Schimenti and Bucan, 1998). This can greatly simplifies the
analysis because only a particular and easily recognized class of G2
offspring are phenotyped. Generating the highly specialized lines of mice
that are required to this type of screen is still difficult.
There are significant advantages of using deletion stock in a
mutagenesis screen that counterbalance the difficulties of generating
mice. The major advantage is that recessive mutations can be detected in
the second generation (Rinchik et al., 1990). The reduction in breeding
costs relative to G3 and G4 designs is a major advantage. No other regions
of the genome, except the male's X and Y chromosomes, are “homozygous” for
off-target mutations, so mutational variance due to recessive alleles is
restricted to the target interval. One minor disadvantage is that the
genetic background of the G2 progeny will be a mixture of the genomes of
the mutagenized male and the genome of the deletion stock. The DELBank
Project deletions are carried on a (129/SvJae x C57BL/6J)F1 background.
The G2 progeny will therefore combine alleles from 129 and C57BL/6 with
whatever strains are used to generate heterozygous carriers of mutations.
The high genetic variance of this type of G2 can be viewed as a positive
feature—mutations that are detected will probably have good penetrance on
a variety of genetic backgrounds. But the downside of this genetic
complexity is that subtle effects of recessive mutations will be much more
difficult to detect.
Another factor to consider in comparing consomic and deletion
methods is the relative intensity of mutagenesis that must be employed to
compensate for the length of the target interval. Targeting a 5 cM
interval typically involves using a relatively high mutagen dosage.
Strains that tolerate high doses or ENU are therefore preferred. In a
consomic screen in which the target is 10 to 20 times longer, a lower dose
may be employed. As a result, even a strain such as C57BL/6 that is
extremely sensitive to ENU may be useful.
Drosophila Precedents. The consomic procedure outline here also
has precedents that trace back to a series of experiments by Dobzhansky
and colleagues (e.g., Dobzhansky et al., 1959) in which single chromosomes
of wild populations of fruit flies were introgressed into special
laboratory strains to examine the load of recessive lethal alleles carried
by natural populations. Recombination in fruit flies can be suppressed,
and given a laboratory strain with the right visible markers it is
relatively simple to follow the assortment of chromosomes in a three
generation cross. In several impressive studies, Mackay and colleagues
(Mackay et al., 1992; Lyman et al. 1996; Anholt et al., 1996) have
exploited chromosome substitution techniques to study effects of induced
mutations. Mutated chromosomes were introgressed into isogenic backgrounds
and the effects on fecundity, sensory bristle number, and olfactory
behavior were assayed. In these studies a complex breeding scheme
exploiting visible markers and tagged P-elements (Fig. 1 of Lyman et al.,
1996) was used to filter progeny and generate test class animals (G4
through G6). They then examined large numbers of nearly isogenic mutants
to detect subtle QTL-like effects. P-elements were subsequently mapped at
high-resolution by in situ hybridization. In a consomic screen using mice,
visible markers are replaced by molecular markers made visible using
simple PCR protocols. There are many specific differences between the
Drosophila substitution mutagenesis screen (see Fig. 1 of Lyman et
al., 1996) and the consomic procedure described here, but the key goals
are sharedÑminimize extraneous genetic variance while maximizing the
efficiency with which induced mutations can be mapped. As in
Drosophila, there is still extraneous mutational variance in the mouse
consomic screen. The source of this variance in flies is not yet clearly
understood (Lyman et al., 1996), but in consomic mice most of this
variance will be associated with the potentially large load of off-target
mutations. These mutations can only be removed by repeatedly backcrossing
G2 non-recombinants to the inbred recipient strain.
In comparing the merits of targeted approaches that rely either on
consomic strains or deletion lines it will be critical to assess the
likely yield of mutants as a function of effort and cost. In general, the
smaller the interval that is targeted, the lower the yield. In a consomic
screen, the target is an entire chromosome, and the yield will be
relatively high. However, this design does require four generations; a
negative when compared to a simple two-generation design. The yield of
mutations is also a function of the sensitivity of tests. Here the
advantage of screening entire litters of homozygotes on a relatively
uniform genetic background may be a persuasive advantage of the consomic
This research was supported by grant RO1 EY06627 from the National Eye
Institute and RO1 NS35485 from the National Institute of Neurological
Disorders and Stroke. I thank Kathryn Graehl for helpful discussion and
for editing the text and Figure 1. My thank to Drs. Rosemary Elliott,
Lorraine A. Flaherty, Jean-Louis Guénet, Eugene Rinchik, John Schimenti,
Benjamin Taylor, and David W. Threadgill for comments and corrections.
Anholt RRH, Lyman RF, Mackay TFC (1996) Effects of single P-element
insertions on olfactory behavior in Drosophila melanogaster.
Brown SDM, Peters J (1996) Combining mutagenesis and genomics in the
mouse—closing the phenotype gap. Trends Genet 12:433–435.
Brown SDM, Nolan PM (1998) Mouse mutagenesis-systematic studies of
mammalian gene function.
Hum Mol Genet 7:1627–1633.
Chalfie M, Au M (1989) Genetic control of differentiation of the
Caenorhabditis elegans touch receptor neurons. Science 243:10273–1033.
Dietrich W, Katz H, Lincoln SE, Shin HS, Friedman J, Dracopoli NC,
Lander ES (1992)
A genetic map of the mouse suitable for typing intraspecific crosses.
Genetics 131:423–447. <http://carbon.wi.mit.edu:8000/cgi-bin/mouse/index>
Dohzhansky T, Spassky B (1954) Genetics of natural populations. XXII. A
comparison of concealed variability in Drosophila prosaltans with
that in other species. Genetics 39:472–487.
Dohzhansky T, Levene H, Spassky G, Spassky N (1959) Release of genetic
variability through recombination. III. Drosophila prosaltans.
García-Añoveros J, Corey DP (1997) The molecules of mechanosensation.
Annu Rev Neurosci 20:567Ð594.
Hudgins CC et al., (1985) Studies of consomic mice bearing the Y
chromosome of the BXSB mouse. J Immunol 134:3849–3854.
Hrabé de Angelis M, Balling R (1998) Large scale ENU screens in the
mouse. Genetics meets genomics.
Mutation Research 400: 25–32.
Justice M (1998) Mouse germline mutagenesis. In: Mouse genetics and
transgenics: a practical approach (Jackson I, Abbott C, eds). pp XXX-XXX.
New York: Oxford UP.
Justice MJ, Zheng B, Woychik RP, Bradley A (1997) Using targeted large
deletions and high-efficiency N-ethyl-N-nitrosourea mutagenesis for
functional analyses of the mammalian genome.
Kacser H, Burns JA (1981) The molecular basis of dominance. Genetics
Kasarskis A, Manova K, Anderson KV (1998) A phenotype-based screen for
embryonic lethal mutations in the mouse.
Proc Natl Acad Sci 95:7485–7490.
Lawrence PA (1992) The making of a fly. London: Blackwell.
Lawrie NM, Tease C, Hulten MA (1995) Chiasma frequency, distribution
and interference maps of mouse autosomes. Chromosoma 104:308–314.
Lufkin T, Lohnes D, Mark M, Dierich A, Forry P, Gaub MP, LeMeur M,
Chambon P (1993) High postnatal lethality and testis degeneration in
retinoic acid receptor alpha mutant mice. Proc Natl Acad Sci USA
Luo J, Pasceri P, Conlon RA, Rossant J, Giguere V (1995) Mice lacking
all isoforms of retinoic acid receptor beta develop normally and are
susceptible to the teratogenic effects of retinoic acid. Mech Dev
Lyman RF, Lawrence F, Nuzhdin SV, Mackay TFC (1996) Effects of single
P element insertions on bristle number and viability in
Drosophila melanogaster. Genetics 143:277–292.
Lynch M, Walsh B (1998) Genetics and analysis of quantitative traits.
Marker PC, Seung K, Blande AE, Russell LP, Kingsley DM (1997) Spectrum
of Bmp5 mutations from germline mutagenesis experiments in mice.
Mendelsohn C, Lohnes D, Decimo D, Lufkin T, LeMeur M, Chambon P, Mark M
(1994) Function of the retinoic acid receptors (RARs) during development
(II). Multiple abnormalities at various stages of organogenesis in RAR
double mutants. Development 120:2749–2771.
Morel L, Mohan C, Yu Y, Croker BP, Tian N, Deng A, Wakeland EK (1997)
Functional dissection of systemic lupus erythematosus using congenic mouse
strains. J Immuno 158:6019–6028.
Muller HJ (1927) Artificial transmutation of the gene. Science
Nadeau JH, Sankoff D (1997) Comparable rates of gene loss and
functional divergence after genome duplications early in vertebrate
evolution. Genetics 147:1259–1266.
Nolan PM, Kapfhamer D, Bucan M (1997) Procedures for the identification
of dominant behavioral mutations in mice using ENU mutagenesis. I:
Methods: A companion to methods in enzymology 13:379–395.
Pickard GE, Sollars PJ, Rinchik EM, Nolan PM, Bucan M (1995)
Mutagenesis and behavioral screening for altered circadian activity
identifies the mouse mutant, Wheels. Brain Res 705:255–266.
Rinchik EM, Carpenter DA, Selby PB (1990) A strategy for fine-structure
functional analysis of a 6– to 11–centimorgan region of mouse chromosome 7
by high-efficiency mutagenesis.
Proc Nat Acad Sci USA 87:896–900.
Rinchik EM (1991) Chemical mutagenesis and fine-structure functional
analysis of the mouse genome.
Trends Genet 7:15–21.
Russell WL, Kelly PR, Hunsicker PR, Bangham JW, Maddux SC, Phipps EL
(1979) Specific-locus test shows ethylnitrosourea to be the most potent
mutagen in the mouse. Proc. Nat. Acad. Sci. USA 76: 5918–5922.
Russell WL, Hunsicker PR, Raymer GD, Steele MH, Stelzner KF, Thompson
HM (1982) Dose response curve for ethylnitrosurea specific-locus mutations
in mouse spermatogonia. Proc Natl Acad Sci USA 79: 3589–3591.
Schimenti J, Bucan M (1998) Functional genomics in the mouse:
phenotype-based mutagenesis screens.
Genome Res 8:698Ð710.
Shedlovsky A, King T, Dove W (1988) Saturation germ line mutagenesis of
the murine t-region including a lethal allele at the quaking locus. Proc
Natl Acad Scie USA 85:180-184.
Suda Y, Matsuo I, Aizawa S (1997) Cooperation between Otx1 and Otx2
genes in developmental patterning of rostral brain. Mech Dev 69:125–141.
Vogel E, Natarajan AT (1979) The relatiion between reaction kinetics
and mutagenic action of mono-functional alkylating agents in higher
eukaryotic systems. I. Recessive lethal mutations and translocations in
Drosophila. Mutation Res 62:51Ð100.
Williams RW (1998) Neuroscience meets quantitative genetics: using
morphometric data to map genes that modulate CNS architecture. The 1998
Short Course in Quantitative Neuroanatomy (Morisson J, Hof P, eds) pp.
66-78. Washington: Society for Neuroscience. (http://www.nervenet.org/papers/ShortCourse98.html).
Wright S (1977) Evolution and genetics of populations, Vol 3. Chicago:
Univ Chicago Press.
You Y, Bergstrom R, Klemm M, Lederman B, Nelson H, Ticknor C, Jaenish
R, Schimenti JC (1997a) Chromosomal deletion complexes in mice by
radiation of embryonic stem cells.
Nat Genet 15:285–288.
You Y, Browning VL, Schimenti, JC (1997b) Generation of
radiation-induced deletion complexes in the mouse genome using embryonic
Zhou G, Strom RC, Giguere V, Williams RW (1998) Modulation of retinal
cell populations and eye size in retinoic acid receptor knockout mice. Soc
Neurosci Abst 24:1033.
The MRC Mutagenesis Programme, Harwell, England. <http://www.mgu.har.mrc.ac.uk/mutabase/>
The German Human Genome Project Mouse ENU Mutagenesis Program.
DELBank Project. Links to the DELBank project and John Schimenti's
Joy of Cloning protocol manual. <http://lena.jax.org/~jcs/>
Maja Bucan's Lab at the University of Pennsylvania. A short
description of mutagenesis screens for behavioral abnormalities.
Tennessee Genomics Consortium. Links to genomics and mutagenesis
programs in Tennessee (Oak Ridge National Laboratory, Vanderbilt
University, University of Tennessee, Meharry Medical School, and St.
Jude Children's Research Hospital). <http://www.nervenet.org/enu/enu.html>
A really quick way to get sufficient DNA for a few PCR reactions.
Since 4 Sept 98