Evidence for complex multigenic inheritance of radiation AML susceptibility in mice
revealed using a surrogate phenotypic assay
F.Darakhshan1, C.Badie, J.Moody, M.Coster, R.Finnon,
P.Finnon, A.A.Edwards, M.Szłuin ´ska, C.J.Skidmore1,
K.Yoshida2, R.Ullrich3, R.Cox and S.D.Bouffler?
Radiation Effects Department, Health Protection Agency, Radiation
Protection Division, Chilton, Didcot, Oxfordshire, OX11 0RQ, UK,
1School of Animal and Microbial Sciences, University of Reading,
Whiteknights, PO Box 228, Reading, Berkshire, RG6 6AJ, UK,
2NIRS, Inage-Ku, Chiba-shi 263, Japan and3Department of Environmental
and Radiological Health Sciences, Colorado State University, Fort Collins,
CO 80523, USA
?To whom correspondence should be addressed. Tel: þ44 1235 822648;
Fax: þ44 1235 833891;
The mapping of genes which affect individual cancer risk is
an important but complex challenge. A surrogate assay of
susceptibility to radiation-induced acute myeloid leukae-
mia (AML) in the mouse based on chromosomal radiosen-
sitivity has been developed and validated. This assay was
applied to the mapping of radiation-induced AML risk
modifier loci by association with microsatellite markers.
A region on chromosome (chr) 18 with strong association is
identified and confirmed by backcross analysis. Additional
loci on chrs 8 and 13 show significant association. A key
candidate gene Rbbp8 on chr18 is identified. Rbbp8 is
shown to be upregulated in response to X-irradiation in
the AML sensitive CBA strain but not AML resistant
C57BL/6 strain. This study demonstrates the strength of
utilizing surrogate endpoints of cancer susceptibility in the
mapping of mouse loci and identifies additional loci that
may affect radiation cancer risk.
Cancer is a disease in which both environmental and inherited
factors affect the incidence in individuals and populations.
Twin studies have estimated that heritable factors can con-
tribute up to 42% of the risk of spontaneous prostate cancer
and perhaps ?20% of leukaemia risk (1). While it is evident
that highly penetrant genes contribute significantly to cancer
risk in some individuals, e.g. BRCA1 and BRCA2 carriers (2),
it is clear that the major proportion of risk in populations is
attributable to multiple genes of low penetrance (3,4). The
identification of these common but low penetrance suscepti-
bility alleles presents a major challenge in cancer genetics.
Modelling studies have indicated that genes of low penetrance
may significantly affect the distribution of breast cancer risk in
human populations (5) and evidently the individual risk will be
influenced. The difficulties in the unambiguous identification
of a specific gene as a cancer susceptibility modifier are
demonstrated in the recent publication implicating the Aurora
2 encoding Stk6/STK15 as a skin cancer modifier (6). In con-
trast, identifying genomic regions harbouring cancer risk
modifiers is becoming simpler in the mouse (7). Most studies
of human cancer susceptibility have focused on spontaneous
cancers; however, cancers induced by factors such as ionizing
radiation will almost certainly be similarly affected by multi-
ple genetic factors. Studies using experimental animals, mice
in particular, are of great importance for identifying modifiers
of induced cancer risk.
Ionizing radiation is a human carcinogen (8) and acute
myeloid leukaemia (AML) features prominently in the cancers
seen in the primary radiation-exposed study population, the
Japanese A-bomb survivors (9). For experimental studies,
mouse models of radiation-induced AML have been developed
(10–12) and one of the most extensively studied of these is the
CBA inbred strain (13). In CBA mice AML presents following
a mean latent period of ?18 months after a whole body acute
X-ray exposure with a maximal incidence of 25–30% at 3 Gy
(13). The spontaneous incidence of AML is very low, 51/
1000 (14), giving confidence that radiation is the causal agent
in the vast majority of tumours studied. Chromosome (chr) 2
deletions are a consistent feature of radiation-induced AMLs
in CBA mice (15,16), and the transcription factor Pu.1/Sfpi1
has been identified as a key target gene, one copy of which is
lost and the other frequently mutated in radiation AMLs (17).
Furthermore, the early induction of elevated frequencies of
chr2 aberrations in bone marrow is a feature of the CBA strain,
but not the AML resistant strain, C57BL/6 (18). In this paper,
the utility of chr2 radiation sensitivity, as a surrogate endpoint
for AML sensitivity, has been investigated. All types of aber-
ration were included in the analysis despite the more specific
association of chr2 deletions with AML. This approach was
selected as certain AMLs present with translocations and some
of these evolve into deletions (15,19). Thus, many or all types
of chr2 aberrations may be associated with AML. Having
established the radiation sensitivity of chr2 in a range of inbred
strains of mouse a novel strategy for the mapping of loci
controlling this trait was developed. The method is essentially
a microsatellite association mapping exercise which exploits
the public database of microsatellite polymorphism informa-
In principle this mapping method is analogous to that
of Grupe et al. (20) with the exception that microsatellite
polymorphisms are exploited rather than single nucleotide
polymorphisms (SNPs). Mapping was refined by further
microsatellite typing and confirmation sought through a
limited backcross analysis.
The picture that emerged from these mapping exercises was
that multiple candidate loci controlling chr2/AML sensitivity
could be identified and that the allelic influence of individual
loci may depend upon other factors in the genetic background
of mice. In this respect chr2/AML sensitivity is similar to other
mouse cancer susceptibilities (21,22). The most strongly asso-
ciating locus is located to a ?1.5 Mb region on the proximal
Abbreviations: AML, acute myeloid leukaemia; chr, chromosome; LOH,
loss of heterozygosity; Tm, melting temperatures; RT–qPCR, real-time–
quantitative PCR; SNP, single nucleotide polymorphism.
Carcinogenesis vol.27 no.2#Oxford University Press 2005; all rights reserved.
Carcinogenesis vol.27 no.2 pp.311–318, 2006
Advance Access publication August 19, 2005
by guest on June 1, 2013
region of chr18. Among the candidate genes in this region is
Rbbp8 encoding CtIP which appears to play a role in several
key cellular functions with relevance to cancer, chief amongst
these in the present context is a potential role in chromatin
remodelling which might be the basis of the specific radiosen-
sitivity of chr2. The expression of Rbbp8 is differentially
regulated in bone marrow cells of the chr2/AML sensitive
CBA and chr2/AML resistant strains following in vivo X
Materials and methods
Mouse strains and irradiation
CBA/H and C3H mice were obtained from MRC, Harwell. Strains DBA/2, A,
AKR, C57BL/6, BALB/c, SJL and CBA/Ca were obtained from Harlan
SeraLab, Loughborough UK. NOD mice were a gift from Prof. J.Todd,
Cambridge Institute for Medical Research, UK. NON and LP were supplied
by the Jackson Labs, Bar Harbor, USA. RFM mice were obtained from the
colony held at the National Institute for Radiological Sciences, Chiba Japan.
The experiments on mice that were carried out in the USA and Japan were
performed in the respective countries according to institutional and national
guidelines. All other animal procedures conformed to the UK Animals (Sci-
entific Procedures) Act, 1986 and were conducted under project licences PPL
30/1169 and PPL 30/1782. Mice were irradiated with a standard whole body
3 Gy dose of 250 KVp X-rays at a dose rate of 0.5–1 Gy/min.
Direct bone marrow metaphase chromosome preparations were made from
unirradiated controls and irradiated mice at 24 h post-irradiation as detailed in
Bouffler et al. (18). For inbred strains and F1 hybrids 3–5 mice were sampled.
Chromosome painting by fluorescence in situ hybridization (FISH) of chrs 1, 2
and 3 was performed by one of the three procedures. Originally a standard
single colour FISH protocol (18) was employed. This was superseded by single
colour FISH with tyramide detection. This latter method followed similar pre-
hybridization and hybridization steps as the standard protocol with the excep-
tion that only 2 ml of the probe diluted in the hybridization buffer was used.
Washing also followed the standard protocol. For detection the TSA system kit
(NEN Life Sciences, Beaconsfield, UK) was used in accordance with the
manufacturer’s recommendations, following an initial incubation of slides
for 30 min at 37?C with 1:200 diluted rabbit anti-FITC conjugated with
horse radish peroxidase (DAKO, Ely, UK). Latterly a three colour FISH
protocol was used (23).
Aberrations were scored in painted chromosomes using a system similar to
that of Tucker et al. (24), i.e. each colour junction between painted and
unpainted (or between differently painted) chromosomes scored 1. All types
of aberration yielding colour junctions were scored (translocations, insertions,
dicentrics, multi-coloured fragments and chromatid exchanges). In addition
where there was a distinct length difference between homologous chromo-
somes (?25%) a score of 1 was recorded, these were assumed to be deletions.
A few intra-homologue aberrationswere observed,where this wasunequivocal
those ‘junctions’ were also scored. The number of colour junctions for each
chromosome was enumerated separately. Care was taken to ensure that the
scoring system was consistent between the single colour and multi-colour
FISH protocols. For the statistical analysis of heterogeneity in aberration
score between chromosomes x2-tests were employed. The distribution of
aberrations among cells was analysed using a dispersion index test (25) and
any overdispersion of aberrations among cells was factored into the x2-tests.
Computer comparison of microsatellite sizes
Initial mapping of candidate loci exploited the publicly available microsatellite
polymorphism data (http://www-genome.wi.mit.edu/cgi-bin/mouse/index).
Whole genome data from release 16 was downloaded into Excel files. The
logical operators of Excel were used to identify microsatellite loci where
strains of one phenotypic group had identical size and strains of the other
phenotypic group had a divergent microsatellite size.
DNA was extracted from spleensby proteinaseK/RNase digestionfollowed by
phenol–chloroform extraction (inbred strains and F1 hybrids) and ear clips
(backcross mice). Microsatellite sizes were analysed following PCR ampli-
fication using MapPairs primers (ResearchGenetics, Paisley, UK) as described
previously (26). PCR products were separated on 3% w/v agarose or 8%
acrylamide gels. DNA run on acrylamide gels was stained with SYBR Green
I (Molecular Probes, Paisley, UK).
The strength of statistical association between genotype and phenotype was
determined using Fisher’s exact tests as previously described (27).
Quantitative PCR analysis
RNA was extracted from the bone marrow of mice exposed to 3 Gy X-rays
in vivo at 1–24 h post-exposure. Duplicate CBA/H and C57BL/6 mice were
used at each time point. Expression of Rbbp8 was investigated by real-time–
quantitative PCR (RT–qPCR) using an iCycler iQ (Bio-Rad, Hemel
Hempstead, UK) and iQ SYBR Green supermix with fluorescein (Bio-Rad).
Hprt and Cables-1 were used as reference genes as their expression was
expected to be comparable with Rbbp8 (28,29) (www.ncbi.nlm.nih.gov/
UniGene). This was confirmed using experiments comparing the baseline
expression levels of b-actin, Gapdh and Hprt. Hprt expression was most
similar to that of Rbbp8. As Cables-1 locates adjacent to Rbbp8 on chr18
(Figure 2) it was also a reference for possible common cis-acting promotor
mechanisms. Primer pair design was aided by Primer3 (30) (http://frodo.wi.
mit.edu) and primers were selected to span cDNA exon–exon junctions and
have similar melting temperatures (Tm). These were demonstrated to amplify
from cDNA only and an optimum Tmwas established from gradient experi-
ments (data not shown). Primer sequences are shown in Table I. A 25 ml
reaction volume was used, and 40 cycles of a three step amplification reaction
were employed (95?C 5 min; 40? (95?C 15 s, 57?C 30 s and 72?C 30 s); 95?C
1 min). This was followed by an 80 step melt curve analysis from 55 to 95?C to
confirm amplification specificity. Samples were analysed in triplicate and a
10? dilution series using Rbbp8 primers and the Rbbp8 PCR product as target
monitored reaction dynamics across the range from 6 ? 102to 6 ? 108copies.
Efficiency of target amplification and reference amplification were equal
over a 100-fold concentration range. Data analysis was performed using a
Microsoft Excel Macro implementation (Bio-Rad) of algorithms outlined by
Vandesompele et al. (31).
Radiation-induced chromosome aberration analysis
Aberration scores for chrs 1, 2 and 3 observed in bone marrow
cells from 12 inbred strains of mouse 24 h following in vivo
3 Gy X-irradiation are given in Table II. Chromosome aberra-
tions were very rare in unirradiated control bone marrow sam-
ples. Control aberration scores for all strains were 0 apart from
AKR and DBA/2, where scores of 1 in 900 and 300 cells,
respectively, were recorded. The results presented in Table II
confirm and extend previous findings (18). Aberration scores
in chrs 1, 2 and 3 were statistically indistinguishable in strains
C57BL, NON, NOD, A, AKR and DBA/2. However in strains
BALB/c, CBA/H, CBA/Ca, SJL, LP, C3H and RFM signifi-
cantly higher scores were observed in chr2 by comparison with
chrs 1 and 3. This is reflected in the chr2 sensitivity ratio,
where ratios of ?1 imply equi-sensitive chromosomes while
ratios of ?1.45 indicate chr2 sensitivity. The majority of the
strains in the chr2 sensitive group are known to be sensitive to
the induction of AML by radiation. Radiation-induced AML
Table I. Primer sequences used for quantitative PCR analysis
GeneForward primer Reverse primer
HPRT - exons 2/3
Cables-1 - exons 8/9
Rbbp8 - exons 11/12
F.Darakhshan et al.
by guest on June 1, 2013
has been documented in CBA/H and CBA/Ca (32,33), C3H
(11), RFM (34), SJL (35) and BALB/c (36). In contrast
radiation-induced AML has not been reported for any of the
strains in the non-chr2 sensitive group. Thus the relative fre-
quency of chromosomal aberrations scored in bone marrow
cells 24 h following 3 Gy in vivo X-irradiation appears to be a
valid surrogate for AML sensitivity. These data additionally
predict that LP should be sensitive to the induction of AML.
Aberration scores averaged over the three chromosomes also
showed substantial variation, compare for example C57BL/6
and A with NOD and CBA/H. This characteristic is not another
reflection of chr2 sensitivity as high scoring non-chr2 sensi-
tivities were identified (NOD for example), as were low scor-
ing chr2 sensitivities (RFM for example). This characteristic of
overall chromosomal radiation sensitivity is the subject of
In the F1 reciprocal crosses between C57BL/6 and CBA/H
(CBB6F1, B6CBF1) chr2 sensitivity appears to act as an at
least partially dominant genetic trait with no clear evidence of
differential inheritance from male or female (Table II). AMLs
can be induced in C57BL ? CBA hybrids, albeit at a some-
what lower frequency than in the inbred CBA parent (37).
Thus the behaviour of F1 hybrids in terms of radiation AML
sensitivity and chr2 sensitivity are consistent.
Mapping by association with microsatellites
The initial mapping strategy adopted to identify candidate
regions associated with chr2/AML sensitivity exploited the
publicly available microsatellite data for inbred mouse strains.
Given the origins of laboratory mouse strains (38) it was
predicted that genome regions harbouring chr2/AML sensitiv-
ity candidate genes would share common microsatellite alleles
in the chr2/AML sensitive strains and the microsatellite alleles
at these locations would be divergent in strains of dissimilar
phenotype. Microsatellite allele comparisons were carried out
by the computer-based analysis of mouse microsatellite data
provided at http://www-genome.wi.mit.edu/cgi-bin/mouse/
index. In principle this approach is analagous to that adopted
by Grupe et al. (20) but using microsatellite variation rather
than SNP variation. Initial tests of the validity of the method
were carried out, which involved attempting to predict the
locations of the mouse coat colour genes agouti, brown and
albino. Taking the observable coat colours of strains A, AKR,
BALB/c, C3H, C57BL/6, DBA/2 and LP and what may be
readily inferred of hidden coat colour genes through standard
crosses with tester strains, the strains were grouped according
to status at the agouti (a), brown (b) and albino (c) loci.
Computer comparison of microsatellite alleles was then car-
ried out for each locus searching for microsatellite loci where
strains within one group carried an identical allele that was
different in all members of the other group. For each of the
coat colour loci 3–6 clusters of microsatellite loci fitting the
criteria were identified along with several more dispersed
individual microsatellite loci. In each case a region spanning
?2–3 cM, including the locus under investigation plus 2–5
false positives, was identified. Thus, even with a limited
number of strains, this approach appeared to be successful in
reducing significantly the portion of the genome in which it
would be necessary to search for the genes of interest.
The false positives require that an independent confirmation
will be required in situations where new loci are being identi-
fied and mapped. With this caveat in mind, this approach was
applied to the chr2/AML sensitivity dataset. Of the 12 strains
for which phenotype information had been obtained, the whole
genome microsatellite data were available for 9 (A, AKR,
BALB/c, C3H, C57BL/6, DBA/2, LP, NOD and NON). These
Thus the chr2 sensitive group consisted of C3H, BALB/c and
LP, while A, AKR, C57BL/6, DBA, NOD and NON com-
prised the non-sensitive chr2 resistant group. Computer com-
parisons of microsatellite alleles between and within these
groups were carried out requiring loci identical within but
divergent between each group to be identified. The microsatel-
lite loci highlighted in this way are given in Table III.
Following these initial ‘in silico’ comparisons, microsatel-
lite typing of the strains not represented in the MIT/WIGR
database (CBA, SJL, RFM) was carried out for the loci given
in Table II. At this time all inbred strains were genotyped at
Table II. Variation in 3 Gy X-ray induced chromosome aberration scores for inbred strains and F1 hybrids between C57BL and CBA/H
Strain Painting protocola
Aberration score per 100 cellsVMR ? SEb
Sensitivity ratio for chr2d
8.16 ? 1.3
13.62 ? 2.2
18.75 ? 2.4
7.80 ? 1.9
6.67 ? 1.5
13.06 ? 2.3
8.33 ? 1.7
13.65 ? 2.5
17.94 ? 3.0
9.33 ? 1.8
6.75 ? 1.4
12.39 ? 3.0
6.10 ? 2.1
17.50 ? 3.3
10.10 ? 1.9
7.44 ? 1.1
12.81 ? 2.2
17.75 ? 2.4
8.26 ? 2.0
9.88 ? 1.7
21.25 ? 2.5
18.27 ? 2.3
19.90 ? 2.8
33.59 ? 4.1
23.33 ? 3.1
16.50 ? 2.2
20.56 ? 4.3
13.16 ? 3.2
19.62 ? 4.0
19.78 ? 2.4
11.94 ? 2.0
12.53 ? 7.2
16.75 ? 2.4
8.81 ? 1.7
7.51 ? 1.7
17.29 ? 2.4
15.85 ? 3.0
11.11 ? 3.0
22.14 ? 3.6
17.33 ? 2.9
11.25 ? 1.8
9.38 ? 2.6
4.86 ? 2.0
9.40 ? 2.2
12.90 ? 2.1
1.46 ? 0.06
1.84 ? 0.07
1.56 ? 0.07
1.12 ? 0.05
1.12 ? 0.09
1.14 ? 0.08
1.40 ? 0.08
1.67 ? 0.08
1.58 ? 0.09
1.85 ? 0.08
1.99 ? 0.07
1.58 ? 0.10
1.22 ? 0.11
1.36 ? 0.07
1.43 ? 0.10
aPainting protocol employed for strain: S, single colour; T, three colour.
bVariance to mean ratio, all chromosomes ? standard error.
cx2P-value testing for heterogeneity in aberration score between chrs 1, 2 and 3.
dchr 2 sensitivity ratio ¼ 2? (score per 100 cells for chr2)/(score per 100 cells for chr1 þ score per 100 cells for chr 3). At least 3 mice scored
144–400 cells per data point.
Genetics of radiation AML susceptibility
by guest on June 1, 2013
these loci to eliminate any confounding due to variation in
allele size between that quoted in the database and that in the
strains actually used in this study. Results are given in
Table IV. Manenti et al. (27) recommend that a Fisher’s
Exact Test ?log P-value of 42 be taken as an indication of a
significant association between a marker and the phenotype.
Under this criterion, the most significantly associating region
was found to be on chr18, apparently centring on D18Mit146.
A more detailed screen of markers in this region of chr18
?logP-values suggestive of the association were also obtained
for D3Mit319, D4Mit248 and D8Mit178 (Table IV). Several
discrepancies in allele size between the MIT/WIGR database
and the strains type tested here were noted, most dramatically
for D1Mit304, which did not appear to be polymorphic in the
animals tested here. For those markers most strongly associat-
ing (D18MIT146 and D18MIT67) either a unique allele asso-
ciating with AML resistance that is larger than alleles in the
AML sensitive group or the allele sizes associating with resis-
tance are all smaller than the alleles found in the sensitive
group. Thus there is a size relationship between alleles within
phenotypic groups. This is expected from the probable evolu-
tionary relationships between large and small alleles.
In an attempt to seek confirmation of the above mapping
exercise, a limited (CBA/Ca ? C57BL/6) ? C57BL/6 back-
cross analysis was carried out. A total of 36 first and second
generation backcross mice were phenotyped and then geno-
typed for the chr18 region. The range of chr2 sensitivity ratios
in these mice was 0.43–2.4. Using the criterion of a chr2
sensitivity ratio of ?1.45, indicating the sensitive phenotype,
8 mice were classified as chr2 sensitive and the remaining
28 as chr2 resistant. Of the 8 sensitive mice 7 were homozy-
gous for C57BL/6 alleles in the chr18 region, while 11 of the
28 chr2 resistant mice were C57BL/6 homozygotes for the
D18Mit markers. A Fisher’s exact test of association between
the D18Mit markers and the sensitive phenotype gives a
?logP-value of 1.7, which is a significant association on the
basis of the definitions proposed by the Complex Traits Con-
sortium (39). Unlike the previous analysis using inter-strain
comparison, in this case the C57BL/6 alleles appear to asso-
ciate with chr2 sensitivity. Genotyping for the other markers
identified in the previous analysis (Table IV) in 22 first gen-
eration backcross mice did not identify any regions with a
stronger association. However, ?logP-values suggestive of
significant association were found for D8Mit178 (?log
P-value of 1.44) and for D13Mit23 (?log P-value of 1.21).
While the finding of apparently differing allelic influences
of the chr18 region on chr2/AML sensitivity using the two
approaches is surprising, it is not without precedent. Tripodis
et al. (21) identified 30 modifier loci of chemically-induced
lung carcinogenesis and the allelic influence of many of these
was found to be dependent on the genotype of another locus.
Thus, the same genotype was found to be associated with
resistance in one case and susceptibility in another, depending
upon the genotype at another locus.
Chr18 loss of heterozygosity in AMLs
Loss of heterozygosity (LOH) for the chr18 region was inves-
tigated in AMLs. In some circumstances cancer risk modifiers
are frequently lost in the tumours to which they predispose (7).
DNAs from 22 AMLs induced by X-irradiation of F1 hybrid
mice (28,37) were examined for LOH at the chr18 markers
shown in Figure 1. No evidence of LOH was obtained (data not
Identification of candidate genes in the chr18 region and the
homologous human region
Given the evidence from two independent approaches that the
region on chr18 around markers D18Mit146 and D18Mit67
influences chr2/AML sensitivity, known or predicted can-
didate genes mapping to the region were identified from the
NCBI, Ensembl and UCSC mouse map viewers (Figure 2).
This region of the mouse chr18 is syntenic with human 18q11.
This region has been identified as being involved in structural
aberrations in 322 cases of cancer (http://cgap.nci.nih.gov/
Chromosomes/RecurrentAberrations), and amongst these are
24 cases of AML and 12 cases of refractory anaemia.
While several of the known or predicted genes within
this region might be associated with cancer susceptibility,
Table III. Microsatellite loci identified by initial comparisons based on
chr2 sensitivity phenotype grouping
Comparison Loci identified
LP, C3H, BALB/c identical
A, AKR, C57BL/6, DBA/2, NOD, NON
A, AKR, C57BL/6, DBA/2, NOD, NON
LP, C3H, BALB/c divergent
Fig. 1. Partial map of chr 18 showing markers checked for association with
chr2/AML sensitivity, strength of association (Fisher’s exact test ?log
P-value) and associating alleles. Map information was taken from
F.Darakhshan et al.
by guest on June 1, 2013
Rbbp8 (retinoblastoma binding protein 8, also known as CtIP,
CTBP interacting protein) is a particularly strong candidate. It
has been proposed that the chr2 sensitivity phenotype is driven
by chromatin structural features specific to the chr2/AML
matin remodelling through its interaction with Brca1 (41).
Furthermore Rbbp8 is phosphorylated by ATM following an
ionizing radiation exposure of human cells (42) and this phos-
phorylation depends upon BRCA1 (43). Indirectly Rbbp8 may
influence chromatin due to both Rbbp8 and polycomb proteins
Pc1 and Pc2 interacting with C-terminal binding protein 1,
Quantitative PCR analysis demonstrates that Rbbp8 expres-
sion in bone marrow is upregulated in response to in vivo X-
irradiation in CBA mice. This upregulation persists for a
period of 2 h. It is over time-scales of this order that chromo-
somal aberrations form. In contrast no upregulation of Rbbp8
is observed in C57BL/6 over the same time course. Thus
differential Rbbp8 expression in response to radiation may
underlie the phenotype described and lends further support to
the potential involvement of Rbbp8: it should be noted that the
expression of Cables-1 is not affected by irradiation of either
of the mouse strains. Therefore it is unlikely that there are
common cis-acting promotor mechanisms acting in this region
strains(40). Rbbp8is implicatedin chro-
Table IV. Experimentally determined microsatellite allele sizes at candidate loci and strength of association with the phenotype
Locus Allele size (bp)Allele associationa
Fisher’s test ?log P-value
Chr2 resistant strains Chr2 sensitive strains
C57A AKRDBA NONNOD C3H BALBLP CBA SJL RFM
aAllele(s) most strongly associating with chr2 sensitive (S) or non-sensitive (R) phenotype;—no strongly associating allele.
Fig. 2. Simplified scheme showing some of the known and predicted genes
locating to the chr18 region harbouring a chr2/AML sensitivity gene.
Table V. Human orthologous regions, AML involvement and candidate
genes for regions of interest
Mad3, Nsd1, Mox2(Hox8)
aAML, acute myeloid leukaemia; RA, refractory anaemia;
MDS, myelodysplastic syndrome; BLL, bilineage leukaemia; from
bFrom NCBI, Ensembl and UCSC genomics resources.
Genetics of radiation AML susceptibility
by guest on June 1, 2013
Candidate genes in other chromosomal regions
The inbred strain and backcross analyses identified associa-
tions between chr2 sensitivity and markers on chrs 3, 4, 8 and
13 in addition to chr18. Genomic and DNA sequence databases
were screened for candidate genes and human orthologous
regions were identified. These are given in Table V. It is
notable that all human orthologous cytogenetic regions are
recorded as having involvement in AML. This is most notable
for the chr8 region. Additionally most of the identified regions
carry reasonable candidate genes for involvement in suscepti-
bility and oncogenesis. Perhaps the most striking gene in the
current context is Nsd1 (Nuclear receptor binding Su-var,
Enhancer of zeste and trithorax domain protein 1), which
was identified by virtue of involvement in a t(5;11) in child-
hood AML (46,47). The protein is involved in transcriptional
regulation through interaction with the ligand binding domain
of the androgen receptor (48), possibly through chromatin
structural modification. Further analysis will be required
to identify definitively the genes involved in chr2/AML
sensitivity in the identified chromosomal regions.
The identification of loci which modify individual susceptibil-
ity is a key challenge in cancer research. Genetic factors have
been estimated to account for up to ?40% of individual risk
for a variety of sporadic cancer types (1,49). Modelling studies
suggest that cancers develop predominantly in genetically
predisposed individuals. Pharoah et al. (5) estimate that
?50% of sporadic breast cancers develop in the most suscep-
tible 12.5% of women and 80% of these tumours will occur in
the most susceptible 50% of the female population. These
and several other lines of evidence point to a major role of
multiple cancer risk modifying genes of relatively low
penetrance acting together to determine individual risk.
Experimental animal studies have a major role to play in
identifying these many cancer risk modifiers and increasingly
sophisticated tools are available for this purpose (7). Recently
the first successful application of the haplotype analysis in
identifying a cancer risk modifier in human and mouse skin
has been published (6).
Mouse models of the radiation-associated haematological
malignancy, AML, are available and inter-strain variation in
susceptibility to radiation-induced AML is apparent. In this
study, this inter-strain variation in AML susceptibility has
been exploited to map candidate modifier loci by genetic
association analysis. Previous work suggested that AML sen-
sitive CBA/H mouse bone marrow cells carried higher than
expected burdens of chr2 aberrations in comparison with
aberrations on chrs 1 and 3 shortly after whole body X-ray
exposure (18). In contrast, this phenotype is not observed in
the AML-resistant C57BL/6 inbred strain. A more complete
survey of the correlation between AML susceptibility and
bone marrow cell chr2 sensitivity is reported here. All
known AML sensitive strains also expressed the chr2 sensitiv-
ity phenotype, while known AML resistant strains did not.
Thus, the chr2 sensitivity assay is a valid surrogate for AML
Genome-wide comparison of polymorphic microsatellite
loci was exploited as a method for genetic association. The
easy availability of datasets make this an attractive and acces-
sible method. The principle of this method is essentially the
same as that described by Grupe et al. (20) for SNP/phenotype
association. The ancestory of laboratory mouse inbred strains
does not favour the use of SNPs for such association studies
due to the presence of a mosaic high/low SNP structure (50).
Testing of the microsatellite association method with known
mapped colour coat phenotypes/genes suggested that the
method is capable of reducing the proportion of the genome
that requires a detailed analysis for identifying the genes of
Using the microsatellite association method, a primary
computer-based screen, including the strains for which allele
size information is available (C57BL/6, A, AKR, DBA, NOD,
NON, C3H, BALB/c and LP), identified potential regions of
interest identified on chrs 1, 2, 3, 4, 5, 8, 11, 13 and 18
(Table III). The addition of new data on microsatellite allele
sizes from strains CBA, SJL and RFM indicated strongest
association in the D18Mit146-D18Mit229 region. The ?log
P-value from Fisher’s exact tests approached 3 for D18Mit146,
a highly significant association (P 5 0.001) as defined by the
Complex Trait Consortium (39). Regions identified on chrs 3,
4 and 8 reached the level of significant association (i.e. P 5
0.05 or ?log P-value 41.3). Using a backcross mapping
strategy and AML incidence as a phenotype Boulton et al.
(51) identified significant associations between AML suscep-
tibility and regions in chrs 1, 2, 4, 6 and 13. Thus there is
partial overlap between the chromosomes identified as har-
bouring AML modifiers in the present study and those identi-
fied in the previous investigation of radiation-induced AML
modifiers. In the Boulton et al. (51) study the chrs 1 and 6
regions were the most strongly associated. Despite some con-
cordance of the chromosomes implicated in harbouring modi-
fiers in the two studies, the distances between peak associated
markers islarge, except in the case of chr13, where markers are
separated by ?5 Mb.
Notwithstanding the differences in associated regions iden-
tified in the present study and that of Boulton et al. (51) both
come to the similar conclusion that multiple loci influence
radiation AML susceptibility. The difficulties of interpretation
of studies with similar aims identifying different regions are
widely recognized (39). The finding that multiple loci affect
radiation AML/chr2 sensitivity is not surprising in the light of
the extensive Tripodis et al. (21) study of modifiers of
chemically-induced mouse lung carcinogenesis, which estim-
ated 60 genes influencing susceptibility. Tripodis et al. (21)
also found that modifiers could exert opposing phenotypic
effect depending on genetic background. A similar phenom-
enon was observed in the present study for the chr18 region.
Furthermore, similar phenomena were reported for the AML
susceptibility loci previously reported (51). In the limited
backcross experiments, C57BL/6 alleles associated with chr2
sensitivity. Fisher’s tests indicated significant association
(?log P ¼ 1.7) for the chr18 region. Confirmation of signifi-
cant association at the chrs 8 and 13 loci was also obtained
(?log P ¼ 1.44 and 1.21, respectively).
The absence of LOH in the chr18 region in radiation-
induced AMLs suggests that susceptibility is conferred by a
mechanism distinct from that operating in familial retinoblas-
toma, for example. In such a situation, high frequency LOH of
one allele would be predicted. As this is not observed the locus
must be acting on other target genes or regions.
Within the chr18 region of interest, Rbbp8 stands out as a
key candidate gene. This gene encodes CtIP, the CtBP
interacting protein. CtIP interacts with multiple protein
F.Darakhshan et al.
by guest on June 1, 2013
partners, such as CtBP (52), Brca 1 (42), Ikaros (53) and Rb
(54). CtIP has been suggested to act as a tumour suppressor in
pancreatic cells (41), and the human Rbbp8 encoding chromo-
somal segment is lost or altered in cancers including AML and
refractory anaemia (http://cgap.nci.nih.gov/Chromosomes/
RecurrrentAberrations). The mouse Rbbp8 gene reacts
differentially to radiation in bone marrow cells from the AML
sensitive CBA strain and the AML resistant C57BL/6 strain
(Figure 3). Rbbp8 is upregulated uniquely in CBA mouse bone
marrow at 1–2 h post-3 Gy whole body X-irradiation. It is over
time-scales of this order that chromosomal aberrations form.
This provides evidence of a difference in response of Rbbp8 in
AML sensitive and resistant strains thus supporting the
involvement of Rbbp8 in post-irradiation chr2 sensitivity and
possibly AML sensitivity. Evidently further work is required
to confirm this suggestion.
Each of the other chromosomal regions identified in the
present mapping study gain support for an involvement in
AML susceptibility by virtue of the human orthologous
regions involvement in somatic rearrangements in AML
(Table V). Genomic database analysis allows the identification
of potential candidate genes within the regions of association.
Credible candidates can be identified but significant effort will
be required to provide direct evidence for involvement.
Boulton et al. (51) suggest that the haemopoietic stem cell
regulator Scfr1 located on chr1 may represent a key radiation
AML modifier. Evidently, alterations in stem cell number
will influence susceptibility. The present study suggests that
early acting factors which determine the nature of genetic
rearrangements following DNA damage might also play a
role. Previous work on AML chr2 breakpoints revealed
strong breakpoint clustering (40) and the genomic architecture
of the region suggested a role of chromatin remodelling in
determining the breakpoint clustering. Formal testing of the
direct radiation sensitivity of this site is required. However, a
model may be suggested in which strain-specific variation in
chr2 chromatin structure driven by CtIP or interacting factors
modifies radiation AML risk in mouse. The identification of
the human orthologous region, 18q11, as being involved in
human AML raises the possibility of similar mechanisms
acting in humans.
The data presented here serve to highlight the complexity of
the genetics of AML susceptibility. The use of a surrogate
endpoint for AML susceptibility, chr2 sensitivity, allowed a
more rapid phenotype analysis. An additional advantage is the
reduction in the experimental animal numbers required.
Microsatellite association is also shown to be a valid and
simple method to map to a crude scale at least cancer risk
modifier loci and indeed any other quantitative trait loci.
The authors thank Graham Bailey for assisting with statistical analyses. The
authors are grateful to Prof. John Todd for providing NOD mice and MRC,
Harwell for use of radiation facilities. This work was funded by EU contracts
MAGELLANS (FIGH-CT-1999-00035) and RISC-RAD (FI6R-CT-2003-
508842).F.D. wassupportedby an extra-muralresearchgrant to the University
of Reading from the National Radiological Protection Board.
Conflict of Interest Statement: None declared.
Environmental and heritable factors in the causation of cancer—analyses
of cohorts of twins from Sweden, Denmark and Finland. N. Engl. J. Med.,
2.Irving,M., Elmslie,F. and Berg,J. (2002) Genetics of breast cancer. Int. J.
Clin. Pract., 56, 677–682.
3.Ponder,B.A.J. (1990) Inherited predisposition to cancer. Trends Genet., 6,
4.Peto,J. and Houlston,R.S. (2001) Genetics and the common cancers.
Eur. J. Cancer, 37, S88–S96.
5.Pharoah,P.D.P., Antoniou,A., Bobrow,M., Zimmern,R.L., Easton,D.F. and
Ponder,B.A.J. (2002) Polygenic susceptibility to breast cancer and
implications for prevention. Nat. Genet., 31, 33–36.
6.Ewart-Toland,A., Briassouli,P., deKoning,J.P. et al. (2003) Identification
of Stk6/STK15 as a candidate low-penetrance tumor-susceptibility gene in
mouse and human. Nat. Genet., 34, 403–412.
7.Demant,P. (2003) Cancer susceptibility in the mouse: genetics, biology
and implications for human cancer. Nat. Rev. Genet., 4, 721–734.
8.International Agency for Research on Cancer. (2000) IARC monographs
on the evaluation of carcinogenic risks to humans, Vol 75 Ionizing
radiation. Part 1: X- and gamma–radiation, and neutrons. IARC Scientific
Publications No.75, IARC, Lyon.
9.Pierce,D.A., Shimizu,Y., Preston,D.L., Vaeth,M. and Mabuchi,K. (1996)
Studies of the mortality of atomic bomb survivors. Report 12, Part I.
Cancer: 1950–1990. Radiat. Res., 146, 1–27.
10.Upton,A.C., Wolff,F.F., Furth,J. and Kimball,A.W. (1958) A comparison
of the induction of myeloid and lymphoid leukaemias in X-radiated RF
mice. Cancer Res., 18, 842–848.
11.Hayata,I., Seki,M., Yoshida,K., Hirashima,K., Sado,T., Yamagiwa,J. and
Ishihara,T. (1983) Chromosomal aberrations observed in 52 mouse
myeloid leukaemias. Cancer Res., 43, 367–373.
(1988) Deletion of chromosome 2 is an early event in the development
Fig. 3. Quantitative PCR analysis of Rbbp8 (shaded bars) and Cables-1
(clear bars) expression in CBA (upper panel) and C57BL/6 (lower panel)
mouse bone marrow cells. Two mice of each strain were irradiated with
3 Gy X-rays in vivo and samples taken at 1–24 h. Expression levels in each
sample were assessed in triplicate. Bars represent the means obtained from
the two independent animals ? standard errors.
Genetics of radiation AML susceptibility
by guest on June 1, 2013
of radiation-induced myeloid leukaemia in SJL/J mice. Leukaemia, 2, Download full-text
13.Mole,R.H., Papworth,D.G. and Corp,M.J. (1983) The dose–response for
X-ray induction of myeloid leukaemia in male CBA/H mice. Br. J.
Cancer, 47, 285–291.
14.Major,I.R. (1979) Induction of myeloid leukaemia by whole-body
single exposure of CBA male mice to X-rays. Br. J. Cancer, 40,
15.Breckon,G., Papworth,D. and Cox,R. (1991) Murine radiation myeloid
leukaemogenesis: a possible role for radiation sensitive sites on
chromosome 2. Genes Chrom. Cancer, 3, 367–375.
Chromosomal abnormalities in neutron-induced acute myeloid leukaemias
in CBA/H mice. Radiat. Res., 146, 349–352.
17.Suraweera,N., Meijne,E., Moody,J. et al. (2005) Mutations of the PU.1 Ets
domain are specifically associated with murine radiation-induced, but not
human therapy-relatedacute myeloid
Chromosome 2 hypersensitivity and clonal development in murine
radiation acute myeloid leukaemia. Int. J. Radiat. Biol., 72, 181–189.
19.Bouffler,S.D., Breckon,G. and Cox,R. (1996) Chromosomal mechanisms
in murine radiation acute myeloid leukaemia. Carcinogenesis, 17,
Ahluwalia,M.K., Higuchi,R. and Peltz,G. (2001) In silico mapping of
complex disease-related traits in mice. Science, 292, 1915–1918.
Complexity of lung cancer modifiers: mapping of thirty genes and twenty
five interactions in half of the mouse genome. J. Natl. Cancer Inst., 93,
22.Balmain,A. (2002) Cancer as a complex genetic trait: tumor susceptibility
in humans and mouse models. Cell, 108, 145–152.
23.Goytisolo,F.A., Samper,E., Martin-Caballero,J., Finnon,P., Herrera,E.,
Flores,J.M., Bouffler,S.D. and Blasco,M.A. (2000) Short telomeres result
in organismal hypersensitivity to ionizing radiation in mammals. J. Exp.
Med., 192, 1625–1636.
(1995) A proposed system for scoring structural aberrations detected by
chromosome painting. Cytogenet. Cell Genet., 68, 211–221.
25.Papworth,D.G. (1970) Appendix 8 to Savage,J.R.K., Sites of radiation-
induced chromosome exchange. In Ebert,M. and Howard,A. (eds) Current
topics in Radiation Research. Vol. VI, pp. 187–191.
Bouffler,S., Finnon,P., Meijne,E., Huiskamp,R. and Cox,R. (1999)
Molecular mapping of chromosome 2 deletions in murine radiation-
induced AML localizes a putative tumour suppressor gene to a 1.0 cM
region homologous to human chromosome segment 11p11–12. Genes
Chrom. Cancer, 24, 95–104.
27.Manenti,G., Stafford,A., De Gregorio,L., Gariboldi,M., Falvella,F.S.,
Avner,P. and Dragani,T.A. (1999) Linkage disequilibrium and physical
mapping of Pas1 in mice. Genome Res., 9, 639–646.
28.Zhang,X., Ding,L. and Sandford,A.J. (2005) Selection of reference genes
for gene expression studies in human neutrophils by real-time PCR. BMC
Mol. Biol., 6, 4.
29.de Kok,J.B., Roelofs,R.W., Giesendorf,B.A., Pennings,J.L., Waas,E.T.,
Feuth,T., Swinkels,D.W. and Span,P.N. (2005) Normalization of gene
expression measurements in tumour tissues: comparison of 13 endogenous
control genes. Lab. Invest., 85, 154–159.
30.Rozen,S. and Skaletsky,H. (2000) Primer3 on the WWW for general users
and for biologist programmers. In Krawetz,S. and Misener,S. (eds)
Bioinformatics Methods and Protocols: Methods in Molecular Biology.
Humana Press, Totowa, NJ. pp. 365–386.
31.Vandesompele,J., De Preter,K., Pattyn,F., Poppe,B., Van Roy,N.,
De Paepe,A. and Speleman,F. (2002) Accurate normalization of real-
time quantitative RT-PCR data by geometric averaging of multiple
internal control genes. Genome Biol., 3, research 0034.1–research
32.Major,I.R. and Mole,R.H. (1978) Myeloid leukaemia in X-ray irradiated
CBA mice. Nature, 272, 455–456.
33.Rithidech,K.N., Bond,V.P., Cronkite,E.P. and Thompson,M.H. (1993) A
specific chromosomal deletion in murine leukaemic cells induced by
radiation with different qualities. Exp. Haematol., 21, 427–431.
Bowles,N.D. and Storer,J.B. (1976) The influence of dose and dose-rate
on the incidence of neoplastic disease in RFM mice after neutron
irradiation. Radiat. Res., 68, 115–131.
35.Resnitzky,P., Estrov,Z. and Haran-Ghera,N. (1985) High incidence
of acute myeloid leukaemia in SJL/J mice after x-irradiation and
corticosteroids. Leukaemia Res., 9, 1519–1528.
36.Mole,R.H. (1986) Radiation-induced acute myeloid leukaemia in the
mouse: experimental observations in vivo with implications for hypotheses
about the basis of carcinogenesis. Leukaemia Res., 10, 859–865.
37.Meijne,E., Huiskamp,R., Haines,J. et al. (2001) Analysis of loss of
heterozygosity in lymphoma and leukaemia arising in F1 hybrid mice
locates a common region of chromosome 4 loss. Genes Chrom. Cancer,
Festing,M.F. and Fisher,E.M. (2000) Genealogies of mouse inbred strains.
Nat. Genet., 24, 23–25.
39.Members of the Complex Trait Consortium (2003) The nature and
identification of quantitative trait loci: a community’s view. Nat. Genet., 4,
40.Finnon,R., Moody,J., Meijne,E., Haines,J., Clark,D., Edwards,A., Cox,R.
and Silver,A. (2002) A major breakpoint cluster domain in murine
radiation-induced acute myeloid leukaemia. Mol. Carcinogen., 34,
41.Wong,A.K.C., Ormonde,P.A., Pero,R. et al. (1998) Characterization of
42.Li,S., Ting,N.S.Y., Zheng,L., Chen,P.-L., Ziv,Y., Shiloh,Y., Lee,E.Y.-H.P.
and Lee W.-H. (2000) Functional link of BRCA1 and ataxia telangiectasia
gene product in DNA damage response. Nature, 406, 210–215.
43.Foray,N., Marot,D., Gabriel,A.,
Perricaudet,M., Ashworth,A. and Jeggo,P. (2003) A subset of ATM and
ATR-dependent phosphorylation events require the BRCA1 protein.
EMBO J., 22, 2860–2871.
44.Sewalt,R.G.A., Gunster,M.J., vander Vlag,J., Satijn,D.P.E. and Otte,A.P.
(1999) C-terminal binding protein is a transcriptional repressor that
interacts with a specific class of vertebrate Polycomb proteins. Mol. Cell.
Biol., 19, 777–787.
45.Kagey,M.H., Melhuish,T.A. and Wotton,D. (2003) The polycomb protein
Pc2 is a SUMO E3. Cell, 113, 127–137.
46.Jaju,R.J., Haas,O.A., Neat,M. et al. (1999) A new recurrent translocation
t(5;11)(q35;p15.5) associated with del(5q) in childhood acute myeloid
leukaemia. Blood, 94, 773–780.
47.Jaju,R.J., Fidler,C., Haas,O.A. et al. (2001) A novel gene, NSD1 is fused
to NUP98 in the t(5;11)(q35;p15.5) in de novo childhood acute myeloid
leukaemia. Blood, 98, 1264–1267.
48.Wang,X., Yeh,S., Wu,G., Hsu,C-L, Wang,L., Chiang,T., Yang,Y., Guo,Y.
and Chang,C (2001) Identification and characterisation of a novel
androgen receptor coregulator ARA267-alpha in prostate cancer cells.
J. Biol. Chem., 276, 40417–40423.
49.Peto,J. and Mack,T.M. (2000) High constant incidence in twins and other
relatives of women with breast cancer. Nat. Genet., 26, 411–414.
50.Wade,C.M., Kulbokas,E.J.I.I., Kirby,A.W., Zody,M.C., Mullikin,J.C.,
structure of variation in the laboratory mouse genome. Nature, 420,
51.Boulton,E., Cole,C., Knight,A., Cleary,H., Snowden,R. and Plumb,M.
(2003) Low penetrance genetic susceptibility and resistance loci impli-
cated in the relative risk of radiation-induced acute myeloid leukaemia in
mice. Blood, 101, 2349–2354.
52.Schaeper,U., Subramanian,T., Lim,L., Boyd,J.M. and Chinnadurai,G.
(1998) Interaction between a cellular protein that binds to the
C-terminal region of adenovirus E1A (CtBP) and a novel cellular protein
is disrupted by E1A through a conserved PLDLS motif. J. Biol. Chem.,
53.Koipally,J. and Georgopoulos,K. (2002) Ikaros-CtIP interactions do not
require C-terminal binding protein and participate in a deacetylase-
independent mode of repression. J. Biol. Chem., 277, 23143–23149.
54.Fusco,C., Reymond,A. and Zervos,A.S. (1998) Molecular cloning and
characterization of a novel retinoblastoma-binding protein. Genomics, 51,
Received December 24, 2004; revised August 1, 2005;
accepted August 2, 2005
F.Darakhshan et al.
by guest on June 1, 2013