This review
deals with the largest set of rat recombinant inbred (RI) strains and
summarizes past and recent accomplishments with this platform for
genetic mapping and analyses of divergent and complex traits. This
strain, derived by crossing the spontaneously hypertensive rat,
SHR/Ola, with a Brown Norway congenic, BN-Lx, carrying
polydactyly-luxate syndrome, is referred to as HXB/BXH. The RI strain
set has been used for linkage and association studies to identify
quantitative trait loci for numerous cardiovascular phenotypes,
including arterial pressure, stress-elicited heart rate, and pressor
response, and metabolic traits, including insulin resistance,
dyslipidemia and glucose handling, and left ventricular hypertrophy.
The strain's utility has been enhanced with development of a new
framework marker-based map and strain distribution patterns of
polymorphic markers. Quantitative trait loci for behavioral traits
mapped include loci for startle motor response and habituation, anxiety
and locomotion traits associated with elevated plus maze, and
conditioned taste aversion. The polydactyly-luxate syndrome Lx mutation has allowed the study of alleles important to
limb development and malformation phenotypes as well as teratogens. The
RI strains have guided development of numerous congenic strains to test
locus assignments and to study the effect of genetic background. Although these strains were originally developed to aid in studies of
rat genetic hypertension and morphogenetic abnormalities, this rodent
platform has been shown to be equally powerful for a wide spectrum of
traits and endophenotypes. These strains provide a ready and available
vehicle for many physiological and pharmacological studies.
quantitative trait loci; gene map; arterial pressure; heart rate; stress response; malformation phenotype
 |
INTRODUCTION |
THIS REVIEW DEALS WITH THE largest
available set of rat recombinant inbred (RI) strains and summarizes
past and recent accomplishments with this platform for genetic mapping
and analyses of markedly divergent and complex traits. These strains,
derived by gender-reciprocal crossing of a spontaneously hypertensive
rat, SHR/Ola, with the BN-Lx, Brown Norway (BN) congenic carrying
polydactyly-luxate syndrome (PLS), is referred to as HXB/BXH, denoting
their origins and reciprocal derivations. The HXB/BXH RI strain set has
been and remains a powerful tool for mapping quantitative trait loci (QTL) for complex phenotypes; its utility has been enhanced with the
recent development of a new framework marker-based map and strain
distribution patterns (SDPs) of polymorphic markers from our laboratory
(Jirout M, Krenova D, Kren V, Breen L, Pravenec M, Schork NJ, and
Printz MP, unpublished observations). Although researchers originally
envisioned that this rat platform would permit studies of rat genetic
hypertension and morphogenetic abnormalities associated with PLS, the
strains have turned out to be equally powerful for a wide spectrum
of traits and endophenotypes associated with many disorders. With
the development of new SDPs, publication of the rat genome sequence
(http://www.hgsc.bcm.tmc.edu/projects/rat/assembly.html), and
development of maps for rat single nucleotide polymorphisms (see
http://snp.cshl.org/), this rat platform has significant future
potential for physiological, behavioral, developmental, and
pharmacogenomic studies.
| |
DEVELOPMENT OF THE HXB/BXH STRAIN SET |
RI and congenic rodent strains constitute an assembly of unique
model organisms for genomic research (68, 69). These
genetic animal models have been exploited to investigate a wide variety of diseases (33) and complex traits (16).
Because of the long-term focus by geneticists on the murine species,
this species, and not the rat, has been the choice for RI
strain development with several sets of murine RI developed and
characterized. Fewer rat RI strain sets have been reported, and the
HXB/BXH strain set is, to the best of our knowledge, currently the
largest Rattus-based RI strain set. Development of the
HXB/BXH strain set commenced in 1982, jointly and cooperatively by
Vladimir Kren at the Institute of Biology of Charles University in
Prague, Czech Republic, and by Michal Pravenec at the Institute of
Physiology of the Czech Academy of Sciences, by using two widely
divergent inbred strains, SHR/Ola and BN-Lx/Cub.
Prior studies had led to the development of an inbred Wistar strain,
the PD/Cub, from outbred rats that exhibited in 1969 a spontaneous
mutation (Lx) predisposing to PLS (7, 34,
reviewed in Ref. 65). To better examine the
genetic origins of PLS, the trait was transferred onto the genetic
background of the normotensive Brown Norway rat (BN/Cub) to form the
congenic BN-Lx/Cub (34, 66). The morphological phenotype
associated with the Lx mutation was visually apparent, and
genetic transfer of this trait could therefore be tracked well before
the advent of modern molecular genotyping. However, unknown to the
investigators at that time, the PD/Cub carried other traits, documented
only recently, such as hypertriglyceridemia (72) and
insulin resistance (65), which have enhanced the
experimental utility of the HXB/BXH strain set for human diseases. The
original BN rat, Rattus norvegicus, was inbred from wild
rats captured in the United States in 1917. Breeding pairs of the
inbred BN strain were transferred to the Institute of Biology at
Charles University (Cub) in 1964 and brother-sister bred to
maintain homozygosity. The genetic homogeneity of the BN/Cub inbred
strain was verified by total genome scan (66). Transfer of
the PLS trait from PD/Cub onto the genetic background of the BN yielded
the BN-Lx/Cub congenic strain, a model of genetic determination of PLS
coded for by a major gene, Lx, the phenotypic manifestation
of which was strongly influenced by minor modifying genes from the
genetic background on which it expresses itself. The Lx
mutation was subsequently mapped to a differential segment of
chromosome 8 of PD origin in close linkage with other genes important
to the widespread utility of the RI strains, specifically, Drd2, Ncam,
and these apolipoprotein clusters: Apoa1, Apoa4, Apoc3, and ApoaV
(40, 43, 64, 66).
The SHR strain originated from an outbred Wistar colony trait selected
by Okamoto and Aoki (46) for spontaneous development of hypertension as young adults and inbred to form the SHR/Kyoto as a
potential genetic model of human essential hypertension. Breeder
pairs were transferred to Hanover, Germany, and subsequently to Prague
as the SHR/Ola. The selection of the SHR/Ola for the new HXB/BXH strain
set was deliberate, to enable future studies of the genetic basis of
this rodent model of human essential hypertension. The selection of the
BN-Lx was also deliberate because the BN is distantly related to albino
Wistar rats, which optimize the frequency of genetic polymorphisms
between the strains, facilitating genetic mapping. In addition,
carrying the PLS mutation permitted segregation analyses of this trait
and studies of the influence of varying genetic backgrounds. Finally,
BN and BN-Lx are normotensive strains that permit mapping of alleles
for the normal control of arterial pressure. Because the two overt
traits, hypertension and PLS, were considered to be independent, there
was minimal concern at the time of trait interference and that
assumption has largely been borne out. At the time that the intercross
was begun, little information existed regarding other traits contained within the progenitor's genomes; however, we now know that
polymorphisms in genes involved in metabolic disturbances, behavior,
developmental processes, and response to toxico- and teratogenic
stimuli were all contained within the genomes of the progenitors. As
discussed below, the assembly and segregation of genes led to
strain-dependent trait expression even when no significant differences
existed between the progenitors.
The construction of RI strains is direct and well described
(68): two highly inbred progenitor strains, as genetically
distant as possible, are mated to produce F2 hybrids. Any F2 hybrid
individual will carry a unique combination of genes because of both the
independent segregation of maternal and paternal chromosomes and
recombinations between homologous chromosomes during gametogenesis in
F1 hybrids. Independent segregation of maternal and
paternal chromosomes and recombination events during meiosis
result in a unique and practically irreproducible combination of genes
in any F2 individual. Subsequent inbreeding of randomly chosen pairs of
F2 animals and brother-sister mating for at least 20 generations yield
individual, relatively homozygous RI strains. These individual
strains carry a pattern of unique paternal-maternal gene combinations,
similar to the F2 animal. In addition, development of the HXB/BXH
strain set utilized gender reciprocal crossing, which provided two sets
of strains differing in the source of mitochondrial DNA and the Y chromosome. RI strains have several advantages over single-generation intercross or backcross progeny: 1) being inbred they
exhibit homozygosity at all loci; 2) all individuals of a
particular RI strain are identical replicas so that studies may be
replicated, permitting verification of any experimental result by
independent testing of genetically identical animals; 3)
phenotyping and genotyping data are cumulative; and 4)
studies may be conducted during development, pre- and postnatal, as
discussed below and shown in studies of target organ damage from
chronic pressure or volume overload (29).
The derivation of the HXB/BXH set started with gender-reciprocal
crosses between SHR/Ola and BN-Lx/Cub. The HXB set (H representing SHR
and of female gender, B representing male BN-Lx) was bred at the Czech
Academy of Sciences (Ipcv) so that all descendants carry mitochondrial
DNA of SHR origin with the Y chromosome (RNOY) from the male BN-Lx/Cub
strain. Originally, brother-sister inbreeding developed 26 HXB/Ipcv
strains; however, five strains (HXB/Ipcv 9, 14, 16, 19, and 30) were
lost after inbreeding due to poor breeding performance. The BXH set was
derived at Charles University from female BN-Lx/Cub and male SHR/Ola so
that the BXH set carries mitochondrial DNA of BN origin, whereas males
carry the Y chromosome from SHR/Ola. Altogether, 32 RI strains
presently exist that are all well beyond 50 generations of inbreeding,
21 HXB/Ipcv, and 11 BXH/Cub, with DNA preserved from all of the
original 36 strains for map construction.
The original colonies have remained in Prague, whereas a second,
pathogen-free colony has been established through embryo rederivation
in La Jolla at the University of California, San Diego
(40). Both colonies, Prague and La Jolla, are checked by
genotype and phenotype analyses. Frozen embryos are being preserved in
both Prague and La Jolla to prevent loss through aberrant breeding, infection, genetic contamination, or disaster. In addition to the two
sets of RI strains, there continues to be active congenic (39) and transgenic (56) strain development
to test putative QTL and to explore complex phenotypes. Within the past
6 yr, a complementary set of 10 RI strains (known as PXO) has been
developed with SHR.Lx and BXH2 strains as progenitors (see
DEVELOPMENTAL TRAITS WITHIN THE HXB/BXH AND PXO PLATFORMS;
Ref. 37). This set carries many of the disease
alleles common to the HXB/BXH but with a diminished portion of BN/Cub
genetic background and with homozygosity of the Lx locus,
facilitating further analyses of alleles that modify expression of PLS.
The overt physical characteristics of the HXB/BXH RI strain set are
detailed in Table 1.
| |
HXB/BXH GENETIC MAP AND SDPS OF
POLYMORPHIC MARKERS |
The power and precision of meiotic mapping of alleles for
complex traits depend on the accuracy of the genetic map for the species and strain under study and on the specific traits and models
selected. One of the most important attributes of RI strains is that
genotyping (like phenotyping) is cumulative (68), thereby facilitating both the initial construction and subsequent refinement of
SDPs of polymorphic markers and traits. The former can facilitate locus
mapping, whereas the latter can improve statistical measures of traits.
However, errors present in the cumulative data set can rapidly
attenuate mapping power and/or bias linkage results. Commencing in
1986, the original genetic characterization of this RI strain set,
essential to QTL mapping, was gradually developed (52-55). The application of minisatellite markers,
amplified fragment length polymorphism markers (47), and
microsatellite polymorphisms (51, 54) rapidly expanded the
SDPs and permitted identification of many putative QTL. However, the
map was still incomplete, and its power was weakened due to unlinked
markers on many chromosomes and the fact that most markers were not
commercially available and their map positions were not defined with
high accuracy to a single reference map.
With completion of the rat map (70) with high marker
position reliability (framework markers with LODs >3), a reference rat
map was available that could facilitate a reanalysis of the RI strain
set. This effort was undertaken in La Jolla and Prague, and the
resulting linkage map for the 20 autosomes was developed by Jirout et
al. (32) with the use of 245 microsatellite markers, predominantly framework, yielding 8,726 individual genotypes. The map
spans a total length of 1,789 cM, which is in good agreement with the
reference (70) and other published maps. This new map covers the rat genome contiguously and completely with an average intermarker distance of ~8.0 cM. The map for chromosome 1 is provided (Fig. 1A),
illustrating linkage of all markers; Fig. 1B provides a
sampling of the framework marker-based SDPs (32). The
original (55) and new (32) table of SDPs may
be accessed at http://www.ratmap.gen.gu.se.
View larger version (49K):
[in this window]
[in a new window]
|
Fig. 1.
A: example of correspondence between
original map for chromosome 1 (left) and new map
(right) based on framework markers. B: example of
framework marker-based strain distribution patterns (SDPs) for
chromosome 2. H, spontaneously hypertensive rat (SHR) genotype; B,
BN-Lx genotype. The table includes both existing and extinct strains.
Code: #H = HXB#; #B = BXH#.
|
|
With the new contiguous map and SDPs, an opportunity was available to
analyze the genetic composition of the strains important to QTL mapping
(32). A genetic similarity matrix indicated that the
progenitors' polymorphic microsatellite markers are evenly distributed
over all but 4 of the available 31 strains: strains BXH8 and BXH08,
which are genotypically similar due to derivation commencing several
generations after the F2 cross, and strains HXB3 and HXB15 (similarity
~78%) for unknown reasons (32). The evidence for
nonsyntenic association across the strains was also assessed by Jirout
et al. (32) as nonsyntenic loci showing allelic association may lead to false-positive signals (73). Three
examples of two different chromosomal areas with identical sequence
were identified, as well as three chromosome pairs with mirror image in
sequence. The former could lead to multiple QTL assignment, whereas the
latter could also yield an apparently opposite allelic influence. With
these analyses, the new SDPs provide enhanced validity for QTL
identification. In pilot studies (Jirout and Printz, unpublished
observations), the new SDPs significantly decrease the number of
strains needed to map loci (significant or highly significant LOD
values), provided that the trait measurements are accurate with
relatively low variance.
| |
CARDIOVASCULAR AND METABOLIC TRAITS WITHIN THE HXB/BXH PLATFORM |
Arterial pressure.
It would be expected that trait measures of arterial pressure would be
continuously distributed across the HXB/BXH RI strains because genetic
hypertension is a complex multigenic disorder and the SHR/Ola was one
of the progenitors. Systemic arterial pressure, although seemingly easy
to measure, is a very challenging phenotype as it is environment
sensitive (both dietary and through behavioral influences), reflects
the controlling influence of a large number of complex integrated
systems, exhibits diurnal variation, and is sensitive to many physical
characteristics of the animals, including body mass (58).
Although arterial pressure and hypertension were early traits in
mapping efforts, the results were far from convincing. We now know that
one problem was due to the available linkage map and SDPs (Jirout and
Printz, unpublished observations). Nevertheless, there were successes,
and these early results on the genetics of hypertension and other risk
factors for cardiovascular disease with the use of the HXB/BXH strain set, along with proof of locus identification through congenic development, have recently been addressed (41, 42, 50). For this reason, we will focus here only on recent and unpublished findings.
Initial arterial pressure phenotyping of progenitors and RI strains by
Kunes and colleagues (45, 52) documented a continuous distribution of systolic and diastolic arterial pressures among the RI
strains, suggestive of multigenic influences on the traits. On the
basis of this initial phenotyping and with the use of the older SDPs,
linkage and association analyses suggested several putative arterial
pressure loci, specifically on chromosomes 1 (associated with marker
Rt6), 2 (associated with D2N35), 4 (associated with marker Il6), 19 (associated with marker
D19Mit7), and 20 (associated with marker D20Utr1)
(37). These analyses were not optimized owing to
uncertainties in the map, the use of acute arterial pressure
measurements, and environmental effects on arterial pressure;
nevertheless, through the use of a maximum number of strains,
associations were identified. Although the exact positions of the loci
remain to be determined, many of the chromosomal assignments drew
subsequent confirmation through congenic strain development where
(large) chromosomal segments from the BN or BN-Lx were transferred onto
the SHR genetic background. However, it has also become clear that each
of these chromosomes likely contain multiple arterial pressure loci, as
has been shown recently for chromosomes 1 (28), 2 (4), and 10 (27). Furthermore, in the case of
chromosome 2 (4), both pressor and depressor loci are
close in proximity to one another.
To minimize environmental influences on arterial pressure measurements
and to examine the effects of diurnal variation and dietary NaCl
loading, we undertook to reanalyze all the RI strains in La Jolla using
radiotelemetry (58). In this study, which was modeled
after a typical clinical paradigm, samplings of systolic pressure,
diastolic pressure, heart rate, and activity were obtained over a 5-s
window, every 5 min, 24 h, for 12 wk. The first 6 wk consisted of
"baseline" measures with normal dietary NaCl intake, weeks
7 through 9 consisted of an 8% NaCl intake, whereas
weeks 10 through 12 was a return to baseline salt
intake. Over the entire 12 wk, rats were maintained in their home cages
in a specially instrumented vivarium room with human interruption only
to the extent of normal vivarium room operation. The data set, which largely occupies one server, is presently undergoing linkage
analysis and mining; however, much new and exciting information has
already been obtained. First, confirming the findings of Pravenec et
al. (52), a continuous strain distribution of arterial
pressures was evident with BN-Lx lowest and SHR/Ola highest (Fig.
2). However, not evident in Fig. 2 is
that the arterial pressures measured by radiotelemetry were
significantly lower than those measured by direct catheterization
(45, 52), likely reflecting the fact that the latter
measurements were taken under stressful conditions. There is,
nevertheless, a rank-order correlation between the two measurements. Second, the strain order distributions for
systolic and diastolic pressures are not identical, indicating that
different genes are likely influencing the two measures, a not
unexpected result given that a multiplicity of physiological systems
control these two measures of arterial pressure. Third, some strains
exhibit marked diurnal rhythms of arterial pressure or heart rate,
analogous to human "dippers," whereas other strains exhibit minimal
rhythm analogous to human "nondippers" (Fig. 2). Note that in the
measure of systolic pressure the two progenitors constitute the
extremes of the trait-strain distribution, as would be predicted.
Fourth, the arterial pressure of some strains is markedly sensitive to dietary intake levels of NaCl (analogous to salt sensitivity), whereas
other strains are quite resistant (salt insensitive). We documented
that the systolic pressure of the BN-Lx was salt sensitive, whereas
diastolic pressure was far less salt sensitive. Although normotensive,
during high salt intake, other RI strains exhibited lower systolic
pressure than the BN-Lx. The SHR/Ola remains the strain with highest
arterial pressures during both normal and high salt intake. Although
the progenitors tend to be the extremes in trait-strain distributions,
transgressive variation was evident for other cardiovascular traits.
These results indicate that genetic determination of arterial pressure
exhibits multigenic complexity, with influences from both the
environment and the genetic background of the strain, constituting
significant confounds for genomics of arterial pressure.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 2.
Illustration of continuous strain-dependent distribution
of systolic arterial pressure (SAP) during active and rest periods. SAP
averages are from a 3-wk period during baseline analysis (see text).
Values reflect either pressures during the rest light period
( ) or active dark period ( ). Note the
diurnal variation is present in some but not all strains. The BN-Lx has
the lowest SAP, whereas the SHR/Ola has the highest SAP. RI,
recombinant inbred.
|
|
It has been argued (44) that the use of inbred congenic
strains will facilitate identification of alleles that control arterial pressure and likely contribute to the hypertension syndrome. It is
certainly correct that congenic development is necessary for testing of
putative QTLs and to facilitate identification of candidate genes;
however, studies conducted with the HXB/BXH RI strain set by our
laboratories and others would argue that gene-trait relationships are
disrupted when the locus under test is placed onto an artificial, uniform genetic background. In such a case, an entirely new form of
genetic hypertension may have been created, while some endogenous genetic forms may have been lost. We would propose that RI strains, varying normally in genetic background, rather than congenic strains, offer the best opportunity for identifying those alleles that will more
likely have maximal influence on resting and active arterial pressures
in the population and thereby would likely be significant contributors
to genetic hypertension.
Metabolic traits.
In 1996, Bottger et al. (17) reported their linkage study
with the HXB/BXH RI strains in which QTLs for cholesterol and phospholipid phenotypes were identified. Using a genome scan of 534 polymorphic markers and high-cholesterol diets, they reported a locus
on chromosome 19 for HDL2 cholesterol. Subsequent congenic studies
identified suggestive loci on chromosomes 4 and 8 for serum
triglycerides (66). Since then, there has been a flurry of
papers examining several metabolic traits with the use of the RI
strains and/or congenics derived from them. As discussed
(66), it is now clear that the origins of these
cardiovascularly important metabolic phenotypes derive from all three
progenitors: PD/Cub, SHR/Ola, and BN/Cub (57, 65, 66). As
shown in Ref. 57, the metabolic traits
examined were continuously distributed across the RI strain set,
indicative of multigenic determination. Additionally, in several of the
traits, evidence of transgressive variation was found, with the
progenitors not constituting the extremes of the trait-strain
distribution. We have documented a similar phenomenon for
cardiovascular and behavioral traits (see below), whereas Zidek et al.
(74) have shown it for several reproductive traits.
Metabolic QTLs were mapped to chromosomes 3 and 17 for the
insulin-to-glucose ratio and to chromosome 7 for the intraperitoneal glucose tolerance test (57). Suggestive loci were also
identified and confirmed through congenic development to be present on
chromosome 19 for HDL2 cholesterol (57) and chromosome 8 for serum triglycerides (66).
Aitman et al. (2, 3) reported on the utility of the SHR
strain for studies of glucose and lipid metabolic traits. This was
followed in 1999 (1) with the report that studies had led the authors to focus on a locus in chromosome 4 that they identified, through gene expression studies and congenic and radiation hybrid mapping, to be close to a candidate gene, CD36, a fatty acid
translocase. They identified polymorphisms within the gene and proposed
that altered CD36 may be involved in insulin resistance and
dyslipidemia. In the past 3 yr, there has been a flurry of studies on
CD36, which go beyond the content of this review. It is worth
mentioning that the progenitor strains of the HXB/BXH set (SHR/Ola,
BN/Cub, and PD/Cub), as well as CD36 targeted congenic and transgenic strains derived from them, have been successfully exploited in pharmacogenetic analysis of the antidiabetic drugs, thiazolidinediones, and studies of their effects on carbohydrate and lipid metabolism (60, 63, 64). The results from these studies highlight
both the role of CD36 and the modifying effect of the genetic
background on drug effect.
| |
BEHAVIORAL, AIRPUFF STARTLE, AND STRESS-RELATED TRAITS WITHIN THE
HXB/BXH PLATFORM |
One of the most exciting new research areas for this RI strain set
lies in studies of complex behavioral traits and endophenotypes. The
HXB/BXH strain set was predicted to have traits for cardiovascular function (particularly arterial pressure) and for musculoskeletal abnormalities, and the finding of important metabolic traits was a
bonus; however, it is now apparent that this RI strain platform is also
of great value for behavioral traits. It is likely that, similar to
findings for metabolic traits, the utility of the strains for
behavioral traits stems from alleles present in all three progenitors
and the interaction of the genetic backgrounds. Our behavioral studies
with the RI strain set originated from studies of sensoriautonomic
coupling that used the airpuff and acoustic startle response (19,
21, 61). In a repeated airpuff trial paradigm with normotensive
Wistar-Kyoto (WKY/lj-cr), BN/hsd, and SHR (SHR/lj-cr) rats, the stress
component (pressor and tachycardia response) was elicited on every
trial, exhibited minimal habituation, and was strain dependent
(21). The orienting response bradycardia was observed in
adult WKY/lj-cr and BN/hsd animals, predominantly only on trials
1-3, and exhibited rapid habituation. In contrast, SHR
animals, from both the La Jolla colony and commercially derived, generally failed to exhibit any bradycardia response, only tachycardia on the initial airpuff trials (21). In these and
subsequent studies, evidence showed a genetic influence on both the
stress (sympathetic) and the orienting response (parasympathetic)
components of the airpuff startle reflex (20, 21, 30, 49,
59).
Because the SHR and BN (or WKY/lj-cr) strains exhibited marked trait
differences in the airpuff cardiovascular and behavioral components of
the startle response, the RI strain set gave us an opportunity to seek
stress-related QTLs for both the cardiovascular and behavioral (e.g.,
motor) traits associated with the airpuff startle reflex. Our
laboratory (31) used our standard chronic catheterization
procedure to examine a subset of the RI strains; with the new SDPs,
QTLs were identified for both cardiovascular and behavioral traits
(Fig. 3). We believe that this study is the first to report a QTL for the startle reflex and for loci controlling heart rate. Jaworski et al. (31)
identified two significant QTLs for the early trial bradycardia
associated with the orienting response: one on chromosome 2 (LOD 2.9, D2Rat62-D2Rat247) and a second on chromosome 3 (LOD 2.1, peak D3Rat20),
together accounting for nearly 50% of the trait variance. Since the
initial analyses, two additional loci have been confirmed with
significant or suggestive LOD scores but smaller contributions to the
variance; however, combining all four accounts for over 70% of the
variance. For airpuff startle stress tachycardia responses, we
identified two significant QTLs on chromosomes 1 (LOD 3.08, D1Rat287-D1Rat292) and 10 (LOD 2.4 D10Rat26-D10Rat267). A stress
arterial pressure QTL (based on the trait, airpuff-elicited change in
arterial pressure) was also identified on chromosome 6 (LOD 2.5, D6Rat80-D6Rat171), a chromosome not previously associated with genetic
control of arterial pressure (31). The finding that QTLs
for these divergent heart rate responses, elicited by an environmental
startling stimulus, are located on different chromosomes likely
reflects a multiplicity of complex control systems for both normal and
stress-dependent trait expression. Important to recognize is that the
loci for these heart rate responses are located on chromosomes that
also contain arterial pressure loci. This raises the potential for linkage, thereby potentially predisposing to enhanced cardiovascular morbidity or mortality. To examine this question further, the HXB/BXH
RI strain set, and associated congenics, provides a ready and available
genetic and physiological platform.
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 3.
Two-dimensional matrix illustrating chromosomes
containing quantitative trait loci (QTLs) with LOD scores either
"above significant" (black squares) or "above suggestive" (gray
squares) for traits associated with the airpuff startle paradigm. Trait
code is as follows: bMAP, basal mean arterial pressure (MAP); bHR,
basal heart rate; 1.9dMAP, averaged over all trials change in MAP; 1.9 dHR, averaged over all trials change in heart rate; t1-t5dMAP1.9,
change in MAP on trials 1-5; t1-t5dHR1.9, change in HR
on trials 1-5; t1 motor, averaged motor response on
trail 1; t1 latency, average latency response on trial
1; motor, averaged motor response over all trials; latency,
averaged latency over all trials; AVE PEAK, averaged peak response over
all trials; T1PEAK, peak motor response on trial 1; aveBW, averaged
body weight.
|
|
| |
ANXIETY-RELATED TRAITS AND ELEVATED PLUS MAZE |
On the basis of our findings with airpuff startle, our laboratory
also undertook an examination of the RI strains by using various
behavioral traits. Using elevated plus maze, Conti et al.
(22) characterized strains for several measures of anxiety and/or locomotion: percent time spent in the open arms, percent entries
into the open arms, number of entries into both the open and closed
arms, and number of entries only into the closed arms. SHR/Ola showed
both greater percent time in the open arms and percent entries into the
open arms than the BN-Lx progenitor, which was expected based on
published studies of heightened measures of anxiety in this strain;
however, it was found that there were no significant differences
between progenitor strains on the other two measures. All four traits
exhibited a continuous trait-strain distribution with significant
differences among the RI strains on each of the four phenotypes,
consistent with multigenic determination of the measures. With the use
of the new SDPs, significant QTLs were identified for anxiety-like
phenotypes on chromosomes 2, 5, 6, 7, and 17 and for a trait reflecting
both locomotion and anxiety on chromosomes 2, 5, and 6. Finally, traits
primarily associated with locomotion mapped to chromosomes 3, 8, and
18. For all but one strain examined, the presence of PLS did not appear to influence strain performance. In one strain, locomotion appeared to
be a confound to performance, and this strain was not included in the
linkage analyses. Interestingly, in an independent study (based on a
cross between BN/hsd and WKY/lj-cr), a QTL for prepulse inhibition of
the acoustic startle response, a measure of sensorimotor gating, was
mapped by Palmer et al. (48) to a nearby locus on chromosome 2.
| |
QTLS FOR STARTLE-ELICITED BEHAVIORAL TRAITS |
As discussed above, both behavioral and cardiovascular traits are
expressed in response to startling stimuli. Because the startle reflex
is a stressor and may elicit measures related to fear, anxiety, and
emotion, analyses of physiological and behavioral responses to a
startling stimulus may provide insight into a number of human
disorders, including anxiety, schizophrenia, and posttraumatic stress
disorder as well as fear-mediated cardiovascular arrhythmia. Using the
airpuff paradigm and only 23 RI strains, Jaworski et al. (Ref.
31 and unpublished observations) found
significant strain effects on peak startle amplitude, startle latency,
and percent startle habituation. Composite interval mapping
(32) identified the first QTLs for several of these
behavioral traits. Significant QTLs were found for peak startle
amplitude on chromosomes 7, 17, and 20, whereas startle latency mapped
to chromosomes 1, 4, and 9 and startle habituation mapped to
chromosomes 1 and 9. Because several of these loci are "near" those
found for plus maze anxiety traits (see above), the
possibility exists that they represent anxiety-related alleles. Such
QTLs and the RI strain platform should facilitate finding candidate
genes important to these behavioral endophenotypes.
| |
LOCI FOR CONDITIONED TASTE AVERSION |
Bielavska et al. (6) used the HXB/BXH RI strain set
to examine conditioned taste aversion by using lithium chloride as the
aversive substance. They observed a continuous distribution of values
across the RI strain set, with the BN-Lx progenitor exhibiting a
stronger aversion than the SHR progenitor, whereas seven strains
exhibited even stronger taste aversion values than BN-Lx. The SHR value
was on the opposite extreme. An association of conditioned taste
aversion (CTA) was found with two loci, one was mapped on
chromosome 2, around D2Cebr11s4 and D2Arb24, and one on chromosome 4, D4Cebrp149s8. The likelihood distribution on chromosome 2 was broad;
therefore, the relationship, if any, between this locus and that found
for other behavioral or stress loci (discussed above) cannot as yet be
determined. However, it is unlikely that these dissimilar traits
identify the same locus. As reported by the authors, the chromosome 2 CTA locus was supported by examining an SHR congenic carrying an
introgressed segment of BN-Lx chromosome 2.
It is of interest that a small number of chromosomes appear to contain
a collection of loci for different traits all related to behavioral
performance or behavior-autonomic coupling. This may be coincidental or
a reflection of a clustering of alleles involved in a common goal. The
HXB/BXH RI strain set provides a platform for examining this issue at
several levels of investigation.
| |
DEVELOPMENTAL TRAITS WITHIN THE HXB/BXH AND PXO PLATFORMS |
There is a substantial difference in the genetic determination
between the original two model diseases integrated into the experimental model of RI strains. Spontaneous hypertension from the SHR
strain is genetically determined by many factors that contribute in
varying degrees to the resulting blood pressure phenotype, thereby SHR
is multigenic and multifactorial. On the contrary, the PLS traits are
genetically determined by one major gene, Lx, the
inheritance of which is basically Mendelian but modified in its
expressivity and penetrance by polygenic influences of the genetic
background (7, 34, 35, 38). Because the morphometric
traits exhibit continuous distribution and are determined by allelic
contributions from the genetic background, these loci are QTLs. The
extreme variability of the PLS fixed in individual strains of both
HXB/BXH and PXO sets of RI strains, as well as congenic strains, is
therefore due to the interaction of the major gene, Lx, with
specific combinations of polymorphic morphogenetic genes of BN or SHR origin.
The Lx mutation was randomly distributed through the HXB/BXH
set of inbred strains, permitting studies of the genetic determinants of limb development. The vertebrate limb has been generally recognized as a model system for the study of the mechanisms that control pattern
formation during development (18, 25). Limb malformation mutants were also proposed to be a useful tool for studying mechanisms of normal development by Eteson et al. (26). Tabin
(71) proposed a molecular genetic model of limb
development involving complex interactions among retinoids, homeoboxes,
and growth factors. It is now generally accepted that genetic
determinants of human limb development find considerable homology of
mechanism with animal model systems. The HXB/BXH RI strains, along with
the new PXO RI set (37) and congenic strains that
carry leg paw malformations as a model of genetically determined
disease, facilitate genomic studies for understanding the complex
interactive processes underlying genetic control of development and
also genetic origins of similar human pathologies.
The original malformation phenotype of PLS in outbred Wistar rats
(34) consisted of preaxial polydactyly and zeugopod
affliction of the hind feet, which imparted a "luxate" appearance
to the carriers. During introgression of PLS onto the genetic
background of inbred SHR and BN strains, it was found that the disease
phenotype was coded for by a major gene designated Lx, the
phenotypic performance of which was a function of the genetic
background. On the SHR genetic background, the Lx gene
performance was completely recessive and the only manifestation of PLS
in homozygous SHR.Lx congenic animals was preaxial polydactyly of the
hind feet without significant zeugopod affliction. On the contrary, on
the BN genetic background, the Lx gene inheritance was
semidominant, with 60% penetrance in +/Lx heterozygotes
that carry preaxial polydactyly of the hind feet. Homozygous
Lx/Lx BN-Lx congenic animals exhibit polyphalangy up to
polydactyly of the front feet and preaxial polydactyly of the hind feet
with prominent affliction of the zeugopod. Such a substantial
difference in the phenotypic expression of the Lx mutant
gene between SHR and BN genetic backgrounds strongly indicates polymorphisms in many modifying genes involved in morphogenetic processes. The PLS phenotype was randomly fixed in 21 of 36 originally derived RI strains and currently resides in 19 of the surviving 32 strains. Morphometric analysis has demonstrated that any individual RI
strain with PLS represented a unique malformation phenotype, provided
that all the morphometric traits were taken together (36).
To expand the utility of the HXB/BXH RI strain set for developmental
analyses of limb malformation genetics, an additional set of strains
was developed. Progenitors for the new set of strains were two strains
exhibiting the most extreme malformation phenotypes: the SHR.Lx
congenic and the BXH2 RI strain. From these progenitors, 10 complementary strains were derived (39) and designated PXO to refer to hind feet polydactyly in SHR.Lx and oligodactyly in BXH2.
These strains were inbred through brother-sister mating and are
currently over 30 generations. A genetic map of the PXO set has been
derived consisting of more then 200 polymorphic framework markers
arranged according to the new HXB/BXH map (32) with residual heterozygosity being <5% (Kemlink D, Jeoabkova M, Janku D,
Krenova D, and Kren V, unpublished observations). The
malformation phenotypes exhibit similar variability of malformation
expression as the original HXB/BXH sets and, together with the original
set of strains, should prove valuable in future studies to identify genes responsible for phenotype variability.
A detailed morphometric digital image analysis of both the HXB/BXH and
the PXO strain sets was performed by measuring 12 selected parameters
and 2 indexes of the zeugopod (36) of 4-mo-old animals (Fig. 4). It was found that most of the
traits examined exhibited continuous distributions across affected RI
strains and progenitors, as illustrated for tibia length (Fig.
5). Interstrain statistical comparisons
with the 14 trait measures supported the finding that genetic
background strikingly modified expression of the Lx mutation (Table 2). There were no significant
differences between SHR.Lx and SHR, reflecting the moderating effect of
the SHR genetic background on Lx expression.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 4.
Identification of 14 dimensional measures of bone
geometry used for analysis of malformation phenotype across strains and
for QTL analyses.
|
|
View larger version (41K):
[in this window]
[in a new window]
|
Fig. 5.
Continuous distribution of tibia length in sampled
HXB/BXH and PXO RI strains (RIS) as well as progenitors.
|
|
Association analysis was conducted and indicated several significant
and suggestive associations between morphometric traits and gene
markers. With the use of the new framework-based map and SDPs for
HXB/BXH (32), several QTLs were identified, some found to
be closely linked with genes involved in morphogenetic pathways, for
example, D10Rat116, which mapped to 45.38 cM, closely linked (according
to UNIGENE) to Aldh3a2. To test these associations, double congenic
strains have been initiated by introgressing segments of SHR
chromosomes 2 and 4 onto the BN-Lx strain background. Each double
congenic strain should thus carry a differential segment of RNO8, with
the major PLS gene Lx and a second differential segment of
SHR origin with putative loci modifying the PLS malformation phenotype.
| |
USE OF THE HXB/BXH STRAIN SET FOR STUDIES OF TERATOGEN GENETICS |
Multigenic control of the physiological effect of teratogens and a
combined interaction with malformation mutant genes in mice were
reported as early as the 1960s (23, 24). Bila and Kren
(8, 9) developed a test for teratogenicity by using PLS
congenic strains on varying genetic backgrounds. This test exploited
the interactions of the teratogen with the Lx mutation, as
well as the influences of the genetic background on gene-teratogen interaction. When pregnant females carrying heterozygous
+/Lx fetuses were administered a teratogen, such as
bromodeoxyuridine or cyclophosphamide, on fetal days
11-13, the critical period for limb development,
newborns exhibited paw affliction with polydactyly up to oligodactyly
and variably afflicted zeugopod. The same dose of teratogen given to
females with +/+ fetuses was largely without effect, i.e.,
these females develop normodactylous progeny. Thus the presence of the
Lx mutant allele sensitized the carrier to the malformation
effects of the administered teratogenic substance. However, because the
effect of the mutant allele and the entire genome interacted in
phenotypic expression, background alleles either suppressed or enhanced
the teratogen influence. In the case of the SHR genome, the effect is
suppression, whereas with the BN genetic background there is enhanced
effects; thus the results argued for the presence of both positive and
negative modifying loci. Further application of the test yielded
positive results with drugs with known teratogenic, mutagenic, or
immunosuppressive potential (8-11, 15).
The teratogenicity test was also used for the characterization of a
class of antiviral acyclic nucleotide analogs, specifically 9-(2-phosphonomethoxyethyl)adenine (PMEA), which interacted with the
mutant allele to produce preaxial polydactyly of the hind feet, whereas
the 1-S-(3-hydroxy-2-phosphonomethoxyethyl)cytosine (HPMPC)
exhibited embryolethal effects (15). The teratogenic effect of thalidomide (12) was established after the
administration of the drug to females carrying +/Lx
heterozygous fetuses, whereas no malformations were ascertained in
fetuses homozygous in the standard allele. In heterozygous fetuses, the
Lx allele penetrance was increased up to 87% (10,
12). The genotype dependence of thalidomide teratogenicity
indicated the usefulness of a genetically defined system of RI and
congenic strains (8, 12).
The HXB/BXH strains have also been used to explore the teratogenic
effects of retinoic acid (RA) (13, 14). Vitamin A is known
to have a key role in embryonal development, and its metabolite, RA,
has been implicated in anterior-posterior patterning in vertebrate embryos (62). RA, acting directly or indirectly, affects a
complex network of receptors, transcription factors, and cytokines, and this has been reinforced by recent gene expression studies
(5). From 532 genes evaluated, 27 genes were identified as
being unquestionably direct targets of RA, whereas hundreds of genes
were influenced indirectly. To examine alleles involved in the action
of RA, studies were conducted in PLS carrying HXB/BXH RI strains and
congenics by administration of the RA to pregnant females
(14). It was found that RA administration resulted in a
significantly decreased body weight of all fetuses irrespective of
their genotype. The embryolethal effect of RA administration was most
pronounced in fetuses with the homozygous BXH2 genome (Fig.
6), where more than 90% of fetuses died
(14). Further studies have established that the SHR genome
not only ameliorated the Lx gene phenotypic manifestation but also buffered the teratogenic effect of RA. In this regard, the
SHR.Lx congenic strain and BXH2 RI strain appear to be the two extremes
in the sensitivity to RA teratogenic action, which parallels their
sensitivity to PLS expression. These findings would argue that both the
Lx mutation and varying genetic background within the
HXB/BXH and PXO RI strains are important determinants of the complex
interplay of genetic factors in RA teratogenesis.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 6.
Percentage of fetal resorptions resulting from retinoic
acid treatment of pregnant dams, by strain. Strains studied are from
both the HXB/BXH and PXO RI sets.
|
|
| |
SUMMARY AND CONCLUSION |
The rat has been the primary model organism for comparative
physiological and pharmacological studies, and a wealth of information has accumulated over the years in the literature. Only a few years ago,
a genetic map for the rat to facilitate genomic studies was not available (67); however, all that has changed.
Although the rat has been the organism of choice for physiologists and pharmacologists and the mouse has been the organism of choice for
geneticists (67), integration of the sciences in this era of genomics has blurred the distinction between these species. Popularity of murine strains among physiologists and pharmacologists over the past 5 yr is largely attributable to the advent of gene targeting afforded by the unique L129 strain. Without question, for
studies seeking to model or study a single gene product on a single
genetic background, genetically engineered mouse models are today the
logical choice; nevertheless, this may need to be reexamined in the
near future because of advances in nuclear transfer. However, if the
goal is to define and thoroughly understand human and comparative
genomics, it can be argued that the rat remains a more logical model
organism for study. This position is based not only on the vast
literature but also on the availability of a large number of
well-defined inbred strains, unique genetically informative model
constructs, including a rapidly growing number of congenic substrains
and the subject of this review, the HXB/BXH RI strain set. In addition,
the rat genetic map rivals the mouse and the recently announced rat
genomic sequence makes the rat an ideal comparative model for human disorders.
The HXB/BXH RI strain set along with congenics and the new PXO RI set
has been shown to have utility in five major research areas:
cardiovascular, metabolic, behavioral, developmental, and toxicological. Additionally, the strains can be starting points for new
constructions designed to address specific questions and or gene
targets. Lastly, with today's technology and the accumulated knowledge
of inbred rat strains facilitating selection of future progenitors, the
construction of new rat RI strain sets can be accomplished
expeditiously. As pointed out by Blizard (16) through genetic correlation analysis, the use of RI strains provides an alternative method to physiological or pharmacological methods to
explore relationships between different domains in the behavioral sciences. Recombinant inbred methodology should enhance studies of the
relationships between complex processes, and this application should be
considered separate from its use in gene mapping (16).
Preparation of this review was supported by National Heart, Lung,
and Blood Institute Grant HL-35018 (to M. P. Printz) and Grants
Grant Agency of the Charles University 38/01 and Grant Agency of the
Czech Republic 204/98/K015 (to V. Kren).
Address for reprint requests and other correspondence:
M. P. Printz, Dept. of Pharmacology 0636, Univ. of
California San Diego, La Jolla, CA 92093-0636 (E-mail:
mprintz{at}ucsd.edu).
TOP
ABSTRACT
INTRODUCTION
a new mutant of the Norway rat.
Folia Biol (Praha)
24:
369-370,
1978.
