Minggu, 03 Juli 2016

Genetic Modifiers in Rodent Models of Obesity

source: http://www.cimedicalcenter.com/


Individuals affected with the same genetic disorder often differ in their clinical
presentation. This effect is evident in the intrafamilial variability observed in weight gain
and glycemic status in syndromic diseases such as Bardet-Biedl and Alström syndromes,
in which all affected family members carry the same mutation (1, 2). Intrafamilial
variability in disease phenotypes may be due to environmental influences, genetic
modifier loci, or a combination of these factors. In addition, interfamilial variability may
be due to allelic differences at specific loci.

Whereas much work has been done on the environmental influences, e.g., diet or
exercise on development of obesity and type II diabetes, the role of genetic modifiers is
gaining prominence. The phenotypic effects of modifier genes on the manifestation of a
primary disease mutation can arise from the modifier’s action in the same or in a parallel
biological pathway as a disease gene. The effect can be enhancing, causing a more severe
mutant phenotype, or suppressive, reducing the mutant phenotype even to the extent of
completely restoring the wild-type condition. Modifier genes can also alter the pleiotropy
of a given disease, resulting in different combinations of traits. In addition, for any given
genetic disorder, alleles of multiple modifier genes may act in combination to create a
final, cumulative effect on the observed phenotype. The latter situation may be especially
true for complex disease traits such as obesity and type II diabetes, for which it has been
extremely difficult to identify underlying genes in the human population.

Studying and identifying genetic modifier loci can yield new insights into the
biological pathways in which Mendelian disease genes act and through which they cause
disease phenotypes. For example, knowing the molecular basis of a genetic modifier may
improve diagnosis and treatment of disease, perhaps by defining a particular subgroup
within the disease population. In addition, the identification of modifier genes may lead
to new treatments either by providing additional information about the genetic
contributions to the phenotype for which treatment may already be available or by
pointing to additional steps in a biological pathway that may be more amenable to

In obesity, environmental influences have typically been emphasized over genetic
causes for the phenotype. Examples of modifier genes in human studies are not abundant;
their existence can, however, be inferred by the finding of association of obesity
subphenotypes with particular alleles of genes that have been implicated in obesity. In
animal models of obesity caused by a mutation in a single gene, it can be shown that
modifier genes influence phenotypic expression. The fact that most obese individuals do
not develop non-insulin-dependent diabetes mellitus (NIDDM), whereas most patients
afflicted with NIDDM are obese, can be interpreted to mean that obesity (and obesity
genes) are necessary but not sufficient for the development of NIDDM and that NIDDM
susceptibility genes may act as modifiers of obesity genes. Better known are the effects
of modifier genes in causing non-insulin-dependent type II diabetes in obesity models. In
mouse models, such as C57BLKS-Lepob/Lepob, the obesity mutation is necessary but not
sufficient for the development of diabetes. In this review we will focus on the role of
genetic modifiers in rodent models of obesity and diabetes and provide some examples of
reported modifier gene action.


The earliest documented gene/gene interaction in an obesity pathway was found between
the mouse mutations obese (ob) (3) and diabetes (db) (4). C57BL/6-ob/ob and C57BL/6-
db/db mice are hyperphagic, hypometabolic, and massively obese. Because of the
phenotypic resemblance of the two mouse strains, Coleman and Hummel undertook a
series of parabiosis experiments. Connecting the blood supplies of ob/ob mice with those
of wild-type mice caused the ob/ob mice to loose weight. Parabiosis of db/db mice to
wild-type mice led to starvation and weight loss in the wild-type mice, while the db/db
mice maintained their body weight. And finally, parabiosis of ob/ob and db/db mice led
to weight loss in the obese ob/ob mice. Apparently, ob/ob mice were lacking a bloodborne
factor that prevented obesity that wild-type mice possessed, and db/db mice could
not respond to that factor and overproduced it. From these results, Coleman concluded that the ob locus might encode a satiety hormone and the db locus, its receptor. This
prediction was proven correct by the identification of the obese gene as Lep, the
adipocyte-secreted hormone, leptin (5), and diabetes as the leptin receptor (Lepr) gene,
acting primarily in the hypothalamus (6, 7).

The example of the interaction between Lepob and Leprdb is unusual in that it was not
demonstrated by genetic means. More typically, gene/gene interactions are discovered by
observing the suppression of a phenotype in double-mutant mice or by the appearance of
a phenotype in compound heterozygous animals.

An example for the former is the discovery of an interaction between the coat color
mutations yellow (Ay) at the agouti locus on chromosome (Chr) 2 and the nonallelic
mahogany (Atrnmg) and mahoganoid (Mgrn1md) loci on Chrs 2 and 16, respectively. In
addition to a yellow coat color, yellow mice develop obesity due to the ectopic
expression in the hypothalamus of agouti signal protein (ASP), a melanocortin receptor
antagonist normally expressed only in the skin (8,9). In order to determine where
mahogany and mahoganoid lie with respect to agouti signaling in a genetic pathway,
Miller and colleagues created double-mutant animals (10). They found that homozygosity
for either the mahogany or the mahoganoid mutation suppressed the effects of Ay on coat
color as well as on obesity, suggesting that mg and md act downstream of agouti to
interfere with agouti signaling. Both mutant genes have been identified by positional
cloning. The mahogany gene codes for the membrane protein, attractin (11,12), which
may act as a low-affinity receptor for the agouti protein (13). Mahoganoid codes for a
novel RING-containing protein with E3 ubiquitin ligase activity, which may function in
protein turnover (14,15).

In some cases, enough is known about a molecular pathway to test directly for
interactions between genes by creating double-mutant mice. Neuropeptide Y (NPY) is an
orexigenic peptide that stimulates feeding when injected into the third ventricle of the
hypothalamus (16). Administration of leptin suppresses hypothalamic expression and
release of NPY; NPY is elevated in leptin-deficient mice (17,18). Mice deficient in NPY,
however, show no abnormality in feeding behavior (19) and only a slight increase in
body weight (20), suggesting the existence of additional pathways controlling feeding in
mice. By generating mice that were deficient for both leptin and NPY, Erickson and
coworkers showed that the lack of NPY in these animals attenuated the obesity normally
observed in Lepob/Lepob mice by reducing their food intake and increasing their energy
expenditure (21). This indicated that NPY is a major effector in leptin signaling.
The last example shows, in particular, that a candidate gene approach to gene/gene
interactions can provide important confirmation of hypotheses regarding biological
pathways. With our rapidly increasing knowledge about gene function, this approach will
gain more importance in the future.


As seen earlier, the identification of modifier genes can be an important component of
understanding biological pathways that lead from a primary mutation to a disease
phenotype. Whereas a candidate gene approach is limited by our knowledge of gene
function, a reverse genetic approach—going from a phenotype to the causative
Genetic modifiers in rodent models of obesity 3
underlying gene—does not require prior knowledge about the interacting genes, and
could be a powerful method for identifying novel pathways. A major stumbling block for
the identification of genetic modifiers of obesity and type II diabetes in humans will be
the difficulty of mapping these loci in the face of the huge genetic heterogeneity in the
human population. And whereas chromosomal localization of modifier loci in large
human families segregating for monogenic diseases that cause obesity and type II
diabetes may be feasible, a real problem will be moving from a general map position to a
narrow enough region that fine physical mapping and gene identification can commence.
Here again, we may be able to use the available monogenic mouse obesity models to
gain insight into the pathways that influence obesity and hyperglycemia. The most
straightforward approach to mapping modifier genes in the mouse is to carry out crosses
between inbred strains that carry the disease-causing mutation and in which a difference
in phenotype is observed. A schematic for this approach is shown in Fig. 1, in which the
obesity phenotype of mice deficient in leptin (homozygous Lepob/Lepob mice) is modified
by the C57BL/6J (B6) and BALB/cJ genetic backgrounds (22). B6-Lepob/Lepob mice
develop an early onset, severe obesity, whereas BALB/cJ-Lepob/Lepob mice are reported
to be obese but lighter than B6-Lepob/Lepob. This suggests the presence of genes in the
BALB/cJ background that moderate weight gain. When F1 offspring from a mating
between B6-Lepob/Lepob and BALB/cJ-+/+ are intercrossed, the modifier genes should
segregate in the F2 population. Because a modifier gene itself does not in most cases
produce a phenotype, only F2 animals that are homozygous for the primary diseasecausing
mutation (in the case of a recessively inherited disease) will show phenotypic
variation and thus be informative for the analysis. In those F2 animals, standard
quantitative trait locus (QTL) analysis methods can be used to map the modifier loci (23).
In order not to confuse background QTLs with modifier genes, the F2 animals that do not
carry the disease mutation should be examined for variation in the trait of interest. If there
is variation, then the background QTLs should also be mapped to distinguish the loci that
affect the trait independent of the disease mutation from the true modifier loci. It should
be pointed out that the primary mutation does not necessarily have to lead to phenotypic
differences in the two parental strains used in the modifier cross. Occasionally, modifier
genes are unmasked only by the interaction of the two genetic backgrounds in the
segregating F2 population (24). This is the case, for example, for the fat mutation:
Although the body weights of C57BLKS-Cpefat/Cpefat mice do not differ much from those
of HRS-Cpefat/Cpefat, in the F2-Cpefat/ Cpefat population from a (C57BLKS×HRS) F1-
Cpefat/+ intercross, body weights vary from normal to severely obese (25).
Although less difficult in the mouse, identification of genetic modifiers of obesityrelated
traits for which the chromosomal locations have been mapped may still be
challenging, especially if more than one gene is contributing to the modification of the
phenotype. If a major modifying locus is found (explaining >40% of the phenotypic
variance), then conventional

fine-structure mapping in a large F2 intercross combined with progeny testing can be
used to narrow the genetic interval sufficiently to proceed with positional cloning (26). In
cases where multiple loci contribute to the phenotypic variance, it may be necessary to
construct congenic lines to isolate individual modifier loci (23,27). If the phenotypic
effect of the modifier locus in the congenic line is greater than that of the nongenetic
variation, then the line can be used in crosses for fine-resolution mapping, as in the case
of the major modifier. Once a high-resolution map has been obtained, conventional
positional cloning techniques may be applied (26). Currently available methods such as
gene expression microarray analysis may be combined with the use of congenic lines to
directly identify a misregulated modifier allele, or to point to the misregulation of a
pathway in which the modifier gene plays a role (28).


The first recognition of modifier genes in obesity research dates back to the study of the
Lepob and Leprdb mutations on different genetic backgrounds. Coleman and his coworkers
noted that B6-ob/ob and B6-db/db mice became obese but remained diabetes-free.
However, when placed on the related C57BLKS (BKS) inbred strain background, both
mutations led to severe diabetes (29,30). This indicates that the BKS genetic background
is diabetogenic, i.e., BKS carries alleles of diabetes susceptibility genes that are necessary
but not sufficient for the development of overt diabetes. These diabetes susceptibility
alleles have to interact with obesity mutations such as Lepob and Leprdb to cause
hyperglycemia. Although the major diabetes modifiers in BKS have yet to be mapped
(31), modifiers of leptin action have been reported in other mouse and rat strains. In
addition, obesity and diabetes modifiers have been reported for different obesity

4.1. Lep and Lepr Mutations
Although the existence of genetic background modifiers affecting glycemic status in the
context of leptin receptor mutations was first recognized in mice, the first published
mapping studies were carried out in the rat model. The Zucker fatty rat carries a
Gln269Pro mutation in the leptin receptor that leads to obesity, hyperinsulinemia, and
glucose intolerance. The animals, however, are, normoglycemic (32). In contrast, the
same mutation when transferred onto the WKY strain background causes obesity,
hyperinsulinemia, and hyperglycemia (33). Chung and colleagues used this strain
difference to map NIDDM susceptibility loci in an F2 intercross between animals of the
WKY and 13M strains homozygous for the Leprfa mutation (34). Significant
genotype/phenotype associations were found on rat Chr 1 for pancreatic morphology, on
Chr 12 for body weight, and on Chr 16 for plasma glucose levels. It is interesting to note
that a number of obesity/diabetes related traits have been mapped to the same region of
Chr 1 in other rat models (35) and in the homologous region on mouse Chr 19 (36,37).
Whether these loci represent variations in the same gene remains an open question until
the genes are cloned. The identification of these loci, however, promises new insights
into the reasons for pancreatic failure in type II diabetes.

Similar to the findings for Lepr mutations, phenotypes of leptin mutations can also be
modified by strain background. Apart from the original observation of hyperglycemia in
BKS-Lepob/Lepob mice versus normoglycemia in B6-Lepob/Lepob [30], modifications of
body weight, insulin levels, and glucose levels have been reported in the BALB/cJ (22)
and the BTBR strain backgrounds (36). In Lepob/Lepob F2 offspring from an intercross of
(B6×BTBR) F1-Lepob/+ mice, Stoehr et al. were able to map three loci controlling insulin
and glucose levels on Chrs 2, 16, and 19 (36). Interestingly, it is the B6 allele on Chr 19
that contributes to increased plasma glucose levels, yet B6-Lepob/Lepob mice are protected
from diabetes. Susceptibility contributions from an overall resistant background are not
uncommon (24), and in this case the resistance of B6 to overt diabetes can be attributed
to an interaction between the loci on Chrs 19 and 16. BTBR alleles on Chr 16 are
necessary to unmask the deleterious effects of the B6 allele on Chr 19 (36).
Genomics and proteomics in nutrition 6

4.2. Tub Mutation
Mice homozygous for the tubby mutation (tub) are a model for sensory loss/obesity
syndromes such as Alström syndrome (25,38). Tubby mice develop late onset obesity
with insulin resistance, early onset retinal degeneration, and neural hearing loss [39–41].
The tubby phenotype is due to a loss-of-function mutation in the novel Tub gene, a
member of the small gene family encoding tubby-related proteins (TULPs) [42–44]. The
biochemical function of theTULPs is not fully understood. Roles as transcription factors
(45), as intermediates in insulin signaling (46), and in intracellular transport [47–49] have
been proposed. Identification of genetic modifiers of the different phenotypes observed in
tubby mice could provide additional clues to the pathways involved, and so may lead to
further insights intoTUB function.
The first modifier of a tubby phenotype to be identified was moth1, the modifier of
tubby hearing 1. In an F1 intercross following a cross between B6-tub/tub and AKR/J, it
was observed that F2 offspring homozygous for the tub mutation varied widely in their
hearing ability from normal hearing to profound deafness. Hearing was quantified
electrophysiologically by measuring auditory brainstem response in the F2-tub/tub
population, and a major QTL, moth1,was mapped to Chr 2 (50). In the absence of the tub
mutation, i.e., in the wild-type B6 strain, this locus has no effect on hearing. Standard
positional cloning techniques were used to identify moth1 as an allele of the geneencoding
microtubule-associated protein 1A (Map1a) (26). The B6 Map1a allele,
associated with hearing loss, carries 12 amino acid alterations and an Ala-Pro repeat
length polymorphism compared to the protective AKR allele. It was shown that these
polymorphisms lead to a weaker binding of the B6 MAP1A protein than of the AKR
variant to the postsynaptic density protein PSD95. That Map1a is indeed moth1 was
confirmed by a transgenic rescue experiment showing that B6-tub/tub mice carrying a
protective 129P2/OlaHsd allele of Map1a have nearly normal hearing. MAP1A has been
shown to be important in trafficking of vesicles and organelles, and PSD95 is a major
component of the synaptic cytoarchitecture. The identification of the moth1 modifier has,
therefore, genetically established that synaptic architecture and intracellular transport are
relevant toTUB function.

Although moth1 has provided more functional information, the findings are
compatible with both the transcription factor and the transport hypotheses of TUB
function. There are still additional modifiers to be identified that may yield further insight
into TUB function. Apart from hearing ability and vision loss, adiposity and plasma
levels of glucose, insulin and lipids also show variation in F2 progeny of crosses between
B6-tub/tub and AKR (25,51), (A.Ikeda, personal communication, 2002). A genome-wide
scan using 57 microsatellite markers distributed at about 30-cM intervals was performed
on 43 female and 37 male F2-tub/tub mice. Several statistically significant and suggestive
QTLs have been found for body and fat pad weights as well as for plasma insulin levels
on Chr.6 (p<4.5×10−6, p<2.6×10−6, p<2.6×10−6 respectively), for plasma cholesterol on
Chr.8(p<2.6×10−6),and for plasma glucoselevelsonChr.4(p< 3×10−4).

4.3. Cpefat Mutation
The mouse fat mutation is a complex model for obesity and type II diabetes (38). The
underlying defect is a mutation in the carboxypeptidase E (Cpe) gene (52), which codes
Genetic modifiers in rodent models of obesity 7
for an enzyme responsible for the final proteolytic processing step of prohormone
intermediates, such as those for insulin and proopiomelanocortin (53). Because a large
number of neuro/endocrine peptides are affected by the Cpefat mutation, the etiology of
obesity and diabetes in the mutant mice is not clear. The identification of modifier genes
in this case should point to pathways that are critical for the expression of a particular

Cpefat is a typical disease gene, i.e., it is necessary but not sufficient for the
development of obesity, type II diabetes, and related metabolic disorders. On the HRS/J
(HRS) inbred strain background, Cpefat/Cpefat mice exhibit early onset hyperinsulinemia
followed by postpubertal obesity without hyperglycemia. In contrast, on the C57BLKS/J
(BKS) genetic background, Cpefat/Cpefat mice become hyperglycemic as well as obese
and hyperinsuli-nemic. In order to map the susceptibility loci responsible for modifying
obesity and diabetes associated traits, Cpefat/Cpefat male progeny from a large F2
intercross between BKS.HRS-Cpefat/Cpefat and HRS-+/+ mice were characterized both
genetically and phenotypically. All traits measured—body weight, adiposity, fat pad
weights, plasma glucose, insulin, triglycerides, and HDL and non-HDL levels—showed a
large variance in the F2 population, indicating the action of modifier loci (24). A
genomewide scan was carried out on 282 male Cpefat/Cpefat F2 progeny, and four major
modifier QTLs for Cpefat were detected. Three loci for hyperglycemia (find2, find1,
findc) were mapped on Chrs 5, 19, and 2, and one locus for adiposity (fina1) on Chr 11
(54). Interestingly, at find1 it is the HRS allele that contributes to hyperglycemia,
indicating that there must be another, as yet unidentified, HRS locus that counteracts
diabetes development to maintain normoglycemia in HRS-Cpefat/Cpefat mice.


Although obesity appears to be a prerequisite for the common forms of human NIDDM,
mutations in genes that are more directly associated with diabetes can also be used to
further define the pathways that lead to NIDDM.
Mutations in the insulin receptor can be used to model insulin resistance. Kido et al.
showed that the effect of reduced insulin receptor activity in animals heterozygous for an
Insr-targeted allele are dependent on the genetic background (55). Male 129S6/SvEvTac-
/Insrtm1Dac/+ are hyperinsu-linemic and have slightly elevated plasma glucose levels
compared to B6-Insrtm1Dac/+ mice. Two loci controlling plasma insulin levels were found
on Chrs 2 and 10, and both susceptibility alleles are contributions from the resistant B6

The genetic complexity of type II diabetes comes to light when a compound
heterozygous mouse model is created by adding a defect in insulin receptor substrate 1,
Irs1, to the defect in the insulin receptor in the model described earlier. In this case, it is
the doubly mutant B6 mouse that is hyperinsulinemic and diabetic, whereas the same
mutations on a 129 strain background cause only a mild elevation in insulin and no
hyperglycemia (56). It is possible that the Irs1 defect unmasks the effects of B6
susceptibility alleles, but in addition to those described in the previous example, other
loci are at play. In F2 double-heterozygous mice from a (B6×129) F1-Insrtm1Dac/+
intercross, Almind et al. mapped one significant and one suggestive locus associated with
Genomics and proteomics in nutrition 8
hyperinsulinemia to Chrs 14 and 12, respectively, and one locus for hyperleptinemia to
Chr 7 (57). The last locus acts synergistically with that on Chr 14 to increase
hyperinsulinemia and with the Chr 12 locus to increase hyperglycemia.


From the examples given, it is clear that genetic modifiers play a role in the phenotypic
variation observed in obesity and type II diabetes. Modifiers have been shown to affect
the age of onset, severity, rate of disease progression, and presence or absence of a
particular disease phenotype.
Whereas some modifier effects have been localized to a chromosomal region, many
have yet to be mapped, and cloning of modifier genes is still a difficult task. In humans,
this is in part due to genetic heterogeneity, in terms of both the large number of obesity
genes that may exist and the high levels of variation among the genetic backgrounds upon
which obesity mutations reside. In addition, modifier effects may occur as a result of
several genes modulating a disease phenotype, and unless there is a significant
contribution from one locus, they may be difficult to isolate. It can be argued that the
approach of studying obesity modifier loci in the mouse and then determining whether
those genes play a similar role in humans may be the most efficient means of identifying
genetic modifiers. The availability of the complete human and mouse genome sequences
is aiding greatly in this quest. Advances in determining the expression patterns of all
genes in the genome help to prioritize candidates in the vicinity of mapped modifier
genes. Large-scale gene expression analysis using microarrays may identify genes that
are coregulated by the modifiers and possibly define novel pathways. Consequently, the
rate at which modifiers are identified will increase in the near future.
Although few modifier genes have been identified to date, these have yielded
additional information about the pathways in which the primary mutation acts, and have
provided new experimental avenues toward understanding the pathological effects of the
primary disease genes. Finally, the elucidation of modifier genes associated with
attenuation of obesity and prevention of progression to type II diabetes may lead to
exploration of new therapeutics aimed at increasing the activity of modifiers in affected

authors: Yun Wang, Patsy M.Nishina, and Jürgen K.Naggert
The Jackson Laboratory, Bar Harbor, Maine, U.S.A.

Tidak ada komentar:

Posting Komentar