EPIGENETICS AND EPISTASIS
Epistasis Among Genes Is a Universal
Phenomenon in Obesity: Evidence From
Rodent Models
Craig H. Warden, PhD, Nengjun Yi, PhD, and Janis Fisler, PhD
Rowe Program in Genetics, the Department of Pediatrics and Section of Neurobiology,
Physiology and Behavior, and the Department of Nutrition, University of California, Davis,
California, USA; and the Department of Biostatistics, Section on Statistical Genetics,
University of Alabama, Birmingham, Alabama, USA
INTRODUCTION
Identification of genes that cause spontaneous mendelian obesity
in mice began a new age of obesity studies with the ability to
discover mechanisms underlying the biological causes of obesity.
Studies in humans quickly confirmed that orthologs of mouse
obesity genes also cause monogenic (mendelian) forms of human
obesity. However, the six genes that cause mendelian forms of
human obesity (leptin, leptin receptor, proopiomelanocortin, pro-
hormone convertase, melanocortin receptor-4, and peroxisome
proliferator-activated receptor-
␥
) each have insignificant effects on
common human obesity.
1
Common obesity is inherited as a poly-
genic (quantitative) trait influenced by many genetic and environ-
mental variables. Thus, alleles of genes causing common forms of
human obesity are unlikely to produce mendelian pedigrees show-
ing segregation of mutations with obesity. Genes for common
obesity will produce small increases in obesity risk ratios or will
influence obesity by more complex routes, such as gene ϫ gene
epistasis and gene ϫ environment interactions.
The identification of genes underlying complex traits such as
obesity, diabetes, and hypertension has been less successful than
for simple mendelian traits. One of the reasons for this is the
presence of epistasis that can mask the effects of one locus on
another locus and reduce the power to identify either locus. Ep-
istasis is generally defined as the interaction between two or more
genes or their mRNA or protein products to influence a single trait.
Non-additive genetic effects and interactions between genes con-
tribute considerably to the phenotypic variance of body weight and
obesity, and an understanding of this network of interactions is
essential for the understanding of the genetic and environmental
contributions to obesity.
2
Human family and adoption studies have
estimated that genetic influences explain 25% to 75% of variation
in body mass index. Studies of twins, however, place that estimate
higher, at 50% to 90% of variance. Segal and Allison
3
argued that
the twin estimates are likely correct and reflect the non-additive
nature, due to a combination of dominance and epistasis, of the
genetic effects on obesity because common environmental fac-
tors may influence estimation of the genetic effects. Indeed, in
a recent review, Moore
4
hypothesized that “epistasis is a ubiq-
uitous component of the genetic architecture of common human
diseases and that complex interactions are more important than
the independent main effects of any one susceptibility gene.” It
is essential to study multiple genes simultaneously to examine
these interactive effects.
Although only a small fraction of papers dealing with the
underlying biology of obesity mention epistasis, we present evi-
dence that gene ϫ gene interactions are common. We then discuss
methods that can be used to find epistatic genes by quantitative
trait locus (QTL) mapping of naturally occurring alleles in exper-
imental animal models.
EPISTASIS IS COMMON AND HAS POWERFUL
INFLUENCES ON OBESITY
It is widely recognized that mouse genetic background alters the
phenotypes of spontaneous mutants and knockouts. When mice
carrying a mutation or knockout are backcrossed for approxi-
mately 10 generations to a different background mouse strain, the
results are strains that are congenic for the mutation or knockout.
That is, the congenic and background strains of mice are com-
pletely identical, except for the mutation or knockout and a small
surrounding chromosomal region derived from whatever donor
strain originally carried the mutation or knockout.
5
Strain depen-
dence of the mutation phenotype is common, demonstrating that
other genes modify the phenotype resulting from mutation of
knockouts with functionally different alleles in the different
strains. The interaction may be the formal epistasis of gene ϫ gene
interactions (or protein ϫ protein), or it may result because the
mutation or knockout is part of a pathway or network that influ-
ences a trait. We do not list all studies showing background gene
effects on obesity knockouts and spontaneous mutations because
this list would be too lengthy for this review, but we do provide a
few pertinent examples to illustrate the phenomenon.
That the diabetes syndromes of the obese (Lep
ob
) and diabetes
(Lepr
db
) mutants are determined by the genetic background on
which the mutation is expressed has been known for more than
30 y. Both Lep
ob
and Lepr
db
mutations have been bred onto the
C57BL/6J and C57BL/Ks backgrounds (and also on others), with
the result that different strains have the same mutations in leptin or
leptin receptor, but most other genes in the two strains have
functionally different alleles. Both of these mutations on the
C57BL/6J background result in compensatory hyperplasia of islet
cells and hyperinsulinemia, whereas these mutations on the
C57BLKs background result in severe depletion of insulin produc-
ing

-cells and much earlier death.
6,7
Impairment in the thermo-
genic response to cold is also a reflection of the varying back-
ground genome on which these mutant genes are expressed.
8
Genetic background interacts with the Lepr
db
mutation to modu-
This work was supported by DK52581 (C.H.W.).
Correspondence to: Craig H. Warden, PhD, Rowe Program in Genetics,
University of California, Davis, CA 95616, USA. E-mail: chwarden@
ucdavis.edu
0899-9007/04/$30.00Nutrition 20:74–77, 2004
©Elsevier Inc., 2004. Printed in the United States. All rights reserved. doi:10.1016/j.nut.2003.09.013