The serotonin receptor gene HTR5A affects plasma triglyceride
levels in humans and mice
Y. Zhang1, E. Smith1, L. Martin2, Shama Mirza1, J. Charlesworth3, E. Moses3, J. Blangero3, J. Lazar1,
Z. Lazarova1, A.H. Kissebah1, and M. Olivier1
1Medical
College of Wisconsin, Milwaukee, WI 2Childrens Hospital, Cincinnati, OH
³Southwest Foundation for Biomedical Research, San Antonio, TX
Abstract
Functional study of the causal variant (EMSA)
Serotonin (5-HT) has been implicated in the development of obesity, but has not been shown to regulate lipid
metabolism or plasma lipid levels. One of the genes for a 5-HT receptor, HTR5A, is located in a quantitative trait
locus on human chromosome 7q36 (LOD = 3.7) linked to plasma triglyceride levels in a family-based obese cohort of
2,207 individuals of Northern European descent. Initial fine-mapping using 1000 SNPs across the 5 Mb interval
uncovered association between sequence variants in the gene and plasma triglyceride (TG) levels. Resequencing of all
exons and 3 kb of the promoter region uncovered 13 SNPs within the 20 kb gene interval. Association analysis
demonstrated that four of these SNPs were associated with TG levels (p<0.002). Subsequent Bayesian quantitative
trait nucleotide (BQTN) analysis identified a single SNP (rs3734967) with significant posterior probability (p=0.59),
suggesting that this variant may be causal. This SNP resides in the putative promoter region about 2 kb upstream of
the start codon of HTR5A. This variant accounts for ~5% of the observed linkage in the QTL region in our cohort.
Individuals homozygous for the G allele have on average 3.7 % higher TG levels when compared to individuals
homozygous for the A allele.
Since HTR5A is exclusively expressed in the brain, we tested the effect of the two alleles of rs3734967 on transcription
factor binding using two human glioma cell lines U87 and D54 in electrophoretic mobility shift assays. Both cell lines
demonstrated significantly altered binding of nuclear proteins to the A allele when compared to the G allele of the
variant. Mass spectral analysis revealed differential binding of multiple DNA binding proteins including
transcription activator and signal transducers (STAT1 and STAT3) and ATP-dependent DNA helicases.
Finally, plasma TG levels of htr5r knockout mice (B6.129-Htr5atm1Dgen) are significantly lower (p=5×10-4) when
compared to heterozygous littermates at 6 weeks of age, suggesting that HTR5A affects plasma TG levels in both
humans and mice.
These results, for the first time, implicate serotonin in obesity-related lipid homeostasis, and may allow the
development of alternative pharmacological lipid-lowering therapies to prevent the cardiovascular complications of
obesity.
QTL Analysis
A genome-wide linkage screening of 503 Caucasian families (2,207 individuals) revealed a QTL on chr7q36. Dense
mapping of this 5 Mb region followed by association analysis showed HTR5A gene accounted for 26% of the
observed linkage. Bayesian quantitative trait nucleotide (BQTN) analysis suggests promoter SNP rs3734967 of
HTR5A may be a causal variant for its associated dyslipidemia (posterior probability=0.59).
XRCC2
GALNT11
INSIG1
DPP6
FLJ42291
ACTR3B
PAXIP1
MLL3
FLJ16734
HTR5A
LOC155100
EN2
A
G
Identifying differentially bound proteins by Mass
Spectrometry (MS)
Differentially bound proteins
identified by MS. Mass spectral
analysis of proteins bound
differentially in EMSA revealed
three rs3734967 G allele-specific
transcription related proteins:
STAT3, RECQ1 (ATP-dependent
DNA helicase Q1) and PARP1
(poly[ADP-ribose] polymerase I.
A allele
G allele
78
42
37
0
0
3
Proteins with
known role in
transcription
UBE3C
SHH
PRR8
AK124544
AK124321
RBM33
Electrophoretic mobility shift assay of
HTR5A promoter SNP in human brain
cells. Oligos of DNA sequences from
HTR5A promoter region centered with
either allele of rs3734967 (A or G) were
incubated with total nuclear proteins from
cultured U87 Glioma cells. Boxed areas
were excised from the gel and prepared
for protein identification by mass
spectrometry (see below).
LMBR1
RNF32
NOM1
Analysis of HTR5A Knockout Mice
HLXB9
Fasting plasma levels of triglycerides in mice
with different genotypes of HTR5A. 8-week old
mice with heterozygous (+/-) or homozygous (-/-)
genotype of HTR5A were measured for plasma
triglyceride levels and compared to reported
levels for parental wild type mice. Htr5a
knockout mice have lower (p=10×10-5) TG levels
than mice with one copy of the gene.
4
LOD
3
2
1
0
0
50
100
cM
150
Association Analysis of HTR5A
10%
5%
26%
2%
HTR5A
*
** ** *** ** **
Results and Conclusions
1. Dense mapping and association analysis identifies HTR5A as a gene in the
QTL region affecting plasma TG levels.
2. Bayesian QTN analysis reveals three statistically causal sequence variants.
3. Functional analysis of rs3734967 using EMSA shows differential binding
patterns of alternative alleles with nuclear factors from human brain Glioma
cells, and three putative DNA binding proteins responsible for this
difference.
(above): Haploview of the proximal region of
HTR5A on chr7 using genotype data from
Hapmap consortium. Colored bars on the top
illustrates the statistically calculated linkage
contribution to observed QTL in TOPS cohort.
(right): All investigated SNPs in HTR5A region.
Baysian quantitative trait nucleotide (BQTN)
analysis shows three variants might be causal.
4. Htr5a knockout mice showed significantly lower levels of plasma levels of
triglycerides than heterozygous mice.
Acknowledgments
The authors would like to thank the TOPS members for their participation in the study, and
Regina Cole and Milena Zelembaba for technical support. This work was supported by NIH
grant HL74168 (M.O.).