Published in Volume
119, Issue 1
(January 5, 2009)J Clin Invest.
Copyright © 2009, American Society for Clinical
Rare loss-of-function mutations in ANGPTL family members
contribute to plasma triglyceride levels in humans
1Donald W. Reynolds Cardiovascular Clinical Research Center and Eugene
McDermott Center for Human Growth and Development, University of Texas Southwestern
Medical Center, Dallas, Texas, USA.
2Howard Hughes Medical Center,
University of Texas Southwestern Medical Center, Dallas, Texas, USA.
3Department of Statistical Sciences, Southern Methodist University,
Dallas, Texas, USA.
4Genomics Division, Lawrence Berkeley National
Laboratory, Berkeley, California, USA.
5US Department of Energy Joint
Genome Institute, Walnut Creek, California, USA.
6Human Genetics Center
and Institute for Molecular Medicine, University of Texas Health Science Center,
Houston, Texas, USA.
7Center for Human Nutrition, University of Texas
Southwestern Medical Center, Dallas, Texas, USA.
Address correspondence to: Helen H. Hobbs or Jonathan C. Cohen, Eugene
McDermott Center for Human Growth and Development, University of Texas Southwestern
Medical Center, 5323 Harry Hines Blvd., Dallas, Texas 75390, USA. Phone: (214)
648-6724; Fax: (214) 648-7539; E-mail: firstname.lastname@example.org
(H.H. Hobbs); email@example.com (J.C. Cohen).
First published December 15, 2008
Received for publication August 12,
2008, and accepted in revised form November 5,
The relative activity of lipoprotein lipase (LPL) in different tissues controls the
partitioning of lipoprotein-derived fatty acids between sites of fat storage (adipose
tissue) and oxidation (heart and skeletal muscle). Here we used a reverse genetic
strategy to test the hypothesis that 4 angiopoietin-like proteins (ANGPTL3, -4, -5,
and -6) play key roles in triglyceride (TG) metabolism in humans. We re-sequenced the
coding regions of the genes encoding these proteins and identified multiple rare
nonsynonymous (NS) sequence variations that were associated with low plasma TG levels
but not with other metabolic phenotypes. Functional studies revealed that all mutant
alleles of ANGPTL3 and ANGPTL4 that were associated
with low plasma TG levels interfered either with the synthesis or secretion of the
protein or with the ability of the ANGPTL protein to inhibit LPL. A total of 1% of
the Dallas Heart Study population and 4% of those participants with a plasma TG in
the lowest quartile had a rare loss-of-function mutation in ANGPTL3,
ANGPTL4, or ANGPTL5. Thus, ANGPTL3, ANGPTL4, and
ANGPTL5, but not ANGPTL6, play nonredundant roles in TG metabolism, and multiple
alleles at these loci cumulatively contribute to variability in plasma TG levels in
In individuals consuming Western diets, more than 100 grams of triglycerides (TGs) are
transported each day from the liver and small intestines to peripheral tissues. The
partitioning of TG between sites of storage (adipose tissue) and oxidation (primarily
heart and skeletal muscle) is determined by the relative activity of lipoprotein lipase
(LPL), an enzyme located on the lumenal surfaces of capillaries. LPL catalyzes the
hydrolysis of the lipoprotein TG, releasing FFAs, which are taken up by adjacent
LPL activity is regulated at the transcriptional and posttranslational levels. The
hormonal and nutritional milieu of tissues modulates transcription of the
LPL gene (1). LPL is regulated at
the posttranscriptional level by 2 angiopoietin-like proteins (ANGPTLs), ANGPTL3 and
ANGPTL4 (2), which belong to a family of 7
structurally similar secreted proteins (ANGPTL1–ANGPTL7). The ANGPTL
proteins contain a signal sequence followed by a helical domain predicted to form a
coiled coil and a globular, fibrinogen-like domain at the C terminus (3). Both ANGPTL3 and ANGPTL4 inhibit LPL activity in
vitro and in vivo (4–6), and mice lacking Angptl3 or
Angptl4 have increased LPL activity and reduced levels of plasma TG
Although ANGPTL3 and ANGPTL4 both inhibit LPL activity, the 2 proteins have different
patterns of expression. ANGPTL3 is expressed almost exclusively in liver (9), an organ that expresses little or no LPL in
adults (1), and is presumed to function as a
circulating inhibitor of LPL. In contrast to ANGPTL3, ANGPTL4 is expressed in multiple
tissues, with the highest level of expression in mice being in adipose tissue (10). Originally, this member of the ANGPTL family
was referred to as “fasting-induced adipocyte factor,” since its
expression is highly induced by fasting (10). It
has been proposed that ANGPTL4 inhibits LPL activity in adipose tissue to reroute fatty
acids away from fat to muscle and other tissues during fasting (2).
A third member of the ANGPTL family, ANGPTL6 (also referred to as
“angiopoietin-related growth factor”) has also been implicated
in energy metabolism and lipid partitioning. Genetic deletion of
Angptl6 in mice is associated with significant (>80%) embryonic
lethality (11). Surviving
Angptl6–/– mice are markedly
obese and hyperinsulinemic and accumulate significant amounts of TG in liver and
skeletal muscle. Circulating TG levels are not altered in these mice, but the levels of
cholesterol and FFA in serum are increased. An ANGPTL6 transgene under
the control of a β-actin promoter was expressed at high levels in multiple
tissues (brown fat, heart, and skeletal muscle) and resulted in reduced white adipose
tissue mass and resistance to diet-induced obesity (11).
The physiological roles of these ANGPTL proteins in humans have been inferred largely
from studies in mice. Recently, we used a population-based resequencing strategy to
examine the metabolic role of ANGPTL4 in humans (12). By resequencing the coding region and proximal intronic regions of
ANGPTL4 in a multiethnic sample of 3,551 individuals, we showed that
sequence variations that alter an amino acid (nonsynonymous [NS] sequence variations) in
ANGPTL4 were more prevalent in individuals with TG levels in the lowest quartile than in
the highest quartile (P = 0.016). One variant (E40K), which was present
in approximately 3% of Americans of mixed European descent, was associated with
significantly lower plasma levels of TG and LDL cholesterol (LDL-C) and higher levels of
HDL cholesterol (HDL-C) in the Atherosclerosis Risk in Communities (ARIC) study and the
Copenhagen City Heart Study (12). These findings
confirmed that ANGPTL4 is involved in TG metabolism in humans and revealed that the
protein also plays roles in the metabolism of HDL and LDL, which was not apparent from
studies in genetically modified mice.
To glean insights into the physiological roles of the other ANGPTL family members in
humans, we resequenced ANGPTL3, ANGPTL5, and ANGPTL6
in a large multiethnic population. We complemented these studies with cell-based and in
vitro assay studies to examine the effects of the mutations identified on protein
synthesis, secretion, and function. We found that 1% of participants in the Dallas Heart
Study (DHS) and 4% of those with plasma TG in the lowest quartile had a rare
loss-of-function mutation in ANGPTL3, -4, or
-5. Functional studies showed that the mutant alleles of
ANGPTL3 and ANGPTL4 had major detrimental effects
on the synthesis, secretion, or function of the protein.
ANGPTL mRNA levels in human tissues. As a first test of the hypothesis that multiple members of the ANGPTL family regulate
TG metabolism in different tissues, we examined the levels of mRNA from each gene in
48 human tissues (Figure 1). Expression of
ANGPTL3 was largely restricted to liver, consistent with the pattern seen previously
in mice (9). The levels of
ANGPTL4 mRNA were also highest in liver, with the next highest level
being in the pericardium. The level of the ANGPTL4 transcript in adipose tissue was
only 10% that found in liver, despite fat having the highest expression level of
ANGPTL4 in mice (10). Low levels of
ANGPTL4 transcript (<10% of liver) were also present in
the adrenal glands, lung, pancreas, and placenta, with only trace amounts detected in
Expression of ANGPTL3, ANGPTL4,
ANGPTL5, and ANGPTL6 in human tissues. Quantitative real-time PCR was used to determine mRNA levels of the 4
ANGPTL family members in commercial cDNA arrays of 48 tissues
prepared from normal humans (Origene). The 21 tissues in which the signal was
detected are shown. Each bar represents the average of triplicate measurements
expressed as a fraction of the Ct value obtained from the tissue expressing the
highest levels of mRNA for that gene.
ANGPTL5 was most highly expressed in adipose tissue, with the bronchus,
epididymis, and vena cava having the next highest levels of expression. The transcript was identified at very low levels in many other tissues. A previous
study reported that ANGPTL5 was expressed most strongly in heart
(13), but those authors did not examine
mRNA from adipose tissue. Although we detected ANGPTL5 mRNA in the heart, it was
present at only 20% of the level found in adipose tissue.
Finally, ANGPTL6 was expressed at the highest level in the liver and
at much lower levels in other tissues.
Thus, 3 of the 4 ANGPTLs analyzed in this study (ANGPTL3,
ANGPTL4, and ANGPTL6) were most highly expressed
in the liver, and 2 of the 4 family members were expressed in adipose tissue
(ANGPTL4 and ANGPTL5). This finding is
consistent with the hypothesis that these genes coordinate fuel trafficking and
homeostasis in response to changes in energy demands.
Multiple sequence variants in the coding regions of ANGPTL3, ANGPTL5, and
ANGPTL6. To develop a comprehensive inventory of sequence variations in the coding regions of
ANGPTL3, ANGPTL5, and ANGPTL6,
we sequenced the exons of the 3 genes in the DHS, a multiethnic, probability-based
population (including 1,870 African Americans, 1,045 individuals of mixed European
descent, and 601 Hispanics) as previously described (12). A total of 255 variants were identified, most of which were rare:
more than half (155/255) were found in only a single individual, and 86% (220/255)
had a minor allele frequency (MAF) below 1% (Supplemental Table 1; supplemental
material available online with this article; doi:
10.1172/JCI37118DS1). NS variants were more common than synonymous
variants in all 3 genes (97 vs. 27 variants). The density of coding sequence variants
was similar in the 3 genes: 1 sequence substitution per 33 nucleotides in
ANGPTL3 and ANGPTL5 and 1 substitution per 30
nucleotides in ANGPTL6.
To determine the phenotypic effects of sequence variations in the 3 genes, we
stratified the DHS population by race, sex, and trait level for metabolic phenotypes
associated with TG removal from the circulation (plasma TG level), TG accumulation
(BMI and hepatic TG content), and indices of energy homeostasis (fasting blood
glucose, fasting insulin, homeostatic model assessment — insulin
resistance [HOMA-IR]). In addition, we tested for association with systolic blood
pressure, diastolic blood pressure, and plasma levels of cholesterol, TG, HDL-C, and
LDL-C. We used a strategy similar to that employed to assess the effect of rare
sequence variations in other genes on quantitative traits (14, 15). For each
phenotype we compared the number of individuals with NS sequence variants in the top
and bottom quartiles (Table 1; ANGPTL4 was
included for comparison). Any sequence variation present in both the top and the
bottom quartiles was eliminated from the analysis.
Numbers of individuals with NS sequence variations identified in
ANGPTL3, ANGPTL4, ANGPTL5, and
ANGPTL6 in the upper or lower quartile of the distribution of
selected traits in the DHS
Rare and common sequence variations in ANGPTL3 are associated with reduced plasma
TG levels. A total of 35 NS sequence variations were identified in ANGPTL3
(Supplemental Table 2). An excess of sequence variants in the lower quartile for
plasma TG levels (14 vs. 5 variants) approached the nominal significance threshold
(P = 0.064; Figure 2A). All
sequences likely to be loss-of-function alleles (frameshift 122 [Fs122], FsQ192, and
FsK455) were in the lowest quartile of TG levels, suggesting that decreased levels of
circulating ANGPTL3 are associated with reduced plasma levels of TG. One of the
missense mutations associated with a low plasma TG level, K63T, is predicted to
disrupt a heparin binding motif
(61VHKTKG66) that is required for LPL
inhibition in mice (4).
Schematic representation of ANGPTL3 with positions of NS sequence variations
identified in the upper and lower quartiles of TG distribution in the DHS. (A) The deduced 460–amino acid ANGPTL3 protein has the
characteristic features of angiopoietins: a signal peptide (SP), an extended
helical domain predicted to form coiled coils (CC), and a globular fibrinogen
homology domain (FBG-like domain) at the C terminus. An excess of NS sequence
variations was found in lower quartile of TG distribution compared with the upper
quartile (14 vs. 5 variations; P = 0.064). Coiled coil domains
were predicted using COILS (http://www.ch.embnet.org/software/COILS_form.html),
and the fibrinogen-like domain sequence was obtained from NCBI
Δ, deletion of an amino acid. (B) A common allele of
ANGPTL3 (M259T) present in 10% of African Americans was
associated with lower plasma levels of TG in 2 the DHS and the ARIC study. The
variant was not associated with HDL-C or BMI. M/M, M/T, and T/T refer to predicted
amino acids at position 259 of ANGPTL3.
To further examine the relationship between sequence variation in
ANGPTL3 and plasma TG levels, we tested for association between
plasma TG and the common SNPs at the locus (MAF >1%). None of these SNPs was
associated with plasma TG levels (data not shown), except for M259T. M259T was
present at appreciable frequency among African Americans (MAF = 5%) and was
significantly associated with plasma TG levels (P = 0.006) in this
ethnic group (Supplemental Table 3 and Figure 2B). The allele was rare among individuals of mixed European descent (MAF,
0.1%). To validate this association, we assayed the M259T SNP in African Americans in
the ARIC study (16) (Supplemental Table 3).
This analysis confirmed that the M259T SNP was significantly associated with plasma
levels of TG (P = 0.014). Taken together, these data indicate that a
spectrum of sequence variants in ANGPTL3 contributes to variation in
plasma TG levels.
The number of NS sequence variants in ANGPTL3 was similar in the
upper and lower quartiles of the distribution for BMI (P = 0.235)
and hepatic TG content (P = 0.773). Similarly, the common allele
associated with plasma TG levels (M259T) was not associated with either parameter. A
significant excess of NS sequence variants was found among individuals in the lowest
quartile for blood glucose levels (16 vs. 3 variants; P = 0.002) in
the DHS, but in both the DHS and in ARIC, the M259T variant was not associated with
blood glucose, insulin, or with any of the other metabolic parameters examined
(Supplemental Table 3). Taken together, these findings indicate that sequence
variation in ANGPTL3 was associated with plasma levels of TG, but
not with other indices of fat accumulation or metabolism, although we cannot exclude
the possibility that some rare variants in ANGPTL3 affect plasma glucose levels.
Excess of rare sequence variations in ANGPTL5 in individuals with low plasma
levels of TG. A similar strategy was used to analyze the metabolic effects of sequence variations
in ANGPTL5 (Figure 3A), a gene
of unknown function that is not expressed in mice (13). A significant excess of NS variants was found in the lowest quartile of
TG levels (n = 9) compared with the highest quartile
(n = 1) (P = 0.022). No frameshift or nonsense
mutations in this gene were identified in the lowest quartile, but a single mutation
in a consensus splice donor site (IVS8+1) was found. Only 1 NS variant (T268M) in
ANGPTL5 had a MAF of greater than 1%, and this variant was not
consistently associated with plasma TG levels in the DHS or in ARIC (data not shown).
Sequence variations in ANGPTL5 were not associated with any of the
other metabolic phenotypes examined.
Schematic representation of ANGPTL5 and ANGPTL6 with positions of NS variants
identified in the upper and lower quartiles of the TG distribution in the DHS. (A) Individuals carrying NS sequence variations in
ANGPTL5 were more prevalent in the bottom quartile than in the
top quartile (9 vs. 1 individuals, respectively; P = 0.012).
(B) A corresponding analysis of ANGPTL6 revealed
no significant difference in allele frequencies (6 vs. 13 individuals;
P = 0.107). (C) Number of subjects with plasma TG
levels in the upper quartile (n = 878) and the lower quartile
(n = 897) of the DHS who had nonsense, missense, and splicing
mutations in ANGPTL3, ANGPTL4,
ANGPTL5, and ANGPTL6. The difference in numbers
of subjects in the upper and lower quartiles is due to differences in the number
of individuals with plasma TG levels at the thresholds for the quartiles. IVS,
Sequence variations in ANGPTL6 are not associated with plasma TG levels. In contrast to ANGPTL3, ANGPTL4, and
ANGPTL5, in which NS variants were significantly associated with
plasma TG levels, we found no evidence of association between sequence variants in
ANGPTL6 and plasma TG levels. The number of individuals with NS
variants in ANGPTL6 did not differ significantly in the upper and
lower quartiles of the TG distribution (Figure 3B). Two NS variants in ANGPTL6 had a MAF of greater than 1%
(R96P and R358C). Neither of these variants was associated with TG levels in the DHS
(data not shown). A statistically significant excess of NS variants in
ANGPTL6 was found in upper quartile of plasma cholesterol levels
when compared with the lowest quartile of cholesterol levels, but not in the upper
quartile of LDL-C or HDL-C levels (Table 1).
Figure 3C summarizes the relationship between
plasma TG levels and sequence variations in 3 of the ANGPTL family members analyzed
in this paper (ANGPTL3, ANGPTL5, and ANGPTL6), together with data from our prior
analysis of ANGPTL4. The numbers of individuals with NS variants in the top and
bottom quartiles for plasma TG are given for each gene. As was observed for ANGPTL3
and ANGPTL5, sequence variations in ANGPTL4 were more common among individuals with
plasma levels of TG in the lowest quartile compared with the highest quartile.
The majority of NS variants in ANGPTL3, ANGPTL4, and ANGPTL5 associated with low
plasma TG levels interfere with protein secretion. Several of the sequence variants in ANGPTL3, ANGPTL4, and ANGPTL5 that were found in
the lowest quartile of plasma TG levels were nonsense, frameshift, or splice-site
mutations (Figures 2 and 3). This finding suggested that loss-of-function alleles of these
genes are associated with low plasma TG levels. The remainder of the mutations found
in the lowest quartile of TG levels were missense mutations. To determine whether
these mutations also interfere with protein function, we generated cDNA expression
constructs for each mutant allele and compared the expression and secretion of the
mutant proteins with wild-type ANGPTL in cultured human embryonic kidney (HEK293A)
cells. Immunoblot analysis of the cell lysates and the medium are shown in Figure
4A. Whereas wild-type ANGPTL3 was readily
detected in the medium, 5 of the 9 missense mutations present in the lowest TG
quartile, but none of the 5 missense variants in the high-TG group abolished
secretion of ANGPTL3 from cells.
Effects of sequence variations on the synthesis and secretion of ANGPTL3,
ANGPTL4, and ANGPTL5. (A) ANGPTL3, (B) ANGPTL4, and (C) ANGPTL5.
Wild-type and mutant forms of ANGPTL3, ANGPTL4, and ANGPTL5 were expressed in
HEK293A cells, and immunoblotting was performed on the cell lysates and medium
using an anti-V5 mAb as described in Methods. Calnexin was used as a loading
control. This experiment was repeated 3 times with similar results.
↓TG, sequence variations found in individuals with TG in the
≤ 25th percentile; ↑TG, sequence
variations found in individuals with TG in the ≥ 75th percentile;
—, vector only.
Similarly, when the same experiment was performed to examine the NS sequence variants
that we previously identified in ANGPTL4 (12), 5 of the 7 missense alleles had either a partial or complete
deficit in ANGPTL4 secretion (Figure 4B).
Finally, in contrast to wild-type ANGPTL5 or the allele containing the missense
mutation found in the high TG group (I233V) 3 of the 7 missense mutations in
ANGPTL5 associated with a low plasma TG level failed to be
secreted from the cells (Figure 4C).
Thus, of the 23 missense mutations in the 3 ANGPTL family members that were found in
individuals in the lower quartile of plasma TGs, more than half (n =
13) impaired protein secretion, presumably by interfering with the proper folding of
the protein. These mutations were located almost exclusively in the highly structured
fibrinogen-like domains located at the C terminus of the proteins (Figure 2).
Effects of ANGPTL3 and ANGPTL4 mutations on LPL-mediated hydrolysis of TG in
vitro. Previously, ANGPTL3 and ANGPTL4 were shown to inhibit LPL activity in vitro (6, 17).
ANGPTL4 appears to disrupt catalytically active LPL dimers (17). To determine whether the ANGPTL variants
identified in the lowest quartile of plasma TG levels that were secreted normally had
a reduced ability to suppress LPL activity, we tested the effect of the mutant
proteins on LPL activity in vitro. Addition of conditioned medium from cells
expressing ANGPTL3 reproducibly suppressed LPL activity by more than 50% (Figure
5A). In contrast, conditioned media from
cells expressing ANGPTL3-259T (Figure 5A) or the rare ANGPTL3 alleles associated with low
plasma levels of TG (Figure 5B) failed to
suppress LPL activity.
Effects of sequence variations in ANGPTL3 and ANGPTL4 on LPL activity. Wild-type or mutant forms of ANGPTL3 (A and B), ANGPTL4
(C), and ANGPTL6 (D) were expressed in HEK293A cells,
and conditioned medium was collected and concentrated. The effect of increasing
concentrations (range of 20-fold) of protein was evaluated for wild-type ANGPTL3
(A) and ANGPTL6 (D) as described in Methods.
Conditioned medium from cells expressing wild-type ANGPTL4 consistently suppressed
LPL activity by more than 90% (Figure 5C). These
data are consistent with prior data showing that ANGPTL4 has more potent inhibitory
effect on LPL activity than does ANGPTL3 (5,
18). Conversely, when conditioned medium
containing equivalent quantities of mutant ANGPTL4 proteins that were found in the
low TG group was added to the lipase assay, no suppression of LPL activity was
observed. Furthermore, the mutant ANGPTL4 proteins failed to suppress LPL even when
added at concentrations 10-fold higher than those at which the wild-type protein
completely inhibited the enzyme (data not shown). The inhibitory effects of both
ANGPTL3 and ANGPTL4 were specific for the salt-sensitive component of post-heparin
plasma lipase activity (data not shown). Both proteins had negligible effects on the
salt-resistant lipase (hepatic lipase) activity.
ANGPTL5 was expressed at much lower levels than was ANGPTL4 and ANGPTL3, and we were
unable to obtain comparable levels of this protein in the media. Therefore, we were
unable to examine the effects of this protein on LPL activity. ANGPTL6 was
efficiently secreted from cells but did not inhibit LPL activity at any of the
concentrations tested (Figure 5D).
The mutations in ANGPTL3 and ANGPTL4 that interfered with their ability to inhibit
LPL activity were in the N-terminal region of the protein, which is consistent with
the observation of Sukonina et al. (17) that
the N-terminal portion of ANGPTL4 interacts with the enzyme.
Comparison of in vitro and in silico analysis of ANGPTL mutations. A total of 31 NS variants were identified in the low-TG group and 8 in the high-TG
group (Table 2). Of the 31 variants in the
low-TG group, 8 (26%) introduced a premature termination codon or altered a consensus
splice site, 13 interfered with secretion of the protein from cells (Figure 4), and 6 resulted in proteins that were secreted
but failed to inhibit LPL activity. Thus, for ANGPTL3 and ANGPTL4, all variants found
only in the low-TG group severely compromised the function of the protein. Three of
the 7 mutations in ANGPTL5 prevented expression or secretion; we were not able to
assess the effects of the remaining 4 mutations on LPL activity.
Observed and predicted functional effects of the NS variants in ANGPTLs in the top
and bottom quartiles of the DHS
Multiple in silico programs have been developed to predict the functional effects of
sequence variations, although these perform with variable success (19). Our analysis allowed us to assess the
utility of 2 computer programs, PolyPhen (20)
and SIFT (21), which predict the functional
consequences of amino acid substitutions. PolyPhen correctly predicted 10 of the 19
mutations (53%), whereas SIFT predicted 12 of the variations (63%). Of the 7 variants
that showed no functional defect, 2 were predicted to be deleterious by PolyPhen and
3 were predicted to be deleterious by SIFT. Thus, the false positive rates were 29%
and 43% for PolyPhen and SIFT, respectively.
Cumulative analysis of rare NS sequence variations in ANGPTL3, ANGPTL4, ANGPTL5,
and plasma TG levels. To assess the cumulative effects of rare variants in the 3 genes in the DHS, we
pooled the data for ANGPTL3, ANGPTL4, and
ANGPTL5 and found a greater than 4-fold excess of individuals
with NS sequence variations in the low-TG group (n = 36) compared
with the high-TG group (n = 8) (Figure 6). The total number of synonymous sequence variations (n
= 8) was identical in the high-TG and low-TG groups. Although each variant was rare
(MAF, <1%), taken together, 1 of every 24 participants (4%) in DHS with TG
levels below the 25th percentile had a NS variant in 1 of these 3 genes.
Cumulative frequency of NS and synonymous sequence variations in
ANGPTL3, ANGPTL4, and
ANGPTL5 in the upper and lower quartiles of plasma TG
distribution in the DHS. The number of individuals in each quartile was n = 878 (upper
quartile) and n = 897 (lower quartile).
In this study we used resequencing in a well-phenotyped population of unselected
individuals to define the physiological roles of ANGPTL proteins in humans. The major
finding was that multiple loss-of-function alleles in ANGPTL3 and
ANGPTL5, together with those previously reported in
ANGPTL4 (12), are associated
with lower plasma levels of TG, suggesting a common role for these proteins in the
metabolism of TG-rich lipoproteins. One percent of participants in the DHS (and 4% of
those with a TG in the lowest quartile) had a rare sequence variant in 1 of these 3
genes that was associated with low plasma levels of TG. Most of the sequence variants
identified in this study prevented (or markedly reduced) secretion of the mutant
proteins from cells. Resequencing also revealed an allele of ANGPTL3
that was common in African Americans and was significantly associated with plasma TG
levels in 2 large populations. The sequence variants associated with low plasma TG
levels were not associated with the amount of fat in the body or in the liver or with
indices of glucose metabolism. Taken together, these findings indicate that
ANGPTL3, ANGPTL4, and ANGPTL5
regulate the uptake of TG-derived fatty acids from the circulation but do not have a
major impact on the overall accumulation of TG in the body, on the partitioning of TG
between the liver and peripheral tissues, or on the efficiency of glucose utilization.
Genetic manipulation of mice is a powerful and widely used strategy to assess the
physiological roles of genes. The finding that a reduction in plasma TG levels is the
major consequence of loss-of-function mutations in ANGPTL3 and
ANGPTL4 both in humans and in mice affirms the value of mouse models
for metabolic studies. However, other phenotypes observed in
Angptl-knockout mice were not recapitulated in humans with
loss-of-function alleles in these genes. In mice, genetic deletion of
Angptl3 is associated with an approximately 50% reduction in plasma
levels of HDL-C, perhaps due to increased activity of endothelial lipase, an enzyme that
hydrolyzes HDL phospholipids (22, 23). Since ANGPTL3 inhibits endothelial lipase in
vitro, it has been proposed that ANGPTL3 normally suppresses endothelial
lipase–mediated catabolism of HDL, resulting in higher plasma levels of HDL
(22, 24). In support of this model, inactivation of Angptl3 in
vivo, either genetically or enzymatically, is associated with a
substantial reduction in HDL phospholipid and HDL-C levels in mice (22, 24). In
the present study, loss-of-function mutations in ANGPTL3 were not
associated with a decrease in plasma levels of HDL-C. Thus, our data are not consistent
with a significant role for ANGPTL3 in determining plasma levels of HDL-C in humans.
Angptl6–/– mice are obese and have
both hepatic steatosis and hyperinsulinemia (11).
We found no evidence that genetic variability in ANGPTL6 contributes to
BMI, hepatic TG content or fasting glucose or insulin levels in humans. The apparent
differences in the phenotypic effects of Angptl6 mutations in mice and
humans may be due to loss-of-function alleles at this locus being recessive in humans.
We did not identify any individuals in the DHS with mutations in both
ANGPTL6 alleles. Alternatively, only a single line of
Angptl6 knockout mice has been described (11), and some of the phenotypic effects reported in that mouse line
may be due to factors unrelated to the inactivation of Angptl6. These
findings underscore the essential role for human studies in validating observations from
The association between sequence variations in ANGPTL5 and plasma TG
levels demonstrates that the reverse genetic strategy employed in this study can provide
insights into the physiological roles of genes even without a priori information from
physiological studies or observations in model systems. ANGPTL5 was
initially identified during a large-scale sequencing study of a human fetal brain
library (25). No ortholog of
ANGPTL5 has been identified in mice, and functional analysis of the
protein is limited to a single report in which recombinant ANGPTL5 was shown to support
ex vivo expansion of human cord blood hematopoietic stem cells (26). The observation that loss-of-function mutations in
ANGPTL5 are associated with low plasma levels of TG suggests that
ANGPTL5 also plays a role in the metabolism of TG-rich lipoproteins. We believe that
this finding provides the first insight into the physiological function of ANGPTL5 in
Recently, 2 large genome-wide association studies reported that plasma levels of TG were
associated with multiple sequence variants spanning a 250-kb interval on chromosome 7
that contains ANGPTL3, DOK7, and USP1
(27, 28). The specific variants responsible for the associations were not identified.
The ANGPTL3 alleles identified in the present study are unlikely to
explain these associations, since all the sequence variations found in individuals of
mixed European descent, the population represented in the genome-wide association
studies, were rare. The only common ANGPTL3 allele
(ANGPTL3-259T) associated with plasma TG levels was largely
restricted to African Americans and had a frequency of 0.1% in individuals of mixed
European descent. Taken together, these findings indicate that a spectrum of rare and
common sequence variants in ANGPTL3 contributes to variation in plasma
TG levels in the general population. No sequence variants in ANGPTL4 or
ANGPTL5 reached genome-wide significance in either study.
The observation that sequence variations in ANGPTL3,
ANGPTL4, and ANGPTL5 contribute independently to
plasma TG levels indicates that the 3 proteins are not functionally redundant. Since
ANGPTL3 and ANGPTL4 are both expressed primarily in
the liver in humans, and loss-of-function mutations in the 2 genes are associated with
very similar lipoprotein phenotypes, it is possible that the 2 proteins function as a
heterodimer. Both proteins contain coiled coil domains as well as a fibrinogen domain at
the C terminus, and prior studies have shown that ANGPTL4 forms higher-order oligomers
(29) and that oligomerization is required for
activity (30). Coexpression of the 2 proteins in
cultured cells (including cultured hepatoctyes) failed to reveal any evidence that they
form a stable complex (data not shown), but additional studies in different cell types
will be required to rule out this possibility.
An alternative possibility is that ANGPTL3, ANGPTL4,
and ANGPTL5 perform similar roles but in different tissues or different
metabolic states. This hypothesis is consistent with our finding that
ANGPTL3, ANGPTL4, and ANGPTL5 have
different patterns of expression (Figure 1) and
with previous reports that ANGPTL4 is strongly upregulated by fasting
and is suppressed by refeeding (10), whereas
ANGPTL3 mRNA levels do not change significantly in response to
changes in food intake (18). Further studies will
be required to determine the specific effects of ANGPTL3, ANGPTL4, and ANGPTL5 on the
uptake of TG-derived fatty acids in different tissues and under different nutritional
Interestingly, the mutations we identified in these 4 members of the ANGPTL family were
not evenly distributed among the 3 major ethnic groups in the DHS. The majority of rare
mutations in ANGPTL4 (8 of 13), as well as a common sequence variation
in this gene (E40K) that was associated with lower plasma TG levels (E40K) were found in
individuals of mixed European descent (12).
Conversely, 10 of the 13 alleles in ANGPTL3 identified in the low-TG
group were present in African Americans, and a common allele associated with low plasma
TG levels (M259T) was also largely restricted to this population. In contrast to these 2
genes, in which the mutations clustered in a single ethnic group, the NS sequence
variations identified in ANGPTL5 were found in all 3 ethnic groups in
approximately equal proportion (Figure 3).
Hispanics were over-represented among those individuals in the highest quartile of TG
who had mutations in ANGPTL6 (Figure 3). These ethnic differences in allele frequencies may reflect historic
differences in selective pressures among the 3 ethnic groups.
The finding that approximately 1% of the individuals in this study had a sequence
variant that profoundly altered protein function is consistent with our findings in
other genes (14, 31, 32) and with in silico predictions
from Sunyaev and colleagues that about 1% of the alleles of a typical
500–amino acid protein will contain a deleterious mutation (33). Thus severe loss-of-function mutations, though
individually rare, are collectively common in the population. This observation provides
a strategy for reverse genetics in humans. Historically, the major advances in human
genetics have been made almost exclusively by using forward genetic approaches in which
individuals are selected based on a phenotypic classification and then screened for
causal sequence variations. Reverse genetic approaches have been highly informative in
model organisms, where genetic manipulations can be performed to assess the phenotypic
consequences of inactivation, overexpression, or alteration of the sequence of a gene,
but these approaches have played little role in human genetics. Since it is not feasible
to manipulate genes in humans, human geneticists rely on naturally occurring variations
that are almost invariably ascertained indirectly as a consequence of their associated
phenotypes. Advances in sequencing technology are poised to shift the focus of
resequencing from selected candidate genes to the entire exome, and possibly to the
genome. By resequencing large cohorts of well-characterized individuals, the sequence
variants identified can be tested against multiple traits to determine the roles of
genes to human physiology.
Study populations. The coding regions of the ANGPTL3, ANGPTL5, and
ANGPTL6 gene were sequenced in all DHS participants
(n = 3,551) who underwent phlebotomy (34). The DHS is a population-based probability sample of Dallas
County (52% African American, self-identified as “black;” 29%
individuals of mixed European descent, self-identified as
“white;” 17% self-identified as
“Hispanic;” and 2% of other ethnicities), in which ethnicity
was self-assigned according to US census categories. Only the individuals of mixed
European descent, African Americans, and Hispanics were included in our analysis. The
DHS was approved by the Institutional Review Board of the University of Texas
Southwestern Medical Center, and all subjects provided written informed consent
before participating in the study.
A subset of the genetic associations observed in the DHS were validated in the ARIC
study (16). The ARIC study is a prospective
study of atherosclerosis in 4 communities in the USA (Jackson, Mississippi;
Minneapolis, Minnesota; Forsyth County, North Carolina; and Washington County,
Maryland). A randomly selected cohort of approximately 4,000 individuals ages
45–64 years was selected from each community. The protocol for the study
was approved by the institutional review boards of the Johns Hopkins University, the
University of North Carolina, the University of Minnesota, and the University of
Mississippi, and all participants provided written informed consent that included
consent for genetic studies.
Assay of plasma lipids and lipoproteins. Plasma and lipoprotein cholesterol and TG concentrations were determined
colorimetrically by using commercial enzymatic reagents.
DNA sequencing and genotyping. The exons and flanking introns of ANGPTL3, ANGPTL5,
and ANGPTL6 were sequenced in both directions in the 3,551 DHS
participants, as previously described (12).
All sequence variants identified were verified by manual inspection of the
chromatograms, and missense changes were confirmed by an independent resequencing
Fluorogenic 5′-nucleotidase assays to detect the M259T polymorphism in
ANGPTL3 were developed using the TaqMan assay system (Applied
Biosystems). The assays were performed on a 7900HT Fast Real-Time PCR instrument with
probes and reagents purchased from Applied Biosystems.
Expression of ANGPTL in cultured cells. Expression constructs for human ANGPTL3, ANGPTL4, ANGPTL5, and ANGPTL6 were made in
pcDNA3.1 (Supplemental Table 4) under the control of the cytomegalovirus
promoter-enhancer (pCMV-ANGPTL3-V5, pCMV-ANGPTL4-V5, pCMV-ANGPTL5-V5). A V5 epitope
tag (GKPIPNPLLGLDST) was placed at the C-terminus of each construct. Single base pair
changes were introduced into these constructs using QuikChange (Stratagene). The
presence of the desired mutation and the fidelity of each construct were confirmed by
HEK293A cells were seeded (1 × 105 cells/well) in 6-well
plates and grown in DMEM with 10% fetal calf serum (Cellgro). Expression plasmids (4
μg/well) were used to transfect HEK293A cells using Lipofectamine 2000
(2.5 μl/μg of plasmid DNA) in DMEM according to the
manufacturer’s protocol (Invitrogen). After 48 h, the medium was
collected and centrifuged for 5 min (5,000 g at 4°C) and
the supernatants were collected. The cells were washed twice in PBS and then
incubated in 0.3 ml of 1× RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl,
1% [vol/vol] NP-40, 0.5% [vol/vol] sodium deoxycholate, 0.1 % SDS, and complete mini
EDTA-free protease inhibitor cocktail [Roche]) for 15 min at 4°C. Cells
were harvested by scraping with a rubber policeman and transferred to 1.5-ml tubes
prior to centrifugation for 15 min (15,000 g at 4°C).
Aliquots from the medium and cells were subjected to SDS-PAGE and immunoblot
SDS-PAGE and immunoblot analysis. Protein concentrations in the cell lysates were determined using the Bio-Rad
bicinchoninic acid assay, according to the manufacturer’s protocol.
Equivalent amounts of protein from the cell lysate and medium were added to the
sample loading buffer (final concentration of 1×). The samples were
heated to 95°C for 5 min, size-fractionated on 12% SDS-polyacrylamide
gels at 150 V, and then transferred to nitrocellulose membranes at 100 V for 1 h.
Membranes were incubated in PBST buffer (1× PBS plus 0.1% Tween-20) with
5% dry milk for 60 min at room temperature before addition of the primary antibodies.
Monoclonal anti-V5 antibody (Invitrogen) and a polyclonal antibody to calnexin
(Stressgen) were diluted to 1:5,000 in PBST buffer with 5% dry milk and incubated
with the membranes for 60 min. Membranes were washed 3 times for 10 min each time in
PBST buffer. Horseradish peroxidase–conjugated donkey anti-rabbit IgG or
goat anti-mouse IgG (Pierce Biotechnology) was diluted (1:10,000) in PBST buffer with
5% dry milk and incubated with membranes for 60 min. Membranes were washed 3 times
for 10 min each time in PBST and were visualized using SuperSignal-enhanced
chemiluminescence (Pierce Biotechnology).
In vitro assay of LPL. LPL activity assay was measured using a modification of the procedure described by
Nilsson-Ehle and Schotz (35). Media harvested
from HEK293A cells transfected with ANGPTL expression constructs were concentrated
8-fold and buffer-exchanged into 1× PBS (pH 7.4). The concentrated media
were mixed with 7.5 μl mouse post-heparin plasma, brought up to a volume
of 50 μl with 0.2 M Tris HCl buffer (pH 8.0), and incubated at
20°C or 25°C. Radiolabeled substrate (16.7 μl)
composed of 9,10-3H(N) triolein (American Radiolabeled Chemicals),
triolein and phosphatidylcholine (Sigma-Aldrich), 16.7 μl
heat-inactivated fetal calf serum, and bovine serum albumin (3% in 66.6
μl of 0.2 M Tris HCl buffer [pH 8.0]) were added and the mixtures were
incubated for 15 min at 37°C in the presence or absence of 1 M NaCl,
which inhibits LPL activity (36). The
reactions were terminated by adding heptane/chloroform/methanol (1:1.25:1.41), mixed,
and centrifuged at 3,000 g for 15 min. A 1-ml aliquot of upper (aqueous) phase was
taken into scintillation tubes and counted. The amount of [3H]-fatty acid
released was calculated as previously described (35).
Real-time PCR. The expression of ANGPTL3, -4, -5, and -6 in human tissues was examined using cDNA
prepared from 48 tissues (Human Normal cDNA Panel). The cDNA was standardized to 2 ng
using GAPDH as a calibrator. Oligonucleotides specific to each gene were used to
amplify by PCR in 2× SYBR Green PCR Master Mix (Applied Biosystems) in a
total volume of 66 μl according to the manufacturer’s
instructions. For each gene, the tissue expressing the highest activity was used as a
reference and assigned an arbitrary expression level of 1. The mean of the 3 CT
measurements in each tissue was expressed as a fraction of the level observed in the
Statistics. The prevalence of NS variants in the upper and lower quartiles of the DHS was
compared using Fisher’s exact test. Individuals with diabetes, as defined
previously (12), and subjects on
lipid-lowering medications were not included in the analysis. Men who consumed more
than 30 g (and women who consumed more than 20 g) of alcohol per day were also
excluded from the analyses. Risk factor levels between carriers of each
ANGPTL variant with a MAF of greater than 1% and noncarriers were
compared by ANOVA. For the comparisons of plasma lipid levels between the genotype
groups, we included age, sex, and diabetes mellitus as covariates in the model.
Plasma levels of TG and insulin were log-transformed before analysis.
View Supplemental data
We thank Zifen Wang, Liangcai Nie, Wendy Schackwitz, Anna Ustaszewska, Joel Martin, and
Crystal Wright for excellent technical assistance, Kim Lawson for statistical analysis,
and Susan Lakoski for helpful discussions. This work was supported by a grant from the
Donald W. Reynolds Foundation, a grant from the NIH (RL1- HL-092550 and HL-20948), and
the American Heart Association Texas Affiliate Fellowship (to S. Romeo).
Authorship note: Stefano Romeo and Wu Yin contributed equally to this
Conflict of interest: The authors have declared that no conflict of
Nonstandard abbreviations used: ANGPTL, angiopoietin-like protein; ARIC,
Atherosclerosis Risk in Communities (study); DHS, Dallas Heart Study; Fs, frameshift;
HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; LPL, lipoprotein lipase; MAF, minor
allele frequency; NS, nonsynonymous; TG, triglyceride.
Citation for this article:J. Clin. Invest.119:70–79 (2009). doi:10.1172/JCI37118
Merkel, M., Eckel, R.H., Goldberg, I.J. 2002. Lipoprotein lipase: genetics, lipid uptake, and regulation. J. Lipid Res. 43:1997-2006.
Li, C. 2006. Genetics and regulation of angiopoietin-like proteins 3 and 4. Curr. Opin. Lipidol. 17:152-156.
Yoon, J.C., et al. 2000. Peroxisome proliferator-activated receptor gamma target gene encoding
a novel angiopoietin-related protein associated with adipose differentiation. Mol. Cell. Biol. 20:5343-5349.
Ono, M., et al. 2003. Protein region important for regulation of lipid metabolism in
angiopoietin-like 3 (ANGPTL3): ANGPTL3 is cleaved and activated in vivo. J. Biol. Chem. 278:41804-41809.
Yoshida, K., Shimizugawa, T., Ono, M., Furukawa, H. 2002. Angiopoietin-like protein 4 is a potent hyperlipidemia-inducing factor
in mice and inhibitor of lipoprotein lipase. J. Lipid Res. 43:1770-1772.
Shimizugawa, T., et al. 2002. ANGPTL3 decreases very low density lipoprotein triglyceride clearance
by inhibition of lipoprotein lipase. J. Biol. Chem. 277:33742-33748.
Koishi, R., et al. 2002. Angptl3 regulates lipid metabolism in mice. Nat. Genet. 30:151-157.
Koster, A., et al. 2005. Transgenic angiopoietin-like (angptl)4 overexpression and targeted
disruption of angptl4 and angptl3: regulation of triglyceride metabolism. Endocrinology. 146:4943-4950.
Conklin, D., et al. 1999. Identification of a mammalian angiopoietin-related protein expressed
specifically in liver. Genomics. 62:477-482.
Kersten, S., et al. 2000. Characterization of the fasting-induced adipose factor FIAF, a novel
peroxisome proliferator-activated receptor target gene. J. Biol. Chem. 275:28488-28493.
Oike, Y., et al. 2005. Angiopoietin-related growth factor antagonizes obesity and insulin
resistance. Nat. Med. 11:400-408.
Romeo, S., et al. 2007. Population-based resequencing of ANGPTL4 uncovers variations that
reduce triglycerides and increase HDL. Nat. Genet. 39:513-516.
Zeng, L., et al. 2003. Identification of a novel human angiopoietin-like gene expressed
mainly in heart. J. Hum. Genet. 48:159-162.
Cohen, J.C., et al. 2004. Multiple rare alleles contribute to low plasma levels of HDL
cholesterol. Science. 305:869-872.
Cohen, J., et al. 2006. Multiple rare variants in NPC1L1 associated with reduced sterol
absorption and plasma low-density lipoprotein levels. Proc. Natl. Acad. Sci. U. S. A. 103:1810-1815.
Atherosclerosis Risk in Communities Study. 1987. ARIC manual of operations:
No 2, cohort component procedures. ARIC Coordinating Center, School of Public
Health, University of North Carolina. Chapel Hill, North Carolina, USA.
Sukonina, V., Lookene, A., Olivecrona, T., Olivecrona, G. 2006. Angiopoietin-like protein 4 converts lipoprotein lipase to inactive
monomers and modulates lipase activity in adipose tissue. Proc. Natl. Acad. Sci. U. S. A. 103:17450-17455.
Ge, H., et al. 2005. Differential regulation and properties of angiopoietin-like proteins 3
and 4. J. Lipid Res. 46:1484-1490.
Kotowski, I.K., et al. 2006. A spectrum of PCSK9 alleles contributes to plasma levels of
low-density lipoprotein cholesterol. Am. J. Hum. Genet. 78:410-422.
Sunyaev, S., et al. 2001. Prediction of deleterious human alleles. Hum. Mol. Genet. 10:591-597.
Ng, P.C., Henikoff, S. 2003. SIFT: Predicting amino acid changes that affect protein function. Nucleic Acids Res. 31:3812-3814.
Shimamura, M., et al. 2007. Angiopoietin-like protein3 regulates plasma HDL cholesterol through
suppression of endothelial lipase. Arterioscler. Thromb. Vasc. Biol. 27:366-372.
McCoy, M.G., et al. 2002. Characterization of the lipolytic activity of endothelial lipase. J. Lipid Res. 43:921-929.
Jin, W., et al. 2007. Hepatic proprotein convertases modulate HDL metabolism. Cell Metab. 6:129-136.
Xu, J., et al. 2001. Cloning, expression and characterization of a novel human REPS1 gene. Biochim. Biophys. Acta. 1522:118-121.
Zhang, C.C., Kaba, M., Iizuka, S., Huynh, H., Lodish, H.F. 2008. Angiopoietin-like 5 and IGFBP2 stimulate ex vivo expansion of human
cord blood hematopoietic stem cells as assayed by NOD/SCID transplantation. Blood. 111:3415-3423.
Willer, C.J., et al. 2008. Newly identified loci that influence lipid concentrations and risk of
coronary artery disease. Nat. Genet. 40:161-169.
Kathiresan, S., et al. 2008. Six new loci associated with blood low-density lipoprotein
cholesterol, high-density lipoprotein cholesterol or triglycerides in humans. Nat. Genet. 40:189-197.
Ge, H., et al. 2004. Oligomerization and regulated proteolytic processing of
angiopoietin-like protein 4. J. Biol. Chem. 279:2038-2045.
Ge, H., Yang, G., Yu, X., Pourbahrami, T., Li, C. 2004. Oligomerization state-dependent hyperlipidemic effect of
angiopoietin-like protein 4. J. Lipid Res. 45:2071-2079.
Cohen, J., et al. 2005. Low LDL cholesterol in individuals of African descent resulting from
frequent nonsense mutations in PCSK9. Nat. Genet. 37:161-165.
Fahmi, S., Yang, C., Esmail, S., Hobbs, H.H., Cohen, J.C. 2008. Functional characterization of genetic variants in NPC1L1 supports the
sequencing extremes strategy to identify complex trait genes. Hum. Mol. Genet. 17:2101-2107.
Kryukov, G.V., Pennacchio, L.A., Sunyaev, S.R. 2007. Most rare missense alleles are deleterious in humans: implications for
complex disease and association studies. Am. J. Hum. Genet. 80:727-739.
Victor, R.G., et al. 2004. The Dallas Heart Study: a population-based probability sample for the
multidisciplinary study of ethnic differences in cardiovascular health. Am. J. Cardiol. 93:1473-1480.
Nilsson-Ehle, P., Schotz, M.C. 1976. A stable, radioactive substrate emulsion for assay of lipoprotein
lipase. J. Lipid Res. 17:536-541.
Berbee, J.F., van der Hoogt, C.C., Sundararaman, D., Havekes, L.M., Rensen, P.C. 2005. Severe hypertriglyceridemia in human APOC1 transgenic mice is caused
by apoC-I-induced inhibition of LPL. J. Lipid Res. 46:297-306.
Szczepaniak, L.S., et al. 2004. Magnetic resonance spectroscopy to measure hepatic triglyceride
content: prevalence of hepatic steatosis in the general population. Am. J. Physiol. Endocrinol. Metab. 288:E462-E468.