Environment International 91 (2016) 188–195

Contents lists available at ScienceDirect

Environment International

journal homepage: www.elsevier.com/locate/envint

Gene-gene and gene-environment interactions on risk of male infertility:

Focus on the metabolites
Weiyue Hu a,b,1, Minjian Chen a,b,1, Wei Wu a,b,c,⁎, Jing Lu d, Dan Zhao e, Feng Pan e, Chuncheng Lu a,b,
Yankai Xia a,b, Lingqing Hu c, Daozhen Chen c, Jiahao Sha f,⁎, Xinru Wang a,b,⁎
a
State Key Laboratory of Reproductive Medicine, Institute of Toxicology, Nanjing Medical University, Nanjing 211166, China
b
Key Laboratory of Modern Toxicology of Ministry of Education, School of Public Health, Nanjing Medical University, Nanjing 211166, China
c
State Key Laboratory of Reproductive Medicine, Wuxi Maternal and Child Health Care Hospital Affiliated to Nanjing Medical University, Wuxi 214002, China
d
State Key Laboratory of Reproductive Medicine, Department of Reproduction, Nanjing Maternal and Child Health Care Hospital Affiliated to Nanjing Medical University, Nanjing 210004, China
e
State Key Laboratory of Reproductive Medicine, Department of Andrology, Nanjing Maternal and Child Health Care Hospital Affiliated to Nanjing Medical University, Nanjing 210004, China
f
State Key Laboratory of Reproductive Medicine, Nanjing Medical University, Nanjing 211166, China

a r t i c l e i n f o a b s t r a c t

Article history: Infertility affects about 17% couples, and males contribute to half of the cases. Compared with independent effects
Received 25 December 2015 of genetic and environmental factors, interactions between them help in the understanding of the susceptibility
Received in revised form 22 February 2016 to male infertility. Thus, we genotyped 25 polymorphisms, measured 16 urinary chemical concentrations and ex-
Accepted 23 February 2016
plored interactions between gene-gene and gene-environment in 1039 Han Chinese using metabolomic analysis.
Available online xxxx
We first observed that GSTT1 might interact with GSTM1 (Pinter = 6.33 × 10−8). Furthermore, an interaction be-
Keywords:
tween GSTM1 and 4-n-octylphenol (4-n-OP) was identified (Pinter = 7.00 × 10−3), as well as a 2-order interaction
Male infertility among GSTT1, GSTM1 and 4-n-OP (Pinter = 0.04). Subjects with GSTT1-present and GSTM1-null genotypes were
Gene-gene interaction susceptible to male infertility when exposed to 4-n-OP (OR = 14.05, 95% CI = 4.78–60.20, P = 2.34 × 10−5).
Gene-environment interaction Most metabolites identified were involved in the tricarboxylic acid cycle. In conclusion, it is a novel study of
Metabolomics the interaction on male infertility from the aspect of metabolomics.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction topic of genetic causes, microdeletions and copy number variations in

Infertility is a reproductive disease defined by the failure to conceive
after 12-month regular unprotected sexual intercourse (Zegers-
Hochschild et al., 2009). In China, a latest prospective follow-up study
claimed that the incidence of 12-month infertility was 13.6% among
all eligible couples (Meng et al., 2015). About half of the cases link to a
male factor among infertile couples (Tian et al., 2014). Except those
who can be treated by assisted reproductive technology, there are still
men incapable of fertilization. Thus, it is worthy of identifying factors
potentially correlated with male infertility to prevent and control such
a problem.
Spermatogenesis is a complex process in the area of male reproduc-
tion, and it can be influenced by many factors such as infection, sex hor-
mone imbalance, tobacco smoking, and alcohol use (Diao et al., 2014; La
Maestra et al., 2014; Wang et al., 2009). In the past few years, around the
⁎ Corresponding authors at: State Key Laboratory of Reproductive Medicine, Institute of
Toxicology, Nanjing Medical University, 101 Longmian Road, Nanjing 211166, China.
E-mail addresses: wwu@njmu.edu.cn (W. Wu), shajh@njmu.edu.cn (J. Sha),
xrwang@njmu.edu.cn (X. Wang).
1
These authors contributed equally to the study and they should be regarded as joint
first authors.
the Y chromosome have been proved to be connected with spermato-
genesis failure (Foresta et al., 2001; Lo Giacco et al., 2014). Furthermore,
both traditional targeted approach and high-throughput genome-wide
association study (GWAS) strategy have successfully identified some
susceptibility loci linking with non-obstructive azoospermia (Hu et al.,
2014; Hu et al., 2012). However, such variants merely accounts for a
part of conceivable reasons, and there have been limited studies, for
the purpose of validation, successfully created animal models of male
infertility using single nucleotide polymorphisms (SNPs) identified by
GWA studies.
To decipher the unknown part of male infertility, researchers have
also observed a contribution of the environment. In the scope of daily
life and occupational health, heavy metal is a toxicant reported to asso-
ciate with abnormal semen quality (Meeker et al., 2008; Zeng et al.,
2015). And our previous studies have also assessed the risks of some en-
vironmental endocrine disrupting chemicals (EDCs), like polycyclic aro-
matic hydrocarbons, phenols and phytoestrogens that are related to the
male infertility (Chen et al., 2014; Xia et al., 2013; Xu et al., 2014). Nev-
ertheless, the role of environmental factors in the development of com-
plex diseases is restricted as the genetics. A combination, or the so called
interaction, is considered to be a prospective direction (Koehl, 2015;
Manuck and McCaffery, 2014).

http://dx.doi.org/10.1016/j.envint.2016.02.025
0160-4120/© 2016 Elsevier Ltd. All rights reserved.
 W. Hu et al. / Environment International 91 (2016) 188–195
Gene-gene and gene-environment interactions are usually applica-
ble to estimate the synergistic or antagonistic effects between both fac-
tors. And interactions have been explored to play essential roles in
cancers, heart diseases, diabetes, etc. (Fedeles et al., 2011; Langenberg
et al., 2014). Hauser et al. (2005) also identified a joint effect related
to sperm motility in a study of 303 people. Given the limited and specific
subfertile study population, the power of study like that is not convinc-
ing. For the purpose of supplying the lack of investigations on male in-
fertility, we recruited 420 fertile men as well as 619 infertile cases in
Han Chinese, examined gene-gene-environment interactions between
two groups, and tried to deepen the comprehension of such interactions
with a metabolomic analysis in the present study.
2. Methods
2.1. Study population and sample collection
Male infertility was characterized with at least two-year infertility
history in our research. 1713 infertile men and 710 fertile men were in-
vited to participate in the study consecutively from affiliated hospitals
of Nanjing Medical University between 2005 and 2010. Of those, 1182
men (case = 700, control = 482) consented to participate and signed
an informed consent. A questionnaire was utilized to collect character-
istic information. All subjects claimed that their life styles and exposures
had not been changed for several months before sample collection. After
completion of the questionnaire, the examiner measured the height and
weight of each participant. Males with abnormal sexual and ejaculatory
functions, immune infertility, medical history of risk factors for infertil-
ity and other known causes related to male infertility, such as genetic
disease, were excluded from the study. Although subjects were request-
ed to donate a urine sample for the measurement of urinary EDCs con-
centrations, a 5 mL peripheral blood for genotyping and an ejaculate for
semen analysis, some of them left no blood or semen. Hence, we includ-
ed 1039 subjects (case = 619, control = 420) in this study at last.
Among these 619 infertile cases, we collected 391 semen samples
which were obtained in private by masturbation, and routine semen
analyses were carried out according to World Health Organization
guidelines. Standard measurements we used were sperm concentration
(normal ≥ 15 × 106 spermatozoa/mL), progressive motility (nor-
mal ≥ 32%) and sperm morphology (normal ≥ 14%). Besides semen sam-
ples, we also kept urine samples at − 20 °C while blood samples at
−80 °C until measurement. The study protocol and informed consent
were approved by the Institutional Review Board of Nanjing Medical
University prior to the study. All activities involved in this study were
done under full compliance with government policies and the Helsinki
Declaration.
2.2. Measurement of urinary EDC concentrations
EDCs included in the study were bisphenol A (BPA), pentachlorophenol
(PCP), benzophenone-3 (BP-3), triclosan (TCS), 4-tert-octylphenol
(4-t-OP), 4-n-octylphenol (4-n-OP), 4-n-nonylphenol (4-n-NP),
2,3,4-trichlorophenol (2,3,4-TCP), 2,4,5-trichlorophenol (2,4,5-
TCP), secoisolariciresinol (SEC), enterodiol (END), enterolactone
(ENL), naringenin (NAR), genistein (GEN), daidzein (DAI) and equol
(EQU). We measured urinary phenol and phytoestrogen concentrations
using ultra high performance liquid chromatography–tandem mass
spectrometry (UPLC–MS/MS) and gas chromatography–mass spec-
trometry (GC–MS) with sensitive methods as previously described
(Chen et al., 2013; Xia et al., 2013). Limits of determination (LOD)
were 0.36 ng/mL (BPA), 0.41 ng/mL (PCP), 0.04 ng/mL (BP-3),
0.9 ng/mL (TCS), 0.34 ng/mL (4-t-OP), 0.02 ng/mL (4-n-OP),
0.02 ng/mL (4-n-NP), 0.28 ng/mL (2,3,4-TCP), 0.15 ng/mL (2,4,5-TCP),
0.09 ng/mL (SEC), 0.04 ng/mL (END), 0.04 ng/mL (ENL), 0.02 ng/mL
(NAR), 0.09 ng/mL (GEN), 0.03 ng/mL (DAI) and 0.12 ng/mL (EQU). Uri-
nary creatinine (CR) concentration was analyzed with an automated
189
chemistry analyzer (7020 Hitachi, Tokyo, Japan) for correcting the
variation caused by fluctuated urine concentration and dilution.
2.3. Genotype identification
We identified variants in phase I and phase II enzyme genes in rela-
tion to the metabolism of phenols and phytoestrogens. All those select-
ed single nucleotide polymorphisms (SNPs), located in exons or UTRs,
have been reported minor allele frequencies (MAF) of N0.05 in the
Han Chinese population. In the case of multiple SNPs which were in
linkage disequilibrium, only one was selected. Finally, we selected 23
potential functional polymorphisms in metabolic enzymes: CYP1A1
rs1048943 and rs4646422; CYP1B1 rs2855658 and rs9341266; CYP2B6
rs3760657, rs2054675, rs707265, and rs1042389; CYP2C8 rs1058932;
CYP2C9 rs4918758; CYP2C19 rs3814637, rs4986894, and rs3758581;
CYP2E1 rs2031920; CYP2S1 rs3810171 and rs338583; NAT1 rs7845127
and rs10888150; NAT2 rs1799930, rs1799931 and rs4646246;
SULT1E1 rs4149525 and rs3736599. In addition, we also included 2
null polymorphisms in GSTT1 and GSTM1 here. SNPs were genotyped
using TaqMan SNP Genotyping Assays (Biosteed, Nanjing, China) or
GenomeLab SNPstream high throughput 12-plex genotyping platform
(Beckman Coulter, Fullerton, CA)(Qin et al., 2014), and null polymor-
phisms were identified using PCR reactions(Wu et al., 2013). For quality
control, 10% of the samples were randomly genotyped again, and the
repeatability was 100%.
2.4. Metabolomic analysis
We randomly selected 31 urine samples to perform metabolomic
analysis. The mixture of 200 μL urine and 400 μL methanol was collected
in a 1.5 mL Eppendorf tube. After centrifugation at 16,000 g for 15 min,
supernatant was obtained and then dried in a centrifugal concentrator.
The residue was reconstituted in 400 μL ddH2O. Metabolomic analysis
was based on an accurate and sensitive method previously reported
by Blasco et al. (2013). Briefly, LC-HRMS analysis was performed on
an UPLC Ultimate 3000 system (Dionex), coupled with a Q-Exactive
mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The
system was controlled by Xcalibur 2.3 (Thermo Fisher Scientific). Dur-
ing the full-scan acquisition which ranged from 70 to 1050 m/z, the in-
strument operated at a 70,000 resolution. A multistep gradient had
mobile phase A of 0.1% formic acid in ultra-pure water and mobile
phase B consisting of acetonitrile acidified with 0.1% formic acid; the
gradient operated at a flow rate of 0.4 mL/min over a run time of
15 min. All samples were analyzed randomly in case of the bias from
the injection order. And identification was based on the accurate mass
and the retention time of metabolites compared with those of
standards.
2.5. Statistical analyses
There was a total of 1039 subjects available in this study. All statisti-
cal analyses were conducted using R 3.1.2 (Foundation for Statistical
Computing, Vienna, Austria). A P-value of ≤0.05 was considered as the
threshold for statistical significance. In statistical hypothesis testing, a
type I error is the incorrect rejection of a true null hypothesis, and it is
more likely to occur when considering a set of inferences simultaneous-
ly. To prevent this error, we used a statistical technique called the
Bonferroni correction to get a higher significance threshold (P-
value ≤ 1.67 × 10−4) for individual comparisons, which compensated
for the number of inferences being made. In descriptive analyses, we ex-
plored the mean age, BMI and constituent ratios of smoking status, alco-
hol consumption and tea consumption in all subjects. Values of urinary
EDC concentrations below LODs were imputed as LOD/2. Additionally,
the creatinine correction was used to balance variations of urinary dilu-
tion in samples, so exposure value in this study was presented as con-
centrations per gram of creatinine. To examine potential interactions
190 W. Hu et al. / Environment International 91 (2016) 188–195

on male infertility, we performed a multivariable logistic regression be- Table 2

tween genotypes and chemicals adjusted by age and BMI, and com- Distribution of urinary EDCs levels (μg/g creatinine) in the study population (n = 1039).

pared changes in nested models with or without an interaction term EDC Geometric mean 10th 25th 50th 75th 90th
using likelihood-ratio test. Analyses related to metabolomics were per- BPA 0.638 0.127 0.243 0.551 1.327 3.283
formed in SIMCA-P 13.0 (Umetrics, Umea, Sweden) and MetaboAnalyst PCP 0.401 0.104 0.160 0.336 0.806 1.851
3.0 (http://www.metaboanalyst.ca/). Orthogonal partial least squares BP-3 0.080 0.010 0.018 0.058 0.271 1.050
discriminant analysis (OPLS-DA) was used to test the difference be- TCS 2.189 0.256 0.481 1.313 6.831 40.598
4-t-OP 0.241 0.076 0.114 0.207 0.421 0.962
tween groups as a whole. And the volcano plot was carried out to iden-
4-n-OP 0.022 0.005 0.009 0.019 0.045 0.120
tify those distinguished compounds. 4-n-NP 0.020 0.005 0.008 0.016 0.040 0.087
234-TCP 0.185 0.063 0.091 0.164 0.309 0.669
245-TCP 0.112 0.035 0.053 0.097 0.198 0.428
3. Results
SEC 3.402 0.301 1.554 4.317 12.048 32.534
END 9.468 0.278 4.328 14.762 47.267 126.693
3.1. Characteristics of the study population ENL 35.942 1.969 8.814 62.523 194.620 453.452
NAR 10.840 1.561 4.315 10.994 38.537 113.767
In this study, we recruited 619 infertile cases and 420 fertile men. GEN 50.202 4.384 13.516 68.723 200.211 382.035
DAI 57.463 5.028 17.109 80.472 193.204 421.603
Characteristics of all participants were given in Table 1. The mean age EQU 5.712 0.032 0.947 5.672 92.991 435.648
of the control group (29.99 ± 3.49 years) was statistically older than
All values below LOD were imputed as LOD/2, and concentrations are given as μg/g
that of the case group (28.81 ± 4.44 years), and so was the BMI (con-
creatinine.
trols: 23.73 ± 3.33 kg/m2 and cases: 23.26 ± 3.19 kg/m2). However,
there were no significant differences of smoke, alcohol or tea consump-
tion between these two groups. there were other marginally significant genotype pairs, like
rs1058932 × GSTT1-null and rs1799931 × GSTT1-null (Supplementary
Table 1), we did not take them into the following analyses.
3.2. Distributions of urinary EDC concentrations

Distributions of all these 16 EDCs in 1039 participants were present-
ed in Table 2. Among phenols, BPA, BP-3 and TCS have good detection
rates. Due to the Chinese eating habits, detection rates of
phytoestrogens were all above 90% except EQU and END (data not
shown). Geometric means of EDCs corrected by creatinine were
0.638 μg/g (BPA), 0.401 μg/g (PCP), 0.080 μg/g (BP-3), 2.189 μg/g
(TCS), 0.241 μg/g (4-t-OP), 0.022 μg/g (4-n-OP), 0.020 μg/g (4-n-NP),
0.185 μg/g (2,3,4-TCP), 0.112 μg/g (2,4,5-TCP), 3.402 μg/g (SEC),
9.468 μg/g (END), 35.942 μg/g (ENL), 10.840 μg/g (NAR), 50.202 μg/g
(GEN), 57.463 μg/g (DAI) and 5.712 μg/g (EQU) respectively.
3.3. Gene-gene interactions on male infertility
3.4. Gene-environment interactions on male infertility
To examine whether different genotypes of GSTT1 and GSTM1 could
modify effects of 16 EDCs on the male infertility, we divided participants
into low- and high-exposed groups depending on the medians of uri-
nary chemical concentrations. Then we combined both glutathione S-
transferases (GSTs) with each chemical respectively, and only the signif-
icant interaction between GSTM1-null and 4-n-OP was identified
(Fig. 2). While people harbouring the deletion polymorphism of
GSTM1 exposed to high-level 4-n-OP, the risk of male infertility was
2.42 times larger than that in the reference (OR = 2.42, 95% CI =
1.45–4.12, P = 8.85 × 10−4).

There were 25 polymorphisms belonging to 15 genes included in the

present study. Given the multiple comparisons, we used the Bonferroni
correction to adjust P values of interactions between each two polymor-
phisms with a statistical threshold P that is equal to 1.67 × 10−4. As
shown in Fig. 1, only the pair (GSTT1 × GSTM1) presented a significant
interaction on male infertility (Pinter = 6.33 × 10−8). Deletions of both
GSTT1 and GSTM1 were hazardous for male infertility (OR = 1.85, 95%
CI = 1.21–2.84 and P = 4.89 × 10−3). However, the genotype (GSTT1-
present × GSTM1-null) was found to be the most dangerous factor
(OR = 7.02, 95% CI = 3.87–13.56 and P = 8.01 × 10− 10). Although
3.5. Gene-gene-environment interactions on male infertility
Due to the above mentioned interactions, we tried to explore wheth-
er there was a 2-order interaction among GSTT1 × GSTM1 × 4-n-OP. As
shown in Fig. 3, a synergistic interaction do exist (Pinter = 0.04). Consis-
tent with the foregoing outcomes, people with GSTT1-present and
GSTM1-null genotypes were susceptible to male infertility when ex-
posed to high-level 4-n-OP (OR = 14.05, 95% CI = 4.78–60.20, P =
2.34 × 10−5).

Table 1
Characteristics of the study population (n = 1039).

Control (n = 420) Case (n = 619) Group 1 (n = 16)a Group 2 (n = 15)a

Age (years, mean ± sd) 29.99 ± 3.49 28.81 ± 4.44⁎ 30.01 ± 4.02 27.73 ± 2.67
BMI (kg/m2, mean ± sd) 23.73 ± 3.33 23.26 ± 3.19⁎ 23.84 ± 3.12 22.49 ± 3.52
Smoking status (n, %)
Never 206 (51.2%) 311 (51.3%) 9 (56.3%) 7 (46.7%)
Ever 196 (48.8%) 295 (48.7%) 7 (43.8%) 8 (53.3%)
Drinking status (n, %)
Never 519 (54.6%) 317 (52.3%) 7 (43.8%) 7 (46.7%)
Ever 182 (45.4%) 289 (47.7%) 9 (56.3%) 8 (53.3%)
Tea consumption (n, %)
Never 224 (73.9%) 254 (72.2%) 11 (78.6%) 11 (73.3%)
Ever 79 (26.1%) 98 (27.8%) 3 (21.4%) 4 (26.7%)
⁎P ≤ 0.05 when compared between control and case groups.
a
Group 1 and Group 2 respresent subjects randomly chosen to do metabolomic analysis with GSTT1-present × GSTM1-present × low-exposed 4-n-OP and GSTT1-present × GSTM1-null ×
high-exposed 4-n-OP respectively.
 W. Hu et al. / Environment International 91 (2016) 188–195 191

Fig. 1. Interaction between GSTT1 and GSTM1 on male infertility. (A) Adjusted ORs and 95% CIs of subjects with different GSTT1 and GSTM1 genotypes for male infertility. (B) Distribution of
subjects with different GSTT1 and GSTM1 genotypes. GSTs+: GSTs-present and GSTs−: GSTs-null.

3.6. Metabolomic changes in the susceptible population
Metabolomics was useful for understanding the reason why people
possessing genotypes of GSTT1-present and GSTM1-null were less adap-
tive to the adverse effect of 4-n-OP. Thus, we analyzed urine samples of
16 participants from the reference group of GSTT1-present × GSTM1-
present × low-exposed 4-n-OP (Group 1) as well as 15 men from the
susceptible group of GSTT1-present × GSTM1-null × high-exposed 4-n-
OP (Group 2). Characteristic features of these people were presented
in Table 1, and there was no significant difference between such two

Fig. 2. Interaction between GSTM1 and 4-n-OP on male infertility. (A) Adjusted ORs and 95% CIs of subjects with different GSTM1 genotypes and urinary 4-n-OP concentrations for male
infertility. (B) Distribution of subjects with different GSTM1 genotypes and 4-n-OP concentrations. GSTM1+: GSTM1-present, GSTM1−: GSTM1-null, 4-n-OP−: low-exposed 4-n-OP and 4-n-
OP+: high-exposed 4-n-OP.
192 W. Hu et al. / Environment International 91 (2016) 188–195
Fig. 3. Interaction among GSTT1, GSTM1 and 4-n-OP on male infertility. (A) Adjusted ORs and 95% CIs of subjects with different GSTs genotypes and urinary 4-n-OP concentrations for male
infertility. (B) Distribution of subjects with different GSTs genotypes and 4-n-OP concentrations. GSTs+: GSTs-present, GSTs−: GSTs-null, 4-n-OP−: low-exposed 4-n-OP and 4-n-OP+: high-
exposed 4-n-OP.
groups. It was demonstrated a homogeneous status in population which
was suitable for detecting differential metabolic profiles between
groups. The OPLS-DA model clearly distinguished two groups (Fig. 4a),
showing that there were some severe metabolic changes happened
in susceptible participants. Then we used volcano plot, a combination
of fold change analysis and t-test, and VIP scores in the first two compo-
nents to discover those significantly changed compounds (Fig. 4b–d).
Pantothenol, 1H-indole-3-acetamide, ortho-hydroxyphenylacetic acid, di-
hydroxyacetone phosphate, tryptophanol, uracil, N-formyl-L-methionine,
L-malic acid and N-acetyl-L-alanine were 9 metabolic compounds finally
discovered. And 6 of them were involved in the metabolic process of
tricarboxylic acid cycle (TCA-cycle) (Fig. 5).
4. Discussion and conclusions
Infertility affects about one in six couples, and male contributions
can be found in approximately half of the cases. Analyses of genetic
pathogeny of the male infertility exploded in the past 10 years, and
our previous studies revealed many genes involved in the process of
spermatogenesis such as mismatch repair genes, cytochrome P450 fam-
ily genes, GSTs, etc. (Hu et al., 2011; Ji et al., 2012; Wu et al., 2013). How-
ever, with the fadeout of GWAS heat, attention to the disease
susceptibility resulted from polymorphisms reduces either. Researchers
have realized that complex diseases like cancer and obesity cannot be
caused by a single genetic factor, and so is the male infertility.
In the present study, we analyzed the joint effects of gene poly-
morphisms and environmental chemicals to make up a deficiency
of single-factor investigation in the past, and results suggested a pre-
liminary evidence in relation to male infertility. Among xenobiotic
metabolism related genes we studied, only GSTT1 and GSTM1
showed a significant interaction corrected by multiple comparisons.
It may be caused by large fragment deletions, and strong polymor-
phisms like insertion and deletion indicates more drastic changes
in gene function than SNPs. GSTs comprise a family of phase II metabolic
enzymes which participate in the detoxification to defense exogenous
toxic chemicals that lead to the imbalance between reactive oxygen spe-
cies (ROS) and antioxidants, and eventually oxidative stress in the body.
It has been demonstrated that the lack of functional proteins caused by
GSTT1- and GSTM-null genotypes associates with an increased susceptibil-
ity to diseases due to oxidative stress (Bohanec Grabar et al., 2009). As for
male infertility, the role of sperm oxidative stress is widely recognized, and
treatment with oral antioxidants reverses a part of infertile conditions
(Gharagozloo and Aitken, 2011). Thus, patients with deficiency in func-
tional GSTs could be subjected to the risk of male infertility at last.
As a kind of exogenous chemicals, 4-n-OP is a degradation product of
octyphenol ethoxylates that are the most widely used surfactants in
the marketplace (Liao and Kannan, 2014). Investigations on 4-n-OP
have revealed a crucial connection to the reproductive toxicity
(Bonefeld-Jorgensen et al., 2007; Song et al., 2014). Our previous acute
toxicity test of another OP, 4-tert-OP, also indicates an adverse effect
in testicles in vivo. In fact, the actual exposure level of 4-n-OP is low
enough compared with other chemicals like BPA in China (Chen et al.,
2013). The powerful reproductive toxicity in such a low dosage, howev-
er, is more worthy of our attention. And observations of a synergistic in-
teraction between GSTs and 4-n-OP promoted our further research of
potential mechanisms involved in.
Metabolomics is a newly emerging field of “omics” research con-
cerned with the small molecule metabolites in systems biology. It can
provide an overview of the global biochemical events, distinguish healthy
and sick statuses, and even predict the risk of disease in a few years
(Tomita and Kami, 2012; Wang et al., 2011). Application of metabolomics
in male infertility widens due to the requirement of energy metabolism
during spermatogenesis and maturation of sperm, and some previous
 W. Hu et al. / Environment International 91 (2016) 188–195 193

Fig. 4. Metabolomic analysis reveals potential mechanisms of interaction on male infertility. (A) OPLS-DA score plot of subjects with GSTT1-present × GSTM1-present × low-exposed 4-n-
OP and GSTT1-present × GSTM1-null × high-exposed 4-n-OP. The OPLS-DA model shows good distinction between two groups. (B) Volcano plot with fold change ≥2 and P of t-tests ≤0.1.
The red circles represent features above the threshold. (C and D) Important features identified in components 1 and 2 by OPLS-DA. The colored boxes on the right indicate the relative
concentrations of the corresponding metabolite in each group. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

researches have indicated the usefulness of metabolomics in male infer-
tility (Shen et al., 2013; Zhang et al., 2014a; Zhang et al., 2014b).
Therefore, we firstly carried out a metabolomic analysis to clarify the
potential mechanisms of interactions in this study. Consistent with our
anticipation, most identified compounds are involved in the process of
TCA-cycle, which is known to be associated with male infertility (Dias
et al., 2014). Uracil is one of the four nucleobases in the nucleic acid of
RNA, and the methylation of uracil produces thymine, another compo-
nent in DNA. Rapid active demethylation process of sperm DNA dis-
closes a potentially regulatory role of altered uracil value in infertile
cases (Guz et al., 2014). As an alcohol analog of pantothenic acid (vita-
min B5), panthenol is quickly oxidized to pantothenate in organisms.
The physiological roles of vitamin B5 on testicular function have been
investigated in rats (Yamamoto et al., 2009). Several parameters of
sperm motility, as well as plasma concentrations of testosterone and
corticosterone, significantly changed in the vitamin B5-free group
showing its essential role in testicular endocrinology. Biosynthesis of
acetyl-CoA requires the involvement of cysteine and pantothenate,
and it occupies the core position in TCA-cycle which comprises a series
of chemical reactions and generates energy through the oxidation of ac-
etate derived from carbohydrates, fats and proteins (Kay and
Weitzman, 1987). In addition, L-malic acid, an intermediate of the
TCA-cycle along with fumarate, is important to the production of energy
in the body during both aerobic and anaerobic conditions. Such a bal-
anced system of energy metabolism is critical for the maintenance of
normal reproductive function. GSTM1, however, takes part in the gluta-
thione metabolic pathway and impacts the generation of cysteine. Peo-
ple with GSTM1-null genotype may have insufficient capacity to keep
such a balanced system in the presence of 4-n-OP, and eventually result
in the increased susceptibility to male infertility.
To summarize, it is a novel attempt to interpret the interaction be-
tween genes and the environment from the aspect of metabolomic pro-
file. Limitations of this study include: 1. the sample size here may not be
enough to detect a 2nd-order interaction (gene × gene × environment);
2. we did not find a reasonable explanation for the higher risk of GSTT1-
present than that of GSTT1-null. Thus, we will continue to further our
study of the interaction, and results need to be confirmed by more
in vitro and in vivo functional studies.
194 W. Hu et al. / Environment International 91 (2016) 188–195
Fig. 5. Metabolic pathway of TCA-cycle involves six significantly changed metabolites. Pantothenol, ortho-hydroxyphenylacetic acid, uracil, N-formyl-L-methionine, L-malic acid and N-
acetyl-L-alanine were 6 metabolites identified based on both volcano plot and VIP scores. The box plots show concentrations of corresponding metabolites in each group. Compounds
in rectangular boxes, synthesized by 6 metabolites, were key nodes in TCA-cycle.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.envint.2016.02.025.
Competing financial interests
The authors declare that they have no competing financial interests.
Acknowledgements
This work was supported by National Natural Science Foundation of
China (Nos. 81330067, 81573182, 81302457, 81401213, 81402713),
Jiangsu Natural Science Foundation (Nos. BK20130894, BK20140909),
China Postdoctoral Science Foundation (Nos. 2013M531388, 2014M560435,
2015T80569), University Natural Science Research Project in Jiangsu Province
(Nos. 13KJB330002, 14KJB330001), Practice and Innovation Training
Programs for Jiangsu Province College Students (Nos. 201310312051X,
201410312031Y), and the Priority Academic Program for the Development
of Jiangsu Higher Education Institutions (Public Health and Preventive
Medicine).
References
Blasco, H., Corcia, P., Pradat, P.F., Bocca, C., Gordon, P.H., Veyrat-Durebex, C., et al., 2013.
Metabolomics in cerebrospinal fluid of patients with amyotrophic lateral sclerosis:
an untargeted approach via high-resolution mass spectrometry. J. Proteome Res. 12,
3746–3754.
Bohanec Grabar, P., Logar, D., Tomsic, M., Rozman, B., Dolzan, V., 2009. Genetic polymor-
phisms of glutathione S-transferases and disease activity of rheumatoid arthritis. Clin.
Exp. Rheumatol. 27, 229–236.
Bonefeld-Jorgensen, E.C., Long, M., Hofmeister, M.V., Vinggaard, A.M., 2007. Endocrine-
disrupting potential of bisphenol A, bisphenol A dimethacrylate, 4-n-nonylphenol,
and 4-n-octylphenol in vitro: new data and a brief review. Environ. Health Perspect.
115 (Suppl. 1), 69–76.
Chen, M., Tang, R., Fu, G., Xu, B., Zhu, P., Qiao, S., et al., 2013. Association of exposure to
phenols and idiopathic male infertility. J. Hazard. Mater. 250-251, 115–121.
Chen, M., Zhou, K., Chen, X., Qiao, S., Hu, Y., Xu, B., et al., 2014. Metabolomic analysis re-
veals metabolic changes caused by bisphenol A in rats. Toxicol. Sci. 138, 256–267.
Diao, R., Fok, K.L., Chen, H., Yu, M.K., Duan, Y., Chung, C.M., et al., 2014. Deficient human
beta-defensin 1 underlies male infertility associated with poor sperm motility and
genital tract infection. Sci. Transl. Med. 6 (249ra108).
Dias, T.R., Alves, M.G., Silva, B.M., Oliveira, P.F., 2014. Sperm glucose transport and metab-
olism in diabetic individuals. Mol. Cell. Endocrinol. 396, 37–45.
Fedeles, S.V., Tian, X., Gallagher, A.R., Mitobe, M., Nishio, S., Lee, S.H., et al., 2011. A genetic in-
teraction network of five genes for human polycystic kidney and liver diseases defines
polycystin-1 as the central determinant of cyst formation. Nat. Genet. 43, 639–647.
Foresta, C., Moro, E., Ferlin, A., 2001. Y chromosome microdeletions and alterations of
spermatogenesis. Endocr. Rev. 22, 226–239.
Gharagozloo, P., Aitken, R.J., 2011. The role of sperm oxidative stress in male infertility
and the significance of oral antioxidant therapy. Hum. Reprod. 26, 1628–1640.
Guz, J., Gackowski, D., Foksinski, M., Rozalski, R., Olinski, R., 2014. Comparison of the ab-
solute level of epigenetic marks 5-methylcytosine, 5-hydroxymethylcytosine, and
5-hydroxymethyluracil between human leukocytes and sperm. Biol. Reprod. 91, 55.
Hauser, R., Williams, P., Altshul, L., Calafat, A.M., 2005. Evidence of interaction between
polychlorinated biphenyls and phthalates in relation to human sperm motility. Envi-
ron. Health Perspect. 113, 425–430.
Hu, W., Yang, H., Sun, J., Zhang, Q., Yang, J., Lu, L., et al., 2011. Polymorphisms in CYP1B1
modify the risk of idiopathic male infertility with abnormal semen quality. Clin.
Chim. Acta 412, 1778–1782.
Hu, Z., Xia, Y., Guo, X., Dai, J., Li, H., Hu, H., et al., 2012. A genome-wide association study in
Chinese men identifies three risk loci for non-obstructive azoospermia. Nat. Genet.
44, 183–186.
Hu, Z., Li, Z., Yu, J., Tong, C., Lin, Y., Guo, X., et al., 2014. Association analysis identifies new
risk loci for non-obstructive azoospermia in Chinese men. Nat. Commun. 5, 3857.
Ji, G., Long, Y., Zhou, Y., Huang, C., Gu, A., Wang, X., 2012. Common variants in mismatch
repair genes associated with increased risk of sperm DNA damage and male infertil-
ity. BMC Med. 10, 49.
Kay, J., Weitzman, P., 1987. Krebs' Citric Acid Cycle: Half a Century and Still Turning (Sym-
posium); Research Books.
Koehl, M., 2015. Gene-environment interaction in programming hippocampal plasticity:
focus on adult neurogenesis. Front. Mol. Neurosci. 8, 41.
La Maestra, S., De Flora, S., Micale, R.T., 2014. Does second-hand smoke affect semen qual-
ity? Arch. Toxicol. 88, 1187–1188.
Langenberg, C., Sharp, S.J., Franks, P.W., Scott, R.A., Deloukas, P., Forouhi, N.G., et al., 2014.
Gene-lifestyle interaction and type 2 diabetes: the EPIC interact case-cohort study.
PLoS Med. 11, e1001647.
Liao, C., Kannan, K., 2014. A survey of alkylphenols, bisphenols, and triclosan in personal
care products from China and the United States. Arch. Environ. Contam. Toxicol. 67,
50–59.
 W. Hu et al. / Environment International 91 (2016) 188–195
Lo Giacco, D., Chianese, C., Sanchez-Curbelo, J., Bassas, L., Ruiz, P., Rajmil, O., et al., 2014.
Clinical relevance of Y-linked CNV screening in male infertility: new insights based
on the 8-year experience of a diagnostic genetic laboratory. Eur. J. Hum. Genet. 22,
754–761.
Manuck, S.B., McCaffery, J.M., 2014. Gene-environment interaction. Annu. Rev. Psychol.
65, 41–70.
Meeker, J.D., Rossano, M.G., Protas, B., Diamond, M.P., Puscheck, E., Daly, D., et al., 2008.
Cadmium, lead, and other metals in relation to semen quality: human evidence for
molybdenum as a male reproductive toxicant. Environ. Health Perspect. 116,
1473–1479.
Meng, Q., Ren, A., Zhang, L., Liu, J., Li, Z., Yang, Y., et al., 2015. Incidence of infertility and
risk factors of impaired fecundity among newly married couples in a Chinese popu-
lation. Reprod. BioMed. Online 30, 92–100.
Qin, Y., Du, G., Chen, M., Hu, W., Lu, C., Wu, W., et al., 2014. Combined effects of urinary
phytoestrogens metabolites and polymorphisms in metabolic enzyme gene on idio-
pathic male infertility. Arch. Toxicol. 88, 1527–1536.
Shen, H., Xu, W., Zhang, J., Chen, M., Martin, F.L., Xia, Y., et al., 2013. Urinary metabolic bio-
markers link oxidative stress indicators associated with general arsenic exposure to
male infertility in a Han Chinese population. Environ. Sci. Technol. 47, 8843–8851.
Song, W.T., Wang, Z.J., Liu, H.C., 2014. Effects of individual and binary mixtures of estro-
gens on male goldfish (Carassius auratus). Fish Physiol. Biochem. 40, 1927–1935.
Tian, M., Bao, H., Martin, F.L., Zhang, J., Liu, L., Huang, Q., et al., 2014. Association of DNA
methylation and mitochondrial DNA copy number with human semen quality. Biol.
Reprod. 91, 101.
Tomita, M., Kami, K., 2012. Cancer. systems biology, metabolomics, and cancer metabo-
lism. Science 336, 990–991.
Wang, R.S., Yeh, S., Tzeng, C.R., Chang, C., 2009. Androgen receptor roles in spermatogen-
esis and fertility: lessons from testicular cell-specific androgen receptor knockout
mice. Endocr. Rev. 30, 119–132.
195
Wang, T.J., Larson, M.G., Vasan, R.S., Cheng, S., Rhee, E.P., McCabe, E., et al., 2011. Metabo-
lite profiles and the risk of developing diabetes. Nat. Med. 17, 448–453.
Wu, W., Lu, J., Tang, Q., Zhang, S., Yuan, B., Li, J., et al., 2013. GSTM1 and GSTT1 null poly-
morphisms and male infertility risk: an updated meta-analysis encompassing 6934
subjects. Sci. Rep. 3, 2258.
Xia, Y., Chen, M., Zhu, P., Lu, C., Fu, G., Zhou, X., et al., 2013. Urinary phytoestrogen levels
related to idiopathic male infertility in Chinese men. Environ. Int. 59, 161–167.
Xu, B., Chen, M., Ji, X., Mao, Z., Zhang, X., Wang, X., et al., 2014. Metabolomic profiles de-
lineate the potential role of glycine in gold nanorod-induced disruption of mitochon-
dria and blood-testis barrier factors in TM-4 cells. Nanoscale 6, 8265–8273.
Yamamoto, T., Jaroenporn, S., Pan, L., Azumano, I., Onda, M., Nakamura, K., et al., 2009. Ef-
fects of pantothenic acid on testicular function in male rats. J. Vet. Med. Sci. 71,
1427–1432.
Zegers-Hochschild, F., Adamson, G.D., de Mouzon, J., Ishihara, O., Mansour, R., Nygren, K.,
et al., 2009. International Committee for Monitoring Assisted Reproductive Technol-
ogy (ICMART) and the World Health Organization (WHO) revised glossary of ART
terminology, 2009. Fertil. Steril. 92, 1520–1524.
Zeng, Q., Feng, W., Zhou, B., Wang, Y.X., He, X.S., Yang, P., et al., 2015. Urinary metal con-
centrations in relation to semen quality: a cross-sectional study in China. Environ. Sci.
Technol. 49, 5052–5059.
Zhang, J., Huang, Z., Chen, M., Xia, Y., Martin, F.L., Hang, W., et al., 2014a. Urinary metab-
olome identifies signatures of oligozoospermic infertile men. Fertil. Steril. 102, 44–53
(e12).
Zhang, J., Mu, X., Xia, Y., Martin, F.L., Hang, W., Liu, L., et al., 2014b. Metabolomic analysis
reveals a unique urinary pattern in normozoospermic infertile men. J. Proteome Res.
13, 3088–3099.