Metformin chlorination byproducts in drinking water exhibit marked toxicities of a potential health concern

Abstract

Metformin (MET), a worldwide used drug for type 2 diabetes, has been found with the largest amount by weight among all drugs in aquatic environment, including the drinking water systems where this emerging micropollutant is inevitably transformed during chlorination process. Whether MET chlorination byproducts Y (C4H6ClN5) and C (C4H6ClN3) exist in drinking water remains unknown. Although MET has health-promoting properties, whether or how its chlorination byproducts affect health is still uncharacterized. Here we reveal that MET and byproduct C are present in worldwide drinking water with the highest doses detected for MET and C as 1203.5 ng/L and 9.7 ng/L respectively. Under simulated chlorination conditions, we also demonstrate that both byproducts can be increasingly produced with increment of MET concentration, suggesting a hidden threat on the safety and sustainability of global water supply. Through systematic evaluations, we demonstrate that MET chlorination byproducts Y and C exhibit toxicities instead of genotoxicity to live worms and human HepG2 cells at millimolar doses. Moreover, both byproducts are harmful to mice and particularly Y at 250 ng/L destroys the mouse small intestine integrity. Unprecedentedly, we unveil boiling and activated carbon adsorption as effective alternative solutions that may be in urgent demand globally for removing these byproducts from drinking water.

Full text

Environment International 146 (2021) 106244
Available online 3 November 2020
0160-4120/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Metformin chlorination byproducts in drinking water exhibit marked
toxicities of a potential health concern
Runshuai Zhang
a
,
b
,
c
,
1
, Yuanzhen He
d
,
e
,
f
,
1
, Luxia Yao
a
,
b
,
c
,
1
, Jie Chen
a
,
b
,
c
, Shihao Zhu
a
,
b
,
c
,
Xinxin Rao
g
, Peiyuan Tang
g
, Jia You
a
,
b
,
c
, Guoqiang Hua
g
, Lu Zhang
d
,
e
,
f
, Feng Ju
d
,
e
,
f
,
*
,
Lianfeng Wu
a
,
b
,
c
,
*
a
Key Laboratory of Growth Regulation and Translational Research of Zhejiang Province, School of Life Sciences, Westlake University, Hangzhou, Zhejiang, China
b
Westlake Laboratory of Life Sciences and Biomedicine, Hangzhou, Zhejiang, China
c
Institute of Basic Medical Sciences, Westlake Institute for Advanced Study, 18 Shilongshan Road, Hangzhou 310024, Zhejiang Province, China
d
School of Engineering, Westlake University, Hangzhou, Zhejiang, China
e
Institute of Advanced Technology, Westlake Institute for Advanced Study, 18 Shilongshan Road, Hangzhou 310024, Zhejiang Province, China
f
Key Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, Zhejiang, China
g
Institute of Radiation Medicine and Fudan University Shanghai Cancer Center, Shanghai Medical School, Fudan University, Shanghai 200032, China
ARTICLE INFO
Handling Editor: Dr. Frederic Coulon
Keywords:
Metformin
Metformin chlorination byproduct
Drinking water
Toxicity
Public health
ABSTRACT
Metformin (MET), a worldwide used drug for type 2 diabetes, has been found with the largest amount by weight
among all drugs in aquatic environment, including the drinking water systems where this emerging micro-
pollutant is inevitably transformed during chlorination process. Whether MET chlorination byproducts Y
(C
4
H
6
ClN
5
) and C (C
4
H
6
ClN
3
) exist in drinking water remains unknown. Although MET has health-promoting
properties, whether or how its chlorination byproducts affect health is still uncharacterized. Here we reveal
that MET and byproduct C are present in worldwide drinking water with the highest doses detected for MET and
C as 1203.5 ng/L and 9.7 ng/L respectively. Under simulated chlorination conditions, we also demonstrate that
both byproducts can be increasingly produced with increment of MET concentration, suggesting a hidden threat
on the safety and sustainability of global water supply. Through systematic evaluations, we demonstrate that
MET chlorination byproducts Y and C exhibit toxicities instead of genotoxicity to live worms and human HepG2
cells at millimolar doses. Moreover, both byproducts are harmful to mice and particularly Y at 250 ng/L destroys
the mouse small intestine integrity. Unprecedentedly, we unveil boiling and activated carbon adsorption as
effective alternative solutions that may be in urgent demand globally for removing these byproducts from
drinking water.
1. Introduction
Metformin (MET), as the rst-line therapy for type 2 diabetes (T2D),
is one of the most prescribed medications in the world. It is anticipated
to be consumed in a much greater amount, due to the rapid global in-
crease in T2D prevalence and the additionally discovered benets of
MET for cancer prevention (Knowler et al., 2002), women’s infertility
treatment ([2]) and lifespan extension (Soukas et al., 2019; Wu et al.,
2016). Typically, MET is prescribed with a starting dosage of 500 mg,
twice a day for weeks, then increased up to a total of 2,550 mg per day
dependent on the tolerance of patients. A latest survey reported that
MET was ranked as the 4th most prescribed drug in 2020 and its pre-
scribing rate increased by 44.2% from 54.5 million in 2006 to 78.6
million in 2017 in the U.S. (ClinCalc). Meanwhile, China consumes
approximately 786 metric tons of MET annually in recent years (Yan
et al., 2019). It is noted that MET is not metabolized by human and gets
almost 100% excreted unmodied (Rena et al., 2017), resulting in
increasing concerns on the potential ecotoxicology of MET and its
* Corresponding authors at: Key Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou,
Zhejiang, China (F. Ju). Key Laboratory of Growth Regulation and Translational Research of Zhejiang Province, School of Life Sciences, Westlake University,
Hangzhou, Zhejiang, China (L. Wu).
E-mail addresses: jufeng@westlake.edu.cn (F. Ju), wulianfeng@westlake.edu.cn (L. Wu).
1
These authors contributed equally to this work.
Contents lists available at ScienceDirect
Environment International
journal homepage: www.elsevier.com/locate/envint
https://doi.org/10.1016/j.envint.2020.106244
Received 19 September 2020; Received in revised form 23 October 2020; Accepted 23 October 2020

Environment International 146 (2021) 106244
2
byproducts after release into the aquatic systems (Caldwell et al., 2019;
Lena Stütz, 2019). Recently, MET present in surface water has been
found to be problematic to wild shes, causing more aggressive behavior
(MacLaren et al., 2018). In fact, MET has been widely found with the
largest amounts among all drugs in waterways and considered as an
emerging pollutant (Ussery et al., 2019). Although MET could be
removed at efciencies ranging from 84% −99% in wastewater treat-
ment plants (Ju et al., 2019), it is still widely detected in surface water
and drinking water worldwide (Caldwell et al., 2019; Scheurer et al.,
2012), necessitating the study of the potential impacts of MET and its
derivatives on the safety and sustainability of drinking water supply and
human health.
At the beginning of our previous MET study in 2012 (Wu et al.,
2016), we incidentally observed an instant reaction of MET with hy-
pochlorite regularly used for Caenorhabditis elegans synchronization in
the worm eld (Fig. S1). However, the reaction and its yielded two
chlorination byproducts of MET: Y (yellow, C
4
H
6
ClN
5
, 159.58 g/mol)
and C (colorless, C
4
H
6
ClN
3
, 131.56 g/mol) were brought to light by the
elegant work of Armbruster D et al in 2015 (Armbruster et al., 2015).
Chlorine is commonly used for both regional and household drinking
water disinfection worldwide owing to its ease of application, low cost,
and high efciency in protecting water from live germ contamination
(Lantagne and Clasen, 2012). We argue that if MET, continuously
released and potentially accumulated in surface water (Blair et al., 2013;
Elliott et al., 2018), is inevitably transformed into chlorination
byproducts of any potential toxicity during chlorination process, it will
likely bring a widespread health threat to all consumers of the water
containing such byproducts. MET is thus far the safest drug for diabetes
treatment, although it rarely causes toxicity or lactic acidosis when
overdose in humans (Florez, 2017). In a recent study, a crude fraction of
MET byproducts was found with genotoxicity in bacterial Salmonella
typhimurium (Lena Stütz, 2019). However, whether or how much the
pure MET chlorination byproducts Y and C may induce genotoxicity to
animals and affect health of animals remains completely unknown.
Theoretically, these byproducts are increasingly generated and released
into the household taps with more MET present in water sources, but
how much they are present in drinking water has not been monitored
and paid enough attention yet.
In the current study, we rstly puried the MET chlorination
byproducts Y (90.8% of purity) and C (99.5% of purity), traced their
present concentrations in drinking water globally and systematically
evaluated toxicities of these puried compounds in animal models and
human cells. We examined water samples from household taps of mul-
tiple countries and found widespread presences of MET and its chlori-
nation byproduct C in both tap water and drinking water sources.
Strikingly, we noticed that neither of MET byproducts induces geno-
toxicity to nematode worms, but both are markedly toxic to worms and
cultured human cells. Moreover, both byproducts are harmful to mice,
and the byproduct Y at potentially achievable doses in tap water
remarkably destroys the integrity of the mouse intestinal epithelium in a
mechanistic manner likely through the inhibition of intestinal epithe-
lium self-renewal. Continuous disruption of such integrity is known as
basis for numerous gut diseases (Mittal and Coopersmith, 2014) and will
induce serious and irreversible effects on health if not stopped.
Together, we demonstrate the MET chlorination byproducts as a hidden
and serious threat to global health and wellbeing. Fortunately, our
exploration on solutions for removing these hazards from drinking water
demonstrates boiling and powered activated carbon adsorption as
effective ones, which are alternative mitigation measures urgently
needed for global implementation.
2. Materials and methods
2.1. Synthesis and purication of MET chlorination byproducts Y and C
MET chlorination byproducts were prepared according to a protocol
modied from the previous study by Armbruster D et al (Armbruster
et al., 2015). Briey, a series of steps were carried out including chlo-
rination reaction of MET (Sigma, PHR1084), extraction, rotary evapo-
ration concentration, column chromatography purication and
recrystallization. In terms of the potential heat-labile property of the
byproduct Y, low-temperature recrystallization (4 ℃ in the dark) was
additionally applied to get the compounds Y and C of maximal purity,
followed by separation of the solid–liquid phase and further removal of
water through freeze-drying (-60 ◦C, 12 h). The purities of synthesized
byproducts Y (bright orange powder) and C (colorless powder) were
determined by nuclear magnetic resonance (NMR), as 90.8% and 99.5%
respectively. The synthesized Y and C were then used as standards for
mass spectrometry analysis of their concentrations in global surface
water and drinking water samples, and for the treatment of nematode
worms, cultured cells, and mice. Consistent with the previous observa-
tion that Y is partly converted to C during preparation (Armbruster et al.,
2015), a certain amount of C signal (4.5%) was detected in the purity
analysis of Y, which is also reected in the same results of mass spec-
trometer analysis (Fig. S2) and explains why the purity of Y is lower
(90.8% vs. 99.5%).
2.2. Water sample collection
The tap water sample or its sources from various lakes or rivers were
collected from different cities of China, the U.S., Japan, Korea, and
Philippines from June to September in 2019. All water samples were
collected freshly and refrigerated at −20 ℃ for later transportation and
storage (14 days as the maximal period from collection to analysis).
Under such transportation and storage conditions, MET (no changes
observed in untreated wastewater at −20 ℃ for 21 days (Yan et al.,
2019) and C are likely to be stable according to their documented
properties (Armbruster et al., 2015), but Y could be decomposed with
the temperature and matrix components within water (Armbruster et al.,
2015). Samples were ltered with 0.22
μ
m mixed cellulose (MCE)
membrane prior to analysis by liquid chromatography-triple quadruple
tandem mass spectrometry (UPLC-MS/MS).
2.3. UPLC-MS/MS trace analysis
UPLC-MS/MS (RxionLC
TM
SCIEX 6500+, QTRAP, SCIEX, U.S.) in the
positive electrospray ionization (ESI+) mode was used to measure the
trace amounts of MET and its byproducts Y and C. MET separation was
achieved on an BEH Amide analytical column (50 mm ×2.1 mm, 1.7
μ
m; Waters, U.S.) using a mixture of 0.1% formic acid aqueous solution
containing 5 mM ammonium acetate (eluent A) and acetonitrile (eluent
B: ACN). Gradient elution was set up as: 0 min (95% ACN); hold: 1 min
(95% ACN); ramp: 4 min (40% ACN); hold: 5 min (40% ACN); ramp: 5.5
min (95% ACN); stop: 7 min (95% ACN). Y and C in the collected
samples were separated by use of a reversed phase column (ZORBAX
Eclipse plus C18, 50 mm ×2.1 mm, 1.8
μ
m; Agilent, U.S.) with a
gradient of 95% eluent A (ultrapure water with 0.1% formic acid), hold
for 1 min, rise to 90% eluent B (Methanol plus 0.1% formic acid) in 2
min, hold for 1 min, rise to 95% eluent A in 0.5 min and hold for 2 min
(total time for 7 min). The UPLC was operated at a ow rate of 0.4 mL/
min, with injection volume of 20
μ
L and column temperature at 40 ◦C.
The optimized mass spectrum conditions were run as follows: source
temperature at 500 ◦C; IonSpray voltage at 2,000 V; curtain gas at 35 psi;
Ion source gas 1 and 2 at 55 psi and 50 psi, respectively. The MET was
quantied based on the external calibration using the MET (Sigma,
PHR1084) as standard and its recovery ranged from 98% to 126% with
quantication based on external calibration and recovery. The limits of
detection and quantication (LOD and LOQ) for MET were 2 ng/L and 5
ng/L, respectively. Our synthesized byproducts Y and C were used as
standards to analyze water samples collected. The recovery of the
byproduct C from the water samples was 86.9% (mean), and its LOD and
LOQ were determined as 0.2 ng/L and 0.5 ng/L, respectively. The
R. Zhang et al.

Environment International 146 (2021) 106244
3
quantication of Y was achieved by multiple reaction monitoring
(MRM) chromatogram of C (132.2 to 71.0; the retention time of C is at
2.47 min) as it was also present at the retention time of Y (2.70 min)
after passing through the ion source (Fig. S2). The signal decomposition
from Y to C appeared to be linearly correlated well. Thus, the LOD and
LOQ of Y were determined as 72.6 ng/L and 90.8 ng/L respectively
(Fig. S2).
2.4. Disinfection assay with water containing MET
To simulate the formation of MET byproducts Y and C during chlo-
rination, NaOCl (Aladdin, Shanghai, China, Cat. S101636) was used to
react with MET in ultrapure water (Milli-Q system) in triplicates. In
brief, 1.5 mL NaOCl solution with 100 mg/L active chlorine was added
into 50 mL glass tube containing 48.5 mL of metformin solution in the
range of 4
μ
g/L (currently reported level in the drinking water source of
this study) to 640
μ
g/L. The chlorine concentration was within the limit
of Standards for Drinking Water Quality in China (GB 5749–2006,
2006). The reaction was carried out and on stirring plate in dark at low
speed for 30 min to ensure the enough reaction time. At the end of re-
action (2 h), the reaction mixtures were quenched with 1
μ
L, 10 g/L
sodium thiosulfate (Na
2
S
2
O
3
) followed by transferring into 1.5 mL
brown sample vials. Collected samples were stored in the dark at −20 ℃
and analyzed within 3 days.
2.5. Cell culture and cell viability test
HepG2 (liver hepatocellular carcinoma) cells were cultured in Dul-
becco’s Modied Eagle’s Medium-High Glucose (DMEM) medium con-
taining L-glutamine (4 mM; Sigma) and 1 mM sodium pyruvate solution
(100 mM; Sigma), 10% fetal bovine serum (FBS) and 1% pen-
icillin–streptomycin. For the cytotoxicity test, the cells were seeded in
24 wells plates at 5 ×10
4
cells per well overnight before drug treatment.
The cell viability was determined by trypan blue staining after the 12-
hour treatment of the tested compounds at indicated dose.
2.6. Worm culture
The wild-type C. elegans strain N2 was maintained at 20 ℃ on
nematode growth media (NGM) plates that were seeded with Escherichia
coli (E. coli) OP50 as a food source. Toxicity tests were conducted with L4
larvae in M9 buffer with different concentrations of MET (Sigma,
PHR1084), sodium (meta) arsenite (Sigma, S7400), and the byproduct Y
or C.
2.7. Mice
Female C57BL/6 mice at 8–10 weeks old were used for acute and
chronic toxicity tests. For the acute toxicity test, the compounds were
diluted in PBS and administered intraperitoneally. Animals were
monitored and recorded on survival and behavior changes for 7 days.
For chronic toxicity test, compounds were delivered through drinking
water for one month. Drinking water containing byproduct Y or C was
prepared freshly every day. Mice were maintained and handled in
accordance with institutional guidelines, and all animal procedures
were approved by the Institutional Animal Care and Use Committee of
Westlake University.
2.8. Genotoxicity test in C. elegans
The strategy for testing genotoxicity of byproducts Y and C (1 mM) is
designed according to the classical protocol for C. elegans mutagenesis
by ethyl methanesulfonate (EMS) (Wu et al., 2016). Phenotypes,
including development, abnormality in shape and behavior, and mor-
tality, were analyzed and recorded throughout the whole process.
2.9. Tissue preparation for immunohistochemistry (IHC)
Dissected intestinal tissues were thoroughly rinsed twice before
being xed in 10% neutral-buffered formalin (Leagene, DF0111) for 24
h and embedded in parafn blocks. Parafn sections (5
μ
m) were stained
with hematoxylin/eosin (H&E; Merck, Darmstadt, Germany). For
immunohistochemistry (IHC), sections were boiled in 0.1 M citrate
buffer (PH 6.0) 30 min for antigen retrieval. The boiled sections were
incubated with Ki67 antibody (Abcam, ab16667; dilution 1:500) over-
night at 4 ◦C in a humidied chamber followed by detection employing
UltraSensitiveTM SP (Mouse/Rabbit) IHC Kit (Maxim, KIT-9720) and
DAB kit (Maxim, DAB-1031) according to manufacturer’s instructions.
All stained sections were scanned using TissueFAXS microscope (Tis-
sueFAX plus; TissueGnostics, Vienna, Austria).
2.10. Boiling test
100 mL Y or C solutions, at a concentration of 2,509.5 ng/L and
790.4 ng/L respectively, in 250 mL beakers were heated by an electric
water heater. Beakers were sealed with foil to minimize evaporation loss
of solutions. The samples were immediately collected into 1.5 mL brown
sample vials after boiling, and frozen for storage at −20 ◦C before
analysis. Triplicated experiments were conducted, and concentrations of
Y and C before and after boiling were determined via UPLC/MS/MS.
2.11. Activated carbon adsorption
Two forms of activated carbon with different particle sizes (200 mesh
and 1–2 mm; Huanyu Carbon Industry CO., LTD. China) were evaluated
as the potential sorbents of MET chlorination byproducts according to
the manufacturer’s instructions. Both powdered and granular activated
carbon were dried in oven at 80 ◦C for 5 h to remove water. In each test,
100 mg activated carbon was added to 30 mL Y solutions (2,509.5 ng/L)
or C solutions (790.4 ng/L) for adsorption followed by 24-hour equili-
bration in a constant temperature shaker (25 ◦C). At the end of equili-
bration, activated carbon was removed from Y or C solutions with 0.22
μ
m MCE membrane lters. The concentrations of compound Y or C
remaining in solutions were subsequently determined by UPLC-MS/MS.
All reactions were carried out in brown glass vials.
2.12. Statistical analysis
For cell viability tests, data were obtained from 3 biological repli-
cates. For the Ki67 staining assay, at least 5 images for each condition
were quantied and analyzed. Experimental differences were analyzed
using one-way ANOVA followed by Dunnett’s multiple comparisons test
in Prism 7.0 d software. Values represent mean ±standard error of the
mean (SEM). P value ≤0.05 is considered of signicance.
3. Results
3.1. Current contamination and projected future accumulation of MET
and its chlorination byproducts in the global drinking water
Currently, MET is widely detected in higher concentrations in surface
water (8.4–34,000 ng/L) worldwide, including many as drinking water
sources (Fig. 1a). To address whether its chlorination byproducts exist in
drinking water, we synthesized the two byproducts Y and C with an
optimized method from the reported synthesis method (Armbruster
et al., 2015), including low-temperature recrystallization and freeze-
drying steps (see methods), and rstly puried them as detection stan-
dards for the following dose assessment. We employed UPLC-MS/MS
and the puried standards to examine water samples from household
taps worldwide, especially where surface water is used for a drinking
water source. Markedly, we found that the current MET concentration
ranges from 5.1 to 1,203.5 ng/L, and byproduct C ranges from 1.3 to 9.7
R. Zhang et al.

Environment International 146 (2021) 106244
4
ng/L in urban drinking water from multiple countries, including China
and the U.S. (Fig. 1b and Table S2). The detection limit of byproduct Y is
relatively high by UPLC-MS/MS (72.6 ng/L) possibly because Y is labile
under the high temperature of the ion source of the mass spectrometer
(Fig. S2) and as reported by others (Armbruster et al., 2015).
Although the current concentrations of Y in tap water samples fall
below the detection limit of the UPLC-MS/MS method established in this
study, possibly due to its detection limit on the UPLC-MS/MS system, C
has been widely detected in tap water from multiple countries and even
found in a lake water from China at a concentration of 3.5 ng/L, sug-
gesting that the MET byproduct C has entered into the water cycling
from household taps to surface water resources (Table S2). Accumu-
lating evidence shows that MET typically exceeds 1
μ
g/L in surface
water worldwide (Fig. 1a) and is particularly high in some surface
waters in the U.S. (Elliott et al., 2018), China (Kong et al., 2015) and
Canada (de Solla et al., 2016) as 34
μ
g/L, 20
μ
g/L and 10
μ
g/L,
respectively. Unfortunately, it is noteworthy that many countries,
including the U.S. and China, take surface water for drinking water
generation without bank ltration or any other underground passage
process, which are thought to be efcient ways to reduce MET from
surface water (Scheurer et al., 2012). To predict the presences of these
chlorination byproducts in tap water with increasing MET levels, we
carried out simulated disinfection experiments and found steady growth
of Y and C with increased addition of MET during chlorination process
(Fig. 1c and Table S3). Thus, theoretically, these byproducts in tap water
are likely to achieve health-threatening doses over time, revealing a
hidden threat on the safety and sustainability of global drinking water
supply.
Fig. 1. Global view of the MET occurrence in surface water and levels of its chlorination byproducts present in drinking water. (a) Concentrations of MET
currently reported in surface water worldwide. The black spots on the map represent the reported sampling sites in the corresponding countries. The reported
concentration and year are provided and labeled with related references in each box, where the three spots are highlighted in orange with the highest concentration
of MET reported in surface water. The superscripts stand for corresponding references listed in Table S1. (b) MET byproduct C is widely detected in household tap
water. The bar graph represents the results of three water samples from the same place of the top three C-rich cities examined. The error bar represents the SEM. (c) A
simulation experiment was carried out to depict the theoretical production of MET chlorination byproducts with a xed concentration of chlorine (0.3%) regularly
used in water treatment and an increasing addition of MET. The byproduct Y is colored in red and C is colored in light blue. Pearson correlation R
2
and correlation P
value (one-tail) are calculated and provided separately. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of
this article.)
R. Zhang et al.

Environment International 146 (2021) 106244
5
3.2. MET chlorination byproducts are markedly toxic to live animals and
human cells
To conrm whether chlorination has transformed MET into geno-
toxic byproducts, we used the rstly puried Y and C at 1 mM (Y =1.6
×10
5
μ
g/L, C =1.3 ×10
5
μ
g/L), which are twofold higher doses than
that used in the previous study (~ 0.4 mM) (Lena Stütz, 2019), to treat
the parent generation (P0) of nematode worms, but did not nd any
abnormal F1 and F2 progenies among over 15,000 animals per condi-
tion, indicating that these compounds may not be able to induce geno-
toxicities to the animals (Fig. 2a). However, we repeatedly observed
evident general toxicities of Y and C to the worms in the genotoxicity
assay (Fig. 2b) and the conrmed survival assay (Fig. 2c). Strikingly, we
found both byproducts at 1 mM exhibit similar or higher toxicity than
MET does at 100 mM (1.29 ×10
7
μ
g/L) to C. elegans (Fig. 2c), which
indicates that chlorination has turned MET into 100 times more toxic
compounds to live animals C. elegans. Researchers frequently use very
high doses and long periods of treatment of MET to induce the effect of
interest and seldom consider MET as a toxic reagent. For instance, re-
searchers use over 10 mM (1.3 ×10
6
μ
g/L) in cells (Kalender et al.,
2010), 50 mM (6.6 ×10
6
μ
g/L) in C. elegans (Wu et al., 2016; Cabreiro
et al., 2013) or 150 mg/kg injected intraperitoneally in mice (Ludman
and Melemedjian, 2019) generally for 24 or 48 h in cells and days or
weeks in animals (Wu et al., 2016; Kalender et al., 2010; Ludman and
Melemedjian, 2019). To dene the toxicity level of MET chlorination
byproducts, we compared them with arsenic, a well-known poisonous
compound ever being neglected in drinking water and eventually
resulting in the most serious public health issue in human history
(Schmidt et al., 2016). Strikingly, we found that both MET chlorination
byproducts are much more toxic than arsenic when all employed at 2
Fig. 2. MET chlorination byproducts exhibit marked toxicity, instead of genotoxicity, to nematode worms and cultured human HepG2 cells. (a) Workow
for the genotoxicity test of MET chlorination byproducts Y and C in Caenorhabditis elegans. P0 means the parent animals treated by Y or C (1 mM, Y =1.6 ×10
5
μ
g/L,
C =1.3 ×10
5
μ
g/L) for 6 h, while F1 and F2 stand for the rst- and second-generation progenies of treated P0 animals respectively. The evaluations were conducted
daily, and abnormal phenotypes compared to vehicle group (Veh) were only observed around 14 h post Y/C treatment. (b) Development variations were observed in
the P0 generation treated by Y or C, shown in images, and quantied in the bar gure. (c) Survival rate of larvae L4 worms treated with MET byproducts Y or C (1
mM, Y =1.6 ×10
5
μ
g/L, C =1.3 ×10
5
μ
g/L; and 2 mM, Y =3.2 ×10
5
μ
g/L, C =2.6 ×10
5
μ
g/L) and As (1 mM, 1.3 ×10
5
μ
g/L; 2 mM, 2.6 ×10
5
μ
g/L) at 50 or 100
times lower concentrations than MET (100 mM, 1.3 ×10
7
μ
g/L) for 12 h. (d) Treatment of MET byproducts Y and C (0.1 mM, Y =1.6 ×10
4
μ
g/L, C =1.3 ×10
4
μ
g/L;
and 0.2 mM, Y =3.2 ×10
4
μ
g/L, C =2.6 ×10
4
μ
g/L) for 12 h resulted in remarkable deaths of HepG2 cells, which are comparable to the death rates induced by the
same doses of As treatment (0.1 mM, 1.3 ×10
4
μ
g/L;0.2 mM, 2.6 ×10
4
μ
g/L), while MET at 20 mM (2.6 ×10
6
μ
g/L) did not cause signicant cell deaths. It is worth
mentioning that HepG2 cells, displaying many genotypic features and metabolic manners of human liver, are widely considered as one of the best in vitro systems for
predicting human toxicity potential of compounds of interest (O’Brien, 2014). The cell viability was determined by trypan blue staining. The results are presented as
the mean SEM of three biological replicates. ** P <0.01, *** P <0.001, **** P <0.0001 from one-way ANOVA analysis followed by Dunnett’s multiple comparisons
test. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
R. Zhang et al.

Environment International 146 (2021) 106244
6
mM doses, while the byproduct C is 5 times more toxic than arsenic to
nematode worms (Fig. 2c).
To assess the cytotoxicity of MET chlorination byproducts, we chose
the human hepatoma carcinoma HepG2 cell, which is widely used in
MET studies (Zang et al., 2004) and cytotoxicity tests (Gerets et al.,
2012). In line with the worm results, we found that MET at 20 mM (2.6
×10
6
μ
g/L) reduced the viability of HepG2 cells slightly, whereas at 0.1
mM (Y =1.6 ×10
4
μ
g/L, C =1.3 ×10
4
μ
g/L) both Y and C induced
marked cell death and nearly killed all cells at a concentration of 0.2 mM
(Y =3.2 ×10
4
μ
g/L, C =2.6 ×10
4
μ
g/L) as measured by trypan blue
exclusion assay after 12 h of treatment (Fig. 2d). To also have an evident
sense of the toxicity level of these byproducts in human cells, we
compared them with arsenic and found that both MET chlorination
byproducts exert similar or even higher toxicity than arsenic to HepG2
cells at the same doses (Fig. 2d). The 50% lethal concentrations (LC50)
of Y and C were titrated as 116.5
μ
M (1.9 ×10
4
μ
g/L) and 90.9
μ
M (1.2
×10
4
μ
g/L) respectively (Fig. S3), which are consistently found com-
parable to the documented arsenic’s LC50 at 100
μ
M (2.0 ×10
4
μ
g/L) in
HepG2 cells (Gerets et al., 2012).
3.3. MET chlorination byproducts exhibit deleterious effects to mice and
their small intestine integrity
To evaluate the toxicity of Y and C at the physiological level, we
Fig. 3. Both MET byproducts Y and C are harmful to mice, and the byproduct Y destroys the mouse small intestine integrity. (a) Seven-day survival rate of
mice after single dose intraperitoneal injection (I.P.) of MET, Y and C at varied doses labeled next to the compound names. For instance, MET150 means the injected
dose of MET at 150 mg/kg body weight of the treated mouse. Phosphate-buffered saline (PBS) served as the vehicle for the injected compounds and the injection
control. The number of mice (n) injected in each group is provided in parentheses. Detailed observations of the mice after injection and more acute toxicity groups are
attached in Fig. S4 and Table S4. (b) Single-dose injection of Y at doses of 1.25 mg/kg and higher led to severe mouse small intestine abnormalities. Surviving mice
were examined for tissue or organ alterations by different doses of Y on day 7 after injection. Small intestines of representative treated groups were isolated to show
the abnormalities. In the chronic response test for (c) and (d), all mice that survived from drinking water with byproduct Y (0.25, 2.5, and 25
μ
g/L) or without for one
month were dissected for evaluation by hematoxylin-eosin (H&E) staining and Ki67 immunohistochemistry (IHC) staining at the end of exposure. Images are
representative of 8 animals for each group. (c) A month of exposure to Y at 2.5
μ
g/L and 25
μ
g/L caused marked morphological changes in the small intestine crypts,
circled in white dashed lines. (d) Byproduct Y inhibits the proliferation of crypt cells in the jejunum in a dose-dependent manner. Quantication of the Ki67-positive
cells in each crypt was conducted and is shown next to the IHC images. Bars represent the SEM (n =100–150 crypts from 8 mice). **** P <0.0001 by one-way
ANOVA followed by Dunnett’s multiple comparisons post hoc test.
R. Zhang et al.

Environment International 146 (2021) 106244
7
examined the mouse response to these compounds acutely via a single
intraperitoneal injection and chronically through one month of drinking
water. In the acute test, we found that all 8 mice injected with 150 mg/
kg body weight MET, a regular dose used in MET research (Ludman and
Melemedjian, 2019), survived through the 7 day test period and did not
show any obvious abnormalities, while 6 of the 8 mice injected with the
same dose of C trembled the body, secreted white substances immedi-
ately after injection, and died within 2 days (Fig. 3a, S4 and Table S4).
However, all 8 mice survived from the 100 mg/kg test of C. In a dramatic
contrast, injection of Y at 50 mg/kg led to the deaths of all 8 mice within
2 days and 3 out of 8 deaths at 10 mg/kg 7 days postinjection (Fig. 3a
and Table S4), which indicates the more potent toxicity of Y than that of
C in mice. Moreover, we noticed that the intestines of mice were
swollen, dark and bloody after injection with doses higher than 10 mg/
kg Y, and swelling was even observed at the low level of 1.25 mg/kg
(Fig. 3b and S4), suggesting that Y may attack the mouse small intestines
specically and remarkably.
Chronic response tests were performed to determine whether low
doses of Y and C could cause adverse effects in mice through daily
exposure in drinking water for one month. After one month of drinking
the water containing 2.5 mg/L C, the mice did not show any obvious
abnormal behaviors (data not shown) or intestinal morphology changes
as determined by histological staining (Fig. S5a). However, consistent
with the acute test results, we observed that the byproduct Y caused
morphological changes in the crypts at the jejunum of the small intestine
(Fig. 3c), where the majority of nutrients are absorbed (Kiela and
Ghishan, 2016), without affecting the normal growth of tested animals
(Fig. S5b), after one month of treatment in drinking water at a dose of
0.25
μ
g/L. As documented, most cells at the base of small intestine
crypts are stem cells that normally turn over every 3 to 4 days for cell
renewal of the epithelium (Hua et al., 2017; Clevers, 2013). To deter-
mine whether Y induces intestinal epithelial layer changes by affecting
cell proliferation of crypt cells, we conducted an immunostaining assay
against the protein Ki67, a marker of cell proliferation (Scholzen and
Gerdes, 2000). According to Ki67 staining, Y inhibits the proliferation of
cells at the base of the small intestine crypt in a dose-dependent manner
(Fig. 3d), which implicates that Y impedes self-renewal of the intestinal
epithelium by blocking cell proliferation. The continuous disruption of
the intestinal epithelium by the byproduct Y is likely to induce serious or
irreversible threats to health and wellbeing.
3.4. Solutions for removing MET chlorination byproducts from water
Given that the evident toxicities of MET byproducts to nematode
worms and cells are comparable or even higher than arsenic and the
byproduct Y impairs the integrity of the mouse small intestine when
administered in drinking water, we should not experience any epide-
miological disaster like the one caused by arsenic contamination in
drinking water, which has been recognized as a historical mistake by
ignoring arsenic in drinking water quality control (Smith et al., 2002).
The reason why we have not seen the causality of potential health risks
from MET chlorination byproducts in drinking water may be thanks to
our means to use tap water, such as boiling the water before drinking
that is regularly employed in many countries, including China. We found
that boiling makes the byproduct Y completely undetectable and C
slightly reduced from water (Fig. 4) and turns the byproduct Y, even at a
highly lethal dose, into other currently unknown but safe byproduct(s)
to mice (50 mg/kg Y, I.P., Fig. S6). The reduction of harms to mice after
water boiling implies that boiling could be one option to avoid the po-
tential risks from drinking the water containing MET chlorination
byproducts, especially for the harmful Y.
To prevent MET chlorination byproducts in tap water from reaching
the harmful levels, we exploited the possibility that activated carbon
adsorption, which is commonly used in water treatment, might help to
remove these compounds from water. Activated carbon adsorption is a
well-established approach in removing disinfection byproducts
generated in water treatment due to its well-known effectiveness and the
ease of activated carbon regeneration (Chiu et al., 2012; Cuthbertson
et al., 2019). We found that powdered activated carbon adsorbs
byproducts Y and C more effectively than the granular form does
(Fig. 4), the results of which can be leveraged to guide water treatment
routes and develop additional processes for completely removing Y and
C from drinking water.
4. Discussion
MET has been widely recognized as a panacea to promote health,
especially for the growing population of T2D patients, while our ndings
here reveal that hypochlorite transforms this panacea drug into toxic
players (byproducts Y and C) with demonstrated threats to the health of
life over time. Concluding from the historical lessons of arsenic
contamination in drinking water, Smith AH et al. highlighted that
“delaying action in an attempt to be thorough in research and long-term
planning can be a mistake” (Smith et al., 2002; Smith et al., 2000). Our
current study necessitates the emergent need to globally pay attention to
the MET byproducts widely present in chlorinated drinking water.
Previous lessons have taught us that drinking water containing 50 µg/L
arsenic would cause 1 in 100 people to die from arsenic-related cancers,
including lung, bladder, and skin cancers (Smith et al., 2000). It is
noteworthy that both MET chlorination byproducts Y and C are highly
nitrogenous substances, which is a typical trait of bladder carcinogens
identied to date (Diana et al., 2019). Whether these previously
neglected MET byproducts are potential bladder carcinogens, or
carcinogen precursors needs to be explored.
To our surprise, MET byproduct Y induced signicant toxicity to
mice in both the acute test and the chronic treatment, especially
attacking the mouse gut. The reason why compound Y targets the gut
remains elusive, but collective evidence has proposed the gut as a major
action site of MET, increasing glucose uptake and lactate production
within the intestine (McCreight et al., 2016). More importantly, MET has
been found to be absorbed mostly in the small intestine and highly
accumulated in the jejunum peaks (Bailey et al., 2008). The difference
between Y and C in the toxic effects may be partly contributed by their
notable differences of structural moiety, as triazole-derivate structure of
Y with an active chlorimine moiety is implicated to induce more toxicity
compared to the chloroorganic nitrile structure of C (Armbruster et al.,
2015). However, whether or how its byproduct C affects the health of
mice and wild animals still needs to be further explored, as it is stable
and has been widely detected in tap and surface water. Future studies
are also encouraged to address the potential limitations of this study: 1)
in order to avoid potential degradations of the MET chlorination
Fig. 4. MET chlorination byproducts can be removed from drinking water
by boiling and activated carbon adsorption. Boiling makes byproduct Y
completely undetectable, while powdered activated carbon (PAC) and granular
activated carbon (GAC), can effectively remove both MET byproducts Y and C
from water. MET byproducts Y and C in water were analyzed by UPLC-MS/MS
prior to and after treatment, including boiling and adsorption through PAC or
GAC. N.D. stands for no detectable signal above signal noise.
R. Zhang et al.

Environment International 146 (2021) 106244
8
byproducts during transportation and storage, future drinking water
samples worldwide should be evaluated timely and locally; 2) what we
have evaluated remain limited in growth inhibition test, not covering
the byproducts’ effect on reproduction and animals of varied ages or
conditions; 3) we would consider our mouse data as a preliminary
reference for future study to determine whether the chemical’s effect on
mouse gut is directly caused by the tested compound or other factors
associated with the compound exposure; 4) all our mouse experiments
were conducted within limited time periods compared to the year-by-
year exposure to (drinking) the water containing MET, Y, and C; 5)
the joint toxicities of MET, Y and C that might be more toxic than any
single one of them are not evaluated in the current study.
Although the current doses of MET present in drinking water may not
directly cause safety concerns to humans, the potential threats of its
chlorination byproducts cannot be neglected. Because they were found
to be toxic to live animals C. elegans and mice at doses generally safe for
MET administration, and have been traced in household tap water
consumed daily by billions of people globally, including young and sick
people. In the current study, we demonstrated that boiling and activated
carbon adsorption are feasible solutions to reduce the MET chlorinated
byproducts from water, which can be used timely to stop the growing
health threats to humans and wildlife from daily exposure of these
byproducts. Strategies to remove MET from the environment were
intensively explored (Scheurer et al., 2012; Poursat et al., 2019; Ju et al.,
2019), but very few effective ones have been suggested, including bank
ltration or underground passage (Scheurer et al., 2012) and biodeg-
radation by certain species of bacteria (Ju et al., 2019). Alternatively,
additional, better drug options to replace MET or reduce MET con-
sumption are also in emerging demand and important for the sustain-
ability of global drinking water supply.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgments
We thank Jinheng Pan, Shan Feng, Nanjia Zhou, Hang Shi and
Xiaohuo Shi for the facility support and discussions. We are grateful to
all people volunteered for assistance on water sampling. We thank
Alexander A. Soukas, Dangsheng Li, Ling Li, Yigong Shi, Tian Xu, and Li
Deng for critical discussions.
Funding Sources
This work was supported by institutional funds from the Westlake
University / Westlake Institute for Advanced Study (L.W. and F.J.), by
the Young Scientists Fund of the National Natural Science Foundation of
China, grant number 51908467 (F.J.), by the National Natural Science
Foundation of China, grant number 31670858 (H.G.).
Author Contributions
R. Z., Y. H., L. Y., F. J. and L. W. designed the experiments. R. Z., L. Y.,
J. C., S. Z., X. R., P. T. and Y. J. performed the cell and mouse experi-
ments. Y. H. and L. Z. synthesized the compounds and performed water
sample analysis. J. Y. and G. H. mentored the mouse experiments and
analyzed the results. R. Z., Y. H., L. Y., F. J. and L. W. wrote the
manuscript. F. J. and L.W. supervised the project.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.envint.2020.106244.
References
Knowler, W.C., Barrett-Connor, E., Fowler, S.E., Hamman, R.F., Lachin, J.M., Walker, E.
A., Nathan, D.M., 2002. Diabetes Prevention Program Research, G., Reduction in the
incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med
346 (6), 393–403.
[2] Fertil Steril 108 (3), 2017, 426–441.
Soukas, A.A., Hao, H., Wu, L., 2019. Metformin as Anti-Aging Therapy: Is It for
Everyone? Trends Endocrinol Metab 30 (10), 745–755.
Wu, L., Zhou, B., Oshiro-Rapley, N., Li, M., Paulo, J.A., Webster, C.M., Mou, F.,
Kacergis, M.C., Talkowski, M.E., Carr, C.E., Gygi, S.P., Zheng, B., Soukas, A.A., 2016.
An Ancient, Unied Mechanism for Metformin Growth Inhibition in C. elegans and
Cancer. Cell 167 (7), 1705–1718 e13.
ClinCalc The ClinCalc DrugStats Database.
Yan, J.H., Xiao, Y., Tan, D.Q., Shao, X.T., Wang, Z., Wang, D.G., 2019. Wastewater
analysis reveals spatial pattern in consumption of anti-diabetes drug metformin in
China. Chemosphere 222, 688–695.
Rena, G., Hardie, D.G., Pearson, E.R., 2017. The mechanisms of action of metformin.
Diabetologia 60 (9), 1577–1585.
Caldwell, D.J., D’Aco, V., Davidson, T., Kappler, K., Murray-Smith, R.J., Owen, S.F.,
Robinson, P.F., Simon-Hettich, B., Straub, J.O., Tell, J., 2019. Environmental risk
assessment of metformin and its transformation product guanylurea: II. Occurrence
in surface waters of Europe and the United States and derivation of predicted no-
effect concentrations. Chemosphere 216, 855–865.
Lena Stütz, P.L., 2019. Wolfgang Schulz and Rudi Winzenbacher Identication of
genotoxic transformation products by effect-directed analysis with high-performance
thin-layer chromatography and non-target screening. Journal of Planar
Chromatography 32 (3), 173–182.
MacLaren, R.D., Wisniewski, K., MacLaren, C., 2018. Environmental concentrations of
metformin exposure affect aggressive behavior in the Siamese ghting sh, Betta
splendens. PLoS One 13, (5), e0197259.
Ussery, E., Bridges, K.N., Pandelides, Z., Kirkwood, A.E., Guchardi, J., Holdway, D.,
2019. Developmental and Full-Life Cycle Exposures to Guanylurea and Guanylurea-
Metformin Mixtures Results in Adverse Effects on Japanese Medaka (Oryzias
latipes). Environ Toxicol Chem 38 (5), 1023–1028.
Scheurer, M., Michel, A., Brauch, H.J., Ruck, W., Sacher, F., 2012. Occurrence and fate of
the antidiabetic drug metformin and its metabolite guanylurea in the environment
and during drinking water treatment. Water Res 46 (15), 4790–4802.
Armbruster, D., Happel, O., Scheurer, M., Harms, K., Schmidt, T.C., Brauch, H.J., 2015.
Emerging nitrogenous disinfection byproducts: Transformation of the antidiabetic
drug metformin during chlorine disinfection of water. Water Res 79, 104–118.
Lantagne, D.S., Clasen, T.F., 2012. Use of household water treatment and safe storage
methods in acute emergency response: case study results from Nepal, Indonesia,
Kenya, and Haiti. Environ Sci Technol 46 (20), 11352–11360.
Blair, B.D., Crago, J.P., Hedman, C.J., Klaper, R.D., 2013. Pharmaceuticals and personal
care products found in the Great Lakes above concentrations of environmental
concern. Chemosphere 93 (9), 2116–2123.
Elliott, S.M., Brigham, M.E., Kiesling, R.L., Schoenfuss, H.L., Jorgenson, Z.G., 2018.
Environmentally relevant chemical mixtures of concern in waters of United States
tributaries to the Great Lakes. Integr Environ Assess Manag 14 (4), 509–518.
Florez, J.C., 2017. The pharmacogenetics of metformin. Diabetologia 60 (9), 1648–1655.
Mittal, R., Coopersmith, C.M., 2014. Redening the gut as the motor of critical illness.
Trends Mol Med 20 (4), 214–223.
Kong, L., Kadokami, K., Wang, S., Duong, H.T., Chau, H.T.C., 2015. Monitoring of 1300
organic micro-pollutants in surface waters from Tianjin. North China. Chemosphere
122, 125–130.
de Solla, S.R., Gilroy, E.A., Klinck, J.S., King, L.E., McInnis, R., Struger, J., Backus, S.M.,
Gillis, P.L., 2016. Bioaccumulation of pharmaceuticals and personal care products in
the unionid mussel Lasmigona costata in a river receiving wastewater efuent.
Chemosphere 146, 486–496.
Kalender, A., Selvaraj, A., Kim, S.Y., Gulati, P., Brule, S., Viollet, B., Kemp, B.E.,
Bardeesy, N., Dennis, P., Schlager, J.J., Marette, A., Kozma, S.C., Thomas, G., 2010.
Metformin, independent of AMPK, inhibits mTORC1 in a rag GTPase-dependent
manner. Cell Metab 11 (5), 390–401.
Cabreiro, F., Au, C., Leung, K.Y., Vergara-Irigaray, N., Cocheme, H.M., Noori, T.,
Weinkove, D., Schuster, E., Greene, N.D., Gems, D., 2013. Metformin retards aging in
C. elegans by altering microbial folate and methionine metabolism. Cell 153 (1),
228–239.
Ludman, T., Melemedjian, O.K., 2019. Bortezomib and metformin opposingly regulate
the expression of hypoxia-inducible factor alpha and the consequent development of
chemotherapy-induced painful peripheral neuropathy. Mol Pain 15,
1744806919850043.
Schmidt, S.A., Gukelberger, E., Hermann, M., Fiedler, F., Grossmann, B., Hoinkis, J.,
Ghosh, A., Chatterjee, D., Bundschuh, J., 2016. Pilot study on arsenic removal from
groundwater using a small-scale reverse osmosis system-Towards sustainable
drinking water production. J Hazard Mater 318, 671–678.
Zang, M., Zuccollo, A., Hou, X., Nagata, D., Walsh, K., Herscovitz, H., Brecher, P.,
Ruderman, N.B., Cohen, R.A., 2004. AMP-activated protein kinase is required for the
lipid-lowering effect of metformin in insulin-resistant human HepG2 cells. J Biol
Chem 279 (46), 47898–47905.
Gerets, H.H., Tilmant, K., Gerin, B., Chanteux, H., Depelchin, B.O., Dhalluin, S.,
Atienzar, F.A., 2012. Characterization of primary human hepatocytes, HepG2 cells,
and HepaRG cells at the mRNA level and CYP activity in response to inducers and
their predictivity for the detection of human hepatotoxins. Cell Biol Toxicol 28 (2),
69–87.
R. Zhang et al.

Environment International 146 (2021) 106244
9
O’Brien, P.J., 2014. High-content analysis in toxicology: screening substances for human
toxicity potential, elucidating subcellular mechanisms and in vivo use as
translational safety biomarkers. Basic Clin Pharmacol Toxicol 115 (1), 4–17.
Kiela, P.R., Ghishan, F.K., 2016. Physiology of Intestinal Absorption and Secretion. Best
Pract Res Clin Gastroenterol 30 (2), 145–159.
Hua, G., Wang, C., Pan, Y., Zeng, Z., Lee, S.G., Martin, M.L., Haimovitz-Friedman, A.,
Fuks, Z., Paty, P.B., Kolesnick, R., 2017. Distinct Levels of Radioresistance in Lgr5(+)
Colonic Epithelial Stem Cells versus Lgr5(+) Small Intestinal Stem Cells. Cancer Res
77 (8), 2124–2133.
Clevers, H., 2013. The intestinal crypt, a prototype stem cell compartment. Cell 154 (2),
274–284.
Scholzen, T., Gerdes, J., 2000. The Ki-67 protein: from the known and the unknown.
J Cell Physiol 182 (3), 311–322.
Smith, A.H., Lopipero, P.A., Bates, M.N., Steinmaus, C.M., 2002. Public health. Arsenic
epidemiology and drinking water standards. Science 296 (5576), 2145–2146.
Chiu, C.A., Westerhoff, P., Ghosh, A., 2012. GAC removal of organic nitrogen and other
DBP precursors. J Am Water Works Ass 104 (7), 41–42.
Cuthbertson, A.A., Kimura, S.Y., Liberatore, H.K., Summers, R.S., Knappe, D.R.U.,
Stanford, B.D., Maness, J.C., Mulhern, R.E., Selbes, M., Richardson, S.D., 2019. Does
Granular Activated Carbon with Chlorination Produce Safer Drinking Water? From
Disinfection Byproducts and Total Organic Halogen to Calculated Toxicity. Environ
Sci Technol 53 (10), 5987–5999.
Smith, A.H., Lingas, E.O., Rahman, M., 2000. Contamination of drinking-water by
arsenic in Bangladesh: a public health emergency. Bull World Health Organ 78 (9),
1093–1103.
Diana, M., Felipe-Sotelo, M., Bond, T., 2019. Disinfection byproducts potentially
responsible for the association between chlorinated drinking water and bladder
cancer: A review. Water Res 162, 492–504.
McCreight, L.J., Bailey, C.J., Pearson, E.R., 2016. Metformin and the gastrointestinal
tract. Diabetologia 59 (3), 426–435.
Bailey, C.J., Wilcock, C., Scarpello, J.H., 2008. Metformin and the intestine. Diabetologia
51 (8), 1552–1553.
Poursat, B.A.J., van Spanning, R.J.M., Braster, M., Helmus, R., de Voogt, P., Parsons, J.R.,
2019. Biodegradation of metformin and its transformation product, guanylurea, by
natural and exposed microbial communities. Ecotoxicol Environ Saf 182, 109414.
Ju, F., Beck, K., Yin, X., Maccagnan, A., McArdell, C.S., Singer, H.P., Johnson, D.R.,
Zhang, T., Burgmann, H., 2019. Wastewater treatment plant resistomes are shaped
by bacterial composition, genetic exchange, and upregulated expression in the
efuent microbiomes. ISME J 13 (2), 346–360.
R. Zhang et al.