ww.sciencedirect.com

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0

Available online at w

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

Effect of wastewater colloids on membrane removal ofantibiotic resistance genes

Maria V. Riquelme Breazeal, John T. Novak, Peter J. Vikesland, Amy Pruden*

Via Department of Civil and Environmental Engineering, Virginia Tech, Blacksburg, VA 24061, USA

a r t i c l e i n f o

Article history:

Received 21 June 2012

Received in revised form

21 September 2012

Accepted 22 September 2012

Available online 4 October 2012

Keywords:

Antibiotic resistance genes

Wastewater colloids

Microfiltration

Ultrafiltration

* Corresponding author. Tel.: þ1 540 231 398E-mail addresses: mvpr@vt.edu (M.V. Riq

colostate.edu, apruden@vt.edu (A. Pruden).0043-1354/$ e see front matter ª 2012 Elsevhttp://dx.doi.org/10.1016/j.watres.2012.09.044

a b s t r a c t

Recent studies have demonstrated that wastewater treatment plants (WWTPs) signifi-

cantly alter the magnitude and distribution of antibiotic resistance genes (ARGs) in

receiving environments, indicating that wastewater treatment represents an important

node for limiting ARG dissemination. This study examined the potential for membrane

treatment of microconstituent ARGs and the effect of native wastewater colloids on the

extent of their removal. Plasmids containing vanA (vancomycin) and blaTEM (b-lactam)

ARGs were spiked into three representative WWTP effluents versus a control buffer and

tracked by quantitative polymerase chain reaction through a cascade of microfiltration and

ultrafiltration steps ranging from 0.45 mm to 1 kDa. Significant removal of ARGs was ach-

ieved by membranes of 100 kDa and smaller, and presence of wastewater colloids resulted

in enhanced removal by 10 kDa and 1 kDa membranes. ARG removal was observed to

correlate significantly with the corresponding protein, polysaccharide, and total organic

carbon colloidal fractions. Alumina membranes removed ARGs to a greater extent than

polyvinylidene fluoride membranes of the same pore size (0.1 mm), but only in the presence

of wastewater material. Control studies confirmed that membrane treatment was the

primary mechanism of ARG removal, versus other potential sources of loss. This study

suggests that advanced membrane treatment technology is promising for managing public

health risks of ARGs in wastewater effluents and that removal may even be enhanced by

colloids in real-world wastewaters.

ª 2012 Elsevier Ltd. All rights reserved.

1. Introduction plants can significantly alter the magnitude and distribution of

Antibiotic resistance and, particularly, multidrug resistance, is

an increasingly critical problemaffecting humanhealth (Davies

and Davies, 2010). Globally, antibiotic resistance has been re-

ported tobewidespread(Allenetal., 2010) andontherise (Zhang

et al., 2006). Although most attention has been focused on

nosocomial (i.e., hospital acquired) infections, environmental

factors are also important contributors to the transport and

spread of antibiotic resistance in the community (Larson, 2007).

Recent studies have demonstrated that wastewater treatment

0.uelme Breazeal), jtnov@v

ier Ltd. All rights reserve

antibiotic resistance genes (ARGs) in receiving environments

(Storteboom et al., 2010). Given that antibiotic resistant bacteria

(ARB) and ARGs of clinical concern have been documented to

arise from environmental sources (Martinez, 2008), there is

growing interest in limiting dissemination pathways between

humans and the environment (Baquero et al., 2008). Thus,

wastewater treatment plantsmay represent an important node

for limiting the spread of antibiotic resistance.

The presence of low levels of ARB and ARGs in wastewater

is inevitable and several studies have documented their

t.edu (J.T. Novak), pvikes@vt.edu (P.J. Vikesland), apruden@engr.

d.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0 131

proliferation and persistence in wastewater treatment plants

(Auerbach et al., 2007; Borjesson et al., 2010; Kim et al., 2010;

Szczepanowski et al., 2009; Zhang et al., 2009). Traditionally

the objective of disinfection following water or wastewater

treatment is to kill or inactivate bacterial cells. However, to

address the challenge of antibiotic resistance, DNA removal or

destruction is ideal prior to discharge, reuse, or other appli-

cations. This is because ARGs are typically carried on highly

transmissible DNA elements, such as plasmids and integrons,

which can remain functional and be assimilated by down-

stream bacteria. Because of this ability of ARGs to transcend

their bacterial host, ARGs themselves are considered to be the

primary “contaminant” of concern (Pruden et al., 2006). Once

released, ARGs can persist in the environment (Biyela et al.,

2004; Storteboom et al., 2010), and have even been observed

to establish and proliferate in drinking water biofilms

(Schwartz et al., 2003; Xi et al., 2009). Although the effect of

some disinfection techniques on the fate of ARGs has been

investigated in a few studies (Munir et al., 2011; Oncu et al.,

2011), the fate of these microconstituents after wastewater

disinfection and release is still largely unknown.

The fate of ARGs through membrane removal processes is

of particular interest as membranes are growing in popularity

for enhancing sustainability of water resources and address-

ing concerns for other emerging contaminants (Dolar et al.,

2009, 2012; Munir et al., 2011; Wintgens et al., 2004). The

present study is the first to our knowledge to examine the

potential formembrane removal of ARGs.Membrane filtration

processes are commonly used for the removal of a wide

variety of contaminants, but most would not be expected to

impart significant removal of DNA based on its molecular

properties. However, wastewater effluents are complex in

their chemistries and could significantly affect membrane

removal of DNA. Thus, the focus of this study was on the

potential for the advanced membrane treatment of ARGs

within real-world wastewater effluent matrices.

Of specific interest to this study was the effect of DNA

interactions with wastewater effluent colloidal particles. DNA

is well known to interact with clay minerals and various soil

colloidal particles (Cai et al., 2007, 2006a, 2006b; Crecchio and

Stotzky, 1998; Nguyen et al., 2010; Nguyen and Elimelech,

2007a,b; Romanowski et al., 1991; Saeki et al., 2010). Consid-

ering that significant amounts of suspended colloidal particles

are present in wastewaters, these interactions represent

a promising and unexploited approach for enhanced removal

of ARGs from wastewater. However, most of the available

studies concerning DNAecolloid interactions have been

carried out in simplified or artificial systems (Mitra et al., 2001;

Nguyen and Elimelech, 2007a,b; Romanowski et al., 1991);

thus, there is a need for studies that emphasize the behavior

and fate of DNA in situ. In addition, the relationship between

the removal of ARGecolloid complexes and membrane pore

size has not been investigated in natural systems.

The overall objective of this project was to investigate the

effect of wastewater colloids on the removal of ARGs by

membrane filtration. It was hypothesized that colloidal

interactions with DNA would enhance membrane removal of

ARGs. To test this hypothesis, plasmids containing vanA and

blaTEM ARGs were spiked into three representative WWTP

effluents and a control buffer, and subjected to a cascade of

microfiltration (MF) and ultrafiltration (UF) steps ranging from

0.45 mm to 1 kDa. The vanA and blaTEM ARGs encode resistance

to vancomycin and b-lactam antibiotics (e.g., ampicillin),

respectively, with the former used typically as a last-resort

antibiotic when other treatments fail. Other potential sour-

ces of DNA loss, such as degradation, were also monitored to

verify that observed ARG removal was due to the membrane

treatment and also to examine the potential for wastewater

colloids to protect ARGs from deoxyribonuclease (DNase),

a naturally-occurring DNA-degrading enzyme. The effect of

membrane composition (alumina versus polyvinylidene

fluoride) was compared for the 0.1 mm pore size. Building

a knowledge base of the effect of membrane treatment on

ARGs is critical for enabling comprehensive management

strategies for emerging contaminants inwastewater effluents.

2. Materials and methods

A cascade of MF and UF steps was used to compare the

removal of ARGs from colloid-containing wastewater effluent

versus colloid-free aqueous buffer samples. Effluents from

three representative WWTPs were employed in order to test

the occurrence of DNAecolloid interactions in a complex

environmental matrix, as well as the reproducibility of the

results. Tris buffer (pH 8) was used as colloid-free material in

order tomaintain DNA stability in the controls throughout the

experiment, as low salt concentrations could cause DNA

instability, while high salt concentrations can promote the

sorption of the genetic material to the glass containers. The

key characteristics of the WWTPs selected for this study and

effluent quality are summarized in Table 1. An additional

experiment was carried out to further explore the potential for

DNAecolloid interactions to protect ARGs from degradation

by DNase.

2.1. Plasmid solution preparation

The pCR� 4-TOPO� plasmids (Invitrogen, Carlsbad, CA) con-

taining the ARGs of interestwere used to spike the appropriate

test solutions. This plasmid contains blaTEM and was modified

to contain vanA. A 732 bp portion of the vanA gene was PCR-

amplified from vancomycin-resistant Enterococcus genomic

DNA using previously described primers (Dutka-Malen et al.,

1995). The 25 mL PCR amplification reaction contained 1�Master Taq reaction buffer (5 Prime, Gaithersburg, MD), 1�Taq Master PCR enhancer (5 Prime), 1.5 mM Mg2þ solution (5

Prime), 0.05 mM of each deoxynucleoside triphosphate

(Promega, Madison, WI), 0.2 mM of each primer, 1.75 U Taq

DNA polymerase (5 Prime), 1 mL template DNA, and autoclaved

nanopure water to a final volume of 25 mL. The thermal cycle

consisted of an initial 3 min denaturing step at 95 �C; followed

by 50 cycles of a 30 s denaturing step at 95 �C, a 30 s annealing

step at 54 �C, and a 30 s extension step at 72 �C; and a 7 min

final extension step at 72 �C. The vanA PCR productwas ligated

to the pCR� 4-TOPO� plasmid and cloned using the TOPO TA

cloning kit (Invitrogen), following the manufacturer’s

instructions. Plasmids were extracted using the Qiagen

Plasmid Mega kit (Qiagen, Valencia, CA) and quantified by

qPCR targeting blaTEM and vanA ARGs.

Table 1 e General WWTP and effluent characteristics and effluent sampling dates.

WWTP Capacity(MGD)

Influent composition Key processesb TSSb

(mg/L)pHb Otherb

(mg/L)Collection

date

A 6 Residential, industrial Conventional,a UV disinfection 7.7 7.1 DO (8.1)

BOD (19.6)

4/21/11

B 9 Residential, hospital,

industrial

Conventional,a chlorine disinfection 2 7 DO (9.5) 3/1/11

C 6 Residential, hospital,

high Na2SO4 industrial

Conventional,a nitrification,

chlorine disinfection

7 6.7 BOD (<5) 4/7/11

An additional sampling event occurred on 1/30/12 from all WWTPs, for DNase-degradation test.

a Conventional treatment: primary clarification, aeration, secondary clarification, and disinfection.

b WWTP and effluent characteristics obtained by phone from WWTP operators.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0132

2.2. WWTP effluent collection and initial filtration

Grab samples of final effluent were collected from each of the

test wastewater treatment facilities (WWTPs A, B, or C) at

most 6 h before the experiment was carried out. Effluent

collection occurred during cool (2e15 �C) working week

mornings, avoiding rain or snow events taking place during

the previous 24 h. The effluent was immediately transported

to the lab where it was kept at 4 �C prior to use. As

a pretreatment to eliminate particles above the colloid size

range, the effluent was vacuum-filtered through a 1.2 mm

pore-sizemixed cellulose estermembrane (Millipore, Billerica,

MA) using a sterilized Nalgene� polysulfone filter holder

(Thermo Fisher Scientific, Rochester, NY).

2.3. Cascading filtration experiments

Each of the WWTP effluents was subjected to cascading

filtration experiments and compared with appropriate

controls: (1) Autoclaved 10 mM Tris buffer pH 8 (blank); (2)

Autoclaved 10mMTris pH 8 spikedwith vanA gene-containing

plasmids to a final concentration of 1 � 107 to 1 � 108 plasmid

copies/mL based on blaTEM gene concentrations (no-colloid

control); (3) filtered WWTP effluent (background); and (4)

filtered WWTP effluent spiked with vanA gene-containing

plasmids to a final concentration of 1 � 107 to

1� 108 plasmid copies/mL based on blaTEM gene concentrations

(wastewater colloid treatment). Four 1000-mL flasks were fil-

led with 600 mL of corresponding solution and lightly agitated

for 4e5 h using a Burrell wrist action shaker (Burrell Scientific,

Pittsburg, PA) to enable DNA interactions with wastewater

colloids prior to filtration. 10-mL subsamples were collected

from each of the four flasks during this stage to test for

alternative sources of DNA loss.

Sequential cascading filtration steps were carried out for

each of the four flask contents. First, 500 mL were filtered

through the membrane with the largest pore size (0.45 mm-

pore size membrane). Next, 10-mL sub-samples of the filtrate

were pipetted into previously baked 30-mL Qorpak (Bridge-

ville, PA) clear graduated glass bottles and immediately frozen

at �80 �C for subsequent lyophilization and DNA extraction.

Filtrate sub-samples originating from the unspiked WWTP

effluent were also collected for colloid characterization and

stored at 4 �C. The remaining water/buffer filtrate was then

filtered through the subsequent membrane (0.1 mm-pore size).

The processwas repeated for each of the filtrates, applying the

membranes in order of decreasing pore size:

0.45 mm/ 0.1 mm/ 100 kDa/ 10 kDa/ 1 kDa. Tominimize

carryover, the flask contents were processed in the following

order: (1) buffer blank, (2) unspiked background wastewater,

(3) spiked no-colloid control buffer, (4) spiked wastewater

containing colloids. The latter was processed after the 4-h

period allotted for DNAecolloid interactions to occur.

MF was carried out using 0.45 mm and 0.1 mm pore size

Durapore hydrophilic polyvinylidene fluoride (PVDF)

membranes (Millipore, Billerica, MA), and a Nalgene� poly-

sulfone filter holder (Thermo Fisher Scientific, Rochester, NY).

To compare the effect of membrane type, Anodisc 47 alumina

0.1 mm pore size membranes (Whatman GmbH, Germany)

were compared to the Durapore membrane of the same pore

size. UF was carried out using 100 kDa, 10 kDa, and 1 kDa

MWCO Ultracel regenerated cellulose membranes (Millipore,

Billerica, MA) and an 8200 Ultrafiltration Stirred Cell (Milli-

pore). UF filtrates flowed into baked glass flasks shielded by

baked aluminum foil. Because the UF membranes were reus-

able, they were cleansed between usages using DNA Away

(Molecular Bio Products, San Diego, CA) followed by a thor-

ough rinsing step that included filtering water through the

membrane for at least 10 min. A list of the membrane types

and pore sizes is provided in Table S1.

2.4. DNA stability in pretreated wastewater and buffer

To determine the extent to which membrane filtration

accounted for removal of ARGs in the above experiments,

changes in gene concentrations in pretreated wastewater and

buffer were monitored over the time allotted for DNAecolloid

interactions prior to filtration. A 10-mL sample was collected

immediately after plasmid spiking, and 2e3 additional

samples were collected during the following 4e5 h. Gene

fractions at times t > 0 were calculated with respect to the

initial concentration measured at t ¼ 0 h.

2.5. DNase-degradation experiment

The potential for colloideDNA interactions to protect ARGs

from degradation was investigated using DNase-degradation

experiments. All three WWTP effluents were sampled on the

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0 133

same day and compared in parallel to the colloid-free Tris

buffer control. 500mL of either Tris buffer (pH 8) or the filtered

effluents were collected in fluorinated HDPE Nalgene bottles

(Thermo Fisher Scientific, Rochester, NY) and spiked to

a concentration of 1� 107 plasmids/mL, and placed in a shaking

incubator at 125 rpm and room temperature for 4 h to allow

DNAecolloid interactions. Sub-samples were collected at 0, 2,

and 4 h to test for background DNA losses. After 4 h, 200 mL of

the contents of each bottle (i.e., buffer orWWTP effluent) were

split into two new bottles (100 mL each), one of which was

spiked with 10.2 U of RQ1 RNase-free DNase (Promega) and

subject to 2.5 mM MgCl2 and 0.5 mM CaCl2 at 37 �C, and the

other only subject to the same salt and temperature conditions

used to activate theDNase. The bottleswere incubated at 37 �Cand shaken at 125 rpm. This approach allowed direct

comparison of the effect of the DNase versus the background

salt and temperature conditions required to activate the

DNase. 10 mL sub-samples were collected after light manual

vortexing, immediately after salt and/or DNase addition (i.e.,

shortly after hour 4), and at 5 and 7 h from the beginning of the

experiment. All collected samples were immediately stored at

�80 �C for further analysis.

2.6. DNA concentration and extraction

All samples were concentrated by lyophilization using

a FreeZone Plus 2.5 L Benchtop Freeze Dry System (Labconco,

Kansas City, Missouri). Isolation of DNA originating from

WWTP effluent samples was carried out using the FastDNA

Spin Kit for Soil (MP Biomedicals, Solon, OH) as recommended

in the kit manual. Isolation of DNA originating from buffer

samples was carried out using the PowerClean DNA Cleanup

Kit (MoBio Laboratories, Carlsbad, CA).

2.7. DNA quantification by qPCR

Quantification of vanA and blaTEM was carried out using real-

time quantitative polymerase chain reaction (qPCR) on

a CFX96 Touch� Real-Time PCR Detection System (Bio-Rad,

Hercules, CA). 1:10 DNA extract dilutions were analyzed in

triplicate reactions containing 40 nM of each primer, 1�SsoFast� EvaGreen� Supermix (Bio-Rad), 1 mL template DNA,

and sterile nanopure to a final volume of 10 mL. The primers

used for amplification of the vanA gene were vanAstF

(Bockelmann et al., 2009) and vana3RP (Volkmann et al., 2004).

The blaTEM gene was amplified using previously described

primers (Bibbal et al., 2007). The temperature programs con-

sisted of a 95 �C initial denaturing step for 30 s; 40 cycles of

denaturing at 95 �C for 5 s, annealing and extension at 60 �Cfor 5 s, and an additional extension step at 72 �C for 5 s for

vanA amplifications only. A melt curve was incorporated in

every reaction to verify specificity by ramping the tempera-

ture from 65 �C to 95 �C by 0.5 �C every 5 s. A calibration curve

was constructed for each set of reactions using at least five

standards. The qPCR standards were prepared by serially

diluting M13 PCR products obtained from cloned positive

controls (vanA or blaTEM). Concentrations were determined by

gel densitometry and converted to gene copies using the

relationship described by Pei et al. (2006). A limit of detection

(LOD) was calculated for each gene by averaging gene

concentrations in 10 reaction blanks and multiplying that

value by 3. The background gene concentrations detected in

each analytical blank was subtracted from the measured

sample concentration. Each individual measurement greater

than the detection limit was then multiplied by the dilution

factor (10) and the extraction volume, and divided by the

volume of each sample in microliters (1 � 104 mL) to obtain

the final concentration in gene copies/mL of each sample.

The three analytical measurements per sample were then

averaged to obtain the estimated concentration for that

sample.

2.8. Colloid characterization

Non-purgeable organic carbon concentrations weremeasured

using the Shimadzu TOC-VCSN analyzer (Shimadzu Scientific

Instruments, Columbia, MD) and potassium hydrogen

phthalate (KHP) as the standard. Polysaccharide concentra-

tions were measured by the phenolesulfuric acid method

(Dubois et al., 1956) using dextrose as standard. Protein

concentrations were measured by the Lowry method (Lowry

et al., 1977) using bovine serum albumin (BSA) as the stan-

dard. Briefly, five reagents were prepared as followse Reagent

1: 143 mM NaOH and 270 mM Na2CO3; Reagent 2: 57 mM

CuSO4; Reagent 3: 124 mM Na-tartrate; Reagent 4: mix of

reagents 1, 2, and 3 at a ratio of 100:1:1 (made daily); and

Reagent 5 contained folin ciocalteus reagent diluted 5:6 with

distilled water. 1-mL samples were added by sequential

addition and mixing of Reagents 4 and 5, respectively. The

samples containing the reagents were incubated for 45 min at

room temperature before measuring the absorbance at

750 nm.

2.9. Statistical analyses

Statistical analyseswere performed using R statistics software

(www.r-project.org). Statistical differences between the

membrane-removal of ARGs in the presence versus the

absence of colloids were calculated using a two-way analysis

of variance (ANOVA). The two-way ANOVAwas applied to the

log of the filtrate gene fraction from spiked WWTP effluent or

spiked buffer as the dependent variable. The filtrate gene

fraction was calculated as C/C0, where C was the gene

concentration in the filtrate and C0 was the average gene

concentration in the unfiltered samples used to test for

degradation. The presence/absence of colloids and the

membrane pore size or type were independent variables. The

analyses were repeated individually for each of the genes

monitored. To test the significance of the DNA degradation in

the water or buffer samples prior to treatment (i.e., prior to

filtration or application of DNase-degrading conditions),

sampling times after the addition of the plasmid spike were

binned as A (1e2 h), B (2.5e3 h), and C (4e5 h). A one-way

ANOVA test was applied to the log of the gene fractions

present in the water or buffer samples for each bin. Statistical

significance was set at a ¼ 0.05 ( p < 0.05). Correlation analysis

between the filtrate fraction of each of the colloidal compo-

nents and the logarithm of the filtrate gene fraction was

conducted using R, applying a one-sided Pearson correlation

test.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0134

3. Results

3.1. Membrane removal of ARGs and effect ofwastewater colloids

Sequential filtration across decreasing membrane pore sizes

resulted in increasing removal of plasmid-associated ARGs

(Fig. 1). Specifically, filtration through the 0.45 and 0.1 mm

pore-size microfiltration membranes resulted in less than

1-log removal of ARGs, either in the presence or absence of

colloidal material. On the other hand, filtration through the

100, 10, and 1 kDa membranes resulted in an average of 0.9,

3.6, and 4.2 e log reductions across all no-colloid controls and

ARGs, and 1.7, 4.9, and >5.9 (most concentrations were below

the LOD) e log reductions across all WWTP effluents and

ARGs. A summary of the log reductions achieved across each

membrane and for each WWTP effluent is shown in Table 2.

As expected, membrane pore size had a significant effect on

the removal of ARGs ( p < 0.001). Further, the presence of

colloidal material in the aqueous matrix was found to have

a significant effect on the removal of blaTEM ( p ¼ 0.004) and

Fig. 1 e Effect of membrane pore size on ARG removal from spike

represent standard deviation of analytical triplicates. 1 kDa dat

WWTP effluent source.

vanA ARGs ( p ¼ 0.002). Additional analysis revealed that as

membrane pore size decreased, the influence of colloids on

the removal of ARGs became more apparent (Table 3).

Specifically, the effect of the presence of colloidal material in

the removal of blaTEM and vanA ARGs became significant at

10 kDa ( p < 0.05 and <0.01, respectively), and reached

a maximum at 1 kDa membrane pore size ( p < 0.001 and

<0.01, respectively).

3.2. Accounting for other sources of DNA loss

blaTEM and vanA ARGs were monitored over 4e5 h to account

for alternative sources of DNA loss, such as degradation

(Fig. 2). ARG losses were found not to be significant ( p > 0.05)

in either wastewater or buffer samples, suggesting that any

observed DNA removal during filtration experiments was

attributable to retention by the membranes either by size

exclusion of the DNA itself, exclusion of DNAecolloid

complexes, or interactions with the membrane material. The

gene concentrations measured in the unspiked buffer blank

confirmed absence of contamination during the experiments.

dWWTP effluents and no-colloid buffer controls. Error bars

a not shown was below LOD. Letters on the left indicate

Table 2 e Log reduction of plasmid-associated ARGs across all WWTPs and ultrafiltration membranes.

WWTP Log reduction of blaTEM gene Log reduction of vanA gene

100 kDa 10 kDa 1 kDa 100 kDa 10 kDa 1 kDa

A 1.1 4.2 a 1.2 4.3 a

B 2.4 4.9 a 2.3 4.7 a

C 1.6 5.5 a 1.7 5.8 5.9

No-colloid

control (average)

1 3.7 4.2 0.9 3.5 4.2

a Gene concentrations were below the LOD.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0 135

ARG concentrations in unspikedWWTP effluent sampleswere

tracked to monitor existing gene concentrations and, if

possible, to characterize the behavior of native sorbed DNA.

Background concentrations of blaTEM and vanAwere generally

very low and near the LOD. Specifically, average blaTEM and

vanA concentrations in unspiked 1.2 mm-filtered effluents

were <LOD and 1.1 gene copies/mL (WWTP A), 31.2 and

7.4 gene copies/mL (WWTP B), and <LOD and 2.2 (WWTP C),

respectively. These averages were determined based on the

ARG concentrations in the unspiked background samples that

were monitored for alternative sources of degradation.

3.3. Effect of colloids on degradation by DNase

The blaTEM and vanA gene fractions measured in the aqueous

samples before and after the application of DNase and DNase-

activating conditions (T and salt) are shown in Fig. S1. Under

the conditions of these experiments, DNA interactions with

wastewater colloids were not found to protect ARGs from

DNase degradation. Conditions were confirmed to be

amenable to DNase activity, as apparent by the significant

difference in ARG concentrations when DNase was present or

absent. However, no significant difference was observed

between the end point concentrations in the presence/

absence of colloidal material.

Table 3 e Relative influence of colloids and membranepore size on the removal of ARGs.

Membranepore size

Gene Colloids Membrane

0.1 mm blaTEM * e

vanA e e

100 kDa blaTEM e ***

vanA e ***

10 kDa blaTEM * ***

vanA ** ***

1 kDa blaTEM *** *

vanA ** *

Symbols represent significance of p-values obtained from two-way

ANOVA test: p > 0.05 (e), p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). The

two-way ANOVA was applied to the log of the filtrate gene fraction

of adjacentmembranes. The filtrate gene fractionwas calculated as

C/C0, where Cwas the gene concentration in the filtrate and C0 was

the average gene concentration in the unfiltered samples used to

test for degradation.

3.4. Colloid characterization

TOC, protein, and polysaccharidemeasurements are provided

in Fig. S2. Results of one-sided Pearson correlation analysis of

the filtrate fraction of each of the colloidal components and

the logarithm of the filtrate gene fraction are shown in Fig. S3.

ARG fractions in the filtrates were found to significantly

correlate with all three colloidal components. The highest

correlation was observed between ARGs and proteins [r ¼ 0.78

and 0.81 for blaTEM and vanA ( p < 0.001)]. Pearson correlation

coefficients were slightly strengthened when considering the

sum of the protein and polysaccharide components [r ¼ 0.79

and 0.82 for blaTEM and vanA, respectively ( p < 0.001)].

3.5. Comparison between PVDF and aluminamembranes

A comparison of ARG removal achieved by MF (0.1 mm-pore

size) hydrophilic PVDF Durapore� (Millipore, Billerica, MA)

versus alumina Anodisc 47� (Whatman GmbH, Germany)

filtration of the WWTP effluents is presented in Fig. 3. There

was less than 1-log removal of plasmid-associated ARGs from

the no-colloid control by either membrane. While less than 1-

log ARG removal was also observed across all wastewater

samples and ARGs when the Durapore� PVDF membrane was

applied, significant removal (3.2 log; p < 0.05) was observed

when the alumina Anodisc 47� membrane was applied. The

blaTEM and vanA log removals achieved using the alumina

membranewere 3.2 and 3.1 forWWTPA, 2.2 and 2.1 forWWTP

B, and 4.5 and 4.2 for WWTP C, respectively. The combined

effect of colloids and the alumina membrane was also found

to be significant ( p< 0.05). There was no significant difference

in the removal of TOC, protein, or polysaccharides by the two

membrane types ( p > 0.05).

4. Discussion

The results of this study suggest that membrane treatment is

a promising technology for the removal of ARGs from treated

wastewater effluents. An encouraging findingwas that greater

removal was achieved than expected. While the molecular

weight of the spiked plasmid was w2600 kDa, in reality, the

practical effective size of DNA for membrane retention is

smaller than predicted by molecular weight. This is because

DNA is a long, thin, and flexible molecule that could hypo-

thetically squeeze length-wise through a membrane pore

Fig. 2 e Stability of vanA and blaTEM ARGs in spiked WWTP effluents and no-colloid buffer controls. Error bars represent

standard deviation of analytical triplicates, but are smaller than the symbols in most cases. Values in parentheses indicate

minimum (in gene copies/mL), maximum (in gene copies/mL), and coefficient of variation. Letters on the left indicate WWTP

effluent source.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0136

(Marko et al., 2011; Travers, 2004). Correspondingly, the 1 kDa

membranes did not completely exclude passage of the

plasmid in the spiked buffer control (Fig. 1). The overall find-

ings highlight the importance of DNA conformation and

behavior, which can vary widely depending on water chem-

istry and composition.

To confirm that DNA removal was actually a result of

membrane filtration, ARG concentrations were monitored to

account for alternative sources of DNA loss, such as biodeg-

radation or sorption to the glass container. Interestingly, the

ARG concentrations did not change significantly, which

supports the conclusion that removal was driven by the

membranes. These control experiments also bring to light

the overall persistence of ARGs in treated wastewater

effluent. This result was somewhat unforeseen as the

extracellular DNA was expected to be more susceptible to

biodegradation given the numerous enzymes and other

cellular components present within a wastewater effluent

matrix. This observation further emphasizes the potential

importance of advanced treatment of wastewater effluents to

remove ARGs that can be transported and protected by larger

molecules.

Previous studies have noted the persistence of ARGs under

environmentally-relevant conditions, highlighting the poten-

tial for DNA protection through interactions with environ-

mental substrates. Specifically, the presence of ARB and ARGs,

including vanA, has been reported in drinking water sources,

treated drinking water, and treated wastewater (Auerbach

et al., 2007; Pruden et al., 2006; Schwartz et al., 2003; Xi et al.,

2009; Zhang et al., 2009; LaPara et al., 2011). The present

study found trace levels of vanA in all three WWTP effluents.

In addition, several studies have previously shown that water

disinfection processes such as chlorination can select for ARB

(Armstrong et al., 1982; Murray et al., 1984; Shrivastava et al.,

2004).

Previous studies have shown that DNA can be protected

against DNase degradationwhen it is bound to othermaterials

(Table 4). Contrary to these studies, the wastewater colloid-

associated ARGs remained susceptible to DNase degradation.

It is worth noting that the DNase degradation experiments in

Fig. 3 e Comparison of DNA removal between 0.1 mm PVDF

and alumina membranes. Error bars associated with

WWTPs represent standard deviations of analytical

measurements. Error bars associated with the control

represent standard deviations of experimental

measurements done independently in parallel with the

experiments involving each of the WWTPs.

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0 137

this study were performed under conditions of excess

enzyme, and it is not certainwhether the DNAwas completely

sorbed to suspended material in the wastewater. Further

investigations of varying concentrations of DNA and DNase,

including environmentally-relevant concentrations, within

the wastewater matrix would be of interest.

Importantly, this study demonstrated that membrane

removal of ARGs is actually enhanced in the presence of

Table 4 e Summary of materials observed to interact with andtransformation ability of sorbed DNA.

Material Sorption Protection againstDNase degradation

T

Organic and

inorganic clays

þ þ

Soil colloids

and particles

þ þ

Silica/sand þ þ

Charcoal þ a

Humic acids þ þNOM þ a

a Data not available.

wastewater. Filtration experiment results revealed a signifi-

cant effect of 1.2 mm-filtered wastewater effluent matrix on

the removal of blaTEM and vanA genes present on the spiked

plasmids. This effect is presumably a result of interactions

with colloidal material present in the effluent, and is sup-

ported by the correlations observedwith the TOC, protein, and

polysaccharide fractions present in the filtrates. Previous

reports provide evidence of interactions between DNA and

organic and inorganic materials that are likely present in

wastewaters and also in natural waters (Table 4). However,

these studies were carried out only in simplified artificial

systems. Thus, the results of the present study provide insight

into the potential for membranes to remove ARGs from actual

wastewater effluents. Correlations with TOC, proteins, and

polysaccharides highlight the possibility that the interactions

taking place in the wastewater are complex and involve

agglomeration with various materials rather than the simple

interaction between DNA and discrete particles. The overall

results indicate that wastewater colloideARG interactions

could facilitate ARG removal by larger pore size membranes,

thus reducing the energy requirement for membrane filtra-

tion. The findings are congruent with those of a recent study

that demonstrated significantly higher ARB and ARG removals

from wastewater treated with a membrane bioreactor

compared to conventional treatment (Munir et al., 2011). This

finding further substantiates the hypothesis of DNAecolloid

interactions because, based on the molecular properties of

DNA, a conventional MBR system would not be expected to

significantly remove extracellular ARGs.

Effluents from three WWTPs were collected and used as

natural colloid-containingmedia in this study.According to the

literature, DNA has been found to bind to a variety of organic

and inorganic materials including different soil colloidal

particles, different types of clay, and NOM (Cai et al., 2007,

2006a, 2006b; Crecchio and Stotzky, 1998; Nguyen et al., 2010;

Nguyen and Elimelech, 2007a,b). However, DNA sorption

behavior not only varieswith type of sorbent, but alsowith pH,

ionic strength, type of cations present and DNA conformation

(Cai et al., 2006a; Mitra et al., 2001; Nguyen et al., 2010; Nguyen

and Elimelech, 2007b; Saeki et al., 2010). For this reason, it is

not possible to generate adsorption isotherms that would be

consistent across effluents. Even within one WWTP effluent,

/or protect DNA against enzymatic degradation, and

ransformationability

References

þ Cai et al. (2007), Cai et al. (2006a),

Cai et al. (2006b), Khanna and Stotzky (1992)

þ Cai et al. (2007), Cai et al. (2006a),

Cai et al. (2006b), Saeki et al. (2010)a Lorenz and Wackernagel (1987), Mitra et al.

(2001), Nguyen et al. (2010), Nguyen and

Elimelech (2007b), Romanowski et al. (1991)a Mitra et al. (2001)

þ Crecchio and Stotzky (1998)a Nguyen et al. (2010)

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0138

the composition would likely change seasonally or even daily.

Based on adsorption times reported in the literature (Mitra

et al., 2001) and on typical WWTP retention times, a 4-h sorp-

tion timewaschosen toallow theDNAto interactwith colloidal

material in the wastewater effluent. Recent studies have indi-

cated trends of DNA transmission through UF membranes at

different ionic strengths, pH, andpressure (Arkhangelsky et al.,

2008; Latulippe and Zydney, 2008). Thus, the ionic strength of

the no-colloid control and the working pressure for all UF

membranes were kept constant throughout this study.

Comparisonof themembrane compositionprovided further

insight intomechanisms of ARG removal. The removal of ARGs

was most significantly enhanced by the combination of the

alumina membrane and the presence of wastewater material.

Interestingly, reductions observed in TOC, protein, and poly-

saccharidecolloidal components followingaluminamembrane

filtration (Fig. S2)were not significant ( p> 0.05).WhileDNAhas

been previously reported to interact with alumina and

aluminumspeciesat intermediatepH(Karliketal., 1980;Maand

Yeung, 2010), the removal of DNA in the buffer control was

negligible (Fig. 3). Although this finding suggests that DNA-

ealumina interactionsmay not be a significant factor affecting

the enhanced DNA removal by alumina membranes, it is also

possible that DNA retention was enhanced by DNAealumina

interactions within the wastewater matrix, but perhaps not

necessarily by the presence of colloidal material. One of the

reported attributes of the Anodisc aluminamembrane is that it

is made of a non-deformable material with no crossover

between individual pores. It is possible that although the

average reported pore size of the PVDF membrane was equiv-

alent (0.1 mm), either the actual range of pore sizes was higher

and reached pore sizes above 0.1 mm, or there were crossovers

among the pores. It is also possible that the PVDF membrane

was slightly deformable, thus allowing the transmission of

larger particles. Theseobservationshighlight the importanceof

investigating thedifferencesamongpotentialmembranesprior

to wider-scale use, as certain membranes could have the

potential to remove a wider range of contaminants under

similar or even lower energy demands.

Fouling is an important factor to consider for any

membrane treatment technology. Membranes subjected to

colloids andwastewater effluents are expected to foul as these

materials build up at the membrane surface (Ang et al., 2011).

The present study was purposefully not designed to examine

the implications of membrane fouling on ARG removal;

however, based upon the observed enhancement in ARG

removal in the presence of wastewater colloids it can be

hypothesized that ARGs will be more efficiently removed by

membranes that are partially fouled with organic material

than pristine membranes alone. This latter hypothesis,

however, requires additional experimental testing under

a variety of operating conditions.

5. Conclusions

Overall, this study demonstrates that membranes are

a promising treatment technology for the removal of ARGs

fromwastewater effluents. Fourmain findings are highlighted

from this study:

� ARGs could be effectively removed by membrane filtration,

with best performance observed for 10 kDamembranes (UF)

and smaller. Membrane removal of ARGs was better than

expected based on the molecular properties of DNA.

� ARGs spiked into the pre-filteredwastewater effluentmatrix

remained stable during the 4-h monitoring period, indi-

cating that ARGs can persist in wastewater effluents and

supporting the need to investigate treatment technologies.

� Associations of ARGswith natural colloidalmaterial present

in wastewater effluent actually resulted in enhanced

removal of ARGs. These results are encouraging, and

suggest that cost and energy requirements for ARG treat-

ment may be less than would be predicted by molecular

properties alone.

� Membrane material and other characteristics also have an

impact on ARG removal and should be optimized in future

studies.

Membrane treatments are likely to grow in their imple-

mentation for advancing water reuse and the results provide

important baseline information for the development of

a comprehensive water sustainability strategy. In particular,

membranes have been demonstrated to be effective for the

removal of other emerging contaminants, such as pharma-

ceuticals and personal care products. Benefits of membrane

treatment could be extended to ARGs as WWTPs consider

upgrades. Considering the growing problem of antibiotic

resistance and the demonstrated links between WWTPs and

ARGs in receiving environments, incorporating membrane

treatment of WWTP effluents may be advantageous for

minimizing potential risk to public health.

Acknowledgments

Financial support for this project was provided by the Water

Environment Research Foundation Project INFR8SG09, by the

NSF CBET CAREER award #0547342, and by the Institute for

Critical Technology and Applied Science award TSTS-11-26.

The findings do not reflect the opinions of the sponsors.

Special thanks to Dr. Dan Gallagher at Virginia Tech for his

insights regarding statistical analyses and Dr. Tony Fane of

the Singapore Membrane Technology Centre for consultation

in experimental design.

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.watres.2012.09.044.

r e f e r e n c e s

Allen, H.K., Donato, J., Wang, H.H., Cloud-Hansen, K.A., Davies, J.,Handelsman, J., 2010. Call of the wild: antibiotic resistancegenes in natural environments. Nature Reviews Microbiology8 (4), 251e259.

Ang, W.S., Tiraferri, A., Chen, K.L., Elimelech, M., 2011. Foulingand cleaning of RO membranes fouled by mixtures of organic

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0 139

foulants simulating wastewater effluent. Journal of MembraneScience 376 (1e2), 196e206.

Arkhangelsky, E., Steubing, B., Ben-Dov, E., Kushmaro, A.,Gitis, V., 2008. Influence of pH and ionic strength ontransmission of plasmid DNA through ultrafiltrationmembranes. Desalination 227 (1e3), 111e119.

Armstrong, J.L., Calomiris, J.J., Seidler, R.J., 1982. Selection ofantibiotic-resistant standard plate-count bacteria duringwater treatment. Applied and Environmental Microbiology 44(2), 308e316.

Auerbach, E.A., Seyfried, E.E., McMahon, K.D., 2007. Tetracyclineresistance genes in activated sludge wastewater treatmentplants. Water Research 41 (5), 1143e1151.

Baquero, F., Martinez, J.L., Canton, R., 2008. Antibiotics andantibiotic resistance in water environments. Current Opinionin Biotechnology 19 (3), 260e265.

Bibbal, D., Dupouy, V., Ferre, J.P., Toutain, P.L., Fayet, O.,Prere, M.F., Bousquet-Melou, A., 2007. Impact of threeampicillin dosage regimens on selection of ampicillinresistance in Enterobacteriaceae and excretion of blaTEM genesin swine feces. Applied and Environmental Microbiology 73(15), 4785e4790.

Biyela, P.T., Lin, J., Bezuidenhout, C.C., 2004. The role of aquaticecosystems as reservoirs of antibiotic resistant bacteria andantibiotic resistance genes. Water Science and Technology 50(1), 45e50.

Bockelmann, U., Dorries, H.H., Ayuso-Gabella, M.N., deMarcay, M.S., Tandoi, V., Levantesi, C., Masciopinto, C., VanHoutte, E., Szewzyk, U., Wintgens, T., Grohmann, E., 2009.Quantitative PCR monitoring of antibiotic resistance genesand bacterial pathogens in three European artificialgroundwater recharge systems. Applied and EnvironmentalMicrobiology 75 (1), 154e163.

Borjesson, S., Matussek, A., Melin, S., Lofgren, S., Lindgren, P.E.,2010. Methicillin-resistant Staphylococcus aureus (MRSA) inmunicipal wastewater: an uncharted threat? Journal ofApplied Microbiology 108 (4), 1244e1251.

Cai, P., Huang, Q., Chen, W., Zhang, D., Wang, K., Jiang, D.,Liang, W., 2007. Soil colloids-bound plasmid DNA: effect ontransformation of E. coli and resistance to DNase Idegradation. Soil Biology & Biochemistry 39 (5), 1007e1013.

Cai, P., Huang, Q., Zhang, X., Chen, H., 2006a. Adsorption of DNAon clay minerals and various colloidal particles from anAlfisol. Soil Biology & Biochemistry 38 (3), 471e476.

Cai, P., Huang, Q.Y., Zhang, X.W., 2006b. Interactions of DNA withclay minerals and soil colloidal particles and protectionagainst degradation by DNase. Environmental Science &Technology 40 (9), 2971e2976.

Crecchio, C., Stotzky, G., 1998. Binding of DNA on humic acids:effect on transformation of Bacillus subtilis and resistance toDNase. Soil Biology & Biochemistry 30 (8e9), 1061e1067.

Davies, J., Davies, D., 2010. Origins and evolution of antibioticresistance. Microbiology and Molecular Biology Reviews 74 (3),417e433.

Dolar, D., Kosutic, K., Pavlovic, D.M., Kunst, B., 2009. Removal ofemerging contaminants of industrial origin by NF/RO e a pilotscale study. Desalination and Water Treatment 6 (1e3),197e203.

Dolar, D., Zokic, T.I., Kosutic, K., Asperger, D., Pavlovic, D.M., 2012.RO/NF membrane treatment of veterinary pharmaceuticalwastewater: comparison of results obtained on a laboratoryand a pilot scale. Environmental Science and PollutionResearch 19 (4), 1033e1042.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F.,1956. Colorimetric method for determination of sugars andrelated substances. Analytical Chemistry 28 (3), 350e356.

Dutka-Malen, S., Evers, S., Courvalin, P., 1995. Detection ofglycopeptide resistance genotypes and identification to the

species level of clinically relevant Enterococci by PCR. Journal ofClinical Microbiology 33 (1), 24e27.

Karlik, S.J., Eichhorn, G.L., Lewis, P.N., Crapper, D.R., 1980.Interaction of aluminum species with deoxyribonucleic-acid.Biochemistry 19 (26), 5991e5998.

Khanna, M., Stotzky, G., 1992. Transformation of Bacillus subtilisby DNA bound on montmorillonite and effect of DNase on thetransforming ability of bound DNA. Applied andEnvironmental Microbiology 58 (6), 1930e1939.

Kim, S., Park, H., Chandran, K., 2010. Propensity of activatedsludge to amplify or attenuate tetracycline resistance genesand tetracycline resistant bacteria: a mathematical modelingapproach. Chemosphere 78 (9), 1071e1077.

LaPara, T.M., Burch, T.R., McNamara, P.J., Tan, D.T., Yan, M.,Eichmiller, J.J., 2011. Tertiary-treated municipal wastewater isa significant point source of antibiotic resistance genes intoDuluth-Superior harbor. Environmental Science & Technology45 (22), 9543e9549.

Larson, E., 2007. Community factors in the development ofantibiotic resistance. Annual Review of Public Health 28,435e447.

Latulippe, D.R., Zydney, A.L., 2008. Salt-induced changes inplasmid DNA transmission through ultrafiltrationmembranes. Biotechnology and Bioengineering 99 (2),390e398.

Lorenz, M.G., Wackernagel, W., 1987. Adsorption of DNA to sandand variable degradation rates of adsorbed DNA. Applied andEnvironmental Microbiology 53 (12), 2948e2952.

Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1977.Protein measurement with Folin phenol reagent. CurrentContents (1), 7.

Ma, C.B., Yeung, E.S., 2010. Entrapment of individual DNAmolecules and nanoparticles in porous alumina membranes.Analytical Chemistry 82 (2), 654e657.

Marko, A., Denysenkov, V., Margraft, D., Cekan, P., Schiemann, O.,Sigurdsson, S.T., Prisneet, T.F., 2011. Conformational flexibilityof DNA. Journal of the American Chemical Society 133 (34),13375e13379.

Martinez, J.L., 2008. Antibiotics and antibiotic resistance genes innatural environments. Science 321 (5887), 365e367.

Mitra, A., Chakraborty, P., Chattoraj, D.K., 2001. Kinetics ofadsorption of DNA at solideliquid interfaces. Journal of theIndian Chemical Society 78 (10e12), 689e696.

Munir, M., Wong, K., Xagoraraki, I., 2011. Release of antibioticresistant bacteria and genes in the effluent and biosolids offive wastewater utilities in Michigan. Water Research 45 (2),681e693.

Murray, G.E., Tobin, R.S., Junkins, B., Kushner, D.J., 1984. Effect ofchlorination on antibiotic-resistance profiles of sewage-related bacteria. Applied and Environmental Microbiology 48(1), 73e77.

Nguyen, T.H., Chen, K.L., Elimelech, M., 2010. Adsorption kineticsand reversibility of linear plasmid DNA on silica surfaces:influence of alkaline earth and transition metal ions.Biomacromolecules 11 (5), 1225e1230.

Nguyen, T.H., Elimelech, M., 2007a. Adsorption of plasmid DNA toa natural organic matter-coated silica surface: kinetics,conformation, and reversibility. Langmuir 23 (6), 3273e3279.

Nguyen, T.H., Elimelech, M., 2007b. Plasmid DNA adsorption onsilica: kinetics and conformational changes in monovalentand divalent salts. Biomacromolecules 8 (1), 24e32.

Oncu, N.B., Menceloglu, Y.Z., Balcioglu, I.A., 2011. Comparison ofthe effectiveness of chlorine, ozone, and photocatalyticdisinfection in reducing the risk of antibiotic resistancepollution. Journal of Advanced Oxidation Technologies 14 (2),196e203.

Pei, R.T., Kim, S.C., Carlson, K.H., Pruden, A., 2006. Effect of riverlandscape on the sediment concentrations of antibiotics and

wat e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 3 0e1 4 0140

corresponding antibiotic resistance genes (ARG). WaterResearch 40 (12), 2427e2435.

Pruden, A., Pei, R.T., Storteboom, H., Carlson, K.H., 2006.Antibiotic resistance genes as emerging contaminants: studiesin northern Colorado. Environmental Science & Technology 40(23), 7445e7450.

Romanowski,G., Lorenz,M.G.,Wackernagel,W., 1991.AdsorptionofplasmidDNA tomineral surfaces andprotection againstDNaseI.Applied and Environmental Microbiology 57 (4), 1057e1061.

Saeki, K., Kunito, T., Sakai, M., 2010. Effects of pH, ionic strength,and solutes on DNA adsorption by andosols. Biology andFertility of Soils 46 (5), 531e535.

Schwartz, T., Kohnen, W., Jansen, B., Obst, U., 2003. Detection ofantibiotic-resistant bacteria and their resistance genes inwastewater, surface water, and drinking water biofilms. FEMSMicrobiology Ecology 43 (3), 325e335.

Shrivastava, R., Upreti, R.K., Jain, S.R., Prasad, K.N., Seth, P.K.,Chaturvedi, U.C., 2004. Suboptimal chlorine treatment ofdrinking water leads to selection of multidrug-resistantPseudomonas aeruginosa. Ecotoxicology and EnvironmentalSafety 58 (2), 277e283.

Storteboom, H., Arabi, M., Davis, J.G., Crimi, B., Pruden, A., 2010.Tracking antibiotic resistance genes in the South Platte riverbasin using molecular signatures of urban, agricultural, andpristine sources. Environmental Science & Technology 44 (19),7397e7404.

Szczepanowski, R., Linke, B., Krahn, I., Gartemann, K.H.,Gutzkow, T., Eichler, W., Puhler, A., Schluter, A., 2009.

Detection of 140 clinically relevant antibiotic-resistance genesin the plasmid metagenome of wastewater treatment plantbacteria showing reduced susceptibility to selectedantibiotics. Microbiology-Sgm 155, 2306e2319.

Travers, A.A., 2004. The structural basis of DNA flexibility.Philosophical Transactions of the Royal Society of LondonSeries A-Mathematical Physical and Engineering Sciences 362(1820), 1423e1438.

Volkmann, H., Schwartz, T., Bischoff, P., Kirchen, S., Obst, U.,2004. Detection of clinically relevant antibiotic-resistancegenes in municipal wastewater using real-time PCR (TaqMan).Journal of Microbiological Methods 56 (2), 277e286.

Wintgens, T., Gallenkernper, M., Melin, T., 2004. Removal ofendocrine disrupting compounds with membrane processesin wastewater treatment and reuse. Water Science andTechnology 50 (5), 1e8.

Xi, C.W., Zhang, Y.L., Marrs, C.F., Ye, W., Simon, C., Foxman, B.,Nriagu, J., 2009. Prevalence of antibiotic resistance in drinkingwater treatment and distribution systems. Applied andEnvironmental Microbiology 75 (17), 5714e5718.

Zhang, R.F., Eggleston, K., Rotimi, V., Zeckhauser, R.J., 2006.Antibiotic resistance as a global threat: evidence fromChina, Kuwait and the United States. Globalization andHealth 2.

Zhang, Y.L., Marrs, C.F., Simon, C., Xi, C.W., 2009. Wastewatertreatment contributes to selective increase of antibioticresistance among Acinetobacter spp. Science of the TotalEnvironment 407 (12), 3702e3706.