RESEARCH ARTICLE

Metagenomic Analysis of the Abundance and Composition of

Antibiotic Resistance Genes in Hospital Wastewater in Benin,
Burkina Faso, and Finland
Melina A. Markkanen,a,b Kaisa Haukka,a,g Katariina M. M. Pärnänen,a Victorien Tamegnon Dougnon,c
Isidore Juste O. Bonkoungou,d Zakaria Garba,e Halidou Tinto,e Anniina Sarekoski,a Antti Karkman,a,b Anu Kantele,b,f,g
Marko P. J. Virtaa,b

a Department of Microbiology, University of Helsinki, Helsinki, Finland

b Multidisciplinary Center of Excellence in Antimicrobial Resistance Research, University of Helsinki, Helsinki, Finland
c Research Unit in Applied Microbiology and Pharmacology of Natural Substances, Polytechnic School of Abomey-Calavi, University of Abomey-Calavi, Abomey-Calavi,
Benin
d Department of Biochemistry and Microbiology, University Joseph Ki-Zerbo, Ouagadougou, Burkina Faso
e Clinical Research Unit Nanoro, Institute for Research in Health Sciences, National Center for Scientific and Technological Research, Ouagadougou, Burkina Faso
Meilahti Vaccine Research Center MeVac, Department of Infectious Diseases, University of Helsinki and Helsinki University Hospital, Helsinki, Finland
f

g Human Microbiome Research Program, Faculty of Medicine, University of Helsinki, Helsinki, Finland

Marko P. J. Virta and Anu Kantele contributed equally to this work. Author order was determined alphabetically.

ABSTRACT Antibiotic resistance is a global threat to human health, with the most
severe effect in low- and middle-income countries. We explored the presence of antibi-
otic resistance genes (ARGs) in the hospital wastewater (HWW) of nine hospitals in Benin
and Burkina Faso, two low-income countries in West Africa, with shotgun metagenomic
sequencing. For comparison, we also studied six hospitals in Finland. The highest sum of
the relative abundance of ARGs in the 68 HWW samples was detected in Benin and the

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

lowest in Finland. HWW resistomes and mobilomes in Benin and Burkina Faso resembled
each other more than those in Finland. Many carbapenemase genes were detected at
various abundances, especially in HWW from Burkina Faso and Finland. The blaGES genes,
the most widespread carbapenemase gene in the Beninese HWW, were also found in
water intended for hand washing and in a puddle at a hospital yard in Benin. mcr genes
were detected in the HWW of all three countries, with mcr-5 being the most common
mcr gene. These and other mcr genes were observed in very high relative abundances,
even in treated wastewater in Burkina Faso and a street gutter in Benin. The results
highlight the importance of wastewater treatment, with particular attention to HWW.
IMPORTANCE The global emergence and increased spread of antibiotic resistance
threaten the effectiveness of antibiotics and, thus, the health of the entire popula-
tion. Therefore, understanding the resistomes in different geographical locations is
crucial in the global fight against the antibiotic resistance crisis. However, this infor-
mation is scarce in many low- and middle-income countries (LMICs), such as those in
Editor Mariana Castanheira, JMI Laboratories
West Africa. In this study, we describe the resistomes of hospital wastewater in Benin
Copyright © 2023 Markkanen et al. This is an
and Burkina Faso and, as a comparison, Finland. Our results help to understand the open-access article distributed under the terms
hitherto unrevealed resistance in Beninese and Burkinabe hospitals. Furthermore, the of the Creative Commons Attribution 4.0
International license.
results emphasize the importance of wastewater management infrastructure design
Address correspondence to Marko P. J. Virta,
to minimize exposure events between humans, HWW, and the environment, prevent- marko.virta@helsinki.fi, or Melina A. Markkanen,
ing the circulation of resistant bacteria and ARGs between humans (hospitals and melina.markkanen@helsinki.fi.
community) and the environment. The authors declare no conflict of interest.
Received 1 November 2022
KEYWORDS antibiotic resistance, West Africa, hospital wastewater (HWW), Accepted 3 January 2023
carbapenemase, colistin resistance, metagenomes

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 1

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

T he global antibiotic resistance crisis has multifaceted effects on human and animal
health and comes with substantial economic losses (1). Due to limitations in diag-
nostic testing in low-resource settings in low- and middle-income countries (LMICs),
broad-spectrum antibiotics are often used empirically without microbiological verifica-
tion of the causative pathogen or its sensitivity to different antibiotics (2). In addition,
unregulated access to antibiotics results in self-medication for humans and animals in
these countries (3–5).
Acquired, potentially mobile antibiotic resistance genes (ARGs) have a pronounced clini-
cal relevance and impact on the current antimicrobial resistance (AMR) problem (6, 7).
Furthermore, the genetic context of the ARG (e.g., whether under a strong promoter or
not) influences its expression and the resulting resistance phenotype (8). Class 1 integrons
are strongly linked to the dissemination of clinically relevant acquired ARGs (9, 10).
Although integrons are not mobile as such, multidrug resistance gene cassettes carried by
integrons can be transferred to new hosts: for example, via plasmids (11). The intI1 and
qacED genes (genes encoding the integron integrase and quaternary ammonium com-
pound resistance) are typically used as markers for class 1 integrons (10, 12).
The use of broad-spectrum antibiotics has increased in clinical practice as a con-
sequence of the increased prevalence of extended-spectrum b -lactamase-producing
Enterobacteriaceae (ESBL-PE) (13–15). The carbapenem resistance-encoding genes
blaGES, blaIMP, blaKPC, blaNDM, blaOXA-48, blaOXA-58, and blaVIM, which are typically carried
by plasmids, have emerged and spread around the world during the past 3 decades
(16–19). Colistin is a last-resort antibiotic used for treating infections caused by mul-
tidrug-resistant and extensively drug-resistant bacteria, such as those resistant to
carbapenems (20). The rapid emergence of colistin resistance mediated by mcr genes
threatens the efficacy of colistin in clinical use (21, 22).
Although AMR is a global concern, the crisis dramatically affects LMICs, such as
those in West Africa (2, 23–26). Lack of research data is one major factor hindering
tackling the AMR problem in these countries (2, 23, 27, 28). Despite these gaps in re-
sistance surveillance data, it is well known that the level of AMR is elevated in LMICs,
including African countries (14, 24, 29, 30). In contrast, in Northern European countries,
such as Finland, AMR occurrence is among the lowest globally, both in the community

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

(30, 31) and in health care settings (32).
Hospital wastewater (HWW) from health care facilities is at the frontline of AMR emer-
gence and spread due to the frequent use of antibiotics and the presence of immuno-
compromised patients (33). Thus, we set out to investigate the AMR situation and the
characteristics of the resistomes in nine hospitals in two West African countries, Benin
and Burkina Faso, where prior data were scarce (14). For comparison, we analyzed sam-
ples from six hospitals in Finland, where the level of AMR was expected to be low (32).
We used shotgun metagenomic sequencing to obtain a holistic view of the resistomes,
mobilomes (a set of ARGs and mobile genetic elements [MGEs], respectively), and micro-
bial communities present in the studied environments.

RESULTS
General features of resistomes, mobilomes, and taxonomical compositions in
HWW and other water samples from Benin, Burkina Faso, and Finland. We studied
hospital wastewater (HWW) collected from Benin, Burkina Faso, and Finland (Table 1;
see Data Set S1, Sheets 1 and 2, in the supplemental material; see Fig. S1 and S2 in the
Supplemental Data Repository, https://data.mendeley.com/datasets/9wxb37t49z/1). In
addition, 11 non-HWW water samples, such as those collected from rivers and near or
within the hospital environment, were analyzed to obtain a more comprehensive
understanding of the ARG prevalence in Benin and Burkina Faso. On average, 31 mil-
lion sequence reads per sample were analyzed. HWW from Benin showed the highest
and HWW from Burkina Faso the second-highest abundance of all detected ARGs nor-
malized to bacterial 16S rRNA genes (the sum of the relative abundance of ARGs). On
the other hand, the sum of the relative abundance of ARGs was the lowest in HWW
from Finland (Fig. 1A). Also, the lowest diversity of ARGs was observed in HWW from

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 2

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

TABLE 1 Sample information

Sample type and IDa Country Category N Hospital/area Collection date (day/mo/yr) HWW
HWW hospital samples
BH01–BH09 Benin BENN HWW A 7 A 27/11/19 Yes
BH27–BH39 BENN HWW B 11 B 29/11/19 Yes
BH44–BH50 BENN HWW C 5 C 9/12/19 Yes
BH58–BH61 BENN HWW D 3 D 11/12/19 Yes
BFH1–BFH4 Burkina Faso BF HWW E 2 E 22/11/19 Yes
BFH6–BFH12 BF HWW F 6 F 28/11/19 Yes
BFH13–BFH15 BF HWW G 3 G 28/11/19 Yes
BFH16–BFH28 BF HWW H 10 H 4/12/19 Yes
BFH29–BFH41 BF HWW I 13 I 12/12/19 Yes
FH1 Finland FI HWW J 1 J 20/1/20 Yes
FH2 FI HWW K 1 K 20/1/20 Yes
FH3 FI HWW L 1 L 20/1/20 Yes
FH4–FH6 FI HWW M 3 M 20/1/20 Yes
FH7 FI HWW N 1 N 23/1/20 Yes
FH9 FI HWW O 1 O 28/1/20 Yes

Other samples
BH11 Benin BENN well water A (drinking) 1 A 27/11/19 No
BH13 BENN street gutter B 1 B 27/11/19 No
BH48 BENN puddle at yard C 1 C 9/12/19 No
BH52 BENN hand-washing C 1 C 9/12/19 No
BSE100 BENN river P (drinking) 1 P 4/12/19 No
BSE74 BENN river Q (drinking) 1 Q 4/12/19 No
BSE79 BENN river R (drinking) 1 R 4/12/19 No
BSE93 BENN tap water S (drinking) 1 S 4/12/19 No
BFH26 Burkina Faso BF exit after biological treatment 1 H 4/12/19 No
BFH27 BF wetland receiving treated HWW 1 H 4/12/19 No
BFH42 BF receiving river after WWTP 1 T 12/12/19 No
aHWW samples from hospitals in Benin (n = 26), Burkina Faso (n = 34), and Finland (n = 8) are denoted at the top, while the other samples are noted at the bottom.

Finland and the highest in Burkina Faso (see Fig. S3A in the Supplemental Data
Repository).
According to the Kruskal-Wallis test, there were significant (P , 0.005) country-wise

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

differences in the sums of relative abundances of ARGs (Fig. 1A) and class 1 integron
genes (intI1) (Fig. 1C), but not when no particular type of MGE was specified (Fig. 1B). The
significant differences were investigated further using the Wilcoxon rank sum test, where

FIG 1 Sum of the relative abundances of (A) ARGs, (B) MGEs, and (C) class 1 integrons (intiI1) in HWW samples from Benin, Burkina Faso, and Finland. The
gene counts were normalized to 16S rRNA gene counts and gene lengths. Country medians are shown as horizontal lines, and the interquartile ranges
(25th and 75th quartiles) as box plot hinges. The horizontal lines represent the highest and lowest values. Outliers are defined as values higher or lower
than 1.5 times the upper or lower quartiles, respectively, and denoted with a text label referring to the category represented by that sample. The box plots
were drawn after excluding the outliers. The comparisons between countries were computed using the pairwise Wilcoxon rank sum test, where the P
values were adjusted for multiple testing using the Benjamini-Hochberg algorithm. The significance levels are as follows: *, P # 0.05; **, P # 0.01; ***, P #
0.001.

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 3

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

FIG 2 Principal-component analysis (PCA) showing the significant dissimilarities of (A) resistomes, (B) taxonomical composition Metaphlan3, and (C)
mobilomes in HWW from Benin, Burkina Faso, and Finland. Count data were transformed using centered log ratio transformation (clr). Confidence ellipses
are drawn for visualization and represent 95% confidence levels.

the P values were adjusted for multiple testing using the Benjamini-Hochberg algorithm.
The differences in intI1 followed a similar pattern in country-wise comparisons to ARGs
(Fig. 1C). Contigs where multiple ARGs, such as carbapenemase or ESBL variants of blaGES
and quinolone resistance genes (qnrVC), were located in proximity to each other were
identified (Fig. S4 in the Supplemental Data Repository). These contigs might indicate gene
cassettes carried by class 1 integron elements as previously reported for blaGES and qnrVC
in various ARG combinations (8, 34, 35). In contrast to intI1 and qacED, no significant corre-
lations were observed between ARGs and int2 or int3 (Fig. S5C and D in the Supplemental
Data Repository), except with HWW from Burkina Faso (Fig. S5D in the Supplemental Data
Repository).
Compositional data analysis (CoDa) methods were applied to investigate the ordination
of the samples from the different countries by their resistome, mobilome, and taxonomical

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

composition with respect to each other. Centered log ratio (clr) transformation, which uses
the geometric mean of the sample vector as reference (36), was applied to transform the
count data for the ordinations. The significance of the distances between samples from
country pairs was calculated on untransformed count data using Aitchison distance, which
corresponds to Euclidean distances between clr-transformed sample abundance vectors
(37). HWW resistomes from Benin and Burkina Faso resembled each other and formed clus-
ters distinct from the resistomes from Finland (permutational multivariate analysis of var-
iance [PERMANOVA]; P , 0.001; Benin versus Finland [BENN-FI], R2 = 0.211, Burkina Faso
versus Finland [BF-FI]; R2 = 0.160; Benin versus Burkina Faso [BENN-BF], R2 = 0.0801)
(Fig. 2A; Table S1A). Country (BENN-FI, BENN-BF, and BF-FI) explained less of the variance
between taxonomical compositions than variance between resistomes (PERMANOVA; P ,
0.001; BENN-FI, R2 = 0.178; BF-FI, R2 = 0.104; P , 0.005; BENN-BF, R2 = 0.0441) (Fig. 2B;
Table S1B). Similar to resistomes, distinct country-wise clusters were seen in the ordinations
of mobilomes (PERMANOVA; P , 0.001; BENN-FI, R2 = 0.196; BF-FI, R2 = 0.103; P , 0.005;
BENN-BF, R2 = 0.0478) (Fig. 2C; Table S1C). When non-HWW water samples were included
in the analysis, these samples seemed to be located approximately within the clusters of
their respective countries (PERMANOVA; P , 0.001) (Table S1D to F; Fig. S6 in the
Supplemental Data Repository). Since the HWW resistomes of Benin and Burkina Faso dif-
fered notably from the resistomes of Finland, the drivers for these differences were subse-
quently investigated.
Significantly differentially abundant ARGs. ARGs significantly differentially abun-
dant in HWW from each country-wise comparison were investigated. For that, analysis
by the ANOVA-Like Differential Expression tool for high-throughput sequencing data
(ALDEx2) was performed with additive log ratio (alr) transformation (36). alr transformation

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 4

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

FIG 3 Significantly differentially abundant ARGs in HWW in pairwise comparisons in Benin versus Finland, Burkina Faso versus Finland, and Benin versus
Burkina Faso were defined using ALDEx2 (36) with 16S rRNA as the reference gene. The difference-between (diff.btw) values represent the difference in alr-
transformed values of each ARG in samples from the compared countries. For instance, in panel A, significantly differentially abundant ARGs for Benin
(green dots above the blue dotted line) have negative diff.btw values (x axis), while ARGs that were significantly differentially abundant in Finland (blue
dots above the blue dotted line) have positive values. The dotted line represents a P value of ,0.05 of the Wilcoxon rank sum test, where the P values
were adjusted for multiple testing using the Benjamini-Hochberg algorithm (wi.eBH). These P values were log10 transformed for the figure.

uses a single component (here, the 16S rRNA counts) as the reference for analyzing all indi-
vidual components.
In the comparisons Benin versus Finland and Burkina Faso versus Finland, ARGs sig-
nificantly differentially abundant in HWW from Finland were fewer than those from
Benin or Burkina Faso (Fig. 3A and B) (BENN [n = 258]-FI [n = 33]; BF [n = 257]-FI
[n = 29]). In addition, the ARGs that were significantly differentially abundant in HWW
from Finland in both comparisons (Benin versus Finland and Burkina Faso versus
Finland) were primarily the same (Table 2; Table S1A and B). blaOXA-211-like genes and
erm(B) macrolide resistance genes were characteristic of HWW from Finland (Table 2
[and see Table S1A to C for all results]). Those ARGs that were characteristic of HWW

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

from Benin included tetracycline resistance gene tetA, quinolone resistance genes
qnrVC, and ESBL gene blaVEB (Table 2; and Table S1A). For Burkina Faso, significantly dif-
ferentially abundant ARGs included blaCMY genes, trimethoprim resistance genes of the
gene family dfrA15, and genes of blaOXA-10- and blaOXA-46-like gene families (Table 2;
Table S1B).
The comparison between HWW from Benin and Burkina Faso revealed fewer dif-
ferentially abundant ARGs between these countries (BENN [n = 118]-BF [n = 53])
than in the comparisons with Finnish HWW. In addition, the volcano plot in Fig. 3C
is more skewed toward the center (diff.btw value of zero), thus referring to less
drastic differences in this comparison. These results support the previous notion
that the resistomes in HWW from Benin and Burkina Faso were more similar to each
other than those from Finland (Table 2; Table S1A to C). However, ESBL genes blaBEL
and blaCMY and carbapenemase genes of the blaOXA-58-like gene family, were characteris-
tic of HWW from Burkina Faso (Table 2; Table S1C). Instead, different aminoglycoside and
lincosamide resistance genes, such as lnu(F) and lnu(C), were characteristic of HWW from
Benin (Table 2; Table S1C).
As a further notion, those blaOXA variants that were significantly differentially abun-
dant in HWW from Finland compared to Benin and Burkina Faso were predominantly
those that are intrinsically carried by some specific species and encode carbapene-
mases (e.g., blaOXA-211-like genes) (Table 3). Instead, the blaOXA variants characteristic for
HWW from Benin and Burkina Faso were those that are typically acquired and, further-
more, do not encode carbapenemase activity (e.g., blaOXA-5 and blaOXA-10) (Table 3). For
example, blaOXA-5-like genes, significantly differentially abundant in HWW from Benin,
are typically carried by class 1 integrons (38).

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 5

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

TABLE 2 Top 10 differentially abundant ARGs in HWW from Benin, Burkina Faso, and
Finlanda
Comparison wi.eBH Diff.btw Gene Gene family
Benin vs Finland
Benin 0.000032 25.811741 tet(A)_1 tet(A)
0.000033 25.267743 tet(A)_6 tet(A)
0.000034 24.932008 tet(A)_4 tet(A)
0.000034 24.842077 tet(A)_5 tet(A)
0.000037 27.475163 qnrVC4_1 qnrVC4
0.000042 26.688637 blaVEB-2_1 blaVEB
0.000043 26.521372 blaVEB-6_1 blaVEB
0.000045 26.600232 blaVEB-7_1 blaVEB
0.000047 26.629523 blaVEB-1_1 blaVEB
0.000058 28.506581 lnu(F)_3 lnu(F)
Finland 0.000065 5.930829 blaOXA-373_1 blaOXA-211-like
0.000068 4.334039 erm(B)_18 erm(B)
0.000079 5.493227 blaOXA-212_1 blaOXA-211-like
0.000085 4.834230 erm(B)_21 erm(B)
0.000113 4.735091 blaOXA-309_1 blaOXA-211-like
0.000118 5.320211 erm(B)_10 erm(B)
0.000189 3.499903 erm(B)_9 erm(B)
0.000225 5.338762 blaOXA-334_1 blaOXA-211-like
0.000407 4.719647 blaOXA-299_1 blaOXA-299-like
0.000409 4.664600 blaOXA-281_1 blaOXA-211-like

Burkina Faso vs Finland

Burkina Faso 0.000016 27.153927 blaCMY-4_1 blaCMY-150
0.000018 26.185994 dfrA15_2 dfrA15
0.000031 26.923325 blaOXA-101_1 blaOXA-10-like
0.000039 25.625515 dfrA15_4 dfrA15
0.000047 25.795837 blaCMY-2_1 blaCMY-150
0.000051 25.223876 blaVEB-6_1 blaVEB
0.000051 25.261070 blaOXA-56_1 blaOXA-10-like
0.000066 24.173189 tet(A)_1 tet(A)
0.000068 25.930935 blaCMY-121_1 blaCMY-150
0.000072 25.478843 blaOXA-46_1 blaOXA-46-like
Finland 0.000043 5.262960 blaOXA-334_1 blaOXA-211-like

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

0.000055 5.481699 blaOXA-299_1 blaOXA-299-like
0.000057 5.261032 blaOXA-309_1 blaOXA-211-like
0.000071 6.062276 blaOXA-212_1 blaOXA-211-like
0.000229 5.378873 blaOXA-373_1 blaOXA-211-like
0.000268 5.311696 blaOXA-281_1 blaOXA-211-like
0.000513 5.474025 blaOXA-280_1 blaOXA-211-like
0.001580 7.159558 blaOXA-211_1 blaOXA-211-like
0.001721 4.643516 erm(B)_10 erm(B)
0.002353 3.711949 erm(B)_6 erm(B)

Benin vs Burkina Faso

Benin 1.64E208 25.4719564 blaOXA-129_1 blaOXA-5-like
1.97E206 25.3861927 aac(69)-IIc_1_NC aac(69)-Iic
8.19E206 23.0755306 blaOXA-256_1 blaOXA-10-like
4.10E205 25.3303508 aph(20)-Ib_1 aph(20)-Ib
5.43E205 25.244107 aac(69)-Im_1 aac(69)-Im
0.00020678 23.4867524 lnu(C)_1 lnu(C)
0.0002553 22.2747178 tet(C)_3 tet(C)
0.00030175 22.3241004 tet(C)_2 tet(C)
0.00051131 24.0449182 aph(20)-Ib_2 aph(20)-Ib
0.00051266 22.256968 tet(C)_1_NC tet(C)
Burkina Faso 0.00091224 2.91935432 tet(39)_1 tet(39)
0.00151264 4.1710954 blaBEL-1_1 blaBEL
0.002193 2.53504117 cmlB1_1 cmlB
0.00328808 3.4417663 blaOXA-58_1 blaOXA-58-like
0.00556017 2.5513906 dfrB5_1 dfrB1
(Continued on next page)

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 6

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

TABLE 2 (Continued)
Comparison wi.eBH Diff.btw Gene Gene family
0.00999705 2.36292741 blaCMY-15_1 blaCMY-150
0.01017209 2.18229534 blaCMY-95_1 blaCMY-150
0.0104837 3.39104568 blaBEL-3_1 blaBEL
0.01139229 2.93238914 blaOXA-397_1 blaOXA-58-like
0.01498384 1.9958258 blaCMY-94_1 blaCMY-150
aThediff.btw values represent the median difference in the alr-transformed count data between the compared
HWW. The P values of the Wilcoxon rank sum test, where the P values were adjusted for multiple testing using
the Benjamini-Hochberg algorithm, are indicated in the table as “wi.eBH.” The table was sorted by the wi.eBH
value before subdividing the entries into the top 10 differentially abundant ARGs by country: Benin, Burkina
Faso, or Finland. The ARGs were clustered into gene families (Gene family column) based on 90% shared
sequence identity using CD-HIT (80). In the case of blaOXA genes, the gene family naming followed the scheme
by Naas and colleagues (46).

Carbapenemase genes. The presence of seven acquired carbapenemase genes

(blaGES, blaIMP, blaKPC, blaNDM, blaOXA-48, blaOXA-58, and blaVIM) was analyzed in detail. These
ARGs were selected as their putative resistance phenotype is often associated with com-
plex infections with minimal treatment options (16, 19, 39), especially in LMICs (2, 25).
The highest relative abundance of these carbapenemase genes was observed in Finnish
hospital J (Fig. 4). On the other hand, hospital M in Finland was nearly free from these
carbapenemase genes (Fig. 4). In HWW from Benin, the carbapenemase genes in the

TABLE 3 Differentially abundant blaOXA variants in HWW from Benin, Burkina Faso, and Finlanda
Comparison for ALDEx2 Cluster name in BLDB Acquired/intrinsicb Host species if intrinsic Carbapenemase activity
Benin vs Finland
Benin blaOXA-5-like A
blaOXA-10-like A
blaOXA-2-like A
blaOXA-347 A and I Bacteroides spp.
blaOXA-229 I Acinetobacter bereziniae
blaOXA-46-like A

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

blaOXA-1-like A
blaOXA-209-like A
Finland blaOXA-296-like I Acinetobacter bohemicus
blaOXA-58-like I Acinetobacter baumannii Yes
blaOXA-211-like I Acinetobacter johnsonii Yes
blaOXA-299-like I Acinetobacter bouvetii
blaOXA-427-like A Yes

Burkina Faso vs Finland

Burkina Faso blaOXA-10-like A
blaOXA-46-like A
blaOXA-2-like A
Finland blaOXA-296-like I Acinetobacter bohemicus
blaOXA-211-like I Acinetobacter johnsonii Yes
blaOXA-299-like I Acinetobacter bouvetii

Benin vs Burkina Faso

Benin blaOXA-5-like A
blaOXA-10-like A
blaOXA-347 A and I Bacteroides spp.
blaOXA-2-like A
blaOXA-209-like A
blaOXA-46-like A
blaOXA-1-like A
Burkina Faso blaOXA-10-like A
blaOXA-58-like I Acinetobacter baumannii Yes
aThe ARGs significantly differentially abundant in each country are denoted by country. The data concerning the origin of the blaOXA gene, whether it is intrinsic or acquired,
was retrieved from Beta Lactamase Database (BLDB) (46). However, we used the expression “intrinsic” instead of “natural” due to the misleading connotation of the word
“natural” in the context of antibiotic resistance.
bA, acquired; I, intrinsic.

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 7

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

FIG 4 Relative abundances of carbapenemase genes in relation to 16S rRNA gene counts in HWW from Benin (left, green background), Burkina Faso
(center, red background), and Finland (upper right, blue background), as well as various other water sources in Benin (center right, green background) and
Burkina Faso (bottom right, red background). Note the differences in the y axis scales between the figures for HWW and non-HWW samples. Only variants
known to encode carbapenemases were screened (46) (Table S3A).

blaGES gene family seemed to dominate over other carbapenemase genes in all four hos-
pitals (Fig. 4). blaGES genes were also detected in the water puddle surrounding the sur-
gery room septic tank at a Beninese hospital yard (hospital C) (Fig. 4; Fig. S2E in the
Supplemental Data Repository) as well as in the water intended for hand washing in the
same hospital (Fig. 4; Fig. S2F in the Supplemental Data Repository). Also, the street gut-
ter, located ;100 m away from another Beninese hospital, was contaminated by blaGES
carbapenemase genes (Fig. 4).
In contrast to the homogeneity of carbapenemase genes in HWW from Benin, most of
the other carbapenemase genes were present in HWW from Burkina Faso and Finland at var-
ious prevalence levels (Fig. 4). blaIMP, blaNDM, and blaVIM were mainly detected in Burkinabe
and Finnish HWW samples, and the latter was significantly differentially abundant in HWW
from Burkina Faso (Fig. 4; Table S2B). blaOXA-48 was detected in two HWW samples from
Burkina Faso and not at all or in very low relative abundances in samples from elsewhere
(Fig. 4). Instead, blaOXA-58-like genes were present in the majority of Burkinabe and Finnish
HWW samples but only in a few Beninese samples (Fig. 4). The detection of blaKPC genes was
restricted to one Finnish HWW sample (Fig. 4). The samples collected from natural waters
and other drinking waters showed only low relative abundances of these carbapenemase

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 8

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

FIG 5 Relative abundances of mcr genes in relation to 16S rRNA gene counts in HWW from Benin (left, green background), Burkina Faso (center, red
background), and Finland (upper right, blue background), as well as various other water sources in Benin (center right, green background) and Burkina
Faso (bottom right, red background). Variants were clustered based on 90% similarity in their sequence identity using CD-HIT (80) (Table S3B).

genes (Fig. 4). However, some blaOXA-58-like were detected in tap water used for drinking in
Benin (Fig. 4).
In Burkina Faso, blaGES carbapenemase genes were detected in HWW, which had
gone through the biological treatment in similar relative abundances as in some of the
samples of untreated HWW from the same hospital (hospital H) (Fig. 4 [note the differ-
ences in the plot scales]). Also, blaVIM was found both in the untreated and treated
water of that hospital (Fig. 4). Lower relative abundances of the studied carbapene-
mase genes were observed in wetlands and rivers receiving treated HWW in Burkina
Faso (Fig. 4). However, the spectrum of these carbapenemase genes, namely, blaOXA-58-
like, blaVIM, and blaGES, reflected the ones detected in the HWW in Burkina Faso (Fig. 4).
Mobile colistin resistance (mcr) genes. mcr genes were detected in several HWW
samples in Benin, Burkina Faso, and Finland (Fig. 5). mcr-5 (variants mcr-5.1 and mcr-
5.2) (Table S3B) was the most common of the mcr genes as they were found in HWW
from all except two hospitals in Burkina Faso (hospitals E and G) and two hospitals in
Finland (hospitals K and L) among the three countries (Fig. 5). In the few samples
selected for metagenomic assembly, this gene was found to be located within a Tn3-
like element (Fig. S7 in the Supplemental Data Repository).
In addition to the HWW samples, high relative abundances of mcr-5 genes were also
detected in the immediate and more distant surroundings of the hospitals in Benin and

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 9

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

Burkina Faso. A very high relative abundance (2.80  1022) of mcr-5 was observed in the
biologically treated HWW from hospital H (Fig. 5). In the wetland receiving treated HWW
from the same hospital, the relative abundance was lower (4.30  1024) but still high, con-
sidering that the sample represented a larger body of natural water. In Benin, mcr-5 was
detected in river water in a distant village and a street gutter near a hospital (hospital B)
(Fig. 5). Furthermore, for hospital C, the relative abundance of mcr-5 was greater in a pud-
dle near the septic tanks in the hospital yard (1.99  1023) than the average in the actual
HWW of that hospital (5.24  1024) (Fig. 5).
The next most prevalent mcr genes were mcr-3.1, mcr-10, and mcr-7, and similarly to
mcr-5, they were also present in non-HWW water samples, such as the water intended for
hand washing in the Beninese hospital C (Fig. 5). The gene mcr-2 was not detected at all,
and the lowest average relative abundance was detected for the gene mcr-1 (Fig. 5).
Taxonomical compositions. The taxonomical compositions of bacteria present in septic
tanks and sumps in Benin and Burkina Faso differed from those found in hospital sewers in
Finland. Genera belonging to the Bacteroidales family showed significantly different abundan-
ces in HWW from different countries and were characteristic of the HWW from Benin and
Burkina Faso (Fig. S8 and S9 in the Supplemental Data Repository). Other significantly differen-
tially abundant taxa in HWW from Benin compared to Finland included Chloracidobacterium,
Geobacter, Aminomonas, Flexilinea, Desulfovibrio, and genera of Synergistetes (Table S4A; Fig.
S8A in the Supplemental Data Repository). The first two mentioned were characteristic also of
the HWW from Burkina Faso (Table S4B; Fig. S8B at Supplemental Data Repository). In addi-
tion, in many HWW samples from Burkina Faso, the relative abundances of Pseudomonas and
Acinetobacter were high (Fig. S9 in the Supplemental Data Repository), and the difference in
their abundance in HWW from Burkina Faso versus Benin was significant (Table S4C; Fig. S8C
in the Supplemental Data Repository).
Coprococcus, Enterococcus, Lactococcus, Streptococcus, Trichococcus, Tessaracoccus, Delftia,
Raultella, and genera of the Proteobacteria family, were all significantly differentially abundant
in HWW from Finland in comparison to Benin and Burkina Faso (Table S3A and B; Fig. S8A
and S8B in the Supplemental Data Repository). Also, the genera that were significantly differ-
entially abundant for HWW in Finland and not Benin were more commonly from the phylum
Pseudomonadota (synonym Proteobacteria), which are typically considered the major contrib-

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

utors to the spread of ARGs carried by plasmids, integrons, or other MGEs (8, 40) (Fig. S8 in
the Supplemental Data Repository).
A great variation was seen in the top 11 taxa in the non-HWW samples (Fig. S9 in the
Supplemental Data Repository). The biologically treated HWW from Burkina Faso showed
high relative abundances of Aeromonas and Pseudomonas. In contrast, high relative abun-
dances of Bacteroides and Klebsiella were detected in the street gutter water located near
hospital B (Fig. S9 in the Supplemental Data Repository). In some samples, the top 11 taxa
were present only in low relative abundances. In contrast, the single genus Polynucleobacter
dominated over other genera in a few sampled natural waters receiving treated wastewater
in both Benin and Burkina Faso (Fig. S9 in the Supplemental Data Repository). The relative
abundance of Acinetobacter in tap water used for drinking in Benin reached a similar level to
that in many HWW samples (Fig. S9) in the Supplemental Data Repository, which possibly
explained the finding of blaOXA-58-like genes in this sample mentioned earlier (Fig. 4).

DISCUSSION
We characterized the bacterial community composition, resistome, and mobilome of 60
HWW and 11 other water samples from Benin and Burkina Faso and compared them to 8
HWW samples from Finland. Due to the lack of systematic AMR surveillance in these West
African countries, available data on bacterial resistance are patchy and heterogeneous
(1, 2, 22, 26, 29). Thus, the magnitude of the resistance problem and the specific ARG reser-
voirs are yet to be unraveled. This is the first study investigating hospital wastewater from
Benin and Burkina Faso using a shotgun metagenomic approach.
Interestingly, the ARGs observed in HWW from Benin were the highest in number
but probably less clinically important (16, 19, 39) than those in the HWW from Burkina

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 10

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

Faso and Finland. While carbapenemase genes blaGES, blaIMP, blaNDM, blaOXA-48-like,
blaOXA-58-like, and blaVIM were detected at various abundances in at least one sample in
HWW from Burkina Faso and Finland, blaGES was the predominant carbapenemase in
HWW from Benin. blaGES genes were also present in the water puddle at the Beninese
hospital’s yard and in water intended for visitors’ hand washing in the hospital. Thus,
one route of transmission of these ARGs within hospitals might be via hands contami-
nated by hand washing water, which might explain the high prevalence of these genes
in hospitals. Jacobs and colleagues have also drawn attention to the potentially inferior
microbiological water quality in similar hand washing water tanks, which are very com-
mon in West Africa (2).
We speculate that the dominance of blaGES carbapenemase genes might have been
caused by selection pressure due to the presence of antibiotic residues in the HWW.
However, as carbapenem antibiotics are used less in West African countries than in
North America and Southern and Central Europe, due to their high price (15), we sug-
gest that compounds from other antibiotic classes could have driven the selection. In
fact, blaGES genes are typically carried by class 1 integrons, in which they may be
coupled with multiple other ARGs (41). Thus, we speculate that class 1 integrons (12)
and coselection phenomena (25, 42) have a role in the dominance of blaGES in HWW in
Benin. That is, in the example of metagenome-assembled contigs in which quinolone
ARGs (qnrVC) and blaGES genes seemed to be located near each other, putatively car-
ried by the class 1 integron gene cassette, the selection pressure targeted to the quino-
lone ARG would enrich both genes even in the absence of the target substrate for
blaGES. However, there is a high sequence similarity among blaGES variants, which
include those conferring ESBL and carbapenem resistance. Therefore, we acknowledge
the possibility that the putative lack of specificity related to the read mapping might
have resulted in some misidentifications between the carbapenemase- and ESBL-
encoding variants of blaGES genes. Nonetheless, ESBL-producing bacteria are very com-
mon in African countries (43) and cause major challenges for infection control (28) as
carbapenem antibiotics are less available.
Although the lowest sum of the relative abundance of ARGs was detected in HWW
from Finland, some of the seven carbapenemase genes showed higher relative abundan-

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

ces in Finland than in HWW of the surveyed West African hospitals. For the occurrence of
blaOXA genes, the abundances of specific species present in the different HWW collection
systems might have played a role, as blaOXA variants that encode carbapenemases and are
typically intrinsically carried by certain Acinetobacter species (44, 45) were characteristic of
Finnish HWW. Instead, those blaOXA variants that are typically acquired and mainly encode
more-narrow-spectrum b -lactamases were significantly differentially abundant in HWW
from Benin and also Burkina Faso (46), although the occurrence of the seven carbapene-
mase genes was not homogenous among the Finnish samples. For example, blaKPC was
limited to a single HWW sample from Finland (hospital J) and no blaKPC gene was found
from Benin or Burkina Faso. This finding aligns with local and global reports and systematic
reviews (47–50), indicating that blaKPC carbapenemase genes are spreading more profusely
in Europe and North America than in Africa. Moreover, the two hospitals J and K, both part
of the University Central Hospital of Helsinki, the capital of Finland, showed the highest
sum of relative abundance of the seven carbapenemase genes among all HWW samples in
this study. These results are somewhat surprising as, to date, Finland is known as one of
the countries with the lowest level of bacterial resistance in Europe and globally (51), and
the carbapenemase-producing strains detected in Finnish hospitals are relatively rare and
typically associated with international travel or hospitalization (52).
One factor likely explaining our findings of the distinctive features in the HWW resis-
tomes in Benin and Burkina Faso compared to Finland was the differences in the waste-
water collection systems between the studied countries. In Finland, the hospital toilet
waters containing human fecal material are directed to HWW, while for the majority of the
Burkinabe and especially Beninese HWW studied here, this was not the case. Additionally,
in septic tanks (Benin and Burkina Faso), the water remains stagnant, possibly giving rise to

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 11

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

anaerobes, unlike the Finnish HWW, which flows through the system. Significant preva-
lence of anaerobic genera, such as Geobacter, Aminomonas, Flexilinea, and Desulfovibrio, as
well as genera of Synergistetes and Bacteroidales observed in HWW from Benin and Burkina
Faso and not in Finland, is in line with this speculation except for the last-mentioned taxa,
which are typical human gut commensals (53). These findings also align with the previous
reports on bacterial genera typically found in soil and aquatic environments (54) and previ-
ously described to dominate HWW from Benin and Burkina Faso (55). Genera of the faculta-
tively anaerobic bacteria, such as the lactic acid bacterium Aeromonas, and of the anaero-
bic bacterium Bifidobacterium (56), which are considered typical human gut microbes (53),
were significantly differentially abundant in the Finnish HWW instead of HWW of Benin or
Burkina Faso.
Compared to many high-income countries, such as those in Northern Europe, anti-
biotic usage in agricultural (4, 5) and clinical (2) settings differs (15) and is less con-
trolled—or even unregulated—in many African countries. For example, while banned
in many other countries, the use of colistin as a feed additive is allowed in many LMICs
(57), including in Africa (5). mcr-5 was the most commonly detected mcr gene in the
HWW in our study, contrary to previous reports on the prevalence of mcr genes glob-
ally (21) and in Africa (22, 58). This difference may be due to the methodologies used
to screen for colistin resistance. Other than the best-known mcr genes (mcr-1, -2, and
-3), other mcr gene variants are rarely targeted when screening for mcr genes using
conventional PCR (59). Therefore, our study shows that to obtain a more realistic view
of mcr genes in Africa, screening should be conducted for a broader set of different
mcr genes, as was recently done by Ngbede and colleagues (60).
The mcr-5 gene detected was embedded in a Tn3-like element, similar to previous
reports for Salmonella enterica (61, 62) and Escherichia coli (63) plasmids and the chro-
mosome of Cupriavidus gilardii (61). These Tn3-like elements are flanked by inverted
repeats, which enable translocation and putatively a broad host range for the mcr-5-
harboring element (61–63). Based on our results, the occurrences of mcr-5 and other
mcr genes in Benin and Burkina Faso were not restricted to HWW septic tanks. These
genes were also detected in water intended for the hand washing of visitors to the
hospital. The high prevalence of mcr-5 among various samples in this study raises

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

questions about its origin and ecology: whether it is of clinical origin or intrinsically car-
ried by some environmental bacteria. However, the highest relative abundances of
mcr-5 were observed in samples influenced by human activity, and all four rivers in the
remote village in Benin were free from this gene.
In Burkina Faso, the relative abundance of mcr-5 in HWW, even after biological treatment,
was high. Although we could not confirm the association between the mcr-5 genes detected
in the treated and the hospital-associated wastewater in the Burkinabe hospital, our study
suggests the inability of the currently used wastewater treatment processes in Burkina Faso
to remove the mcr genes. Wastewater treatment systems have indeed previously been
described as inadequate in Benin and Burkina Faso (64). The situation appears to be espe-
cially critical in Cotonou, Benin, where the hospitals involved in the study did not use any
kind of wastewater treatment. Groundwater in the city is extracted from the shallow aquifer,
which is polluted due to unauthorized waste deposits, inadequate toilets, pit latrines, and
septic tanks prone to leakage and hydraulic failure (65). This enables the continuous circling
of mcr genes and other ARGs as the inadequately treated water is released into natural
waters and used for various purposes by local people. Furthermore, in Africa, untreated
wastewater is commonly used for irrigation in urban agriculture, possibly enabling the dis-
semination of ARGs to fresh produce (66, 67). Our study shows that untreated or even
treated hospital wastewater, which can leak into groundwater and the environment, may
carry clinically hazardous antibiotic-resistant bacteria or resistance genes.
Based on our results from nine hospitals, we can state that clinically important
ARGs are circulating in Beninese and Burkinabe hospitals and their surroundings. There
were differences in the HWW collection systems in Benin and Burkina Faso compared
to Finland. This, among other factors, seems to be reflected to some extent in the

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 12

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

taxonomical compositions and, therefore, resistomes found in these HWW. However,

taxonomy explained the variance in the resistomes from the different countries only
partially, at least at the genus level. Although there were fewer dissimilarities in the
resistomes between HWW from Benin and Burkina Faso than in the comparisons with
Finland, it is important to consider the differences among West African countries in
AMR surveillance.

MATERIALS AND METHODS

Sample description. Hospital wastewater (HWW) samples were collected in Benin from four hospi-
tals (hospitals A to D; n = 26) and in Burkina Faso from five different hospitals (hospitals E to I; n = 34) in
November and December 2019. In Finland, HWW samples were collected from six different hospitals
(hospitals J to O; n = 8) in January 2020. The various wards, clinics, and other units typically had their
own septic tanks or sumps in the Beninese and Burkinabe hospitals. In Burkina Faso, the samples were
mainly from septic tanks or sewers in the hospital area. In Benin, none of the hospitals was connected to
a sewer system, and the samples were from septic tanks or sumps (unstructured wastewater wells),
which were never emptied to our knowledge. In most cases, the toilet water was not directed into these
sumps. For comparison, 11 non-HWW water sources were sampled. These included samples from the
tap [BENN tap water S (drinking); n = 1] and river waters [BENN river P, Q, R (drinking); n = 3] used for
drinking in a remote countryside village in the community of Savalou in central Benin as well as drinking
water from a well located in a 100-m distance from hospital A in Benin [BENN well water A (drinking);
n = 1]. Street gutter water near hospital B (in a 100-m distance) (BENN street gutter A; n = 1) and a water
puddle at hospital C yard next to HWW septic tank (BENN puddle at yard C; n = 1) and a tank distributing
water for hand washing in hospital C in Benin (BENN hand-washing C; n = 1) were also sampled. In
Burkina Faso, samples from biologically treated HWW from hospital H (BF exit after biological treatment
H; n = 1) and a wetland receiving this water (BF wetland receiving treated HWW H; n = 1) were collected.
Additionally, one sample was collected from wastewater treated in a local wastewater treatment plant
(WWTP) and destined for a river in Burkina Faso (BF receiving river after WWTP; n = 1) (Table 1). Detailed
sample descriptions are provided in the Data Set S1, Sheets 1 and 2, in the supplemental material.
Illustrative pictures and a map indicating sample collection regions are shown in Fig. S1 and S2 in the
Supplemental Data Repository, https://data.mendeley.com/datasets/9wxb37t49z/1.
DNA extraction and metagenomic sequencing. Water samples were collected into 1- L bottles and
transported to a laboratory on ice. They were kept at 14°C until processed within 24 h. A volume of 50
to 100 mL was filtered through a 0.2-m m-pore polycarbonate filter (Whatman; GE Healthcare Life
Sciences) using a portable vacuum pump (Millivac-Mini vacuum pump XF54; Millipore, Merck). DNA was
extracted from the filters using the Qiagen Dneasy PowerWater DNA kit following the manufacturer’s
instructions. The concentration and quality of the extracted DNA were determined with a NanoDrop
spectrophotometer. Altogether, 79 samples were subjected to shotgun metagenomic sequencing using

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

Illumina Novaseq6000 with Nextera XT library preparation at the Institute of Biotechnology, University
of Helsinki.
Bioinformatic analyses. All quality control and read mapping analyses were run using an in-house
Snakemake (v.5.3.0) (68) workflow. Briefly, the quality control steps included in the workflow were performed
using FastQC (v.0.11.8) (69) and MultiQC (v.1.9) (70), with adapter and low-quality read removal using Cutadapt
(v.2.7) (71) (parameters -O 10 -m 30 -q 20). Nucleotide sequence reads were mapped using Bowtie2 (v.2.4.1)
(72) (parameters -D 20 -R 3 -N 1 -L 20 -I S,1,0.50) against the ResFinder database (v.3.2; downloaded on 28 June
2020) (73). The reads were sorted and filtered using SAMtools (v.1.9) (74), such that the reads mapping as pairs
or alone were calculated as a single count. Mobile genetic elements were identified by mapping the reads with
Bowtie2, similarly to the procedure described above, against the MobileGeneticElementDatabase (56) (https://
github.com/KatariinaParnanen/MobileGeneticElementDatabase; downloaded on 28 June 2020), consisting of
2,714 unique MGE sequences, including transposons, integrons of classes 1, 2, and 3, the integron-associated
disinfectant resistance gene qacED, and standard insertion sequences (ISs) and ISs with insertion sequence com-
mon regions (ISCRs).
Taxonomic profiling was performed using both Metaphlan3 (v.3.0.1) (75) and Metaxa2 (v.2.2.1) (76).
Metaphlan3 was run to achieve both relative abundances and “absolute abundances” defined by the pro-
gram developers using the parameter -t rel_ab_w_read_stats. The outputs from the former were used for
ordinations and the latter for diversity analyses with vegan package (v2.6.2) (77, see below). In both cases,
the merged abundance tables were modified so that only taxa that were identified to species level were
included in the downstream analyses. However, knowing the limitations of species-level taxon identifica-
tion using shotgun metagenomics (78), the results were presented only at the genus level after taxon
agglomeration by the function tax_glom from the phyloseq package (v.1.40.0) (79) prior to analyses.
The counts for bacterial 16S rRNA from Metaxa2 were used to normalize ARGs and MGE counts to
obtain relative abundances. Gene lengths were taken into account in the normalization. Three HWW
samples were collected in duplicate (Data Set S1, Sheet 2). These replicates were not included in the sta-
tistical analyses but were used to evaluate the selected methods for detecting resistance. The sums of
the relative abundances of ARGs in the replicated sample pairs were highly similar to each other (Fig.
S10 in the Supplemental Data Repository).
All resistance genes found in the ResFinder database (73) were clustered into gene families based on
90% similarity in their sequence identity using CD-HIT (v.4.8.1) (80). Data available in the Beta-Lactamase
DataBase (BLDB) (46) were used to confirm the carbapenemase activity of the variants of blaOXA gene

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 13

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

families as well as other carbapenemase genes studied here (Table S3A). For instance, only blaGES var-
iants known to encode carbapenemases were included in the visualization (Fig. 4), while those encoding
ESBL phenotypes were not. Similarly, ARG clustering of 90% shared nucleotide identity was applied to
group mcr variants (Table S3B).
To study the genetic environment of ARGs, a subset of samples was assembled into contigs with MEGAHIT
(v.1.2.8) (81) with parameters –min-contig-len 1000 -m 32000000000. The anvi’o (v.7) (82) and Bandage (83) pro-
grams were applied to visualize the genetic environments (Fig. S5 and S7 in the Supplemental Data
Repository). For the mcr-5 gene, one to two mcr-5-positive samples from all studied countries were selected for
the assembly and the analysis of the genetic background. The same samples were used to search putative inte-
gron-carried multidrug resistance gene cassettes. Those putative gene cassettes with multiple ARGs encoded
by the same fragment are represented in Fig. S4 in the Supplemental Data Repository.
Statistical analyses. (i) General. All statistical analyses described below were performed for the 68
HWW samples (Table 1; Data Set 1, Sheet 1). The sum of the relative abundances for ARGs, MGEs, and
intI1 was obtained using 16S rRNA gene counts and gene lengths to normalize the count data. As the
data did not fulfill the assumptions of normality, a Kruskal-Wallis test from the stats package (v.4.2.0)
(84) was applied to study whether the differences between countries were significant. Pairwise Wilcoxon
rank sum tests adjusted by Benjamini-Hochberg from the stats package (v.4.2.0) (84) were performed to
determine which comparisons were significant for ARGs and intI1. Pearson correlations for the relative
abundances of ARGs and MGEs were computed using package ggpubr (v 0.4.0.999) (85, see below).
(ii) Compositional analyses. Next-generation sequencing (NGS) data are compositional as they contain
only relative information (37). By ignoring this compositional nature of NGS data, results conducted by tradi-
tional normalization methods may suffer from technical artifacts due to sequencing depth limitations (86).
The ANOVA-Like Differential Expression tool for high-throughput sequencing data (ALDEx2) (v.1.28.0) (36)
was applied to study the divergent features of the resistomes in HWW from each studied country. ALDEx2
handles the compositionality of the data by applying suitable data transformations. To study the differentially
abundant ARGs in this study, additive log ratio (alr) transformation was used with the 16S rRNA gene as the
denominator gene. First, the ARG count data were split, so that pairwise comparisons between countries
were possible (e.g., Benin versus Finland). The command aldex.clr(df, conditions, denom = ref) excludes the
features with zero count in all samples and performs the alr transformation with the selected reference gene.
The significance of the comparisons was tested using the Wilcoxon rank test [command aldex.ttest(x, paired.t-
est = FALSE, verbose = FALSE)], in which Benjamini-Hochberg corrections control false-positive identifications.
Finally, the effect sizes and the within- and between-condition values were estimated with the command
aldex.effect. ALDEx2 (36) was also applied to study the differentially abundant taxa between the HWWs from
different countries. For that, centered log ratio transformation, which uses the geometric mean of the sample
vector as the reference, was applied to Metaphlan3 count data (generated using the parameter -t rel_ab_-
w_read_stats), and the significance of the comparisons was tested similarly as described above.
For principal-component analysis (PCA) ordinations, clr-transformed ARG, MGE, and taxon counts
(generated by Metaphlan3 with parameter -t rel_ab_w_read_stats) were visualized using microViz
(v.0.9.1) (87) with the command count_data %.% tax_transform(“clr”) %.% ord_calc(method = “PCA”)
%.% ord_plot(color = “country”). For PCA, taxa were fixed to the genus level as described earlier. The

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

significance of the distances between samples from country pairs was calculated using Aitchison dis-
tance on the untransformed count data by microViz (87) [command count_data %.% dist_calc(“aitchi-
son”) %.% dist_permanova(variables = c(“country”), n_perms = 9999, seed = 12345)].
All statistical analyses were performed in RStudio (v.4.2.0), and the results were visualized using
ggplot2 (v.3.3.6) (88) and patchwork (v.1.1.1) (89). Vector maps were drawn using the packages rnatura-
learth (v.0.1.0) (90), ggspatial (v.1.1.6) (91), and maps (v.3.4.0) (92).
Ethical permission. The ethical permissions for the project were received from Comité National
d¨Ethique pour la Recherche en Santé under the Health Ministry in Benin and Comité d¨Ethique pour la
Recherche en Santé under the Health Ministry in Burkina Faso.
Data availability. The data for this study have been deposited in the European Nucleotide Archive
(ENA) at EMBL-EBI under accession no. PRJEB47975. They will be public upon article publication. All cus-
tom codes used for the analyses are available from https://github.com/melinamarkkanen/AMRIWA upon
article publication. Analysis scripts were provided for the reviewers prior to publication.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.
TABLE S1, XLSX file, 0.02 MB.
TABLE S2, XLSX file, 0.1 MB.
TABLE S3, XLSX file, 0.01 MB.
TABLE S4, XLSX file, 0.02 MB.
DATA SET S1, XLSX file, 0.1 MB.

ACKNOWLEDGMENTS
We thank the staff in all of the participating hospitals for their collaboration in
sampling. We also thank the laboratory personnel of the Research Unit in Applied
Microbiology and Pharmacology of Natural Substances at the University of Abomey-
Calavi in Benin and the Clinical Research Unit of Nanoro (CRUN) in Burkina Faso for

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 14

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

sample processing. Finally, we acknowledge CSC, IT Center for Science, Finland, for
providing the computational resources for the study.
This work was supported by the Academy of Finland (grant no. 346125, 318643, and
316708) and the University of Helsinki HiLife Grand Challenges funding. Open access
funded by Helsinki University Library.

REFERENCES
1. World Health Organization. 2016. Global action plan on antimicrobial re- 19. Nordmann P, Poirel L. 2014. The difficult-to-control spread of carbapene-
sistance. https://www.who.int/publications/i/item/9789241509763. mase producers among Enterobacteriaceae worldwide. Clin Microbiol
2. Jacobs J, Hardy L, Semret M, Lunguya O, Phe T, Affolabi D, Yansouni C, Infect 20:821–830. https://doi.org/10.1111/1469-0691.12719.
Vandenberg O. 2019. Diagnostic bacteriology in district hospitals in sub- 20. Falagas ME, Kasiakou Sofia K. 2005. Colistin: the revival of polymyxins for
Saharan Africa: at the forefront of the containment of antimicrobial resist- the management of multidrug-resistant Gram-negative bacterial infec-
ance. Front Med (Lausanne) 6:205. https://doi.org/10.3389/fmed.2019.00205. tions. Clin Infect Dis 40:1333–1341. https://doi.org/10.1086/429323.
3. Belachew SA, Hall L, Selvey LA. 2021. Non-prescription dispensing of anti- 21. Ling Z, Yin W, Shen Z, Wang Y, Shen J, Walsh TR. 2020. Epidemiology of
biotic agents among community drug retail outlets in sub-Saharan Afri- mobile colistin resistance genes mcr-1 to mcr-9. J Antimicrob Chemother
can countries: a systematic review and meta-analysis. Antimicrob Resist 75:3087–3095. https://doi.org/10.1093/jac/dkaa205.
Infect Control 10:13. https://doi.org/10.1186/s13756-020-00880-w. 22. Anyanwu MU, Odilichukwu C, Okpala R, Chah KF, Shoyinka VS. 2021. Preva-
4. Butcher A, Cañada JA, Sariola S. 2021. How to make noncoherent problems lence and traits of mobile colistin resistance gene harbouring isolates from
more productive: towards an AMR management plan for low resource live- different ecosystems in Africa. Biomed Res Int 2021:6630379. https://doi.org/
stock sectors. Humanit Soc Sci Commun 8:287. https://doi.org/10.1057/ 10.1155/2021/6630379.
s41599-021-00965-w. 23. Collignon P, Beggs JJ, Walsh TR, Gandra S, Laxminarayan R. 2018. Anthro-
5. Van TTH, Yidana Z, Smooker PM, Coloe PJ. 2020. Antibiotic use in food pological and socioeconomic factors contributing to global antimicrobial
animals worldwide, with a focus on Africa: pluses and minuses. J Glob resistance: a univariate and multivariable analysis. Lancet Planet Health 2:
Antimicrob Resist 20:170–177. https://doi.org/10.1016/j.jgar.2019.07.031. e398–e405. https://doi.org/10.1016/S2542-5196(18)30186-4.
6. Martínez JL, Coque TM, Baquero F. 2015. What is a resistance gene? Rank- 24. van Boeckel TP, Pires J, Silvester R, Zhao C, Song J, Criscuolo NG, Gilbert
ing risk in resistomes. Nat Rev Microbiol 13:116–123. https://doi.org/10 M, Bonhoeffer S, Laxminarayan R. 2019. Global trends in antimicrobial re-
.1038/nrmicro3399. sistance in animals in low-and middle-income countries. Science 365:
7. Zhang AN, Gaston JM, Dai CL, Zhao S, Poyet M, Groussin M, Yin X, Li LG, van eaaw1944. https://doi.org/10.1126/science.aaw1944.
Loosdrecht MCM, Topp E, Gillings MR, Hanage WP, Tiedje JM, Moniz K, Alm EJ, 25. Laxminarayan R, Duse A, Wattal C, Zaidi AKM, Wertheim HFL, Sumpradit
Zhang T. 2021. An omics-based framework for assessing the health risk of anti- N, Vlieghe E, Hara GL, Gould IM, Goossens H, Greko C, So AD, Bigdeli M,
microbial resistance genes. Nat Commun 12:4765. https://doi.org/10.1038/ Tomson G, Woodhouse W, Ombaka E, Peralta AQ, Qamar FN, Mir F,
Kariuki S, Bhutta ZA, Coates A, Bergstrom R, Wright GD, Brown ED, Cars O.
s41467-021-25096-3.
2013. Antibiotic resistance—the need for global solutions. Lancet Infect
8. Nielsen TK, Browne PD, Hansen LH. 2022. Antibiotic resistance genes are
Dis 13:1057–1098. https://doi.org/10.1016/S1473-3099(13)70318-9.
differentially mobilized according to resistance mechanism. Gigascience
26. Iskandar K, Molinier L, Hallit S, Sartelli M, Hardcastle TC, Haque M, Lugova H,
11:giac072. https://doi.org/10.1093/gigascience/giac072.
Dhingra S, Sharma P, Islam S, Mohammed I, Naina Mohamed I, Hanna PA,
9. Karkman A, Berglund F, Flach CF, Kristiansson E, Larsson DGJ. 2020. Pre-
Hajj SE, Jamaluddin NAH, Salameh P, Roques C. 2021. Surveillance of antimi-
dicting clinical resistance prevalence using sewage metagenomic data.
crobial resistance in low- and middle-income countries: a scattered picture.

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

Commun Biol 3:711. https://doi.org/10.1038/s42003-020-01439-6.
Antimicrob Resist Infect Control 10:63. https://doi.org/10.1186/s13756-021
10. Gillings MR, Gaze WH, Pruden A, Smalla K, Tiedje JM, Zhu YG. 2015. Using
-00931-w.
the class 1 integron-integrase gene as a proxy for anthropogenic pollu-
27. Ambler RP. 1980. The structure of beta-lactamases. Philos Trans R Soc
tion. ISME J 9:1269–1279. https://doi.org/10.1038/ismej.2014.226.
Lond B Biol Sci 289:321–331. https://doi.org/10.1098/rstb.1980.0049.
11. Li Y, Yang X, Zhang J, Yang S, Zhang S, Chen M, Xue L, Ding Y, Zeng H, Gu
28. Bernabé KJ, Langendorf C, Ford N, Ronat JB, Murphy RA. 2017. Antimicrobial
Q, Zhang Y, Wei X, Wang J, Wu Q. 2021. Molecular characterisation of anti-
resistance in West Africa: a systematic review and meta-analysis. Int J Antimi-
microbial resistance determinants and class 1 integrons of Salmonella crob Agents 50:629–639. https://doi.org/10.1016/j.ijantimicag.2017.07.002.
enterica subsp. enterica serotype Enteritidis strains from retail food in 29. Ouchar Mahamat O, Kempf M, Lounnas M, Tidjani A, Hide M, Benavides JA,
China. Food Control 128:108191. https://doi.org/10.1016/j.foodcont.2021 Carrière C, Bañuls A-L, Jean-Pierre H, Ouedraogo A-S, Dumont Y, Godreuil S.
.108191. 2021. Epidemiology and prevalence of extended-spectrum b -lactamase-and
12. Gillings MR. 2014. Integrons: past, present, and future. Microbiol Mol Biol carbapenemase-producing Enterobacteriaceae in humans, animals and the
Rev 78:257–277. https://doi.org/10.1128/MMBR.00056-13. environment in West and Central Africa. Int J Antimicrob Agents 57:106203.
13. Ventola CL. 2015. The antibiotic resistance crisis. Part 1: causes and https://doi.org/10.1016/j.ijantimicag.2020.106203.
threats. P T 40:277–283. 30. Hendriksen RS, Munk P, Njage P, van Bunnik B, McNally L, Lukjancenko O,
14. World Health Organization. 2021. Global Antimicrobial Resistance and Use Röder T, Nieuwenhuijse D, Pedersen SK, Kjeldgaard J, Kaas RS, Clausen PTLC,
Surveillance System (GLASS) report 2021. https://www.who.int/publications/ Vogt JK, Leekitcharoenphon P, van de Schans MGM, Zuidema T, de Roda
i/item/9789240027336. Husman AM, Rasmussen S, Petersen B, Global Sewage Surveillance Project
15. World Health Organization. 2018. WHO report on surveillance of antibiotic Consortium, Amid C, Cochrane G, Sicheritz-Ponten T, Schmitt H, Alvarez JRM,
consumption: 2016–2018 early implementation. World Health Organization, Aidara-Kane A, Pamp SJ, Lund O, Hald T, Woolhouse M, Koopmans MP, Vigre
Geneva, Switzerland. https://apps.who.int/iris/bitstream/handle/10665/277359/ H, Petersen TN, Aarestrup FM. 2019. Global monitoring of antimicrobial re-
9789241514880-eng.pdf?sequence=1&isAllowed=y. sistance based on metagenomics analyses of urban sewage. Nat Commun
16. Suay-García B, Pérez-Gracia MT. 2019. Present and future of carbapenem- 10:1124. https://doi.org/10.1038/s41467-019-08853-3.
resistant Enterobacteriaceae (CRE) infections. Antibiotics 8:122. https://doi 31. Pärnänen KMM, Narciso-da-Rocha C, Kneis D, Berendonk TU, Cacace D,
.org/10.3390/antibiotics8030122. Do TT, Elpers C, Fatta-Kassinos D, Henriques I, Jaeger T, Karkman A,
17. Poirel L, Marqué S, Héritier C, Segonds C, Chabanon G, Nordmann P. Martinez JL, Michael SG, Michael-Kordatou I, O'Sullivan K, Rodriguez-
2005. OXA-58, a novel class D b -lactamase involved in resistance to car- Mozaz S, Schwartz T, Sheng H, Sørum H, Stedtfeld RD, Tiedje JM, Giustina
bapenems in Acinetobacter baumannii. Antimicrob Agents Chemother 49: SVD, Walsh F, Vaz-Moreira I, Virta M, Manaia CM. 2019. Antibiotic resist-
202–208. https://doi.org/10.1128/AAC.49.1.202-208.2005. ance in European wastewater treatment plants mirrors the pattern of clin-
18. Poirel L, Weldhagen GF, Naas T, de Champs C, Dove MG, Nordmann P. 2001. ical antibiotic resistance prevalence. Sci Adv 5:eaau9124. https://doi.org/
GES-2, a class A b -lactamase from Pseudomonas aeruginosa with increased 10.1126/sciadv.aau9124.
hydrolysis of imipenem. Antimicrob Agents Chemother 45:2598–2603. 32. European Centre for Disease Prevention and Control. 2020. Antimicrobial re-
https://doi.org/10.1128/AAC.45.9.2598-2603.2001. sistance in the EU/EEA (EARS-Net)—annual epidemiological report 2019.

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 15

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

https://www.ecdc.europa.eu/en/publications-data/surveillance-antimicrobial 49. van Duin D, Doi Y. 2017. The global epidemiology of carbapenemase-pro-
-resistance-europe-2019. ducing Enterobacteriaceae. Virulence 8:460–469. https://doi.org/10.1080/
33. Buelow E, Rico A, Gaschet M, Lourenço J, Kennedy SP, Wiest L, Ploy M-C, 21505594.2016.1222343.
Dagot C. 2020. Hospital discharges in urban sanitation systems: long- 50. Mushi MF, Mshana SE, Imirzalioglu C, Bwanga F. 2014. Carbapenemase
term monitoring of wastewater resistome and microbiota in relationship genes among multidrug resistant Gram negative clinical isolates from a
to their eco-exposome. Water Res X 7:100045. https://doi.org/10.1016/j tertiary hospital in Mwanza, Tanzania. Biomed Res Int 2014:303104.
.wroa.2020.100045. https://doi.org/10.1155/2014/303104.
34. Luk-In S, Pulsrikarn C, Bangtrakulnonth A, Chatsuwan T, Kulwichit W. 51. WHO Regional Office for Europe and European Centre for Disease Preven-
2017. Occurrence of a novel class 1 integron harboring qnrVC4 in Salmo- tion and Control. 2021. Surveillance of Antimicrobial Resistance in Europe,
nella Rissen. Diagn Microbiol Infect Dis 88:282–286. https://doi.org/10 2020. Executive Summary. https://www.who.int/europe/publications/i/item/
.1016/j.diagmicrobio.2017.03.016. 9789289056298.
35. Xia R, Guo X, Zhang Y, Xu H. 2010. qnrVC-like gene located in a novel com- 52. Räisänen K, Lyytikäinen O, Kauranen J, Tarkka E, Forsblom-Helander B,
plex class 1 integron harboring the ISCR1 element in an Aeromonas punc- Grönroos JO, Vuento R, Arifulla D, Sarvikivi E, Toura S, Jalava J. 2020. Mo-
tata strain from an aquatic environment in Shandong Province, China. lecular epidemiology of carbapenemase-producing Enterobacterales in
Antimicrob Agents Chemother 54:3471–3474. https://doi.org/10.1128/ Finland, 2012–2018. Eur J Clin Microbiol Infect Dis 39:1651–1656. https://
AAC.01668-09. doi.org/10.1007/s10096-020-03885-w.
36. Fernandes AD, Reid JNS, Macklaim JM, McMurrough TA, Edgell DR, Gloor 53. Yatsunenko T, Rey FE, Manary MJ. 2012. Human gut microbiome viewed
GB. 2014. Unifying the analysis of high-throughput sequencing datasets: across age and geography. Nature 486:222–227. https://doi.org/10.1038/
characterizing RNA-seq, 16S rRNA gene sequencing and selective growth nature11053.
experiments by compositional data analysis. Microbiome 2:15. https://doi 54. Thompson LR, Sanders JG, McDonald D, Amir A, Ladau J, Locey KJ, Prill RJ,
.org/10.1186/2049-2618-2-15. Tripathi A, Gibbons SM, Ackermann G, Navas-Molina JA, Janssen S,
37. Gloor GB, Macklaim JM, Pawlowsky-Glahn V, Egozcue JJ. 2017. Micro- Kopylova E, Vázquez-Baeza Y, González A, Morton JT, Mirarab S, Zech Xu
biome datasets are compositional: and this is not optional. Front Micro- Z, Jiang L, Haroon MF, Kanbar J, Zhu Q, Jin Song S, Kosciolek T, Bokulich
biol 8:2224. https://doi.org/10.3389/fmicb.2017.02224. NA, Lefler J, Brislawn CJ, Humphrey G, Owens SM, Hampton-Marcell J,
38. Ma L, Li AD, Le Yin X, Zhang T. 2017. The prevalence of integrons as the Berg-Lyons D, McKenzie V, Fierer N, Fuhrman JA, Clauset A, Stevens RL,
carrier of antibiotic resistance genes in natural and man-made environ- Shade A, Pollard KS, Goodwin KD, Jansson JK, Gilbert JA, Knight R, Earth
ments. Environ Sci Technol 51:5721–5728. https://doi.org/10.1021/acs.est Microbiome Project Consortium. 2017. A communal catalogue reveals
.6b05887. Earth’s multiscale microbial diversity. Nature 551:457–463. https://doi
39. Tacconelli E, Carrara E, Savoldi A, Kattula D, Burkert F. 2017. Global priority .org/10.1038/nature24621.
list of antibiotic-resistant bacteria to guide research, discovery, and devel- 55. Oyem IM, Oyem HH, Atuanya EI. 2020. Molecular characterisation of bac-
opment of new antibiotics. http://remed.org/wp-content/uploads/2017/ teria strains of septic tank sewage samples from related sites in Delta and
Edo States of Nigeria using 16S rRNA denaturing gradient gel electropho-
03/lobal-priority-list-of-antibiotic-resistant-bacteria-2017.pdf. Accessed 3
resis (DGGE). Afr J Environ Sci Technol 14:132–138. https://doi.org/10
February 2019.
.5897/AJEST2020.2851.
40. Rice EW, Wang P, Smith AL, Stadler LB. 2020. Determining hosts of antibi-
56. Pärnänen K, Karkman A, Hultman J, Lyra C, Bengtsson-Palme J, Larsson
otic resistance genes: a review of methodological advances. Environ Sci
DGJ, Rautava S, Isolauri E, Salminen S, Kumar H, Satokari R, Virta M. 2018.
Technol Lett 7:282–291. https://doi.org/10.1021/acs.estlett.0c00202.
Maternal gut and breast milk microbiota affect infant gut antibiotic resis-
41. Araújo S, Sousa M, Tacão M, Baraúna RA, Silva A, Ramos R, Alves A, Manaia
tome and mobile genetic elements. Nat Commun 9:389. https://doi.org/
CM, Henriques I. 2021. Carbapenem-resistant bacteria over a wastewater
10.1038/s41467-018-06393-w.
treatment process: carbapenem-resistant Enterobacteriaceae in untreated
57. Olumuyiwa Olaitan A, Dandachi I, Alexandra Baron S, Daoud Z, Morand S,
wastewater and intrinsically-resistant bacteria in final effluent. Sci Total Envi-
Rolain JM. 2021. Banning colistin in feed additives: a small step in the

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

ron 782:146892. https://doi.org/10.1016/j.scitotenv.2021.146892.
right direction. Lancet Infect Dis 21:29–30. https://doi.org/10.1016/S1473
42.  ski W, Korzeniewska E, Harnisz M, Drzymała J, Felis E, Bajkacz S. 2021.
Zielin
-3099(20)30915-4.
Wastewater treatment plants as a reservoir of integrase and antibiotic resist-
58. Olowo-Okere A, Yacouba A. 2020. Molecular mechanisms of colistin re-
ance genes—an epidemiological threat to workers and environment. Envi-
sistance in Africa: a systematic review of literature. Germs 10:367–379.
ron Int 156:106641. https://doi.org/10.1016/j.envint.2021.106641. https://doi.org/10.18683/germs.2020.1229.
43. Anago E, Ayi-Fanou L, Akpovi CD, Hounkpe WB, Agassounon-Djikpo 59. Kneis D, Berendonk TU, Heß S. 2019. High prevalence of colistin resistance
Tchibozo M, Bankole HS, Sanni A. 2012. Antibiotic resistance and geno- genes in German municipal wastewater. Sci Total Environ 694:133454.
type of beta-lactamase producing Escherichia coli in nosocomial infec- https://doi.org/10.1016/j.scitotenv.2019.07.260.
tions in Cotonou, Benin. Ann Clin Microbiol Antimicrob 14:5. https://doi 60. Ngbede EO, Poudel A, Kalalah A, Yang Y, Adekanmbi F, Adikwu AA,
.org/10.1186/s12941-014-0061-1. Adamu AM, Mamfe LM, Daniel ST, Useh NM, Kwaga JKP, Adah MI, Kelly P,
44. Figueiredo S, Bonnin RA, Poirel L, Duranteau J, Nordmann P. 2012. Identi- Butaye P, Wang C. 2020. Identification of mobile colistin resistance genes
fication of the naturally occurring genes encoding carbapenem-hydrolys- (mcr-1.1, mcr-5 and mcr-8.1) in Enterobacteriaceae and Alcaligenes faecalis
ing oxacillinases from Acinetobacter haemolyticus, Acinetobacter johnsonii, of human and animal origin, Nigeria. Int J Antimicrob Agents 56:106108.
and Acinetobacter calcoaceticus. Clin Microbiol Infect 18:907–913. https:// https://doi.org/10.1016/j.ijantimicag.2020.106108.
doi.org/10.1111/j.1469-0691.2011.03708.x. 61. Borowiak M, Fischer J, Hammerl JA, Hendriksen RS, Szabo I, Malorny B.
45. Périchon B, Goussard S, Walewski V, Krizova L, Cerqueira G, Murphy C, 2017. Identification of a novel transposon-associated phosphoethanol-
Feldgarden M, Wortman J, Clermont D, Nemec A, Courvalin P. 2014. Iden- amine transferase gene, mcr-5, conferring colistin resistance in d-tartrate
tification of 50 class D b -lactamases and 65 Acinetobacter-derived cepha- fermenting Salmonella enterica subsp. enterica serovar Paratyphi B. J Anti-
losporinases in Acinetobacter spp. Antimicrob Agents Chemother 58: microb Chemother 72:3317–3324. https://doi.org/10.1093/jac/dkx327.
936–949. https://doi.org/10.1128/AAC.01261-13. 62. Borowiak M, Hammerl JA, Deneke C, Fischer J, Szabo I, Malorny B. 2019. Char-
46. Naas T, Oueslati S, Bonnin RA, Dabos ML, Zavala A, Dortet L, Retailleau P, acterization of mcr-5-harboring Salmonella enterica subsp. enterica serovar
Iorga BI. 2017. Beta-lactamase database (BLDB)—structure and function. J Typhimurium isolates from animal and food origin in Germany. Antimicrob
Enzyme Inhib Med Chem 32:917–919. https://doi.org/10.1080/14756366 Agents Chemother 63:e00063-19. https://doi.org/10.1128/AAC.00063-19.
.2017.1344235. 63. Guo S, Tay MYF, Thu AK, Seow KLG, Zhong Y, Ng LC, Schlundt J. 2019. Conju-
47. Olalekan A, Onwugamba F, Iwalokun B, Mellmann A, Becker K, Schaumburg gative IncX1 plasmid harboring colistin resistance gene mcr-5.1 in Escherichia
F. 2020. High proportion of carbapenemase-producing Escherichia coli and coli isolated from chicken rice retailed in Singapore. Antimicrob Agents Che-
Klebsiella pneumoniae among extended-spectrum b -lactamase-producers in mother 63:e01043-19. https://doi.org/10.1128/AAC.01043-19.
Nigerian hospitals. J Glob Antimicrob Resist 21:8–12. https://doi.org/10.1016/ 64. Peace Hounkpe S, Codjo Adjovi E, Crapper M, Awuah E. 2014. Wastewater
j.jgar.2019.09.007. management in Third World cities: case study of Cotonou. Benin J Environ
48. Manenzhe RI, Zar HJ, Nicol MP, Kaba M. 2015. The spread of carbapene- Protect 5:387–399. https://doi.org/10.4236/jep.2014.55042.
mase-producing bacteria in Africa: a systematic review. J Antimicrob Che- 65. Totin HS, Ernest A, Odoulami L, Edorh PA, Boukari M, Boko M. 2013.
mother 70:23–40. https://doi.org/10.1093/jac/dku356. Groundwater pollution and the safe water supply challenge in Cotonou

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 16

HWW Resistomes in Benin, Burkina Faso, and Finland mSphere

town, Benin (West Africa). IAHS-AISH Proc Rep 361:191–196. https://doi 77. Oksanen J, Blanchet GF, Friendly M, Kindt R, Legendre P, McGlinn D,
.org/10.13140/RG.2.1.2668.0168. Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E,
66. Bougnom BP, Zongo C, McNally A, Ricci V, Etoa FX, Thiele-Bruhn S, Wagner H. 2022. vegan: Community Ecology Package. R package version
Piddock LJV. 2019. Wastewater used for urban agriculture in West Africa 2.5–7. https://cran.r-project.org/package=vegan.
as a reservoir for antibacterial resistance dissemination. Environ Res 168: 78. Thomas AM, Segata N. 2019. Multiple levels of the unknown in microbiome
14–24. https://doi.org/10.1016/j.envres.2018.09.022. research. BMC Biol 17:48. https://doi.org/10.1186/s12915-019-0667-z.
67. Bougnom BP, Thiele-Bruhn S, Ricci V, Zongo C, Piddock LJV. 2020. Raw waste- 79. McMurdie PJ, Holmes S. 2013. Phyloseq: an R package for reproducible
water irrigation for urban agriculture in three African cities increases the interactive analysis and graphics of microbiome census data. PLoS One 8:
abundance of transferable antibiotic resistance genes in soil, including those e61217. https://doi.org/10.1371/journal.pone.0061217.
encoding extended spectrum b -lactamases (ESBLs). Sci Total Environ 698: 80. Fu L, Niu B, Zhu Z, Wu S, Li W. 2012. Sequence analysis CD-HIT: acceler-
134201. https://doi.org/10.1016/j.scitotenv.2019.134201. ated for clustering the next-generation sequencing data. Bioinformatics
68. Mölder F, Jablonski KP, Letcher B, Hall MB, Tomkins-Tinch CH, Sochat V, 28:3150–3152. https://doi.org/10.1093/bioinformatics/bts565.
Forster J, Lee S, Twardziok SO, Kanitz A, Wilm A, Holtgrewe M, Rahmann 81. Li D, Liu CM, Luo R, Sadakane K, Lam TW. 2015. MEGAHIT: an ultra-fast sin-
S, Nahnsen S, Köster J. 2021. Sustainable data analysis with Snakemake. gle-node solution for large and complex metagenomics assembly via suc-
F1000Res 10:33. https://doi.org/10.12688/f1000research.29032.1. cinct de Bruijn graph. Bioinformatics 31:1674–1676. https://doi.org/10
69. Andrews S. 2010. FastQC: a quality control tool for high throughput .1093/bioinformatics/btv033.
sequence data. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/. 82. Eren AM, Kiefl E, Shaiber A, Veseli I, Miller SE, Schechter MS, Fink I, Pan JN,
70. Ewels P, Magnusson M, Lundin S, Käller M. 2016. Data and text mining Mul- Yousef M, Fogarty EC, Trigodet F, Watson AR, Esen ÖC, Moore RM,
tiQC: summarize analysis results for multiple tools and samples in a single Clayssen Q, Lee MD, Kivenson V, Graham ED, Merrill BD, Karkman A,
report. Bioinformatics 32:3047–3048. https://doi.org/10.1093/bioinformatics/ Blankenberg D, Eppley JM, Sjödin A, Scott JJ, Vázquez-Campos X, McKay
btw354. LJ, McDaniel EA, Stevens SLR, Anderson RE, Fuessel J, Fernandez-Guerra
71. Marcel M. 2011. Cutadapt removes adapter sequences from high-throughput A, Maignien L, Delmont TO, Willis AD. 2021. Community-led, integrated,
sequencing reads. EMBnet J 17:10–12. https://doi.org/10.14806/ej.17.1.200. reproducible multi-omics with anvi’o. Nat Microbiol 6:3–6. https://doi
72. Langmead B, Salzberg SL. 2012. Fast gapped-read alignment with Bowtie .org/10.1038/s41564-020-00834-3.
2. Nat Methods 9:357–359. https://doi.org/10.1038/nmeth.1923. 83. Wick RR, Schultz MB, Zobel J, Holt KE. 2015. Bandage: interactive visualization
73. Bortolaia V, Kaas RS, Ruppe E, Roberts MC, Schwarz S, Cattoir V, Philippon of de novo genome assemblies. Bioinformatics 31:3350–3352. https://doi
A, Allesoe RL, Rebelo AR, Florensa AF, Fagelhauer L, Chakraborty T, .org/10.1093/bioinformatics/btv383.
Neumann B, Werner G, Bender JK, Stingl K, Nguyen M, Coppens J, Xavier 84. R Foundation for Statistical Computing. 2022. R: a language and environment
BB, Malhotra-Kumar S, Westh H, Pinholt M, Anjum MF, Duggett NA, for statistical computing. https://www.R-project.org/. https://www.r-project.org/.
Kempf I, Nykäsenoja S, Olkkola S, Wieczorek K, Amaro A, Clemente L, 85. Kassambara A. 2022. ggpubr: ‘ggplot2’ Based Publication Ready Plots. https://
Mossong J, Losch S, Ragimbeau C, Lund O, Aarestrup FM. 2020. ResFinder cran.r-project.org/package=ggpubr.
4.0 for predictions of phenotypes from genotypes. J Antimicrob Chemo- 86. Quinn TP, Erb I, Gloor G, Notredame C, Richardson MF, Crowley TM. 2019.
ther 75:3491–3500. https://doi.org/10.1093/jac/dkaa345. A field guide for the compositional analysis of any-omics data. Giga-
74. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, science 8:giz107. https://doi.org/10.1093/gigascience/giz107.
Abecasis G, Durbin R, 1000 Genome Project Data Processing Subgroup. 87. Barnett D, Arts I, Penders J. 2021. microViz: an R package for microbiome
2009. The Sequence Alignment/Map format and SAMtools. Bioinformatics data visualization and statistics. J Open Source Softw 6:3201. https://doi
25:2078–2079. https://doi.org/10.1093/bioinformatics/btp352. .org/10.21105/joss.03201.
75. Beghini F, McIver LJ, Blanco-Míguez A, Dubois L, Asnicar F, Maharjan S, 88. Wickham H. 2022. ggplot2: elegant graphics for data analysis. https://
Mailyan A, Manghi P, Scholz M, Thomas AM, Valles-Colomer M, Weingart ggplot2.tidyverse.org.
G, Zhang Y, Zolfo M, Huttenhower C, Franzosa EA, Segata N. 2021. Inte- 89. Pedersen TL. 2020. patchwork: the composer of plots. https://cran.r-project
grating taxonomic, functional, and strain-level profiling of diverse micro- .org/package=patchwork.
bial communities with bioBakery 3. eLife 10:e65088. https://doi.org/10 90. South A. 2017. rnaturalearth: world map data from Natural Earth. https://

Downloaded from https://journals.asm.org/journal/msphere on 02 February 2023 by 41.138.89.207.

.7554/eLife.65088. cran.r-project.org/package=rnaturalearth.
76. Bengtsson-Palme J, Hartmann M, Eriksson KM, Pal C, Thorell K, Larsson 91. Dunnington D. 2022. ggspatial: spatial data framework for ggplot2. https://
DGJ, Nilsson RH. 2015. metaxa2: improved identification and taxonomic paleolimbot.github.io/ggspatial/.
classification of small and large subunit rRNA in metagenomic data. Mol 92. Becker RA, Wilks AR, Brownrigg R, Minka TP, Deckmyn A. 2021. maps: draw
Ecol Resour 15:1403–1414. https://doi.org/10.1111/1755-0998.12399. geographical maps. https://cran.r-project.org/package=maps.

Month YYYY Volume XX Issue XX 10.1128/msphere.00538-22 17