The Role of the Natural Aquatic Environment in the Dissemination of Extended Spectrum Beta-Lactamase and Carbapenemase Encoding Genes: A Scoping Review

1 The natural aquatic environment is a significant contributor to the development and circulation 2 of clinically significant antibiotic resistance genes (ARGs). The potential for the aquatic 3 environment to act as a reservoir for ARG accumulation in areas receiving anthropogenic 4 contamination has been thoroughly researched. However, the emergence of novel ARGs in the 5 absence of external influences, as well as the capacity of environmental bacteria to disseminate 6 ARGs via mobile genetic elements remain relatively unchallenged. In order to address these 7 knowledge gaps, this scoping literature review was established focusing on the detection of two 8 important and readily mobile ARGs, namely, extended spectrum beta-lactamase (ESBL) and 9 carbapenemase genes. This review included 41 studies from 19 different countries. A range of 10 1 Room 2008, Clinical Science Institute, National University of Ireland Galway, Costello Road, Shantalla, Galway, Ireland. H91V4AY 2 different water bodies including rivers (n=26), seawaters (n=6) and lakes (n=3), amongst others, 11 were analysed in the included studies. ESBL genes were reported in 29/41 (70.7%) studies, while 12 carbapenemase genes were reported in 13/41 (31.7%), including joint reporting in 9 studies. The 13 occurrence of mobile genetic elements was evaluated, which included the detection of integrons 14 (n=22), plasmids (n=18), insertion sequences (n=4) and transposons (n=3). The ability of 15 environmental bacteria to successfully transfer resistance genes via conjugation was also 16 examined in 11 of the included studies. The findings of this scoping review expose the presence 17 of clinically significant ARGs in the natural aquatic environment and highlights the potential 18 ability of environmental isolates to disseminate these genes among different bacterial species. As 19 such, the results presented demonstrate how anthropogenic point discharges may not act as the 20 sole contributor to the development and spread of clinically significant antibiotic resistances. A 21 number of critical knowledge gaps in current research were also identified. Key highlights 22 include the limited number of studies focusing on antibiotic resistance in uncontaminated aquatic 23 environments as well as the lack of standardisation among methodologies of reviewed 24 investigations. 25

different water bodies including rivers (n=26), seawaters (n=6) and lakes (n=3), amongst others, 11 were analysed in the included studies. ESBL genes were reported in 29/41 (70.7%) studies, while 12 carbapenemase genes were reported in 13/41 (31.7%), including joint reporting in 9 studies. The 13 occurrence of mobile genetic elements was evaluated, which included the detection of integrons 14 (n=22), plasmids (n=18), insertion sequences (n=4) and transposons (n=3). The ability of 15 environmental bacteria to successfully transfer resistance genes via conjugation was also 16 examined in 11 of the included studies. The findings of this scoping review expose the presence 17 of clinically significant ARGs in the natural aquatic environment and highlights the potential 18 ability of environmental isolates to disseminate these genes among different bacterial species. As Antibiotic resistance is recognised as a major threat to public health. Bacteria utilise a range of 33 mechanisms to evade the effects of antibiotics leading to challenges in clinical infection 34 treatments. These range from non-specific processes including increased efflux pump activity or 35 downregulation of porin channels, to the production of enzymes that specifically target and 36 inactivate antibiotics (Peterson and Kaur, 2018). As bacteria continue to adapt to the presence of 37 antibiotics through the acquisition of antibiotic resistance genes (ARGs) via mobile genetic 38 elements (MGEs), available antibiotics are becoming less effective. Due to the growing 39 limitation of treatment options, older antibiotics such as colistin, which can cause negative side 40 effects (Morrill et al., 2015), are being employed to treat infections caused by bacteria that are 41 resistant to last resort antibiotics. 42 The increase in the proportion of serious infections associated with extended spectrum beta-43 lactamase (ESBL) and carbapenemase producing organisms is a significant clinical concern. 44 ESBL enzymes have evolved by point mutations occurring in beta-lactamase genes such as 45 blaTEM-1, blaTEM-2 and blaSHV-1 (Shaikh et al., 2015). The Ambler classification is used to 46 categorise these beta-lactamases based on their amino acid sequence (Ambler, 1980). Many of 47 2 Methods 146

Research Question and Database Queries 147
The following research question was formulated to focus and direct the scoping review: The field tag 'TS' was applied to the Web of Science database to focus the search string on the 159 topic of the articles. The search string was adapted to the Scopus database using TITLE-ABS-160 KEY to identify the search terms in titles, abstracts and keywords. Relevant subject areas and 161 source types were also selected and applied in each database to limit the numbers of irrelevant 162 articles retrieved. All articles obtained from the searches were exported to Endnote and 163 duplicates were removed. 164 Given the limited number of studies that tested water bodies reported to be strictly "free" from 188 contamination, it was decided that 'urban' waters would be included if investigations did not The genomic detection of ESBL and/or carbapenemase genes. 201

Additional Sources
ii) The genomic detection of MGEs or demonstration of transferability of resistance 202 genes by experimentation. 203 In the case of articles where information on the variant of the beta-lactamase gene detected using 204 genotypic methods was lacking, phenotypic screening such as antibiotic susceptibility testing 205 was used to determine if the genes detected were true ESBL/ carbapenemase producers. As an 206 example, the blaSHV and blaTEM genes have some variants allowing for classifications as ESBL 207 producers (blaTEM-3 and blaSHV-12) and some which are classified simply as beta-lactamases 208 (blaTEM-1, blaSHV-1). 209 3 Results 210

Screening Stages for Article Inclusion 211
A total number of 1415 articles were identified and subject to the first screening stage following 212 application of the search string to the 5 main databases including PubMed, Scopus, Web of 213 Science, MEDLINE and EMBASE (Figure 1). An additional 5 peer-reviewed articles were 214 identified from grey literature queries placing the total number of articles at 1420. Following 215 initial title and abstract screening, 1313 articles were excluded, with a total of 107 articles 216 subject to full text review. This was narrowed down to 34 following exclusion of 73 articles that 217 failed to meet pre-established inclusion/exclusion criteria. The most common exclusion factor 218 was local presence of point source discharges (n=24). A recurrent feature during this screening 219 phase was the lack of detail provided in the results of several investigations which often resulted 220 in article exclusion. Several investigations failed to link the results presented to specific site 221 types analyzed (e.g. polluted, pristine), but instead provided a general summary of all resistance 222 genes detected (e.g. Dhawde et al., 2018). In other cases, due to the extensive nature of the data 223 generated, primarily a feature of investigations employing metagenomics, ARGs were often 224 summarised to an overly generic level. In some instances, only the quantity of beta-lactamase 225 resistance genes that were present in a given sample were discussed rather than providing details 226 on the different types of resistance genes or their location within the chromosome or mobile 227 genetic element (Garner et al., 2016). As such, it was not possible to determine the types of beta-228 lactamase genes detected. Bibliography screening of the 34 included papers following the pre-229 defined inclusion/ exclusion criteria identified an additional 14 potentially suitable articles. 6 of 230 these papers were included for analysis increasing the total number of articles for inclusion to 40.

Data Extraction 237
All 41 identified articles were subject to data extraction based on a number of pre-established 238 data fields (Table 2). Primary data extraction fields included (i) country of origin for sample(s) 239 analysed, (ii) type of sample analysed (e.g. seawater/river/lake), (iii) types of mobile genetic 240 elements detected (e.g. plasmids, integrons, etc.), (iv) the application of conjugation transfer and 241 whether it was successful at transferring the ARGs of interest, and (v) ESBL and/or 242 carbapenemase gene variants detected. A range of data fields supplementing the information 243 presented in Table 2 and pertaining local environments (e.g. local discharges downstream of 244 sampling points), bacterial species detected and further information on the types of mobile 245 genetic elements detected (e.g. plasmid incompatibility groups, integron class) are outlined in 246 Supplementary

Summary Analysis of Included Studies 256
The geographical distribution of included investigations is shown in Figure 2. Overall, studies 257 derived from nineteen different countries across five continents. As such, the data presented is  Table 2). These bacteria are often associated 328 with clinical infections and as a result, some methodologies employed a selection bias using 329 selective broths and agars, (see Supplementary Table 3). A range of natural water bodies were 330 investigated in the reviewed studies, however, the majority of studies focused on lotic systems 331 (n=29; 70.7%). The high incidence of studies based on these systems may be associated with the 332 smaller dimensions of rivers/streams in comparison to larger marine and coastal water bodies, 333 equating to a lesser dilutional effect on viable antibiotic resistant organisms. Additionally, lotic 334 systems are often located within urban and agricultural landscapes, with river-sourced water 335 often used as a domestic water supply and thus representing a potential pathway for antibiotic 336 resistance transmission to humans. As indicated above, based on the lack of studies investigating 337 "pristine" aquatic environments, water bodies labelled as 'urban' were also included in the 338 review protocol if no local point sources were described.

Research highlights and recommendations 461
The lack of standardised methodology adopted among identified investigations represents a 462 significant knowledge gap and a challenge for the interpretation of collated data. In particular, 463 issues with method sensitivity and the lack of valid comparisons to analyse reported results are 464 apparent. As such, a strong argument is made for future investigations to adopt a more 465 standardised methodological approach that is sensitive enough to detect low levels of antibiotic 466 resistance genes. Similarly, future research should also take associated human risks into 467 consideration, (e.g. potential consumption volumes during recreational activities and infectious 468 dose of organisms). Evidently, methodology criteria applied in this review attempted to 469 standardise articles with highly variable collection volumes and processing techniques as much Of key relevance is the high prevalence of phenotypic screening methods employed in the 503 majority of reviewed investigations. This factor limited the detection of antibiotic resistance 504 genes to those that reside within viable bacteria that are easily cultured. However, reportedly less 505 than 1% of environmental bacteria can be easily cultured using standard laboratory techniques 506 (Allen et al., 2010). Only four included studies performed PCR/sequencing of the sample without 507 prior culturing of organisms (Supplementary Table 3) which enabled detection and analysis of 508 fastidious bacteria. Overall, current research tends to exclude less clinically significant and 509 fastidious environmental bacteria, which as a result, may go unnoticed as harbouring MGEs 510 capable of spreading antibiotic resistance elements. Ideally, simultaneous application of both 511 phenotypic culture methods and molecular analysis of environmental samples would need to be 512 employed to generate complete resistance profiles inclusive of genetic composition as well as 513 phenotypic expression. 514

5
Conclusions 515 The results presented highlight the importance of aquatic environments as substantial reservoirs 516 of ESBL and carbapenemase ARGs. As such, point contaminant sources may not be the sole 517 contributors to the presence of antibiotic resistance in the aquatic environment. Additionally, 518 collated data serves to demonstrate the potential for interspecies transference of ARGs among 519 environmental bacteria to potentially pathogenic species. The following recommendations are 520 made based on their potential to contribute to our current understanding of both the prevalence 521 and risk factors associated with ARGs in aquatic environments: 522 • More investigations are required involving routine screening of antibiotic resistance 523 genes in water bodies, particularly those considered to be disassociated from direct point 524 contaminant sources. 525 • A highly sensitive and standardised methodology which enables valid and robust 526 comparisons among study outcomes. 527 • Incorporation of genomic screening and culture-based analyses aiming to mitigate the 528 selective bias imposed by culturing. 529 • Detailed reporting in investigations, particularly in terms of study site characteristics (e.g.