Metagenomic Insight into Microbiome and Antibiotic Resistance Genes of High Clinical Concern in Urban and Rural Hospital Wastewater of Northern India Origin: a Major Reservoir of Antimicrobial Resistance

ABSTRACT India is one of the largest consumers and producers of antibiotics and a hot spot for the emergence and proliferation of antimicrobial resistance genes (ARGs). Indian hospital wastewater (HWW) accumulates ARGs from source hospitals and often merges with urban wastewater, with the potential for environmental and human contamination. Despite its putative clinical importance, there is a lack of high-resolution resistome profiling of Indian hospital wastewater, with most studies either relying on conventional PCR-biased techniques or being limited to one city. In this study, we comprehensively analyzed antibiotic resistomes of wastewater from six Indian hospitals distributed in rural and urban areas of northern India through shotgun metagenomics. Our study revealed the predominance of ARGs against aminoglycoside, macrolide, carbapenem, trimethoprim, and sulfonamide antibiotics in all the samples through both read-based analysis and assembly-based analysis. We detected the mobile colistin resistance gene mcr-5.1 for the first time in Indian hospital sewage. blaNDM-1 was present in 4 out of 6 samples and was carried by Pseudomonas aeruginosa in HWW-2, Klebsiella pneumoniae in HWW-4 and HWW-6, and Acinetobacter baumanii in HWW-5. Most ARGs were plasmid-mediated and hosted by Proteobacteria. We identified virulence factors and transposable elements flanking the ARGs, highlighting the role of horizontal gene transmission of ARGs. IMPORTANCE There is a paucity of research on detailed antibiotic resistome and microbiome diversity of Indian hospital wastewater. This study reports the predominance of clinically concerning ARGs such as the beta-lactamases blaNDM and blaOXA and the colistin resistance gene mcr and their association with the microbiome in six different Indian hospital wastewaters of both urban and rural origin. The abundance of plasmid-mediated ARGs and virulence factors calls for urgent AMR crisis management. The lack of proper wastewater management strategies meeting international standards and open drainage systems further complicates the problem of containing the ARGs at these hospitals. This metagenomic study presents the current AMR profile propagating in hospital settings in India and can be used as a reference for future surveillance and risk management of ARGs in Indian hospitals.

A ntimicrobial resistance (AMR) is a global public health emergency. In 2019, an estimated 5,000,000 deaths were associated with bacterial AMR, exerting the highest burdens in low-and middle-income countries (LMIC) (1). Exacerbating the problem, a rise in AMR has been accompanied by a reduction in the number of new antibiotics approved for human use (2,3). The World Health Organization's "One Health" approach emphasizes the close association of human, animal, and environmental health (4). Reliable and accurate surveillance is critical for characterizing the risk of AMR in a given region, tracking the spread of specific antibiotic resistance genes (ARG) geographically and over time, identifying new ARGs, and supporting preventative measures and interventions against multidrug-resistant (MDR) pathogens.
Hospitals are important reservoirs and vectors of AMR, where the frequent and persistent use of antimicrobials selects for MDR pathogens that can cause hospitalacquired infections (5,6). Hospital wastewater (HWW) receives liquid medical waste, and the excrement of patients, visitors, and health care professionals, and as such, contains hazardous chemicals, pharmaceutical residues, and human pathogens (7). In turn, it contains a greater diversity and abundance of ARGs than other wastewater systems (8). Additionally, in many settings, particularly in LMICs, HWW is released into municipal sewage systems without treatment (7). This, coupled with open drainage systems and inefficient wastewater treatment plants, can lead to the dissemination of HWW contents, including antimicrobial residues and MDR pathogens, into the local environment and community (9). Metagenomic surveillance of HWW using short-read nextgeneration sequencing data is an efficient approach for assessing the overall AMR burden in a given hospital (10)(11)(12). Unlike traditional culture-and PCR-based approaches, which are laborious and limited to selected taxa and ARGs, metagenomics can quantify thousands of species, ARGs, and virulence factors (VFs) from a single sample at a relatively low cost. Further, surveillance of sewage instead of samples from human patients has the additional advantage of being easily obtained and analyzed without ethical review. However, current research on AMR in HWW has mainly focused on highincome countries, with limited data reported for LMICs (13).
India is one of the largest consumers and producers of antibiotics and is a hot spot for AMR (14). Antibiotic usage in hospitals in Asia is comparatively higher than in European hospitals (15). A study based on data on patient-level antibiotic consumption from 209 surveys and 284,045 children (aged ,5 years), collected over 19 years and covering 101 countries, reported a dramatic surge in the median national antibiotic usage in India from 48% in the year 2000 to 67% in 2018. The lack of awareness about judicious antibiotic usage and the detrimental consequences of "over the counter" medication has caused a notable increase in the consumption of fluoroquinolone and third-generation cephalosporin in India from 2000 to 2018 (16). According to the WHO, by the year 2023, 60% of the total antibiotic consumption in a country should be constituted by WHO-Access group antibiotics, the antibiotics that are efficacious against commonly encountered susceptible pathogens, and there are low chances of resistance development against these antibiotics. In 2015, only 30% of the total antibiotic consumption was covered by the WHO-Access group of antibiotics in India (17). Several factors favor the spread of AMR in Indian hospitals such as over-the-counter drug availability, low doctor-to-patient and nurse-to-patient ratios, antibiotics prescription by informal health care providers, and lack of infection prevention and control guidelines (18)(19)(20). India is also the hub for the manufacturing and distribution of generic antibiotics for global use, and this industrial-scale manufacturing and its associated waste likely further contribute to the development and spread of antimicrobial resistance (21). The high usage of antimicrobials in Indian hospitals undoubtedly selects for AMR. However, there are limited data on the ARG burden in Indian HWW, with previous reports limited to select ARGs and culturable bacteria and/or to a single city (22,23).
In this study, we use shotgun metagenome sequencing to characterize the microbiomes and resistomes of HWW from six hospitals in five cities across northern India. To the best of our knowledge, this is the first study from India, comprised of samples from rural and urban hospitals, all of which were equipped with bed facilities and admitted severe infection cases. Our findings suggest that the diversity and abundance of AMR in Indian hospitals are greater than previously thought, which poses an immediate potential risk to patients and the surrounding communities. Wastewater bacteria Cloacibacterium and Flavobacterium and environmental bacteria Pseudomonas were the predominant genera in HWW-1 to HWW-3, whereas Acinetobacter was dominant in HWW-4 to HWW-6. The Shannon diversity index showed similar microbiome diversity in all the samples (Fig. 3G).
Prevalence of ARGs of urgent clinical concern in Indian HWW resistome. In total, we identified 183 unique AMR determinants using ShortBRED (25). HWW-1 to HWW-3 had lower ARG abundances, quantified as reads per kilobase million (RPKM), compared to HWW-4 to HWW-6 ( Fig. 4A). However, the relative abundance of ARGs to specific antibiotic drug classes was comparable across sampling sites (Fig. 5A). We detected the lowest abundance of ARGs in HWW-3, and this is also reflected in as a lower ARG richness, but the Shannon diversity of all HWW samples was comparable (Fig. 5B).
The WHO and other literature have identified several ARGs as being of urgent clinical concern because they have been associated with antibiotic treatment failure in hospitals and/or are widespread on MGEs (26). Many of these were identified in all 6 HWW samples, namely: the aminoglycoside-modifying enzymes aac(69), aac(3), aph (39), and aph(6), the carbapenemase bla OXA , the beta-lactamase bla TEM , the trimethoprim resistance gene dfrA1, the macrolide-lincosamide-streptogramin (MLS) resistance gene Erm, and the sulfonamide resistance gene sul1 (Fig. 4B). Further, most of these genes are also among the top 10 most abundant ARGs in each sample (Fig. 4C), indicating that they are not only present but highly prevalent in Indian HWW, and by extension in the hospitals themselves. In addition, numerous other clinically important ARGs were identified in a subset of HWW samples (Fig. 4B), highlighting the severity of AMR propagating in the Indian HWW environment.
Other ARGs of rising concern detected include the tetracycline-inactivating enzyme tet(X), which has been previously identified in one isolate from India (27,28); however, its presence in five of six HWW samples suggests it may be more prevalent than previously thought. We also detected mobile colistin resistance genes mcr-3 and mcr-5 in HWW-6, which represents the first time mcr-5 was identified in Indian hospital wastewater (29).
To study ARG variants and genetic contexts, we assembled metagenomic reads and identified ARGs using the contig-based tool, RGI (Resistance Gene Identifier) (30). Although homologous ARGs belong to the same gene family, different variants exhibit substantially different risks in terms of host range, mobility potential, and ecological distribution (26). ARG abundances and compositions were consistent among this approach and ShortBRED (Fig. 5C). RGI allowed for the accurate detection of several clinically relevant high-risk ARG variants. Out of 212 unique ARG variants encompassing 43 major antibiotic classes, several were clinically concerning ARGs (53/212), often associated with MDR infections in humans (Fig. 4D). Most of these high-risk ARGs were more prevalent in HWW-4 to HWW-6, as expected owing to high ARG richness in these samples (Fig. 5C). Aminoglycoside resistance (18% to 47%) and macrolide resistance (7.9% to 36.5%) were the top two drug resistance class in all the samples, followed by carbapenemase and sulfonamide resistance. bla OXA was the most prevalent carbapenemase across all the samples (3.6% to 11.5%) except HWW-6. In HWW-6, bla RSA-1 was the dominant carbapenemase (9.1%). Since smaller fragments of ARGs may remain undetected, we used fARGene to validate the prevalence of ARGs by reconstruction of contigs (23). Among the detected ARGs, only four beta-lactamase genes were not identified earlier by RGI, namely, bla CARB-16 and bla VIM-38 in HWW-4 and bla CMY-59 and bla OXA-4 in HWW-6. Several ARGs of urgent clinical concern are endemic to India (31,32). bla NDM , a carbapenemase conferring resistance against most b-lactam antibiotics, was first identified in a Swedish patient returning from New Delhi but in a short span of time was detected in several outbreaks around the world (33). Several studies reported multiple bla NDM variants in J.N.M.C. hospital wastewater (HWW-1), but due to lack of surveillance at other sites of study, this is the first time that we are reporting the presence of bla NDM-1 in HWW-2, HWW-4, HWW-5, and HWW-6 ( Fig. 6A to C) (34)(35)(36). bla NDM-1 often spreads through plasmid-mediated horizontal gene transfer (HGT) (37), and in this study, bla NDM-1 was predicted to be harbored by plasmid in HWW-2.
Aminoglycosides are typically not included in the first line of antibiotic treatment in many clinical scenarios, yet the high abundance of aminoglycoside resistance genes in HWW-2, HWW-4, HWW-5, and HWW-6 was surprising. As aminoglycosides are used against bacteria already resistant to beta-lactams and fluoroquinolones, the predominance of aminoglycoside resistance genes represents the presence of MDR and pandrug-resistant bacteria in the hospital environment (38). Colistin is a last resort antibiotic for treating MDR infections, and our detection of the mcr-5.1 gene, likely harbored on an E. coli IncX1 plasmid, is worrisome ( Fig. 6D) (29). ARG co-occurrence and association with MGEs. Assembly-based analysis revealed the coexistence of multiple ARGs, which may cause a severe threat to the treatment of clinical infections. We identified ARGs that were present on the same contig (Fig. 7). In HWW-1, five pairs of coexisting ARGs were identified in which three pairs had resistance toward the same drug class. Among the cooccurring resistance genes belonging to different drug classes, bla GES-1 , a carbapenemase, was associated with AAC(6')-II, an aminoglycosidase and qacEdelta1 (acridine dye resistance) with sul1 (sulfonamide resistance gene). The resistance markers belonging to three different drug classes, ANT (30)-IIa, bla OXA-10 , and cmlA5, were encoded by the same contig in HWW-2, whereas in HWW-4 one contig carried seven different glycopeptide resistance genes along with one macrolide resistance gene. bla NDM-1 was coassociated with BRP(MBL) in HWW-4, HWW-5, and HWW-6 ( Fig. 6A to C). In HWW-6, bla OXA-21 coexisted with cmlB, AAC(6')-IIa, and qacEdeltaI.
Co-occurrence of ARGs with plasmids and bacterial taxa. Plasmids play a key role in bacterial ecology and evolution, particularly in regard to the spread of ARGs  (39). Therefore, we next sought to identify ARGs associated with plasmids. ARGs in all the HWW samples were predominantly mediated by plasmids, including several important carbapenemases, such as bla OXA variants, bla NDM-1 , bla VIM variants, and bla IMP variants. In HWW-5 and HWW-6, a significant proportion of ARGs was carried by plasmids (78% and 72% in HWW-5 and HWW-6, respectively). Exploring the bacterial host of an ARG is a crucial question to be addressed for understanding their emergence and diversification (40). Hence, we identified the association of ARGs with bacterial taxa. As expected, based on their predominance in the composition of the HWW microbiome, ARGs were more associated with Proteobacteria with an average content of 83.8% (range, 77.9% to 91.9%) followed by Bacteroidetes. The ARGs were mostly carried by the genus Pseudomonas, and P. aeruginosa was the predominant species in all samples except HWW-4 where the genus Acinetobacter and species A. baumannii were most prominent. Both P. aeruginosa and A. baumanii ESKAPE pathogens often cause severe nosocomial infections leading to high mortalities (41,42). We detected multiple carbapenemase genes such as bla NDM , bla OXA , bla DIM , bla IMP , bla VIM , and bla GES carried by these species. The taxonomic identification of contigs carrying bla NDM-1 showed that it was carried by P. aeruginosa in HWW-2, K. pneumoniae in HWW-4 and HWW-6, and A. baumanii in HWW-5. Colistin resistance gene mcr-5.1 in HWW-6 was detected in E. coli.
Further, we explored the cooccurrence patterns between the abundance of the top 10 ARGs across all the samples and the associated bacterial abundance at phylum and genus levels. There was a significant Spearman's rank correlation (Spearman's r = 0.81;1; P , 0.05) between microbial diversity and ARG diversity (Fig. 8). We used network analysis to explore cooccurrence patterns between ARG subtypes and bacterial taxa. Several studies have hypothesized that nonrandom cooccurrence patterns between ARGs and microbial taxa are indicators for possible host information for ARGs if ARGs and coexisting microbial taxa have significantly similar abundance trends (Spearman's r . 0.8; P , 0.05) (43,44). A total of 20 genera and 3 phyla Proteobacteria, Bacteroidetes, and Actinobacteria were identified as possible hosts of ARGs. Proteobacteria was the host of aminoglycoside resistance genes [aadS, ANT(39')-IIa, and APH(6)-Id], macrolide resistance genes (ErmF, mphG, and msrE), sulfonamide resistance gene (sul2), beta-lactamase (bla VEB-9 ), and qacEdelta1.
Virulence factor distribution. Virulence factors are the agents of ecological connectivity that play an important role in the delocalization of AMR genes across niches through the formation of biofilms and increased infectivity. Twitching motility proteins mainly belonging to type IV pili and the flagellar proteins were prevalent in most HWW samples (Table 3). Bacteria, especially Pseudomonas spp., with twitching motility proteins have greater cytotoxicity toward the epithelial cells, enhanced transmission to other organs, and higher virulence. The flagellar proteins are responsible for adhesion, biofilm formation, and modulation of the immune system of eukaryotic cells (45). General secretion pathway proteins (gspC, gspF, gspG, gspH, gspI, gspK, and gspM) were most abundant in HWW-2 followed by type IV pili and twitching motility proteins (pilG, pilH, and pilI). In all the HWW samples, most virulence factors were associated with Pseudomonas except in HWW-2. In HWW-2, Escherichia carried most of the virulence factors. HWW-2 represented a distribution of unique VFs compared to other samples. Most of them were E. coli-specific factors like dispersin, type III secretion system effectors, general secretion pathway proteins, and E. coli common pilus chaperones. The presence of alginate biosynthesis, alginate regulation, and alginate biosynthesis virulence factors enhances the capability of bacteria to produce biofilm. The biofilm is  highly proficient in transferring AMR and has an innate tolerance toward antibiotics (46). The general secretion pathway proteins have been identified as a major target for the treatment of infections and to curb the growing antibiotic resistance (47).
Metagenome-assembled genomes. To better study the content of individual genomes and the genetic contexts of ARGs, we constructed metagenome-assembled genomes (MAGs). We recovered a total of 167 draft MAGs of various quality: 26 highquality, 65 medium-quality, and 76 low-quality drafts (Fig. 9A). The most commonly recovered genera were Raoutella (13/167), Comamonas (11/167), and Acinetobacter (8/ 167), all Proteobacteria. Of the total 84 unique genera identified, 54 were unique to just 1 MAG. Next, we screened the 26 high-quality MAGs for known ARGs in silico. We identified a total of 20 unique ARGs in 18/26 MAGs (Fig. 9B). Most of these MAGs (12/18) encoded at least one beta-lactamase, including bla OXA carbapenemase and bla PER extended-spectrum b-lactamases (ESBLs).

DISCUSSION
We have shown that Indian HWW samples have a high abundance and diversity of clinically relevant ARGs potentially hosted by high-priority pathogens. We collected samples from six tertiary-care hospitals in north India with various patient loads, located in urban (HWW-1, -2, -4, and -5) and rural areas (HWW-3 and -6) ( Fig. 1 and  Table 1). Earlier studies have been limited to single hospitals (48). We used shotgun metagenomics to analyze our HWW samples, which has the advantage of quantifying thousands of genes from culturable as well as nonculturable taxa simultaneously. PCR-based approaches may have greater sensitivity to low-abundance ARGs due to targeted amplification (49), but they are limited in the ARGs able to be detected. For example, a study comparing culture-based approaches and metagenomic methods showed that the culture-based technique isolated bacteria from 104 out of 539 clinical samples and therefore, captured resistance against 8 antibiotic types in only 16.17% of the clinical samples. Contrary to the culture method, metagenomic analysis identified 1,573 species and 885 ARG subtypes in the hospital sewage (50). Hospital wastewater surveillance is useful for monitoring antibiotic-resistant bacteria and the ARG load in the hospital environment (51). In our study, we found high ARG richness (fragments per kilobase million [FPKM]) in hospital wastewater. A study reported a comparably lower abundance of ARGs in urban wastewater compared to     our findings (11). In India, hospital wastewater treatment is inadequate and the hospital wastewater mostly gets released into the public sewer network increasing the probability of ARG dissemination to humans, especially to the poor who often reside near the drains (22,51). Owing to its high abundance, the phyla Proteobacteria carried most of the ARGs, constituting approximately 85% of the ARG abundance (range, 78% to 84%) in terms of FPKM, followed by Bacteroidetes (HWW-1 to HWW-4), Firmicutes (HWW-5), and Actinobacteria (HWW-6) (Fig. 3 to 5). The taxonomic compositions of our HWW samples are similar to that reported from a single hospital in the city of Mumbai, located in western India (48). They also reported the dominance of Proteobacteria, followed by Bacteroidetes and Firmicutes. Interestingly, they reported Acinetobacter as the dominant genus (30%), which is also true of our HWW samples from New Delhi (HWW-4 = 26% and HWW-5 = 18%); however, Acinetobacter was not among even the top three most abundant genera in the other samples ( Fig. 2 and 3). Pseudomonas was among the top four genera (3% to 9%), carrying maximum ARG abundance (23% to 38%) in all the samples except HWW-4. In congruence with the aforementioned study, Acinetobacter predominantly carried 23% ARG abundance in HWW-4 of New Delhi (48). The species carrying most of the ARGs were P. aeruginosa across all the samples, A. baumanii in HWW-4, and P. putida in HWW-6. All these species are high priority, nosocomial pathogens with a high rate of ARG dissemination, mostly involved in hospital-acquired infections (52,53). The opportunistic pathogens A. baumanii and P. aeruginosa carry natural intrinsic resistance toward multiple drug classes, including beta-lactamases and other carbapenemases. They are not eliminated by wastewater treatment either (54,55). P. putida is an environmental Gram-negative bacterium. It is rarely a causative agent for human diseases, but there have been reports of serious infections and outbreaks from time to time. Sometimes it functions as an exchange platform for ARGs, spreading ARGs to more pathogenic species like P. aeruginosa (56).
We generally expected HWW from the largest hospitals to have the highest ARG burdens due to increased patient load. Surprisingly, the samples from hospitals with the largest number of beds (HWW-1 = 2,400 beds and HWW-2 = 1,050 beds) had among the lowest ARG abundances. Conversely, HWW-4 and HHW-6 had 4 to 6 times higher ARG abundance but with a fraction of the number of beds (HWW-4 = 650 and HWW-6 = 300). This could be due to improved antibiotic stewardship and waste disposal practices. However, we note that "hospital size" is not definitive because many hospitals in India over admit 3 to 4 times the total bed capacity, and these true numbers may not be reported (22).
Several studies have reported the injudicious use of broad-spectrum antibiotics in Indian hospitals and, consequently, high incidences of beta-lactamase (carbapenemase) resistance (57,58). We also detected predominant resistance against broadspectrum antibiotics with a leading abundance of aminoglycoside resistance in all the samples except HWW-1 and HWW-3 where macrolide resistance was most abundant (Fig. 4A and 5) (48). Carbapenem and sulfonamide resistance was among the top five ARG-enriching drug classes in all the samples. Earlier, in Indian HWW, the sul1 sulfonamide resistance gene was reported as the most abundant (11.4%); we also identified sul1 in all HWW samples but it was among the top 10 most abundant ARGs only in HWW-2 and HWW-6 (2.2% and 9.6%, respectively) ( Fig. 4) (48). The ARG composition listed in the top 10 ARGs replicated the abundance of resistance evaluated at the drug class level with 8 aminoglycoside resistance genes, 6 macrolide resistance genes, and 3 carbapenemases. Among the top 10 ARGs, the beta-lactamases bla OXA-10 , bla GES-1 , and bla RSA-1 were dominant. Multiple incidences of bla OXA outbreaks in Indian hospitals have been reported in the past, which correlates with bla OXA being the most abundant carbapenemase in all our samples (59,60). The bla OXA-10 and GES-type ESBLs have been earlier reported as the most abundant beta-lactamase in one of the Mumbai hospitals (48). The bla NDM-1 is endemic to India and is often flanked by transposable elements. The bla NDM-1 was found prevalent among P. aeruginosa, K. pneumoniae, and A. baumanii (Fig. 6A to C). The use of colistin used as the last-resort antibiotic in extreme clinical cases of MDR and extensively drug-resistant (XDR) infections also leads to the emergence of mcr variants. The detection of plasmid-mediated mcr-5.1 in HWW-6 ( Fig. 6D) for the first time in Indian hospital sewage raises concern for future healthcare systems and is an alarming signal for the upcoming antibiotic apocalypse when no antibiotic will work. MAGs construction and analysis also revealed at least one beta-lactamase, including bla OXA carbapenemase and bla PER ESBLs. Most of the ARGs identified in our study can efficiently transmit to different species of bacteria through horizontal gene transfer, and hence, their spread to the local population has become a challenge for healthcare workers. The concentration of several ARGs like blaOXA-1, blaOXA-10, and blaTEM-1 increases with wastewater treatment procedures (61). Plasmid carried the maximum ARGs in all the samples reflecting the high probability of AMR spread through hospital wastewater.
In this study, we identified several virulence factors mostly associated with the general secretion pathway, motility, and alginate biosynthesis often involved in biofilm formation (Table 3). These VFs aid the innate resistance against antibiotics (45,47,62).
Our study adds further evidence that India is facing an "AMR pandemic" and needs urgent national surveillance for assessing the ARG risk. Hospital wastewater transports high-risk clinical threats to the public sewer system and facilitates their dissemination to the public through HGT (61). The common practice of open drainage systems and inadequate sanitation measures in most parts of India (63) may lead to infection spread with antibiotic-resistant bacteria and outbreaks in the community as well as a hospital setting. Likely due to our limited sample size, we did not find significant dissimilarities in ARGs and microbial diversity with geographical variations. Overall, the ARG diversity in our samples was in concordance with one of the earlier studies on Indian hospital wastewater samples, carried out in Mumbai (48).
This explorative study shows that Indian HWW contains an abundance of high-risk ARGs that are present in mobile genetic elements and are carried by high-priority pathogens. Antimicrobial risk management strategies should be immediately implemented in hospitals. Efficient wastewater treatment strategies meeting international standards should be implemented in hospitals, as well as areas downstream including the public sewage systems where the HWW are deposited.

MATERIALS AND METHODS
Sample collection and processing. Wastewater samples were collected from six hospitals located in different regions of northern India from December 21, 2019, to March 21, 2021 (Table 1). From the main sewage pipeline receiving effluents from every other pipeline in the hospital, multiple samples were collected from adjacent points in sterile bottles and pooled into one sample of 2 liters of unfiltered sewage water. Samples were stored on ice without additives or DNA stabilizers. In a sterile environment, samples were vortexed and 50 mL was collected and then centrifuged again at 7,000 Â g to separate cell pellets and water. The pellet was stored at -20°C until DNA extraction.
DNA was extracted using the DNeasy PowerSoil kit (Qiagen) as per the manufacturer's instructions. Extracted DNA was quality checked using NanoDrop 2000 Spectrophotometer (ThermoFisher Scientific), to check for RNA and protein contaminants. Further, to validate the quantitative estimation of the extracted DNA, 50 ng of extracted DNA and Lambda DNA/HindIII Marker (SM0102; ThermoFisher Scientific) was loaded on a 1% agarose gel stained with ultrapure ethidium bromide (ThermoFisher Scientific) and electrophoresed by running the gel at 80 V for 1 h. Finally, the gel was imaged in a SmartView Pro 1100 Imager System (Major Science). Sample DNA concentrations were quantified using Qubit dsDNA HS assay kit (Thermo Fisher Scientific), as per manufacturer's instructions, using a Qubit 3.0 Fluorometer (Thermo Fisher Scientific) (64).
DNA library preparation and sequencing. One hundred nanograms of intact DNA was enzymatically fragmented using Covaris targeting the 250-bp fragment size, followed by end repair to convert the overhangs into blunt ends. To the adenylated fragments, loop adapters were ligated and cleaved with a uracil-specific excision reagent (USER) enzyme. The samples were further purified using AMPure beads. DNA was enriched by PCR with six cycles using NEBNext Ultra II Q5 master mix, Illumina universal primer, and sample-specific octamer primers. The amplified products were cleaned by using AM pure beads to remove unused primers. The final DNA libraries were eluted in 15 mL of 0.1Â TE buffer, and concentrations were quantified using the Qubit DNA HS assay kit and a Qubit.3 Fluorometer.
Library quality was assessed using DNA 5000 ScreenTape in an Agilent 4150 Tape Station system. Here, 1 mL of the library was mixed with 5 mL sample buffer, vortexed, and then spun to collect the sample to the bottom of the strip. The strip was then loaded into the Agilent 4150 Tape Station instrument. The library qualification criteria were the presence of a broad peak in the range of 200 bp to 1,000 bp, with an average size of 350 bp in the Agilent 4150 TapeStation system, the Qubit concentrations above 2 ng/mL or 10 nmol/L and library devoid of primer, adapter, and larger size peaks.
Taxonomic classification. For taxonomic classification, the reads were quality filtered using fastp v.0.20.1 (69). These clean reads were taxonomically classified using Kraken2 with the NCBI nonredundant nucleotide database as a reference (70). Kraken hits with a relative abundance of ,0.1% reads were filtered out.
Resistance via mutation in genes (e.g., resistance to antifolate drugs via mutations in dhfr, resistance to rifamycin via mutation in rpoB); 4.
The relative abundance of AMR gene families was quantified by mapping reads to the filtered set of marker sequences using shortbred_quantify.py. ShortBRED hits were filtered out if they had counts less than 2 or a mean RPKM , 0.001.
Ethics approval and consent to participate. Not applicable. Preapproval from the ethical committee is not required in India to work on hospital sewage samples.
Data availability. All metagenomic sequencing data are available at the National Center for Biotechnology Information (NCBI) database with BioProject accession number PRJNA682952 and SRA accession numbers SRR13227005, SRR13227004, SRR13227003, SRR13227002, SRR15384560, and SRR15384559.

Antibiotic Resistance in Indian Hospital Wastewater
Microbiology Spectrum