intI1 gene abundance from septic tanks in Thailand using validated intI1 primers

ABSTRACT Antimicrobial resistance (AMR) poses a serious global health threat, and wastewater treatment (WWT), including septic tanks, is a source of AMR. In Thailand, antibiotics are unregulated, and septic tanks are commonly used. Yet, their impact on the spread or mitigation of AMR is unknown. We monitored household and healthcare conventional septic tanks (CST) and household solar septic tanks (SST) in Thailand using the class 1 integron-integrase (intI1) gene abundance as a proxy for AMR. A systematic review of the literature found 65 intI1 primers. We evaluated the coverage and specificity of each, including a new MGB TaqMan primer-probe, against clinical and environmental intI1, intI1-like, and non-intI1 databases. The three best primers were selected, laboratory validated for DNA and mRNA quantification, and used to quantify septic tank intI1 gene abundance. No primer set could distinguish between intI1 and intI1-like sequences. While primer choice did not affect gene abundance of the same sample (P-value > 0.05), sometimes when comparing the same samples quantified by different primers, statistical differences were observed for one but not the other primer set. This may lead to different interpretations of AMR risk. Irrespective of primers or reactor type intI1 gene abundance was greatest in influent > effluent > sludge. intI1 gene abundance was lowest in the effluent of the SST-household < CST-household < CST-healthcare. 31% to 42% of intI1 was removed by the CST-household tank, indicating while septic tanks remove some intI1 they remain a source to the surrounding environment. Toward the goal of achieving standardization across studies, we recommend the F3-R3 primer for intI1 quantification. IMPORTANCE Antimicrobial resistance is a global crisis, and wastewater treatment, including septic tanks, remains an important source of antimicrobial resistance (AMR) genes. The role of septic tanks in disseminating class 1 integron, and by extension AMR genes, in Thailand, where antibiotic use is unregulated remains understudied. We aimed to monitor gene abundance as a proxy to infer potential AMR from septic tanks in Thailand. We evaluated published intI1 primers due to the lack of consensus on optimal Q-PCR primers and the absence of standardization. Our findings confirmed septic tanks are a source of class 1 integron to the environment. We highlighted the significance of intI1 primer choice, in the context of interpretation of risk associated with AMR spread from septic tanks. We recommend the validated set (F3-R3) for optimal intI1 quantification toward the goal of achieving standardization across studies.

The occurrence of AMR via mutation and subsequent vertical gene transfer or acquisition of AMR genes via horizontal gene transfer (HGT) is an inevitable natural phenomenon in the evolution of microbes (2,3).Nonetheless, recent global challenges including extensive consumption and misuse of antimicrobials, particularly antibiotics, in clinical settings, agri-and aqua-culture, and their subsequent release to the envi ronment, have given rise to the emergence and rapid dissemination of AMR genes among bacteria, including microbes of clinical importance, and the environment (2).Consequently, high global mortality, as a result of patient treatment failure, has been associated with AMR-related infections (700,000 deaths in 2014) (4).Moreover, the global death toll from AMR-related infections has been projected to increase to 10 million deaths per year by 2050 surpassing death from cancer, assuming no change to the current trends/policies, coupled with an economic burden of 100 trillion US Dollars (4).
Wastewater treatment (WWT), including decentralized treatment systems such as septic tanks, receives significant amounts of antibiotics from human and animal waste (30%-90% of antibiotics are excreted in urine and feces (5)) and are now recognized as important reservoirs for AMR creating hotspots for transfer and subsequent release to the environment (3,6,7).The selective pressure introduced by the often multiple, low-level, sub-inhibitory concentrations of antimicrobials found in wastewater (WW), promotes AMR gene acquisition among microbes via HGT and selection for AMR bacteria.WWT and septic tanks are unable to effectively remove these (3,8,9), resulting in increased AMR genes and bacteria discharged directly to the environment, contribu ting significantly to the global burden (10).The global AMR burden from wastewater is further exacerbated in the Global South due to the high prevalence of extensive antibiotic usage-propelled by poor regulations on usage, ineffective or lacking WWT, coupled with increasing populations and rapidly expanding megacities.
The necessity to tackle AMR discharge from WWT to the environment requires a comprehensive understanding of the role of WWT in the dissemination of AMR to the environment.This understanding will create unique opportunities to implement key strategies to mitigate AMR spread, and, in turn, allow for the safeguarding of global public health.This knowledge can be informed by sensitive, accurate detection, quantification, and tracking of AMR genes from the source (e.g., WWT) to the environ ment.However, multiple AMR genes exist within WWT.Monitoring numerous AMR genes simultaneously is a major challenge (8), particularly if a rapid assessment is needed.Similarly, monitoring one or a subset of AMR genes is not ideal, as selected AMR gene(s) may be absent (8).Previously, the clinical class 1 integron (CL1-integron) integrase gene (intI1) was proposed as a proxy for inferring potential AMR, which circumvents multiple monitoring limitations, by acting as a proxy for potential AMR pollution (8).intI1 was proposed as a proxy because it is linked to genes that confer resistance to antibiotics, disinfectants, and heavy metals; it is found in diverse taxonomic groups of pathogenic and non-pathogenic bacteria and can move across taxa via HGT due to its physical linkage to mobile genetic elements (MGEs) such as plasmid and transposons; its abundance can rapidly change in response to external pressures such as the presence of antibiotics; selection pressures imposed by recent human activities have resulted in the emergence of the highly conserved clinical intI1 variant (8); and the elevated presence of which in the environment indicates pollution and potential hotspot for AMR transfer (8,11).
Currently, molecular approaches, specifically real-time quantitative PCR (Q-PCR), have emerged as the methods of choice for AMR and CL1-integron detection and quantifica tion in the environment.By far, the most prevalent approach for detecting or quantifying the CL1-integron is the amplification of the intI1 gene at the 5′ conserved segment (CS) across diverse ecological niches including engineered systems, for example, WWTs (12)(13)(14) and natural ecosystems such as sediments (15,16).While targeting the intI1 gene provides no information about the structure beyond the 5′ CS (17), quantification of the intI1 gene as an initial screening to infer potential AMR contamination within complex environments is invaluable and a useful initial screening approach.However, within the literature, numerous primers targeting the intI1 gene are available (see Table S1) and different sets are used across different studies.The current lack of standardiza tion prevents cross-study comparisons and limits the current understanding of AMR in the environment.As such, selecting optimal intI1 primers with both high coverage and specificity suitable for environmental monitoring is a challenge.Moreover, several primers have been designed based on the highly conserved clinical intI1 gene sequen ces (≥98% protein similarity to each other), and the extent to which these primers target the less conserved intI1 gene variants (<98% protein similarity) found also in environmental samples (8,(17)(18)(19) and on the chromosome non-pathogenic Betaproteo bacteria which carries gene cassettes not associated with AMR genes (19) has yet to be determined.As such, a comprehensive and comparative evaluation of published intI1 primers to determine their coverage and specificity against clinical and environmental intI1 sequences to identify a consensus optimal intI1 primers for monitoring AMR within environmental samples is urgently needed (20).
With this need identified, we undertook to review, evaluate, and then apply intI1 primers to quantify the gene across a suite of wastewater samples from septic tanks in Thailand.Specifically, we compare the recent solar septic tank (SST) technology currently implemented in some areas of Thailand (7,21) to that of conventional septic tanks (CST) treating household and healthcare wastewater.The SST technology differs from CST primarily by the incorporation of a central disinfection chamber containing a heated copper coil connected to a passive solar heat collection system installed on the roof of the toilet block served by the SST (7,21).The heat from the central chamber (50°C-60°C by design) promotes partial pasteurization as the effluent passes through the chamber prior to discharge.Effluent water quality is improved by reducing microbial biomass including potential pathogens, and by extension reduction in the microbial load should reduce the AMR burden on receiving water bodies.Moreover, the in-tank temperature is raised by the passive transfer of heat from the chamber to the rest of the tank, thus promoting enhanced microbial degradation of both retained solids (sludge) and soluble compounds (7).As such, we hypothesize that intI1gene abundance would be lower in the SST than in the CST sludge and effluent owing to the enhanced treatment caused by the increased temperature.
To address this methodological knowledge gap and our hypothesis, a systematic review of the literature was undertaken to obtain published intI1 primers followed by a comprehensive in silico analysis of primer coverage and specificity against a curated database of clinical and environmental intI1 sequences to select the best-performing primers.A subset of the best-performing primer sets was used to quantify the intI1 gene abundance from 30 septic tank wastewater samples comparing conventional (healthcare and household wastewater) and solar septic tank (household wastewater), with intI1 specificity validated by Illumina MiSeq.We further confirmed the suitability of the primers to quantify intI1 gene transcripts.Thus, we propose a validated intI1 primer set for quantification of genes and transcripts from environmental samples toward the goal of achieving standardization across intI1 studies.

A systematic review of literature and alignment of primers to intI1 reference sequence
A systematic review of >3,000 peer-reviewed publications was conducted to retrieve intI1 primer and probe sequences across a range of settings including clinical and environmental, for example, agricultural and human-impacted settings including WWTPs.For this, the "Web of Knowledge" database (https://www.webofscience.com/; last assessed 04/10/2022) was searched using the term "Class 1 integron." Only published articles in the English language were considered.In all, 3,266 published articles were subsequently recovered.The intI1 primer sequences from the respective literature were either retrieved in the main text or from the accompanying supplementary material.

Database construction and curation
The integron-integrase database by Zhang et al. (20), consisting of 922 and 2,462 intI1 gene and integron-integrase (intI) of other class protein sequences, respectively (herein referred to as non-intI1 database) (Fig. 1), was employed for the analysis of primers.While the intI of other class databases was mostly populated with protein sequences from other integron-integrase classes, it also contained a number of XerCs integrases (n = 78) and transposases protein sequences (n = 66) as recently reported by Roy and co-workers (24).In this study, however, the inclusion of these protein sequences within the non-intI1 database is not of significance, as the goal was to confirm that analyzed intI1 primer sets were unable to amplify sequences within this database via in silico testing, thus confirming their specificity.
The IntI1 protein database was curated by discarding duplicate protein sequences (n = 1) from the 922 intI1 protein sequences (Fig. 1).Retained IntI1 protein sequences were then compared to the reference IntI1 protein sequence of pVS1 plasmid (AAA25857.1)using NCBI BlastP, to ensure intI1 nucleotide sequences used for in silico assessment of primer and probe sequence coverage were indeed intI1 sequences as suggested by Roy et al. (24).Furthermore, protein sequences whose percentage identity to the reference pVS1 IntI1 plasmid protein sequence was ≥98%, were characterized as IntI1 sequence, while sequences whose protein similarity to the reference pVS1 IntI1 plasmid protein sequence were <98% were categorized as IntI1-like protein sequences (24).Moreover, the protein sequence identified as IntI1-like was manually checked to ensure the percentage similarity score to the pVS1 protein sequence reported by NCBI was not due to missing sequence caused by the alignment of a partial sequence to a complete length sequence.As such, protein sequences (n = 2; WP_058137959.1 and WP_058135314.1)incorrectly identified as IntI1-like were added to the IntI1 protein database (Fig. 1).
Finally, three intI1 gene nucleotide sub-databases (SDB1, SDB2, and SDB3) were created for robust primer analysis based on the criterion specified in Table 1.
SDB1 (n = 104) contained full-length intI1 sequences ≥ 1,000 bp, confirmed by the presence of a start and stop codon; SDB2 (n = 144), contained full-length intI1 sequences ≥ 900 bp confirmed by the presence of a start codon and stop codon.Sequences within SDB1 are all present in SDB2.The final intI1 sub-database (SDB3, n = 503) contained both complete and partial sequences (Fig. 1; Table 1).All sequences within SDB1 and SDB2 were also present within SDB3.The intI1-like (<98% similarity to pVS1 on protein level) sub-database contained both complete and partial-length sequence (n = 15; Fig. 1; Table 1).The non-intI1 database contained 1,540 integrase sequences of other classes (Fig. 1).
In parallel, the non-intI1 sequence sub-database was constructed from the 2,462 intI of other class protein sequences by applying a ≥300 amino acid length thresholds (900 bp nucleotide length) to filter out shorter-length protein sequence (Fig. 1; Table 1).Retained protein IDs for the intI1, intI1-like, and non-intI1 sequences were then used to manually obtain the nucleotide sequences from NCBI in Fasta format.
To summarize, intI1 sequences from this study were defined as intI1 protein sequen ces whose percentage identity shared a ≥ 98% similarity to pVS1 intI1 plasmid pro tein sequence (AAA25857.1),while intI1-like sequences were defined as intI1 protein sequences sharing a < 98% similarity to pVS1 intI1 plasmid protein sequence (Fig. 1).

Primer evaluation
Published intI1 primers were analyzed as primer pair (Table S1), using Primer Prospector (25), to evaluate coverage and specificity against constructed integrase sub-databases protein sequence (n = 1) was discarded.Retained protein sequences were compared to the reference IntI1 protein of pVS1 plasmid (AAA25857.1)using NCBI BlastP to identify true intI1 sequences.IntI1 protein sequences with a ≥98% identity to the pVS1 protein sequence were classified as IntI1 protein sequences.
Conversely, IntI1 protein sequences with a <98% identity to the pVS1 protein sequence were classified as IntI1-like sequences.Three intI1 gene nucleotide sub-databases (SDB1, SDB2, and SDB3) were finally constructed based on criteria specified in Table 1 and were used to evaluate the coverage of primers.intI1-Like (n = 15) and non-intI1 (n = 1540) sub-databases were used to evaluate the specificity of primers.* Indicates removal of 1 protein sequence (CP006631.1)from the 921 non-duplicate intI1 protein sequence, due to no similarity score to the IntI1 pVS1 protein sequence generated, as a result of low sequence similarity.
# Indicates the two (WP_058137959.1 and WP_058135314.1)IntI1 protein sequences incorrectly identified as IntI1-like protein sequences by the low similarity score generated by NCBI following alignment to the pSV1 protein sequence due to these sequences being partial length sequences.(Fig. 1).The analyze_primers.pyfunction with the default settings on Primer Prospec tor was used to generate an alignment profile file for each primer against unaligned individual nucleotide sequence in each test sub-databases.For each primer alignment to a nucleotide sequence, a weighted score (WS) was given.

Overall WS was calculated as
Non-3′ mismatches * 0.4 per mismatch +3′ mismatches * 1.0 per mismatch +Final 3′ base mismatch * 3.0 per mismatch +non-3′ gaps * 1.0 per gap +3′ gap * 3.0 per gap.The first five bases of the primer and the target sequence were defined as the 3′ end and thus, mismatches within these bases were termed 3′ mismatches.The remaining bases of the primer and the target sequence were defined as the non-3′ end.Therefore, mismatches within these non-3′ end bases were regarded as non-3′ mismatches.Gaps in the alignment of the primer and the target sequence in the first five bases were termed 3′ gaps while gaps in the alignment for the remaining primer and template sequence were known as non-3′ gaps.The lower the WS, the better the compatibility between the primer and target DNA sequence, a 0 score indicates perfect alignment.Primer Prospector will force a primer sequence to bind anywhere within the target sequence even if the primer binding site is unavailable to generate a WS for the primer.As such, the primer binding orientation of each analyzed primer pair was checked for each sequence using R (R Core team 2022).In addition, the seqnir package (26) in R was used to load the DNA/protein sequences.The mean WS for the forward and reverse primer for the primer pairs with the correct binding orientation was noted and a WS plot of each primer set was generated using the "ggplot2" R package (27).A detailed description of the primer evaluation is in supplementary material 2.1.
In the case of primer pairs that incorporated a TaqMan hydrolysis probe, the primer-probe-binding-orientation (forward, probe, and reverse) against each unaligned sequence was first verified, for each unaligned sequence by checking the hit positions of the forward, probe, and reverse primer sequence in R. Unaligned sequences with correct primer-probe orientation were subsequently retained and analyzed in the manner as described in supplementary material 2.1.

Design of a new intI1 primer set and TaqMan-minor-groove binder (TaqMan-MGB) probe
To improve intI1 sequence coverage and specificity for Q-PCR the intI1 primer set, F3-R3 (Rosewarne et al., (28), Table S.I) was modified to generate a new intI1 primer incorporat ing an MGB TaqMan probe set (intI1 DF-DR, Table S.I) following guidelines for primerprobe design outlined by McKew and Smith (29).An MGB probe of 15 bp was designed using Primer Express software (version 3.0.1;Applied Biosystems).A detailed protocol of the MGB probe design can be found in supplementary material 2.2.Primer and probe sequences were BLAST searched (BLASTN) to validate the sequence specificity.Specificity and coverage of the newly designed primer and probe set were tested as detailed above with Primer Prospector.

Optimization of selected primer sets for Q-PCR
The amplicon produced from selected primers for laboratory validation was assessed in silico first using sequences within SDB1 and then in the laboratory by endpoint PCR.Selected intI1 primer sets that resulted in the correct size amplicon were further optimized for Q-PCR assays (Table 2).Q-PCR standard curves were constructed by amplifying a synthetic intI1 gene fragment (Integrated DNA Technologies) containing the primer binding site for all selected primers (Fig. S2).The insert fragment was amplified by PCR using T7 forward (5′-TAATACGACTCACTATAGGG-3′) and M13 reverse (5′-CAGGA AACAGCTATGAC-3′) primers.Reaction volume and condition in supplementary material 2.3.Resulting amplicons were purified, and size selected with the Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA) per the manufacturer's recommendation, using a 1:1 ratio of beads volume to PCR product volume, and eluted in a 25 µL volume nucleasefree water.Purified products were quantified fluorometrically using Qubit (Invitrogen, according to the manufacturer's recommendations), and the gene copy number was determined using EndMemo DNA copy number calculator (http://endmemo.com/bio/dnacopynum.php).The purified concentrated stock was subsequently diluted to 10 9 copies/μL, followed by a 5-, 10-fold serial dilution (10 7 -10 3 copies/μL) for amplification by Q-PCR.A standard curve was obtained by plotting the average of each triplicate threshold cycle (Cq) against the log10 of standard concentration (copies/μL).Standard curve descriptors including efficiency, slope, y-intercept, and R 2 are reported (30).

Solar and conventional tank sampling
Two household solar septic tank (SST; SST01 and SST07) units and three conventional septic tank (CST; two household tanks and one healthcare tank) units, operational within the Pathum Thani province and Samut Prakan province, Thailand, were sampled between April 2018 and September 2019 (Table 3).
The SST and the household CST units (CST-P3 and CST-J6) have a 1,000 L total working capacity, while the healthcare CST units (CST-HC2; herein referred to as CST-HC) has a 2,000 L total working capacity each.Each tank was buried to approximately 1.5 m below ground level; with the tank surface (lid) at ground level, and so exposed to atmospheric temperatures (7).The sampling approach used is described in supplementary material 2.4 and is outlined elsewhere (7).In total, 100 mL of effluent and 40 mL sludge was sampled from the SST and CSTJ7 and CST-HC, while 100 mL of influent was also collected from CST-P3.40 mL of sludge was sampled from each reactor.All samples were pelleted for DNA extraction.For the sludge, DNA was exacted from 0.5 g pellet.The months for sampling the SST were selected based on the highest recorded internal temperature of the 12-month sampling campaign conducted.

DNA extraction
From each sample, DNA extraction was performed with the DNeasy PowerSoil Kit (Qiagen), in accordance with the manufacturer's instructions.Integrity of extracted genomic DNA was assessed via agarose gel electrophoresis and DNA concentration was quantified fluorometrically using the Qubit (Invitrogen) according to the manufacturer's instructions.

Q-PCR quantification of intI1 gene from wastewater
intI1 genes were quantified from septic tank wastewater samples from Thailand (Table 3) using optimized Q-PCR conditions for the three selected intI1 primer pairs (DF-DR, F3-R3, and F7-R7).For each primer set, Q-PCR amplification was carried out in a 20 µL volume reaction using 2 µL (1:50 diluted) template DNA.Reaction volume, conditions, primer sequences, and probe type for the three selected optimal intI1 primer pairs are detailed in Table 2. Triplicate/duplicate no template control (NTC) was included for each primer set.Reactions were performed on the Bio-Rad CFX96 Touch Real-Time PCR Detection System and analyzed with the Bio-Rad CFX Manager 3.1 software.Melt curve analysis was performed, for the SYBR Green assay, from 65°C to 95°C with 0.5°C increments every 5 s, and a single peak was confirmed to ensure assay specificity.Statistical analyses were performed in R (R Development Core Team, 2008).Two-way analysis of variance (ANOVA) followed by a Turkey HSD post hoc test was employed to compare intI1 gene abundance for each of the sample types (influent, sludge, and effluent) and reactor type (CST and SST) for each primer set.Finally, a Pearson correla tion analysis was applied to calculate the relationship between the abundance of intI1 detected between each primer set.

MiSeq amplicon sequencing
The specificity of the selected intI1 primer sets (DF-DR, F3-R3, and F7-R7) used to quantify the intI1 gene from septic tank sludge and wastewater was confirmed by Illumina MiSeq amplicon sequencing of the intI1 gene from 31 wastewater samples (Table 3) using the optimized endpoint PCR conditions outlined in Table 2.A two-step PCR was performed to barcode samples as detailed previously (32,33).A detailed description of the method is provided in supplementary material 2.5.

Bioinformatics
Primer sequences were used to extract the intI1 gene from the resulting reads, particularly for shorter primer pairs, using the Cutadapt algorithm (34).Abundance tables were then generated by constructing amplicon sequencing variants (ASVs) using the Qiime2 pipeline and the DADA2 algorithm (35) with details given at https:// github.com/umerijaz/tutorials/blob/master/qiime2_tutorial.md.Constructed ASVs were blast searched on NCBI, and the closest hit sequences were retrieved for each ASV.The phylogenetic distance between sequences was investigated.First, a multiple sequence alignment of ASV sequences retrieved NCBI sequences, complete length intI1 and intI1-like and an intI3 (class three integrase gene; nucleotide ID: AY219651.1)sequence was done using MAFFT (36) for each primer set.Aligned sequences were visualized in BioEdit (version 7.0.5.3) (23) and trimmed to retain only aligned regions without gaps.Phylogenetic trees were constructed using a maximum likelihood approach with a generalized timer-reversible substitution model implemented in RAxML version 8 (37).Consensus trees were calculated after 1,000 bootstrapping permutations.
The phylogenetic tree of the trimmed and aligned sequence, for each primer pair, was constructed with RAxML (38).A heat tree of the constructed ASVs, after log2 transforma tion of ASV abundance per sample for each primer set, was mapped to analyzed samples, colored, and visualized using the ggtree package (39).The tip of the tree was colored based on the sequence isolation source.

Sample collection, filtration, and DNA/RNA co-extraction
As the septic tank wastewater samples were previously collected and only DNA extracted and stored at −80°C, they were not suitable for RNA analysis (33).Therefore, we tested the suitability of the optimized primer sets to detect intI1 mRNA using freshly collected environmental samples of river water collected from the Kevin River, Glasgow (UK), to determine whether intI1 mRNA transcripts could be quantified in receiving water bodies.In total, 3 L of surface water was collected in March and April 2022 and filtered through a sterile glass microfiber filter (FisherBrand MF200; retention 1.2 µm) and onto a 0.22-µm Sterivex filter.Filters were immediately extracted from or frozen at −80°C for later use.
DNA-RNA co-extraction was carried out according to the protocol previously described (33,40,41), with a minor modification to the bead-beating time (45 s) as outlined by Lim et al. (42).Detailed description of this method is outlined in sup plementary material 2.6.Briefly, RNA was prepared from the DNA-RNA co-extract by DNase treating with Turbo DNase Kit (Ambion) in accordance with the manufacturer's recommendation, with modification to the incubation time and volume of DNase added as previously described (33), 1 µL DNase volume was added and incubated at 37°C for 1 hour, followed by further addition of 1 µL DNase and a re-incubation at 37°C for another hour.Detailed protocol is found in supplementary material 2.7.

RT-Q-PCR quantification of inti1 genes and transcripts from river water
Q-PCR DNA standard curve was constructed as above (see above section Optimization of selected primer sets for Q-PCR).For each primer set, intI1 cDNA and DNA Q-PCR amplification was carried out in a 20 µL volume reaction using 2 µL (1:2 and/ 1:5 diluted) template DNA/cDNA.In addition, two priming strategies, genespecific (GS) and/or random (RH) priming, were used to reverse transcribe intI1 mRNA to cDNA.Detail of this approach is provided in supplementary material 2.7.Q-PCR volume, conditions, primer sequences, and probe type for the three selected optimal intI1 primer pairs are the same as specified in Table 2. Reactions were performed on the Bio-Rad CFX96 Touch Real-Time PCR Detection System and analyzed with the Bio-Rad CFX Manager 3.1 software.Melt curve analysis was performed, for the SYBR Green assay, from 65°C to 95°C with 0.5°C increments every 5 s, and a single peak was confirmed to ensure assay specificity.Standard curve descriptors including efficiency, slope, y-intercept, and R 2 are reported (Table 4).

Evaluation of primers for coverage
In total, 64 different intI1 primer sets, including 4 TaqMan primer-probe sets, were retrieved from the systematic review (Table S1).In addition, the primer and probe set designed in this study were included in the analysis, resulting in 65 primers evaluated (Table S1).Primers were initially aligned against the reference P. aeruginosa plasmid pVS1 nucleotide sequence (M73819.1)(Fig. S1) and renamed for ease of identification Table S1).Next, primers were aligned against the SDB1 (full-length intI1 database) to ensure binding sites were present (in forward or reserve orientation) and that the expected amplicon size would be generated.From this, 10 primer sets were discarded, which included two sets (F61-R61 and F64-R64) that were not intI1 primers (Table S2).The F61-R61 primer set targeted the aadA1a aminoglycoside adenylyl transferases gene (43,44), while the F64-R64 primer pair targeted the class two integron-integrase gene (intI2) (45).In addition, these primer sets (F61-R61 and F64-R64) aligned poorly to the reference P. aeruginosa pVS1 intI1 nucleotide sequence (data not shown) and had no hit (High WS) with complete length intI1 sequences within SDB1 (Table S2).
Five primer sets analyzed in this study incorporated a TaqMan probe (Table S5), two of which (DF-DR and F7-R7 sets) were among the best-performing primer sets.Of these, the DF-P-DR primer-probe set, designed in this study, consistently produced the highest number of amplicons at 0 WS across the three intI1 sub-databases with 102 (98%), 142 (99%), and 494 (99%) of sequences amplified within SDB1, SDB2, and SDB3, respectively (Table S5).In addition, allowing for a single non-3′ mismatch (WS = 0.4) between primer and probe, resulted in all sequences with the correct primer-binding orientation to be amplified across the three intI1 sub-databases (Table S5).

Evaluation of primers for specificity
The primer sets were tested for specificity against the intI1-like (n = 15) and non-intI1 (n = 1540) sub-databases, respectively (Fig. S5; Table S3).Here, the aim was for the primers to amplify the least amount of non-target sequence reflected by a higher forward and reverse primer WS for sequences where primers bind in the correct orientation.The 10 best-performing primer sets identified above were focused on.
For the best-performing primer sets, the number of sequences with correct primerbinding orientation ranged from 67% to 100% and 41% to 65% for the intI1-like and non-intI1 sub-databases, respectively, with 57%-80% and 0% of these correct primer-binding orientation sequences amplified at 0 mismatch in intI1-like and non-intI1 sub-databases, respectively (Fig. S5, Table S3).
Of these best-performing sets, the F16-R16 primer set amplified the highest number of intI1-like amplicons (n = 11, 79%) at 0 mismatch and was removed, while the primer pair F57-R57 amplified the lowest number of intI1-like sequence (n = 7) (Fig. S5A, Table S3).The incorporation of a TaqMan probe generally improved primer specific ity; however, two of the primer sets that incorporated a TaqMan probe (DF-P-DR and F7-P-R7) both amplified intI1-like sequences.The number of intI1-like amplicons amplified by the DF-P-DR (n = 9) and F7-P-R7 (n = 8) primer-probe sets at a 0 WS were similar.Of note, while the 10 best-performing was focused on, the other primer sets analyzed (see section Evaluation of primers for coverage) also amplified intI1-like sequences (Fig. S5A; Table S3).
Next, the nine remaining primer sets were tested against the non-intI1 sequences (Fig. S5B and C).None of the primers amplified the non-intI1 sequence at a 0 mismatch.In general, primers only produced amplicons from the non-intI1 database with very high weighted scores (sum of forward and reverse primer mean WS ranged: 8.39-11.6)(Fig. S5B and C; Table S3).However, the primer pairs F1-R1 (WS: 2) and F13-R13 (WS: 3.2) performed worst, having the lowest WS required to amplify at least one non-intI1 target.As such, were removed from further analysis, leaving seven sets (DF-DR, F3-R3, F7-R7, F31-R31, F35-R35, F57-R57, and F60-R60) to be considered the best overall performing intI1 primer sets in terms of coverage and specificity.

Recommendation of optimal primer sets for in situ laboratory validation and in silico amplicon size distribution
From the initial 65 primer sets, seven (DF-DR, F3-R3, F7-R7, F31-R31, F35-R35, F57-R57, and F60-R60) were identified that had high coverage in our intI1 database, but low specificity to the non-intI1 database, indicating they are good primer sets targeting a broad range of intI1 targets while discriminating against non-intI1 sequences.Two published sets (F3-R3 and F7-R7) were selected (Table S3), as they not only had the highest WS required to amplify non-intI1 target [i.e., needed the highest number of mismatches to target the sequence; F3-R3-WS: 5.2 (2 non-intI1 amplicons), F7-R7-WS 5.2: (six non-intI1 amplicons)] but also had short amplicons (100-200 bp), making them ideal for both Q-PCR and high-throughput amplicon sequencing.In addition, each of these selected primer sets targeted a different region of the intI1 gene and was commonly used within the literature (28,31,(46)(47)(48)(49). F7-R7 incorporated a TaqMan probe.The primer and probe set, DF-P-DR, designed in this study, were also included resulting in three intI1 primer sets selected for laboratory validation.
Each of the standard curves from all three primer sets had high efficiencies which ranged from 91.29% to 95.7%, y-intercepts of 35.71 to 39.6, the slope of −3.43 to −3.55 and a No Template Control Ct from undetected to 36.9 (Table 2).
Although a similar overall pattern of intI1 gene abundance was observed, the F7-R7 primer set was the only primer set that reported no statistical difference (P-value > 0.05) in gene abundance between samples (influent, sludge, and effluent) and reac tors (CST-household, CST-healthcare, and SST-household), while only DF-DR primer set showed significantly higher intI1 gene abundances in the effluent of the CST healthcare than all other samples (Fig. 2).
SST-household units incorporated an internal pasteurization effect and were therefore expected to have lower intI1 gene abundance in the effluent.intI1 gene abundance per ng of DNA was lower in effluent than in both the CST-household and CST-healthcare tanks for all three primer sets (Fig. 2).However, these differences were only statistically significant for DF-DR primer se (P-value = 0.005) between SST-house hold and CST-healthcare effluent (Fig. 2).Nonetheless, the lower intI1 gene abundance quantified in the solar septic tank (SST-household) effluent, albeit only statistically lower in the CST-healthcare for the DF-DR primer set.
Although intI1 gene copies in the sludge of the reactors were lower than the effluent, these intI1 gene copies were still high, albeit that the gene abundance between the three reactors was marginally different but not statistically significant (P-value > 0.05), regardless of the primer set (Fig. 2).However, depending on the primer set used, the reactor with the higher abundance and thus likely to contribute most to the environment changed.
The DF-DR primer set reported the CST-healthcare (1.35 × 10 4 ± SD8.45 × 10 3 copies/ng DNA) to be the higher contributor of intI1 gene to the environment via sludge and the SST-household unit (7.86 × 103 ± SD1.43 × 10 3 copies/ng DNA) to be FIG 2 Impact of primer choice on the quantification of intI1 gene copies from CST-household, CST-healthcare, and SST-household septic tank wastewater reactors, and three wastewater sample types (influent, effluent, and sludge).Results of the two-way ANOVA analysis showed a statistically significant difference in intI1 gene copies quantified between reactor types and sample types.For each primer set, a boxplot sharing the same letter indicates no statistically significant difference at P-value > 0.05, while the boxplot with different letters indicates a statistically significant difference at P-value < 0.05.A statistically significant difference in intI1 gene abundance between primer sets for the same sample was not observed (P-value > 0.05; see supplementary table s.vi).X icon indicates mean intI1 copy number/ng DNA.The black dot indicates the data outlier.the lowest of the three reactors (Fig. 2A), while primer set F3-R3 also indicated the CST-healthcare sludge as the higher contributor (9.58 × 10 3 ± SD5.27 × 10 3 copies/ng DNA) of CL1-intgeron to the environment via sludge, but reported the CST-household (7.30 × 103 ± SD3.11 × 10 3 copies/ng DNA) as the least contributor (Fig. 2B).Finally, the F7-R7 primer set revealed the CST-household unit sludge (8.42 × 10 3 ± SD4.10 × 10 3 copies/ng DNA) to be the greater contributor of intI1 to the environment and the SST-household (8.06 × 10 3 ± SD1.59 × 10 3 copies/ng DNA) as the least contributor (Fig. 2C).As such, the SST-household in general had the lowest intI1 gene abundances in sludge when primer sets DF-DR and F7-R7 was used, but not when the F7-R7 primer set is used (Fig. 2).
In summary, primer sets used did not change the overall pattern of intI1 gene abundances nor did it result in statistical difference (P-value > 0.05) in intI1 gene abundance for the same sample type (influent, sludge, and effluent) quantified with the different primers.However, comparing samples within the same primer set did sometimes result in statistical differences between samples., which could alter the interpretation of the risk of intI1 gene abundances, and, in turn, AMR pollution to the environment.similar to the intI1 sequence from activated sludge, although with a low bootstrap value (Fig. 3A).
For primer set F3-R3, a single cluster was observed, supported by a 95% bootstrap value containing ASV-1, 3, and 4.These clustered with unknown and known intI1 from Tannery effluent and activated sludge as well as known intI1-like sequences of clinical origin (Fig. 3).While ASV-1 was present in all samples, ASV-3 and 4 were only detected in the CST-household effluent (CST-P3_08-19).Outside of this cluster was ASV-2, highly similar to intI1 from Acinetobacter baumannii, a clinical pathogenic bacterium.ASV-2 was present in both the CST-household and SST-household tanks sludge and effluent but only present in one CST-healthcare sludge sample (CT-HC_09-19) (Fig. 3B).As primer sets DF-DR and F3-R3 targeted the same region of the intI1 (Fig. S1), the ASVs generated by each primer set (DF-DR and F3-R3-ASV-1; DF-DR-ASV-2 and F3-R3-ASV-3; DF-DR-ASV-3 and F3-R3-ASV-4) had 100% sequence similarity to each other but only ASV-1 from each primer set showed a 100% sequence similarity when aligned against the full-length intI1 (C) intI1 primer sets.Generated ASVs coupled with known and unknown intI1 within SDB1 (n = 104), best hit NCBI sequences, and intI1-like sequences were aligned with Mafft, trimmed to only aligned region with no gaps, and phylogenetic tree constructed using the RAxML with 1,000 bootstrap permutations.The number at a node represents a bootstrap value > 50% (from 1,000 permutations).The bootstrap value at node < 50 is not shown.The class 3 integron-integrase (intI3) gene (nucleotide ID: AY219651.1),which on the protein level, shared a 60.74% similarity to the pVS1 protein sequence (AAA25857.1)was used as the outgroup.The color of tree tips indicates the isolated source of sequence/ ASVs generated by the primer set.Heatmap shows log2 fold abundance (mean number of ASVs-DF-DR:5.1955× 10 4 , F3-R3: 4.6602 × 10 4 and F7-R7: 3.6684 × 10 4 ; Table S7) of detected ASVs within each wastewater sample.CTP3 and CTJ6 samples originated from two independent CST-household reactors.CT-HC sample was from a CST-healthcare tank.ST01 and ST07 are two independent SST-household units.The sampling month and year are indicated by the format month_year (i.e., 06_19 = June 2019).CST, conventional septic tank; SST, solar septic tank.nucleotide sequences (pVS1, M73819.1).In addition, F3-R3-ASV-2 did not align with the full-length intI1 with a 100% similarity.
Primer set F7-R7 detected 11 ASVs separated into two clusters.Within cluster I, ASV-1 was present in all samples, clustered with ASV-8, 4, 7, 11, 10, 5, and 6 and was detected in CST-household (sludge), CST-healthcare (sludge and effluent), and SST-household (sludge and effluent) reactors.It clustered with known intI1 from sources such as hospital sewage and Tar-Pond but also intI1-like sequences.Clustering was not supported by a high bootstrap value.Within cluster II, ASV-9 and 2 were detected in CST-household influent, CST-healthcare sludge and effluent, and SST-household effluent samples and clustered with unknown and known intI1 sequence, as well as intI1-like sequence, found in environmental source such as sediment and wastewater biofilm.However, clustering was not supported by a high bootstrap value (<50%) (Fig. 3).A final ASV (ASV-3), again only detected in the CST-household effluent (CST-P3_08-19), clustered outside the main group, but was not supported by a high bootstrap value (<50%) (Fig. 3C).
In summary, intI1 diversity showed all samples to be dominated by a single ASV-1.It was highly similar to intI1 from clinical and environmental samples; however, intI1-like samples also clustered with it.CST-household had the highest richness with primer set DF-DR and F3-R3, but a different picture arose with the F7-R7 primer set with the CST-healthcare effluent having the highest intI1 diversity.

Laboratory validation of selected intI1 primers to quantification intI1 mRNA transcript from environmental samples
As the detection of intI1 DNA does not infer integrase activity, each of the validated primer sets was tested for their ability to quantify intI1 mRNA transcripts.For this, fresh river water samples were used as the quality of the RNA extracted wastewater nucleic acids may be of poor quality due to long-term storage (33), although the quality was not measured.For each primer set, the reverse transcriptase reaction was carried out with random hexamers (RH) and genespecific (GS) primers as previous work showed increased specificity with genespecific priming (50).In addition, Q-PCR quantifications were carried out with (Fig. 4C) and without the probes (i.e., SYBR Green) (Fig. 4A and B; Table 4).All primer sets successfully quantified intI1 DNA and mRNA from river water, with intI1 gene abundances greater than intI1 transcripts (Fig. 4).
As previously shown (50), genespecific priming was more efficient than random hexamer priming.The F7-R7 primer set did not work as a TaqMan probe assay but worked in a SYBR green assay.It should be noted that, while higher intI1 transcript copies per ng DNA were quantified by the F3-R3 SYBR green assay (1 in two dilution), direct comparison to DF-DR and F7-R7 cannot be made as they were not done on the same sample.The aim here was simply to demonstrate that the primer sets were able to quantify intI1 mRNA transcripts.In summary, the primer sets tested are appropriate to quantify intI1 mRNA transcripts from environmental samples, using both genespecific and random hexamer priming, albeit the TaqMan probe chemistry must be swapped to SYBR green chemistry if the F7-R7 primer set is to be used.

DISCUSSION
Accurate quantitative data are key to inform evidence-based management strategies and policies to reduce the global AMR burden.Quantitative approach, alongside unified methodologies to enable comparison among data sets, is a powerful tool to enable this.The clinical class 1 integron (CL1-integron) integrase gene (intI1) has been proposed as a proxy for inferring potential AMR (8).The first step to investigating the potential for this is to select appropriate primers; however, our systematic literature review revealed over 65 intI1 primer sets with little consensus on the best primer to use.Through in-silico testing of the published primer sets, in addition to the design of an optimized primer set in this study, we selected three intI1 primer sets for laboratory validation and further testing of their specificity on septic tanks from Thailand associated with healthcare and household usage to investigate their contribution in disseminating CL1-intgerons to the environment.This included a novel solar septic tank designed with internal heating ranging from 39°C to 63.6°C in the disinfection chamber.From the 65 primers in the literature, three were selected-two published primer sets, F3-R3 (28) and F7-R7 (31) which have been extensively applied to survey CL1-integron abundance in a range of ecological settings including WWT (48,49) and agricultural settings (46,47) and a newly designed primer, DF-DR, modified from the F3-R3 primer set and an MGB probe added to increase specificity showed good coverage and specificity.All were successful PCR and RT-Q-PCR assays.
To confirm their specificity, MiSeq amplicon sequencing of the short amplicons was undertaken from the Thai septic tank samples.While the diversity of the amplicons was low, likely reflecting the short amplicon length, the intI1 gene was ubiquitous in our samples supporting previous findings where it was dominant in polluted environ ments such as WWT (8,51).The ASVs generated from each primer set were highly similar to intI1 and intI1-like sequences obtained from known and unknown bacteria, which were isolated from a range of clinical settings (e.g., human commensal) and environmental sources (e.g., wastewater-activated sludge) (Fig. 3).Interestingly, a few of the intI1-like sequence characterized were from known bacterial species isolated within clinical context including human feces and bloodstream (Fig. 3).This observa tion challenges the well-established knowledge that intI1 sequence recovered within clinical settings has identical/nearly identical protein (≥98% protein identity) (24) and/ nucleotide sequence (99%-100% nucleotide identity) (18) to each other.As such, this FIG 4 IntI1 DNA and mRNA transcript quantified from a river water sample by the three selected intI1 primer sets (DF-DR, F3-R3, F7-R7).The reverse transcriptase reaction for each primer set was performed with random hexamers (RH) and genespecific (GS) primers.In addition, TaqMan assays were carried out with (C) and without the probes (i.e., SYBR Green) (A, B).NT denotes not-tested, and ND denotes non-detected.
implies that intI1-like sequences can also be present within clinical settings and not just restricted to environmental settings as originally thought.
Our sequence results also highlight that the primer sets show that it was not possible to distinguish between intI1 and the lesser conserved intI1 variants (intI1-like, <98% protein similarity) that have been shown to coexist within these settings and similar environments (8,18).These less conserved CL1-integron integrases (intI1-like) have been found, for example, on the chromosome of non-pathogenic Betaproteobacteria isolated from biofilms and soil, and the entrained gene cassettes encoded other functions rather than AMR (19).intI1-like may therefore not contribute to AMR but will contribute to intI1 Q-PCR signal.None of the primers, not even with the addition of a TaqMan or MGB probe, were able to distinguish between intI1 and intI1-like.As such, quantified intI1 gene abundance could potentially be overestimated.However, designing new primers over longer region but still suitable for Q-PCR, capable of distinguishing both variants can be a challenge.This is because the intI1-like protein sequence identity between bacteria species can vary when compared to the reference intI1 P. aeruginosa reference sequence (pVS1, AAA25857.1).For example, the IntI1-like protein sequence from Salmonella enterica subsp (KMJ40944.1),a gamma-proteobacteria and Comamonas thiooxydans beta-proteobacteria (WP_012838479.1)shared 87.8% and 92.8% identity to the reference pVS1 IntI1 protein sequence, respectively.As such, the varying conserved region shared between the IntI1, and IntI1-like protein sequence variant makes it a challenge to design a primer that exclusively distinguish both variants.
The potential contributions of intI1-like abundance to the overall abundance of intI1 gene quantified via Q-PCR suggest that intI1 abundance may not be an adequate or reliable proxy for inferring overall AMR abundance.Therefore, other potential proxies such as the qacEΔ1 (confer antiseptic resistance) or VanA (confer vancomycin resistance) (52) should be investigated for reliable estimation of overall AMR abundance in polluted environments.
This work has also shown the impact of using different primers on the interpretation of the findings and, in turn, our understanding of the risk of AMR.While across our septic tanks, the three best-performing primer sets revealed the same overall trends (Fig. 2), they did on occasion change the statistical difference between samples.For example, there were statistically higher intI1 gene abundances in the effluent than the sludge of the CST-healthcare unit when quantified with one (DF-DR) of the three primer sets but no difference when using the other two (F3-R3 and F7-R7) (Fig. 2).Depending on the primer set used, our understanding of the role of wastewater in the dissemination of CL1-integron and entrained AMR gene to the wider environment differed, highlighting the need for primers standardization whether comparisons and environmental meaning are to be gained from the large body of literature and work currently being undertaken in this area.With this in mind, from the work carried out validating and comparing the primer sets, we arrived at three very good primer sets, albeit with the lack of specific ity for intI1.As the addition of the TaqMan and MGB probes did not offer increased specificity, we recommend F3-R3 primer set and SYBR green assay (28).This primer set has previously been extensively used in the literature to survey CL1-integrons from a wide-ranging environment (49) and here we have further demonstrated their suitability to quantify mRNA also.For this, a genespecific RT-Q-PCR assay performed best as previously demonstrated (50).
The combination of quantitative PCR and amplicon sequencing approach offers a rapidly targeted and costeffective alternative, in contrast to shotgun metagenomics.This approach permits reliable and accurate profiling of functional genes from various environment, and is, therefore, highly recommended in future studies.

Ecological risk assessment of septic tanks in contributing to intI1 gene abundance to the environment
Comparing the abundance of intI1 genes (copies/ng DNA) among the different septic tanks, we showed that they were higher in the effluent compared to sludge, for all three reactors (CST-household, CST-healthcare, and SST-household), irrespective of the intI1 primer set used, with the highest gene abundance quantified in the conventional healthcare (CST-healthcare) effluent (Fig. 2).This finding was consistent with a previ ous study that reported higher intI1 relative gene abundance (normalized abundance to the 16S rRNA copies) in hospital effluent compared to urban or municipal WWTP effluent (48).Healthcare institutions are among the primary consumers of antimicrobials particularly antibiotics (48).As such, stronger selective pressures are imposed within the bacteria communities, which, in turn, drives the acquisition of resistance genes carried within key vectors such as CL1-integron, to ensure their survival from the constant threat of antimicrobials within theWWT system.
Of the three reactors, lower intI1 gene abundance per ng of DNA was quantified in the household solar septic tank (SST) samples (sludge and effluent) compared to the conventional tanks (CST-healthcare and CST-household) by two (DF-DR and F7-R7) of the primer sets while the third (F3-R3) only quantified lower intI1 gene copies in the SST-household effluent and not the sludge sample (Fig. 2).Nonetheless, this implies that the increased temperature potentially plays a role in reducing CL1-integron from WWT and thus, the abundance entering the environment.This finding agrees with our proposed hypothesis of decreased intI1 gene abundance as a result of increased temperature driving enhanced wastewater treatment.Although the target internal temperature (50°C-60°C) within the solar tank was not consistently achieved, our finding is consistent with the recent study by Zhang and colleagues (53), who investigated removal of CL1-integron and entrained AMR genes from anaerobic digestors operated at higher (thermophilic: 55°C) and lower (mesophilic: 35°C and 25°C) temperatures and reported statistically lower intI1 gene abundance and removal at higher temperature.In addition, statistically lower 16S rRNA gene abundance was reported at the higher temperature, coupled with a lower relative abundance of AMR gene cassettes, albeit slightly higher ARG subtypes were detected with the higher temperature.
Although typified by poor treatment performance (7), the conventional household tank was able to reduce intI1 gene abundance in the effluent from the influent by 31.21% to 42.33%, depending on the intI1 primer set used.This finding is consistent with a previous study by Chen and Zhang (13), although Chen and co-worker reported higher intI1 removal in the effluent from the influent (estimated around 1.9 to 2.3-log removal) for two of the three onsite domestic WWT associated with single-family usage investigated than observed in this study.In the third onsite, domestic WWT associated with single-family use enrichment of intI1 gene abundance was reported.However, the better removal from the two tanks in their study (13) may be due to the additional secondary treatment incorporated into the tank, such as ecofilter, constructed wetland, and multi-soil layering, prior to discharge to the environment which was not done in our study.
WWT sludge represents an additional source of CL1-integron and entrained AMR genes to the environment, particularly if improperly managed (i.e.improperly disposed of without further treatment), which further exacerbates the global AMR burden (54).In the Global south regions such as Thailand and Vietnam, only 10%-20% of the fecal sludge generated are estimated to be adequately disposed of, while the vast majority are discharged directly to the environment (54).With the high intI1 abundance quantified in the sludge for the three reactors, coupled to the already high abundance in the effluent, we found that, on average, 1.22x10 5 to 1.48 × 10 5 , 8.41 × 10 4 to 1.1 × 10 5 , and 7.73x10 4 to 9.4 × 10 4 intI1 gene copies per ng DNA (depending on primer set), enters the environ ment via the CST-household, CST-healthcare, and SST-household, respectively.This is significant when taking into account the proportion of global population (2.7 billion people) estimated to be served by onsite decentralized WWT including septic tanks (55).Thus, highlighting septic tanks as an important source of CL1 to the environment, further supports the broader knowledge that WWT in general, is a major source of CL1-integrons and entrained resistance genes to the environment.For the CST-household tank with accessible influent sample, while the load of intI1 was decreased from the influent, the abundance of intI1 quantified in the sludge and effluent by the different primer sets represent a significant source of intI1to the environment and therefore, emphasizes the need to optimize the conventional septic tank for AMR removal.
The increased abundance of CL1-intgerons entering the natural environment from WWT coupled with a slow decay rate [intI1 halve-life estimated ≥ 1 month in soil (56)] increases the risk of acquisition and dissemination into broader bacteria taxa, especially clinically relevant human pathogenic bacteria including Acinetobacter baumannii (57), Proteus mirabilis (58,59), and P. aeruginosa (60,61).

Conclusions
This present study has provided insight into the importance of primer choice especially in the context of validating the intI1 as a suitable proxy for AMR pollution, and the need for standardization across studies to comprehensively understand the role in which wastewater plays in disseminating CL1-intgerons and by extension AMR genes to the environment.Further work is needed to determine whether the intI1 is indeed a suitable proxy for overall AMR gene abundances.
Moreover, we showed septic tank decentralized wastewater, particularly the conventional healthcare tank (CST-healthcare), can be a significant source of CL1 integron to the environment via the effluent and sludge if the sludge is directly applied to the environment without undergoing additional treatments, such as wetlands, to reduce the intI1 gene load.Thus, supports growing evidence that WWT in general is a major source of CL1-intgerons and associated resistance genes to the wider environment which further exacerbates the global burden from AMR.

TABLE 1
Criterion for construction of integrase sub-

databases Sub_databases ID Criteria Number of sequences within sub_database Nucleotide sequence length (bp) Beginning start codons Ending stop codons
a SDB, sub-database.

TABLE 2
Primers and probe sets selected and optimized for Q-PCR to quantify intI1 gene copies from wastewater a

Primer ID Sequence (5′-3′) Orientation Target (length) Assay type Q-PCR Experimental cycle condition Q-PCR standard curve descriptors Reference Efficiency (%) R 2 Slope Intercept NTC
(Continued on next page) Full-Length Text Applied and Environmental Microbiology November 2023 Volume 89 Issue 11 10.1128/aem.01071-237 TABLE 2 Primers and probe sets selected and optimized for Q-PCR to quantify intI1 gene copies from wastewater a (Continued) Primer ID Sequence (5′-

TABLE 3
Selected sample timepoint for each septic tank investigated a,b a INF = influent; EFF = effluent; SLG = sludge; SST = solar septic tank; CST = conventional septic tank.b Excluded from intI1 Q-PCR quantification due to insufficient sample volume but was included in MiSeq amplicon sequencing.

TABLE 4 intI1
mRNA transcripts copies/ng DNA a