Water samples were collected in a sterile falcon tube from hospital drainage system and Buriganga river. In total ten and twenty water samples were collected from hospital drainage and Buriganga river respectively. Collected samples were transferred aseptically to the laboratory and processed by diluting four times with distilled water and then stored in a refrigerator at 4˚C. Samples were grown in selective Salmonella-Shigella-agar (SS agar) media. After overnight incubation, 20 bacterial colonies were selected for further experimental analysis. All samples were collected in the period of 2015 to 2016.

Characterization of the bacterial isolates was carried out using biochemical tests for instance Motility Test, Catalase Test, Gelatin Hydrolysis Test, Indole Test, Urease Test, Triple Sugar Iron Agar Test. All the tests were performed according to the protocol of ASM Microbe Library (http://www.microbelibrary.org). And 16S rRNA gene based polymerase chain reaction (PCR) was performed using 16S rRNA gene primes for identification of the isolated bacteria (after sequencing the PCR product).

The Kirby-Bauer disc diffusion test was carried out with the isolated bacteria to determine the sensitivity or resistance towards various commonly used antibiotics (n = 11). The concentration of antibiotics those were used in this study are Ampicillin (AMP) 25 μg; Nalidixic acid (NA) 30 μg; Tetracycline (TE) 30 μg; Kanamycin (K) 30 μg; Doxycycline (DO) 30 μg; Streptomycin (S) 10 μg; Ciprofloxacin (CIP) 5 μg; Penicillin (P) 10 μg; Cotrimoxazole (COT) 25 μg; Gentamicin (GEN) 10 μg & Chloramphenicol (C) 30 μg. Clustering of bacterial strains based on the antibiotic susceptibility pattern was performed by using K-means clustering algorithm.

In Silico analysis is an important medium for initial identification and characterization of bacterial strain. In this analysis, different Bioinformatic tools were utilized to identify the isolated bacterial strain and to study CRISPR-cas system of isolated Enterobacteriacaea.

Basic Local Alignment Search Tool (BLAST) is an online program of National Center for Biotechnology Information (NCBI) (https://blast.ncbi.nlm.nih.gov/Blast.cgi). To identify the bacterial isolates and other similar organism having similar sequence BLAST program was used.

This site acts as a gateway to publicly accessible CRISPRs database and software (http://crispr.i2bc.paris-saclay.fr/). It enables the easy detection of CRISPRs in locally-produced data and consultation of CRISPRs present in the database. This web site is the product of an original work by Ibtissem Grissa (PhD thesis Paris University), improved by Christine Drevet, and presently developed by David Couvin. CRISPR locus information was derived from this database.

Molecular Evolutionary Genetics Analysis (MEGA) is computer software for conducting statistical analysis of molecular evolution and for constructing phylogenetic trees (Tamura et al., 2013). It includes many sophisticated methods and tools for phylogenomics and phylomedicine. MEGA was used for the comparative analysis of molecular sequence data was used for reconstructing the evolutionary histories of species and inferring the nature and extent of selective forces shaping the evolution of genes and species.

In total seven and thirteen bacterial strains were isolated from Buriganga river and medical waste water respectively. The identification pathogenic bacterial strains in the Buriganga river points towards the possibility of contaminating their river water medical sewage containing pathogenic bacterial strains from nearby hospital. Because hospitals always deal with numerous patients with infectious disease, so the drainage system of hospitals might be contaminated with pathogens and could be a good source for antibiotic resistant bacteria. Another site for sample collection was Buriganga river, because this river water is contaminated with the sewerage water dumped by numerous river side industries and hospitals. After isolating bacteria through culturing in Salmonella Shigella (SS) agar media, various biochemical tests were performed to identify the bacterial isolates. Results showed most of the bacterial isolates were positive for the test of motility (N = 13), catalase (N = 20), urease production (N = 18), and capacity to gelatin hydrolysis (N = 12) but all were negative for gram stain and indole production.

Antibiogram of the isolated bacteria were shown in a HeatMap (Figure 1) where the strains were divided into two Groups A & B. The strains of Group A showed less resistance against various antibiotics than Group B. But Strain-5 was a hospital isolate which showed most resistant against the used antibiotics, therefore this strain didn’t include into those two groups. The bacterial strains of Group A except strain 13 and 14 were isolated from Buriganga river whereas Group B strains were isolated from hospital drainage system. Results indicated that multi-drug resistance bacteria were more prevalent in hospital drainage system as

compared to Buriganga river. Previously it was found that lactose fermenting Enterobacteriaceae isolated from clinical samples were more resistant to antibiotic than environmentally isolated lactose fermenting Enterobacteriaceae [19]. It is possible that the bacteria isolated from hospital drainage system may originate from infected patient of those hospitals. And these bacteria were previously exposed to antibiotics during treatment. Therefore, it is not unlikely that due to prior exposure to antibiotics these bacterial strains developed antibiotics resistance [20] . Whereas the bacterial strains isolated from Buriganga river may have remained unexposed to antibiotics and show less antibiotic resistance.

The next question was to determine the percentages of bacterial strains which showed resistance to multiple antibiotics. Therefore, a pie chart was created that confirmed 35% of bacterial strains were resistant to four different antibiotics. The percentage of resistance against multiple antibiotics among the isolated bacterial strains is presented in Pie chart (Figure 2). All the bacterial isolates show multiple resistance patterns and one strain found to be resistant against ten among the eleven antibiotics.

The Basic Local Alignment Search Tool (BLAST) was used to identify the unknown query sequence (16S rRNA gene) that match best with the known sequence present in NCBI databases. The sequence of 16S rRNA gene of Strain 1 to Strain 12 bacteria nucleotide were blasted using online blast tool in National Center for Biotechnology Information (NCBI; Altschul et al., 1997; http://www.ncbi.nlm.nih.gov/blast). The 16S rRNA gene sequences of 12 bacterial strains have been deposited in the GenBank data base and accession numbers have been obtained. The name alone with the accession number obtained from the BLAST results of bacterial isolates are given in Table 1. Their source and name of susceptible antibiotics are also given.

For species identification of bacteria molecular approach is needed. 16S rRNA gene is considered as the gold standard in identifying bacterial strains by molecular methods. 16S rRNA gene sequence is the universal genetic marker for identification of bacterial strains as all it is most common housekeeping gene in all bacteria [21] [22] . Genomic DNA of the bacterial isolates were extracted using alkaline lysis buffer. Identification of the bacterial isolates were conducted by Polymerase chain reaction (PCR) based assay of 16S rRNA gene and DNA

*DMC = Dhaka Medical Collage; AMP = Ampicillin; NA = Nalidixic acid; TE = Tetracycline; K = Kanamycin; DO = Doxycycline; S = Streptomycin; CIP = Ciprofloxacine; P = Penicillin; COT = Cotrimoxazole; GEN = Gentamicin; C = Chloramphenicol.

extracted from the isolated strains were amplified by using 2 universal primers 27F and 1492R (Figure 3). 1500 base pair [21] sequence of 16S rRNA gene was sequenced by Sanger sequencing method after performing PCR. Sequencing of first 12 isolated bacterial strains (Strain 1 to Strain 12) was done. Molecular identification reveals the isolated bacteria belongs to Proteus & Citrobacter bacterial strain. These bacteria are more prevalent in the environmental samples as they are mostly found in the human fecal excretion [23] . And samples for this study were collected from drainage system and river. Highly similar nonrodent nucleotide BLAST results of the sequences show that Strain 8, 9 & 12 includes Citrobacter freundii bacterial species and Strain 1, 2, 3, 4, 5, 6, 7, 10 & 11 includes Proteus mirabilis bacterial species. Homology between the bacterial isolates were found when a phylogenetic tree was constructed using the 16S rRNA gene sequences. This suggests Citrobacter freundii and Proteus mirabilis poses similar type of 16S rRNA gene.

A phylogenetic tree was constructed using MEGA6 software. Twelve sequences of 16S rDNA from the sequencing result were used to construct phylogenetic

tree. Phylogenetic relations among the bacterial strains are shown in the constructed phylogenetic tree (Figure 4).

To identify CRISPR locus in the bacterial isolates (Proteus & Cirobacter), PCR was performed using two CRISPR locus primers CRISPR-1, and CRISPR-2. Results confirmed that there was no CRISPR locus present in the isolated bacterial strains after performing PCR followed by agarose gel electrophoresis (Figure 5). As Most of the bacterial isolates showed antibiotic resistance, therefore the lack of CRISPR locus may facilitate the acquisition of antibiotic resistance cassette in these bacterial strains. On the contrary, we cannot rule out the possibility that the bacteria might have atypical CRISPR locus that didn’t match with our universal primers sequence or they may have CRISPR locus not in the genomic DNA but in their plasmid.

Acquiring new genes are important for bacteria to become antibiotic resistance. Bacteria acquire new genes by horizontal gene transfer [12] . CRISPR-cas has recently been identified in many bacteria and archeaea as adaptive immune system that prevent acquisition of extra chromosomal genetic elements into them. So, there may be an inverse correlation between antibiotic resistance and CRISPR locus in bacteria. As CRISPR-cas system play potential role in limiting antibiotic resistance in the bacteria by preventing phage mediated horizontal gene transfer of antibiotic resistance genes in bacteria as previous studies showed that the bacteria capable of incorporating phage DNA sequence as spacers into the CRISPR loci and become resistant to that phage infection [15] . That’s why a

statistical study was conducted for CRISPR locus and molecular approach was used for identification of CRISPR locus in isolated isolates bacterial.

In total genome and plasmid sequences of 14 and 7 bacterial strains from Citrobacter and Proteus species respectively were searched for the presence of putative CRISPR locus. CRISPR finding tool as implemented in CRISPRdb was utilized to annotate potential CRISPR locus.

Surprisingly most of the analyzed genomes were found to be devoid of any CRISPR locus. Only two Citrobacter strains (Citrobacter koseri ATCC BAA-895 and Citrobacter rodentium ICC168) out of 14 was observed to harbor CRISPR locus (Table 2). In case of Proteus three out of five strains (Proteus mirabilis

GCF_001281545, Proteus mirabilis GCF_001640985 and Proteus mirabilis GCF_002055685) were found to be positive for CRISPR locus (Table 2). Interestingly no CRISPR locus was identified in the plasmid rather all the strains harbor CRISPR locus in the chromosomal DNA. All together these results strengthen the assumption that CRISPR locus in Citrobacter and Proteus strains are of chromosomal origin and may not be present in plasmids. The low number of CRISPR locus in the Citrobacter and Proteus strains as obtained from database reinforced our hypothesis that Citrobacter and Proteus strains analyzed in the current study may well lack the active CRISPR locus all together and for this reason no CRISPR locus was identified by PCR. Therefore our results can serve as a solid platform to further in depth association study between CRISPR locus and antibiotic resistance.