IDENTIFICATION OF Bacteroides spp. FROM DUCKS USING 16S rRNA GENE PCR ASSAY: PRELUDE TO ITS APPLICATION IN MICROBIAL SOURCE TRACKING

Fecal pollution monitoring using fecal indicator bacteria has many limitations. Alternatively, using 16S rRNA gene Bacteroides-based polymerase chain reaction (PCR) showed promise because Bacteroides spp. are a major part of the gut microbiota. In the Philippines, duck backyard farming waste management is unmonitored. Hence, the objective of this study was to isolate Bacteroides DNA from duck feces and to establish genetic relatedness that can be used to develop new microbial source tracking markers. Fifty (50) duck fecal samples were collected. Of the 50, 20 were subjected to total DNA extraction and the remaining 30 were used for the isolation of Bacteroides. A total of 32 Bacteroides were isolated that included 6 B. nordii and 26 B. ovatus. A phylogenetic tree showing the host sources of Bacteroides generated five clusters. Four clusters came from this study, while the fifth cluster was composed of sequences from a study in Japan. This shows an intraspecies difference between Bacteroides species in the Philippines and Japan. This difference is a result of variations in the gut microbiome such as host species, diet, and geography. Results of this study will expand the present sequence database of Bacteroides from ducks that can be used in identifying sources of fecal pollution in the environment.


INTRODUCTION
Water monitoring of fecal indicator bacteria (FIB) such as Escherichia coli and enterococci, is routinely used for determining levels of fecal pollution. However, the use of FIB has its limitations: (1) the major gut microbiota are not fecal coliforms but obligate anaerobes such as Bacteroides spp. and clostridia; (2) although E. coli shows specificity to fecal sources of contamination, the same cannot be said for other fecal coliforms since there are coliforms of environmental origin; (3) the presence of FIB exhibits poor correlation with the presence of pathogens in contaminated water, thus providing erroneous information; and (4) the presence of FIB does not indicate the origin of the fecal pollutant. If the origin of fecal contamination is correctly identified, more efficient mitigation plans can be put in place (Savichtcheva and Okabe, 2006;Oun et al., 2017). In view of these limitations of FIB-based monitoring, a new method known as microbial source tracking (MST) was developed. MST is based on the principle that there are fecal microorganisms that are more strongly associated with their specific hosts and thus may be used to indicate host-specific contamination (Waso et al., 2018). MST has been used in many countries to determine and monitor the water quality of several large bodies of water (Staley et  Among the MST methods developed, the most commonly used is the libraryindependent method (LIM). This method is based on the detection of hostspecific genetic biomarkers (e.g. 16S rRNA and virulence genes). LIM goes by the concept that genetic biomarker sequences are not only specific to fecal bacteria, such as Bacteroides spp., but also specific to the host species, thus allowing discrimination among suspected sources (Bernhard and Field, 2000;Rivera and Rock, 2011). The most established MST biomarkers are Bacteroides-based biomarkers. Bacteroides spp. are the most diverse group of microorganisms in the gut of most animals. They have shown to have very high level of genetic host-specificity because they tend to coevolve highly with their hosts (Bernhard and Field, 2000). Additionally, there have been several Bacteroides-based MST biomarkers developed to discriminate human fecal sources from other fecal animal sources. These Bacteroides-based MST biomarkers were designed to target specific diagnostic sequences within the Bacteroides 16S rRNA gene present in feces from various animals (Layton et al., 2006). Some of the widely used Bacteroides-based MST biomarkers are for humans (HF183,HuBac,BacHum,BacH), pigs (Pig-1-Bac, Pig-2-Bac), cows (BoBac, BacBov, BacCow), dogs (BacCan), geese (CGOF1-Bac, CGOF2-Bac), muskrats (MuBa01), and ducks (DuckBac) (Kobayashi et al., 2013;Koraijkic et al., 2014). Each of these MST biomarkers has shown varying degrees of sensitivity and specificity based on where the samples were collected. Hence, the need to establish MST biomarkers that would target more specific areas within the target sequence for better accuracy. The necessity to create new MST biomarkers has become even more apparent because the gut microbiome communities exhibit variations across different countries and latitudes (Yatsuneko et al., 2012;Suzuki and Worobey, 2014). For example, human MST biomarker HF183 was the best performing biomarker in the United States (California), France, Ireland, Portugal, the United Kingdom, Australia (Queensland), and Bangladesh, but it was the human MST biomarker BacHum that was found to be the best for Kenya and India. By contrast, human MST biomarker B. theta was the best performing biomarker in Darwin, Australia (Gawler et al., 2007;Ahmed et al., 2008;Jenkins et al., 2009). Hence, even places within the same geographical area (Europe [i.e. UK and Ireland] and Australia [Queensland and Darwin]) would still show differences in the performance of MST biomarkers. Therefore, the objective of this study was to determine the genetic relatedness of Bacteroides spp. (uncultured and cultured) from ducks. The results of this study will expand the present sequence database of Bacteroides from ducks that can be used in identifying sources of fecal pollution in the environment. This database can also be used in the development of more specific duck-associated MST biomarkers that will better differentiate duck Bacteroides in the Philippines and other locations that share similar duck farming condition. This study is limited to collection of droppings from farmed ducks. Moreover, all fecal samples were collected from caged ducks, although free-ranged duck farming is also common in the Philippines.
Fecal pollution monitoring using fecal indicator bacteria has many limitations. Alternatively, using 16S rRNA gene Bacteroides-based polymerase chain reaction (PCR) showed promise because Bacteroides spp. are a major part of the gut microbiota. In the Philippines, duck backyard farming waste management is unmonitored. Hence, the objective of this study was to isolate Bacteroides DNA from duck feces and to establish genetic relatedness that can be used to develop new microbial source tracking markers. Fifty (50) duck fecal samples were collected. Of the 50, 20 were subjected to total DNA extraction and the remaining 30 were used for the isolation of Bacteroides. A total of 32 Bacteroides were isolated that included 6 B. nordii and 26 B. ovatus. A phylogenetic tree showing the host sources of Bacteroides generated five clusters. Four clusters came from this study, while the fifth cluster was composed of sequences from a study in Japan. This shows an intraspecies difference between Bacteroides species in the Philippines and Japan. This difference is a result of variations in the gut microbiome such as host species, diet, and geography. Results of this study will expand the present sequence database of Bacteroides from ducks that can be used in identifying sources of fecal pollution in the environment.

Fecal sample collection
A total of 61 fecal samples were collected between February 2019 and June 2019 from the province of Laguna and Metro Manila in the Philippines. Fecal samples were collected from ducks (n = 50), chickens (n = 5), cows (n = 3), and dogs (n = 3). Freshly collected feces were transported on ice to the laboratory within 24 h. All samples were stored at -20 °C until use. Duck fecal samples were collected from those found within the duck cages. Care was taken to make sure that there was no cross-contamination of fecal samples. With the assistance of farm workers, when a duck dropped feces, we were immediately informed to collect them at once. The same procedure was done in collecting chicken and cow droppings from poultry houses and cow farms, respectively. Dog droppings were collected from a local pet shop.

Total DNA extraction from fecal samples
Genomic DNA were extracted directly from 20 fecal samples of ducks, 5 from chickens, 3 from cows, and 3 from dogs using Qiagen QIAamp Stool Mini-kit (Qiagen Sciences, MD, USA) following the manufacturer's protocol. All extracted DNA were stored at -20 °C until use.

Culture and isolation of Bacteroides spp.
The remaining 30 duck fecal samples were used for the isolation and culture of Bacteroides spp. Decimal serial dilution was done by adding 1 g of fecal sample to 9 ml of normal saline water (0.9% NaCl). Of the first two dilutions (10 -1 and 10 -2 ), 0.1 ml aliquots were spread plated in duplicates using Bacteroides Bile Esculin (BBE) agar (Laboratories Conda S.A., Spain). Plates were incubated for 4 days at 37 °C in an anaerobic condition using the Thermo Scientific TM Anaeropack TM anaerobic gas generator (ThermoFisher Scientific, USA). Preliminary indication of growth was observed when the brownish agar color of BBE agar turned to black. Distinct gray, circular, raised colonies were selected and re-streaked into fresh BBE agar for purification and incubated at 37 °C for 3-4 days in an anaerobic condition. Streaking for isolation, well-isolated colonies were sub-cultured every 3 weeks. Extraction of DNA from pure Bacteroides cultures was done using the boiling method (Garcia et al., 2015). The resulting suspension was vortexed and placed in a dry bath at 100 °C for 15 min. The resulting DNA extracts were kept at -20 °C until use.

Sequencing and phylogenetic analyses
PCR amplicons were sent to Macrogen, South Korea for purification and sequencing. Partial sequences (670 bp) were compared to similar sequences that are available in the National Center for Biotechnology Information (NCBI) using a BLAST Search. Sequences were aligned using MEGA X (Version 10.0.4). Additional Bacteroides sequences from other ducks were downloaded from GenBank with accession numbers AB666138, AB666140, AB666141, AB666152, and AB666154. Chimera sequences in the partial 16S rRNA gene sequences of Bacteroides were checked using Decipher version 2.11.03 (Wright et al., 2012) and rectified. Afterward, sequences were submitted to GenBank and were assigned the corresponding GenBank accession numbers. Using MEGA X (Version 10.0.4), a phylogenetic tree was constructed using the Neighbor-Joining (NJ) method. Bootstrap analyses of 1000 replicates (Tamura-Nei) were done to approximate the confidence of the trees. The sequence of Phocaecola abscessus (EU694176), a species within the family Bacteroidaceae to which Bacteroides also belongs, was used as an outgroup.

Isolation and molecular identification of Bacteroides spp.
A total of 32 presumptive Bacteroides isolates were grown in culture using BBE agar at an anaerobic condition. All isolates consistently exhibited a gray, raised, and circular colonies with a change in agar color from brown to black indicative of the biochemical reaction of Bacteroides to esculin (Figure 1).  All 32 isolates were identified as Bacteroides when cross-referenced with the NCBI database. A total of 6 out of 32 Bacteroides from farmed duck feces were identified to belong to B. nordii with acceptable % similarities of 99.35%-99.84%. This was supported by the phylogenetic tree ( Figure 3) showing that the NCBI-identified B. nordii did group with the reference sequence of B. nordii with high bootstrap value of 99%. On the other hand, 26 of the 32 isolates were identified as B. ovatus (97%-99%). In the constructed phylogenetic tree ( Figure  3), all the isolates identified through NCBI as B. ovatus formed low bootstrap values to the reference sequence of B. ovatus. Thus, in such case of high NCBI percent similarity but showed low bootstrap values to the reference strain, the 26 Bacteroides isolates were simply registered and submitted to NCBI as Bacteroides spp.

Figure 3
Phylogenetic relationships of 32 Bacteroides isolates from duck feces using 16S rRNA gene sequences with reference sequences of valid Bacteorides species using Neighbor-Joining (NJ) at 1000 replicates. Values >60% are only shown. Phocaeicola abscessus was used as an outgroup.

Duck (host)-origin phylogenetic analysis
The 16S rRNA gene, from the extracted total DNA of duck fecal smaples and Bacteroides isolates, were amplified and sequenced. Additional 16S rRNA gene sequences were downloaded from NCBI such as those from humans, cows, pigs, chickens, goats, and fish. The phylogenetic tree shows that the sequences clustered into several clades representing Bacteroides species coming from different animal sources (Figure 4).  (Chang et al., 2003).

Figure 4 Phylogenetic relationships of cultured and uncultured
Unfortunately, the problem with such industry is the unregulated or underregulated fecal waste disposal into small bodies of water that may eventually end up in larger bodies of water allowing exposure of the general populace. Some waterfowls cater pathogens such as enterococci, E. coli, Campylobacter spp., Salmonella spp., and Cryptosporidium spp. Ducks, for example, have been shown to harbor E. coli, enterococci, Campylobacter spp., and Cryptosporidium spp. with a prevalence of 95%, 100%, 29%, and 1.3%, respectively (Moriarty et al., 2011) , 2014). Oocysts of Cryptosporidium were detected from duck feces and high fecal coliform levels attributed to the permanent mallard duck population in a beach in Madison, Wisconsin, USA that caused its closure (Fleming and Fraser, 2001). In the study of Ramirez-Martinez et al., (2018), using next generation sequencing, duck feces contained sequences from Herpesviridae, Adenoviridae, Retroviridae, and Myoviridae viral families. These studies increase our understanding of ducks as reservoir of pathogenic microorganisms and provide the basis for studying and better monitoring of the transmission of pathogens from wild animals to humans.
In this study, we detected Bacteroides through the amplification of 16S rRNA gene from total DNA extraction and culture method from fecal samples.
Bacteroides are Gram negative, non-spore-forming, non-motile, and strictly anaerobic bacteria. They were targeted for isolation because they are known to be consistently present in high numbers in gut and fecal microbiota (Rivera and Rock, 2011). Due to their anaerobic nature, they have little capacity for growth outside the gut of their hosts, making them more suitable for MST since they provide a more accurate count in terms of the degree of fecal pollution (Rivera and Rock, 2011;Kabiri et al., 2016). Moreover, there is a strong correlation between the presence of Bacteroides and the presence of fecal pathogens such as E. coli, Salmonella, and Campylobacter as opposed to detecting the presence of traditional FIB that does not show a similar correlation Ahmed et al., 2016). Therefore, due to the characteristics of Bacteroides spp., a majority of LIM are Bacteroides-based and are designed to target specific sequences within the 16S rRNA gene to differentiate the different sources of contamination by either conventional PCR or quantitative PCR (Hussein et al., 2014).
In the phylogenetic analysis, clusters A-D came from Philippine duck fecal samples and their respective nodes/branches formed closer groups with one another, while cluster E (Japan sequences) had a branch distinctly separate from the Philippine duck clusters. This shows an intraspecies difference between Bacteroides species in the Philippines and in Japan. This difference may depend on geographical location since clusters A-D were composed of samples collected from the Philippines, and cluster E was composed of samples from Japan. This may affect the sensitivity of MST biomarkers developed in Japan and may not be applicable to Philippine environment. Another feasible reason for this grouping is the diet of the ducks. In this study, the Philippine ducks were kept in cages and fed with the same feeds twice a day. By contrast, the DuckBac biomarker designed in Japan came from wild ducks that are free to roam and get exposed to a plethora of food that they can consume. Studies revealed that Bacteroides spp. have a high level of host-specificity because they adapt to the gut conditions of their hosts and even have the tendency to coevolve with their host species. Moreover, diet produces a strong selective pressure making gut microbiomes distinct between populations ( (Coyne et al., 2014). Bacteroides poses no direct hazard when they are outside the host due to their anaerobic nature, however, within the host gut, Bacteroides can affect through HGT other bacteria or transient bacteria that can persist outside the body, to become antibiotic resistant. Thus, this study has laid the groundwork for the need to include ducks in the monitoring of fecal contamination using the Bacteroides directly isolated from ducks. The sequences acquired in this study will serve as basis for constructing duck-specific MST biomarkers.

CONCLUSIONS
Ducks have been shown to harbor pathogens that are zoonotic, hence making them suitable for developing MST biomarkers (Moriarty et al., 2011). Bacteroides spp. are the preferred bacteria for the development of MST biomarkers because of their many advantages over the traditional FIB place (Savichtcheva and Okabe, 2006;Oun et al., 2017). Although culture-based procedures take time, they provide better phenotypic characterization of bacterial isolates. Culture-based methods in conjunction with molecular and metagenomic techniques can help in strengthening the NCBI database since bulk of Bacteroides information is generated through metagenomics. The combination of methods provides a higher confidence level in the final selection of biomarkers. Afterall, metagenomics still needs the functional verification that only culturebased methods can provide. Metagenomics provides the target while culturebased methods provide the proof (Guo et al., 2014). It is recommended that since 16S rRNA gene sequences from ducks are already stored in GenBank, primer design of duck-specific MST biomarkers should be forthcoming. Future studies on antibiotic susceptibility patterns of Bacteroides from ducks can also be done. Detection of antibiotic resistance genes and Bacteroides virulence genes can also be studied since there is a gap on this field of research on Bacteroides in animals.