Microbiome and Biocatalytic Bacteria in Monkey Cup (Nepenthes Pitcher) Digestive Fluid

Tropical carnivorous plant, Nepenthes, locally known as “monkey cup”, utilises its pitcher as a passive trap to capture insects. It then secretes enzymes into the pitcher fluid to digest the insects for nutrients acquisition. However, little is known about the microbiota and their activity in its pitcher fluid. Eighteen bacteria phyla were detected from the metagenome study in the Nepenthes pitcher fluid. Proteobacteria, Bacteroidetes and Actinobacteria are the dominant phyla in the Nepenthes pitcher fluid. We also performed culturomics approach by isolating 18 bacteria from the Nepenthes pitcher fluid. Most of the bacterial isolates possess chitinolytic, proteolytic, amylolytic, and cellulolytic and xylanolytic activities. Fifteen putative chitinase genes were identified from the whole genome analysis on the genomes of the 18 bacteria isolated from Nepenthes pitcher fluid and expressed for chitinase assay. Of these, six clones possessed chitinase activity. In conclusion, our metagenome result shows that the Nepenthes pitcher fluid contains vast bacterial diversity and the culturomic studies confirmed the presence of biocatalytic bacteria within the Nepenthes pitcher juice which may act in symbiosis for the turn over of insects trapped in the Nepenthes pitcher fluid.

and 616 species were identified by metagenome analysis. The dominant bacteria phyla in Nepenthes pitcher fluid sample H1 were Proteobacteria followed by Bacteroidetes, Actinobacteria, Verrucomicrobia and Planctomycetes (Fig. 1). A total of 27% of the sequencing read does not match to any DNA sequence in the RDP database. Thus, these reads were listed as "unclassified" (Fig. 1).
Acidobacteria which is commonly found in environment with lower pH, was not the dominant bacterial group in this slightly acidic Nepenthes pitcher fluid (pH6). It contributed only 0.3% of the total bacteria in Nepenthes fluid. A total of three members from the phylum Acidobacteria were identified within the Nepenthes fluid, namely Terriglobis saanensisi (98%), Acidobacterium capsulatum (0.9%) and Candidatus Koribacter versatilis (0.8%).
Bacteria Identity. Eighteen bacteria were isolated from Nepenthes pitcher fluid sample H1. These bacteria were identified by MALDI-TOF MS Biotyper analysis on the bacterial protein profile and the phylogenetic analysis of the bacteria 16S rDNA gene. The bacteria identity obtained from both methods is listed in Table 1. The DNA sequences of 16S rDNA genes were deposited in GenBank with the accession numbers listed in Table 1.
Biocatalytic Activity of the Bacterial Isolates. We conducted the amylolytic, proteolytic, cellulolytic, xylanolytic and chitinolytic screenings and the results are shown in Table 2. Bacillus sp. strains H1a and H1m, P. aeruginosa strain H1l, Sphingobacterium sp. strain H1ai, S. marcescens strain PH1a and H1q, Pseudomonas strain PH1b, Myroides odoratimimus strain H1bi and Microbacterium paraoxydans strain DH1b possessed proteolytic activity, specifically caseinase activity, as indicated by the formation of cleared zones on the skim milk agars.
The ability of the 18 isolates in the degradation of polysaccharide of different degree of complexity was tested using starch, cellulose and xylan as substrates. Among the tested isolates, Bacillus strains H1a, H1m and Sphingobacterium sp. strain H1ai shows amylolytic activity. The degradation of the iodine/starch complex by the bacteria released iodine into the medium forming yellow zone around the colony on the dark blue background.
On the other hand, the cellulose degradation by cellulolytic bacteria was identified by the Congo red discoloration which resulted in the formation of light yellow zone or clear zone around the colony on cellulose agar 9 . Yellow zone surrounding the bacterial colony on cellulose agars was observed upon inoculation of Bacillus sp. strains H1a and H1m, Sphingobacterium sp. strain H1ai, Pseudomonas sp. strains H1h and PH1b, M. odoratimimus strain H1bi, S. marcescens strain PH1a and M. paraoxydans strain DH1b, indicating cellulolytic activity.
The screening for xylanase activity was conducted by fluorescence-based protocol with EnzChek Ultra Xylanase Assay. The xylanase substrate was labeled with fluorescence dye as described by the manufacturer.
Pseudomonas sp. Xylanase-producing bacteria cleaved the xylosidic linkages of the xylanase substrate and released the xylose(s) as well as the flurophore from the mentioned xylanase substrate. The fluorescence level was compared with the positive control and the fluorescence reference control. Results show that Pseudomonas sp. strains H1h, P. aeruginosa strain H1l, Sphingobacterium sp. strain H1ai, and Pseudomonas sp. strains PH1b exhibited xylanolytic activity. In this study, β-N -acetylglucosaminidase, chitobiosidase and endochitinase activities were tested by supplying bacterial isolates with chito oligosaccharides (4-nitrophenyl N-acetyl-β -D-glucosaminide, 4-nitrophenyl N,N' diacetyl-β -D-chitobioside or 4-nitrophenyl β -D-N,N' ,N"-triacetylchitotriose) with nitrophenol added to the substrate's terminal non reducing end. The chitinase produced by the bacterial isolates hydrolysed the β (1-4) glycosidic bond of the chitinase substrate supplied released p-nitrophenol into the solution. The p-nitrophenol was ionized by the stop solution (sodium carbonate) into p-nitrophenylate ion which appeared as yellow color to human eye and absorbed at the absorption 405 nm.

Bacterial Genome.
Genomic study was carried out on each of the bacterial isolated from the Nepenthes fluid. The estimated genome size, average coverage, G + C content, number of contigs and GenBank accession number of each bacteria are listed in Table 3. The size of bacteria genome in this study range from 3.1 to 7.4 Mbps and the genome G + C content ranges from 34% to 71.3% (Table 3). This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession number of AYME00000000-AYMV00000000. The version described in this paper is version AYME01000000-AYMV01000000.

Chitinase Gene Determination.
A total of 15 bioinformatically predicted chitinase-encoding genes were identified from the mentioned bacterial genomes and the sequences were deposited at DDBL/EMBL/GenBank with the accession numbers KT921876-KT921890 (Supplementary Table S2).
The amino acid sequences of the predicted genes were compared against NCBI-nr database followed by categorization of sequences according to members of glycoside hydrolases (GH). Thirteen of these predicted chitinase genes were grouped under the GH 18 family chitinase ( Fig. 2A)   family chitinase (Fig. 2B). The predicted chitinase gene sequences from S. marcescens H1q and PH1a were identical. Therefore, only chitinase genes from S. marcescens H1q were selected for subsequence studies. Similarly, chitinase gene sequences of S. fonticola strain H1n and strain H1w were identical. Hence only chitinase gene from S. fonticola H1n was selected for further studies. The selections were based on alphabetical order of the strain name. These predicted chitinase genes activity were validated empirically where the predicted chitinase genes were heterologously cloned and expressed in Escherichia coli BL21 followed by chitinase assay on these transformants. A total of 6 out of 11 transformants exhibited chitinolytic activity. β-N-acetylglucosaminidase cloned from contig 15 of K. oxytoca H1g and contig 36 of S. fonticola H1n showed hydrolysis of 4-nitrophenyl N-acetyl-β-D -glucosaminide and thus releasing p-nitrophenol at 1.58 and 1.45 nmole min −1 ml −1 , respectively (Supplementary Table S1). The transformants which carried the chitobiosidase-encoding genes of Pseudomonas H1h, P. aeruginosa H1l, Bacillus sp. H1m, and S. marcescens H1q were capable to hydrolyse the β-1,4-glycosidic bond of N,N' diacetyl-β-D -chitobioside at 3.70, 2.47, 4.56 and 1.97 nmole min −1 ml −1 , respectively (Supplementary Table S1). Lastly, the transformant which carried the chitinase genes of Pseudomonas H1h, P. aeruginosa H1l, Bacillus sp. H1m, and S. marcescens H1q hydrolysed the β-1,4-glycosidic bond of 4-nitrophenyl β-D -N,N' ,N"-triacetylchitotriose and thus releasing the p-nitrophenol at the rate of 0.99, 1.81, 0.83 and 0.93 nmole min −1 ml −1 , respectively (Supplementary Table S1).

Discussion
Several literatures have reported the low bacterial diversity within the Nepenthes pitcher fluid 3,10 . The low bacterial diversity is often associated with the low pH of the Nepenthes juice and the presence of antibacterial and antifungal compounds in the Nepenthes digestive fluid which created an unfavourable environment for bacteria's growth 2,3,5 . It is noteworthy that most studies conducted on Nepenthes were based on the cultured bacteria, therefore the actual bacterial diversity is yet to be discovered especially the uncultured bacteria 3,11 . Contrary to this, the diverse microbial distribution observed in this study suggested that conventional culture-dependent methods have indeed undermined the microbial diversity in the pitcher fluids [12][13][14] . In addition to this, 27% of the bacteria from Nepenthes pitcher fluid sample H1 were unclassified, as these sequences do not match to any of the sequences in the database. This suggests that the microbial diversity of Nepenthes pitcher fluid has a relativity higher complexity. However, the microbiota community within this quadrant remains unknown at the moment due to limitation of resources. The absence of these sequences from the database is because these bacteria were either yet to be discovered or still in the progress of characterization. Therefore, their 16S rDNA sequences are not available in the database.
Our data show that the dominant bacterial phyla in this Nepenthes pitcher fluid are Proteobacteria, Bacteroidetes, Actinobacteria, Verrucomicrobia and Planctomycetes. Former studies have reported their role in the decomposition of organic material in various environments including the gut of eukaryotes. For example, the phylum Proteobacteria (Enterobactericeae) decomposes monosaccharides and recycle nutrient of forest soil 15 . On the other hand, the phylum Bacteroidetes which is the major composition of human guts microbiota plays an important role in breaking down complex glycan, while the phylum Actinobacteria degrades starch and amylopectin in human gastrointestinal 16,17 . Verrucomicrobia, which was found in the guts of sea cucumber and termites, has been reported to be the key player in degrading complex polysaccharide 18 . The presence of these polysaccharides-degrading bacteria suggest there is a high probability that these microbes play a crucial role in the carbon recycling system within the Nepenthes pitcher. These microbes convert complex carbohydrates into simple sugar for the absorption of the plant via the pitcher organ. However, further study is needed in order to determine their contribution within Nepenthes digestive fluid.
The culture-dependent approach was conducted to confirm the presence of bacteria in the Nepenthes pitcher fluid. MALDI-TOF MS identification was adopted for rapid bacterial identification as it has short turn around time. Results of both identification methods, 16S rDNA and MALDI-TOF biotyping, agreed with each other ( Table 1), suggests that the environmental bacterial identification by MALDI-TOF MS Biotyper is reliable. Hence, we would like to recommend the usage of MALDI-TOF MS biotyping for environmental bacterial identification owing to its fast turnover, low operating cost and simple workflow 19,20 .
In general, most of the Nepenthes live on barren soil that lacks of nitrogen source, which is an essential macronutrient for plant growth 21 . The absorption of nutrients from Nepenthes captive by the pitcher stimulates the nutrient uptake activity via root, and thus the later activity promotes the growth of the plant 22 . This interconnect mechanism display the importance of the pitcher for the vital plant growth 22 .
In this work, screening of the bacterial biocatalytic activities shows that a total of 12 bacterial isolates were able to degrade complex carbohydrate such as starch, xylan and cellulose. The degradation of complex polysaccharides such as tree bark, plant debris, and leaves that fallen into the Nepenthes pitcher fluid releases carbon as well as nitrogen into the Nepenthes pitcher fluid, thus nourishing the plant as well as the microbes inside the fluid 8 .
The Nepenthes pitcher serves as a passive trap that lures insects, which are organisms armoured by fibrous chitin and filled with protein. Therefore, the degradation of the insect cadavers involves complex proteolytic and chitinolytic activities 23 . The nitrogenous protein stored within the insect fibrous chitin and proteins embedded within the insect's exoskeleton made up of 10% of insect's biomass 24-26 . Among the 18 bacterial isolates, 9 proteases-producing and 4 chitinases-producing bacteria strains were identified as the potential contributors in insect decomposition processes. Previous study has indicated 61.5% ± 7.6% of the nitrogen within Nepenthes plant contributed by the decomposition of insects 27 .
Previous studies have confirmed that Nepenthes secretes class I, III, and IV chitinases into its digestive pitcher fluid 6,28,29 . The upregulation of the Nepenthes chitinase gene expression in the presence of insect or chitin and the ability to hydrolyse β -1-4 glycosidic bond of chitin has led to the suggestion that the plant chitinases plays an important role in insect degradation in Nepenthes pitcher 6,7,29 . The exoskeleton of insect is the main captive in Nepenthes pitcher which is made up of primarily chitin, a polymer of N-acetyl-β-D -glucosamine 30,31 . The chitin degradation involves the cleavage of its β -1-4 linkage from the non-reducing end by exochitinase and the random cleavage of chitin β -1-4 linkage within the chitin chain by the endochitinase 30,[32][33][34][35] . The degradation of chitin involved 2 major steps, which is the cleavage of chitin into chitin oligosaccharides followed by the further cleavage to N-acetylglucosamine by chitobiases 31,36 .
In this study, 4 chitin oligosaccharides degradation bacteria were isolated from the Nepenthes digestive fluid. The whole genome analysis on the bacterial genome has determined 15 endo-and exochitinases. These chitinase genes were further classified into GH18 and GH19 families based on their amino acid sequences 37,38 . Before the documentation of GH19 family chitinase gene in Streptomyces grisues HUT6037, GH19 family was only found on higher plants 39 . However, at the point when this report was written, 1799 bacteria GH19 were documented in carbohydrate active enzyme (CAZY) database. Further screening has confirmed the β -N-acetylglucosaminidase, chitobiosidase and endochitinase activities of the chitinase genes. These suggest that the bacteria inhabit in Nepenthes digestive fluid cooperate with the Nepenthes plant for the complete digestion of the insect captive. The Nepenthes chitinase breakdown the chitin in the insect exoskeleton converting the chitin into chitin oligosaccharides, while the bacteria aid in the chitin oligosaccharides degradation, accelerating the nutrient recycling process.
The production of a wide range of biocatalytic enzymes by both the Nepenthes and bacteria inhabit in the Nepenthes pitcher digestive fluid contrives a dynamic environment in which both work in synchrony for the decomposition of insects, benefiting both the plant and its microbiota. This symbiotic system ensures constant essential nutrient supply from Nepenthes pitcher to the entire plant, thus enables the Nepenthes plant to survive and thrive in a nutrient-deplete environment. However, further studies need to be conducted in order to prove this hypothesis. Metagenome DNA Extraction. Total genomic DNA was extracted from 35 ml of Nepenthes pitcher fluid by modified cetyltrimethylammonium bromide (CTAB) DNA extraction protocol. Briefly, cells were mechanically lysed with glass bead followed by chemical lysis with CTAB lysis buffer (100 mM Tris-HCl, 100 mM EDTA, 100 mM K 2 HPO 4 , 1.5 M NaCl, 1.0% (w/v) SDS and 1.0% (w/v) CTAB) 40,41 . Enzymatically lysis was performed by addition of lysozyme (100 μ g/ml) and incubated at 37 °C for 30 min. After Proteinase K (40 μ g/ml) and RNase A treatment, the DNA was purified with phenol/chloroform/isoamyl alcohol mixtures (25:24:1) and precipitated with 70.0% (v/v) ethanol. DNA pellet was rehydrated with 35 μ l nuclease-free water and kept in − 20 °C.
Taxonomic Assignment of Metagenomic Sequences. Sequences were trimmed with CLC genomic workbench and annotated with MG-RAST (v3.3) against RDP database 21,43,44 . Rarefaction curve and alpha diversity were constructed and calculated by comparing the data to RDP database using default setting (maximum e-value cutoff: 1e −5 , minimum identity cutoff: 60.0% and minimum alignment cutoff: 15) in MG-RAST user interface. The bacteria abundance was calculated based on the data generated from MG-RAST.
Culturomics Bacterial Identification. Bacteria were isolated from highland Nepenthes pitcher fluid (sample H1) and maintained using Luria-Bertani (LB) medium at 28 °C. The identity of the culturable bacteria was identified using MALDI-TOF MS (Bruker, Germany) equipped with Bruker FlexControl software version 3.3 and Bruker MALDI Biotyper Real Time Classification (RTC) version 3.1. This bacterial identification was performed according to the direct transfer procedure from Bruker 45 . Protein from the bacterial cell was measured by MALDI-TOF MS and the spectra generated were compared to the reference database for bacterial identification. Results with the log (score) value equal or higher than 2 indicate high confidence identification and is tabulated in Table 1.
The bacterial identities were confirmed molecularly by phylogenetic analysis on their 16S rDNA genes sequences. Bacterial genomic DNA were extracted using QIAamp DNA mini kit according to manufacturer instruction. 16S rDNA genes were amplified with 27F and 1525R primer pair 46 . Molecular identities of the isolates were identified based on the phylogenetic analysis of the 16S rDNA gene sequences using MEGA (v6.06) 47 .
Biocatalytic Assays. Amylolytic activity of the 18 bacterial isolates was screened using diluted LB agar supplied with 0.5% (w/v) soluble starch. The utilization of starch by the bacteria was observed by flooding the starch agar with iodine. Iodine reacts with starch giving a dark blue substrate. Amylolytic bacteria degrade starch and release iodine from iodine/starch complex into the medium which results yellow halo around bacteria colonies on dark blue background.
The bacterial proteolytic activity was screened using diluted LB agar supplied with 5% (v/v) skim milk. Skim milk contributed to the opaque properties on the agar. Clear-zone around the bacterial colony will be observed on the skim milk agar for bacteria that possess proteolytic activity.
Cellulolytic activity was tested with basal salt medium supplied with 0.15% (w/v) microcrystalline cellulose, adjusted to pH 7.0. This agar was stained red with 0.02% (w/v) Congo red. The cellulose degradation by cellulolytic bacteria was identified by the Congo red discolorization which results in the formation of light yellow zone or clear zone around the colony on medium 9 .
Xylanolytic activity was screened using EnzChek ® ultra xylanase assay kit (Invitrogen, USA). The xylanase substrate (synthetic hemicellulose polysaccharides) was tagged with fluorescence dye and supplied by the kit. The structure of the xylanase is proprietary product to the manufacture. Xylanase by Trichoderma viride (Sigma, US) was used as the positive control for the xylanolytic assay. Bacterial samples were prepared by resuspending the cell pellet from 1 ml of bacteria culture into 100 μ l of sterile culture medium. Working buffer (100 μ l) was added to the resuspended cell before loading the bacteria cells (50 μ l) into the 96 wells black plate. The assay was performed according to the protocol provided by manufacture at 25 °C. The presence of xylanase was detected by measuring the fluorescence level and compared to the standard curve. Fluorophore was released from the xylanase substrate when xylanase breaks the xylosidic lingkage of the xylanase. The fluorescence was measured at excitation/emission 360/465 using Infinite ® F200 PRO (Tecan, Switzerland).
Chitinase, in particular, β -N-acetylglucosaminidase, chitobiosidase and endochitinase activities of the bacterial isolates were tested using chitinase assay kit (Sigma, US) as described by manufacture 48 . Substrate solution was prepared by dissolving 1 mg of appropriate substrate (4-nitrophenyl N-acetyl-β -D-glucosaminide, 4-nitrophenyl N,N' diacetyl-β -D-chitobioside or 4-nitrophenyl β -D-N,N' ,N"-triacetylchitotriose) in 1 ml of assay buffer. Chitinase control enzyme was diluted to 10 μ g/ml with phosphate buffered saline before used. Diluted chitinase control enzyme (10 μ l) was added to 90 μ l of substrate solution and this reaction served as positive control in the assay. Blank reaction was prepared by loading 100 μ l of substrate solution to a well in microtitre plate. Each reaction tested by adding 10 μ l of overnight bacteria culture to 90 μ l of substrate solution. The reaction was incubated for 24 hours followed by the addition of 200 μ l of stop solution. The assay was carried out at 28 °C for bacterial culture and 37 °C for E. coli transformant carrying chitinase gene. β -N-acetylglucosaminidase, chitobiosidase and endochitinase producing bacteria hydrolyzed the chitinase substrate and released p-nitrophenol. The p-nitrophenol released was ionized by the stop solution (sodium carbonate) into p-nitrophenylate ion which was yellow color. The yellow tint measurement was taken at the absorption 405 nm using Infinite ® M200 (Tecan, Switzerland) and compared with the reading from chitinase control enzyme.
Bacteria Whole Genome Sequencing. Genomic DNA of the bacterial isolates was extracted using QIAamp DNA mini kit (Qiagen, Germany) as described by the manufacturer. The quality and quantity of the DNA was checked with Nanodrop 2000c (Thermo Scientific, USA) and Qubit 2.0 fluorometer (Invitrogen, USA), respectively. Sequencing libraries were prepared with Nextera sample prep kit (Illumina, USA) and the indexing was performed according to Nextera Index kit (Illumina, USA). Prior to sequencing, the sequencing library was verified with Bioanalyzer DNA high sensitivity chip (Agilent, USA). Subsequently, 12 pM of the pooled sequencing library was loaded into MiSeq sequencing cartridge (2 × 250 bp, 500 cycles) (Illumina, USA). The sequencing was performed on MiSeq (Illumina, USA) sequencing platform.

Bacteria Draft Genome Analysis and Gene
Prediction. The quality of the sequencing raw data was checked with FastQC 49 . Subsequently, the paired-end reads were trimmed and de novo assembled with CLC genomic workbench version 6 (CLC, Denmark) 21 . The coding DNA sequence (CDS) was predicted using prodigal (v2.60) 50 . Draft genomes were annotated by BLASTX against NCBI non-redundant (nr) database 51 . The draft genomes were deposited on DDBJ/EMBL/GenBank. Chitinase Gene Analysis and Cloning. The putative chitinase genes sequences with e-value of 0.00, indicating that the sequence alignment was not occurred by chance, were preceded for further analysis 52 . The low e-value indicates that the match of the query gene against the database was significant. The chitinase genes were group by phylogenetic analysis using MEGA (v6.06) by its amino acid sequences 47 . These genes were synthesized and inserted into pET− 23(+ ) expression vector by service provider, Genscript. Subsequently, the vector was transformed into Escherichia coli BL21 and preceded to chitinolytic assays.