Y./Water 165()47 Limited.China).Paired-end library was prepared with TruSegTM the carbohydrate-active enzymes database(CAZy.accessed on July DNA Sample P 2016).the database (acc ssed on December 2016)and ments.P on August Kit and HiSeq 30004000 SBS Kits. ing Ima Generator (BRIG)(Alikhan et al 2011)and Easyfig ewcnpmdg 0n2.0(Ko9 the umin The o6黑新8 dintohigh-qualitydraft Wagner et a 201 )The relative ssion and pathway able S n were cale 2.3.Metagenomic assembly and binning the 2.6.LC-MS-based metabolomic profing and quntitation analysi s.The pr rief. ed by centntug sing a sonicat hen.preoit The su ich are specific within a phy etic lin 2.4.Phylogenetic analysis of recovered drafg me 3.Results ed to 3.1.Higher nirogen removal rate with ion version oinoo nd or groups wer 25.Metagenomic and 282NNi57 NOT-N n rat The ANO meopeneidn ugh P cantly lower 3.7%0 rtia with a rate of 245.9+12.4 ntilthis point ((d) Limited, China). Paired-end library was prepared with TruSeqTM DNA Sample Prep Kit (Illumina, San Diego, CA, USA). Adapters containing the full complement of sequencing primer hybridization sites were ligated to the blunt-end fragments. Paired-end sequencing was accomplished on Illumina HiSeq 4000 platform (Illumina Inc., San Diego, CA, USA) at Majorbio Bio-Pharm Tech￾nology Co., Ltd. (Shanghai, China) with HiSeq 3000/4000 PE Cluster Kit and HiSeq 3000/4000 SBS Kits. Total RNAs of consortia samples was subjected to an rRNA removal procedure with the Ribo-zero Magnetic kit according to the manufacturer's instruction (Epicentre, an Illumina® company). cDNA libraries were constructed with TruSeq™ RNA sample prep kit (Illumina). The barcoded libraries were paired end sequenced on the Illumina Hiseq 3000 platform at Majorbio Bio-Pharm Tech￾nology Co., Ltd. (Shanghai, China) with HiSeq 4000 PE Cluster Kit and HiSeq 4000 SBS Kits according to the manufacturer's in￾structions (www.illumina.com). Finally, raw metagenomics and metatranscriptomics datasets were deposited in the NCBI Sequence Read Archive database, and accession numbers were listed in Table S1. 2.3. Metagenomic assembly and binning First, raw metagenomic reads were trimmed by stripping adaptor sequences and ambiguous nucleotides with SeqPrep version 1.1 based on default parameters (Prensner et al., 2011). The trimmed sequences were quality filtered with Sickle version 1.33 based on a minimum quality score of 20 and a minimum sequence length of 50 bp (Wu et al., 2016). Next, contigs and scaffolds were assembled individually for each sample via IDBA-UD with default parameters (Peng et al., 2012). Generated scaffolds were binned into draft genomes based on abundance and the tetranucleotide frequency using the MetaBAT version 0.32.5 with the sensitive model and a minimum contig size of 1500 bp (Kang et al., 2015). CheckM version 1.0.7 was employed to assess the completeness and the quality of recovered draft genomes through single-copy marker genes, which are specific within a phylogenetic lineage (Parks et al., 2015). Draft genomes were deposited in GenBank, and accession numbers were listed in Table S2. 2.4. Phylogenetic analysis of recovered draft genomes Phylosift version 1.0.1 was used to construct the phylogenetic tree (Darling et al., 2014). Marker genes of 37 reference genomes were selected. The concatenated protein alignments of the recov￾ered draft genomes and reference genomes were aligned with MAFFT version 7.310 (Katoh et al., 2002) and concatenated again, using homemade scripts. Finally, a maximum likelihood phyloge￾netic tree was conducted using the RAxML version 8.2.11 (Stamatakis et al., 2008). Molecular Evolutionary Genetics Analysis (MEGA) software version 7.0 were used to visualize the phyloge￾netic tree (Kumar et al., 2016). 2.5. Metagenomic and metatranscriptomic analysis First, raw metagenome reads of each sample were mapped to all contigs which were assembled into high-quality draft genomes (completeness  70% and contamination  10%) with the bbmap version 37.75 (minid ¼ 0.95 and ambig ¼ random). Then, the coverage of each bin was calculated by adding the reads which were mapped to the contigs of each bin and normalized by genome size. The open reading frames (ORFs) of each recovered draft genome were annotated through Prodigal version 2.6.3 with the ‘meta’ option for metagenomes, where the minimum nucleotide length was set as 60 (Hyatt et al., 2010). ORFs were then queried against the carbohydrate-active enzymes database (CAZy, accessed on July 2016), the eggNOG database (accessed on December 2016) and Kyoto Encyclopedia of Genes and Genomes pathway database (KEGG, accessed on August 2017) with DIAMOND software, where the blast e value was set as 1e-5 (Buchfink et al., 2014). OrthoANI software was employed to calculate average nucleotide identity (Lee et al., 2016). Genome comparison was visualized via BLAST Ring Image Generator (BRIG) (Alikhan et al., 2011) and Easyfig software (Sullivan et al., 2011). The raw metatranscriptomic reads were trimmed and quality filtered with the SeqPrep and Sickle software respectively. rRNA reads were removed through SortMeRNA version 2.0 (Kopylova et al., 2012) aligning to the SILVA 128 version database (accessed on February 2017). Non-rRNA reads were mapped to all contigs which were assembled into high-quality draft genomes with bbmap software. Read counts of each gene were calculated using htseq-count v0.9.1 with the ‘intersection strict’ parameter (Anders et al., 2015) and normalized as transcripts per million (TPM) values (Wagner et al., 2012). The relative gene expression and pathway expression were calculated by relativizing the median TPM across the draft genome, and by the median TPM value of each reaction in the pathway, respectively. 2.6. LC-MS-based metabolomic profiling and quantitation analysis Metabolic products were extracted from the anammox consor￾tia samples in autotrophic and mixotrophic groups. The protocols used to extract metabolic products were conducted according to those described previously (Guo et al., 2017; Tang et al., 2018b). In brief, anammox consortia suspension was collected by centrifuga￾tion, washed using Phosphate buffer saline (PBS), and sonicated using a sonicator. Then, previously cooled methanol was added to the mixture, and the precipitated protein was removed by centri￾fugation. The supernatants were dried in a nitrogen gas stream. The residues were used to determine metabolite contents via LC-MS￾based metabolomic analysis. All tests were conducted in quadruplicate. 3. Results 3.1. Higher nitrogen removal rate with acetate addition In order to investigate the effect of acetate on the nitrogen removal rate of anammox consortia, batch tests were conducted. Based on nitrogen and chemical oxygen demand (COD) consump￾tion (Fig. 1), 155 min was selected as the sludge sample collection point, for NO3
-N concentrations in mixotrophic groups were significantly (p < 0.05 by t-test) and 40% lower than autotrophic groups at this point, which enabled the determination of discrepant expression genes and metabolic pathways of autotrophic and mixotrophic anammox consortia. At 155 min, average NH4 þ-N removal rates were 311.5 ± 15.7 and 284.0 ± 7.2 mg N/(Ld), and the average NO2
-N removal rates were 384.0 ± 41.9 and 359.2 ± 14.2 mg N/(Ld) in autotrophic group and mixotrophic group, respectively. Importantly, the average NO3
-N accumulation rate in the mixotrophic group was 31.0 ± 19.0 mg N/ (Ld), which was significantly lower than that in the autotrophic group (70.9 ± 8.0 mg N/(Ld)) (p < 0.05 by t-test). The DNO3 - -N/ DNH4 þ-N ratio in the mixotrophic group (0.11 ± 0.06) was signifi- cantly lower compared to that of the autotrophic group (0.23 ± 0.03) (p < 0.05). In mixotrophic group, a 59.0 ± 3.7% COD was degraded by anammox consortia with a rate of 245.9 ± 12.4 mg/(Ld) until this point (Fig. 1(d)). Y. Feng et al. / Water Research 165 (2019) 114974 3