Fusobacterium Genomics Using MinION and Illumina Sequencing Enables Genome Completion and Correction

Bacterial growth and genomic DNA preparation.All strains of Fusobacterium were grown overnight in CBHK (Columbia broth, hemin [5 µg/ml], menadione [0.5 µg/ml]) at 37°C in an anaerobic chamber (90% N₂, 5% CO₂, 5% H₂). Genomic DNA from stationary-phase bacteria was isolated in deionized water (diH₂O) from each strain using a Wizard isolation kit (Promega) and was quantitated using a Qubit fluorimeter (Life Technologies, Inc.).

Short-read Illumina sequencing.Short-read DNA sequencing was carried out at the Genomic Sequence Center at the Virginia Tech Biocomplexity Institute and Novogene (strain F. nucleatum subsp. nucleatum ATCC 25586). For sequencing at Virginia Tech, DNA sequencing (DNA-seq) libraries were constructed using a PrepX ILM 32i DNA library reagent kit on an Apollo 324 NGS library preparation system. Briefly, 150 ng of genomic DNA was fragmented to 400 bp using a Covaris M220 focused ultrasonicator. The ends were repaired, and an “A” base was added to the 3′ end for ligation to the adapters, which have a single “T” base overhang at their 3′ end. Following ligation, the libraries were amplified by 7 cycles of PCR and barcoded. The library generated was validated by the use of an Agilent TapeStation and quantitated using a Quant-iT dsDNA HS kit (Invitrogen) and quantitative PCR (qPCR). The libraries were then pooled and sequenced using a NextSeq 500/550 Mid Output kit V2 (300 cycles) (P/N FC-404-2003) to 2 × 150 cycles. BCL files were generated using Illumina NextSeq control software v2.1.0.32 with real-time Analysis RTA v2.4.11.0. BCL files were converted to FASTQ files, and adapters were trimmed and demultiplexed using bcl2fastq Conversion Software v2.20. Illumina sequencing statistics and genome coverage are detailed in Table S1 in the supplemental material, and the public availability of the data at NCBI is detailed in Table 1.

Long-read MinION sequencing.Purified Fusobacterium genomic DNA was sequenced on a MinION sequencing device (Oxford Nanopore Technologies) using one-dimensional (1D) genomic DNA sequencing kit SQK-LSK108 according to Oxford Nanopore Technologies instructions. Multiplexed samples were barcoded using a 1D native barcoding kit (EXP-NBD103) according to instructions. Briefly, purified genomic DNA was repaired with NEBNext FFPE repair mix (New England Biolabs). A NEBNext Ultra II End-Repair/dA-tailing module was utilized to phosphorylate 5′ ends and add dAMP to the 3′ ends of the repaired DNA. For multiplexed samples, barcodes were ligated to the end-prepped DNA using NEB Blunt/TA master mix (New England Biolabs). Barcoded samples were pooled into a single reaction mixture, and an adapter (Oxford Nanopore Technologies) was ligated to the DNA using NEBNext Quick T4 DNA ligase (New England Biolabs). For single reactions, an adapter (Oxford Nanopore Technologies) was ligated to the end-prepped DNA using NEB Blunt/TA master mix (New England Biolabs). The DNA was purified with AMPureXP beads (Beckman Coulter, Inc., Danvers, MA) following each enzymatic reaction. Purified, adapted DNA was sequenced on an MK1B (MIN-101B) MinION platform with a FLO-min 106 (SpotON) R9.4 or FLO-min 107 (SpotON) 9.5 flow cell using MinKNOW software version 1.7.10 or 1.7.14 (Oxford Nanopore Technologies). After sequencing, Fast5 files were base-called using Albacore version 2.1.7 (Oxford Nanopore) on a Macbook Pro with a 3.3 GHz Intel Core i7 processor. For multiplexed samples, base-called fastq files were demultiplexed based on the ligated barcode using Porechop (https://github.com/rrwick/Porechop) and adaptors were trimmed. Sample preparation and sequencing details are presented in Table S2, and MinION sequencing statistics and genome coverage are detailed in Table S3. As an example of data quality, Fig. 2 shows the long-read coverage obtained using MinION sequences for the F. necrophorum funduliforme 1_1_36S genome.

Genome assembly.Genome assemblies were carried out using Unicycler version 0.4.3 open-source software (10), resulting in complete, single chromosomes for each of the eight sequenced genomes. While both the Illumina and MinION sequencing runs produced far more data than necessary, data sets were split to utilize ample and yet reasonable mean depth of coverage for 1.6-Mb to 3.5-Mb genomes. Prior to assembly, data were not sorted based on base call quality as judged by Phred scoring, as we show in Fig. 1 that the data are of high quality. Using the mean depth of coverage for each genome described in Tables S1 and S2, each genome can be constructed in 2 to 3 h using a standard Macbook Pro laptop (2.8 GHz Intel Core i7). The utility of Unicycler therefore signifies that it is a robust method for researchers without the need for a supercomputer to handle data processing. The details of all final assemblies are shown in Fig. 3, and the public availability of the data at NCBI is detailed in Table 1. For consistent starts to the circular chromosome, each genome was rotated to have gene 1, which encodes the rod-shape-determining protein MreC, in the reverse orientation as seen for the beginning of the F. nucleatum subsp. nucleatum ATCC 25586 reference genome (8).

Open reading frame predictions.Gene predictions for protein-encoding open reading frames were carried out using the bacterium-specific program Prodigal version 2.6.3 via command line on a Mac (15). Genes for tRNA encoding were predicted with Prokka (16) using the KBase server (17). rRNAs were identified using Barrnap (bacterial rRNA predictor) version 0.8 (18). In addition, we used the CRISPRone Web server (19) to identify all CRISPR-associated proteins and arrays, which consist of spacer and repeat regions. Details of each of these components are found on the FusoPortal repository. For each genome, the protein-encoding gene predictions by Prodigal and Prokka were in nearly complete agreement (data not reported). In addition, genome annotation for each genome was performed by NCBI upon data deposition into GenBank (Table 1).

Software and code availability.All software and scripts used in this study have been described and properly referenced in previous Materials and Methods sections.

Technical validation of sequencing reads and whole genomes.Phred quality scores for Illumina sequencing reads were determined using Geneious version 9.1.4, and these data are shown for the F. nucleatum subsp. nucleatum ATCC 25586 genome in Fig. 1A. In addition, MinION read quality was assessed using the software package Pauvre as depicted in Fig. 1B and C. Data for all eight genomes as seen in Fig. 1A can be found on the Fusoportal website (http://fusoportal.org/phred.html).

CheckM (20) on the Kbase server was used to check the quality of each genome using the reduced tree data set setting. Analysis for all genomes can be found on the Fusoportal website (http://fusoportal.org/checkm.html).

Accession number(s).Raw data and completed genomes for each of the eight Fusobacterium strains have been deposited at NCBI under the BioProject, BioSamples, sequence read archives (SRA), and GenBank accession numbers detailed in Table 1.

Data availability.The raw data, genome assemblies, and annotations can be accessed via the NCBI BioProject under accession PRJNA433545, and further details of these files can be found in Table 1. In addition, all of these data are easily accessible in our newly implemented FusoPortal data repository or on our Open Science Framework database (http://osf.io/2c8pv).