Unexpected Relations of Historical Anthrax Strain

ABSTRACT In 1998, it was claimed that an 80-year-old glass tube intentionally filled with Bacillus anthracis and embedded in a sugar lump as a WWI biological weapon still contained viable spores. Today, genome sequencing of three colonies isolated in 1998 and subjected to phylogenetic analysis surprisingly identified a well-known B. anthracis reference strain isolated in the United States in 1981, pointing to accidental laboratory contamination.

there were 5,474,730 high-quality nucleotide positions available for genotypic comparison.
The initial molecular subtyping by canonical SNP typing (6,7) assigned the strains to the A.Br.Ames clade. Multilocus variable-number tandem-repeat analysis (MLVA) with two different established systems (6,8) revealed the same allelic patterns as for the Ames ancestor strain. Subsequent in silico analysis confirmed the laboratory results. Further in silico investigations by whole-genome SNP typing revealed only two nonsynonymous SNPs, at positions 1798709 (A ¡ G) and 4212867 (G ¡ A) compared to the Ames ancestor. No difference was detected in the virulence plasmids, whose complete sequences were also determined. This high identity was unexpected, as the Ames ancestor strain was isolated from a cow in Sarita, TX, in 1981, 64 years after the historical sample's discovery (9). Extensive research after the 2001 Amerithrax letter attacks showed that members of the A.Br.Ames clade were naturally uncommon (10). Phylogeographic analysis was consistent with the derived Ames clade's historical importation into North America, perhaps from China, where many close relatives are found (11).
In particular, the Ames ancestor and its identical sister genome, FTD1004, are very high quality closed and finished sequences that represent two stocks derived directly from the original 1981 Texas B. anthracis isolate. These two stocks were independently established in May 1981 after material was transferred to USAMRIID. Together, they represent the oldest known laboratory stocks of the original Ames strain, established within months of the bovine anthrax case.
Directly diverging from these five genomes ( Fig. 1) is that of strain Porton Down Ames, with 65 unique SNPs. Careful examination of the data argues that these are not sequencing errors but rather represent mutations that occurred during laboratory growth. A similar divergence from other Ames isolates was reported by Read et al. and attributed to mutations that occurred during a plasmid-curing regimen (12,13).
We can only speculate about the direct linkage between the original Texas 1981 Ames strains and the Porton Down sugar isolates of 1998. The Porton Down scientists were the world leaders in anthrax research, and their plasmid-cured Ames strain is still in use in reference laboratories worldwide. Therefore, they were well aware of the value of this historical sample and fumigated the class 3 microbiological safety cabinet with formaldehyde prior to opening the vials to avoid contamination (4). These careful protocols were invoked to cultivate these historical specimens, which would lead to the oldest B. anthracis strains ever described, as previously the oldest samples were from 1954 (14).
It is clear from their study description that the isolation process was more difficult than the standard microbiological methods employed for B. anthracis (4). Because no colonies were apparent from culturing of the original liquid, an 8-day enrichment in liquid broth was used. Plating of the enriched liquid resulted in only four colonies, which were confirmed to be B. anthracis by McFadyean's test and species-specific PCR assays. However, none of these techniques was able to differentiate among B. anthracis strains. The first suitable molecular typing techniques (MLVA) using eight markers was published in 2000 (15), 2 years later, and hence, molecular typing of the isolates was indeed not possible at the time of the report.
Despite secure handling and microbiological safety cabinet cleaning, it seems highly likely that the isolated strains are accidentally isolated contaminants of formerly processed Ames strains. In 1981, the progenitor of the Ames ancestor and FTD1004 was isolated in Texas and then transferred to USAMRIID at Ft. Detrick, MD, United States. In 1982, an Ames culture was transferred from Ft. Detrick to Porton Down (12). Only one phylogenetically informative SNP separates the Porton Down genomes from the USAMRIID Ames genomes (Fig. 1). This high identity and phylogenetic topology argue that the Porton Down Ames strain and Porton Down sugar contaminants were derived from the Ames material sent in 1982.
In the context of spatial separation of analytical and research facilities to avoid contamination, there were no dedicated analytical facilities at the Porton Down defense laboratories in 1998. Therefore, analysis of the sample had to be conducted in a facility used for research activities. This changed in 2006, when the Defence Science and Technology Laboratory implemented a purpose-built reception and screening facility. Culturing of samples is now carried out in a separate facility to mitigate against the possibility of laboratory contamination.
When genome diversity is so low that only a few SNPs differentiate critical isolates, high-quality NGS and careful phylogenetic analyses are needed for unambiguous strain identification. Only this ultradiscriminatory power has unraveled the mystery of the historical B. anthracis sugar lumps, revealing that the spores really did not survive for this extended period. Today, direct metagenomic analysis of the capillary fluid might be capable of generating the whole genome sequence, even if there were no viable spores (16). The investigation of historical pathogen specimens has clearly entered a new era.
Materials and methods. All strains (sugar 2, sugar 3, and sugar 4) were successfully grown on sheep blood agar out of deep-frozen cryobank stocks in a biosafety level 3 (BSL3) area. Strain sugar 1 failed to grow. After incubation overnight at 37°C and visual inspection for contamination, single colonies were picked and diluted. Subsequently, the suspension was filtered with a Millipore 0.1-m filter to obtain spore-free liquids and its nucleic acid was purified with the Qiagen DNA Blood and Tissue kit (Qiagen,  Table S1 in the supplemental material. Anthrax Sugar Cubes ® Hilden, Germany). In accordance with standard procedures, 10% of the volume was plated again onto sheep blood agar and incubated for 14 days without any growth characteristics, prior to transfer of the DNA out of the BSL3 laboratory. Final DNA concentrations were measured with the Qubit 2.0 fluorometric assay (Life Technologies, Inc., Darmstadt, Germany) and analyzed with an Agilent Bioanalyzer.
MLVA was performed with an ABI 3130 Capillary Sequencer as described previously (6,8,17). For in silico MLVA, in-house python scripts were used to determine the corresponding fragment lengths by using the published primer sets.
NGS was performed by the Illumina MiSeq approach. In accordance with the manufacturer's protocol, sequencing libraries were prepared by using 2.0 ng of genomic DNA with the Nextera XT kit (Illumina, San Diego, CA). Genomes were sequenced on an Illumina MiSeq platform with paired-end v 3 chemistry (2 ϫ 300 bp). De novo genome assembly was performed with SPAdes (18) version 3.5.0 and polished by Pilon v.1.3.0 (19).
For SNP identification, the Northern Arizona SNP pipeline was used (http:// tgennorth.github.io/NASP/) (16). This includes the alignment of raw data against the Ames ancestor (NC_007530, NC_007322, and NC_007323) with the BWA-MEM algorithm (20) and SNP calling by the UnifiedGenotyper method in GATK (21,22). To calculate the depth of coverage, raw data were aligned with the Ames ancestor and the per-base depth of coverage was calculated by the GenomeCoverageBed method in BEDTools (23).

ACKNOWLEDGMENT
This work was supported by a grant from the U.S. Department of Homeland Security, Science and Technology Directorate.