Bacillus paranthracis Isolate from Blood of Fatal Ebola Virus Disease Case

Abstract

A Bacillus paranthracis isolate was cultured from the blood of a fatal Ebola virus disease (EVD) case in Liberia and was identified by whole genome sequencing. Although B. paranthracis has only recently been described and is poorly characterized, this case may represent the bacterial co-infection of an EVD patient.

Bacteremia, possibly via gut translocation, might complicate Ebola virus disease (EVD) [1,2,3], and current supportive care guidelines recommend the empiric administration of broad-spectrum antibiotics to all EVD patients [4,5]. However, supporting data are limited as microbiological investigations are not routinely performed in Ebola treatment units (ETUs) during outbreaks. A single case of Gram-negative sepsis was reported in an EVD patient treated in Europe, although the organism was not fully identified, and iatrogenic infection could not be excluded [1]. In one small cohort of 18 EVD patients treated at a military ETU in Sierra Leone, blood cultures were obtained and yielded only one isolate of a coagulase–negative Staphyloccoci skin contaminant [6]. Other, indirect evidence of bacteremia in EVD patients hinges on clinical observation [2] and retrospective unbiased deep sequencing [3].

Here we report on the microbiological and molecular characterization of a bacterial isolate obtained from the retrospective analysis of a stored whole blood sample obtained from a female patient of unknown age admitted to the ELWA-3 ETU in Monrovia, Liberia on 2 November 2014 [7]. The patient tested positive for Ebola virus by quantitative reverse-transcription polymerase chain reaction (cycle threshold = 23.58) four days following symptom onset. Malaria diagnosis was negative. The patient died two days after admission with no further samples collected. This study was designated “not human subjects research” by the NIH Office of Human Subjects Research Protection (OHSRP) and was approved by the University of Liberia-Pacific Institute for Research and Evaluation (UL-PIRE) institutional review board.

Upon triage, a whole blood specimen was collected in an EDTA tube and frozen at −80 °C for long-term storage at biosafety level 4 (BSL-4). The gold-standard culture-based approach for the identification of bacteremia was utilized with a lowered input volume due to limited sample availability. A 200 μL aliquot of the whole blood was inoculated into a BD BACTEC Peds Plus /F culture vial with BD fastidious organism supplement and incubated in a BD BACTEC FX40 (Becton Dickinson, https://www.bd.com/). Logarithmic growth by fluorescence was detected 11.5 h after inoculation. The sample was sub-cultured on non-selective sheep blood agar (SBA) and chocolate agar at 35 °C without supplemental CO₂ and produced heavy monomicrobial growth of dull, gray-white, opaque colonies at 24 h, with weak β-hemolysis evident directly beneath the colonies at 48 h. The isolate was designated RML14492_ELWA-3_3298 (ELWA 3298).

For identification, the colonies were removed from BSL-4 after inactivation in TRIzol [8] (Thermo Fisher, https://www.thermofisher.com) according to standard operating procedures approved by the Institutional Biosafety Committee, as standard biochemical testing for identification was not readily available in BSL-4. Analysis of nucleic acids by 16S rRNA sequencing and purified protein extract by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI–TOF MS) [9] provided only confident genus-level identification within the Bacillus cereus sensu lato group [10]. The lack of strong β-hemolysis on SBA (

Table 1. Comparison with other Bacillus spp.

Bacillus Species/Strain	Hemolysis	Motility	Penicillin	Catalase	PEA § Growth
ELWA 3298, B. paranthracis	weak β *	+ †	R ‡	+	+
B. cereus sensu stricto	strong β	+	R	+	+
B. cereus biovar anthracis	γ	+/−	S/R	+	unknown
B. anthracis	γ	-	S	+	-

Table adapted from American Society for Microbiology’s “Sentinel Level Clinical Laboratory Guidelines for Suspected Agents of Bioterrorism and Emerging Infectious Diseases”, 2017. * β-hemolysis on sheep-blood agar was most evident at 48 h and directly beneath colonies. ^† Motility testing was performed with 2,3,5-triphenyltetrazolium chloride (TTC) motility agar using Escherichia coli DH10B as a positive control and Klebsiella pneumoniae (ATCC 13883) as a negative control. ^‡ Antimicrobial susceptibility testing was performed with Sensititre broth microdilution (Thermo Fisher, https://www.thermofisher.com) and interpreted according to CLSI M45 MIC criteria; the isolate was also resistant to oxacillin, ampicillin, and cephalosporins.^§ Phenylethyl alcohol agar.

) and the geographical proximity of the patient both to human infections of B. anthracis and zoonotic infections of B. cereus biovar anthracis [11] prompted full-genome sequencing for species-level identification, as conventional microbiological testing and other diagnostic assays were not immediately available in BSL-4. A TruSeq DNA Nano kit was used for library preparation and 250 bp paired-end sequencing was performed on a MiSeq sequencer (Illumina, https://www.illumina.com). Raw reads were trimmed, filtered, and coordinate-order sorted using Cutadapt v1.12, FASTX Toolkit v0.0.14, and a custom Perl script. Mira v4.0.2, Velvet v1.2.10, Sequencher (Genecode 5.4.6), and Pilon v1.22 were used for assembly and final polishing. The draft genome de novo assembly resulted in 71 contigs with a total length of 5,681,902 bp, a N50 value of 182,992 bp, and a GC content of 35.2%. This whole genome shotgun project has been deposited at DDBJ/ENA/GenBank under the BioSample identifier SAMN13240933. Following annotation with the NCBI Prokaryotic Genome Annotation Pipeline and average nucleotide identity analysis, the isolate was identified as B. paranthracis as it demonstrated a 97.6% identity with 83% coverage to the type strain of B. paranthracis, Mn5T. B. paranthracis is poorly characterized and was first described in 2017 [12], although current reports describe its isolation from human feces [13] and it may have been responsible for a small emetic and diarrheal outbreak [14]. A limited range of phenotypic testing was performed on the isolate following identification for further characterization (Table 1).

BTyper predicted the ELWA 3298 sequence type (ST) as 1074, based on seven housekeeping genes (glp, gmk, ilv, pta, pur, pyc, and tpi) commonly used for Bacillus cereus sensu lato multilocus sequence typing (MSLT) [15]. The other currently described isolates of B. paranthracis have STs of 26 and 761. MLST phylogenetic analysis, utilizing concatenated sequences of the seven housekeeping genes, was performed. The sequences were aligned with MUSCLE and a maximum-likelihood tree with 1000 bootstraps was generated with PhyML using a general time reversible nucleotide substitution model. The ELWA 3298 isolate fell into clade 1 of the B. cereus sensu lato group [15] and a lineage was formed with the other B. paranthracis STs, as shown in Figure 1. A MLST minimum spanning tree containing the STs for the known B. paranthracis isolates placed ELWA 3298 (ST-1074) closest to the ST-26 B. paranthracis isolate, as they share four of seven alleles, as shown in Figure 2. The Mn5T B. paranthracis type strain isolate, which has an ST of 761, is more distantly related and is part of the ST-205 clonal complex. Isolates across these STs have been cultured from numerous sources and have been the etiological agents of various human diseases, including sepsis [16].

This is the first report, to our knowledge, that provides species-level identification and phenotypic characterization of a bacterium isolated from the blood of an EVD patient. However, although iatrogenic infection can be ruled out in this patient, as the sample was collected at triage, non-anthracis Bacillus spp. are typically regarded as environmental contaminants when isolated from blood [17]. Nevertheless, invasive infections, including sepsis, are increasingly attributed to some non-anthracis Bacillus spp., particularly in immunocompromised patients [17]. Given the ability of Bacillus spp. to readily colonize the gut, both as transient flora and pathogenically [17], a gastrointestinal source in this patient is plausible if true bacteremia was present [1,2]. With the present limitations, however, caution must be taken with such an interpretation and a determination of the clinical significance of this isolate cannot be definitively made. This finding underpins the need for the continued study of possible bacterial co-infections with EVD.