Detection of Bacillus anthracis and Bacillus anthracis-like spores in soil from state of Rio de Janeiro, Brazil

BACKGROUND Bacillus anthracis is the aetiologic agent of anthrax, a re-emerging, septicaemic, haemorrhagic and lethal disease that affects humans, domestic ruminants and wildlife. Plasmids pXO1 and pXO2 are attributes that confer pathogenicity to B. anthracis strains. This bacterium was used as biological weapon in the World Wars and in the biological attack in the United States of America at 2001. B. anthracis is classified as a Tier 1 bioterrorism agent by the Centers for Diseases Control and Prevention. Anthrax is recognised as a re-emerging disease. Several studies concerning the dynamics of B. anthracis cycle in soil revealed that nonpathogenic B. anthracis strains due to lack of pXO2 plasmid are commonly found in some types of soil. OBJECTIVES This study aimed isolation and identification of B. anthracis spores in soil samples of the state of Rio de Janeiro, Brazil. METHODS Phenotypic and genotypic approaches were used to identify isolates including MALDI-TOF/MS, motility test, susceptibility to gamma phage and penicillin, survey for pag and cap genes as surrogates of pXO1 and pXO2 plasmids, respectively, and sequencing of 16SrRNA-encoding gene. Physicochemical analysis of the soil samples were carried out to describe soil characteristics. FINDINGS We observed the presence of one B. anthracis pXO1+ and pXO2- isolated from clay loam soil; one B. anthracis-like strain pXO1+ and pXO2-isolated from loamy sand; and 10 Bacillus spp. strains sensitive to phage-gamma that need better characterisation to define which their species were recovered from loamy sand. MAIN CONCLUSIONS This work showed promising results and it was the first study to report results from an active surveillance for B. anthracis in Brazil.

some soils are more likely to harbour spores than others. B. anthracis is most frequently found in clayey soils rich in organic matter and Ca 2+ , with pH above 6.0 and temperatures above 15.5°C. (6,7) According to Schild et al., (8) the disease occurs all over South America where Brazil share borders (total length of 16,885.7 km) with 10 out of 12 countries. From 2006 until July 2019, Argentina (borders with Rio Grande do Sul, Santa Catarina e Paraná states) reported 144 anthrax outbreaks, Uruguay (borders with Rio Grande do Sul state) 63 outbreaks, Paraguay (borders with Paraná e Mato Grosso do Sul states) 54 outbreaks, Peru (borders with Acre e Amazonas states)18 outbreaks, Bolivia (borders with Mato Grosso e Rondônia states) 38 outbreaks and Colombia (border with Amazonas state) 11 outbreaks. (9) Brazil's vast border extension can make it vulnerable to the clandestine entry of contaminated animals.
In Brazil, anthrax is on the list of diseases that requires animal health protection measures since 1934, requiring the sacrifice of affected animals and mandatory notification. According to the Ministry of Agriculture, Livestock and Supply, the last reported animal case occurred in 2016, in the state of Rio de Janeiro. (10) Regarding to human disease, nine cases of cutaneous anthrax were reported, between 1930-1932, all due to contact with contaminated animals. (11) Since 1928, anthrax in cattle and goats mostly occurs in three distinct areas of Brazil: south and west region of Rio Grande do Sul, where some places were called "cursed fields"; Paraiba river valley in São Paulo and Minas Gerais states; and in the northeast region. Among 72 outbreaks of sudden death in cattle that occurred between 2000 and 2014 in Rio Grande do Sul, seven were identified as caused by B. anthracis. (12) Ten outbreaks of anthrax were confirmed from January 1978 to March 2006 in Brazil occurred in cattle, most non-vaccinated, in the southeastern and southern region of Rio Grande do Sul, in municipalities on the Uruguay border. Lack of vaccination may have been an adjuvant for the occurrence of the disease after exposure to a primary source, such as soil from an old anthrax grave. (8) Several studies on the dynamics of B. anthracis cycle in soil have been carried out. (7,13) However, there is no data available about the occurrence of B. anthracis in soil in Brazil and very little information about Brazilian strains. The aim of this research was to survey B. anthracis in soil samples from Brazil, correlated or not with anthrax cases.

MATERIALS AND METHODS
Soil samples -Soil samples were collected at two different geographical sites in Rio de Janeiro state: in Barra de Guaratiba [three soil samples, two from Fazenda Modelo and one from Army Technology Center (CTEx)], sites not correlated with anthrax cases and in Barra do Piraí, location of the last case of anthrax reported in Brazil, in 2016, a cattle burrier site. Table I shows geographical coordinates of sampling.
Soil samples (20 g) were collected with sterile stainless-steel spatulas at 2 to 5 cm depth. Sampling areas with 50 m 2 were delimited and subsamples were col-lected at the angles (A, B, C and D) and at the central point (E), totalising 100 g in each collection site (Figure). These samples were stored at 4°C until processing.
Physicochemical analysis of soil samples -Physicochemical characterisation of soil samples was performed at the Analytical Centre Laboratory of the Federal Rural University of Rio de Janeiro (UFRRJ). The main features for B. anthracis spore persistence in soil, such as granulometry, Ca 2+ concentration, pH and organic matter content were analysed. Values suggested by the Brazilian Agricultural Research Corporation (EMBRAPA) and the Commission of Soil Fertility of Minas Gerais State (CFSEMG) were used for comparison. (14,15) The definition of soils textural class was based on the Triangle Diagram from the United States Department of Agriculture (USDA). (16) Soil processing -All samples were submitted to heat treatment at 70ºC for 15 min in water bath. Serial dilutions were performed in SATAMP up to 10 -9 . From each serial dilution, 100 µL were transferred and plated with Digralski loops in the culture media and incubated at 33°C for 24 h. Culture media used in this study were: nutrient agar (NA) (Difco™), NA with 0.25 M sodium acetate (NA+) (Difco™), Columbia agar base (Difco™) with 5% of sheep blood (CBA), CBA with 0.25M sodium acetate (CBA+), CBA with 0.5% sodium bicarbonate (CBA++) and PLET agar (Sigma-Aldrich -15 g/L agar, 30,000 U/L polymyxin b, 300,000 U/L lysozyme, 300 mg/L EDTA and 40 mg/L thallium acetate). Samples were homogenised with manual rotational motion for 30 min. Subsequently, 1 g of soil was transferred to 100 mL of sterile buffered saline (0,2 M potassium hydrogen phosphate -SATAMP -pH 7.2) and homogenised at 33°C for 30 min at 150 rpm (Shaker Innova 4080, New Brunswick Scientific Co., USA). Colonies presenting characteristic B. anthracis morphology were selected and submitted to microscopic examination using Gram staining. Gram-positive bacteria presented suggestive characteristics of B. anthracis ("box-shapped" cells with 1-1.5 μm width and 4-10 μm length and subterminal, cylindrical and non-deforming spores) were further characterised as described below.
Identification of isolates suspected of B. anthracis -Isolates presenting B anthracis characteristic colonies and cell morphology were further evaluated for gamma phage and penicillin susceptibility, motility and presence of pag and cap genes by specific PCR and MAL-DI-TOF/MS (Microflex LT mass spectrometer, Bruker, USA) as described below.
B. anthracis (CCGB 1861) was used as positive control and B. cereus (CCGB 0406 LFB-Fiocruz 406) as negative control. Both strains were provided by Collection of Genus Bacillus and Related Genera-CCGB-Fiocruz.
The strains received a numeral code for identification (Table II). The results of the 52 isolates are shown in Table II.
Motility test -To evaluate motility a 1 µL bacterial loop of each isolate was inoculated in 5 ml of nutrient broth with 0.5% of D-(+)-glucose and incubated in anaerobic conditions at 33°C for 20 h. Motility was observed on a fresh exam of broth culture by optical microscopy (x1000) (NiKon, Japan).
Susceptibility to gamma phage and penicillin -For the gamma phage susceptibility assay, 1 µL of 7.9 × 10 9 PFU/mL gamma phage suspension [provided by Public Health Agency (HPA), Porton Down, England] was used. For the penicillin susceptibility test, 10 U.I. Sensibiodisc (CECON, Brazil) were used. Samples grown in CBA were inoculated in nutrient broth and incubated at 33ºC for 24 h. Bacterial suspensions were inoculated into CBA plates using sterile swab. After the inoculation, penicillin disks and a 10-µL drop of gamma phage suspension were placed on the surface and plates were incubated at 37ºC for 24 h in aerobic conditions. Absence of bacterial growth as lysis plates and inhibition zones around the penicillin disk of any size were considered as positive results.
Microbial identification by MALDI-TOF/MS -Mass spectra were acquired using a Microflex LT mass spectrometer and results were analysed using Biotyper software version 3.1 (Bruker Daltonics, USA) containing a bioterrorism biological agent library, including 23 virulent B. anthracis strains and the Sterne strain.
Sample inactivation was performed according to Lasch et al. (17) Briefly, 1 µL of each sample was transferred to a 96-well stainless-steel plate and allowed to dry at room temperature. Then, 1 µL of α-cyano-4hydroxycinnamic acid matrix (HCCA) was added and allowed to dry at room temperature. MALDI-TOF MS analysis was performed in duplicate by linear mode with the following parameters: laser frequency 20 Hz; voltage of ion sources 1 and 2, 20 kV and 18.6 kV, respectively; molecular weight range 2,000 to 20,000 Da. Mass spectra were collected through the sum of 240 laser shots, acquired through four groups of 60 shots cast in randomly selected distinct regions. The spectra of the samples were calibrated internally, using Escherichia coli DH5α (IDQBRN, Brazil) ribosomal proteins as reference. Spectra were processed by Biotyper software in standard operating mode, which performs the research in its reference library in order to get the best match for the sample. Results were scored ranging from 0 to 3.0. Scores below 1.7 were considered unreliable, those between 1.7 and 2.0 were considered possible indicator of genus, those between 2.0 and 2.3 were considered reliable for genus and possible for species identification and those above 2.3 were considered reliable for identification at the species level.  (3,18) DNA extraction was carried out using Instagene matrix kit (Bio-Rad, USA) according to the manufacturer's instructions. PCR reactions were performed as previously described. (17,18) Briefly, multiplex PCR reactions for pag and cap were performed in 25 µL reaction volume containing 2.5 µL of template DNA, 2.5 µL of 10 × PCR Buffer, 0.5 µL of DNTP, 1,25 µL of a 10 µM solution of each primer, 14.38 µL of DNAse free water (Thermo Fisher Scientific, USA) and 0,0125 µL of Taq. For amplification, the following parameters were used: an initial denaturation step of 95°C for 5 min; 30 cycles of 95°C for 1 min; 55°C for 30 sec; 72°C for 30 sec; and a final extension step at 72°C for 5 min. (19,20) PCR products were observed after 1% agarose gel electrophoresis containing ethidium bromide under ultraviolet light.

Evaluation of pag and cap genes in isolates sus
PCR amplicon sequencing -Isolates positive for pag genes were submitted to simplex PCR, as described above, with the pag primers. PCR products were purified using Exosap-IT Express (Thermo Fisher Scientific, USA), according to the manufacture instructions. Amplicons were sequenced in an ABI3730XL automated sequencer (Applied Biosystems). Forward and reverse sequences were aligned and edited using Seqman program (DNAstar, Larsegene, version 7.0) and the data compared to the sequences deposited in GenBank with nBLAST.
16S rRNA gene sequencing -Isolates positive for pag genes and those susceptible to gamma phage were submitted to amplification and sequencing of the gene encoding the 16S rRNA, according to Watts et al., (21)   Products were purified using Exosap-IT Express (Thermo Fisher Scientific, USA), according to the manufacture instructions and sequenced at the DNA Capillary sequencing Facility -SANGER (Fiocruz, Brazil) in an ABI3730XL automated sequencer (Applied Biosystems). Forward and reverse sequences were aligned and edited using BioEdit 7.2 software and data were compared to the sequences in GenBank using nBLAST. The sequences were submitted at GenBank.

Physicochemical characteristics of soil samples -
The textural class of all samples was identified after particle size analysis. Samples BP1 and BP2 were classified as Loam: BP3 as Sandy Clay Loam, CA as Clay loam; CB and CT as Sandy Loam (Table III).
Regarding organic matter content, samples BP1, CB and CT presented high levels (> 3.0 dag/kg or > 30.0 g/dm3), while, BP2, BP3 and CA presented average levels (1.5-3.0 dag/kg or 15-30 g/dm3), according to EMBRAPA. (15) Isolation and identification of B. anthracis suspected bacteria -Bacterial growth was observed after incubation in aerobiosis. After observation of macroscopic characteristics, 369 colonies were Gram stained and 52 isolates were selected based on morphological and staining characteristics, as shown in Table IV.    (Table II and Table V).

DISCUSSION
In this study, one B. anthracis strain (harbouring pXO1, but not pXO2), one strain of B. cereus group harboured a pXO1-like plasmid (99.6% similarity with pag) and 10 Bacillus strains were isolated (Table V) from different soils of Brazil. These 10 isolates might be B. anthracis strains lacking virulence plasmids, which is in agreement with the findings of Kolton et al. (22) The probability of positive results for B. anthracis species cannot be ruled out, as there are reports of such occur- rence in the literature. (23,24) These isolates were recovered from six different culture media. From PLET cultures, a pXO1+ B. anthracis was isolated, from NA, a pXO1+ B. anthracis-like was isolated. These data corroborated what is described in literature, which suggest PLET is the most selective medium for B. anthracis isolation. (6) This study is the first on this subject performed in Brazil. It showed the importance of active surveillance in soil and the correct identification of the isolates for better understanding B. anthracis distribution in nature and for the elucidation of possible outbreaks in Brazil.
Several methods for isolation and identification B. anthracis in soil have been reported, although there is no consensus among the respective studies. A direct method was chosen, because it is a simple and economical protocol for survey B. anthracis in soil samples. Among the soils analysed in this study, the sample presenting characteristics that would favour the prevalence of B. anthracis spores was BP1, in according to Shadomy et al. (25) Interestingly, neither B. anthracis nor B. anthracislike strains were isolated from this sample. In the USA, the bacterium was isolated from soils with neutral to alkaline pH and high concentrations of nitrogen, while in Germany, there was no correlation with the type of soil. These differences indicate that the occurrence depends not only on the type of soil, but also on environmental conditions, the pathogen's life cycle, persistence, ecolo-gy and the ability to lose or acquire virulence attributes. (6,26) B. anthracis (harbouring pXO1, but not pXO2, strain 29) was isolated from the CA soil, a clay loam, with medium acidity, high Ca 2+ and medium organic matter content. B. anthracis was isolated in the former USSR from soils with medium acidity. (26) An isolate (strain 41) from CT soil cultured in NA presented microscopic characteristics of large Bacillus was resistant to penicillin and gamma phage, motile and beta haemolytic and would not be regarded as B. anthracis. However, this isolate harboured a pXO1-like plasmid (99.6% similarity with pag) and might be a B. anthracis-like bacterium. It was described that 7% of 1,000 B. cereus sensu lato isolated from environmental samples (soil, water, insects, plants) were shown to contain pXO1-like and/or pXO2like plasmids. B. mycoides harbouring pXO1-like and/or pXO2-like plasmids is less related than B. cereus or B. thuringiensis. It is relevant to note that B. anthracis-like strains have already been described causing disease with the same symptoms as anthrax in animals and humans, including being lethal. This suggested the occurrence of plasmid transfer events across the different species of the B. cereus group. (27) Species allocated in the B. cereus group have high genetic identity, which makes the differentiation complex, even by means of molecular tests. (28,29) In this group of bacteria, species classification is still based on pheno-typic and genotypic traits, pathogenicity, host preference and ecological niche. (23) Phenotypic tests are essential for B. anthracis identification. However, those considered to be classic phenotypic characteristics may not be present in all B. anthracis strains. Therefore, Kovac et al. (30) proposed phylogenetic classification based on the complete genome sequencing for bacteria belonging to the B. cereus group. The isolation of 10 strains presenting all the phenotypic characteristics of the specie, but not harbouring pXO1 neither pXO2 corroborated what were described in the literature.
Molecular assay in MALDI-TOF MS analysis, the only strain with a score above 2.3, which confirms species identification, was the one isolated from CA soil (strain 29). This isolate was confirmed as B. anthracis by other methods applied in this study. Among the isolates analysed, several were identified only at genus level, while others were identified as non-Bacillus, although the results of physiological tests had identified them as Bacillus spp.
Due to the inherent difficulty in differentiating between species with such a genomic similarity, it is necessary to enlarge the number of spectra in MALDI-TOF library in order to achieve better results for B. anthracis and B. anthracis-like bacteria identification. (30) Although PCR has the potential to infer the presence pXO1 and pXO2 plasmids, the method does not inform the species and the lineage unequivocally. Total genome sequencing has been shown to be the best methodology to study geographical distribution of strains during natural outbreaks or events suspected of bioterrorism. (22,24) 16sRNA sequencing showed that it is difficult to differentiate strains that belong to the B. cereus group. These results are due to the remarkable 99.6% identity percentage what makes them indistinguishable from each other when 16sRNA gene subunit sequencing was made. (23,24,29) These results showed the importance of whole genome sequencing for identification of bacteria belonging to the B. cereus group.
To date, the literature describes that virulent strains of B. anthracis necessarily harbour both plasmids. However, strains of B. cereus harbouring plasmid-like may cause fatal inhalation infection similar to inhalation anthrax, and cases in humans have already been described. Due to Brazil's continental dimensions, a large diversity of soils and climates can be found. The samples analysed represent only a fraction of this diversity. Nevertheless, it is important to highlight that this work showed promising results and it was the first study to report results from an active surveillance for B. anthracis in Brazil.
In conclusion, since livestock is one of the main economic activities in Brazil, large-scale studies involving B. anthracis active surveillance in soil in Brazil should be performed to reduce and/or prevent economic losses, and a complete characterisation of the isolates should be carried out in order to provide accurate information about Brazilian strains. In this work, we isolated B. anthracis (pXO1 + and pXO2-) and B. anthracis-like in soil samples, being the first description of this type in Brazil. This data increases the need to carry out surveillance of these species in the soil. The correct identification of these species is of paramount importance for the knowledge of the pathogen distribution in nature and elucidation of possible outbreaks.