A Pilot Study for the Characterization of Bacillus spp. and Analysis of Possible B. thuringiensis/Strongyloides stercoralis Correlation

Differentiating between Bacillus species is relevant in human medicine. Bacillus thuringiensis toxins might be effective against Strongyloides stercoralis, a nematode causing relevant human morbidity. Our first objective was to evaluate genomic and MALDI-TOF identification methods for B. thuringiensis. Our secondary objective was to evaluate a possible negative selection pressure of B. thuringiensis against S. stercoralis. PCR and Sanger were compared to MALDI-TOF on a collection of 44 B. cereus group strains. B. thuringiensis toxin genes were searched on 17 stool samples from S. stercoralis-infected and uninfected dogs. Metagenomic 16S rRNA was used for microbiome composition. The inter-rate agreement between PCR, Sanger, and MALDI-TOF was 0.631 k (p-value = 6.4 × 10−10). B. thuringiensis toxins were not found in dogs’ stool. Bacteroidota and Bacillota were the major phyla in the dogs’ microbiome (both represented >20% of the total bacterial community). Prevotella was underrepresented in all Strongyloides-positive dogs. However, the general composition of bacterial communities was not significantly linked with S. stercoralis infection. The genomic methods allowed accurate differentiation between B. thuringiensis and B. cereus. There was no association between B. thuringiensis and S. stercoralis infection, but further studies are needed to confirm this finding. We provide the first descriptive results about bacterial fecal composition in dogs with S. stercoralis infection.


Introduction
Bacillus species are spore-forming, Gram-positive bacteria that include Bacillus thuringiensis, B. cereus, B. cytotoxicus, B. anthracis, B. pseudomycoides, B. weihenstephanensis, B. toyonensis, and B. mycoidesare.Being highly similar in genotype and phenotype, these bacteria are classified as part of the B. cereus group in taxonomy [1,2].Unambiguous differentiation between Bacillus species can be relevant for human and animal health, but this is not usually possible with routine diagnostics [3].For instance, B. thuringiensis is used as a biocontrol agent in agriculture, permitting a reduction in the amount of chemical products [4].This is possible through the production of various crystal protein toxins (Cry5, Cry6, Cry12, Cry13, Cry14, Cry21, and Cry55) that exhibit a substantial biological activity against various insects, nematodes, and other pathogenic pests [5][6][7][8].Previous studies reported B. thuringiensis-derived crystal protein Cry5B to be effective against a broad range of gastrointestinal parasitic hookworms [9] and Strongyloides stercoralis [10].In particular, Cry5B was tested in vitro and in vivo on hamsters and dogs, demonstrating good efficacy against Ancylostoma ceylanicum, Ancylostoma caninum, and Necator americanus [9].Regarding S. stercoralis, multiple stages, including the first larval stage (L1s), infective stage (iL3s), free-living adult stage, and parasitic female stage, were all susceptible to Cry5B, as indicated by the impairment of motility and decreased viability in vitro [10].Strongyloides stercoralis is a soil-transmitted helminth that affects around 600 million people globally, causes a neglected tropical disease [11][12][13], and occurs in humans, non-human primates, dogs, cats, and wild canids [11].Its zoonotic potential is under study [14][15][16].Indeed, some authors have considered strongyloidiasis a zoonotic disease, while others have argued that the different hosts carry host specialized populations of S. stercoralis.In particular, clinical manifestations of S. stercoralis in humans and dogs range from asymptomatic to severe infection, with symptoms and signs involving mostly the intestines, respiratory tract, and skin [17].Moreover, a potentially fatal syndrome (hyperinfection/dissemination) can develop; this is associated with immunosuppression in humans, while it can occur also in immunocompetent dogs [17].The implementation of public health strategies for the control of S. stercoralis in endemic areas has been recommended by the WHO [18].For instance, the strategies can include regular parasitological examinations of dogs and inspections of park soil, kennels, and dog shelters.Moreover, the strategies primarily entail mass drug administration of ivermectin, but caution should be paid to the possible emergence of drug resistance, considering the already-existing resistance to this drug observed in veterinary medicine [19,20].Hence, additional strategies that could help to contain environmental contamination with S. stercoralis larvae might be useful for integration with mass drug administration.The primary objective of this study was to assess a MALDI-TOF and genomic lab pipeline for the identification of B. thuringiensis and its genes involved in the synthesis of the Cry toxins, such as cry5Ab, cry5Ac, and cry5Ba genes [21].Our secondary objective was to investigate a hypothetical negative selective pressure of these toxins against S. stercoralis and a possible different bacterial composition in the gut microbiota of S. stercoralis-infected and uninfected dogs.

Study Setting and Population
A total of 44 bacterial strains (Table S1) from our collection were used for the genotyping and MALDI-TOF analyses.In particular, among this collection, 3 strains were used as positive controls for B. thuringiensis (NRRL B-18400, NRRL B-18765, and DSM 2046T) and 6 for B. cereus (LMG 6923T, LMG 12334, LMG 12335, LMG 17615, NCTC 11143, and ATCC 11778).Moreover, in order to explore the potential effect of B. thuringiensis on S. stercoralis, we analyzed 17 dog stool specimens (Table S2), a cohort enrolled in our previous study investigating the epidemiology of strongyloidiasis in Southern Italy [22].The dogs' samples were collected from a kennel and three farms located in Apulia Region and all samples were tested for S. stercoralis infection using both Baermann method and Real-Time PCR (rtPCR), as previously described [22].In particular, three dogs resulted to be infected by S. stercoralis using both methods and another one tested positive only by rtPCR (see Table S2).

Mass Spectrometry
The B. cereus group strains were cultured on Columbia Agar with Sheep Blood (PB5039A Thermofisher, Monza, Italy) and incubated at 37 • C for 16/24 h.The mass spectrometry identification was performed by MALDI-TOF using the instrument Maldi Biotyper MBT smart (Bruker, Milan, Italy) with the MBT IVD Library DB 8326 March 2019 J and the software MBT Compass IVD v 4.1.100(Bruker, Milan, Italy) following the manufacturer's instructions.

DNA Extraction/Purification
For the B. cereus group strains, the total genomic DNA was extracted and purified from a 2 mL overnight culture using the Wizard Genomic DNA purification kit (Promega Corporation, Madison, WI, USA), following the manufacturer's instructions.For the stool samples, the DNA was isolated from 200 mg using a Qiamp Fast DNA stool mini kit (Qiagen, Milan, Italy), according to the manufacturer's instructions.The samples were eluted in 30 µL of elution buffer.The quality and quantity of DNA was analyzed using a NanoDrop One/Onec Spectrophotometer (Thermofisher, Monza, Italy) and a Qubit 4 Fluorometer (Thermofisher, Monza, Italy).The isolated DNA was stored at −80 • C until PCR and sequencing.

Genotyping of B. cereus Group Strains
The B. cereus group strains (n = 44) were first characterized by Bacillus species-specific PCR for a fragment of gyrase B gene (gyrB), followed by Sanger sequencing (Eurofins Genomics, Ebersberg, Germany) as previously described [23].Moreover, the B. cereus group strains (n = 44) and the dog stools (n = 17) were analyzed using specific PCRs for the gene ces [24] coding B. cereus emetic toxin cereulide and for the genes cry5Ab, cry5Ac, and cry5Ba [21] coding for B. thuringiensis Cry toxins.In this analysis, we included the positive controls B. thuringiensis NRRL B-18400 and NRRL B-18765 for cry genes and B. cereus LMG 12,334 for ces.For all PCR experiments, the amplification products were loaded on a 1.5% agarose gel and visualized by exposure to ultraviolet (UV) light.
The Sanger sequences of the gyrB gene were aligned with Clustal X software v2.0 [25], obtaining a final consensus length of 645 nucleotides for the 44 strains under analysis and the type strain of the species B. mycoides DSM 2048T and B. cytotoxicus NVH 391-98T, for which the sequence was retrieved from the NCBI database.The consensus-sequence alignment was imported on MEGA software version 11 [26], and the Neighbor-Joining algorithm was used for the reconstruction of the phylogenetic tree.The evolutionary distances were computed using the Tamura-Nei model with complete deletion.The gyrB sequences of the strain B. subtilis subsp.subtilis BCRC 10255T was retrieved from the NCBI database and used as the outgroup.

Sequencing and Bioinformatic Analysis
To reduce the possible bias of biological/dietary/environmental interferents, we chose to analyze samples collected from dogs belonging to the same kennel, with sampling performed on the same day.For this purpose, among the total dogs cohort (n = 17), the metagenomics analysis was conducted on ten stool samples (three positives and seven negatives for S. stercoralis) collected from dogs living in the same kennel (Table S2).Libraries were prepared following the 16S Metagenomic Sequencing Library Preparation protocol (Illumina, San Diego, CA, USA) in two amplification steps: an initial 35 cycle PCR amplification using 16S rDNA V3-V4-specific PCR primers (16S-341F 5 ′ -CCTACGGGNBGCASCAG-3 ′ and 16S-805R 5 ′ -GACTACNVGGGTATCTAATCC-3 ′ ) and a subsequent amplification that integrates relevant flow-cell binding domains and unique indices (NexteraXT Index Kit, FC-131-1001/FC-131-1002).Libraries were sequenced on a NovaSeq instrument (Illumina, San Diego, CA, USA) using 300 bp paired-end mode.Base calling, demultiplexing, and adapter masking were carried out through the Illumina BCL Convert v3.9.3 (https://emea.support.illumina.com/(accessed on 18 January 2023)).The FASTQ sequences obtained were analyzed firstly using Kraken 2, which examines the k-mers obtained from a sequencing read sample with those produced from the Silva ribosomal RNA Database (release 138.1) available for Kraken 2 [27].The taxonomic abundance for each taxon was estimated through Bracken [28].In addition, the reads generated for the ten stool samples were analyzed through DADA2 version 1.8 [29] by the R 3.5.1 environment.DADA2 was run as described in https://benjjneb.github.io/dada2/tutorial.html(accessed on 20 July 2023), applying the following parameters: trimLeft equal to 30 and the truncLen option set to 270 and 200 for the forward and reverse fastq files, respectively.Taxonomic assignment was performed comparing the amplicon sequence variants (ASVs) predicted from DADA2 against the Silva ribosomal RNA Database (release 138.1) using the function assignTaxonomy and addSpecies.

Statistics
B. cereus and B. thuringiensis cases were summarized for each technique using frequencies and percentages.The agreement between MALDI-TOF, PCR, and Sanger was evaluated using Fleiss' kappa coefficient.S1.The results of Sanger sequencing are shown in Table S1 and Figure 1.Of note, one of the positive controls, B. thuringiensis NRRL B-18765, was misidentified as B. cereus by MALDI-TOF (Table S1).We then compared the data obtained from MALDI-TOF and the specific PCR followed by Sanger only on the B. cereus and B. thuringiensis results (n = 32 strains), excluding positive controls and other Bacillus species of the collection.We found 0.631 k (p-value = 6.4 × 10 −10 ) inter-rate agreement between the three methods.As summarized in Table 1, all three methods identified B. cereus in twenty-one strains and B. thuringiensis in four strains, whilst seven strains resulted in B. thuringiensis by genomic approaches and were misidentified as B. cereus by MALDI-TOF.For a better differentiation, the Bacillus strains were investigated by specific PCRs for the genes cry5Ab, cry5Ac and cry5Ba, which code for the B. thuringiensis toxin Cry5B and for the gene ces coding for the B. cereus emetic toxin cereulide (Table S1).The analysis was performed using control strains B. thuringiensis NRRL B18765 and NRRL B-18400 and B. cereus LMG 12334, confirming the presence of cry5Ab, cry5Ac, and cry5Ba only in B. thuringiensis and the presence of ces only in B. cereus, as expected (Table S1).Apart from the controls described above, no other strains from the collection showed signals for all the analyzed genes (Table S1).For the study's secondary objective, we tested the dogs' stool for the genes cry5Ab, cry5Ac, and cry5Ba, finding no amplification in all samples (Table S3).

16S Illumina
Both Kraken2|Bracken and DADA2 identified Bacteroidota and Bacillota as the most abundant phyla (>20% as maximum value, Figure 2, Tables S5 and S6).In particular, four dogs in the present study were characterized by a reduction in Bacteroidetes, with a consequently increase in the ratio between Bacillota and Bacteroidota, presented by KC1A (dogs with frequent episodes of diarrhea), KC3A (dog manifesting some episodes of diarrhea), KC7A (dog positive for S. stercoralis infection in 2018), and KC6A (dog positive for S. stercoralis infection).The main orders were Bacteroidales, Lactobacillales, and Eubacteriales (>20% as maximum value, Figure 2, Tables S7 and S8).The main genera were Prevotella, Streptococcus, Alloprevotella, Fusobacterium, and Clostridium (>20% as maximum value, Tables S9 and S10).Regarding the order Caryophanales (former Bacillales) and the genus Bacillus, they were mostly identified in the sample KC10A (S. stercoralis negative), with a relative abundance of 2.38% and 2.14%, respectively (Tables S7 and S9).In fact, in all the other samples these taxa were found with a relative abundance lower than 1.5% (Tables S9 and S10), resulting in the least represented bacterial community.This fact could explain why the genus Bacillus and, consequently, the order Caryophanales were not identified in the samples using the approach based on the DADA2 pipeline (Tables S8 and S10).The general composition of the bacterial communities was not directly linked with S. stercoralis infection.However, all S. stercoralis-positive samples (KC4A, KC6A, and KC9A) showed very low percentages of the genus Prevotella, ranging between 0.17% and 1.06% for the data analyzed with the Kraken2/Bracken approach and between 0.21% and 1.31% according to the DADA2 analysis.In addition, one S. stercoralis-negative sample (KC1A) showed similarly low values for Prevotella; interestingly, this sample originated from a dog with frequent episodes of diarrhea (Table S2).
tion.However, all S. stercoralis-positive samples (KC4A, KC6A, and KC9A) showed very low percentages of the genus Prevotella, ranging between 0.17% and 1.06% for the data analyzed with the Kraken2/Bracken approach and between 0.21% and 1.31% according to the DADA2 analysis.In addition, one S. stercoralis-negative sample (KC1A) showed similarly low values for Prevotella; interestingly, this sample originated from a dog with frequent episodes of diarrhea (Table S2).

Discussion
The primary aim of this study was to assess a lab pipeline for the identification of B. thuringensis and its characterization for the presence of genes involved in the synthesis of Cry toxins, such as cry5Ab, cry5Ac, and cry5Ba genes [21].While all six B. cereus controls were correctly identified, one out of the three B. thuringiensis positive controls (NRRL B-18765) was misidentified as B. cereus by MALDI-TOF, highlighting a possible discrepancy with the PCR/Sanger.This observation was deep-rooted extending the analysis to all the collection strains, and the gyrB genomic approach allowed differentiation between B. cereus and B. thuringiensis.

Discussion
The primary aim of this study was to assess a lab pipeline for the identification of B. thuringensis and its characterization for the presence of genes involved in the synthesis of Cry toxins, such as cry5Ab, cry5Ac, and cry5Ba genes [21].While all six B. cereus controls were correctly identified, one out of the three B. thuringiensis positive controls (NRRL B-18765) was misidentified as B. cereus by MALDI-TOF, highlighting a possible discrepancy with the PCR/Sanger.This observation was deep-rooted extending the analysis to all the collection strains, and the gyrB genomic approach allowed differentiation between B. cereus and B. thuringiensis.
We then characterized all the strains by cesand cry5-specific PCRs.As expected, only B. cereus had the genetic determinant for the emetic toxin cereulide (ces), whilst two positive B. thuringiensis controls (NRRL B-18765 and NRRL B-18400) confirmed all cry5 genes.
Of note, although B. cereus and B. thuringiensis have different pathogenicity and applications [30], still there are no available guidelines to reliably distinguish species and strains [3].So far, different approaches such as protein crystallization, pulsed-field gel electrophoresis, and molecular typing were tested [3], but no reliable differentiation has been achieved with good results.A recent work assessed MALDI-TOF mass-spectrometry for differentiating closely related Bacillus species [31].In this work, we chose gyrB as the target gene because it is also the target of the putative species-specific PCR developed by Yamada and colleagues (1999) [23] and we combined in-home PCR, Sanger sequencing, and MALDI-TOF approaches with the instruments and library currently used in our laboratory for a comprehensive identification of the strains.Our results highlight that the genomic approach (gyrB PCR and Sanger) is more accurate for B. thuringiensis identification compared to MALDI-TOF.Based on these results, we proceeded with our secondary objective and we used the approach for genomic characterization in order to investigate the potential presence of B. thuringiensis cry5 genes in dog stool.The data obtained from this analysis did not show a possible correlation between the presence of S. stercoralis infection and B. thuringiensis, as all samples from the dogs were negative for the toxin genes.To the best of our knowledge, to date, there is no evidence of in vivo correlation of B. thuringiensis with S. stercoralis or other parasites and only evidence of efficacy testing following treatment/administration [9,10].Moreover, neither the general composition of the bacterial communities nor specific taxa (at the phylum, order, or genus level) were directly associated with S. stercoralis infection.In addition to gyrBand cry5-specific PCRs, we used a 16S metagenomics approach in order to detect the fecal bacteria composition.To reduce possible bias of biological/dietary/environmental interferents, we chose to analyze samples collected from dogs belonging to the same kennel, with sampling performed on the same day.However, being a small group, some differences in gender, age, physiological, and health status remain among the dogs leading to possible differences in fecal composition as well.By the way, these characteristics were not taken into account in this work.In order to strengthen the analysis, we conducted the bioinformatics examination using two different tools, in order to investigate how different approaches of read analysis can affect the interpretation of bacteria composition and community structure and try to evaluate the agreement between them.Kraken2|Brachen is based on alignment-free k-mer searches against a reference genome library [27], while DADA2 infers amplicon sequence variants (ASVs) and the identification at taxonomic level is based on sequence alignment against a reference database [29].Overall, the two pipelines yielded almost identical compositions and relative frequencies at the phylum level for the samples analyzed, while some differences arose at the order and genus levels specifically for very-low-represented taxa, such as Caryophanales (former Bacillales) and Bacillus.The low sensitivity of DADA2 paired with SILVA for the characterization of samples at the genus level has been recently reported [32].In addition, the authors confirmed the accuracy of the SILVA database for the investigation of the composition of microbial communities.Another aspect to consider is that the outcomes of DADA2 pipelines are directly dependent on the parameters chosen by the users for read filtering and trimming, which can affect the final identification of the taxa [33].
The two pipelines identified the presence of five different phyla among the samples analyzed, where the most abundant were Bacteroidota (former Bacteroidetes [34]) and Bacillota (former Firmicutes [34]).This is consistent with previous reports [35], which characterized the microbiota of 96 healthy dogs.Bacteroidetes and Firmicutes, together with Fusobacterium, represent the most important bacterial phyla in the gastrointestinal tract of dogs [36].In particular, it is widely accepted that the Firmicutes/Bacteroidetes (F/B) ratio has an important role in maintaining intestinal homeostasis [37,38] and the evidence of this study is in accordance with that reported by Chaitman et al. (2020) [39], where the authors observed a significantly lower abundance of Bacteroidetes in dogs affected by acute diarrhea.
The overall composition of the fecal microbiome of the Strongyloides-infected and uninfected dogs enrolled in our study is in line with what is reported in the literature for dogs either under natural or experimental conditions [40,41].In spite of the overall similarities in the composition of the fecal microbiota between Strongyloides-positive and -negative dogs, Prevotella was relatively mostly abundant in the uninfected compared to the infected group.There are no other data in the literature exploring the potential microbiota difference between Strongyloides-infected and uninfected dogs.Limited data have been described in chronic S. stercoralis-infected humans, showing significantly expanded populations of Clostridia and Leuconostocaceae [42] and an increasing presence of the Ruminococcus torques group [43].On the other hand, discordant observations have been reported about Prevotella abundance in humans, with or without helminth infection [44].Putatively the dogs enrolled in this study had S. stercoralis infection of a long duration due to re-infection from the environment, suggesting a potential correlation with Prevotella differential abundance between Strongyloides-infected and uninfected dogs.Some limitations should be highlighted: (i) the small size of the dog cohort, which depended on the availability of samples in the context of our previous study [22] for which they were collected, and (ii) the uninfected dogs were not fully representative of the general canine population.
To conclude, here, we assessed an internal lab pipeline that aimed to characterize a list of B. cereus group species for three cry5 toxin genes and for the differential analysis of B. thuringiensis and B. cereus.Our results suggest that the genomic approach combining specific gyrB PCR and Sanger might be superior for Bacillus identification compared to the MALDI-TOF approach.
The specific PCRs and 16S metagenomics analyses on dog stools showed no significant correlation between B. thuringiensis and the fecal microbiome, although a potential Prevotella differential abundance between Strongyloides-infected and uninfected dogs should be further explored.We provide preliminary descriptive results about fecal bacterial composition in dogs with and without S. stercoralis infection.Further investigations are needed in larger cohorts to investigate whether bacterial toxins might have a role in reducing environmental contamination by S. stercoralis larvae.
MALDI-TOF and the specific gyrB PCR permitted us to identify B. cereus, B. thuringiensis, B. mycoides, B. cytotoxicus, and B. subtilis in the 44 B. cereus strains, as reported in Table

Microorganisms 2024 , 10 Figure 1 .
Figure 1.Neighbor-Joining phylogenetic tree based on the gyrB gene sequence comparison for the Bacillus strains under analysis with the corresponding sequence retrieved for the type strain B. mycoides DSM 2048 T , B. cytotoxicus NVH 391-98 T , and B. subtilis subsp.subtilis BCRC 10255 T .The tree was reconstructed through MEGA11 with the Tamura-Nei model and complete deletion treatment for gaps.The accession numbers of the sequence deposited and/or available in NCBI database were reported in brackets.

Figure 1 .
Figure 1.Neighbor-Joining phylogenetic tree based on the gyrB gene sequence comparison for the Bacillus strains under analysis with the corresponding sequence retrieved for the type strain B. mycoides DSM 2048 T , B. cytotoxicus NVH 391-98 T , and B. subtilis subsp.subtilis BCRC 10255 T .The tree was reconstructed through MEGA11 with the Tamura-Nei model and complete deletion treatment for gaps.The accession numbers of the sequence deposited and/or available in NCBI database were reported in brackets.

Figure 2 .
Figure 2. Phylum-and order-level gut microbiota composition in the fecal samples of 10 dogs.(A) Data obtained with Kraken2/Bracken.(B) Data obtained with DADA2.

Figure 2 .
Figure 2. Phylum-and order-level gut microbiota composition in the fecal samples of 10 dogs.(A) Data obtained with Kraken2/Bracken.(B) Data obtained with DADA2.

Table 1 .
Results of B. cereus and B. thuringiensis identification with MALDI-TOF, PCR, and Sanger methods.n = 32 is the total of analyzed strains.