Discrimination of Bacteria Belonging to Bacillus cereus Group at Species Level by Fourier Transform Infrared Spectroscopy

Abstract

The Bacillus cereus group (B. cereus group) comprises several closely related species that share high genetic similarity but display markedly different phenotypic traits and pathogenic potential. Reliable and rapid discrimination at the species level remains challenging using conventional microbiological and molecular methods. In this study, Fourier Transform Infrared (FTIR) spectroscopy was evaluated as a rapid phenotypic approach to differentiate seven members of the Bacillus cereus sensu stricto (B. cereus s.s.), Bacillus anthracis (B. anthracis), Bacillus thuringiensis (B. thuringensis), Bacillus mycoides (B. mycoides), Bacillus toyonensis (B. toyonensis), Bacillus wiedmannii (B. wiedmannii) and Bacillus weihenstephanensis (B. weihenstephanensis). A collection of 190 isolates previously characterized by whole genome sequencing was analyzed using the IR Biotyper system. Spectral data were processed through multivariate analyses, including principal component analysis and linear discriminant analysis, following a hierarchical classification strategy. FTIR spectroscopy enabled clear discrimination of B. anthracis from other members of the B. cereus group and allowed the separation of several additional species based on distinct spectral signatures. A further discrimination step permitted differentiation between B. cereus sensu stricto and B. thuringiensis, with minimal overlap. These findings demonstrate that FTIR spectroscopy represents a promising and rapid tool for species-level discrimination within the B. cereus group. While the results should be considered preliminary for species represented by a limited number of isolates, this approach shows strong potential as a complementary method to molecular techniques in routine diagnostics in food safety and veterinary microbiology.

1. Introduction

The Bacillus cereus group is composed of about 22 closely related species of Gram-positive, rod-shaped and spore-forming bacteria [1]. These groups of microorganisms attract scientific interest not only because of their roles in human health, particularly foodborne illnesses and zoonotic infections, but also due to their ecological versatility and potential applications in biotechnology.

Accurate species-level discrimination within the B. cereus group is of critical importance in both clinical and epidemiological contexts, as different species are associated with markedly distinct risks and management strategies. In the field of food safety, the ability to distinguish B. cereus s.s. from closely related species is essential due to its well-established role in foodborne intoxications and infections mediated by emetic and diarrheal toxins. In veterinary and public health contexts, the rapid and reliable identification of B. anthracis remains a priority because of its high pathogenicity, zoonotic potential, and implications for biosafety and outbreak response. Moreover, other members of the B. cereus group, such as B. thuringiensis, B. wiedmannii and B. toyonensis, are increasingly recognized for their relevance in environmental monitoring, human health, and food production systems. Consequently, diagnostic approaches capable of resolving species-level differences within this group are essential to support risk assessment, surveillance, and appropriate control measures across diverse application areas.

However, the high genetic and phenotypic similarity of these microorganisms often makes discrimination challenging using traditional microbiological techniques.

Current methods for typing microbial strains often rely on labour-intensive, time-consuming, and expensive techniques such as whole genome sequencing (WGS) analysis, limiting their applicability in rapid-response scenarios. In this context, the IR Biotyper utilizing Fourier transform infrared (FTIR) spectroscopy, offers a novel approach, providing specific spectra for fast strain typing within approximately 3 h after bacterial growth, including sample preparation and spectra acquisition.

FTIR spectroscopy is a well-established and widely used analytical chemistry technique, recently introduced for the characterization of microorganisms relying on the strain-specific absorption patterns of the infrared radiation [2]. Technology leverages the principle of vibrational spectroscopy to detect subtle chemical differences between bacterial strains, facilitating species-specific discrimination [3]. Each microbial strain exhibits a unique IR radiation absorption spectrum, which represents the fingerprint obtained based on the biomolecular components of the cell. The FTIR-based IR Biotyper^® system (IRBT—Bruker Daltonics GmbH & Co. KG, Bremen, Germany) proved to allow a promising novel approach in the field of microbial strain typing [4] with successful applications reported in different microbiology fields, namely food, veterinary and environmental microbiology [5,6,7,8], hospital hygiene [9,10,11,12,13,14,15] and probiotic production [16,17]. In this study, we evaluated the discriminatory power of IRBT to discriminate at the species level some members of the B. cereus group, investigating a cohort of animal and food-related isolates previously characterized with biomolecular methods. Definitively, the aim of this study was to demonstrate the possibility to use IRBT as a fast method for the identification and differentiation of seven species belonging to the B. cereus group, as an alternative to more common biomolecular analyses.

Compared to our previous study focusing on the discrimination between B. anthracis and B. cereus s.s. [18], the present work substantially expands the scope of FTIR spectroscopy application within the B. cereus group. Here, we extend the analysis from a binary species comparison to a multispecies framework, evaluating the discriminatory performance of the IR Biotyper across seven phylogenetically closely related species. Moreover, we introduce a hierarchical, multistep classification strategy that mirrors realistic diagnostic workflows, enabling progressive species differentiation within this complex group. This study therefore represents one of the first systematic assessments of FTIR spectroscopy for broad species-level discrimination within the B. cereus group, supporting its potential role as a complementary tool in food, veterinary, and public health microbiology.

2. Materials and Methods

A total of 190 B. cereus complex isolates were included in this study (n.51 strains of B. cereus s.s. isolated from various food, n.110 strains of B. anthracis, n.2 B. thuringiensis, n.1 B. mycoides, n.7 B. toyonensis, n.18 B. wiedmannii, n.1 B. weihenstephanensis), collected at the Anthrax Reference Center of Italy at the Istituto Zooprofilattico Sperimentale of Puglia and Basilicata. All used strains were previously typed by WGS analysis.

For IRBT analysis, strains were cultured on Tryptone Soy Agar (Liofilchem, Roseto degli Abruzzi, Italy) overnight at 37 °C. A 10 μL loopful of bacterial culture was resuspended in 100 μL of distilled sterile water and incubated at 98 °C for 30 min. Later, 50 μL were taken and added to 50 μL of 70% (vol/vol) ethanol in the tubes of the IR Biotyper kit (Bruker Daltonics, Bremen, Germany), which contain metal cylinders, spitting to obtain a homogeneous suspension. Finally, 15 μL of bacterial suspension was spotted into the IRBT silicon sample plate in three replicates and dried at room temperature. For each sample, three biological replicates (independent bacterial cultures on different days) were analyzed. Quality control was performed for every run using the Infrared Test Standards (IRTS 1 and 2) provided by the manufacturer (Bruker Daltonics, Bremen, Germany).

FTIR spectra were acquired, processed and analyzed by the IRBT system (Bruker Daltonics, Bremen, Germany), using the IRBT Opus and Client v3.1 software. Spectra were acquired in transmission mode in the spectral range of 4000–500 cm⁻¹ (mid-IR), vector-normalized using the Savitzky–Golay algorithm, and the second derivative over 9 datapoints was calculated.

Exploratory data analysis was performed using PCA (principal components analysis) and LDA (linear discriminant analysis), after spectra acquisition. The different wavenumber regions of the IR spectra, corresponding to the absorption of different classes of biomolecules, were investigated, to establish which wavenumbers provide the highest discriminatory power for the desired differentiation [19].

A multistep approach, using different wavenumber regions in each step, was found to be the optimal way to proceed, following a hierarchical approach. In the first step, a binary differentiation between B. anthracis and the other species was performed, as described by Manzulli et al. [18]. In the second step, the further separation of B. cereus s.s./B. thuringiensis from all the other species was achieved, analyzing the spectra in the wavenumber region corresponding to C-H bonds, amidic groups, carbohydrates and fingerprint (3000–2800, 1800–1700 and 1400–700 cm⁻¹). In the third step, B. cereus s.s. was differentiated from B. thuringiensis, using the wavenumber region corresponding to carbohydrates and fingerprint.

In each step, LDA algorithm was used to create predictive models, to assess the suitability of the method as a potential tool to type unknown samples. As LDA is a supervised method, the assignment of a group identifier to define the classes to be differentiated was necessary, as well as the check for overfitting. Each LDA model was built using 50% of the strains (randomly selected) for its training, assigning the species as a group identifier (in this case, the species). The remaining 50% strains, not assigned to any group identifier (species), were used as testing set, to check the robustness of the model. An exception was made for B. weihenstephanensis and B. mycoides, for both of which the only available isolate was included in the training set.

3. Results

3.1. First Step

The IR spectral region between 1300 and 700 cm⁻¹, corresponding to the absorption of polysaccharides and fingerprinting, proved to be the best for the discrimination of B. anthracis/B. cereus, in concordance with our previously published study [18]. Exploratory data analysis performed with PCA/LDA showed that B. anthracis and the other species of the B. cereus group included in this study form two clearly resolved clusters (Figure 1).

The LDA model proved to be very robust and showed a very good separation between B. anthracis and all the other species. The test spectra (represented by crosses) clustered correctly in the group to which they belonged (Figure 2).

3.2. Second Step

The IR spectral region comprising C-H bonds (3000–2800 cm⁻¹), amidic groups (1800–1700 cm⁻¹), polysaccharides (1300–900 cm⁻¹) and fingerprinting (900–700 cm⁻¹) enabled the best discrimination of B. cereus s.s./B. thuringiensis from the other species. Exploratory data analysis performed with PCA/LDA showed that B. anthracis and the other species of the B. cereus group included in this study form two well-separated clusters (Figure 3).

The LDA model shows an equivalent clustering (Figure 4).

3.3. Third Step

The IR spectral region 1300–800 cm⁻¹ (corresponding to the carbohydrates) enabled the discrimination between B. cereus s.s. and B. thuringiensis, except for one B. cereus s.s. strain, which clustered in the B. thuringiensis group (Figure 5 and Figure 6).

4. Discussion

The B. cereus group comprises a complex of closely related bacterial species, which share a high degree of genomic similarity but exhibit remarkably diverse phenotypes and pathogenic behaviours [1]. Discriminating among these species remains a significant challenge in routine diagnostics due to overlapping biochemical profiles and the limitations of conventional assays. In this context, our study demonstrates the efficacy of FTIR spectroscopy as a rapid and reliable method for differentiating species within the B. cereus group. These findings align with previous studies supporting the utility of FTIR spectroscopy for correct and reliable bacterial typing [20].

In our study, multivariate analysis of the spectral data revealed reproducible clustering patterns that were species-specific, suggesting a strong correlation between spectral signatures and taxonomic identity.

Of relevance is the robust discrimination of B. anthracis from all other species included in the analysis. Given the clinical severity and biosafety implications associated with B. anthracis, the availability of a rapid, culture-based, and non-genomic screening method represents a valuable asset for reference and surveillance laboratories. These results are consistent with earlier findings and further support the reliability of FTIR spectroscopy for high-consequence pathogen differentiation.

The ability of FTIR to differentiate between B. cereus and B. thuringiensis, two species historically difficult to distinguish due to their nearly identical 16S rRNA gene sequences and shared ecological niches, is also significant. While whole genome sequencing (WGS) remains the gold standard for taxonomic resolution, it is time-consuming, costly, and not always feasible in routine settings [21]. FTIR offers a complementary solution that is both rapid and more cost-effective for laboratories that are in possession of this instrument. Furthermore, the technique’s high throughput potential and minimal sample preparation requirements make it suitable for integration into diagnostic workflows in clinical, veterinary, and food microbiology laboratories. As antimicrobial resistance and zoonotic threats continue to rise, the need for fast and accurate microbial identification methods becomes increasingly urgent [22]. In this light, FTIR could play a pivotal role in strengthening surveillance and early detection systems, especially for emerging or re-emerging Bacillus pathogens.

While our findings are promising, certain limitations must be acknowledged. FTIR analysis may be influenced by culture conditions, growth phase, and media composition, necessitating standardization across laboratories for reproducible inter-lab comparisons [20]. Additionally, the uneven distribution of isolates across species, particularly for B. mycoides and B. weihenstephanensis, restricts the generalizability of conclusions for these species. Despite this limitation, the hierarchical analytical strategy adopted in this study allowed us to explore the potential of FTIR spectroscopy as a rapid phenotypic screening tool within the B. cereus group. The results demonstrate that, even in the presence of an imbalanced dataset, FTIR spectroscopy can provide informative discriminatory signals, supporting its use as a complementary approach to genomic methods in routine diagnostic and surveillance settings. Future studies based on larger and more balanced strain collections will be essential to validate and extend these findings. Moreover, inter-laboratory validation will be necessary, integrating quantitative performance metrics to further substantiate the diagnostic value of this approach.

5. Conclusions

This study, together with a previously published work, represents a preliminary but significant step toward evaluating the IR Biotyper-based FTIR spectroscopy as a tool for species-level discrimination within the B. cereus group. The results demonstrate that this approach can reliably distinguish B. anthracis from closely related species and can further differentiate several other members of the group using a hierarchical analytical strategy.

Although the findings for species represented by limited numbers of isolates should be interpreted with caution, the overall results support the potential of FTIR spectroscopy as a rapid, cost-effective, and complementary method to molecular techniques such as whole genome sequencing. With further validation on larger and more diverse datasets, and with appropriate standardization, this methodology could contribute meaningfully to routine diagnostics, epidemiological surveillance, and outbreak investigations involving Bacillus species.