MIR spectroscopy as alternative method for further confirmation of foodborne pathogens Salmonella spp. and Listeria monocytogenes

Abstract

Listeriosis and Salmonellosis are two of the most common foodborne diseases. Consequently, an early and accurate detection of Listeria monocytogenes and Salmonella spp. in food products is a critical concern of public health policies. Therefore, it is of great interest to develop rapid, simple, and inexpensive alternatives for pathogen detection in food products. In this study, mid-infrared spectroscopy has been successfully used to confirm Listeria species and the presence of Salmonella isolated from food samples. This methodology showed to be very sensitive and could be a rapid alternative to detect these important pathogens, allowing to obtain results in a few minutes after previous growth in selective media, avoiding the confirmation procedures that delay the achievement of the results for up to 2 days.

Introduction

Nowadays, the microbiological quality of food is a very important factor for food industries and regulatory agencies. The accurate and reliable detection and identification of microorganisms in food is critical in order to detect microorganisms early on to prevent their transmission, avoiding infections and/or food poisoning.

Listeria monocytogenes and Salmonella spp. are some of the most important and common foodborne pathogenic microorganisms, causing public health problems in almost all industrialized countries. The multistate outbreak due to cheese contaminated with L. monocytogenes (Heiman et al. 2016) and the international outbreak due to chia seeds contaminated with Salmonella spp. (Harvey et al. 2017) are recent examples.

The genus Listeria is constituted by six species, of which L. monocytogenes is considered an opportunistic serious foodborne pathogen for humans, causing a disease named listeriosis which is a significant public health hazard (Rocourt and Buchrieser 2007). The identification of Listeria at the species level in routine laboratories is time-consuming, laborious, and expensive (Pusztahelyi et al. 2016). Selective primary and secondary enrichment steps, followed by isolation in selective differential media, and suspect colonies confirmation procedures by biochemical tests are needed (Pusztahelyi et al. 2016). OCLA agar medium allows the visual differentiation of the pathogenic species (L. monocytogenes and L. ivanovii) from the non-pathogenic species (L. innocua). Colonies from pathogenic species present blue to green colour surrounded by an opaque halo that does not exist in the case of the non-pathogenic species. However, ISO 11290-1 (1996) specifies that the biochemical confirmation based on haemolytic activity and carbohydrate fermentation should be performed. L. monocytogenes and L. ivanovii possess haemolytic activity, whereas L. innocua does not have this capacity. Regarding carbohydrate fermentation, it is known that L. monocytogenes and L. innocua use glucose, lactose, and rhamnose in aerobic conditions, while L. ivanovii ferments xylose (Pine et al. 1989). The confirmatory tests extend the detection for more 2 days, being the total identification completed only after 7 days (ISO 11290-1 1996). In addition, despite of the advances in the detection methodologies, the incidence of false-negative results is still a problem (Benetti et al. 2013; Bernardi et al. 2015).

The genus Salmonella includes two species: S. enterica and S. bongori. S. enterica is further divided into six subspecies, including S. enterica enterica, that includes more than 2500 serovars (Tindall et al. 2005). Salmonellosis disease typically resolves in about 6 days and does not require treatment with antibiotics, however, bacteremia occurs in 3–10% of the reported cases (Hoelzer et al. 2011). Cultivation methods for the detection of Salmonella include a non-selective pre-enrichment step, followed by a selective enrichment and posterior plating in selective and differential solid media. Posteriorly, suspect colonies need to be confirmed by biochemical or serological tests as referred in ISO 6579 (2002), Gracias and McKillip (2004). The majority of Salmonella are recognized as non-lactose fermenters (Lac⁻) and hydrogen-sulphide producers (H₂S⁺), thus, confirmatory testing of all H₂S⁺ and/or Lac⁻ colonies is required, which extends the total identification time to 5 days. The majority of the H₂S⁺ and Lac⁻ colonies turn out not to be Salmonella enterica, but related species such as Citrobacter (Brenner et al. 2000).

During the last years, molecular and immunological procedures have been developed to detect the presence of Listeria and Salmonella in food (Yoshida et al. 2016; Su et al. 2016; Aydin et al. 2018). These methods, despite being rapid are yet expensive, requiring high skilled personnel. Moreover, they can be affected by the presence of certain substances in food products, which can lead to an underestimation of bacterial concentration or false negative results (Hu et al. 2009). Additionally, the legislation requires that positive results for the presence of Salmonella and Listeria in foods must be confirmed by traditional culture methods and further confirmation (ISO 11290-1 1996; ISO 6579 2002). Taking these aspects into account, it is very important to develop sensitive and specific faster methods to detect/identify these two foodborne pathogenic microorganisms in order to prevent illnesses associated to food consumption.

Vibrational spectroscopic techniques, infrared spectroscopy (IR) [mid-infrared (MIR) and near-infrared (NIR)] and Raman spectroscopy, have been used since the 1980s as complementary methods for bacteria differentiation owing to their rapid “fingerprinting” capabilities and the molecular information that they can provide (Ellis et al. 2012; Sahar and Dufour 2014; Kammies et al. 2016). These techniques present several advantages in the microbiological classification and identification fields. They are fast (requiring virtually no sample processing), non-destructive, multi-purpose (e.g., detection, enumeration, classification, identification) and discriminating at different taxonomic levels (serotype, strain, species or genus) (Lefier et al. 2006; Das and Agrawal 2011; Ferraro and Krischnan 2012; Moreirinha et al. 2016). Although both MIR and NIR have been used for bacterial identification purposes, NIR bands are 10 to 100 times less intense than the corresponding MIR bands. Upon comparison, MIR spectroscopy has a higher signal/noise ratio, resolution, sensitivity, and output than NIR. Therefore, MIR spectroscopy can easily elucidate the chemical composition as well as the secondary and tertiary structures of a given bacteria (S. Sivakesava et al. 2004; Ellis et al. 2012), facilitating its identification.

Identification of Listeria species and Salmonella serovars using IR spectroscopy has been undertaken previously (Lefier et al. 2006; Rebuffo et al. 2006; Brandily et al. 2011; Romanolo et al. 2015). However, these studies included some time consuming cell pre-treatments, such as growth of the bacteria in liquid medium with subsequent centrifugation and washing with distilled water. In this study, the potential of mid infrared spectroscopy (MIR) to confirm the presence of L. monocytogenes and Salmonella spp. in food products, after isolation in their selective media, was investigated. Moreover, the potential to differentiate Listeria species and to discriminate Salmonella from other bacteria that grew on XLD Salmonella selective medium and could be mistaken for Salmonella (e.g. Shigella flexneri, Pseudomonas aeruginosa and Citrobacter freundii) was also evaluated.

Materials and methods

Bacterial strains and growth conditions

Listeria spp. and Salmonella serovars were isolated from food products (cheeses, sausages and prepared dishes) according to ISO 6579 (2002) and ISO 11290-1 (1996), respectively. Three different species of Listeria were obtained: 3 isolates of L. monocytogenes, 2 isolates of L. ivanovii, and 3 isolates of L. innocua, as well as three serovars of S. enterica: 2 isolates of serovar S. Nottingham, 2 isolates of serovar S. Anatum and 2 isolates of serovar S. Liverpool (Table 1).

Table 1.

Bacteria species/serovars used in this study and respective food sources

Bacteria species/serovar	Food product
Listeria monocytogenes	Soft cheese
Listeria monocytogenes	Minced meat
Listeria monocytogenes	Smoked pork sausage
Listeria ivanovii	Soft cheese
Listeria ivanovii	Soft cheese
Listeria innocua	Bovine meat burger
Listeria innocua	Cured cheese
Listeria innocua	Pasta with bovine meat
Salmonella enterica sv. Nottingham	Cured cheese
Salmonella enterica sv. Nottingham	Cured cheese
Salmonella enterica sv. Anatum	Bovine meat burger
Salmonella enterica sv. Anatum	Hotdog with pork sausage
Salmonella enterica sv. Liverpool	Smoked sausage
Salmonella enterica sv. Liverpool	Cured cheese

Briefly, for the isolation of Listeria spp., a primary enrichment of the samples in half-Fraser broth (1:10), for 24 h at 30 °C was done. Then, a secondary enrichment was performed by transferring 0.1 mL of the previous culture to a tube with Fraser broth (Oxoid Lda., UK) that was incubated for 48 h at 37 °C. The cultures obtained in the primary enrichment were transferred to Listeria identification agar base (PALCAM, Oxoid Lda., UK) or Oxoid chromogenic Listeria agar (OCLA, Oxoid Lda., UK) media. PALCAM and OCLA plates were incubated at 37 °C and examined after 24 h or, if necessary, 48 h to check for the presence of Listeria characteristic colonies. Confirmation of Listeria spp. was carried out by selecting 5 presumed Listeria colonies and plating them in tryptone soya yeast extract agar (TSYEA, Oxoid Ltd., UK). The plates were incubated at 37 °C during 18–24 h. Typical colonies, colourless with an opaque halo, were then confirmed with catalase test. If the morphological and physiological characteristics indicated the presence of Listeria spp. a haemolysis test was performed in Columbia agar 5% (Oxoid Lda., UK) to investigate which species was present. Biochemical API Listeria and RAPIDEC L. monocytogenes identification kits (Biomerrieux, France) were used to confirm the identification of the species.

For Salmonella spp. isolation, a pre-enrichment in buffered peptone water (BPW, Oxoid Ltd., UK) for 18 h at 37 °C was made. Then, 0.1 mL of the previous culture was transferred to Rappaport–Vassiliadis with soya agar (RVS, Oxoid Ltd., UK) and 1 mL to Muller-Kauffmann tetrathionate novobiocin (TTmk, Oxoid Ltd., UK). Samples were incubated for 24 h at 42 °C (RVS) and 37 °C (TTmk) respectively. After incubation, samples were plated in xylose lysine deoxycholate agar (XLD, Merck, Germany) and brilliant green agar (BGA, Merck, Germany) in order to select typical colonies of Salmonella spp. (red with a black centre). The positive colonies were identified by API system (API 20E, Biomerrieux, France). The OMNI-O antiserum test (Bio-Rad, USA) was also used to confirm the presence of Salmonella (Salmonella spp. shows positive agglutination).

Collection strains L. innocua NCTC-11288 and L. monocytogenes NCTC-1194 were also included in order to assess the robustness of the multivariate analysis, being used as an external validation.

Collection strains S. flexneri DSM-4782, C. freundii NCTC-6272 and P. aeruginosa isolated on our laboratory (Vieira et al. 2012) were also used in this study, as they also grow in XLD agar and can be mistaken for Salmonella due to possible colour similarities.

Mid-infrared spectroscopy

Spectroscopic acquisition was performed in an infrared spectrometer (Bruker Alpha Platinum, Germany) with a resolution of 4 cm⁻¹ and 32 scans, in the mid-infrared region (4000-600 cm⁻¹). Analyses were performed in a room with controlled temperature (25 °C) and humidity (29%). Bacterial colonies grew in agar medium, for 18 h at 37 °C: trypticase soy agar (TSA, Merck, Germany) for Salmonella Nottingham and L. monocytogenes preliminary analysis, XLD for Salmonella, Shigella, Citrobacter and Pseudomonas and OCLA for Listeria. The colonies were collected with a loop and placed directly on the crystal of a 2 mm × 2 mm horizontal single attenuated total reflectance (ATR) accessory and were dried under gentle cold air flow for 10 s. At least three replicate spectra were obtained for each sample, obtained from different plates in different days. The sampling accessory was cleaned with ethanol (70%) and distilled water between each measurement.

Statistical analysis

The spectra (obtained in OPUS format) were transferred via JCAMP.DX format to an in-house developed data analysis package (CATS build 97). Principal component analysis (PCA) was used to find the major sources of variability in data, detect outliers and detect the probable presence of clusters. Previous to PCA, the spectra were standard normal deviate (SNV) corrected.

Results and discussion

Listeria species discrimination

In this study, instead of the biochemical confirmation step, MIR was used to assess the species distinction.

MIR spectral profile of Listeria spp

Figure 1 shows the spectral profile of the three bacterial strains of Listeria obtained in OCLA agar. Despite their similarity, some spectral differences were visible in the region between 1800 and 980 cm⁻¹. Consequently, this region was selected to perform the discrimination by principal component analysis.

Fig. 1.

MIR average spectra of L. monocytogenes, L. ivanovii and L. innocua obtained from OCLA agar in the spectral region between 4000 and 500 cm⁻¹

Principal component analysis of Listeria spp

PCA analysis (Fig. 2a) revealed 3 different clusters. Listeria monocytogenes is located on negative PC1, mainly characterized by a peak at 1517 cm⁻¹ (Fig. 2b). Listeria innocua and L. ivanovii are located on positive PC1, both characterized by a broad band between 1700 and 1570 cm⁻¹ and a peak at 1200 cm⁻¹. Listeria innocua is also located at positive PC2, characterized mainly by a peak at 1666 cm⁻¹ and another at 1505 cm⁻¹, and L. ivanovii is located at negative PC2, with the most significant peaks at 1615, 1330, 1240 and 1035 cm⁻¹.

Fig. 2.

Principal component analysis (PC1 vs. PC2) of L. ivanovii, L. monocytogenes and L. innocua spectra in the region between 1800 and 980 cm⁻¹. a Scores scatter plot, b loadings plot profile. L. innocua and L. monocytogenes from NCTC culture collection are surrounded by a circle

In this case, it was possible to discriminate the 3 different studied Listeria species after growth in OCLA agar, with the spectra from the collection strains L. monocytogenes and L. innocua being grouped along with the spectra of the isolated bacteria of the same species.

Regarding the chemical structure of the cell wall of the different species, there is little information. In consequence, it is difficult at this stage to make specific assignments to the discriminatory peaks (Romanolo et al. 2015). However, differences on the surface proteins, which influence the species virulence, would contribute to the found discrimination peaks (between 1700 and 1500 cm⁻¹). This is in accordance with studies showing that surface protein patterns are specific for species and even serovars (Su et al. 2016). Furthermore, our study indicates that not only the proteins may have a significant paper on the species distinction, but also the saccharides and phosphates, that are also present in bacterial extracellular envelope, were responsible for the obtained distribution (peaks between 1300 and 980 cm⁻¹) (Orsini et al. 2000; Davis and Mauer 2010). Taking these results into account, MIR spectroscopy allowed the rapid differentiation of the 3 Listeria studied species without the necessity of performing the biochemical confirmation step, after obtaining the colonies in the selective medium. This could shorten the identification procedure in 2 days.

Salmonella identification

MIR spectral profile of Salmonella and other bacteria

Mid infrared spectra from the isolated bacterial strains of Salmonella, C. freundii, P. aeruginosa and S. flexneri obtained in XLD agar showed similar profiles (Fig. 3). However, as in the case of Listeria species, some spectral differences are visible at the region between 1800 and 980 cm⁻¹, and this was also the chosen region to perform principal component analysis.

Fig. 3.

MIR average spectra of C. freundii, P. aeruginosa, S. flexneri, S. Nottingham, S. Liverpool and S. Anatum obtained from XLD agar in the spectral region between 4000 and 500 cm⁻¹

Principal component analysis of Salmonella and other bacteria

The PCA of the three Salmonella serovars, C. freundii, P. aeruginosa and S. flexneri are shown in Fig. 4. Five distinct groups are visible: S. Anatum and S. Liverpool were grouped together and S. flexneri, S. Nottingham, C. freundii and P. aeruginosa formed distinct groups. S. Anatum, S. Liverpool and S. flexneri were located on negative PC1. C. freundii, S. Nottingham and P. aeruginosa were located on positive PC1. C. freundii on negative PC2 separates from S. Nottingham and P. aeruginosa, both on positive PC2 (Fig. 4a).

Fig. 4.

Principal component analysis (PC1 vs. PC3) of S. enterica serovars (Anatum, Nottingham and Liverpool), S. flexneri and C. freundii and P. aeruginosa spectra in the region between 1800 and 980 cm⁻¹. a Scores scatter plot, b loadings plot profile

According to the loadings plot profile, the most significant peaks that contributed to the distinction of the studied species were located in the region between 1200 and 980 cm⁻¹, region of saccharides and phosphates, compounds that may differ at the surface of the different studied species/serovars (Fig. 4b). As it is known, IR spectra of bacteria provide not only the absorption bands that describe molecular composition of the cells, but many of these bands are also sensitive to structure changes, numerous intra and inter-molecular interactions including the hydrogen bonding pattern, membrane constitution, lipid-protein interactions, and conformational states as the secondary structures of proteins (Lasch and Naumann 2006). So, in this case, it is also difficult to make an accurate assignment of the spectral bands to structural compounds of bacteria external membrane or cell wall.

XLD agar offers the possibility of selecting Salmonella-like colonies because the medium contains indicators of H₂S production (typical of Salmonella spp.) and pH changes (Salmonella spp. exhaust the xylose of the medium and decarboxylate the lysine, altering the pH to alkaline). Shigella spp. are genetically close to Salmonella spp. and Citrobacter spp. (Doyle and Buchanan 2013). Citrobacter generally grows as yellow colonies, different from Salmonella, however, there are some works reporting that C. freundii and other Citrobacter species produced false-positive colonies (Warburton et al. 1994; Coleman et al. 1995). Likewise, P. aeruginosa and Shigella spp. can produce colonies very similar to those produced by some Salmonella species on XLD media, leading to misinterpretation of the results (Maddocks et al. 2002). Taking this into account, MIR can help to overcome these problematics, as in this case these bacteria were easily differentiated with this methodology.

Despite the existence of distinct groups, it was intended to obtain a better separation of Salmonella serovars from S. flexneri. As the latter is distributed in the same quadrant of S. Anatum and S. Liverpool, a PCA with Shigella and Salmonella serovars was performed, and a good separation between Shigella and Salmonella was observed (Fig. 5).

Fig. 5.

Principal component analysis (PC1 vs. PC2) of S. enterica serovars (Anatum, Nottingham and Liverpool) and S. flexneri spectra in the region between 1800 and 980 cm⁻¹. a Scores scatter plot, b loadings plot profile

Salmonella flexneri is separated from the other bacteria in the negative PC2. S. Anatum, S. Liverpool and Shigella are separated from S. Nottingham in the positive PC1 (Fig. 5a). The loadings plot profile (Fig. 5b) shows that the distribution of the species is mainly characterized by peaks in the region between 1700–1500 and 1200–980 cm⁻¹, that may have the contribution of protein, phosphate and saccharide differences between the studied species/serovars.

Conclusion

In general, the studied bacteria were discriminated by peaks corresponding mainly to polysaccharide (1200–980 cm⁻¹) and protein (1700–1500 cm⁻¹) spectral regions. This was expectable, as the cellular surface of the different species and serovars studied differ in these compounds. Some signals, corresponding to DNA phosphates, could also be important in the discrimination of the bacteria.

In this study, mid-infrared spectroscopy has proved to be a very useful method for the confirmation of Salmonella spp. and Listeria spp. after their isolation in selective media. Our data are indicative that MIR can be used instead of the traditional biochemical methods, as it allows a very good discrimination of the analysed Listeria, Salmonella and other foodborne bacteria, by forming well discriminated groups separating the studied species in PCA analysis. Ideally, matching the spectra obtained for bacterial colonies with a previously obtained spectral library of different foodborne bacteria would allow the immediate identification of possible pathogens. Further studies using both MIR and traditional confirmatory methods in different real food samples are needed to evaluate the potential of MIR to confirm the presence of pathogenic bacteria. Moreover, MIR can also be a good substitute for the existent molecular identification methods that despite of being accurate, are expensive, complex and time consuming, delaying the detection of the microorganisms.

The methodology used in this work is very easy to perform (the colonies obtained in specific agar media are directly placed on the MIR sampling accessory without any handling), rapid (a few seconds to obtain a spectrum) and considerably inexpensive, taking in account that MIR spectrometers with very good quality and not expensive are available nowadays. Using this method, the time for the detection of Salmonella spp. and Listeria spp. can be shortened up to 2 days, comparing with traditional methods. Another advantage is to avoid the use of additional expensive chemicals needed for traditional biochemical confirmation procedures. MIR could be a new tool for laboratories and manufacturers to easily screen food pathogens, even though more research work is needed, namely further studies with more species/serovars should be performed to validate this methodology.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.