An ethics waiver (reference number: W-CBP-220422-03) was obtained from the Witwatersrand Human Research Ethics Committee (Medical), University of the Witwatersrand, Johannesburg, South Africa. The study was conducted between March 2022 and August 2022, and involved no human participants or animal research. Bacterial isolates and PIF samples were sourced from the Infection Control Service Laboratory of the National Health Laboratory Service. The PIF samples were tested for contaminants during routine food testing at the Infection Control Service Laboratory before use in artificial contamination experiments. All of the experiments complied with institutional and laboratory biosafety and ethical guidelines. The data generated contained no personal identifiers and were managed following institutional data protection policies.

This study was a laboratory-based validation combining in silico and in vitro approaches to evaluate the performance of primers for a duplex PCR assay that can rapidly detect Cronobacter spp. and C. sakazakii in PIF. The study was conducted at the Infection Control Service Laboratory of the National Health Laboratory Service at the University of the Witwatersrand in Johannesburg, South Africa, between March 2022 and August 2022.

This study used a collection of 27 complete Cronobacter genome sequences (20 C. sakazakii, 2 C. malonaticus, 1 C. turicensis, 1 C. dublinensis, 1 C. universalis, 1 C. muytjensii, and 1 C. condiment), and 20 complete genome sequences of non-Cronobacter foodborne pathogens, including three Salmonella Typhi, two Shigella sonnei, two Listeria monocytogenes, three E. cloacae, two Escherichia coli, three Bacillus cereus, two Klebsiella pneumoniae, and three Staphylococcus aureus. The genome sequences were downloaded as FASTA files from the National Center for Biotechnology Information database (NCBI: https://www.ncbi.nlm.nih.gov) and were used in the in silico PCR analysis. In addition, 11 previously characterised foodborne pathogens and four PIF samples (PIF7069, PIF6748, PIF7070, and PIF6749) from two brands were obtained from the Infection Control Service Laboratory through routine diagnostics; these were used for the in vitro PCR analysis. The foodborne pathogens included three Cronobacter isolates (C. turicensis, C. sakazakii M0020, and C. sakazakii M0022) and eight non-Cronobacter isolates (Salmonella Typhi, S. sonnei, L. monocytogenes, E. cloacae, E. coli, B. cereus, K. pneumoniae, and S. aureus). Routine food testing confirmed that PIF7069, PIF6748, and PIF7070 were free of bacterial contamination, whereas PIF6749 was positive for B. cereus.

A literature search through PubMed, Google Scholar, and Scopus was performed to identify gene targets for the molecular detection of Cronobacter spp. and C. sakazakii. The search was performed using the following terms: ‘Cronobacter sakazakii’, ‘Cronobacter spp.’, ‘PCR’, ‘gene target’, ‘species-specific primers’, ‘genus-specific primers’. Seven gene targets were selected based on published studies that reported their potential diagnostic value in differentiating Cronobacter spp., specificity, sensitivity, and availability of the primer sequences. These seven gene targets included six species-specific genes (cgcA, gyrB, rpoB, fimp1, fimG, lpfA_1) and one genus-specific gene (grxB). The corresponding primer sequences for each gene target were downloaded and used in the in silico PCR analysis (Table 1).

Genome sequences of the 27 Cronobacter species and 20 non-Cronobacter foodborne pathogens were used in the in silico PCR analysis to determine the sensitivity and specificity of the primer sequences for each gene target. The in silico PCR was performed using an online platform available at http://insilico.ehu.es/PCR/, which simulates theoretical PCR amplification based on primer sequences and genome sequences.³⁵ Each primer pair was tested against each of the 47 genome sequences, using default settings to determine the presence or absence of an amplification product matching the expected amplicon size. The in silico PCR analysis provides the predicted start position and size of the amplicon. Results were recorded as positive (+) if the expected product size was detected, and negative (-) if no product was observed. Thereafter, the sensitivity and specificity of each primer set were calculated using Equations 1 and 2. The species-specific primers with the highest sensitivity and specificity for C. sakazakii detection were selected for further validation in vitro.

Cronobacter turicensis, C. sakazakii M0020, and C. sakazakii M0022, and eight non-Cronobacter foodborne pathogens were cultured on nutrient agar (Thermo Fischer Scientific, Waltham, Massachusetts, United States) and incubated overnight at 36 °C. Crude genomic DNA was extracted by suspending two loopfuls of bacterial colonies in 200 µL of tris-ethylenediaminetetraacetic acid (TE) buffer, followed by incubation at 95 °C for 25 min, vortexing, and centrifugation at 12 000 rpm for 3 min. The supernatant was collected and stored at 8 °C until further use. Two singleplex PCR assays targeting the lpfA_1 and grxB genes, were optimised in a 25 µL reaction volume containing 12.5 µL of 2X PCR Master Mix (Thermo Fischer Scientific, Waltham, Massachusetts, United States), 2 µL of DNA, 0.2 µM primers, and nuclease-free water. The PCR cycling conditions were 95 °C for 3 min, 35 cycles of 95 °C for 30 s, 51 °C for 30 s, 72 °C for 60 s, followed by a final extension at 72 °C for 10 min. Amplicons were resolved by gel electrophoresis on a 2% agarose gel stained with GelRed™ (Anatech, Johannesburg, South Africa), using 1X Tris-Borate-EDTA buffer at 110 V, and visualised using the GelDoc system (Bio-Rad Laboratories, Hercules, California, United States).

The duplex PCR assay targeting both lpfA_1 and grxB was optimised under the same conditions, with each primer added at a final concentration of 0.2 µM. In all PCR reactions, C. turicensis and C. sakazakii M0020 were used as positive controls for the grxB and lpfA_1, and nuclease-free water was used for the negative control. The specificity of the duplex PCR assay was evaluated using genomic DNA from C. sakazakii M0022, C. turicensis, and eight non-Cronobacter foodborne pathogens listed above. The specificity of the assay was calculated using Equation 2.

To artificially contaminate the PIF samples with C. sakazakii, 5 µL of a suspension (10⁸ colony-forming units/mL) was added to 10 g of each sample in a sterile glass bottle with 100 mL buffered peptone water, followed by incubation for 6 h at 36 °C. This resulted in an estimated final concentration of approximately 5.0 × 10⁶ colony-forming units/mL in the pre-enrichment. The PIF7070 sample was also inoculated with 5 µL of Salmonella Typhi (10⁸ colony-forming units/mL). After incubation, 40 mL of the pre-enrichment broth was centrifuged at 3000×g for 10 min. The pellet was resuspended in 200 µL of supernatant, and genomic DNA was extracted using the High Pure PCR Template Preparation Kit (Roche Diagnostics GmbH, Mannheim, Germany), followed by PCR amplification using the duplex PCR assay. Powdered infant formula samples PIF7070 and PIF6749 were specifically used to evaluate assay interference. The experiment was performed in duplicate.

All data were collected and recorded using Microsoft Excel® 2010. For both in silico and in vitro PCR assays, amplification results were recorded based on the presence or absence of the expected amplicon size. Similarly, results from the artificially contaminated PIF samples were classified as positive or negative according to band visualisation on a 2% gel electrophoresis. A true positive was defined as a C. sakazakii or Cronobacter spp. yielding a positive amplification with the expected amplicon size. A true negative was defined as a non-Cronobacter foodborne pathogen (or genome sequence) with no amplification, confirming assay specificity. False positives were recorded when amplification occurred in non-Cronobacter spp., and false negatives when the assay failed to amplify the target genes for Cronobacter spp. The results were tabulated, and the sensitivity and specificity were calculated using Equation 1 and Equation 2: Specificity=(true negativestrue negatives+false positives)×100[Eqn 1] Sensitivity=(true positivesfalse positives+true negatives)×100[Eqn 2]

In total, seven primer pairs corresponding to the seven gene targets for Cronobacter spp. and C. sakazakii detection were evaluated against 47 genome sequences using in silico PCR. Species-specific primers targeting the lpfA_1, fimG, and fimp1 genes demonstrated 100% sensitivity and specificity for C. sakazakii detection during in silico PCR. The cgcA and gyrB primers had 100% specificity, with sensitivities of 94.7% and 89.5%, each missing one or two C. sakazakii sequences during the in silico PCR (Table 2). The rpoB primers exhibited the lowest performance, with sensitivity and specificity of 78.9% and 92.6%, whereas the genus-specific grxB primers detected all Cronobacter spp. except C. dublinensis and C. condiment, resulting in 92.3% sensitivity and 100% specificity (Table 2). Based on these results, lpfA_1 and grxB primers were selected for in vitro testing.

Two singleplex PCR assays targeting the lpfA_1 and grxB genes were successfully optimised and effectively detected the respective genes in C. sakazakii M0022, C. sakazakii M0020, and C. turicensis (Figure 1a). Additionally, a duplex PCR assay targeting both genes was optimised, demonstrating simultaneous detection of both the lpfA_1 and grxB genes in C. sakazakii M0022, C. sakazakii M0020, while only detecting the grxB gene in C. turicensis (Figure 1b).

The duplex PCR assay was 100% specific for Cronobacter spp. and C. sakazakii, with successful amplification of the grxB gene in C. sakazakii M0022, C. sakazakii M0020, and C. turicensis, as well as the amplification of the lpfA_1 gene in C. sakazakii M0022 and C. sakazakii M0020, and no amplification in any of the eight non-Cronobacter foodborne pathogens (Figure 2a). Moreover, the lpfA_1 and grxB genes were successfully detected in the PIF samples artificially contaminated with C. sakazakii M0022 (Figure 2b, Table 3), with no interference from the non-Cronobacter foodborne pathogens in PIF samples PIF6749 (containing B. cereus) and PIF7070 (containing Salmonella Typhi). These results demonstrate the assay’s reliability and suitability for detecting Cronobacter spp. and C. sakazakii in PIF, even in the presence of other foodborne pathogens.

This study had a few limitations. The in vitro sensitivity of the duplex PCR assay could not be fully assessed because of the limited availability of C. sakazakii and other Cronobacter spp. isolates, despite efforts to source additional strains from public and private laboratories. However, incorporating in silico validation of the primers using publicly available genome sequences from multiple Cronobacter species allowed us to evaluate primer specificity across diverse species. In silico analysis thus provided a valuable alternative for assay validation. This approach also supports the growing utility of bioinformatics tools in enhancing assay design. Another limitation of this study is that the sensitivity of the assay was determined using high bacterial concentrations, which may not fully represent the variable contamination levels of C. sakazakii typically found in PIF, as contamination may occur at low levels, and accurate detection under such conditions is critical for food safety. Future work should therefore include the determination of the assay’s limit of detection using serial dilutions of C. sakazakii spiked into PIF. While this study focused on conventional PCR to ensure affordability and accessibility in basic laboratory settings, it may not fully meet the throughput and automation needs of modern food safety laboratories. Adapting the assay into a real-time multiplex PCR assay will allow for quantitative detection, improved sensitivity, and reduced turnaround time. Additionally, integrating automated DNA extraction workflows could streamline sample processing and minimise human error, making the assay more suitable for high-throughput environments and outbreak response scenarios. These enhancements would support broader implementation in both resource-limited and well-equipped laboratories, contributing to more efficient monitoring of Cronobacter contamination in infant formula.

This assay combined in silico and in vitro validation of a rapid and affordable duplex PCR assay for the simultaneous detection of Cronobacter spp. and C. sakazakii in PIF. The performance of the assay was not affected by the presence of non-Cronobacter species such as B. cereus and Salmonella Typhi. This duplex PCR assay can facilitate the diagnosis of C. sakazakii in PIF, particularly in resource-limited settings, contributing to improved epidemiological surveillance and screening of Cronobacter spp. in PIF. Its adaptability for routine diagnostics and outbreak investigations further underscores its potential for integration into national food safety monitoring programmes.