Evaluation of the different methods to detect Salmonella in poultry feces samples

Salmonella is one of the most common causes of foodborne outbreaks and infection worldwide. The gold-standard detection method of Salmonella is cultivation. There is a need to investigate rapid and accurate processes with time-consuming cultivation. The study evaluated different approaches to detect Salmonella in poultry feces samples. Poultry farm feces samples from 21 cities in Iran were collected from January 2016 to December 2019. Microbiological cultures, serological assays, and multiplex PCR (m-PCR) were used to detect and characterize Salmonella spp. isolates. Serological assays and m-PCR were used to determine the serogroups A, B, C1, C2, D1, E, H, and FliC. The m-PCR was used to detect seven Salmonella serovars, and a Chi-square test was performed to compare the discriminatory power of the methods. Of 2300 poultry feces samples, 173 (7.5%) and 166 (7.2%) samples were detected as Salmonella spp. by cultivation and m-PCR, respectively. The sensitivity of the molecular method was equal to cultivation at 0.96 (CI = 95%). Assessment of H antigenic subgroups showed the same for both m-PCR and serological tests. Therefore, the matching rate of the two methods for detecting all H antigenic subgroups was 100%. Thus, the relationship between the results obtained from both methods was significant in the contingency table test (P < 0.01). The PCR-based approach confirmed the detection of Salmonella in a shorter period (24–36 h) compared to the conventional microbiological approach (3–8 days).


Introduction
Salmonella is gram-negative, rod-shaped bacteria in the family Enterobacteriaceae. Salmonella is the most common bacterial pathogen associated with foodborne disease in the United States (Gu et al. 2019). Poultry products, including meat and eggs, have been a significant source Communicated by Erko Stackebrandt.
Reza Khaltabadi Farahani and Maryam Meskini are co-first authors. of Salmonella infections (Wang et al. 2020). Salmonella infections account for 93.8 million cases and 155,000 deaths per year worldwide, with several serotypes involved, such as Salmonella enterica serovar Enteritidis, S. enterica serovar Senftenberg, S. enterica serovar Hadar, S. enterica serovar Agona, and S. enterica serovar Typhimurium (Gantois et al. 2008). S. enterica serovar Enteritidis is the most reported in human outbreaks during the last 2 decades (Ghazalibina et al. 2019). There is limited information on the prevalence of foodborne diseases in Iran. Studies have considered small populations and did not show the status of foodborne diseases in Iran. The Centers for Disease Control developed a national guideline for foodborne diseases in 2006 and launched it in the same year (Asl et al. 2015). The country's technical committee has approved the guideline. Its first step is to identify the five most common foodborne illnesses caused by Salmonella, Shigella, E. coli, Staphylococcus toxin and Botulism (Asl et al. 2015). Therefore, Salmonella spp. infection prevention is crucial for poultry health and food processing industries.
Accordingly, the diagnosis and serotyping of Salmonella spp. are critical subjects. Conventional Salmonella detection methods include culturing in a selective medium, followed by colony characterization using biochemical and serological tests (Kasturi 2020). Most laboratories use serotyping as the main phenotyping method for subspecies Salmonella spp. typing and approximately 2600 serotypes have been described according to the Kauffman-White-LeMinor scheme. Included is the somatic antigen (O) determining the group, flagellar antigen (H) determining the serotype, and capsular antigen (K) (Kariuki et al. 2015).
Conventional detection methods are laborious and timeconsuming. Serotyping methods are often ineffective as epidemiological tools because of their low discriminatory capacity for strains with the same serotype or similar biochemical characteristics (Barrow and Neto 2011). Therefore, it is not often possible for research laboratories to detect Salmonella in-house, and isolates are sent to commercial or expert laboratories, which may delay the results (Diep et al. 2019). Consequently, a rapid and sensitive method is required to detect Salmonella spp. and their serovars. Some molecular techniques are widely used to detect Salmonella spp. and serovars. Molecular techniques are used instead of conventional methods because of their reduced time for diagnosis with similar or higher efficiency, increased discriminatory power, simplicity, better standardization, reproducibility, and higher sensitivity and specificity (Malorny et al. 2009;Khaledi and Meskini 2020). The diagnosis time of Salmonella in samples is very critical because, in some situations, especially the clearance of heavy shipments is very important in several respects: first, to observe the cold chain to prevent their spoilage (for example, shipments of several thousand tons of imported meat); secondly, the cost of keeping them in proper conditions at customs is very high, and thirdly, the late clearance of such shipments is important in terms of cargo fluctuations. By shortening the detection time to less than 2 days by molecular method, all the above problems can be overcome (Barrow 1994;Roy et al. 2002).
Molecular techniques based on the amplification of DNA, such as multiplex polymerase chain reaction (m-PCR), have been used to detect Salmonella serotypes (Du et al. 2020). The m-PCR uses pairs of primers that allow the simultaneous detection and identification of specific DNA sequences in the same reaction (Maciorowski et al. 2005). To the best of our knowledge, the comparison of traditional versus m-PCR techniques to detect Salmonella genus, serogroups, and serovars in Iran has not been conducted. Therefore, this study compared the discriminatory power of several methods such as cultivation, serological, and m-PCR to detect Salmonella genus, serogroups and serovar in farm poultry feces samples.

Collection of samples and isolation of Salmonella
Poultry feces samples from five different areas in each poultry farm were collected (Fig. 1). Farms were located in Semnan, Fars, Qazvin, Qom, Yazd, Khorasan Razavi, Mazandaran, Kerman, Alborz, South Khorasan, Kurdistan, Markazi, Isfahan, Kohkiluyeh, Boyerahmad, West Azerbaijan, Golestan, Zanjan, Hamedan, Kermanshah, Khuzestan, and Lorestan, of Iran. Using sterile spoons, samples were collected in sterile zipper bags from January 2016 to December 2019. All samples were transported immediately to the Department of Molecular Microbiology, Pasteur Institute of Iran while maintaining sterile and cold chain conditions.
The PCR was performed in a thermocycler (Eppendorf Thermomixer comfort, Germany). The oligonucleotide sequences used in this study, annealing temperature, and the expected band size are listed in Table 1. The PCR product fragments were analyzed in 2% (w/v) agarose gel by electrophoresis using a 1× TAE buffer. Fragment size was determined by comparison with Gene-Ruler 100 bp DNA ladder (Fermentas, EU).

Multiplex PCR to serovar typing Salmonella
The m-PCR was performed for invA (Styinva-JHO-2), sdf  Figure 3 shows all summarized steps performed in the present study.

Statistical analysis
The statistical analysis of data was conducted using IBM SPSS version 16.00 (SPSS Inc., Chicago, IL, USA). The Chi-square test was used to compare different methods, and P < 0.05 was considered statistically significant.

Salmonella genus and serogroup detection
A total of 2300 poultry feces samples from farms located in 21 cities of Iran were collected from January 2016 to December 2019. The percentage of abundance of different Salmonella serogroups is given in Fig. 4. Among them, 173 (7.5%) samples were detected as Salmonella by cultivation, and 166 (7.2%) samples were detected as Salmonella by amplification of invA gene. Thus, the sensitivity of molecular detection and microbiological cultivation was equal to 0.96 (CI = 95%). Molecular serotyping gave the same results as the antisera approach with A (prt), C1 (wzxC1), and E (wzxE1) serogroups. The concordance of molecular detection of serogroups (Fig. 5) and the serological method for all samples were 100%. Other samples were in different groups, and the matching results of the two molecular and serological serotypes were higher than 93% for all samples. The detection of B (rfb), C2 (wzxC2), and D (tyv) showed that the concordance of molecular serotyping and antisera method was 93.3%, 94.9%, and 94.2%, respectively. Eighteen samples were identified by neither serological nor molecular methods and named unidentified. The D and C2 Salmonella serogroups were the most abundant. The match between molecular serotyping and serology methods for unidentified samples was 94.4%. In addition, 10.4% of the samples were not identified. The accuracy of detecting the flics gene by m-PCR and antigen H by the serological method significantly correlated with the contingency table test (P < 0.01). In the Receiver-Operating Characteristic (ROC) diagram, the sensitivity and specificity of the m-PCR were 92.3% and 95.2%, respectively, compared to the serological method (as gold standard). In evaluating H antigenic subgroups (including Ha, Hb, Hc, and Hj), the same results were obtained with both m-PCR and serological methods for a matching rate of 100%. Therefore, the relationship between methods was significant in the contingency table test (P < 0.01). Figure 6 shows the m-PCR product band to identify the subgroup H1 Salmonella genus (Ha, Hb, Hd, Hj).  Barrow and Neto (2011) invA

Discussion
Salmonella is a life-threatening foodborne zoonotic pathogen with more than 2500 serotypes. Over 95% of the strains cause infections in humans and animals to belong to serogroups A to D (Diep et al. 2019). Identification of Salmonella is necessary for the prevention, surveillance, and control of foodborne diseases. Therefore, there is a need for rapid detection, identification of sources, control of outbreaks, and identification of emerging serotypes of Salmonella. In this study, traditional (culture and serology) and molecular methods were used to detect Salmonella isolates from poultry farms in Iran. Serogroups and serovars were compared to determine the best fast and valid method. In this study, 7.5% (173/2300) of the isolates were identified as Salmonella by culturing and 7.2% (166/2300) were identified by PCR (invA). The current study exhibited a lower prevalence of Salmonella than broiler poultry farms in Bangladesh where prevalence ranged from 23 -38% (35/100; 36/123; 106/503) (Alam et al. 2020). A longitudinal Salmonella surveillance study was conducted in raw chicken meat in Mexico on 1160 samples collected between 2016 and 2018 (Regalado-Pineda et al. 2020). The study revealed a significantly higher prevalence (P < 0.0001) of S. enterica in supermarkets (27.2%, 158/580) than in wet markets (9.0%, 52/580). The prevalence of S. enterica was observed in other regions of the world, it included Venezuela, the USA, Canada, Wales, Australia, Brazil, Belgium, China, Columbia, Ecuador, Portugal, and Spain, where infection levels ranged between 9.5 and 65% (Regalado-Pineda et al. 2020). The lower prevalence of Salmonella observed in the current study could be attributed to the sample size (Persoons et al. 2011), where larger samples were compared in previous studies. The sampling sources could be another factor (Taylor et al. 2018) as the previous studies included various sampling locations and sources such as cloacal swabs, litter, chicken meat, feed. In comparison, the current study only included fecal samples from poultry farms. In addition, the geographical locations of the studies could be another factor that influences the current findings (Shah et al. 2011).
Target genes used in our study were previously validated in several studies using PCR and m-PCR assays to detect Salmonella serogroup A-E (Farahani et al. 2018). The present study implemented m-PCR of 878-897 gene to identify S. enterica serovar Infantis following previous studies where m-PCR was used on the same gene to identify S. enterica serovar Infantis in spiked chicken feces and meat samples. Furthermore, the current study used STM4492 and fliC genes to identify S. enterica serovar Typhimurium and S. enterica serovar Choleraesuis, respectively. These findings are consistent with previous reports where STM4492 was used as a target marker gene to identify S. enterica serovar Typhimurium and exhibited high specificity and differentiation between the Salmonella serovars (McCarthy et al. 2009). Studies have shown that the STM4492 gene discriminated S. enterica serovar Typhimurium from S. enterica serovar Enteritidis in broiler and chicken meat samples (Paião et al. 2013;Saeki et al. 2013). The fliC gene is the other target gene for S. enterica serovar Typhimurium and S. enterica serovar Choleraesuis detection that encodes the phase 1 flagellin protein (H1) which is the most frequently used gene to differentiate Typhimurium serovar from the others (Gürbüz et al. 2018). Researchers at Konya (Turkey) used the fliC gene to isolate S. enterica serovar Typhimurium from chicken meat and giblets   Ahmed et al. (2009) in Egypt, the had gene was used to detect multidrug resistance in Salmonella spp. isolated from diarrheic calves. The P1-P2 primer pair targeted the oriC gene as an internal control in all m-PCR reactions.
The present study found the highest prevalence of S. enterica serovar Enteritidis in fecal samples from poultry in Iran. The lowest prevalence was associated with S. Heidelberg, indicating that live poultry was the source of S. enterica serovar Enteritidis for contamination of raw chicken meat in the primary part of the chain production. The motile Salmonella spp. are mainly associated with food products, and they are the significant causes of salmonellosis in humans (Whiley and Ross 2015). Approximately 7.5% of Salmonella isolates were confirmed as S. enterica serovar Typhimurium in our study. Similar results were found by Barua et al. (2013), where 11% of commercial broiler chicken farm isolates were motile Salmonella and Islam et al. (2016) in Bangladesh, where 15.91% of isolates were S. enterica serovar Typhimurium. Alam et al. (2020) showed that 85.7% of the isolates from Bangladesh were confirmed as motile Salmonella, which is higher than our results. In another study conducted from 154 commercial poultry layer farms in the Southern part of India, a total of 1215 samples containing poultry meat, tissues, egg, and environmental samples were screened for nontyphoidal Salmonella (NTS) serovars. Multiplex-PCR, allele-specific PCR, enterobacterial repetitive intergenic consensus (ERIC) PCR, and pulse field gel electrophoresis (PFGE) revealed 21/1215 (1.73%) samples positive for NTS (Saravanan et al. 2015). Similarly, during disease outbreaks (40-80% mortality) in poultry farms in Lagos, Ogun and Oyo states, Nigeria, PCR and serotyping conducted on chicken organ samples collected at postmortem examinations identified motile Salmonella serotypes primarily represented by S. enterica serovar Zega (34.14%), S. enterica serovar Kentucky (24.32%), S. enterica serovar Herston (16.22%), S. enterica serovar Nima (10.81%), S. enterica serovar Colindale (2.70%), S. enterica serovar Telelkebir (8.11%) and S. enterica serovar Tshiongwe (2.70%) (Mshelbwala et al. 2017).
Bacterial culture-based techniques are time-consuming, laborious, and have a lower discriminatory capacity. Simultaneously, molecular methods such as m-PCR are crucial in detecting, typing, speciating, and classifying Salmonella at the genus level, serogroups, and serovars. The m-PCR assay is a sensitive, reliable, specific, and highly effective diagnostic test for the simultaneous identification of Salmonella and its serogroups and serovars. However, the cultivation-based PCR-dependent technique has certain limitations, such as the less abundant microbes could not be grown easily, and uncultivable microorganisms are not retrieved, resulting in the wrong interpretation of the result. Conversely, the cultivation-independent PCR-dependent technique is more reliable as it involves the PCR of the metagenome directly retrieved from the environment, devoid of any prior cultivation (Ghosh 2015). This system could significantly reduce reliance on tedious conventional serotyping. However, the main issues to be considered are the cost scale-up of these advanced methods and the regulatory necessities. Although the present results are preliminary, the m-PCR assay could offer a valuable alternative to traditional typing methods (culture and serological) to identify and differentiate the most Salmonella spp. in diverse samples. Further investigations should embark on the whole genome sequencing, functional genomics, extraction, and purification of the bioactive compounds from these isolates, which could contribute to understanding the mechanism of infections.