A 16S ribosomal RNA TaqMan real-time reverse transcription PCR for specific detection of Salmonella in blood

Background : The Gram -negative bacterium Salmonella enterica is an important human pathogen causing a huge public health burden worldwide. Reference Salmonella infections (BSI) in patients is based on in vitro blood culture followed by biochemical and serotype identification. We have developed a TaqMan real-time reverse transcription PCR (RT-PCR) assay for the specific detection of Salmonella 16S rRNA molecules. The specificity was confirmed on a bacterial test panel grown in 5 ml BBL TM culture medium broth, comprising of six Salmonella enterica serovars (Typhi, Paratyphi A, Typhimurium, Enteritidis, Heidelberg and Weltevreden) and 10 non-Salmonella bacterial species that are known to cause human BSI. Results : The limit of detection (LOD) of the assay in serial dilutions of purified Salmonella enterica serovar Typhimurium RNA was 40 pg RNA per reaction, corresponding to ~2500 colony forming units (CFU). When applied directly in blood experimentally spiked with Salmonella Typhimurium, the assay showed an LOD of 10 5 CFU/ml which is 100x more sensitive than 16S rDNA detection using the same primers and probe set. In 10 ml whole blood spiked with Salmonella Typhimurium at 5 CFU/ml followed by incubation in an automatic blood culture instrument, the assay detected 16S rRNA after 10 hours incubation compared to 12 hours for 16S rDNA detection. Conclusion : Although prior in vitro incubation of blood samples is required, we here delivered the proof-of-concept of specific detection of Salmonella in clinical specimens by targeting its 16S rRNA molecule. detection, the same assay was applied but with replacement of the volume of reverse transcriptase and Ribosafe RNase inhibitor with RNase-free water and omitting the reverse transcription step (15 minutes at 45°C) from the PCR cycling. All RT-PCR reactions were performed on a Lightcycler 480 instrument (Roche Applied Science, Penzberg, Germany) and fluorescence data were acquisitioned during the extension step. A positive signal of detection was determined if amplification intersected with the threshold within 35 cycles (threshold cutoff criterium); and results are reported in Cp (cross points) values. Samples with Cp ³ 35 cycles or no Cp values recorded were

pg/µL and 10 pg/µL Salmonella Typhimurium SL1344 in triplicate (Additional file 1: Figure S1). When testing the 16 bacterial species included in the study (Table 1) at 1 ng/µl, the assay detected all Salmonella serovars and did not detect any non-Salmonella bacterial species (Additional file 2: Table   S1 and Figure 1). Mean Cp values ranged between 16,09 and 19,22 (Additional file 2: Table S1). In ten-fold serial dilutions of purified Salmonella Typhimurium RNA, the LOD was 10 pg/μl corresponding to 40 pg RNA per RT-PCR reaction (Figure 2). Considering that one bacterial cell contains 16 fg rRNA [18], the observed analytical sensitivity corresponded to ~2500 CFU per reaction.

Limit of detection in human blood
The LOD of the 16S rRNA assay was assessed in human blood spiked with known numbers of Salmonella Typhimurium SL1344 and compared with 16S rDNA detection. We observed an LOD of 10 5 CFU/ml blood for 16S rRNA detection (Cp value µ=32.69, s=0.55) and 10 7 CFU/ml blood for 16S rDNA detection (Cp value µ=31.91, s=1.12) (Additional file 2: Table S1 and Figure 3). We observed a difference of 15 cycles between 16S rRNA (Cp value µ=16.95, s=0.42) and 16S rDNA (Cp value µ=31.91, s=1.12) detections in the blood samples spiked with 10 7 Salmonella cells per ml blood.
There was no amplification detection in negative control samples (Additional file 2: Table S1). The RNA samples did not show DNA contamination when run with PCR without reverse transcription.

Limit and speed of detection in spiked blood cultures
The LOD of the 16S rRNA assay was compared with 16S rDNA detection in time-course measurements of blood culture bottles spiked with 5 CFU/ml blood in triplicate. Salmonella 16S rRNA was detected after 10 hours of incubation (Cp value µ=31.5, s=0.9) and 16S rDNA after 12 hours (Cp value µ=33.52, s=0.6). There was no amplification in the non-spiked blood culture bottle (Additional file 2: Table S1). In none of the blood culture bottles growth was detected by BacT/ALERT at 12 hours of incubation. Our experimental setting did not allow a continuous monitoring of blood culture growth by BacT/ALERT after 12 hours incubation.

Discussion
We report a novel TaqMan real-time reverse transcription PCR (RT-PCR) assay for detection of Salmonella 16S rRNA in biological samples. The assay shows high specificity by detecting all Salmonella serovars included in the study and remaining negative for non-Salmonella bacteria.
Despite the high analytical specificity and repeatability of the 16S rRNA assay, the sensitivity of 10 5 Salmonella cells per ml of blood is too low for diagnostic use without prior in vitro blood culture.
Bacterial concentrations in blood of a patient with Salmonella blood infection can be as low as 1 to 10 CFU per ml of blood. We assessed the required time to detection of blood culture prior to RT-PCR testing for a 10 ml blood sample with 5 CFU per ml blood. While 16S rDNA testing required 12 hours blood culture incubation, the 16S rRNA assay turned positive after 10 hours. BacT/ALERT detected growth in the spiked blood culture bottles >12 hours incubation, but the exact time could not be monitored due to the experimental setting. Our comparative analysis in grown blood cultures confirm our hypothesis that detection of 16S rRNA is more sensitive than detection of 16S rDNA.
Previous PCRs for Salmonella detection in blood have been published. They mainly target Salmonella serovar specific genes and show various values of LOD. For instance, three different studies have applied different single genes (flagellin, clyA, flagellar genes fliC-a and fliC-d) targeted PCR respectively to detect typhoidal serovars in blood (20-23). While the 16S rRNA gene has been previously used as a potential molecular target in Salmonella PCRs (24-26), this is the first assay that targets the highly abundant 16S rRNA molecules. 16S rRNA and its gene has, in addition to highly conserved regions, nine hypervariable regions across bacterial species and its function has likely not changed over a long period of time through the evolution [ The low sensitivity of our designed assay is probably due to the high annealing temperature (Ta) of the primers used in the RT-PCR assay to obtain the required specificity given that Salmonella specific detection is based on single nucleotide polymorphisms (SNPs) in the 16S rRNA sequences. The forward primer shows only 1 SNP with Klebsiella pneumoniae, Enterobacter cloacae and Pluralibacter gergoviae while the reverse primer target sequence has no SNPs in those 3 species. The use of locked nucleic acids (LNA) or peptide nucleic acids (PNA) may be considered to lower the Ta while maintaining the primers' melting temperatures and thus specificity [30,31]. RT-PCR-inhibitors in blood, such as immunoglobulins, haemoglobin and lactoferrin as well as the anticoagulants EDTA and heparin, are known to inhibit reverse transcription and PCR amplification. In addition, commercial blood culture bottles contain media, color indicators, and sometimes antimicrobial removal resins or beads, and may thus also contain RT-PCR inhibitory components. To our knowledge, we applied for the first time the Polaris enrichment technique to bacterial RNA extraction from blood samples and presence of RT-PCR inhibitors in the RNA samples cannot be excluded. Different methods for bacterial RNA extraction from blood and blood cultures should be compared for optimal yield and purity in order to select the most efficient method for 16S rRNA RT-PCR.
Despite the low sample size specially with spiked blood cultures, with our study we deliver the proof of concept of 16S rRNA testing for Salmonella detection in blood and confirm our hypothesis that 16S rRNA detection is more sensitive than 16S rDNA testing. Hereto, further evaluation of the developed 16S rRNA RT-PCR test on larger size of clinical blood samples is needed. The potential value of the 16S rRNA RT-PCR test in clinical practice is that it can detect bacteria in blood samples earlier than conventional blood culture by BacT/ALERT and allows direct identification of Salmonella enterica, omitting biochemical testing for species identification.
Patients frequently use antibiotics prior to blood culture sampling and a major advantage of molecular diagnostics is that cell-free bacterial DNA can be detected as proxy for a bacterial bloodstream infection. Given the relatively long half-life of rRNA after bacterial death [32, 33], we expect that rRNA is an equal marker for bacterial infection as DNA. The same RT-PCR platform could allow development of multiplex 16S rRNA assays detecting in parallel multiple bacterial species responsible of bacterial bloodstream infections. However, targeting 16S rRNA in multiplexed PCRs will require a 16S rRNA based resolution to the species level and this is not always possible for some bacterial species such as E. coli and Shigella. On the other hand, 16S metagenomics can simultaneously detect all bacteria in a given blood sample based on PCR amplification of the 16S ribosomal RNA gene followed by deep sequencing of the PCR amplicons and taxonomic labelling of the sequence reads at genus or species level. However, the technique currently presents various technical limitations such as limited taxonomic resolution and high background signals jeopardizing its implementation in routine patients' diagnosis [11]. Based on the proof-of-concept presented here, 16S rRNA amplicon deep sequencing may be more sensitive than conventional 16S metagenomics at DNA level. Therefore, direct sequencing of 16S rRNA could complement the 16S metagenomics tool set for bacterial identification and decrease false positive hits as RNA represents a marker of viable organisms [33, 34].

Conclusions
We developed a 16S rRNA TaqMan real-time RT-PCR assay for specific detection of Salmonella applicable on grown blood cultures. The assay detects all Salmonella serovars tested in this study and does not cross-react with non-Salmonella bacterial species but requires prior in vitro blood culture incubation to obtain the required sensitivity for clinical diagnosis.

Bacterial strains
Salmonella Typhimurium reference strain ATCC SL1344 was used in this study for assay optimization and 15 additional Salmonella and non-Salmonella bacterial strains were selected for in vitro evaluation of the specificity of the designed primers and probe (Table 1). All strains were grown in 5 ml BBL TM culture medium broth (Becton, Dickinson and Company, Sparks, Maryland, USA) at 37°C aerated at 200 rpm. Bacteria were harvested at the logarithmic phase (OD600 = 0.5) by centrifugation at 4000 rpm, 4˚C, for 20 minutes.  Table S2). 16S rRNA sequences of ≥1450bp and sequence alignment quality of ≥93% (ranked by SILVA database) were retained (Additional file 4: Supplemental information S1). The 16S rRNA sequences were imported into AlleleID and subjected to sequence alignment using ClustalW [36] for the design of Salmonella enterica primers and TaqMan MGB probe sets. The specificity of the top-five ranked sets of primers and probes was further evaluated in silico using Geneious v11.1 (Auckland, New Zealand) and BLAST against 16S rRNA gene sequences of Salmonella enterica and the most closely related Enterobacteriaceae (Additional file 3: Table S2). The primers and probe set with the highest in silico specificity for Salmonella enterica was selected and the sense primer was manually curated to avoid cross-amplification of Pluralibacter gergoviae (Additional file 5: Figure S2). The sequences of the selected and curated sense and anti-sense primers with the 5' FAM reporter and 3' Eclipse quencher labelled probe are presented in Table 2

RT-PCR assay performance
The efficiency and limit of detection (LOD) were determined as performance characteristics of the assay. Salmonella Typhimurium SL1344 RNA at 1 ng/µl, 100 pg/µl and 10 pg/µl were prepared and tested in RT-PCR to determine the efficiency of the primers and probe set. The correlation coefficient (R 2 ) and efficiency were calculated by linear regression of the mean Cp scored against log10 of its concentration in pg/µl. To determine the LOD of the assay, 10-fold serial dilutions of Salmonella Typhimurium SL1344 RNA ranging from 1ng/μl to 1fg/μl were prepared in triplicate and assayed. The LOD was defined as the lowest concentrations of RNA for which all replicates are positive. Throughout the study all experiments were conducted in triplicate to assess repeatability of the assay.

Spiked human blood samples
Salmonella Typhimurium SL1344 cells grown in 5 ml BBL TM culture medium broth to logarithmic phase (OD600=0.5) were pulled through a 0.45x13 mm needle (Terumo, Tokyo, Japan) and the OD was measured from which the corresponding bacterial cells/ml was calculated. 10-fold serial dilutions of bacterial cells were prepared in sterile 1xPBS ranging from 10 8 to 10 cells per ml. A 100 µl of each Salmonella suspension was then spiked in triplicates in 1 ml of healthy human whole blood with EDTA as anticoagulant, resulting in triplicated spiked 1 ml blood samples ranging from 10 7 to 1 bacterial cell(s) per ml blood. The exact number of spiked Salmonella in 1 ml blood was retrospectively determined by colony counting. The Salmonella suspensions at 10 3 , 10 2 and 10 cells per ml 1XPBS buffer were plated by spreading their respective 100 µl on Luria-Bertani (LB) agar plates and incubated overnight at 37°C, followed by colony counting on the next day.

Spiked and grown blood culture samples
Salmonella Typhimurium SL1344 cells grown on an LB agar plate were diluted in PBS to prepare a suspension at OD600=0.5, corresponding approximately to 10 8 CFU/ml, and diluted to 250 CFU/ml in 10 ml sterile physiological saline solution. From this suspension, 800 µl was added to 40 ml defibrinated horse whole blood (International Medical Products, Oudergem, Belgium) to obtain approximately 5 CFU/ml. Ten ml was added to BacT/ALERT adult blood culture bottle (bioMérieux, Marcy-l' Etoile, France) in triplicate and 10 ml horse blood with no Salmonella spiked was used as negative control. The BacT/ALERT bottles were incubated at 35°C for growth in 3D BacT/ALERT automated culture machine (bioMérieux) and at each hour, starting from the 2 nd till the 12 th hour of incubation, two times 1 ml blood culture broth were collected for RNA and DNA extraction.

Supplementary information
Additional file 1: Figure S1. Efficiency of the developed Salmonella 16S rRNA TaqMan real-time reverse transcription PCR.
Additional file 2: Additional file 4: Supplemental information S1. 16S rRNA gene sequences for Salmonella enterica serovars and non-Salmonella bacterial species used in this study, retrieved from the SILVA database.
Additional file 5: Figure S2. Location of the forward primer (Fprimer), probe (TaqMan) and reverse primer (Rprimer) in the alignment of 16S rRNA gene sequences of Salmonella enterica serovars and non-Salmonella bacterial genera and species. Figure 1 Specificity of the Salmonella 16S rRNA TaqMan RT-PCR assay on 6 Salmonella serovars and 10 non-Salmonella bacterial species (Table 1)

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.