A qPCR assay for Bordetella pertussis cells that enumerates both live and dead bacteria

Bordetella pertussis is the causative agent of whooping cough, commonly referred to as pertussis. Although the incidence of pertussis was reduced through vaccination, during the last thirty years it has returned to high levels in a number of countries. This resurgence has been linked to the switch from the use of whole-cell to acellular vaccines. Protection afforded by acellular vaccines appears to be short-lived compared to that afforded by whole cell vaccines. In order to inform future vaccine improvement by identifying immune correlates of protection, a human challenge model of B. pertussis colonisation has been developed. Accurate measurement of colonisation status in this model has required development of a qPCR-based assay to enumerate B. pertussis in samples that distinguishes between viable and dead bacteria. Here we report the development of this assay and its performance in the quantification of B. pertussis from human challenge model samples. This assay has future utility in diagnostic labs and in research where a quantitative measure of both B. pertussis number and viability is required.


Introduction
Whooping cough, or pertussis, is a highly contagious respiratory tract infection of humans caused by the gram-negative coccobacillus Bordetella pertussis. Clinical manifestations of pertussis depend on age and immune status of the host and include a low-grade fever, cyanosis, and paroxysmal cough accompanied by a high-pitched "whoop" [1]. Infants aged less than 1 year old present the highest incidence of pertussis and are also at the greatest risk of severe disease and death [2].
The introduction of vaccination in the early 1950s significantly reduced the incidence of pertussis in developed nations, however the number of reports of pertussis has been Tohama 1 [29]. B. pertussis strains UK48 and UK71 can both be identified with European Nucleotide Archive Accession numbers: ERS176875 and ERS227772, respectively. B. pertussis strain B204 (B1878) and B184 (B2973) were both derived from the Netherlands and can be located with the following NCBI Genbank accession numbers: NZ_CSNV00000000 and NZ_CSRZ00000000, respectively. All strains were cultured on charcoal agar (ThermoFisher Scientific, Oxoid™, Basingstoke, UK) at 37˚C for 3 days for routine culture.

The preparation of heat-killed bacterial cell suspensions
Plate-grown B1917 were resuspended in PBS (ThermoFisher Scientific, Oxoid™, Basingstoke, UK) to an OD 600 = 1.0 (approximately 10 9 cfu/ml). To optimise heat killing, 1 ml aliquots were heat-killed at 80˚C for 1, 3 and 6 minutes in a pre-heated heat block. Aliquots were placed on ice immediately after incubation. Bacterial death was confirmed by the absence of growth after streaking 10 μl of suspension onto charcoal agar plates and incubating at 37˚C for 5 days. To ascertain the integrity of heat-killed cells, samples were subjected to flow cytometry (FACS-CantoII, BD UK Ltd, Wokingham, U.K.). A detergent-lysed sample acted as a positive control for lysis and a sample containing live cells was a positive control for cell integrity.

Genomic DNA Isolation
Genomic DNA (gDNA) was isolated using the GenElute Bacteria Genomic DNA Kit (Sigma-Aldrich, Dorset, UK) according to the manufacturer's instructions and eluted with 200 μl of elution buffer. gDNA was purified from THP-1 cells and human challenge model samples using the QIAmp DNA mini and blood extraction kit (QIAgen, Manchester, UK) as per the manufacturer's protocol. gDNA was quantified using a Qubit 1.0 fluorometer (Invitrogen, Loughborough, UK) according to the manufacturer's instructions.

Establishing linerarity using standard curves
A 10-fold serial dilution of gDNA in nuclease-free water was assayed for qPCR as described previously. A standard curve was automatically generated using StepOnePlus™ Software v2.3 to establish the linearity of the assay and to allow for the absolute quantification of unknowns.

Calculating copy number from Ct values/ DNA concentration
The genome copy number equivalent to the amount of template in a qPCR reaction was calculated using the formula: copy number = 50 x (amount of template in ng � 6.022x10 23

Protocol deposited on protocols.io
The PMA-qPCR method has been deposited on protocols.io and can be accessed here: http:// dx.doi.org/10.17504/protocols.io.bc5niy5e.

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR cells or without THP-1 cells and were then treated with the selected PMA treatment. Non PMA-treated samples were run in parallel.
To determine if eukaryotic cells interfered with the action of PMA on dead bacterial cells, 100,000 live THP-1 cells were combined with different ratios of viable B. pertussis cells and heat-killed B. pertussis cells (final bacterial concentration was 10 6 cfu/ml) in a clear Eppendorf tube, total volume 200 μl. These samples were then subjected to the selected PMA treatment. A non-PMA treated control was included. gDNA was extracted from each sample and used for qPCR.

Statistical analysis
Unpaired T tests, corrected for multiple comparisons, and two-way ANOVA using the Holm-Sidak method was carried out using Prism 8 for macOS Version 8.2.1 to evaluate statistical significance. One-way ANOVA and Dunnett's multiple comparisons test, with a single pooled variance was also used. A p value of <0.05 was defined as statistically significant and is indicated by asterisks.

Ethics
Samples from volunteers participating in the human challenge model study were obtained in accordance with the provisions of the Declaration of Helsinki (1996) and the International Conference on Harmonization Guidelines for Good Clinical Practice. This study is registered with ClinicalTrials.gov: NCT03751514, was reviewed and approved by the South Central-Oxford A Research Ethics Committee (REC reference: 17/SC/0006, 24 February 2017) and the UK Health Research Authority (IRAS project ID: 219496, 1 March 2017). The protocol has been published previously and details written and oral consent received from human participants [23]. It can be found on www.periscope-project.eu.

qPCR provides a lower limit of detection of 2 B. pertussis cells
IS481 is often used as the target for qPCR detection of B. pertussis as it is present at~250 copies per cell in B. pertussis, providing great sensitivity. To develop a PMA-qPCR assay, the sensitivity of qPCR for detection of B. pertussis was tested over a range of template gDNA concentrations. A linear relationship between Ct value and template concentration was observed over the range of 2 to approximately 2.42x10 6 B1917 cells for qPCR (Fig 1). Ct values greater than 35 were considered to be a negative reaction. Thus, the assay is able to detect B. pertussis gDNA equivalent to very few bacterial cells and is linear over a wide range of B. pertussis concentrations.

Heat-killing B. pertussis at 80˚C for 6 minutes maintained the integrity of cells
The ability of PMA to inhibit PCR-amplification from dead B. pertussis was tested using heatkilled B. pertussis B1917. It was envisaged that clinical samples may contain dead, but intact, B. pertussis. Heat-killing may cause cell lysis which would not mimic intact dead cells. Thus, the integrity of cells following heat killing was assessed using flow cytometry. Incubation of B. pertussis suspensions at 80˚C for 6 minutes resulted in 100% killing, but with cells remaining intact and were the conditions used throughout (Fig 2).

Optimisation of PMA treatment
The effect of PMA concentration on inhibition of PCR amplification from dead B. pertussis B1917 was tested (Fig 3). Incubation of heat-killed cells with 50μM of PMA resulted in a

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR 97.42% reduction in PCR signal compared to that generated from untreated samples. Lower levels of PMA also resulted in very similar levels of inhibition (Fig 3).
The optimal conditions for photo-activation of PMA were determined. Incubation under dark conditions for 10 minutes followed by light activation for between 5 and 30 minutes resulted in greater than 99% reduction in PCR signal from dead cells compared to untreated controls. Five minutes of light activation following 10 minutes of darkness resulted in 99.64% reduction in detection of B. pertussis DNA (Fig 4).
From these optimisations, standard conditions of 50μM PMA and incubation in the dark for 10 minutes followed by light activation for 5 minutes were selected as minimal incubation times that achieved high levels of inhibition. Even though 20μM PMA inhibited PCR amplification from dead cells, 50μM PMA was selected as the concentration to use in the assay, as clinical samples will contain cells other than B. pertussis that may sequester PMA, requiring an excess for consistent inhibition of PCR signal from dead B. pertussis. These conditions were tested in four independent assays. An average of 94.15% reduction in PCR signal was observed compared to untreated controls (Fig 5).

The effect of exogenous cells on detection and PMA-mediated inhibition
Clinical samples are likely to contain cells other than B. pertussis, including eukaryotic cells that contain very large amounts of DNA compared to B. pertussis cells. Eukaryotic cells may  (Fig 6).
It was possible that the presence of other cells would interfere with the PMA-mediated inhibition of PCR signal from dead B. pertussis. Thus, the effect of heat-killed or live THP-1 cells on PMA-mediated inhibition of PCR amplification from heat-killed B. pertussis was tested. A 99.94% reduction in PCR signal was observed indicating that THP-1 cells did not prevent PMA-mediated inhibition of PCR signal from dead B. pertussis (Fig 7).
To test the assay's ability to distinguish between viable and dead B. pertussis, in the presence of other cells, a constant number of THP-1 cells were combined with different ratios of heatkilled and viable B. pertussis B1917 cells. The reduction in PCR signal was proportional to the amount of heat-killed cells in each suspension (Fig 8) demonstrating that the assay was able to distinguish viable from dead B. pertussis, even in the presence of human cells.
Collectively, these studies revealed that the THP-1 cells did not interfere with the PMAmediated inhibition of PCR signal from dead B. pertussis or prevent the accurate enumeration of viable B. pertussis cells.

Measuring the viability of B. pertussis during in vitro growth
During development of the assay, it was observed that PMA treatment of live B. pertussis suspensions used as controls consistently reduced the PCR signal compared to untreated samples. This suggested that B. pertussis colonies taken from plate grown cultures contains both live and dead bacteria. To investigate this, and to determine the proportion of live to dead B. pertussis in plate grown cultures over time, suspensions of cells were made of B. pertussis B1917 grown on plates for either 3, 4, 5 or 8 days. The suspensions were treated with PMA and qPCR performed. The percentage of PCR signal observed was compared to untreated controls, Fig 9. B. pertussis is relatively slow growing and many protocols for plate growth involve incubation for 72 hours to achieve visible colonies. However, at this point B. pertussis viability was only 89%. Interestingly, although colony size continued to increase between days 3 and 5, percentage viability decreased to 24%. Further incubation resulted in further loss in viability. Thus, when using plate grown B. pertussis in assays, suspensions will be a mixture of live and dead bacteria, and that enumeration of B. pertussis by plating serial dilutions of a suspension and counting the resulting CFU's will not be a measure of the total number of cells in the suspension.

Use of the assay to enumerate live and dead B. pertussis from human challenge model samples
The assay was developed in order to provide a method for monitoring the colonisation status of participants in a novel human challenge model of B. pertussis colonisation. During development of this model, a group of participants were inoculated with 10 5 CFU of B. pertussis B1917 and daily samples were taken over a 14-day period to monitor colonisation, including nasosorption fluids, pernasal swabs, throat swabs, and nasopharyngeal washes [21].

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR Here, samples from Day 9 post-challenge were tested by PMA-qPCR, and by culture, Fig  10. Samples were split and one portion was treated with PMA. B. pertussis were enumerated using qPCR from PMA-treated and untreated samples. Using culture, 3 out of 5 participants were determined to be colonised [21]. By PMA-qPCR, 4 out of 5 volunteers were deemed to carry viable B. pertussis, with detection from nasal washes, pernasal swabs, nasosorption and throat samples. Nasal washes from Day 11 samples also had detectable viable B. pertussis in 2 of the 5 volunteers using PMA-qPCR and in a third volunteer using culture. (Fig 10E). Pernasal samples from Day 11 revealed detectable B. pertussis in a single volunteer using culture, that was not detected using PMA-qPCR (Fig 10F). PMA-qPCR revealed samples contained both viable and dead B. pertussis, in approximately equal numbers. Interestingly, on Day 16 of sampling, two days after volunteers started azithromycin treatment to eradicate the infection, all but one volunteer was negative for detectable B. pertussis cells. In this volunteer, the PMA-qPCR assay was able to detect low levels of viable and dead B. pertussis, with a higher proportion of dead genomes detected compared to viable genomes, however these low levels of B. pertussis were undetectable by culture (Fig 10G). Nasosorptions from this cohort of volunteers were all culture-negative.

Confirming utility of PMA-qPCR assay by enumerating five additional strains of B. pertussis
To confirm the utility of this assay with other strains outside of the human challenge model, qPCR was performed on gDNA extracted from PMA and non-PMA treated suspensions from the following strains: BP536, UK48, UK71, B204 (B1878), B184 (B2973). B1917 was also enumerated as a control (Fig 11). These suspensions contained B. pertussis cells taken from threeday old plate cultures resuspended in PBS to an OD 600 = 1.0. The suspensions were plated onto agar to confirm the enumeration obtained by the PMA-qPCR assay. The recovery of viable B. pertussis cells from both PMA-qPCR and culture were comparable for all strains, confirming the utility of this assay as a reliable method for enumeration, however, there was a significant increase in the recovery of viable cells from B1917 and BP536 using PMA-qPCR compared to culture. DNA copy numbers were calculated using the mean genome size of 4.1 mb.

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR

Discussion
Ordinarily, the detection and quantification of viable B. pertussis is achieved through culture on laboratory agar. However, the relatively slow growth rate of B. pertussis means that the growth of countable colonies can take between 72-120 hours. The development of a human challenge model for B. pertussis as part of the PERISCOPE project requires that enumeration of viable B. pertussis be achieved in a much shorter time than this, in order to be able follow colonisation closely.
In addition, simple enumeration of viable bacteria within a sample doesn't provide the complete picture. In many scenarios, such as measuring bacterial load in an infection model, it is of great interest to know the total bacterial number as understanding the dynamics of bacterial growth that involves both cell division and cell death is very important. Thus, while traditional qPCR provides a faster detection method for B. pertussis than culture, the modification of a qPCR assay with the introduction of PMA treatment of samples reported here enables both fast detection of B. pertussis and the ability to distinguish viable from dead cells.
Here, we demonstrate that PMA inhibits PCR-mediated amplification from dead B. pertussis and that inhibition of signal from dead cells occurs even in the presence of high numbers of eukaryotic cells. This may be important for the detection of B. pertussis from complex samples that contain a mix of cell types as seen in the human challenge model. Samples obtained from

PLOS ONE
Enumerating viable B. pertussis using PMA-qPCR volunteers that were identified as positive for B. pertussis by culture, were also detected in our initial test of the PMA-qPCR assay. The same volunteers were identified as being negative for B. pertussis by both qPCR and culture, with the exception of a single sample that had low levels of B. pertussis identified only by qPCR. This result demonstrates the high sensitivity of this assay to detect very low levels of viable and dead B. pertussis. Further optimisation studies to determine the exact point in which there is loss in sensitivity when amplifying fewer than 10 3 B. pertussis cells in the presence of THP1 cells would further support the results obtained from the human challenge model. Interestingly, PMA-qPCR detected approximately equal numbers of viable and dead B. pertussis, demonstrating its use to enumerate total bacteria rather than only viable cells. The full results of the human challenge model are published elsewhere [21]. Here we demonstrate that the PMA-qPCR assay allowed for a determination of colonisation status within hours of obtaining the samples compared to days when using culture.
To confirm that the assay can be used with strains other than B1917, we tested five additional B. pertussis strains. Approximately 250 copies of IS481 were found in all isolates of B. pertussis, however, the exact number of copies differs amongst strains within a narrow range https://doi.org/10.1371/journal.pone.0232334.g011 [30]. The number of copies range from 236-272 among the closed genome sequences available for B. pertussis. Thus, this will create some error when performing absolute quantification of strains for which the copy number is not known, but this error is not large (<10%).
The use of IS481 as a target means that there is the chance of cross-reactivity of IS481 with B. holmseii and B. bronchiseptica, although B. holmseii is rarely recovered from nasopharyngeal samples [32]. However, here, this assay was specifically designed to support the human challenge model, therefore using a single qPCR target of IS481 to detect known amounts of B. pertussis B1917 that has been administered to volunteers was appropriate. To extend the use of this assay and to increase specificity, species-specific targets, which are commonly used in many diagnostic labs, should be considered. [30]. These changes will reduce false positives and result in a more sensitive and specific assay for the accurate diagnosis of B. pertussis, as well as help rule out coinfections or pertussis symptoms caused by other Bordetella spp.
The utility of the PMA-qPCR assay has been shown for the human challenge model, but has wider uses. For example, in diagnostic laboratories, where ascertaining if B. pertussis is viable or dead will facilitate whether to pursue culture as a means to obtaining a live culture for characterisation. It is also of use in a range of research and industrial settings enabling investigation of the dynamics of B. pertussis growth by determining both cell division and cell death.