Introduction

Campylobacter has been classified as a leading cause of bacterial gastroenteritis and growth stunting in children living in poverty. Campylobacter jejuni and Campylobacter coli are the predominant Campylobacter species associated with gastroenteritis, both in pediatric populations and in cases of travellers’ diarrhea [1–3]. Additionally, C. jejuni, C. coli and other Campylobacter spp. are detected from infants with asymptomatic Campylobacter infections in low- and middle-income countries (LMICs) where Campylobacter infections are endemic [4,5] and are associated with growth deficits even when they are asymptomatic [6]. The high prevalence of this pathogen globally has been supported by large scale population-based studies that have used nucleic acid-based diagnostics to attribute cases of diarrhoea to this pathogen [7–11].

The performance of culture independent diagnostic technologies (CIDTs), such as quantitative polymerase chain reaction (qPCR), also known as real-time PCR, is particularly important for the detection of Campylobacter spp. in LMICs. Campylobacter spp. are fastidious microorganisms that are challenging to culture and isolate requiring a variety of specific atmospheric and nutritional requirements that are often not easily obtained in LMICs [12]. Moreover, nucleic acid based diagnostic tests have been shown to detect Campylobacter spp. at levels lower than those required for culture based isolation [13]. The target for Campylobacter genus detection is usually the 16S rRNA gene [4,5,13,14] or the chaperonin protein cpn60 [1,15,16] that has been used to detect both common thermotolerant Campylobacters, such as C. jejuni and C. coli and less common thermotolerant and non-thermotolerant Campylobacter species, such as C. upsaliensis, C. concisus and “Candidatus Campylobacter infans.” A variety of genetic targets have been utilized to detect C. jejuni and C. coli. Assay targets may be specific for either species, mapA and hipO for C. jejuni, ceuE and glyA for C. coli [17,18]. Alternatively, a cadF primer-probe set was designed to detect both C. jejuni and C. coli [19]. In large scale studies, a combination of the chaperonin protein cpn60 and cadF primer-probe sets have been used in TaqMan Array cards for simultaneously detecting Campylobacter spp. and C. jejuni/ C. coli, respectively [8,20]. For samples that tested positive for Campylobacter spp. and C. jejuni/ C. coli, the samples were scored as possessing C. jejuni/ C. coli and those that tested positive for Campylobacter spp. and negative for C. jejuni/ C. coli were scored as non-C. jejuni/ C. coli [12,21].

Several studies examining fecal samples from infants in LMICs detected non-C. jejuni/ C. coli Campylobacter at levels as high as 70% of Campylobacter infections, leaving an unacceptably large group of Campylobacter species undesignated at the species level [12,21]. Recently, shotgun metagenome sequencing of stool samples that tested qPCR positive for Campylobacter spp. and negative for C. jejuni/ C. coli (cadF) demonstrated that nearly 64% of the samples possessed either C. jejuni or C. coli, and sometimes both [5]. This major underestimation of the overall prevalence of C. jejuni and C. coli in fecal samples suggested that an improved primer-probe set for the detection of C. jejuni/ C. coli was needed.

A community-based cohort study near Iquitos, Loreto, Peru provided fecal samples from asymptomatic and diarrheal samples to test an alternative rpsKD-based qPCR assay to detect both C. jejuni and C. coli. Here we report the development and assessment of this assay that would have an improved sensitivity relative to cadF for the detection of C. jejuni and C. coli.

Materials and methods

Validation and evaluation of rpsK/rpsD primer-probe assay

In silico validation.

The selected sequences for the rpsK/rpsD primers and probe (Table 1) were compared against 9,000 C. jejuni and C. coli genomes in the pubMLST database (https://pubmlst.org/organisms/campylobacter-jejunicoli) using BLASTN. To ensure that other closely related species are not amplified, we examined by BLASTN against 300 C. upsaliensis and 300 C. lari genomes from the same database.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Primer and probes used to detect Campylobacter spp. and Campylobacter jejuni / Campylobacter coli.

https://doi.org/10.1371/journal.pntd.0012018.t001

In vitro evaluation.

The performance of the new rpsK/rpsD primer-probe assay was evaluated with a set of archived (-70°C) stool samples from children under 24 months of age. All samples were processed using both assay [A] which consisted of the Taqman based multiplex assay to detect Campylobacter spp. (16S rRNA gene) and Campylobacter jejuni / Campylobacter coli using the cadF gene and assay [B] which consisted of the Taqman based multiplex assay to detect Campylobacter spp. (16S rRNA gene) and C. jejuni / C. coli using the newly validated rpsK/rpsD gene fragment. Primer and probe sequences of both assays is presented in Table 1.

Fecal DNA was extracted from 0.2 grams of feces using the QIAamp DNA Stool Mini Kit (Qiagen, Carlsbald, CA), according to the manufacturer’s instructions. A negative control consisting of RNA and DNA free water was used for each extraction set. The final assay consisted of a 25 μL final reaction mixture with 12.5 μL of Environmental Master Mix (2X) (Applied Biosystems, Foster City, CA), forward and reverse primers (0.2 μM), probes (0.1 μM), 1 μL of DNA template and RNase and DNase free water (Ambion, Thermo Fisher Scientific, Waltham, MA, USA). Both qPCR assays were performed on a QuantStudio 7 Flex (Applied Biosystems, Foster City, CA) using the following cycling conditions: 95°C for 10 minutes followed by 45 cycles of 95°C for 15 seconds and 60°C for 1 minutes. Custom manufactured double-stranded synthetic DNA fragments (gBlocks, Integrated DNA Technologies, Coralville, IA, USA) were used as positive controls. Negative template controls (RNase and DNase free water) were included in each amplification reaction.

Standard curves for each marker were prepared using 10-fold serial dilutions of synthetic positive controls (6.0 x 104–6.0 x 10⁰) gene copies/μL. For both assays, a cut-off cycle threshold of 38 was used to determine positivity.

Shotgun metagenomic DNA sequencing.

Fecal samples that were positive for Campylobacter spp. (16S rRNA gene), but negative for C. jejuni and C. coli (either with the cadF primer set or with the rpsK/rpsD primer set) underwent shotgun metagenomic sequencing. DNA sequencing libraries were prepared using Illumina DNA Prep Tagmentation kit (Illumina, San Diego, CA), following the manufacturer’s instructions except for the following changes. The insert size was increased to a range of ~375–1100 bp by reducing the 1st and 2nd volumes of Sample Purification Beads to 40 μl and 11 μl, respectively. This modification resulted in larger inserts compared to the mostly below 300 bp inserts obtained using the manufacturer’s protocol. The final elution volume of the libraries was in 10 μl of Illumina resuspension buffer. Illumina-DNA/RNA UD Indexes Plate A, B, C and D dual index adapters were ordered from Integrated DNA Technologies (Coralville, IA) and used at 1 μM final concentration. Instead of pooling equal volumes, individual libraries were quantified using the KAPA Library Quantification Kit (Roche), since we found qPCR to be a more accurate quantification than using equal volume. Libraries were quantified in 10 μl volume reactions and 90-s annealing/extension PCR, and then pooled and normalized to 4nM. Pooled libraries were re-quantified by ddPCR on a QX200 system (Bio-Rad, Hercules, CA), using the Illumina TruSeq ddPCR Library Quantification Kit and following the manufacturer’s protocols. Libraries were sequenced using a MiSeq Reagent Kit v2 (500-cycles) on a MiSeq instrument (Illumina) at 16 pM, following the manufacturer’s protocols. Average sequence read lengths for each sample was 200 nucleotides. Genomes were assembled using SPAdes assembler 3.15.5 within Geneious Prime 2023.1.2 (Biomatters, Ltd., Auckland, New Zealand). Short read sequence data for all the samples are available at NCBI’s Sequence Read Archive (SRA) and are associated with BioProject Accession PRJNA834762.

Results

Evaluation of cohort samples

A total of 242 human fecal samples were assayed for both Campylobacter spp. and C. jejuni/C. coli by qPCR. In all assays, presence of Campylobacter spp. was determined using a 16S rRNA primer/probe set. The presence of C. jejuni and/or C. coli was determined using C. jejuni/C. coli cadF primer-probe assay [19] or rpsK/rpsD primer-probe assay designed in this study. Table 2A shows results obtained after running all 242 samples with Campylobacter 16S rRNA gene and the cadF gene primer-probe assay. Campylobacter spp. was detected in 19.4% (47/242) of samples, and C. jejuni/ C. coli in 6.6% (16/242) of samples using the cadF assay. Table 2B shows results obtained by running the same samples with the Campylobacter 16S rRNA primer-probe assay and rpsK/rpsD gene primer-probe assay. In this case, Campylobacter spp. was detected in 19.0% (46/242) of samples, and C. jejuni and/or C. coli in 11.2% (27/242) of samples. In other words, the cadF assay identified 34.0% (16/47) of Campylobacter spp. positive samples as C. jejuni/ C. coli, whereas the rpsK/rpsD assay identified 58.7% (27/46) of Campylobacter spp. positive samples as C. jejuni/ C. coli, increasing the attribution of Campylobacter associated diarrheal to C. jejuni and C. coli by over 20%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Contingency Tables of Campylobacter spp (16S rRNA), and Campylobacter jejuni/ Campylobacter coli using the cadF primer probe set and the rpsK/rpsD primer probe set.

Using the cadF primers, 66% (31/47) Campylobacter infections would be assigned to “other Campylobacter” whereas only 41.3% (19/46) Campylobacter infections would be assigned to this category using newly designed rpsK/rpsD primer set. No specimens that were cadF positive were negative using rpsK/rpsD primers.

https://doi.org/10.1371/journal.pntd.0012018.t002

When comparing the performance of the cadF primer and the rpsK/rpsD primer in a single contingency table (Table 2C), we observe that 40.7% (11/27) of rpsK/rpsD positive samples were classified as negative with the cadF primer, while none of the cadF positive samples were classified as negative by the rpsK/rpsD primer.

Assessment of qPCR results

To assess the detection accuracy of the two C. jejuni/ C. coli qPCR assays, we performed shotgun metagenomic sequencing of stool samples. Of the 31 samples that were Campylobacter spp. positive for the 16S rRNA gene target and cadF negative, shotgun metagenomic DNA sequencing identified Campylobacter jejuni and/or Campylobacter coli reads in 10 samples, thus demonstrating a false negativity of 33.3% (10/31) of cadF in the identification of C. jejuni and C. coli in patient samples (Table 3). In respect to species of Campylobacter other than C. jejuni and C. coli, C. concisus was identified in 8 samples (in 5 cases as the only Campylobacter species identified, and once each with C. coli, C. hominis, or C. concisus, respectively). C. hominis was identified only in one sample, C. infans in four samples (once with C. concisus, once with C. upsaliensis and once as the only Campylobacter species identified) and C. upsaliensis in a single sample. Two samples had reads that matched to more than one Campylobacter spp., and 9 samples had no reads that match to any Campylobacter spp. Of these 9 samples, Ct values for 16S rRNA primer-probe set were higher than 37.0. Finally, there were not enough reads for an appropriate analysis to be performed in three samples.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Number of reads detected for Campylobacter jejuni, C. coli, C. upsaliensis, and C. helveticus, C. concisus and C. infans in human fecal samples, and qPCR status.

Samples highlighted in green show highlight 8 of 11 samples that were negative using cadF primers, positive with new rpsK/rpsD primers, and confirmed as having C jejuni or C. coli infection by metagenomic analysis.

https://doi.org/10.1371/journal.pntd.0012018.t003

Of the 19 samples that were Campylobacter 16S rRNA positive and rpsK/rpsD negative metagenomic analysis identified C. jejuni and C. coli in only two samples for a false negativity of 11% (2/19). These two samples had very few reads for C. jejuni and C. coli, and both had a Ct value over 34 in the rpsK/rpsD qPCR assay and 16S rRNA Ct values of 37.3 and 35.4. Finally, of the 11 samples that were cadF negative, yet rpsK/rpsD positive, C. jejuni reads and/or C. coli were identified by the rpsK/rpsD primer set in 8 of the 11 samples. Of the 3 remaining samples, the Ct obtained by the rpsK/rpsD primer set was higher than 36.5. Details of the total number of reads and reads matched to specific Campylobacter spp. species are shown in Table 3.

Discussion

Molecular diagnostics has greatly advanced our understanding of the epidemiology of enteric diseases. However, for certain species, such as Campylobacter, the underrepresentation of whole genomes or extensive genomic data from different geographic areas impairs the development of effective primers and probes. After demonstrating the inadequate performance of the cadF assay [5], we developed a new qPCR assay for the detection C. jejuni/ C. coli based on rpsK/rpsD that reduced the false negative detection results observed with a cadF-based qPCR assay. Overall, we detected Campylobacter spp. in 19% (47/242) of fecal samples from infants under one year that are enrolled in a cohort study in Amazonian Peru. Among these 47 samples the cadF qPCR assay detected C. jejuni/ C. coli in only 16 samples while the rpsK/rpsD qPCR assay detected C. jejuni/ C. coli in 27, a 23% (11/47) increase in detection of attribution of Campylobacter species to C. jejuni/ C. coli. Furthermore, we validated C. jejuni and/ or C. coli in 8/11 of the samples with discordant cadF and rpsK/rpsD results using shotgun metagenomic sequencing. The three samples that lack C. jejuni and/or C. coli sequencing reads are not necessarily indicative of the rpsK/rpsD qPCR providing false negative results. Shotgun metagenomic sequencing does not include any target amplification steps and positive results not only depend on the presence of C. jejuni and/or C. coli but also the levels of other bacterial species in the fecal microbiome that essentially saturate the available reads with non-informative data. Adding a pre-enrichment amplification of broadly cross-reactive primers that amplify across the highly diverse Campylobacter genus may in the future address this technical issue. Here, the shotgun metagenomic sequencing identified the presence of Campylobacter spp. other than C. jejuni/ C. coli including C. concisus and “Candidatus Campylobacter infans” at a prevalence high enough to serve as a reminder that all Campylobacter species from clinical specimens should not be assumed to be C. jejuni or C. coli. In this study, even with improved primers for the detection of C. coli/C. jejuni we found that over 40% of the Campylobacter identified by the 16S RNA gene target were not clearly assigned to C. jejuni/ C. coli by qPCR. This remains an important diagnostic gap. The prevalence of Campylobacter genus positivity was 19% in this population, but only 11.1% were positive to C. jejuni/ C. coli with the newly developed improved primer-probe set described here.

The clinical significance of other Campylobacter species, including ’Candidatus Campylobacter infans’, Campylobacter upsaliensis, and Campylobacter concisus, in relation to the burden of diarrhea in early childhood and its subsequent consequences, requires further investigation through large-scale population-based studies. The development of a multiplex qPCR primer-probe assay capable of detecting other Campylobacter species would be beneficial. The primers described in this article are intended to improve the targeted detection of C. jejuni and C. coli, which are commonly recognized as the most medically important species of Campylobacter present globally. The major limitation of this assay and molecular diagnostics in general is that the diagnostic capacity of hospitals and other health care settings in low and middle-income countries limits the utilization of this primer-probe assay in clinical settings in areas where they are most needed. However, improved diagnostics in the research setting and in large epidemiologic studies still allows improved molecular diagnostics to inform disease burden estimates and prioritize the development of disease specific interventions.

We have previously shown that the detection of C. jejuni/ C. coli using the single gene target cadF is not optimal [4,5]. Specifically, we demonstrated that within fecal samples that were positive for Campylobacter spp. but negative for C. jejuni/ C. coli using the cadF gene target, over 50% of the samples possessed C. jejuni and C. coli sequence reads by shotgun metagenomic sequencing [4,5]. Interestingly, the number of Campylobacter sequencing reads in this current study was lower than the previous studies. In this study, the number of reads range from 2 to 57 while in the previous studies [4,5], the number of reads ranged from 2 to >7000 for C. jejuni/ C. coli. Similarly for non-C. jejuni/C. coli reads, the difference was even wider with reads ranging from 2 to 164 in this study compared to a range of 2 to approximately 70,000 reads [4].

There are a number of reasons that the current samples provided fewer Campylobacter reads. First, the samples from the previous study were all from infants that had medically attended diarrhea while the stools in this in this study were from symptomatic and asymptomatic infants in a cohort study. Second, the age of the infants was distinct, with the previous studies under 2 versus under 1 year old in this study. Finally on our previous study we utilized a different sequencing protocol using KAPA library preparation. This method creates libraries through mechanical shearing, creating greater diversity of reads regarding DNA GC content. In this study, we used Illumina library kits that rely on enzymatic cleavage, which give lower coverage for DNA with low GC content. This led to the hypothesis that the Illumina library kit method would reduce the number of reads from low GC content organisms, such as Campylobacters, during microbiome library construction. However, we repeated the extraction method using KAPA library preparation, obtaining similar read values. and thus, proving similar performance of both library preparation methods.

Although other qPCR assays have been developed to detect C. jejuni and C. coli separately such as hipO for C. jejuni and 23S rRNA for C. coli [24,25], these targets have failed to demonstrate high reproducibility [26]. For this reason, the cadF gene target, which detects both C. jejuni and C. coli was adopted as a more relevant target, as both species produce similar clinical syndromes. There are important implications in the underperformance of the most common target for the detection of the two most medically important species of Campylobacter. Global estimates of disease, particularly those of low-income countries, have relied on cadF diagnostics for causal attribution, and thus disease burdens have been systematically underestimated for C. jejuni and C. coli one of the principal etiologies of bacterial enteritis globally and a vaccine target [8,15,20]. Nucleic acid diagnostics improves the sensitivity of Campylobacter diagnostics compared to culture dependent methods. This work demonstrates improved performance of the rpsK/rpsD target for the identification of C. jejuni and C. coli and demonstrates the utility of metagenomics in validating the assay performance as a useful and practical strategy in assessing the performance of assays of fastidious organisms such as Campylobacter.