Improved molecular detection of Babesia infections in animals using a novel quantitative real-time PCR diagnostic assay targeting mitochondrial DNA

Background Babesiosis is a protozoal, tick transmitted disease found worldwide in humans, wildlife and domesticated animals. Commonly used approaches to diagnose babesiosis include microscopic examination of peripheral blood smears, detection of circulating antibodies and PCR. To screen and differentiate canine Babesia infections many PCR assays amplify the 18S rRNA gene. These sequences contain hypervariable regions flanked by highly conserved regions allowing for amplification of a broad-range of Babesia spp. However, differences in the 18S rRNA gene sequence of distantly related clades can make it difficult to design assays that will amplify all Babesia species while excluding the amplification of other eukaryotes. By targeting Babesia mitochondrial genome (mtDNA), we designed a novel three primer qPCR with greater sensitivity and broader screening capabilities to diagnose and differentiate Babesia spp. Methods Using 13 Babesia mtDNA sequences, a region spanning two large subunit rRNA gene fragments (lsu5-lsu4) was aligned to design three primers for use in a qPCR assay (LSU qPCR) capable of amplifying a wide range of Babesia spp. Plasmid clones were generated and used as standards to determine efficiency, linear dynamic range and analytical sensitivity. Animals naturally infected with vector-borne pathogens were tested retrospectively and prospectively to determine relative clinical sensitivity and specificity by comparing the LSU qPCR to an established 18S rDNA qPCR. Results The LSU qPCR efficiencies ranged between 92 and 100% with the limit of detection at five copies/reaction. The assay did not amplify mammalian host or other vector-borne pathogen gDNA except Cytauxzoon felis (a feline protozoal pathogen). The LSU qPCR assay amplified 12 different Babesia. sp. and C. felis from 31/31 (100%) archived samples, whereas the 18S qPCR amplified only 26/31 (83.9%). By prospective analysis, 19/394 diagnostic accessions (4.8%) were LSU qPCR positive, compared to 11/394 (2.8%) 18S rDNA qPCR positive. Conclusions We have developed a more sensitive qPCR assay with a more expansive range of Babesia spp. detection by targeting a highly conserved region of mtDNA, when compared to an established 18S qPCR.


Background
Babesiosis is a protozoal, tick transmitted disease found worldwide in humans, wildlife and domesticated animals. Dogs can be infected with a wide range of Babesia spp., including B. gibsoni, B. vogeli, B. canis, B. rossi, B. conradae, B. microti-like (also referred to as "Theileria annae" or "B. vulpes") and several large un-named Babesia spp., designated "B. coco" [1][2][3][4][5][6]. Clinical signs of canine babesiosis include thrombocytopenia, anemia, splenomegaly, fever, and can result in death [7]. Methods of diagnosing Babesia infections include microscopic examination of peripheral blood smears, indirect immunofluorescent antibody test to detect circulating antibodies and polymerase chain reaction (PCR) to detect pathogen DNA. Gene targets that have been used to amplify Babesia DNA include 18S ribosomal RNA, betatubulin, heat shock protein 70 (hsp70), thrombospondin related adhesive protein gene (P18 or BgTRAP) and two internal transcribed spacers (ITS1 and ITS2) [8][9][10][11][12][13][14]. Amplifying evolutionarily conserved genes, essential for survival, are often reliable PCR targets for genus-specific screening assays; however, identifying highly conserved regions for primer annealing in close proximity to regions of sequence heterogeneity for species discrimination can be challenging, particularly when including more distant lineages of Babesia spp. To screen and differentiate canine Babesia infections, the Vector-Borne Disease Diagnostic Laboratory at North Carolina State University (VBDDL-NCSU) has utilized a quantitative real-time PCR (qPCR) assay designed to amplify a region of the evolutionarily conserved 18S rRNA gene [15]. Ribosomal DNA sequences contain hypervariable regions, which are frequently used for species-specific amplification and are flanked by highly conserved regions used for broad-range genus amplification. This effectively allows amplification and discrimination of most Babesia species. However, differences in the 18S rRNA gene sequence of the more distantly related clades, which include B. conradae and B. microti-like parasites, make it difficult to design 18S rDNA assays that will amplify all Babesia species while excluding the amplification of other eukaryotes. Therefore, diagnostic laboratories often design separate and specific PCR assays to amplify B. conradae and B. microti-like in dogs where these pathogens are suspected. Screening for a wide range of Babesia species using specific primers for each species creates challenges for high throughput testing and limits the ability of a laboratory to identify "new" species that might infect dogs or other domestic and wild animals. In this study, we set out to design a novel assay with greater sensitivity and broader screening capabilities while retaining the ability to differentiate Babesia spp. This goal was achieved by targeting the Babesia mitochondrial genome (mtDNA).
Like other Apicomplexa, Babesia mtDNA can be present in higher copy numbers than the chromosomal genome and contains evolutionarily conserved genes including cytochrome b (cytb), cyclooxygenase (cox) and large subunit ribosomal DNA (lsu) [16]. Improved sensitivity over the 18S rDNA target has been demonstrated using mtDNA targets in several Apicomplexa PCR assays, including Babesia and Theileria spp. [17][18][19][20][21][22]. To the authors' knowledge this report describes the first single qPCR targeting Babesia mtDNA that amplifies a wide range of Babesia spp. We describe the development and validation of a Babesia genus-specific, three primer qPCR assay targeting the lsu5-lsu4 region of mtDNA. The diagnostic utility of this assay (LSU qPCR) was demonstrated through retrospective and prospective analysis by comparing the sensitivity and specificity to an established 18S rDNA Babesia genus-specific qPCR using blood samples from uninfected and naturally-infected animals.

Samples
Samples of ethylenediamine tetraacetic acid (EDTA)-anticoagulated whole blood specimens from various host animals submitted to the VBDDL-NCSU for research or diagnostic testing were used to test the sensitivity and specificity of this assay. Retrospective testing was performed on archived feline, bovine, canine, equine, and wildlife samples previously characterized as containing Babesia spp. (n = 31) to assess sensitivity, and samples containing a different vector-borne pathogen (n = 13) or samples from uninfected animals (n = 4) were used to assess specificity. Archived DNA samples were previously characterized as uninfected or infected using species-specific PCRs or PCR amplification and sequence analysis of the V4 hypervariable region of the Babesia 18S rRNA gene [8]. Six of the characterized samples were from previously published studies and included a B. rossi sample, three B. conradae samples and two B. microti-like samples from grey and red foxes [23][24][25]. Samples containing non-Babesia vector-borne pathogens were confirmed by PCR amplification and sequencing by the VBDDL-NCSU using species-specific gene targets and included Anaplasma platys, A. phagocytophilum, Bartonella henselae, Cytauxzoon felis, Ehrlichia canis, E. ewingii, Hepatozoon americanum, H. canis, Leishmania infantum, Mycoplasma hemocanis, Neorickettsia risticii, Rickettsia rickettsii, Theileria equi and Trypanosoma cruzi. Prospective testing was performed on canine samples submitted between July 1, 2015 and August 28, 2015 for vector-borne disease testing (n = 394).

Primers
To design primers better able to detect DNA from known and emerging Babesia spp., alignments were made between a wide range of Babesia mtDNA sequences. A region spanning two large subunit rRNA gene fragments (lsu5 and lsu4) conserved among Babesia spp. containing sequence heterogeneity, flanked by areas of high similarity, was identified as a potential new qPCR target ( Fig. 1) [26][27][28]. To develop a new Babesia LSU qPCR (LSU qPCR), three primers (2 forward, 1 reverse; Table 1) (Fig. 2). B-lsu-F and B-lsu-R2 were designed to amplify a~150 bp product from all Babesia spp. in the alignment except B. microti or B. microti-like. Bmic-F was designed to amplify a~230 bp product from B. microti or B. microti-like when used with B-lsu-R2 (Figs. 1, 2). For assay comparisons, an established Babesia genus 18S rDNA qPCR (18S qPCR), utilized by the VBDDL-NCSU for Babesia molecular diagnostic detection, was used to amplify a~200 bp region of the Babesia 18S rRNA gene with Bcommon_F and Bcom-mon_R primers (Table 1). Established Babesia speciesspecific PCRs using primers designed to anneal to a hypervariable region of the 18S rRNA gene and newly designed species-specific cox1 (a gene found on the mtDNA) primers were used to confirm Babesia species in LSU qPCR and 18S qPCR positive samples (Table 1). Initial amplicons generated in both the LSU qPCR and cox1 species-specific qPCRs from samples containing known Babesia spp. were sequenced to confirm primers amplified the correct target DNA for each Babesia species. Amplicons generated during prospective testing from the LSU-qPCR but not the 18S qPCR assay (discordant results) were sequenced.

DNA extraction and PCR conditions
DNA extraction was performed on QIAsymphony SP (Qiagen, Hilden, Germany) with QIAsymphony® DNA Mini Kit (192) (Qiagen) or Qiagen BioRobot® M48 Robotic Workstation with MagAttract® DNA Mini M48 kit (Qiagen) depending on the time of sample submission. DNA was stored at -20°C until PCR analysis. All PCR sample preparations were prepared in a biocontainment hood with UV light decontamination capabilities. The absence of PCR inhibitors was demonstrated by the amplification of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) [8]. Amplification reactions for all PCRs contained 12.5 μl SYBR®Green Supermix (Bio-Rad, Hercules, USA), 5 μl DNA template, primers at various concentrations ( Table 2) and molecular grade water to a final volume of 25 μl. Thermocycler conditions were established based on several factors, which included using temperature cycles that could be run with other NCSU-VBDDL qPCR assays, recommendations by Bio-Rad for use of SYBR®Green Supermix, and calculated melting temperatures of newly designed primers. Furthermore, gradients for primer annealing temperatures and concentrations were performed to identify a combination Fig. 1 Babesia LSU qPCR primers in relation to the mitochondrial DNA genome structure. a a1 Babesia microti (Type-I and Type II orientation) [27] and B. microti-like [28]; b b1 "typical" Piroplasmida [26] Fig. 2 Sequence alignment of the lsu5-lsu4 region of the mitochondrial DNA from representative Babesia spp. and primers designed for use in the Babesia LSU qPCR assay. The source of sequence for Babesia spp. are shown in parentheses (GenBank accession numbers or PCR amplicon sequence). * represents sequences aligned in a reverse complement orientation that amplified target DNA efficiently but did not amplify nonspecific DNA. Thermocycler conditions consisted of an initial denaturation step at 98°C for 3 min, followed by 40 cycles at 98°C for 15 s, 60°C for 15 s (or 62°C for B. microti-like 18S species-specific qPCR), and 72°C for 15 s. Melting temperature (T m ) measurements were made between 65 and 88°C at 0.5 s intervals. All qPCRs included a positive control consisting of either a previously characterized Babesia-infected sample or Babesia plasmid DNA and negative controls including a no-template control consisting of filter-sterilized, molecular-grade water and uninfected dog or cat genomic DNA (gDNA). Newly extracted samples were tested with DNA extraction controls. PCR amplification was performed in a C1000™ Thermal Cycler (Bio-Rad) with CFX96™ Real-Time Detection System. Following PCR, amplicons were analyzed via the quantification cycle (C q ), melt curve shape and T m difference. To validate amplicon size, several PCR products were visualized on a 2.0% agarose gel in 1× TAE with ethidium bromide staining alongside a DNA molecular size marker. Sequencing of products or plasmids was performed by GENEWIZ Inc. (Research Triangle Park, NC) and alignments made with GenBank reference sequences using AlignX software (Vector NTI Suite 6.0, InforMax, Inc.)

qPCR efficiency
Plasmid clones, used as standards for efficiency and analytical sensitivity determination, were constructed using LSU qPCR amplicons from B. vogeli, and B. microti-like template DNA with pGEM-T Easy Vector System (Promega, Madison, WI) as recommended by the manufacturer. Plasmids were sequenced and inserts confirmed using M13R primers. Plasmid copy numbers were calculated assuming an average base pair weight of 650 Da and Avogadro's number (6.022 10 23 ) using the following equation: copy number = (DNA ng amount × 6.022 10 23 molecules/mol)/(length of DNA in base pairs × 1 × 10 9 ng/g × 650 g/mol). Duplicate, serial 10-fold dilutions in uninfected canine gDNA (~10-30 ng/μl) resulted in 50-500,000 copies/reaction of plasmid DNA, and standard curves of quantification cycle (C q ) values were plotted against the logarithm of plasmid copy numbers/reaction. PCR efficiency was estimated through linear regression of the dilution curve (10^(-1/slope)-1) × 100). Coefficients were calculated (R 2 ) using Bio-Rad CFX Manager™ software. Efficiency reactions were performed using both 2 and 3 primer reactions to establish any potential interference by a third primer.

Analytical sensitivity and specificity
Analytical sensitivity was determined by calculating the limit of detection (LOD), defined as the lowest concentration at which 95% of the positive samples were detected. Plasmids diluted in canine gDNA (~10-30 ng/μl) to 1 copy/μl were added to reaction wells resulting in 3 and 5 copies/reaction. Twenty intra-assay replicates for each plasmid concentration were tested and C q ranges determined. All assays were performed using both 2 and 3 primer reactions to establish any potential interference by a third primer. Analytical specificity was evaluated retrospectively using gDNA (~10-30 ng/μl) previously extracted from bovine, canine, equine and feline EDTAwhole blood specimens and tested by the VBDDL, determined to be either uninfected or infected with non-Babesia vector-borne pathogens.

Clinical sensitivity and specificity
To determine relative clinical sensitivity and specificity, the LSU qPCR was compared to an established Babesia 18S qPCR diagnostic assay that has been used for over 12   positive amplicons from the prospective study were sequenced to confirm results and were retested in triplicate with both PCRs to account for the effect of Poisson distribution on samples with low template concentrations.

Statistical analysis
Relative sensitivity and specificity for the 18S qPCR and LSU qPCR were determined by a 2 × 2 table, calculating an estimate of agreement relative to a non-reference standard, either the 18S qPCR or the LSU qPCR [29]. Positive percent agreement (PPA), representing relative sensitivity, was determined by the proportion of nonreference standard positive samples where the index assay is positive [a/(a + c)], and negative percent agreement (NPA), representing relative specificity, was determined by the proportion of non-reference standard negative samples where the index test is negative [d/(b + d)]. Agreements and proportions were reported with their 95% confidence intervals (95% CI), calculated by the modified Wald method [30,31]. Confidence intervals were performed using GraphPad Software (La Jolla, California, USA).

qPCR efficiency, analytical sensitivity and specificity
Amplicon size for each primer pair was visualized on an agarose gel and corresponded to the expected size of 150 bp for most Babesia spp. and~230 bp for B. microti-like. T m ranged from 76 to 77.5. LSU qPCR efficiencies ranged from 92 to 100% with an R 2 of 0.99 and a linear dynamic range of 50-500,000 copies/reaction of plasmid DNA (Table 3, Fig. 3). Results were comparable when using either two primers (Bmic-F with B-lsu-R2 for B. microti-like plasmid template or B-lsu-F with Blsu-R2 for B. vogeli plasmid template) or three primers (Bmic-F, B-lsu-F and B-lsu-R2) ( Table 3). LODs were initially tested at 3 copies/well using 20 intra-assay replicates, with detection ranging from 60 to 90% (Table 3). When the LOD was repeated at 5 copies/well, all reactions showed 100% detection except the 3 primer B. microti-like reaction (95% detection). Neither the LSU qPCR nor the 18S qPCR amplified uninfected mammalian host DNA.
The LSU qPCR did not amplify non-Babesia vector-borne pathogens except for C. felis, and the 18S qPCR amplified C. felis, H. canis and T. equi (Table 4).

Clinical sensitivity and specificity
Of the 31 archived, naturally-infected animal blood samples, previously characterized as containing 12 different Babesia spp. or C. felis, 31/31 (100%) were LSU qPCR positive and 26/31 (83.9%) were 18S qPCR positive ( Table 4). The sequenced LSU qPCR amplicons from 9/ 12 Babesia species and C. felis were 100% identical with GenBank reference sequences. Babesia lengau, B. odocoilei and Babesia (sensu stricto) mtDNA reference sequences were not available in GenBank for comparison and were aligned with the lsu5-lsu4 region from other Babesia mtDNA sequences (Fig. 2). Of the 26 concordant samples, the average C q value difference between the two qPCRs (ΔC q = 18S qPCR C q -LSU qPCR C q ) was 7.1. The LSU qPCR C q was less than the 18S qPCR C q (representing earlier amplicon production) for B. bovis  (Fig. 4). Babesia odocoilei was the only Babesia species in this sample set where the 18S qPCR amplified DNA at a lower C q (ΔC q = -10.7, -12.1 and -7.35) than the LSU qPCR.

Discussion
In this report, we describe the development and validation of a sensitive and specific broad-range molecular diagnostic assay for detecting Babesia infections using a three-primer qPCR assay targeting the mtDNA. Prior to validation of the LSU qPCR, the VBDDL utilized a qPCR targeting the 18S rRNA gene, optimized to detect Babesia species infective to dogs, which included B. canis, B. gibsoni, B. rossi and B. vogeli [15]. Since development of the 18S qPCR assay, emerging Babesia species have been discovered to infect dogs, including B. conradae, B. microti-like and several large un-named Babesia spp., designated B. coco [5][6][7]. Furthermore, diagnostic laboratories, including NCSU-VBDDL, routinely test for emerging Babesia spp. in samples collected from other animals, such as wildlife. At high pathogen DNA quantities, the 18S qPCR can amplify 18S ribosomal DNA from these emerging species; however, amplification efficiency is poor with lower pathogen loads. Other Babesia diagnostic methods include testing with multiple, individual, species-specific PCR assays, reverse line blotting, or nested PCRrestriction fragment length polymorphism analysis. These approaches are either costly, not as conducive to a high throughput platform, or may not detect novel Babesia pathogens [32,33]. Screening diagnostic samples at the genus level using broad-range primers followed by additional analysis of PCR positive samples, such as species-specific PCRs and/or amplicon sequencing to determine the species, supports a high throughput platform, is cost effective, and facilitates the discovery of "new" pathogens. Identifying a DNA target conserved among Babesia that contains hypervariable sequences flanked by highly conserved sequences, Hepatozoon canis (dog) 39.0 85 none where primers do not amplify other eukaryotic DNA can be challenging. The LSU qPCR assay amplifies a region of DNA spanning two large ribosomal subunits, lsu5-lsu4, that is conserved among piroplasmida indicating it is less likely to be deleted or changed due to DNA rearrangements or mutations [26][27][28]. This study demonstrated an increased level of relative sensitivity and a broader range of Babesia spp. detection when the LSU qPCR was compared to an established molecular diagnostic assay. There was improved detection when testing animal samples naturally infected with B. bovis, B. coco, B. conradae, B. lengau, and B. microti-like organisms. While we did not include a C. felis mtDNA sequence (GenBank accession no. KC207821) in the Babesia alignment used to design new primers, the Blsu-F and B-lsu-R2 primers are 100% identical to the corresponding sequences of the C. felis lsu5-lsu4 region and amplified this feline pathogen. For differentiation of C. felis from Babesia spp., the authors recommend either amplicon sequencing or testing the sample with a validated C. felis species-specific PCR [17]. The C q differences (ΔC q = 18S qPCR C q -LSU qPCR C q ) between the 18S qPCR and the LSU qPCR have several possible explanations (Table 1, Fig. 4). It is possible the noted differences were due in part to differing primer efficiencies and not necessarily related to target DNA copy number. Primer alignments using the 18S qPCR primers, Bcommon-F (bp = 24) and Bcommon-R (bp = 29), revealed base pair mismatches were greatest with the Bcommon-F primer in B. conradae (9/24) B. lengau (8/24), B. microti-like (7/24), and C. felis (9/24), while all the other Babesia spp. in the retrospective sample set had bp mismatches ≤ five (Fig. 5). Except for B. microti and B. microti-like, where a different forward primer is used for the LSU qPCR target, the B-lsu-F primer has only two mismatches with B. canis and B. vogeli. However, these mismatches are mitigated with two degenerate primer nucleotides (Fig. 2). The B-lsu-R2 Table 4 Retrospective analysis qPCR results. Retrospective qPCRs were performed simultaneously on previously characterized diagnostic or research samples from mammals that were uninfected (n = 4), naturally infected with Babesia or Cytauxzoon species (n = 31) or other vector-borne pathogens (n = 13) (Continued) Abbreviations: C q quantification cycle, T m melting temperature; −, Babesia was not amplified Fig. 4 Comparison of the ΔC q from retrospective 18S qPCR and LSU qPCR assays performed simultaneously on previously characterized diagnostic or research samples naturally infected with Babesia spp. and C. felis. n/a = sample was negative by 18S qPCR and positive in LSU qPCR; ΔC q = C q 18S qPCR -C q LSU qPCR A large prospective set of samples from dogs suspected of exposure to vector-borne pathogens was used to measure relative sensitivity and specificity between an established diagnostic qPCR and a newly developed qPCR. The newly developed LSU qPCR relative sensitivity was 100% when compared to the established 18S qPCR, while the 18S qPCR relative sensitivity was 58% when compared to the LSU qPCR, suggesting the new assay is more sensitive at detecting Babesia. The LOD for the LSU qPCR assay was determined to be~five copies/reaction while the established Babesia 18S rDNA was determined to be between five and ten copies/reaction (data not included); both LODs were determined using plasmids containing one copy of the target DNA diluted in canine gDNA. Making direct comparisons between LODs with other qPCRs reported in the literature using similar methods of calculation show the LSU and 18S qPCRs to be comparable. One assay targeting B. canis hsp70 calculated a LOD at~ten copies using a probe for detection [12]; another qPCR assay targeting Table 6 Prospective analysis was performed on canine diagnostic samples naturally infected with Babesia spp. and quantification cycles (C q ) were compared between the 19 positive samples detected by Babesia genus assays (18S qPCR or LSU qPCR) and species-specific assays (18S sp-sp or cox1 sp-sp). Results are shown from the original Babesia genus qPCRs, discordant samples that were repeated in triplicate with the Babesia genus qPCRs, and Babesia species-specific qPCRs. All positive samples generated the correct melting temperature (T m ) values (not shown) Genus   B. microti 18S rDNA calculated~3.6 copies per reaction using blood spiked with plasmid DNA [34]. Both the LSU and 18S qPCR assays had high specificities relative to each other, with the 98% specificity reported for LSU qPCR reflecting the positive discordant samples not detected by the 18S qPCR. When comparing results from naturally infected canine samples used for the prospective study, eight samples were PCR negative by the 18S qPCR but positive by the LSU qPCR (Table 7). It is unclear if this difference in sensitivity is related to improved primer binding and subsequent amplification, increased target copy number, or clinical status of the patient at time of sampling (treated or untreated). It's possible mtDNA copy numbers vary among Babesia species or stages of piroplasmida infection. Schreeg et al. detected higher mtDNA copy numbers in cats acutely infected with C. felis when compared to cats infected over a year [17]. It remains unclear if there is a link between the mtDNA copy number and stage of Babesia infection.
Limitations of this study include the inability to calculate true sensitivity and specificity for the new LSU qPCR assay and that the new assay was only compared to one other qPCR assay. However, in both the prospective and retrospective sample sets, the LSU qPCR outperformed the 18S qPCR. While reasons behind this improvement remain unclear, we hypothesis that increased target copy number for the mitochondrial PCR is a contributing factor.

Conclusions
In summary, we have developed a qPCR assay with increased sensitivity to detect a broader number of Babesia spp. The LSU qPCR targets a highly conserved region of the mtDNA spanning the lsu5-lsu4 region. Retrospective and prospective analysis with samples from naturally infected animals highlight the expansive range of Babesia spp. detection and improved relative sensitivity when compared to a current 18S qPCR. Currently the VBDDL has implemented the new LSU qPCR assay for all diagnostic Babesia PCRs and confirms all positive samples with a second species-specific qPCR, targeting either the 18S rRNA or cox1 genes. Samples from ungulates are also screened with the 18S qPCR for improved detection of B. odocoilei.