Bacteria and fungi.

To test the specificity of each qPCR assay, reference material was obtained to cover broad taxonomic orders within each target group (S3 File). A total of 23 bacterial and 23 fungal species were provided from the U.S. Department of Agriculture (USDA) Agricultural Research Service Culture Collection (Peoria, IL, USA) either as live or lyophilized specimens. Live cultures, provided as agar slants, were scraped with loops, and submerged into tubes containing trypticase soy broth. Bacteria were plated onto both trypticase soy agar and blood agar, and fungi were plated on cornmeal agar. Lyophilized cultures were first submerged in trypticase soy broth and swirled with a loop to assist in rehydration. After the pellet dissolved, cultures were plated as above. Plates and broth tubes were incubated at room temperature for approximately 14 days. DNA was isolated from the culture plates and broth tubes using the ZymoBIOMICS^™ DNA Miniprep Kit (Zymo Research, Irvine, CA, USA) following the manufacturer’s protocol [35]. For broth cultures, the broth was thoroughly mixed with a pipette and 1 mL was transferred to a 1.5 mL tube and centrifuged at max speed for 3 min using a Centrifuge 5415 C (Eppendorf, Hamburg, Germany). A volume of 250 µL from the bottom of the tube was pipetted into the ZR BashingBead^™ Lysis Tubes for DNA isolation. Visible colonies from cultures grown on agar were swabbed with BBL^™ CultureSwab^™ EZ swabs (Becton Dickinson, Franklin Lakes, NJ, USA) and swab heads were cut directly into DNA lysis tubes. Samples underwent bead beating for 40 min using speed 7 with the Vortex-Genie 2T vortex (Scientific Industries, Inc. Bohemia, NY, USA). DNA was eluted in 100 µL of ZymoBIOMICS^™ DNase/RNase Free water and steps 11–13 of the manufacturer’s protocol were repeated for a second 100 µL elution.

Plants and arthropods.

Forty-four plant specimens (single leaf for each plant) were collected from live plants at the JC Raulston Arboretum in Raleigh, North Carolina and 22 arthropods were collected live from various locations in Raleigh. Subsamples of plant leaves (~10–35 mg) and single arthropod legs were placed into 1.5 mL microcentrifuge tubes and subsequently bleached in a 5% bleach solution for 5 min and washed three times with sterile water to minimize co-extraction of non-specific DNA [36]. After washing, specimens were placed onto to a thermomixer at 40°C for 20 min to facilitate drying. Tubes were left open and allowed to dry in a biosafety cabinet overnight. DNA from plant leaves was subsequently isolated using the DNeasy Plant Pro Kit (Qiagen) following the manufacturer’s protocol with a minor modification that included a second 50 µL elution. Plant leaves that were extremely tough (e.g., pine needles) were crushed with sterilized plastic pestles (Carolina Biological Supply Co., Burlington, NC, USA) prior to DNA isolation to ensure sufficient cell lysis. The DNeasy Blood and Tissue Kit (Qiagen) was used to isolate arthropod DNA following the manufacturer’s protocol for insects. Arthropod legs or whole arthropods (mealybugs only) were ground using sterile BioMasher^® II Micro Tissue Homogenizers (Kimble Chase, Vineland, NJ, USA) and after the addition of lysis reagents, incubated at 56°C in a ThermoMixer C (Eppendorf) for 2 hours at 300 rpm. DNA was subsequently eluted in two 50 µL elutions.

DNA quantification.

DNA isolated from single source bacteria, fungi, plants, and arthropods was quantified with the dsDNA High Sensitivity Assay (Invitrogen, Waltham, MA, USA) using the Qubit^™ Fluorometer and subsequently diluted to 0.5 ng/µL with sterile water. DNA was stored at −20°C until use.

Annealing temperature optimization.

For each primer pair, a total of seven different sources of DNA were tested to determine the optimal annealing temperature: a) four individual single source species within the target taxon (e.g., four taxonomically diverse plant species for the plant psbA primers), and b) each non-target DNA pool (e.g., fungal primers were tested against bacterial, plant, and arthropod pools). Duplicate reactions for each primer pair and DNA source were completed at 57°C, 58°C, 59°C, and 60°C. These temperatures were selected as: 1) the manufacturer recommends an annealing temperature range of 55–65°C for this master mix, and 2) considering these primers were selected to amplify across broad taxa, annealing should be slightly lower than the standard 60°C but not too low to introduce non-specific amplification.

Primer concentration optimization.

Nine primer concentrations were tested with each primer pair using an annealing temperature of 57°C. For these experiments, only the specific DNA pool for a given primer pair was used (e.g., only plant DNA pool was tested with plant psbA primers). Final concentrations tested were as follows with the forward primer listed first and the reverse primer listed second: 300/300 nM, 300/500 nM, 300/800 nM, 500/300 nM, 500/500 nM, 500/800 nM, 800/300 nM, 800/500 nM, and 800/800 nM [38].

Specificity.

Each primer pair was tested with a) individual single source DNA from species within the target taxon, b) each non-target DNA pool (e.g., fungal primers were tested against bacterial, plant, and arthropod pools), and c) individual single source DNA from seven vertebrate species (black bear, alligator, sheep, rhesus monkey, dog, mourning dove, northern water snake). For all four primer pairs, duplicate reactions for each DNA source were completed at an annealing temperature of 57°C with a final primer concentration of 300 nM for both forward and reverse primers. The number of taxon-specific single source species screened for each primer pair was 47 for plants, 24 for bacteria, 30 for fungi, and 23 for arthropods. Additionally, DNA isolated from 15 surface soil and 15 dust samples were tested with each primer pair. Fungal, plant, and arthropod DNA pools that were utilized in a separate project were also included as a source of mixed starting template, as some of the taxa used in these pools were not included in the list of single source species being tested. These pools consisted of two fungi (baker’s yeast and a yard mushroom), two bacteria (E. coli and a Salmonella species), ten plants (basil, pine, moss, Boston fern, white wild indigo, Columbian monkshood, red bitterberry, henbane, hemlock, and purging croton), and ten arthropods (cockroach, moth, tick, beetle, fly, wasp, cricket, silverfish, spider, and pill bug). Additionally, the ZymoBIOMICS^™ Microbial Community DNA Standard (Zymo Research), which includes eight bacteria and two fungi, was incorporated to test the bacterial primers.

Reproducibility.

Reproducibility experiments encompassed two separate phases: 1) analyst, and 2) master mix. Notably for both phases, different qPCR instruments were used to additionally assess instrument reproducibility. For phase 1, two analysts completed matching experiments (i.e., same reagents and DNA samples) across three qPCR instruments: the QuantStudio 5 (Applied Biosystems; used for the initial experiments previously outlined) and two different Step-One Plus (Applied Biosystems) instruments. Each primer pair was tested with a) four individual single source DNA from species within the target taxon (e.g., four plants for the plant psbA primers), and b) each non-target DNA pool (e.g., fungal primers were tested against the bacterial, plant, and arthropod pools). For phase 2, three different SYBR-based qPCR master mixes were evaluated: PowerTrack^™ SYBR^™ Green Master Mix (Applied Biosystems; used for the initial experiments previously outlined), Luna^® Universal qPCR Master Mix (New England Biolabs, Ipswich, MA, USA), and GoTaq^® qPCR Master Mix (Promega Corporation, Madison, WI, USA). Experiments to evaluate master mix were completed on both a QuantStudio 5 and a Step-One Plus qPCR instrument. DNA pools were only used for phase 2 and individual species testing was not performed. For experiments completed in both phase 1 and 2, duplicate reactions were performed for each DNA sample, and an annealing temperature of 57°C with a final primer concentration of 300 nM each were used.

Standard curve generation & linear dynamic range.

Complete bacterial 16S, fungal 18S, arthropod 16S, and plant psbA sequences were downloaded from NCBI and aligned in CLC Genomics Workbench 20.0.2 (Qiagen) using the alignment tool with the default settings (S4 File). The region to be amplified by each primer pair was first identified in the alignment, then the consensus sequence for each target amplicon was subsequently extracted from each alignment and provided to IDT for synthesis into four separate gBlocks^™ (i.e., synthetic double stranded DNA fragments). To provide some flexibility with primer binding, 15 nucleotides upstream of the forward primer binding site and 15 nucleotides downstream of the reverse primer binding site were included in each consensus sequence for gBlock^™ synthesis at a yield of 250 ng. IDTE (IDT) was used to resuspend each gBlock^™ with a volume of 25 µL for a final concentration of 10 ng/µL and were stored at 4°C. Copy number was determined for each gBlock^™ following an online copy number calculator [39] using the mass of each gBlock^™ and a 1:10 serial dilution was subsequently performed (10¹–10⁸). Each dilution (n = 8) was tested with all primer pairs using 57°C as the annealing temperature with a final primer concentration of 300 nM for both the forward and reverse primers. Each gBlock^™ was also used as DNA template (10⁵ copy number dilution only) for all primer pairs to further assess cross-reactivity and specificity.

Sensitivity.

Sensitivity was assessed using the smallest gBlock standard (10 copies) with 10 replicates to determine if 10 DNA copies could be consistently detected. Each primer pair was tested against each replicate with a final primer concentration of 300 nM each and annealing temperature of 57°C.

Inhibition.

Inhibition experiments were completed to assess the impact of three commonly encountered PCR inhibitors likely to be associated with environmental samples on qPCR [40]. Ethylenediaminetetraacetic acid (EDTA; 0.5 M, pH 8.0; Fisher Scientific, Hampton, NH, USA), humic acid (Fisher Scientific), and tannic acid (Fisher Scientific) were prepared and diluted following Kavlick 2019 which describes the development of a mitochondrial DNA quantification assay for forensic DNA analysis [41]. Stock EDTA was diluted to three concentrations for qPCR testing: 5000 µM, 2500 µM, and 1250 µM. Humic acid was reconstituted using 4 mg of stock powder and 1 mL of nuclease-free water and was subsequently diluted to 40 µg/mL, 20 µg/mL, and 10 µg/mL for qPCR. Tannic acid was reconstituted using 10 mg of stock powder and 10 mL of nuclease-free water and was diluted to four concentrations: 75 µg/mL, 50 µg/mL, 25 µg/mL, and 12.5 µg/mL. All 20 µL qPCR reactions used 2 µL of the diluted inhibitor. Each inhibitor was tested in triplicate using the gBlock^™ standards at both 10³ and 10⁵ copy numbers for each primer pair. Triplicate reactions were also performed for each primer pair using each inhibitor concentration containing no DNA (inhibitor control).

C_q values.

Replicates were averaged together for each individual DNA sample in each experiment and used to generate the summary statistics (mean, standard deviation, median, minimum, and maximum) of the C_q values. The C_q values from the specificity experiments were calculated for each taxon’s primer pair for the: 1) “in” group (e.g., plant primers tested against plant single source species), 2) “out” group (e.g., plant primers used with non-plant DNA pools), and 3) “mixed” group (e.g., plant primers amplifying plant DNA within soil and dust samples). Normality was tested using the Shapiro-Wilk test and found not to hold. The Kruskal-Wallis test was used to compare means between all three groups and Dunn tests with a Bonferroni correction were performed to determine which group differences were statistically significant. For reproducibility testing, summary statistics were determined for each instrument by analyst. Analyst’s C_q values for each instrument were compared using the Wilcoxon signed rank test or paired t-test depending on whether the data was non-normally or normally distributed, respectively. The intraclass correlation coefficient (ICC) based on a single rater and on absolute agreement was utilized to assess reliability [43]. C_q values between instruments (averaged over raters) were compared using Wilcoxon signed rank tests and the ICC was calculated for the three instruments. Summary statistics were calculated for the master mix reproducibility experiments. C_q values from each instrument within each master mix were compared using the Wilcoxon signed rank test (non-normally distributed) or paired t-test (normally distributed) and the ICC across the instruments within each master mix was determined. Following this, master mix C_q values were within each instrument were compared to each other using Wilcoxon signed rank tests and the ICC across master mixes within each instrument was calculated. For each standard curve, the regression equation (C_q = β₀ + β₁ X₁, where the value of X₁ is the power of the dilution) and standard errors were calculated. Summary statistics and reference intervals were determined for sensitivity experiments. Kruskal-Wallis test and Dunn post-hoc tests were also used for the inhibition studies.

Melt curve temperatures.

For specificity experiments, each taxon’s primer pair melt curve temperatures were summarized per group (i.e., in, out, and mixed) as described above. Normality was assessed with Shapiro-Wilk tests and found not to hold, therefore Kruskal-Wallis tests were used to compare the melt curve temperatures and Dunn tests with a Bonferroni correction were applied to determine which of the three groups were significantly different when required. Reference intervals were also calculated for each group including 90% confidence intervals for the locations of the reference internal endpoints. Summary statistics were calculated for reproducibility experiments, and 1) differences between analysts for each instrument were assessed using Wilcoxon signed rank tests and the ICC was calculated, and 2) instruments were compared to each other averaging over analysts using Wilcoxon signed rank tests and the ICC was calculated. Master mix reproducibility was assessed by first determining the summary statistics and comparing the instruments within each master mix using Wilcoxon signed rank tests or paired t-tests, followed by the ICC. Master mixes were then compared within each instrument using paired t-tests and the ICC. Summary statistics and reference intervals were determined for sensitivity experiments. Kruskal-Wallis test and Dunn post-hoc tests were also used for the inhibition studies.

Annealing temperature optimization.

Four target single source species amplified in duplicate for each primer pair were used to determine the optimal annealing temperature (S2 Table). For fungi and plants, the average quantification cycle (C_q) value increased as temperature increased, however this trend was not observed with bacteria and arthropods. With increasing temperature, this trend indicated that less DNA was able to be amplified which can have an impact on determining the true copy number in each sample. The average melt curve temperature and standard deviation for each primer pair across all temperatures tested (57°C, 58°C, 59°C, and 60°C) were as follows (four species each performed in duplicate for each temperature): 1) bacteria – 89.02°C ± 2.14°C, 2) fungi – 79.69°C ± 0.98°C, 3) arthropods – 79.83°C ± 1.79°C, and 4) plants – 81.92°C ± 0.28°C. Considering the temperature consistency of the melt curves, and that fungi and plants had the lowest C_q value at the lowest temperature, 57°C was selected as the annealing temperature for all primer pairs and used in all remaining experiments. If a lower temperature was selected there is an increased risk for non-specific amplification and if a higher temperature had been chosen there is the potential for reduced amplification of true template.

Primer concentration optimization.

Nine primer concentration combinations were tested in duplicate using taxon-specific DNA pools. Average and standard deviation C_q values for each primer pair were as follows (one pool performed in duplicate for each concentration combination): 1) bacteria – 22.47 ± 0.40, 2) fungi – 18.62 ± 0.44, 3) arthropods – 19.61 ± 0.11, and 4) plants – 21.66 ± 0.33. Negative controls were assessed to see if increasing primer concentration had an impact in obtaining a C_q, as negative controls should be undetermined (no DNA present). A total of 18 negative controls were included for each primer pair, and the number with “undetermined” C_q (those with C_q greater than the 40 cycles of PCR) were as follows: bacteria 1/18, fungi –16/18, plants –17/18, and arthropods – 6/18. For those negative controls in which a C_q value was generated, all C_q values for negative controls exceeded 30, with a median C_q of 35.54 for bacteria and 36.55 for arthropods (which exceeded the limit of detection; see Sensitivity). Bacteria’s one undetermined occurred when both primers were 300 nM. The instances of undetermined for arthropods happened when both primers were 300 nM for both replicates, one replicate for 300/500 nM, both replicates for 500/300 nM, and one replicate at 800/500 nM. Since a) the standard deviation between C_q values was low between the primer concentrations tested, and b) that each primer pair negative control for 300/300 nM were undetermined (except for one negative for bacteria), 300 nM for each primer was selected for all targets.

Specificity.

Specificity of each primer pair was assessed using single source species, pools of single source DNA, and mixed soil and dust DNA samples (Fig 1 and S1–S8 Figs). Bacteria – All but one single source sample for bacteria generated C_q values. Notably, multiple melt curve peaks were identified; of 24 single source species screened, 15 had a single melt curve peak, whereas the remaining eight species had double melt curve peaks and one species had a triple melt curve peak. The phyla that encompass the multi-peak melt curves were Actinomycetota (n = 6), Pseudomonadota (n = 1), Verrucomicrobiota (n = 1), and Bacteroidota (n = 1). Species within these phyla have been reported to have multiple 16S rRNA genes within the genome [44] and thus the observation of multiple melt curve peaks could be attributed to gene duplication. The bacterial standard, eight soil samples, and six dust samples produced single melt curve peaks, whereas the remaining seven soil and six dust samples had double melt curve peaks, and even three dust samples had triple melt curve peaks. Fungi – All fungi samples and replicates had generated C_q values, and all melt curve analyses resulted in single peaks except for one fungal species (Coemansia nantahalensis [Kickxellales order]) that had a double melt curve peak for one replicate, but not the second replicate. Arthropods – All arthropod samples and replicates had determined C_q values except one single source sample (mealybug), one soil sample and two dust samples. Single melt curve peaks were produced for all single source species except for a mealybug species (that did not result in a melt curve peak), 11 soil samples, and nine dust samples. Four soil samples and five dust samples had double melt curve peaks and one dust sample had triple melt curve peaks. Plants - C_q values were generated for each sample and its replicate, as well as a single melt curve peak. Only one plant species (Selaginella braunii [Selaginellales order]) had a double melt curve peak at 82.41°C and 84.20°C. Table 1 summarizes the mean, standard deviation, median, minimum, and maximum of C_q values for each primer pair and group.

Download:

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

Table 1. Summary Statistics Using C_q Values for Each Primer Pair.

https://doi.org/10.1371/journal.pone.0343035.t001

Download:

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

Fig 1. Melt curve temperature plots for all in group samples.

Single source specimens for plants (A), fungi (B), bacteria (C), and arthropods (D) underwent melt curve analysis for specificity testing of each primer pair. Each trace is a different sample. The y-axis is the derivative reporter fluorescence (-Rn’) and the x-axis is temperature (°C). The peak of the melt curve is the temperature at which 50% of DNA is double stranded and 50% is single stranded.

https://doi.org/10.1371/journal.pone.0343035.g001

Mean C_q values were compared for all groups (in, out, and mixed) for each primer pair and were found to be statistically different (Table 1). For bacteria, fungi, arthropods, and plants the in groups for each taxon were statistically different from the mixed groups (soil and dust) (Table 1). When comparing in vs. out (e.g., plant primers used with non-plant DNA pools) groups within plants, fungi, and arthropods, the C_q values were significantly different, but were not for bacteria, and only arthropods were significantly different between the out vs. mixed group. It is important to recognize that the out groups should have n = 0; however, these assays are designed to quantify total taxon DNA, and it is possible that DNA from non-target species may be present in target samples (i.e., bacterial species that were co-isolated with plant leaf tissue). S3 and S4 Tables provide summary statistics and comparisons, respectively, of the associated melt curve temperatures. The only statistically significant differences between groups were for plants and bacteria, where plant in vs. mixed groups were different, and bacterial in vs. mixed and in vs. out groups were different. The sample sizes for the out group samples were generally small, which may have been responsible for the lack of statistically significant differences between the out group and in or mixed groups. This is expected as a larger sample size of out groups would indicate the assays are not specific. Additionally, Sanger sequencing was used to confirm that primer pairs were amplifying the intended target if non-specific amplification occurred. For instance, when screening the plant Oxalis crassipes with bacterial primers, amplification occurred. Upon searching the resulting Sanger sequence data against GenBank, the best hit was to the bacterium, Serratia marcescens, likely co-isolated despite completing a 5% bleach wash of the plant tissue prior to DNA isolation. Reference intervals were also calculated for both in and mixed groups for each primer pair, but intervals were unable to be calculated for out groups given small sample sizes (Table 2). These intervals provided an estimate for expected melt curve peaks for each primer pair and whether the sample is single source or mixed.

Download:

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

Table 2. Reference Intervals for Each Primer Pair Based on Melt Curve Temperatures (°C).

https://doi.org/10.1371/journal.pone.0343035.t002

Of note, mixed DNA samples (soil and dust) that were targeted with bacterial primers often had melt curve peaks that do not follow normal peak morphology for dye-based qPCR (i.e., multiple peaks, short and broad) (S1D Fig). Considering many bacteria contain different copy numbers of the 16S rRNA gene [44], we hypothesize that the combination of numerous bacterial species and a range of 16S copy numbers present in environmental samples, impacted the melt curve analysis. Melt curve analysis was key in identifying primer dimer as the temperature was found in the low 60s°C and morphology of these peaks were short and wide compared to true amplicon signal (tall and narrow). We therefore identified that the arthropod 16S primers had a higher incidence of primer dimer, which fit the primer dimer peak morphology, (Fig 1D) compared to the other three assays that had minimal to no primer dimer.

Additionally, seven vertebrate species were also screened using each of the four primer pairs in duplicate. For bacteria, at least one replicate for each species had amplification (as observed by a recorded C_q value), but were all below the sensitivity threshold (see 3.1.8), except the sheep with an average C_q of 18.73. To confirm specificity, each vertebrate species was amplified via standard PCR with the bacterial primers, underwent Sanger sequencing, and the resulting sequences searched against NCBI’s BLAST. The PCR product amplified using sheep DNA returned an uncultured Ruminobacter species and the amplicon product from the alligator DNA was identified as an uncultured Gammaproteobacteria species. The remaining sequences were too messy to query against BLAST, however since these are not sterilized specimens, it is likely that the PCR product contains multiple bacterial species. Surprisingly, the sequence generated from dog DNA returned a BLAST hit for a ncRNA belonging to Canis lupis dingo and mRNA for Canis lupus familiaris; this sequence was however ~35 bp longer than what was visualized on a 2.2% agarose Flash gel (Lonza, Basel, Switzerland). For arthropods, three of the seven vertebrate species did not have any amplification (alligator, sheep, and mourning dove), whereas only one replicate amplified for the northern water snake, and both replicates amplified for the bear, monkey and dog. However, the C_q values were below the sensitivity threshold for bear and snake, but within range for the monkey (31.09) and dog (30.81). As with the bacteria, standard PCR was performed and the resulting amplicons were Sanger sequenced and searched against BLAST for the black bear, rhesus monkey, and dog samples. The black bear sequence was too messy to search most likely due to its low quantity. The dog sequence returned a match to a giraffe species, Giraffa tippelskirchi tippelskirchi, and the rhesus monkey sequence matched to Macaca mulatta; both matched the mitochondrial genome. It is unclear how the dog sample was identified as a giraffe, but notably the generated Sanger sequence was not pristine (i.e., some higher background noise), which may have impacted the BLAST results as the top hit had a percent identity of 93.44%, query coverage of 98%, and e-value of 6e-15. While giraffe was the best match, according to the Organization of Scientific Area Subcommittee’s GenBank Standard, the reported BLAST statistics would not result in species identification at that level [45]. When comparing the melt curves, the melt curve temperatures fall in the range identified in Table 2 (82.85°C for rhesus monkey and 81.32°C for dog). Amplification did not occur for any of the seven vertebrate species with the fungal and plant primers.

Reproducibility – analyst.

Analyst reproducibility was assessed for each instrument using mean C_q value and found to not be significantly different for all three instruments. The intraclass correlation coefficient (ICC) between analysts was determined for each instrument and found to have ICC values of 0.86 [StepOne Plus 1], 0.99 [QuantStudio 5], and 0.97 [StepOne Plus 2]. The ICC of 0.86 is considered good reliability and the ICC of 0.99 and 0.97 are excellent reliability, indicating that the two analysts’ C_q values were in agreement. The mean melt curve temperatures between the two StepOne Plus instruments were also not significantly different, however the mean melt curve temperatures were statistically different for the two analysts (83.16 vs. 83.10°C) using the QuantStudio 5 (V = 273, p < 0.001). This is possible when one analyst consistently produced values that were smaller compared to the other analyst. The statistics applied here indicated that the analysts were consistently different from each other, but the difference was not large (0.06°C). The resulting ICC between analysts for each instrument was 1, demonstrating excellent reliability, despite the minor melt curve temperature differences between analysts when using the QuantStudio 5.

Reproducibility – instrument.

When comparing C_q values, these were found to be significantly different between StepOne Plus 1 and the QuantStudio 5 (V = 50, p < 0.001) and the StepOne Plus 2 and the QuantStudio 5 (V = 364, p < 0.001), but not for both StepOne Plus instruments (V = 289, p = 0.253). The ICC was also calculated across all three instruments and was 0.94. Despite the disagreement between the actual values between the QuantStudio 5 and both StepOne Plus instruments, the ratings are very highly correlated across the three instruments to where any single instrument reliably differentiates among samples in the same way. Mean and standard deviation melt curve temperatures are 83.37°C ± 3.94°C for the StepOne Plus 1, 83.50°C ± 3.95°C for the QuantStudio 5, and 83.30°C ± 3.97°C for the StepOne Plus 2. Instrument reproducibility was found to be significantly different across all three instruments when comparing mean melt curve temperatures: StepOne Plus 1 vs. QuantStudio 5 (V = 76, p = 0.019), StepOne Plus 1 vs. StepOne Plus 2 (V = 389, p = < 0.001), and QuantStudio 5 vs. StepOne Plus 2 (V = 325, p = 0.004). Despite the disagreement between values, the ICC was 1 across all three instruments, demonstrating that any single instrument reliably differentiates among samples in the same way.

Reproducibility – master mix.

Master mix reproducibility was evaluated across two instruments, the QuantStudio 5 and StepOne Plus 1, and three master mixes. Luna^® Universal qPCR Master Mix was found to be significantly different between instruments (V = 192, p = 0.033), however the ICC was determined to be 0.94, indicating that despite the differences in C_q values, the correlation between measures was high. The C_q values obtained using the GoTaq^® qPCR Master Mix were not different between instruments (t(17) = −1.40, p = 0.178) and had an ICC of 0.89 and was considered good reliability. Lastly, the PowerTrack^™ SYBR^™ Green Master Mix was determined to be significantly different between the two instruments (V = 22, p = 0.016), however as seen with the Luna^® Universal qPCR Master Mix, the ICC was 0.85, indicating that despite the differences in C_q values there was good reliability between the two instruments. When comparing the master mixes to each other, they were not significantly different when used on the StepOne Plus 1 and had an ICC of 0.83, which was considered good reliability. For the QuantStudio 5, the GoTaq master mix was significantly different from the Luna (V = 6, p < 0.001) and PowerTrack master mixes (V = 127, p = 0.015). The ICC was calculated across the three master mixes for QuantStudio 5, and was 0.79, demonstrating good reliability despite differences in C_q values.

When comparing melt curve temperatures across the different master mixes, the samples amplified with Luna had a slightly higher mean temperature when using the QuantStudio 5 (82.67°C) compared to the StepOne Plus (82.55°C) (V = 5, p = 0.001). The ICC between the two instruments was 1, indicating that the correlation between the two instruments was high with the Luna master mix. When using the GoTaq master mix, the mean melt curve temperature was higher on the QuantStudio 5 (81.91°C) compared to the StepOne Plus (81.72°C). These results were significant (t(13) = −2.60, p = 0.022), however the ICC of the two instruments when using GoTaq was 0.99 where the correlation between the two instruments was high. The PowerTrack master mix showed no significant differences between the QuantStudio 5 (82.85°C) and StepOne Plus (82.82°C) and the two instruments also had an ICC of 1. When comparing master mixes when using the StepOne Plus, the Luna master mix (81.79°C) was significantly different from the GoTaq master mix (81.43°C; t(12) = 4.91, p < 0.001), but not compared to PowerTrack (81.98°C; t(13) = −1.43, p = 0.176). GoTaq and PowerTrack were also significantly different from each other (t(13) = −3.85, p = 0.002). GoTaq had a lower mean melt curve temperature compared to the other two master mixes. The ICC was 0.98, signifying excellent reliability. Similar observations were seen for the QuantStudio 5, where GoTaq had a lower mean melt curve temperature (82.02°C) compared to Luna (82.36°C) and PowerTrack (82.55°C) master mixes. GoTaq was significantly different from Luna (V = 84, p = 0.08) and PowerTrack (V = 12, p = 0.007), but Luna and PowerTrack are not statistically different (V = 29, p = 0.149). The ICC was also 0.98, showing excellent reliability despite significant differences.

Standard curve generation & linear dynamic range.

Consensus sequences for each gBlock^™ standard are provided in S5 Table with associated length information. Copy number was calculated for each gBlock^™ using the mass of each dsDNA standard, the total amount of DNA in ng, and Avogadro’s constant. A serial dilution starting with 10⁷ copies and ending with 10¹ copies were prepared for each gBlock^™ standard (S9 Fig). The highest standard, 10⁸ copies, contained too much template resulting in reaction failure, and thus was not included in the standard curve. Regression analysis was used to determine the slope and R² values for each standard (Table 3). A range in slopes from −3.58 and −3.10 indicate an ideal reaction, with a slope of −3.3 signifying 100% amplification efficiency [46]. The slopes for these four assays were found to be higher than −3.3 indicating that PCR is less than 100% efficient. The slopes for the bacterial and arthropod assays fall within the acceptable slope range, however the plant and fungal assays have slopes that fall outside of this range (Table 3). Radwan and colleagues, who published the fungal primers for probe-based qPCR, reported their slopes as −3.0472 (single-plex), −3.3034 (multiplex with bacteria), and −4.202 (multiplex with bacteria and archaea) [33]. This disparity could potentially be due to the use of a dye-based assay compared to probe-based, additionally the cycling parameters employed in this research were different compared to those previously reported [33]. Specificity was also confirmed for each gBlock^™ standard using the 10⁵ dilutions with all primer pairs, in which all primer pairs were found to be specific to their gBlock^™ standard.

Download:

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

Table 3. Regression Analysis of gBlock^™ Standard Curves.

https://doi.org/10.1371/journal.pone.0343035.t003

Sensitivity.

Sensitivity was assessed using 10 replicates of the lowest gBlock^™ standard (10¹) for each primer pair. All 40 reactions successfully generated a C_q value and summary statistics can be found in Table 4. The lowest mean C_q was 29.43 for bacteria, followed by 30.26 for arthropods, 34.23 for plants, and the highest C_q of 34.46 for fungi. Mean melt curve temperatures for each primer pair were determined, with arthropods having the lowest temperature at 78.6°C and bacteria the highest at 87.94°C (Table 4). Reference intervals (95%) were also calculated for each standard with respect to melt curve temperature (Table 4). Importantly, the median, minimum, and maximum for each standard were within the reference interval described in Table 2, demonstrating their specificity. These assays exhibit sensitivity down to 10¹ copies and have minimal variation in their melt curve analysis. Importantly, the limit of detection is equivalent to the median C_q values obtained for each primer pair and samples generating C_q values that are higher than those reported should be considered inconclusive or below the limit of detection.

Download:

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

Table 4. Sensitivity Assessment Utilizing the Lowest Standard of Each Primer Pair.

https://doi.org/10.1371/journal.pone.0343035.t004

Inhibition.

EDTA, humic acid, and tannic acid at varying concentrations were used to evaluate inhibition for each primer pair using two gBlock^™ concentrations, 10³ and 10⁵ copies of DNA. Increases in C_q values indicated that the inhibitor used in the qPCR reaction had a negative impact on amplification when compared to the controls. Melt curve temperatures were also analyzed to determine if inhibition increased or decreased the temperature compared to the controls. EDTA did not impact qPCR when using the bacterial and arthropod primers, however C_q values were significantly different for the fungal and plant primers (S6 Table). Specifically, for the fungal primer pair with both gBlock^™ quantities, the 5000 µM EDTA concentration had C_q values that were significantly higher compared to the control. This was also observed for the plant primer pair when using 10⁵ copies of DNA, and for the 10³ copy number samples, statistical significance was observed, but following adjustment for multiple testing, the post-hoc tests could not identify any specific group differences (S6 Table). As for the melt curves for samples treated with EDTA, only the arthropod primer pair with 10⁵ DNA copies was statistically significant, with the 5000 µM samples having a slightly lower temperature compared to the 1250 µM samples. The addition of humic acid resulted in a statistical difference for all primer pairs and gBlock^™ concentrations with the exception of the fungal primers using 10⁵ copies of DNA. Post-hoc testing identified that the C_q values were higher for 40 µg/mL treated samples compared to the control for 1) both gBlock^™ concentrations when using the bacterial primers, 2) the 10³ copies of DNA for the fungal primers, and 3) the 10⁵ copies of DNA for the arthropod primer pair. The 40 µg/mL humic acid treated samples were significantly different from the 10 µg/mL treated samples for the arthropod primers at 10³ copies and for the plant primers at 10⁵ copies. There were not specific differences identified for the plant primers using 10³ copy numbers (S6 Table). With respect to melt curve temperatures, all primer pairs and gBlock^™ concentrations were significantly different. Post-hoc testing identified that the 40 µg/mL treated humic acid samples were slightly lower compared to the 10 µg/mL treated samples for the bacterial primers (10³ copies) and arthropod primers (10⁵ copies). The 40 µg/mL treated samples were also slightly lower compared the control (0 µg/mL of humic acid) for the fungal primers (10³ copies), arthropod primers (10³ copies), and plant primers (10⁵ copies). No specific differences were identified for 10⁵ copy number samples using the bacterial and fungal primers as well as the 10³ copy number samples with the plant primers (S6 Table). Lastly, qPCR reactions with the addition of tannic acid were statistically significant for all gBlock^™ concentrations and primer pairs except for the fungal primers. The 75 µg/mL tannic acid treated samples had higher C_q values compared to the controls when using the bacterial primers (10³ copies) and arthropod primers (10⁵ copies). Similar observations were seen for the 50 µg/mL and 25 µg/mL tannic acid treated samples compared to the control for the plant primers at 10³ copies and 10⁵ copies, respectively. Specific differences were unable to be identified via post-hoc testing for the bacterial primers (10⁵ copies) and arthropod primers (10³ copies) (S6 Table). The only statistical difference observed for melt curve temperatures with tannic acid was the arthropod primer pair using the 10⁵ copy gBlock^™, where the 75 µg/mL tannic acid samples had a slightly lower temperature compared to the 25 µg/mL samples (S6 Table).

When looking across each inhibitor, there was not a consistent pattern that was apparent (e.g., all high inhibitor concentrations have higher C_q values compared to controls for all primer pairs), nor does a specific DNA copy number appear to be affected more by inhibition than the other. Interestingly, not all reactions that had significant C_q values had significant differences observed for melt curve temperatures. It is evident that these assays were able to tolerate various levels of inhibitor, particularly 1250 µM and 2500 µM of EDTA, 10 µg/mL and 20 µg/mL of humic acid, and 12.5 µg/mL of tannic acid. While not all reactions with each primer pair were affected by high inhibitor concentrations, a large proportion resulted in delayed C_q values. If inhibition is suspected, additional DNA purification may be needed. Additionally, other inhibitors that may be encountered (e.g., melanin) should undergo testing to characterize the level of inhibition that may be observed. While advances to DNA isolation kits to remove inhibitors and polymerases to be tolerant to inhibitors have been made, there could be some scenarios in which inhibition could negatively impact quantification results.

In silico analysis.

Each qPCR primer pair was assessed with the in silico primer tool, ecoPCR, for specificity. A total of 14,710 psbA, 22,652 18S, and 26,152 16S eukaryotic RefSeq sequences were downloaded for analysis. Additionally, 38,138 RefSeq bacterial assemblies were downloaded which contained 86,817 complete genome and plasmid sequences. Table 5 describes the results from each in silico analysis. For plants, 81.9% of plant sequences were amplified and had a median forward and reverse primer mismatch of 0 and 2, respectively. A mismatch of 0 is perfect annealing of primer to template DNA, whereas a mismatch of 2 is an imperfect annealing, where two nucleotides of template will not be able to anneal to the primer, which may impact amplification. Only 3.7% of non-plant sequences were amplified and had a median of three mismatches for both the forward and reverse primers. The taxa comprising the 3.7% non-specific amplification at the highest taxonomic ranking included two clades (Sar [n = 1] and Discoba [n = 2]) and seven phyla within the bacteria kingdom (Rhodophyta [n = 5], Pseudomonadota [n = 5], Fusobacteriota [n = 1], Verrucomicrobiae [n = 1], Cyanobacteriota [n = 58], Bacillota [n = 3], Actinomycetota [n = 15]). With bacteria, each assembly downloaded from NCBI may have had more than one genome or plasmid present and thus the number of genomes/plasmids differed from the number of assembly entries. Therefore, amplification success was determined based on number of assemblies, where 97.6% of bacterial assemblies had at least one sequence amplified using ecoPCR. The median number of mismatches was zero for the forward primer and one for the reverse. The ecoPCR output also provided the melting temperature for primers that did not have any degenerate bases. Given the reverse bacteria primer did not contain degenerates, the tool calculated median melting temperature was 50.1°C. The bacterial primers were also tested against the 16S eukaryotic sequences, where 23,427 sequences had been amplified in silico, belonging to 12,002 accession numbers, none of which were arthropod sequences. The median amplicon size was 133 bp, the median melting temperature was 48.64°C, and median forward and reverse primer mismatches were one and two, respectively. Fungi had an amplification success rate of only 29.1% and a median of zero mismatches for both primers. Fungal primers had non-specific amplification at 0.08%, where the phylum Streptophyta (n = 4), kingdom Metazoa (n = 1), and clade Sar (n = 5) were included. Lastly, 97.1% of arthropod sequences were amplified in silico with zero median mismatches for both primers and a median melting temperature of 50.06°C. The non-specificity rate against 16S eukaryotic sequences is 37.2% with a median of two mismatches for both primers, however median melting temperature is 37.41°C. Arthropod primers were also tested against the bacterial genomes and plasmids, where only two sequences were amplified: one from Mycoplasma tauri (3 mismatches for both primers and amplicon size of 143 bp) and the other from a Wolbachia endosymbiont of Corcyra cephalonica (rice moth; 3 mismatches for the forward and 0 for the reverse primers and an amplicon size of 142 bp).

Download:

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

Table 5. Summary of in silico analysis using ecoPCR.

https://doi.org/10.1371/journal.pone.0343035.t005

The plant, bacterial, and arthropod primer pairs demonstrate a high amplification success and high specificity. Since ecoPCR is unable to determine melting temperatures for sequences when using primers with degenerate bases, median temperatures were not calculated for plants and fungi. Considering the higher percentage of non-specific amplification of arthropod primers (37.2%), the median melting temperature was 37.41°C, which is well below the annealing temperature of 57°C, and thus this rate would drastically reduce in vitro, as the annealing temperature should be a few degrees lower than melting temperature. With the fungal primers, only 29.1% of sequences were amplified, which was lower than expected. However, in vitro experiments indicated that these primers amplified all single source species successfully as observed in Radwan and colleagues [33].

As with most scientific studies, limitations are often present. One notable challenge when working with environmental samples is the inability to detect all species that may be present, or the potential of detecting non-target species. The primers selected for these assays were found to be the best performers compared to many others identified from the scientific literature. It is impossible to screen primers in wet-laboratory experiments against every species on Earth to demonstrate 100% specificity. Therefore, there is the possibility that the primer pairs validated in this paper could amplify non-target taxa. Notably, results from in vitro and in silico experiments have indicated that each of the four primer pairs are highly specific, and non-specificity occurrence is low. We have identified low level cross-reactivity with the arthropod primers, which amplified mitochondrial DNA from two vertebrate species, however since there is currently not a qPCR assay on the market to quantify total arthropod DNA, this assay is still an important starting point for quantification. Given the broad range of taxa screened during this validation, the specificity of the arthropod primer pair is still considered high, however researchers should take caution if they suspect cross-reactivity. Implementation of these assays into a laboratory should consider the laboratory’s own validation studies, including specificity, especially if bulk environmental sources other than soil and dust are intended for use. Furthermore, since the plant and fungal assays had standard curve slopes that fell outside of the ideal range, additional validation studies exploring primer annealing temperature effect on PCR efficiency should be undertaken. Given this study only assessed inhibition with target DNA, validations should also include an assessment of non-target DNA in the presence of inhibitors. Future work is needed to modify the arthropod primer pairs to reduce this cross-reactivity, noting that in silico screening of all arthropods and vertebrates is not feasible. Given the 1 ng total input of DNA and low-level amplification observed here (indicated by higher C_q values), it is hypothesized that non-target species would contribute to a lower signal compared to the overall sample concentration of arthropods in a mixed DNA sample of unknown origin, however further investigation is required to support this hypothesis.