Assessing reliability and accuracy of qPCR, dPCR and ddPCR for estimating mitochondrial DNA copy number in songbird blood and sperm cells

Primers and synthetic DNA

We designed a new set of primers targeting a range of passerine species and a fragment of their mitochondrial 16S gene. The primer sequences (Forward 5′-ATTATTGAGCGAACCCGTCTC-3′ and Reverse 5′- TTCACAGGCAACCAGCTATC-3′) were designed following the method described by Amer, Ahmed & Shobrak (2013). In brief, we downloaded multiple mitochondrial 16S sequences from a range of songbird species (infraorder Passerides), covering several genera (Spinus spinus, Emberiza schoeniclus, Coccothraustes coccothraustes, Fringilla coelebs, Pyrrhula pyrrhula, Parus major, Luscinia svecica) from GenBank (http://www.ncbi.nlm.nih.gov) (Table S1). These sequences were then aligned in the Geneious Pro R10 software (https://www.geneious.com) with the Clustal Omega multiple alignment function and consensus sequences for each species were generated. Then, the consensus sequences were aligned, and primers were created using the primer design function. The specificity of the newly designed primers was visually assessed by inspecting the multi-species alignment and confirmed using the Integrated DNA Technologies (IDT) PrimerQuest tool (PrimerQuest™ program; IDT, Coralville, Iowa, USA). Additional validation was conducted using the primer blast function from the NCBI website (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). Following these in silico validation steps, we performed an additional in vitro validation with PCR and ddPCR on DNA extracted from several songbird species: S. spinus, Turdus pilaris, Phylloscopus trochilus, L. svecica, Fringilla montifringilla, P. pyrrhula, Regulus regulus, C. coccothraustes, E. schoeniclus. We used DNA previously extracted from sperm samples and stored at the DNA bank at the Natural History Museum of Oslo (NHMO). Sample information is provided in Table S2. These thorough validation steps, including temperature gradient analysis, were performed to ensure the specificity and optimal amplification of the target fragment.

Additionally, we designed a synthetic DNA fragment of mitochondrial 16S (5′-ATTATTGAGCGAACCCGTCTCTGTGGCAAAAGAGTGGGATGACTTGTTAGTAGTGGTGAAAAGCCAATCGAGCTGGGTGATAGCTGGTTGCCTGTGAA-3′), following a method described by Han et al. (2023). We used the same multiple sequence alignment as described above, and the consensus sequence was created with most frequent nucleotide for each base, containing no ambiguous nucleotides—only A, T, C, G. Then, we tested this sequence against our previously designed primers by using “Test with Saved Primers” function in Geneious Pro R10 software, which ensured the match between primers and the fragment of our gene of interest. Primer forward and reverse sequences, as well as 16S fragment were ordered from Integrated DNA Technologies (IDT, Inc., Coralville, Iowa, USA). We used this synthetic fragment of mtDNA to determine limits of detection and quantification with the three methods used in this study.

Samples

We selected the Eurasian siskin (Spinus spinus) as our study species because they are common breeding birds in Norway and are easily caught at bird feeders. Siskins have relatively long sperm cells (219 µm), carrying long midpieces (190 µm) (Omotoriogun et al., 2020), which could indicate they are assembled from a high number of mitochondria (Knief et al., 2021). Blood and sperm samples used in this study were collected from male birds during the breeding seasons 2021–2022. A total of 10 blood samples and 10 sperm samples were collected from five individuals in 2021 and 13 in 2022 (Table S3). Birds were caught using mist nets at their breeding sites in southern Norway and released at the trapping site immediately after sampling. Sperm samples were collected via cloacal massage in a similar way as described in Wolfson (1952). Two males were sampled twice. The ejaculate was collected in a micro-capillary tube and immediately mixed with phosphate-buffered saline (PBS) (Kleven et al., 2008). This was done to prevent the sperm cells from clumping together and to enable pipetting. Sperm samples were then purified through a series of centrifugation and washing with PBS, and frozen at −80 °C until DNA extraction. Blood samples were collected through brachial venipuncture (10–20 µL) and stored in 96% ethanol. Sampling was conducted in adherence to ethical guidelines for use of animals in research and with permission from all relevant local authorities and approved by the Norwegian Food Safety Authority (permit no. 23294 and 29575) and The Norwegian Environment Agency (permit no. 2021/39021).

DNA was extracted from blood samples using the E. Z. N. A. Tissue DNA kit (Omega Bio-Tek, Inc, Norcross, GA, USA) following the manufacturer’s guidelines. DNA was extracted from sperm samples using the QIAamp DNA Micro Kit (Qiagen, Inc. Valencia, CA, USA) following a protocol for sperm samples as described by Kucera & Heidinger (2018). We made two changes from the protocol; (i) the mixture of the sample and dithiothreitol (DTT) buffer was incubated for 2 h in order to ensure the proper lysis of sperm cells and the release of DNA and (ii) the final elution was carried out with 25 µL of MilliQ water and repeated twice to ensure a maximum DNA yield. Following DNA extraction, total DNA concentration from each sample was measured with Qubit 2.0 fluorometer with the DNA High Sensitivity Assay (Invitrogen, Waltham, MA, USA). Sample concentrations were then normalized to a concentration of 0.5 ng/µL using MilliQ water before further downstream analysis. All DNA extractions were performed in a pre-PCR dedicated laboratory, all the bench surfaces were wiped with 5% deconex solution and absolute ethanol before and after the procedures.

qPCR

Samples were analyzed on a Bio-Rad CFX96 Real-Time System (Bio-Rad Laboratories, Hercules, CA, USA) using the primers described earlier. Quantitative PCR reactions were conducted in a 20 µL final volume containing 10 µL QIAcuity EvaGreen Mastermix (Qiagen, Hilden, Germany), 8.5 µL MilliQ water, 0.25 µL of each primer (10 µM) and 1 µL of template DNA at 0.5 ng/µL (see the sample description above). The qPCR program was as follows: initial denaturation at 95 °C for 2 min 10 s, followed by 40 cycles of denaturation at 94 °C for 15 s, annealing at 55 °C for 15 s and extension at 72 °C for 15 s. This was followed by 1 min at 72 °C and held at 4 °C until the amplified samples were removed from the qPCR machine. The plate with sperm and blood samples also included the standard dilutions which were used to calculate starting concentrations of the DNA samples. The change in fluorescence intensity was measured at the end of each extension step, and results were analyzed using the CFX manager software (version 3.1; Bio-Rad, Hercules, CA, USA). Quantification cycle (Cq) values were determined by manually establishing a threshold line within the exponential amplification phase across both amplification plots.

dPCR

Samples were analyzed on a nanoplate-based QIAcuity Digital PCR System (Qiagen, Hilden, Germany) using the automated workflow provided by the QIAcuity Software Suite (v.2.1). dPCR reactions were performed in a 10 µL final volume consisting of 3.33 µL QIAcuity EvaGreen Mastermix (Qiagen, Hilden, Germany), 5.44 µL MilliQ water, 0.13 µL of forward and reverse primers (10 µM) and 1 µL template DNA at 0.5 ng/µL (see the sample description above). dPCR reactions were then dispensed to 96-well 8.5k QIAcuity Nanoplates (Qiagen, Hilden, Germany), to divide all single samples into 8,500 of individual partitions before end-point PCR. The dPCR plates used in this study resulted in half as many partitions as ddPCR. Reaction conditions were as follows: 2 min initial denaturation at 95 °C, followed by 35 cycles of denaturation for 15 s at 95 °C, 15 s annealing at 55 °C, and 15 s extension at 72 °C, with a final step of 5 min at 40 °C. Results were analyzed using the QIAcuity Software Suite (v.2.1.8.23; Qiagen, Hilden, Germany). Green channel was used to detect the EvaGreen® fluorophore. Exposure duration was 500 ms for sperm and blood samples, and 300 ms for the synthetic mtDNA fragment. The common fluorescence intensity (RFU) threshold to identify positive and negative samples was set at 57 for the analysis of sperm and blood samples, and at 30 for synthetic mtDNA analysis.

ddPCR

Samples were analyzed on a Bio-Rad QX200 ddPCR System. ddPCR reactions were performed in a 20 µL final volume, consisting of 10 µL Bio-Rad ddPCR EvaGreen supermix, 0.25 µL of forward and reverse primers (10 µM), 8.5 µL of MilliQ water and 1 µL template DNA at 0.5 ng/µL (see the sample description above). After vigorous mixing, each reaction was pipetted into the sample well of a DG8 Droplet Generator Cartridge, and 70 µL of Droplet Generation Oil for EvaGreen was added to the oil wells. Droplets were then generated using the QX200 Droplet Generator (Bio-Rad, Hercules, CA, USA), and a final 40 µL volume of droplets for each sample was carefully transferred to a ddPCR 96-well plate, later sealed with pierceable foil using a PX1 PCR Plate Sealer (Bio-Rad, Hercules, CA, USA) before end-point PCR. The specifications of the system ensure the partitioning of the sample into approximately 20,000 droplets, although the final number of partitions can vary between the runs and can result in fewer than 20,000 droplets (see results). PCRs were performed on a BioRad CFX96 Real-Time System (Bio-Rad Laboratories, Hercules, CA, USA). ddPCR conditions were as follows: 10 min at 95 °C, followed by 40 cycles of denaturation for 30 s at 94 °C and annealing at 55 °C for 1 min, with ramp rate of 2 °C/s, followed by 10 min at 98 °C and a hold at 8 °C. Droplets were then read on a QX200 droplet reader (Bio-Rad, Hercules, CA, USA). Quantification data were checked using the Bio-Rad QuantaSoft software (v.1.7.4.0917). Thresholds for positive signals were determined according to QuantaSoft software instructions, and all droplets above the fluorescence threshold (10,000) were counted as positive events, those below it being counted as negative events.

Estimation of the limit of detection and limit of quantification

The MIQE guidelines (Bustin et al., 2009) define the LOD as the lowest concentration at which 95% of the positive samples are detected. The definition of the LOQ varies between different studies, but is usually defined as the lowest concentration at which replicates show a CV of or less than 35% (Forootan et al., 2017; Brys et al., 2021). To establish the LOD and the LOQ, we performed a 10-fold dilution series with the synthetic mtDNA fragment (see the primers and synthetic DNA section). The synthetic DNA was used to avoid nonspecific amplification and to ensure sensitivity of the PCR assays (Han et al., 2023). The dilution series ranged from 2.50E−04 ng/µL to 2.50E−12 ng/µL, and included nine dilution points.

Ten technical replicates of each dilution point were analyzed with each quantification platform, and a minimum of four negative controls were included on each plate. The serial dilution further allowed us to quantify the mtDNA detected with qPCR using the standard curve generated. The LOD for ddPCR, dPCR and qPCR were calculated using dedicated R scripts as in Hunter et al. (2017). We also followed the method outlined by Klymus et al. (2020) to assess the LOD and LOQ for qPCR assays. Thus, the LOD for qPCR was calculated using both methods (Hunter et al., 2017; Klymus et al., 2020) to assess whether they will return similar results. Both scripts resulted in the same LOD for qPCR. The LOQ for ddPCR and dPCR were estimated following the threshold and methods described in Forootan et al. (2017), Brys et al. (2021).

Statistical analyses

In the qPCR analysis, the concentration of copy number per µL was estimated using the standard curve generated from the serial dilution of the synthetic mtDNA fragment (Bustin et al., 2009). In the dPCR and ddPCR assays, the copy number per µL was calculated using the QIAcuity Software Suite and QuantaSoft respectively, using the fraction of positive and negative partitions based on Poisson distribution (Coudray-Meunier et al., 2015). Coefficient of Variation (later referred to as CV) was calculated for quantification results as the ratio between the standard deviation and the mean multiplied by 100. Smaller CV values across replicates indicate better repeatability. All statistical analyses were performed in R v4.3.2 using RStudio/2023.12.0+369 and the packages tidyverse v2.0.0 (Wickham et al., 2019), and vegan v2.6-4 (Oksanen et al., 2022). The methods were compared using several statistical tests: we compared the results of the synthetic mtDNA quantification with dPCR and ddPCR using Welch t-test to allow for unequal variance. We compared the sperm and blood mtDNA quantification results using one-way analysis of variance (ANOVA), followed by Tukey’s Honest Significant Difference test if significant. Plots were generated using ggplot2 v3.4.4 (Wickham et al., 2016).

LOD and LOQ

qPCR Efficiency % ranged from 94.5 to 116.3, had slope from −3.462 to −2.984, had Y intercepts from 7.411 to 38.623, and had R² from 0.968 to 0.976 for synthetic and biologically-derived mtDNA, respectively. For qPCR, the LOD was established at 2.42 copies/µL and LOQ was established at 4.9 copies/µL. dPCR had an LOD at 0.51 copies/µL and LOQ at 2.12 copies/µL, while ddPCR had an LOD at 0.63 copies/µL and LOQ at 1.23 copies/µL (Table S4 and Figs. S1–S6).

Synthetic mtDNA

We investigated the correlation between the measured concentration and the expected concentration of the dilution series (Fig. 2; Table S5). Only dPCR and ddPCR were compared in this step, as the dilution series was used to convert qPCR Cq values into concentrations. Based on the results of the Welch t-test, mtDNA quantification did not significantly differ between dPCR and ddPCR at five of seven dilution points (Table 1). We found that ddPCR had a lower CV than dPCR at five of seven concentrations of synthetic mtDNA.

Sperm and blood mtDNA

The comparison of copy numbers estimated using the three methods (Fig. 3) revealed similar mean sperm mtDNA copy number with dPCR and ddPCR: 310.81 copies/µL and 298.26 copies/µL, respectively, while the copy number measured with qPCR was higher−507.13 copies/µL. The comparison of measured blood mtDNA concentration across the three platforms, showed more differences; mean concentrations measured with qPCR, dPCR and ddPCR were as follows: 17.19 copies/µL, 21.84 copies/µL and 36.14 copies/µL. There was no significant difference in sperm mtDNA quantification across the three methods (ANOVA, p = 0.25). However, blood mtDNA quantification was significantly different (ANOVA, p = 3.33E−06). The one-way ANOVA was followed by Tukey’s HSD test, which revealed no statistically significant difference between qPCR and dPCR (adjusted p = 0.3). In contrast, ddPCR differed significantly from both dPCR and qPCR (adjusted p = 2.1E−04 and 3.7E−06, respectively).

We calculated CV for each individual using the mtDNA copy number measured for each technical replicate across qPCR, dPCR and ddPCR (Fig. 4). Mean CV across methods was <50%. In sperm samples, mean CV was 23% for qPCR, 25% for dPCR, and 9% for ddPCR, while blood samples had a higher CV with qPCR (30%) and dPCR (40%), and lower with ddPCR (7.5%). The CV differed significantly across the three methods in both sperm and blood samples (ANOVA, p = 0.03 and 4.31E−05, respectively). Tukey’s HSD test, applied to the one-way ANOVA results, indicated no significant difference in CVs between qPCR and dPCR for sperm (adjusted p = 0.89) or blood samples (adjusted p = 0.22). In contrast, ddPCR had lower CV than qPCR and dPCR in sperm samples (adjusted p = 0.088 and 0.034, respectively) and in blood samples (adjusted p = 3.24E−03 and 3.45E−05, respectively).

Finally, we explored the relationship between the CV in blood and sperm samples and the mean measured mtDNA concentration across the three methods (Fig. 5) and found that CV decreased as the concentration increased. Lower CVs were obtained when quantifying mtDNA in sperm samples, especially for qPCR and dPCR.