The differentiation of strains or genotypes of viruses can determine and, in many cases, predict the virulence of an infection. In the case of Dengue, for example, it is important to determine the serotype as a method of surveillance in the population to try to prevent hemorrhagic fevers [16, 17]. In the case of LCMV, a different strain or genotype can determine the type of infection: chronic (Clone-13) or acute (Armstrong). For this purpose, a tentacle probe was designed to detect Clone-13 over Armstrong and vice versa. The thermodynamic parameters obtained using DINAMelt and UNAFold applications [12, 13] were used to calculate the equilibrium constants for each one of the hybridization steps of the capture and detection probe described by Satterfield et al. [11], to predict the fraction of probe fluorescing with the target or non-target. These predictions helped determine the most suitable combination of detection and capture sequence among more than ~400 possible sequences that were simulated.
For the probe designed to detect Clone-13 strain, the calculations predicted that the ratio of probe fluorescing with the target would be 54.47% versus 0.53% with the non-target (Table 1). For the probe designed to detect the Armstrong strain the percentages were 31.88% and 2.76% with the target vs the non-target, respectively. With the thermodynamic calculations it could also be determined the increase in the equilibrium constant when it takes into account the hybridization of the capture sequence (Table 1), rather than when it is not present in the probe, as in molecular beacons.
Melting curve and optimization of conditions for qPCR
To confirm the overall melting temperature calculated by DINAMelt and UNAFold of each one of the probes, a melting curve was performed at different DNA concentrations obtained from harvesting virus in BHK21 cells for the target and the non-target in each case. Although the thermodynamic calculations are a good estimate to determine the best differentiation, the prediction is not accurate because as observed in the melting curve for both probes (Fig. 2), the detection of the non-target was considerably higher than 0.053% and 2.76%, respectively. Figure 2a shows that the best range at which the Clone-13 probe differentiates the target (Clone-13) and non-target (Armstrong) DNA is around 40 °C, but annealing temperatures in PCR are known to be best in the 45-60 °C range. For this purpose, qPCRs were performed at 4 different annealing temperatures along this range: 45, 50, 55 and 60 °C. Figure 3a shows raw fluorescence counts of the target (solid lines) and non-target (dashed lines) at each annealing temperature with the Clone-13 probe. After each cycle, it is observed how raw fluorescence is increasing Although at each temperature there is a visible difference between target and non-target, the best differentiation at maximum fluorescence was observed using an annealing temperature of 50 °C with ~1.3 fold of maximum fluorescence.
For the Armstrong probe, Fig. 2b shows the melting curve for this probe and it suggests that the best range at which it can differentiate best target (Armstrong) and non-target (Clone-13) DNA is between 40 and 50 °C, but practical limits on cooling make those impractical and amplification is less reliable at lower temperatures. Four different qPCRs were performed with different annealing temperatures along this range: 42, 45, 48 and 50 °C, and were evaluated for the best differentiation between target and non-target. Figure 3b shows normalized fluorescence counts to the maximum non-target fluorescence at each temperature tested and Fig. 3c shows raw fluorescence counts for the same temperatures. Both graph suggest that the best annealing temperature for this probe is 50 °C, showing ~1.85 fold of maximum fluorescence for the target versus non-target. It is important to mention that for this probe it is not observed the exponential phase in the fluorescence curves, as it is expected in a qPCR curve, in both raw and normalized fluorescence counts (Fig. 3b and c).
For the annealing temperature, all the thermodynamic calculations were based in an assay temperature of 60 °C for each probe, although as observed with the melting curve for each probe (Fig. 2), the melting temperature for each was lower than expected, and by 60 °C the probe was completely dissociated for both the target and the non-target. The best annealing temperature for each probe was chosen because it gave the maximum differentiation between target and non-target at the same DNA concentration.
After determining which was the optimal annealing temperature to differentiate target and non-target DNA, two other parameters were optimized: the optimal initial sample DNA concentration for each reaction and the MgCl2 concentration needed in the reaction. The concentration of MgCl2 is crucial for the activity of the enzyme and even the salt effect on DNA. Three or four concentrations (0.5, 1.5, 3 and 5 mM) were tested for best results, including the concentration suggested to use by the manufacturer of the enzyme employed. Results showed that for the Clone-13 probe the most suitable concentration to use was 5 mM with ~1.4 fold maximum fluorescence of the target vs non-target, and for Armstrong probe, the most suitable concentration to use was 1.5 mM with ~1.5 fold maximum fluorescence of the target vs non-target (Additional file 1: Figure S1).
Finally, to determine the minimal concentration at which the fluorescence is best for target detection, three initial DNA concentrations were evaluated: 0.1, 1 and 10 nM. This experiment help estimate the minimal amount of DNA needed so the probe can differentiate between target and non-target. Figure 4a shows the normalized fluorescence counts obtained with different starting concentrations of sample DNA for the Clone-13 probe and Fig. 4b shows the same but with the Armstrong probe. Both graphs show that differentiation between target and non-target is best when using a sample concentration of 1 nM as a starting DNA concentration. Although 10 nM of DNA also showed good results, the best fluorescence differentiation between target and non-target is using 1 nM of sample DNA. To show reproducibility of the assay, different concentrations of viral cDNA were tested with each probe and 30-35 replicates were run. The minimum concentration (0.5 nM) was chosen according to the limit of detection of the qPCR, which for the fragment amplified is ~0.3 nM. Figure 5 shows that for both probes after 15 cycles, the curves of all concentrations would reach a measurable fluorescence. For the Clone-13 probe there is almost no difference between different concentrations even at low concentrations. For the Armstrong probe, at lower concentrations the CT shifts to a higher cycle due to the initial amount of DNA.
Determination of specificity and sensitivity for each probe
Once the conditions for the probes were optimized, a blind experiment was performed to determine the specificity of each probe. The experiment consisted in measuring the fluorescence by qPCR of 30 samples, 15 target and 15 non-target, that were organized randomly by another scientist, who knew the order of these samples. With the fluorescence results obtained, the samples were assigned as target or non-target compared to a positive and negative control, and afterwards the assignment was compared to the real identity of the samples. A sample was assigned as target if the maximum normalized fluorescence was higher than a threshold: the mean of the non-target maximum normalized fluorescence plus two standard deviations \( \left(\overline{X_{NT}}+2{\sigma}_{NT}\right) \). This was defined as the statistically significant threshold or limit of detection between target and non-target. The experiment was performed in triplicate for each probe. For the Clone-13 probe, 29 of 30 samples were assigned accurately: 14 true positives, 1 false negative, 15 true negatives, and 0 false positives (Additional file 2: Figure S2A). For the Armstrong probe, 25 of 30 samples were assigned accurately: 13 true positives, 2 false negatives, 12 true negatives, and 3 false positive (Additional file 2: Figure S2B).
Quasispecies detection
After the conditions of the qPCR for the probes were optimized and it was shown that the probes could detect specifically each strain in samples with mixtures, serum samples from mice infected with an unknown strain of LCMV, suspected to be Clone-13, were tested to detect Clone-13 or Armstrong strain. Plaque assay was performed with these samples and RNA was isolated from plaques from each sample in triplicate. qPCR with the Clone-13 probe was performed to test for the presence of this strain in the samples. Figure 6 shows the normalized fluorescence of the three samples tested compared to a positive control (target) and a negative control (non-target). It is observed that the maximum fluorescence for the three samples was above the proposed threshold in previous experiments \( \left(\overline{X_{NT}}+2{\sigma}_{NT}\right) \). To compare the results with other detection method, the DNA obtained from the samples was sequenced and confirmed the results obtained with the tentacle probe for Clone-13 detection.
Detection of a single mutation difference in a mixture of strains
Detection of specific viruses even between the same family is essential, especially in clinical samples where samples sometimes present more than one viral infection at a time or can present different quasispecies in the same host. In nature LCMV wild type strain Armstrong is commonly found, but Clone-13 strain is not found frequently even though it causes chronic disease in rodents. To further test the lower limit of detection of Clone-13 strain in a mixture of the other strain (Armstrong), mixtures of target and non-target in different proportions were tested to detect Clone-13 strain using its specific probe. Figure 7 shows the normalized maximum fluorescence of the mixtures of Clone13 and Armstrong samples ranging from 0.1-100% of Clone-13 DNA. The dashed line represents the limit of detection proposed for the blind experiment \( \left(\overline{X_{NT}}+2{\sigma}_{NT}\right) \). It is observed that the probe can detect to a lower limit of detection of approximately 10% of the Clone-13 when it is present in a mixture with Armstrong DNA.