RNA viruses affect most of the population, especially in developing countries, and are the major cause of emerging and re-emerging diseases in humans [1–3]. One of the most important problems with understanding RNA viruses, including fitness and diversity, is their propensity for rapid change due to its high mutation rate because of lack of proofreading property in the encoded RNA polymerase [3–5]. High mutation rate and high replication rate results in quasispecies even in a single host [4–7] and the presence of quasispecies correlates with enhancement of virulence because of the flexibility of the population of the virus to adapt easily with the dynamics of the host [6]. This phenomenon has been observed in poliovirus, where the presence of quasispecies cooperated the entrance of a neovirulent clone in mice [6], and in evasion of immunity of Hepatitis C Virus [7].

Current diagnostic methods can detect the virus directly, indirectly, fluorescence-based emerging technologies and immune-based. Indirect detection is achieved by detection of virus particles (antigens) or RNA. Fluoresce-based are usually molecular procedures that are used to detect RNA using reverse transcriptase PCR (RT-PCR), which is sensitive and specific, but is highly prone to false-positive results due to easy contamination [8, 9]. The most common molecular technique is using a fluorescent dye that binds to dsDNA and using a calibration curve to quantify the amount of DNA. Another fluorescent-based technique is using probes called molecular beacons, which consist in a stem-loop that, when self-hybridized, places the fluorophore on the 5′ end in close proximity to the quencher at the 3′ end, thus quenching the fluorescence. When the molecular beacon encounters its target, it changes to an open conformation and hybridizes with the target leading to the fluorophore to emit a signal. Tentacle Probes are modified molecular beacons because they have a single stranded DNA attached to the stem-loop by a 9-mer chain of polyethylene glycol (PEG). The ~15 nt single stranded DNA is called capture sequence because, as it is implied by its name, it captures the DNA strand and as a consequence it increases the local concentration of the stem-loop around the DNA strand from a range of nM (free probe) to mM (capture probe bound), increasing the probability of the stem-loop to bind to the target [10, 11]. This technology was successfully used for the detection of Yersinia pestis and Bacillus anthracis using the bacterial DNA [10], but it has not been tested with any RNA organism or virus.

In this study, we proposed a fast, cost-effective and sequence-specific molecular method for detection of RNA viruses. For this purpose, a virus model for RNA viruses, Lymphocytic Choriomeningitis Virus (LCMV) was used. The method intends to differentiate two strains of LCMV: Clone-13, which causes a chronic infection in mice, versus the wild type Armstrong strain, which causes an acute infection in mice. The difference in type of infection is due to a single mutation in the S segment of LCMV, C855T, which is a missense mutation observed in an amino acid change in the receptor protein GP1 (F260 L). The proposed method uses Tentacle Probes coupled with qPCR and showed that the designed probes to detect Armstrong and Clone-13 strains could differentiate them even with only a single mutation in the RNA that was obtained both in vitro and in vivo. Results also showed that the Clone-13 Tentacle Probe could detect Clone-13 strain with as little as 10% of presence in a sample with a mixture of both strains.

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 C_T shifts to a higher cycle due to the initial amount of DNA.

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.

Fast and cost-effective detection methods for viruses are currently needed for opportune diagnosis and treatment of viral disease. In this study, we tested the specificity of Tentacle Probes to detect one strain of LCMV that only differs from the other strain in three point mutations, specifically one missense mutation in the S segment. The mutation that was studied causes a change in one amino acid of the receptor protein GP1, which also changes the type of infection induced (chronic vs. acute) and is the main reason why we chose this mutation above the other two mutations present compared with the wild type. The design of the probe for the detection of each strain was following the calculations proposed by Satterfield et al. [11] In here, the calculated thermodynamic parameters (i.e. enthalpy, entropy) with DINAMelt application were used to calculate the equilibrium constants and therefore estimate the probe fluorescing when is bound to the target and non-target (Table 1). The predicted amount of probe fluorescing for both probes did not agree as precisely with what was observed. For the Clone-13 probe, the predicted difference in fluorescence between the target and the non-target was 50% higher, while the fluorescence obtained experimentally was ~40% higher, on average. A similar behavior was observed with the probe designed to detect Armstrong strain. This can be attributed to the fact that the predicted calculations did not take into account any kinetic factors, but only thermodynamic. Kinetic factors, like size of the probe, not only can be as important as the thermodynamic ones, but also can define entirely how much time a step in the qPCR might need in order to achieve the conformation desired in the case of the stem-loop (open or closed) [11, 18]. Nevertheless, the predicted fluorescence of the probe bound to the target or non-target is a good guide to choose the most suitable combination of capture and detection probe that will favor hybridization with the target.

Validation of the method was described in detail by optimizing the qPCR conditions for each probe, including annealing temperature, MgCl₂ concentration and starting sample DNA concentration (Figs. 3, 4 and Additional file 1). The annealing temperature was expected to be higher than the melting temperature of the individual regions (capture and detection probes) because the principles of cooperativity indicate that the affinities of each part of the probe should be slightly weaker than the affinity required to achieve binding of the whole probe. However, as mentioned before, calculations did not take into account important kinetic factors such as the size of the whole probe and each part in separate [18] that might have affected the variance in the experimental annealing temperature compared to the predicted temperature. It is important to mention that the curves observed with the Armstrong probe at all temperatures did not show the exponential region (sigmoidal shape) usually observed with qPCR curves, as it was observed with the Clone-13 probe. This is an indication that the Armstrong probe is not as fit as the Clone-13 probe and is in agreement with the thermodynamic predictions of fluorescence with the target and the non-target, where the differentiation with the Armstrong probe is not as pronounced as the differentiation predicted with the Clone-13 probe (Table 1).

Reproducibility of the assay was observed for different concentrations of each target with the correspondent probe. Clone-13 probe showed high sensitivity at all concentrations compared to the Armstrong probe, which had a higher C_T at lower concentrations (1 and 0.5 nM). However, it is important to highlight that for the Armstrong probe one of the false negatives seems to not have been amplified the amplicon at all (purple flat line). This error could be due to other experimental errors (i.e. some mistake in the master mix preparation) rather than a lack of specificity of the probe because each unknown had only one replicate. As observed with the other experiments, this experiment also reflects the strength of both of the probes, which confirmed the initial predictions of the thermodynamic calculations.

MgCl₂ concentration is another important factor because it has a high contribution in ionic strength of the reaction and therefore influence directly into the thermodynamics. The calculations were performed to use 5 mM of MgCl₂ and 1.5 mM for the Clone-13 and Armstrong probe, respectively. The concentration used for the calculation of the Clone-13 probe was the one recommended by Satterfield et al. and the concentration used for the calculations of the Armstrong probe was the one recommended by the manufacturer to use for an optimal reaction using Platinum Taq polymerase. There was an agreement of the expected optimal concentration of MgCl₂ with the one observed for both the Clone-13 probe and Armstrong probe. This shows that the ionic strength used in the calculations accurately predicted the optimal salt concentration in the reaction.

It was observed that the calculations showed a better differentiation between target and non-target for the Clone-13 probe than for the Armstrong probe (Table 1). The difference between the two concentrations of MgCl₂ for the calculations might be one of the factors for the lower efficiency of the Armstrong probe compared to the Clone-13 probe. This difference in efficiency of detecting one strain or the other can also be attributed to the thermodynamic advantage of the Clone-13 probe due to the presence of the cytosine in the mutation site, instead of the thymine like in Armstrong probe. Cytosine will hybridize with a guanine forming three hydrogen bonds, meaning the base-pairing hybridization will be more stable than the hybridization of the mismatch (anomalous cytosine-adenine pairing), which means that the hybrid between the probe and the target (C-G) is favored by stability. For the Armstrong probe, the thymine from the probe will hybridize with the adenine forming only two hydrogen bonds with the target instead of the hybrid with the non-target mismatch (anomalous thymine-guanine pairing), in which case it would form three hydrogen bonds. Although the probability of the hybridization of the Armstrong probe with the mismatch is low, due to the necessary conversion from the keto-thymine to the enol-thymine conformation, the hybridization of thymine and adenine (Armstrong probe-target) is less favored thermodynamically than the hybridization guanine-cytosine (Clone-13 probe-target), in terms of Gibbs free energy.

Experimentally it was shown that each probe designed could differentiate effectively one strain from the other. The blind tests performed with each probe to determine the specificity and sensitivity gave a sense of how accurate the probe could detect the target strain. The Clone-13 probe showed a better specificity giving results with only 6% of false-negatives. On the other hand, the Armstrong probe showed results with 13% of false-negatives and 20% of false positives. It was expected that the Armstrong probe would have a higher probability of obtaining an inaccurate result, due to the smaller difference of expected fluorescence with the target vs. non-target in the initial calculations. Nevertheless, these percentages could be lower and further experiments with a bigger sample size (>50 samples) could be used to determine the percentage of false-positives and false-negatives more accurately.

This method reduced false-positives compared to common real time PCRs used for other RNA viruses such as Dengue [8] and common PCRs with a difference of only one point mutation between strains, as observed when using specific primers to detect each strain (Additional file 3: Figure S3). The reduction of false-positives with this method is an advantage when using it with clinical samples overall in tropical regions, where many viruses with similar symptoms affect the same population (i.e. Dengue and Zika virus) [16, 19]. Furthermore, detection of viral strains in serum using this method was established when we could accurately detect Clone-13 virus in the serum samples from infected mice (Fig. 6).

Also, the probe tested could effectively differentiate each strain in virus mixing experiments with the two strains. Clone-13 probe can detect down to 10% of Clone-13 diluted in Armstrong (Fig. 7). The detection of one strain in a mix of other strains of viruses has a high impact in a detection method because it simulates the conditions of many clinical samples. This experiment in particular intended to illustrate two important viral phenomena: the presence of quasispecies in the population of viral strains, even in the same host, and co-infection of two viruses in one host. Infections with RNA viruses can often result in the formation of quasispecies of virus strains that can rapidly adapt and can develop resistance to anti-viral drugs or vaccines [4]. Quasispecies of viral strains differ between them in a few mutations, even in only one mutation [2, 4, 6, 20]. Accurate detection of one of these quasispecies over another one can be crucial to determine possible treatment of the infection, the pathogenicity, and the evolution of a determined virus in a population. Another scenario where detection of one viral strain in a mixed sample is essential is in viral co-infections. Co-infection of two unrelated viruses that affect the same population is very common in the “real world” [21]. The presence of two or more viruses in a host can modulate the immune response and/or alter the disease [21, 22]. Therefore, the use of Tentacle Probes and qPCR for the detection of RNA viruses can be widely applied in many situations where high specificity and reduction of false-positives is needed.

Herein, we showed that each Tentacle Probe could detect with specificity each strain studied in samples with a single strain, and, as low as 10% of the target, in samples with a mixture of strains. The thermodynamic calculations performed to determine the most suitable sequence of the probe to detect the target vs the non-target, were accurate to establish the necessary ionic strength and starting DNA sample concentration, but not as accurate in terms of the fluorescence detected by each probe. Nevertheless, the calculations are a good guide to determine the best detection and capture sequence to use.

Although currently there are other methods to detect quasispecies like sequencing, they are still cost-prohibitive. This study describes a fast cost-effective detection method that was designed for differentiating RNA viruses, even between close related viruses or quasispecies of a virus. This method could potentially be useful not only to measure virus diversity, in terms of measuring absence/presence of known variants, but to understand virus evolution, pathogenicity of the virus, and to determine possible treatment of infection.