Quantification of viruses is the cornerstone in virology and it focuses on directly or indirectly counting the infectious viruses or the number of virus particles . Unfortunately, most of the used methods are time-consuming and expensive, as well as potentially require special instrumentation and expertise. We have previously introduced a label-free technique called the Protein-Probe, which has been applied for varying protein-related studies [35,36,37,38]. Using the protein-binding property of the Eu3+-probe, a key component of the Protein-Probe technique, we have now developed a virus particle quantitation assay, exploiting viral surface proteins. Interaction between the Eu3+-probe and viral proteins in the Protein-Probe modulation solution leads to an increase in TRL-signal upon an increase in virus particle concentration (Fig. 1). Linear TRL-signal increase upon virus binding is due to Eu3+-chelate protection when the probe brings the label close to the virus surface. This reduces TRL-signal quenching, which efficiently eliminates all signals from the non-bound Eu3+-probe in the absence of the virus. Here, we demonstrate that the Protein-Probe technique is useful in the determination of total virus count and is applicable to many enveloped and non-enveloped viruses used as models (Table 1).
TEM enables measurement of total virus count and quality estimation of virus stock
One of the main problems in virus research is the lack of a standardized way of reporting virus concentration. Thus, depending on the method and research purpose, given information may vary. Virus stock concentration is often reported as mg/mL, PFU/mL, TCID50/mL, or vp/mL, but since these units are not interrelated, direct comparison of virus stocks with different units is impossible. To develop a method to tackle and understand this problem, we first visualized four different viruses using TEM, to calculate the virus particle count of these stocks and to compare it to calculated values using total protein concentration. We focused on various influenza A viruses, with a known total protein concentration of 1 mg/mL, as these highly purified influenza A viruses have a known conversion factor from mg/mL to vp/mL .
In TEM, full virus particles on the grid appear as structures with a light outer ring and darker interior (Fig. 2). The viral particle concentrations were counted from 90 randomly taken non-overlapping images using three grids of each virus sample. The quantitation calculations yielded the average of 350 vp/image area of 77 µm2. The total grid area covered with the sample solution was 7,306,000 µm2, resulting in the virus particle count of 3.3E7 on the grid. With the 1/5 dilution of the original virus stock, the stock concentration of 1 mg/mL protein content was 1.7E11 vp/mL. This number calculated based on the TEM figures is in relatively close accordance with the calculated value using the published conversion factor and the total protein concentration, giving an estimated value of 1E12 vp/mL . However, still, the observed 6 × difference between TEM and calculated values can be counted as significant in some applications. The observed difference can originate from variations in TEM image analysis and grid preparation, as potentially not all particles applied on the grid bind to the grid as assumed. In addition, notable variation was observed between TEM images, and even viruses were heavily purified, some unidentified particles can still be visualized from the grid. However, despite these complications, and as approximately the same virus count was calculated for all the TEM-imaged influenza A viruses, A/California/07/2009 (H1N1)pdm09, A/Vietnam/1194/2004 (NIBRG-14) (H5N1), A/Hong Kong/1093/99 (H9N2), and A/Singapore/1/57 (H2N2), the TEM count values were applied for all influenza viruses in the Protein-Probe assays. This was done, as the potential error in virus quantification leads, if something, rather to underestimate the Protein-Probe sensitivity.
The Protein-Probe can detect virus particles in culture media
Obtaining high titer virus stocks requires culturing of the virus in cells. Depending on the culture media and purification steps, virus samples may contain cell culture media such as DMEM, Sf-900™ II serum-free media or allantoic fluid, as well as supplements like FBS and antibiotics. Thus, the method must tolerate e.g. high protein concentration to enable virus counting. To this end, the effect of DMEM, DMEM + 5% FBS, Sf-900™ II serum-free media, and allantoic fluid was studied in the presence and absence of the Sendai virus. Virus dilutions were made in 0.1 × PBS, and as a control, the same dilution without a virus was analyzed to monitor the background TRL-signal. As expected, the highest concentrations of DMEM + 5% FBS gave elevated TRL-signal even without the Sendai virus (Fig. 3A). The serum-free Sf-900™ II media and allantoic fluid also affected the measurements, but the effects were far less significant compared to DMEM + 5% FBS. Based on the results, 1:7000, 1:100, and 1:3000 dilutions for DMEM + 5% FBS, Sf-900™ II serum-free media, and allantoic fluid, respectively, were required to completely diminish the effect of media on signal. On the other hand, DMEM without FBS did not have any effect on the Protein-Probe detection, suggesting that FBS had the most significant negative effect. In conclusion, protein-rich solutions seem to have a negative effect on virus detection, which was expected as the Eu3+-probe utilized proteins for its binding.
Sendai virus–spiked samples were clearly detectable with the Protein-Probe method in buffer, DMEM + 5% FBS, DMEM, Sf-900™ II serum-free media, and allantoic fluid, although the background effect of especially 5% FBS and allantoic fluid were significant (Fig. 3). However, after the background signal correction, detection of the Sendai virus was evident using higher than 1000-fold dilution of 5% FBS-supplemented DMEM and allantoic fluid. A hook effect in the background reduced TRL-signals which was observed in DMEM + 5% FBS and allantoic fluid and thus dilution will need to be performed with care to avoid misleading conclusions (Fig. 3B). This effect might be due to insufficiently high total protein concentration in comparison to Eu3+-probe concentration. It also indicates that the detected viral proteins are giving better TRL-signal protection upon Eu3+-probe binding in comparison to proteins in culture media. These results demonstrated that the Protein-Probe method can detect viruses also in complex media with interfering components, such as FBS, and it shows some specificity towards viral surface proteins in comparison to proteins in culture media. This significantly increases the potential of the method for virus particle detection, as virus preparation often contains some impurities (Fig. 2). Of note, OptiPrep™ Density Gradient Medium was found incompatible due to high TRL-signal inhibition most probably caused by iodixanol in the medium (data not shown).
The Protein-Probe enables enveloped and non-enveloped virus particle quantitation
In addition to virus size, the viral surface structure may have an impact on quantification. As the effect of cell culture media was investigated only with enveloped Sendai virus, we next set a larger study to obtain information on the Protein-Probe method both with enveloped and non-enveloped viruses. Titrations were performed with seven influenza A viruses with known total protein concentration (1 mg/mL), four different enveloped and non-enveloped viruses with known vp count, and one infectious influenza A virus with known amount of infectious viruses (PFU/mL) (Table 1).
Formaldehyde-inactivated influenza A viruses (A/Singapore/1/57 (H2N2), A/Texas/50/2012 (H3N2), A/Shanghai/2/2013 (H7N9), A/Vietnam/1194/2004 (NIBRG-14) (H5N1), A/Hong Kong/1/68/164 (H3N2), A/Hong Kong/1093/99 (H9N2), and A/California/07/2009 (H1N1)pdm09) were first assayed to confirm that these viruses are equally detected (Fig. 4A). As expected, no change in the assay linear area (5E6–3E10 vp/mL) or slope was detected, when influenza A viruses were compared. Similarly, the LLD for all influenza A viruses were in the same range, from 7.5E6 to 7.4E7 vp/mL (Fig. 4A). Detected TRL-signal variation and the approximately tenfold difference in the visibility of these viruses may be addressed to variation in the protein load as the original particle count was based on the sample absorbance measurement at 280 nm. As all these viruses were formaldehyde-inactivated and stored in PBS, we also assayed infectious influenza virus A/California/07/2009 (H1N1)pdm09 stored in allantoic fluid. We observed similar linear area and slope values (0.96 inactivated and 1.03 infectious virus) with the Protein-Probe method for both A/California/07/2009 (H1N1)pdm09 virus samples after the background subtraction. Unfortunately, the concentration of the infectious A/California/07/2009 (H1N1)pdm09 virus was given as PFU/mL making the direct comparison with the inactivated virus impossible (data not shown). However, the results indicate that only a relatively small fraction of virus particles (~ 1/16000) are active in the infectious A/California/07/2009 (H1N1)pdm09 stock as calculated from the measured signals for both virus samples, and using the titration curve of the inactivated A/California/07/2009 (H1N1)pdm09. This shows the applicability of the method to enable the assessment of viral stock condition with respect to total virus count. Based on these results, we can also conclude that formaldehyde inactivation of viruses had no effect on detectability in the Protein-Probe assay. In addition to the set of the influenza A viruses, we also measured another enveloped virus, baculovirus AcMNPV, stored in Sf-900™ II serum-free media supplemented with 5% FBS and 10 µg/mL gentamicin. By subtracting the effect of the media, baculovirus was successfully detected with the LLD of 4.3E4 PFU/mL (data not shown).
As the Eu3+-probe is expected to interact with the viral surface proteins and structures, we anticipated finding differences in the detection capacity of enveloped and non-enveloped viruses. Therefore, we studied non-enveloped coxsackie-, adenovirus Ad5/3-D24, and Ad5/LacZ virus to define the limitations of the viral target to the detection. As a control, we measured enveloped Sendai virus (Fig. 4B) having a similar detection limit to influenza A viruses. The data suggests that a higher virus concentration is required for the detection of the non-enveloped virus than the enveloped viruses as the calculated LLDs were 2.8E4 (3.5E6 vp/mL) in 8 µL sample volume for the Sendai virus and for the non-enveloped viruses LLDs varied from 1.9E8 to 1.4E10 vp/mL. Also, as expected, adenovirus Ad5/3-D24 and Ad5/LacZ were detected with the same efficiency, indicating that the same virus control can be used for similar types of viruses to evaluate the virus particle number. Although lower sensitivity was obtained for non-enveloped viruses, these assays demonstrate that the detection and quantification of non-enveloped viruses can be successfully performed with the Protein-Probe. However, no clear explanation of these differences in visibility can be given at this point. We can only speculate that as we used low pH in Protein-Probe solution to prompt the virus detection with a highly negative glutamic acid–rich peptide sequence, the target virus isoelectric point (pI) might play an important role. Unfortunately, pI values determined and calculated for different viruses and viral structures vary significantly [44, 45]. This is further complicated by the presence of polynucleotide-binding regions, and in the case of enveloped viruses, the phospholipid membrane .
In addition to visibility, a distinctive difference was found regarding the linear slope of the assays with non-enveloped and enveloped viruses. As the slope values for the enveloped viruses were from 0.85 to 1.08, indicating a linear response to the increase of the virus concentration, the non-enveloped viruses systematically gave a higher slope ranging from 1.31 to 1.86. This is not well understood but the data suggests that the Eu3+-probe has increased binding/affinity with the non-enveloped viral surface as the concentration of virus increases. This also highlights the need for proper control of each type of virus. As seen clearly with different closely related influenza A viruses, signal level cannot be directly converted to virus particle number, even the slope between these viruses did not change (Fig. 4A). This comes even more significant in the case of non-enveloped viruses, leading to complications and high uncertainty to calculate virus concentration in mixed samples containing multiple virus species.
Currently, Protein-Probe is limited to samples with a virus control having known virus concentration. This is due to the fact that non-enveloped and enveloped viruses are detected differently as these two types of viruses have a diverse outer layer with varying proteins. As non-enveloped virus capsid is protein-based, one could expect the Protein-Probe to prefer these viruses. However, on the contrary, non-enveloped viruses have lower detectability possibly due to capsid’s highly beta sheet ordered and compact structure. Enveloped viruses, on the other hand, contain both lipids and proteins in the outer layer and these proteins are expected to be more exposed for the probe. Protein-Probe binding to a less compact surface enables improved TRL-signal protection, and thus LOD. In addition, results indicate that virus size has an effect on the detectability, as measured for coxsackievirus CVB1-VLP. This small virus has a low protein content per virus particle. This further limits the exact counting of unknown viruses with the Protein-Probe without a specific virus control.