Lentivirus titration by p24 ELISA
All lentivirus samples were titrated by p24 ELISA to confirm the physical titer. The lentivirus particles were diluted in PBS with 10% FBS by serial dilution. After determining the dilution range within the ELISA kit linear detection range, 3 serial dilutions of each sample were used for titration. All 3 dilutions resulted in similar p24 titer (data not shown). Interestingly, the p24 ELISA titration method from our experiments showed different titers compare with other titration methods (Table 2). The discrepancy could be a result of different lentivirus purification methods that lead to the different purity of the final lentivirus or different p24 standards used. Free p24 protein was also measured in the lentivirus sample without the lysing process. Because the p24 protein is covered by viral envelope, only free p24 protein can be detected by ELISA when the lentivirus is intact. ELISA results showed that about 2% of p24 protein is free p24 (free p24: 2.8 × 109 ± 1.5 × 108, total p24: 1.4 × 1011 ± 1.1 × 1010). Calculated from the molecular weight of p24, 1 pg of p24 is equivalent to 2.5 × 107 molecules. Each lentivirus contains about 2000 p24 molecules, therefore, 1 pg of p24 is about 1.25 × 104 lentivirus particles. The equation for lentivirus titer calculation is: p24 amount (pg) × 1.25 × 104 × dilution factor.
Viral RNA extraction efficiency
To determine the lentiviral RNA extraction efficiency, SARS-CoV-2 RNA Fragment 1, a NIST research grade test material (RGTM 10169) was used as a calibrator. After the lentivirus particles were lysed in the lysis buffer, 2 × 106 copies of RGTM 10169 were spiked into the lysis mixture followed by viral RNA extraction as described above. The RNA samples were serially diluted and used as a template for one-step RT-ddPCR. Three assays were used to detect the copy numbers of RGTM 10169 in the direct RT-ddPCR. RNA recovery rate was calculated as the percentage of detected copy numbers versus total spiked in copy numbers. As shown in Fig. 1, the RGTM 10169 RNA recovery rate determined by all 3 assays (Targeting two independent regions of the SARS-CoV-2 nucleocapsid gene (N1, N2) and envelope gene E-Sarbeco (SarE)) was between 86 and 93%. Interestingly, operators 1 and 2 in our lab obtained very similar RNA recovery rate. Sample A showed the highest RNA extraction efficiency (93%) and Sample C showed a relative lower recovery rate (86–89) % (Fig. 1). The extraction efficiency of the spiked-in RGTM 10169 RNA was used to evaluate the recovery rate of released LV RNA genome from the lentiviral particles. The pre-extraction LV RNA genome copy concentration was calculated by using the detected post-extraction LV RNA genome copy concentration normalized to the extraction efficiency of RGTM.
RNA recovery rates from 2 independent operators calculated by spike-in RNA. In each LV sample (A, B, and C), 2 × 106 copies of RGTM 10169 were spiked in after viral particles were lysed with lysis buffer. The RNA recover rates were calculated by percentage of detected copy numbers of each detecting target within the RGTM 10169 RNA out of total spiked in copy numbers. Bars present mean ± SD, n = 3.
Establish one-step direct RT-ddPCR using different primers/probe sets to titrate lentivirus RNA genome
Lentiviral RNA preparation is a step that may cause variations for lentivirus particles titration. One-step direct RT-ddPCR could reduce the sample handling time and the variability introduced by RNA isolation. Five sets of ddPCR assays were designed and used to titrate the LV genome, as shown in Fig. 2 by using a GSK third generation of LV as an example to illustrate the genes and primer locations. We first used a lentivirus DNA sequence as a template to validate the primers/probe assays. As shown in Fig. 3A, all 5 sets of primers/probe can detect lentiviral DNA genome at the same detection level in different dilutions, indicating that the primer/probe sets are equally efficient in amplifying lentiviral DNA. The linear correlation of each primers/probe set was excellent, with R2 > 0.997. We then tested 5 sets of primers/probe for titrating lentiviral RNA genome in the direct RT-ddPCR reactions. The LV samples were serially diluted and used for direct RT-ddPCR without any treatment. All 5 sets of primers/probe detected lentiviral RNA genome without RNA extraction (Fig. 3B,C,D). All 5 sets of primers/probe had near perfect linearity with different dilution of lentivirus in all 3 sources (Fig. 3B,C,D). Interestingly, the highest titer obtained by RRE primers/probe and p24 ELISA titer are within the same order of magnitude.
Schematic diagram showing the locations five sets of ddPCR primers/probe for LV genome quantitation by using a Sample A LV sequence as an example.
Using 5 primers/probe sets to titrate lentivirus by detecting RNA genome copy number. (A) By using lentivirus genome DNA as template, 5 sets of primers/probe were all able to detect the viral genome DNA and showed equivalent PCR efficiency. (B–D) All 5 sets of primers/probe detected lentivirus RNA genome in one-step direct RT-ddPCR reactions. Sample C does not contain GFP gene, therefore there were only 4 assays displayed in Fig. 3D. Input copy number was determined by viral titer provided by the companies. All data presents mean ± SD, n = 3.
Heat inactivation of lentivirus particles increased one-step RT-ddPCR detection sensitivity
We showed that the primers/probe sets can detect lentiviral RNA genome without RNA extraction (Fig. 3B,C,D). Because interaction between the viral RNA genome and capsid protein could interfere with direct digital PCR efficiency, we hypothesized that heat inactivation of the lentivirus can increase the detection level of lentiviral RNA. To test this hypothesis, we treated the lentivirus particles at 56 °C, 60 °C, 65 °C, 70 °C, and 90 °C for 15 min. We found that low temperature treatment did not increase the detection level (56 °C and 60 °C). But 90 °C heat treatment of the lentivirus increased the detection of 5′ LTR more than 17-fold comparing with the reactions from the lentivirus that were not treated (Fig. 4). The detection of the other 4 assays also increased two to four- fold comparing with samples without heat treatment (Fig. 4). We further compared the detection levels of 90 °C treatment for 10 min and 15 min. A slightly higher detection level was found after 10 min of treatment (Fig. 4). A comparison between 90 and 95 °C heat treatment did not show a significant difference in detection level of any primer/probe set (Supplemental Fig. S2). Therefore, we used 90 °C, 10 min heat inactivation for the rest of direct RT-ddPCR assays.
Optimization of one-step direct RT-ddPCR reaction. Lentivirus particles were treated at different temperatures for 10 min to 15 min before one-step direct RT-ddPCR reaction. Bars present mean ± SD, n = 3.
Effect of diluents on RT-ddPCR detection level
We then investigated whether the diluents used to dilute the lentivirus affect RT-ddPCR detection level. Our first hypothesis was that adding proteins such as fetal bovine serum (FBS) to PBS will reduce lentivirus particles binding to the tube wall and increase the RT-ddPCR detection level. To test this hypothesis, we performed serial dilution of Sample C in PBS and PBS containing 10% FBS before heat inactivation and RT-ddPCR. As shown in Fig. 5A–D, there was no difference of LTR detection level at higher concentrations, and a 25% to 52% increase in detection at the lower concentration. For the other 3 assays (Psi, RRE, and WPRE), adding 10% FBS to PBS all increased detection level (39–117%) as shown in Figure. We next tested whether other diluent such as RSS would protect RNA from degradation after RNA releasing by heat inactivation. The lentivirus sample A was treated in 2 different protocols: (a) serially diluted into different diluents then the final diluted sample was heat inactivated at 90 °C for 10 min; or (b) first diluted 10× in different diluents, then heat inactivated at 90 °C for 10 min and then further diluted in RSS containing 20 ng/µL yeast tRNA (tRNA/RSS) to the targeted dilution for RT-ddPCR. As shown in Fig. 5E,F, diluting the lentivirus particles in the diluents before heat inactivation helped prevent the RNA degradation in the solutions. These results also indicated that dilution of the lentivirus samples in tRNA/RSS before heat inactivation resulted in the highest detection level for LTR, Psi, and WPRE amplicons.
Effect of diluent on the one-step RT-ddPCR detection. (A) Comparison of LTR one-step RT-ddPCR results from the LV diluted in PBS and 10% FBS/PBS. (B) Comparison of WPRE one-step RT-ddPCR results from the LV diluted in PBS and 10% FBS/PBS. (C) Comparison of Psi one-step RT-ddPCR results from the LV diluted in PBS and 10% FBS/PBS. (D) Comparison of LTR one-step RT-ddPCR results from the LV diluted in PBS and 10% FBS/PBS. (E) Comparison of one-step RT-ddPCR results of 5 assays from the LV diluted in different diluents before heat treatment. (F) comparison of one-step RT-ddPCR results of 5 assays from the LV diluted in different diluents after heat treatment. All data presents mean ± SD, n = 3.
Measure the integrity of lentiviral RNA genome by one step direct RT-ddPCR
Within the lentivirus particles, there could be empty particles without RNA genome, truncated RNA genome, or intact RNA genome. To measure the lentiviral RNA genome integrity, we first synthesized a GSK lentiviral RNA genome in vitro as calibrator. Dilution linearity and reverse transcription efficiency (RT efficiency) was analyzed in one-step RT-ddPCR and the results are shown in Fig. 6A and Supplemental Fig. S3. All 5 primers/probe sets had very good dilution linearity and the RT efficiency varied from 39.7 to 62.4% (Fig. 6A and Supplemental Fig. S3, Table S1).
LV RNA genome integrity. (A) RT efficiency of 5 primer/probe sets assays. All data presents mean ± SD, n = 3. (B) Comparison of LV RNA genome integrity determined by direct RT-ddPCR and by RT-PCR on extracted RNA using in vitro transcribed RNA as calibrator. Average of 4 experiments, bars present mean ± SD. (C) RNA size analysis by Bioanalyzer on RNA extracted from Sample A.
We utilized the synthesized lentiviral RNA (derived from the GSK sample) as a calibrator to measure the integrity of RNA that was packed inside the lentivirus particles. The detected copy numbers of each element from lentiviral RNA genome were normalized to its RT efficiency, and the element with the highest detection level was assigned as 100% (Fig. 3A and Table S1). For most cases, RRE had the highest detection level. The highest detected element was assigned as 100% for each tested lentivirus and other elements were normalized to the detection level of the highest element. Theoretically, identical detection levels of all elements will be achieved for a LV with full RNA genome after RT efficiency normalization. As shown in Fig. 6B, 5′-LTR was the lowest detectable element, which is about 16.34% ± 8.07% for Sample A (Average results of 4 experiments performed on the same batch of Sample A), 21.83% ± 0.63% for Sample B (data not shown), and 15.94% ± 0.48% for Sample C (data not shown). The percentage of LTR to the highest detected element did not show statistic difference between direct one step RT-ddPCR and RT-ddPCR on RNA extracted from LV particles (Fig. 6B). We further confirmed the RNA sizes on Bioanalyzer by using Agilent RNA 6000 nano kit. The intact sample A lentiviral RNA genome is around 4000 nt. As shown in Fig. 6C, the majority of the RNA isolated from the lentivirus particles is at 25–500 nt sizes. There is only a small amount of RNA showed around 4000 nt. This data indicated that most of the lentivirus particles had truncated RNA genomes, especially in the 5′-LTR region.
Comparison of one step direct RT-ddPCR and RNA extraction RT-ddPCR method
We compared the titer units detected by these RT-ddPCR with and without RNA extraction from viral particles. Interestingly, the RRE primer/probe set had similar titers detected by both methods. For sample A, we detected, we detected 1.05 × 1011 ± 3.61 × 109 from one step direct RT-ddPCR and 1.35 × 1011 ± 1.30 × 108 from RNA extraction sample (Fig. 7 and Table 2). Similarly, we detected 1.69 × 1010 ± 8.19 × 108 and 1.56 × 1010 ± 2.72 × 109 from Sample B, respectively. Slightly higher level of detection was observed from Sample C by one step direct RT-ddPCR at 1.80 × 1010 ± 1.53 × 108 comparing to 9.29 × 109 ± 2.14 × 108 by RNA extraction RT-ddPCR method (Fig. 7, Table 2).
Titer comparison between one step direct RT-ddPCR and RNA extracted from LV using RRE primer/probe set. Direct ddPCR indicates one step direct RT-ddPCR (Data from all three samples). ***indicates P < 0.001; NS, indicates not significant.