Measuring infectious SARS-CoV-2 in clinical samples reveals a higher viral titer:RNA ratio for Delta and Epsilon vs. Alpha variants

Novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants pose a challenge to controlling the COVID-19 pandemic. Previous studies indicate that clinical samples collected from individuals infected with the Delta variant may contain higher levels of RNA than previous variants, but the relationship between levels of viral RNA and infectious virus for individual variants is unknown. We measured infectious viral titer (using a microfocus-forming assay) and total and subgenomic viral RNA levels (using RT-PCR) in a set of 162 clinical samples containing SARS-CoV-2 Alpha, Delta, and Epsilon variants that were collected in identical swab kits from outpatient test sites and processed soon after collection. We observed a high degree of variation in the relationship between viral titers and RNA levels. Despite this, the overall infectivity differed among the three variants. Both Delta and Epsilon had significantly higher infectivity than Alpha, as measured by the number of infectious units per quantity of viral E gene RNA (5.9- and 3.0-fold increase; P < 0.0001, P = 0.014, respectively) or subgenomic E RNA (14.3- and 6.9-fold increase; P < 0.0001, P = 0.004, respectively). In addition to higher viral RNA levels reported for the Delta variant, the infectivity (amount of replication competent virus per viral genome copy) may be increased compared to Alpha. Measuring the relationship between live virus and viral RNA is an important step in assessing the infectivity of novel SARS-CoV-2 variants. An increase in the infectivity for Delta may further explain increased spread, suggesting a need for increased measures to prevent viral transmission.

All samples used in this study were collected during upswings in cases in the region. Figure 4 Supplementary Information Text Extended Methods.

Generation of Viral Stocks
Clinical specimens positive for Alpha, Epsilon or Delta variants were used to isolate virus in Vero E6-TMPRSS2 cells, essentially as in 3 . Viral titer from successful isolates was determined by focus assay and this initial material was used to grow viral stocks by infecting Vero E6-TMPRSS2 cells at an MOI of 0.0005 at 32°C, overlaid with DMEM, 10% FBS, 1% HEPES and 1% Penn/Strep. Viral stocks were harvested when cytopathic effect was visible, and viral variant identification was confirmed by next generation sequencing.

Selection of samples
Clinical specimens were selected on three different days: Mar 25, Aug 3, and Aug 26, 2021. All specimens were nasal swabs in PBS collection kits (Greiner Bio-One cat. #456163) from drive-up community test sites in western and southern Washington. Aliquots from each sample were frozen at -80° C within 54 hours of sample collection (Mean ± SD: Alpha=40.1±10.4; Epsilon=41.4±9.9; Delta=39.3±10.2), and kept frozen until titering. Mutations in Spike indicating variants of concern (L452R and N501Y) were identified in clinical specimens using an RT-ddPCR assay similar to that we have previously described 4 . Specimens containing Alpha, Epsilon and Delta variants were selected by interpreting these results in combination with NGS surveillance done by UWVL in the period of each sample collection. In the week of the March collection, surveillance sequencing identified 31.5% Alpha (Y501), 29.5% Epsilon (R452), and 0.1% Delta; in the weeks of the August collection, samples were 0.4-0.6% Alpha, 0% Epsilon, and 98.1-98.5% Delta (R452) (https://depts.washington.edu/labmed/covid19/#sequencing-information).

Focus forming assay:
Vero E6-TMPRSS2, Huh7.5 or Caco-2 cells were infected, overlaid with 1.2% methylcellulose (Acros cat. #332620010) in DMEM and incubated for 24 h. As previous described 5 , all dilutions were plated in duplicate with the exception of the neat clinical sample for which we only had sufficient material to plate one well per sample. All clinical specimens were titered in Vero E6-TMPRSS2 cells. Foci were detected with a cross-reactive rabbit anti-SARS-CoV N monoclonal antibody (Sinobiological, distributed by Thermo Fisher, Cat. #40143-R001; 1:20,000), a peroxidase-labelled goat anti-rabbit antibody (SeraCare, Cat. #5220-0336; 1:4,000) and peroxidase substrate (SeraCare, Cat. #5510-0030). Plates were imaged on a BioTek ImmunoSpot S6 MACRO Analyzer and foci were counted using an automated virus plaque counter as previously described 6 and manually corrected.

In vitro infectivity assay:
Vero TMPRSS-2, Huh7.5, Caco-2 or Calu-3 cells were infected with stocks of Alpha, Delta and Epsilon viruses at an MOI of 0.1. We collected clarified supernatants at 24 hpi and measured the FFU and viral RNA levels (E gene RNA copies) for each sample.

Statistical Analysis:
Data for Fig. 1A was analyzed with GraphPad Prism 8, using a linear regression analysis to compare the stability of the three variants to each other at each concentration, and determine if the slopes of the lines were significant from each other, or significantly different than 0. P-values for whether slopes were non-zero ranged from p=0.1 to 0.87. The slopes were not significantly different between variants; p=0.8, 0.1, 0.2 for high, medium and low concentrations of virus respectively. Data for Fig. 1B, Fig. 2A-C were analyzed using R 7 . For Fig 1B a linear regression model was fitted on log transformed titers consisting of an intercept, a main cell effect, a main variant effect and a variant-cell type interaction. The variant-cell type interaction coefficients for delta and epsilon (relative to an alpha reference) were not significant. For Fig 2 titers were log transformed, and FFU values below LoD values were set to one-tenth the LoD. Linear regression was employed to model the relationship between titer and CT(E) or CT(sgE). Interaction terms between the CT values and variant were not significant and were dropped in favor of a simpler variant intercept only model, with all variants sharing a common slope.