Fig 1.
Drop-based Microfluidic Platform for Measuring Viral Burst Size from Single-Cell Infections.
(A) IAV (H1N1 or H3N2) was inoculated onto plated A549 cells at an MOI of 0.1 pfu/cell. (B) Single infected cells were encapsulated into 100 μm drops using a flow-focusing device. (C) Drops were incubated at 37°C for 18 hpi to allow for replication of IAV progeny virus. (D) In-drop infections were injected into a microfluidic device that splits 1/8th of the extracellular progeny virus from the host cell and merges these viruses with a continuous flow of multiplexed RT-qPCR mix targeting IAV M gene RNA and cellular β-actin mRNA. (E) Drops were thermocycled in a standard qPCR machine and sampled at discontinuous N. (F) ΔRN was measured serially using flow-based detection. (G) ΔRN values were converted to RNA cpd using dqPCR. (H) IAV burst size distributions of M gene RNA cpd from thousands of single-cell infections (green bars). The mean burst size of each distribution was calculated from a parametric model (gray curve) fitted to our data.
Fig 2.
Droplet Quantitative PCR (dqPCR) Model for Converting Drop Fluorescence to Nucleic Acid Concentration.
(A) M gene RNA (1.71 × 102 cpd) was amplified in 50 μm drops. ΔRN was detected at N = 1, 10, 13, 16, 19, 22, 28, 40 and displayed as the mean (black dots) with error bars representing one standard deviation. Approximately 1000 drops were detected at each cycle. A continuous reference amplification curve (blue solid curve) was reconstructed from discontinuous ΔRN measurements using SCF-E. The EN curve (black dashed curve) was fit to ΔRN at sampled cycle numbers to calculate ΔRN at non-sampled cycle numbers. (B) The efficiency parameters fit from the reference amplification curve (blue solid curve) were used to model 1000 virtual amplification curves (gray dashed curves, subset of 10 shown), creating a high-resolution dilution series called an Amplification Curve Library (ACL), representing 1000 unique M gene RNA concentrations. (C) Standard curves were constructed using ΔRN from all virtual curves at a single N (i.e., cycle 18). The valid range of the standard curve (gray shaded area) is determined by the min and max ΔRN of the reference curve.
Fig 3.
Resolving IAV M Gene RNA Concentrations from Single and Mixed Droplet Populations.
(A) Three known concentrations of M gene RNA, 1.71 × 101 cpd (red), 1.71 × 102 cpd (blue), and 1.71 × 103 cpd (green) were amplified in 50 μm drops. (B) ΔRN was detected at multiple different N (S6 Table) within each group and converted to M gene cpd. M gene cpd measurements were pooled to a single distribution for each group, with means of 2.03 × 101 cpd (red, n = 4,513 drops); 1.31 × 102 (blue, n = 15,899 drops); 1.16 × 103 (green, n = 65,249 drops). (C) The uniformity of each distribution was quantified using G. (D) Three M gene RNA concentrations in drops were collected in a mixed sample. (E) Mixed ΔRN were detected at N = 16–19, converted to M gene cpd, and pooled to a single distribution (grey bars, n = 68,748). Three individual peaks were isolated from the mixed distribution using GMM analysis (black curve) (S7 Table). (F) The heterogeneity of the mixed population compared to the individual populations in (C) was significantly greater with G = 0.639.
Fig 4.
H3N2 and H1N1 Single-Cell Infection Burst Size Distributions.
IAV burst sizes of H3N2 (blue) and H1N1 (red) were sampled at N = 16, 19, 22, 25, and 28 across three biological replicates (S8 Table). (A) Low to high M gene RNA (cpd) from a random sample of drops (n = 100) for H3N2 (blue) and H1N1 (red). (B) Single burst size distributions for H3N2 (blue, n = 9,620) and H1N1 (red, n = 5,585) populations. The measured mean burst sizes were 1.0 × 103 M gene cpd (H3N2) and 4.8 × 102 M gene cpd (H1N1). (C) The H3N2 distribution (G = 0.705) was less uniform than H1N1 (G = 0.586). (D-E) Distributions were fitted to a negative binomial (grey curve, S10 Table). The modeled mean burst sizes were (D) 709 cpd (H3N2) and (E) 385 cpd (H1N1).