Search tips
Search criteria 


Logo of jvirolPermissionsJournals.ASM.orgJournalJV ArticleJournal InfoAuthorsReviewers
J Virol. 2016 February 1; 90(3): 1599–1612.
Published online 2016 January 15. Prepublished online 2015 November 25. doi:  10.1128/JVI.02190-15
PMCID: PMC4719634

High-Throughput Single-Cell Kinetics of Virus Infections in the Presence of Defective Interfering Particles

D. S. Lyles, Editor


Defective interfering particles (DIPs) are virus mutants that lack essential genes for growth. In coinfections with helper virus, the diversion of viral proteins to the replication and packaging of DIP genomes can interfere with virus production. Mounting cases of DIPs and DIP-like genomes in clinical and natural isolates, as well as growing interest in DIP-based therapies, underscore a need to better elucidate how DIPs work. DIP activity is primarily measured by its inhibition of virus infection yield, an endpoint that masks the dynamic and potentially diverse individual cell behaviors. Using vesicular stomatitis virus (VSV) as a model, we coinfected BHK cells with VSV DIPs and recombinant helper virus carrying a gene encoding a red fluorescent protein (RFP) whose expression correlates with the timing and level of virus release. For single cells within a monolayer, 10 DIPs per cell suppressed the reporter expression in only 1.2% of the cells. In most cells, it slowed and reduced viral gene expression, manifested as a shift in mean latent time from 4 to 6 h and reduced virus yields by 10-fold. For single cells isolated in microwells, DIP effects were more pronounced, reducing virus yields by 100-fold and extending latent times to 12 h, including individual instances above 20 h. Together, these results suggest that direct or indirect cell-cell interactions prevent most coinfected cells from being completely suppressed by DIPs. Finally, a gamma distribution model captures well how the infection kinetics quantitatively depends on the DIP dose. Such models will be useful for advancing a predictive biology of DIP-associated virus growth and infection spread.

IMPORTANCE During the last century, basic studies in virology have focused on developing a molecular mechanistic understanding of how infectious viruses reproduce in their living host cells. However, over the last 10 years, the advent of deep sequencing and other powerful technologies has revealed in natural and patient infections that viruses do not act alone. Instead, viruses are often accompanied by defective virus-like particles that carry large deletions in their genomes and fail to replicate on their own. Coinfections of viable and defective viruses behave in unpredictable ways, but they often interfere with normal virus growth, potentially enabling infections to evade host immune surveillance. In the current study, controlled levels of defective viruses are coinfected with viable viruses that have been engineered to express a fluorescent reporter protein during infection. Unique profiles of reporter expression acquired from thousands of coinfected cells reveal how interference acts at multiple stages of infection.


The infection of a cell by a virus produces a mixture of viable and noninfectious progeny particles (1,3). A common class of noninfectious particles has defective genomes, often carrying deletions in essential genes that disable their ability to productively infect cells. However, in coinfections with viable or helper virus, the genomes of these defective particles compete with the viral replication machinery and packaging processes, interfering with infectious virus production in vitro (4, 5), and often reducing virulence in vivo (6, 7). These so-called defective interfering particles (DIPs) have for many decades been observed in laboratory cultures of virtually every class of DNA and RNA virus (4, 8). More recently, DIPs have been isolated and characterized from patients infected with influenza virus (9), individuals infected with dengue virus (10, 11), and birds infected with West Nile virus (12). Moreover, sequencing of patient and natural isolates has contributed to a growing list of diverse viral genomes that carry deletions in essential genes or regulatory sequences, including hepatitis C virus (HCV) (13), polyomavirus BK (14), hepatitis B virus (15), human papillomavirus type 16 (16), and baculovirus (17). Notably, for hepatitis C virus, trans-complementation studies showed that defective HCV genomes could be encapsidated and released from cells as infectious virus-like particles (13), but their interfering effects on helper HCV replication are not yet known.

During laboratory studies of acute infections, DIPs have been found to stimulate the host innate and adaptive immune responses, attenuate the virulence of the infection, and contribute to viral persistence (18,25). It is not yet known what role DIPs play in natural infections. However, owing to their ability to attenuate acute infections and stimulate host immune responses, DIPs are currently being explored and developed as antiviral therapeutics, novel infection control strategies, and vaccines against influenza, HIV, and mumps (26,28). Finally, DIPs pose a challenge to the industrial-scale production of live-attenuated vaccines, because they readily emerge during live-virus cultures (29,31).

To advance our understanding, at a fundamental level, of how the presence of DIPs influences viral growth, infection spread, and ultimately disease pathogenesis, one will need to develop a better quantitative grasp of how virus-DIP interactions within the coinfected host cell affect virus and DIP production. Such an understanding will also facilitate the engineering of DIPs for antiviral strategies and vaccines, as well as enable their control in bioprocess applications. Past studies have used coinfected cell cultures with DIPs and infectious virus to deduce DIP concentrations and elucidate their effects on virus and DIP production (2, 29, 32,38). Vesicular stomatitis virus (VSV), grown on diverse cell cultures, has served for more than 50 years as a model for the study of DIPs (5, 39). Moreover, quantitative mathematical and computational models have been developed to describe the kinetics of VSV replication in cells (40,42), its intracellular interaction with DIPs (43, 44), and its population dynamics (29).

DIPs typically slow the kinetics of virus progeny release from coinfected cells and reduce overall virus yields (24, 29, 32, 34, 37, 44, 45). The effects of DIP dose on virus yield have been elucidated (2, 29, 32, 33, 44) and have served as the basis for quantifying interference activity, assuming that individual DIPs are able to completely shut down virus production in coinfected cells (2, 32, 46). To date, however, relatively little has been revealed about how DIPs affect the intracellular steps that precede release of virus and DIP particles from coinfected cells. Moreover, most studies have obtained average measures from cell populations, which obscure the potentially diverse behaviors of individual cells.

Measures of single-cell behaviors have shown how cell-to-cell variation can quantitatively affect virus infections (47,52), and these measures may be extended to reveal potentially diverse DIP-virus interactions during coinfection. Early analysis of DIPs in single infected cells showed that higher DIP levels shifted distributions of virus yields toward low or nondetected virus yields, providing compelling evidence that single DIPs could indeed completely inhibit virus production from single cells (49). By demonstrating remarkable single-cell diversity at the level of virus yield, this work set the stage for the development of a more comprehensive picture of interference, specifically one that would account for how DIPs affect dynamic properties of virus particle release (Fig. 1) at the single-cell level.

Kinetic parameters of single-cell infection. Measures of a single round of infection within a single cell, taken over time, provide a growth curve, from which characteristic parameters may be extracted. Latent time is the interval between entry of adsorbed ...

Advances in single-cell imaging have enabled real-time tracking of viral activity in single cells (53,55; J. Warrick, A. Timm, A. Swick, and J. Yin, submitted for publication). Here, we coinfected BHK cells with a high multiplicity of infection (MOI) of recombinant VSV expressing red fluorescent protein (RFP) and various VSV DIP concentrations. We tracked the time-dependent expression of RFP-reported viral gene expression in hundreds to thousands of single cells, estimated kinetic parameters for each trajectory, and fit them to a gamma distribution, providing a metric to describe how DIP levels quantitatively influence the diversity of time-dependent infection behaviors. Further, we applied this approach to characterize the behavior of single cells in isolation, where they would be free from the influence of cytokines, growth factors, or direct interactions with other cells. Isolated cell behaviors were then compared with the behavior of single cells that were part of a larger cell population.


Virus and DIP culture.

Baby hamster kidney (BHK-21) cells were cultured at 37°C and 5% CO2 in Eagle's minimum essential medium (MEM) (CellGro) with 1% Glutamax I (Gibco) and 10% fetal bovine serum (FBS) (Atlanta Biologicals). The culture medium was switched to medium with 2% FBS for all virus infections. A recombinant vesicular stomatitis virus (VSV) strain incorporating DsRed-Ex (Clontech) into the fifth genomic position of VSV-Indiana and expressing red fluorescent protein (RFP) (56) was used for infections. To prepare virus stock, BHK-21 cells were infected with plaque-purified virus diluted to 0.001 PFU per cell in a T-75 flask (BD Falcon), incubated for 24 h at 37°C, filtered with a 0.22-mm filter (Millipore), and stored at −80°C. Infectious virus concentrations were quantified by plaque assay (47). DIPs were produced by fixed-multiplicity serial infections using VSV-rWT-DsRed-Ex at 10 PFU/cell, and the passage with the highest DIP concentration (determined by yield reduction assay [44]) was stored at −80°C for later infections (30). No RFP expression was observed in cells infected with only DIP and in cells coinfected with DIPs and the Indiana serotype of vesicular stomatitis virus, VSV-N1 (57), reflecting the loss of a functional RFP gene during DIP generation.

Infections in solution.

The methods for performing infections in the solution phase, rather than on cells adsorbed on a surface, have been previously described (47). Briefly, BHK-21 cells were released from their culture surfaces by treatment with trypsin, counted, and aliquoted into 15-ml tubes. The cell suspensions were cooled on ice and mixed with excess DsRed-Ex VSV (multiplicity of infection [MOI] of 30 [fluctuations in the MOI in the proximity of 30 did not affect the interference by DIPs]) and with different multiplicities of DIPs (MODIP of 0, 0.1, 1, and 10) in the cold for 30 min to enable virus to bind to cell surfaces without allowing fusion and entry. Following the virus adsorption period, the temperatures of the virus-cell solutions were raised to 37°C in a water bath for 5 min to allow for internalization of the attached virus. To remove any excess virus, the cell solutions were centrifuged (1,000 rpm, 4 min), the infection medium was decanted, and the infected-cell pellet was resuspended in fresh medium. This procedure was repeated 3 times to eliminate excess virus and DIPs in the cell suspension. Infected cells were serially diluted in medium containing 2% FBS and 1 μM Hoechst 33342 (AnaSpec, a live-cell nucleic acid stain) or in untreated medium with 2% FBS to appropriate concentrations to be used in various experiments described below.

Single-cell yields.

The infected-cell suspensions were added to wells of 96-well plates at densities of either 13,400 cells per well or 1 cell per well. For infections of isolated single cells, the plates were manually scanned using a bright-field microscope to locate wells containing just one cell. Fluorescence intensity was monitored over time by fluorescence microscopy. The wells containing a single cell or a population of cells were sampled at 24 h postinfection (hpi) and quantified via plaque assay to determine final viral titer.

Effects of cell density.

BHK-21 cells, infected in solution at an MOI of 5, were plated out in 96-well plates at various densities. Samples were collected from 10 wells at each density at 24 hpi, and virus production on a per cell basis was quantified by plaque assay.

Quantification of plaque reduction.

BHK-21 cells infected in solution at an MOI of 30 and at four different MODIP were diluted in uninfected cells (5 × 105 uninfected cell/ml) in MEM with 2% FBS to 20 infected cells/ml. One milliliter of cell suspension was plated into each well on 12-well plates in triplicate. After the cells were incubated for 1.5 h, the supernatant was removed and cells were overlaid with 0.6% (wt/vol) agar. After 20 h of incubation, the agar overlay was removed, and cells were fixed in 4% (wt/vol) paraformaldehyde and 5% (wt/vol) sucrose solution. The fixative remained on the cells for 20 min, and then the cells were rinsed twice with Dulbecco's phosphate-buffered saline (DPBS) and then stained with 2.5% (vol/vol) crystal violet (CV) diluted in 20% ethanol to aid in visualizing plaques. After the CV dried, plaques were counted. The entire experiment was performed in triplicate.

Seeding single cells into a microwell device.

High-throughput, single-cell experiments were performed using a polydimethylsiloxane (PDMS) microwell device of a size approximately the same as a standard glass slide (58; Warrick et al., submitted). The device is divided into 10 regions called “bull's-eyes” due to their design. These bull's-eyes physically separate parallel experiments performed using this device. Each bull's-eye contains ~2,500 microwells. One hour prior to infection, the microwell device was placed on a clean glass slide in an aluminum device holder, exposed to UV for 30 min for sterilization, and moved to a vacuum chamber to degas for 30 min. Once the device was degassed, medium droplets containing 1 μM Hoechst 33342 and HEPES (25 mM) (Sigma-Aldrich) were placed on each bull's-eye and placed in a humidified incubator until cell seeding.

To seed the infected cells into the microwell device, the existing droplets were removed and replaced with 70- to 80-μl droplets of the infected-cell solution in medium containing Hoechst 33342 and HEPES (105 cells/ml). After 30 to 60 s, the droplets were removed swiftly and replaced quickly with fresh medium. The swift removal of the droplets sweeps cells off the top surface of the device but does not disturb the cells that have settled into the microwells. The bull's-eyes were washed with fresh medium twice more in the same manner. The device was sealed by quickly removing the droplets from all bull's-eyes and gently covering with a glass slide while avoiding bubble formation. The top side of the glass slide was covered with a thin layer of 0.1% Tween to prevent condensation on the glass lid. Pressure was applied via an aluminum clamp device to maintain the seal. In this experiment, two bull's-eyes each were seeded with cells infected at one of the four MODIP conditions. The sealed device was placed into the environmental control chamber on the fluorescence microscope and imaged as described below.

Plating cell populations in wells.

BHK-21 cells infected in solution at an MOI of 30 and four different MODIP were diluted in medium containing 2% FBS and Hoechst 33342 and plated at a concentration of 5 × 105 cells/well into 12-well plates in triplicate. After the plate was incubated for 1 h to allow the cells to settle, the plate was placed into the environmental control chamber and imaged over time as described below.

Live-cell microscopy. (i) Monitoring viral activity in microwells and 96-well plates.

The time-lapse microscopy experiments of microwells and low-throughput single-cell yield experiments were performed on an Nikon TE Eclipse 300 microscope fitted with an outer warming chamber at 37°C (InVivo Scientific) and a stage-top incubator chamber (Pathology Devices) at 37°C, 5% CO2, and 85% relative humidity, at a magnification of ×4. Fluorescence illumination was provided by a Chroma PhotoFluor light source and controlled with a Lambda 10-2 optical filter changer. Images of microwells were taken at 20-min intervals for 23 h beginning at 1.5 hpi, and images of 96-well plates were taken at 2-h intervals for 24 h starting at 4 hpi.

(ii) Monitoring viral activity in 12-well plates.

The time-lapse microscopy experiments of single cells within a cell population in 12-well plates were performed using a Nikon Eclipse-Ti microscope fitted with an outer warming chamber at 37°C and a stage-top chamber (both from InVivo Scientific) at 5% CO2, with humidification, at a magnification of ×10. Fluorescence illumination was provided by Lumincor and Nikon filter cubes in the rotating turret. Images were taken at 10-min intervals for 25 h beginning at 1.5 hpi.

(iii) Correction images for quantitative imaging.

For image quantification, calibration images were taken of a uniform-intensity slide (Ted Pella) immediately before or after the time-lapse imaging to correct for uneven light intensity (flat-field correction). The exposure time for the average illumination field image was adjusted to between 1 and 5 ms to provide representative intensity of actual images. The exposure time for the dark-field calibration image was the same as the exposure used in the actual time-lapse imaging of the associated channel.

Image processing workflow and data analysis.

Image organization and processing were done in JEX, a Java-based software for high-throughput image processing and quantification. For microwell images, the microwell locations were identified and numbered on the basis of the maxima in the blue channel (nuclear label, 4′,6′-diamidino-2-phenylindole [DAPI]). Measurements were taken about the maximum signal in each single-cell well using a standard area based on the cell radius. The measurement and cell count data tables were exported and analyzed in MATLAB. Images of 96-well and 12-well plates were processed in the same way, except the image calibration using correction images was done before cell tracking and intensity measurements. Also, single cells in 12-well plates were tracked over time using a JEX script based on a nearest-neighborhood method.

Extraction of infection kinetic parameters from RFP profiles.

After the viral activity in single cells was tracked over time and recorded, the single-cell kinetic profiles lacking continuity due to cell-to-cell fusion and cell mobility were eliminated by selecting cells based on a number of criteria. Cells tracked (based on nuclear label) longer than 7 h were selected for analysis. RFP yields of single-cell kinetic profiles were calculated as the average of the five highest fluorescence intensities over time. If RFP yield of a single cell remained below the limit of detection [LOD = background signal of negative control + (4 × standard deviation of the background noise)] over the infection tracking period, the cell was counted as a nonproducer cell. Among the RFP-positive trajectories, we discarded ones exhibiting large jumps during a short period, defined as minimally 25% change in RFP yield during a 40-min period; based on our direct observation of such events, they were typically artifacts that reflected fusion of an RFP-positive cell with an RFP-negative cell and resulted in a corresponding rapid change in detected fluorescence during two adjacent sampling times. To calculate latent time and rise time, the first rise time point was identified as the start of an at least 1.5-h-long continuous increase in fluorescence intensity without any drop to the background level. The exponential rise rate was calculated by fitting an exponential function (yield = Aerise rate × time) to the data points of the 1.5-h-long initial rise weighted based on the noise levels (~signal0.5) (59). The end of rise period was taken as the time point when the fluorescence intensity reached 0.85 of the RFP yield of the associated single-cell kinetic profile.

Maximum likelihood estimation.

To estimate the parameters of the probabilistic model presented below (see Fig. 7), maximum likelihood estimation was used. The measured infection kinetic parameter distributions obtained from single cells were converted to probability density histograms, and the bin sizes of the histograms were calculated on the basis of the Freedman-Diaconis rule. The measured probability density histograms were used to calculate the model parameters with maximum likelihood, and the likelihood intervals were determined on the basis of the 10% relative likelihood region [likelihood > (0.1 × maximum likelihood)] in two-parameter space (the degree of change in the mean [ka] and degree of change in the variance [kb]) (60). The maximum likelihoods were used to calculate the relative likelihood ratios, which are the maximum likelihood of the chosen model divided by the maximum likelihood of other models. Using relative likelihood ratios, the fit of the chosen model was compared against other tested models.

Mathematical model of the dependence of viral gene expression kinetics on DIP levels in single cells. Added DIPs (black bullets) adsorb to cells (pink) following a Poisson distribution. For a given number of DIP particles per cell (m), the infection kinetic ...

Also, to assess the performance of the chosen model, cross validation analysis was employed, where combinations of experimental results at different MODIP conditions were used to train the model, and the excluded conditions (validation data sets) provided a score for model validation. This score for each combination was calculated by dividing the maximum likelihood to the best maximum likelihood estimates of the validation data sets, which were obtained by fitting a gamma distribution to each experimental condition, independent of the other conditions.

Statistical analysis.

The significance of correlations was determined using the Pearson product-moment correlation coefficient. Statistical significance for differences between measured characteristics of single-cell populations was tested using the Mann-Whitney U test or paired t test as indicated.


Viral yields drop as DIP levels rise.

Cells coinfected with excess viable virus (MOI of 30) and different DIP multiplicities (MODIP) were diluted down to either single isolated cells or populations containing 106 cells per well. Viral yields, determined on a per cell basis, fell with increasing MODIP, reaching a nearly 100-fold drop at an MODIP of 10 (Fig. 2). Moreover, yields from cells in a population were significantly higher than from isolated single cells at MODIP of 0, 1, and 10 (P < 0.05).

Dependence of virus yields on DIP level. Yields of virus production were measured for cells in a population (gray hexagons; three wells at each multiplicity of DIPs [MODIP]) and for isolated single cells (box plots; 15 to 19 cells at each MODIP). Cells ...

Single-cell virus yields correlate with reporter yields.

For 37 isolated single cells coinfected by virus and DIPs, we measured RFP expression over time and quantified endpoint viral yields. Normalized RFP yields correlated with PFU yields (r = 0.856 and P < 0.01); however, DIP inhibitory effects were up to 10-fold greater on PFU yield than on RFP yield, as reflected by the deviation of data from the parity line (Fig. 3). Moreover, at an MODIP of 10, 3 cells out of 19 cells (~16%) and at an MODIP of 1, 1 cell out of 19 cells (~5%) did not produce any infectious virus particles (<5 PFU/cell). However, fluorescence intensity was positive in all of these cells, and only one cell at an MODIP of 10 (~5%) was below detection limits for both RFP and PFU (not shown).

Correlation of virus production with reporter expression in single cells. Reporter (RFP) yields correlate with virus (PFU) yields in isolated single cells (r = 0.856; P < 0.01; n = 33). Cells coinfected with virus (MOI of 30) and various DIP levels ...

Individual cells exhibit diversity of RFP profiles.

Cells were coinfected with virus at high MOIs and different DIP levels, diluted into microwells or plated as a cell population, and monitored. In the absence of DIPs, individual cells in microwells exhibited a broad distribution of RFP expression over time (Fig. 4a, MODIP of 0), with maximum values spanning an 8-to-10-fold range and lysis events apparent in some expression profiles as precipitous drops in RFP levels. At DIP levels overall peaks in RFP expression levels declined, a pattern that was apparent for both cells in microwells and cells in populations (Fig. 4a and andb,b, respectively).

Dependence of RFP reporter expression on DIP levels. RFP reporter expression was tracked in single cells coinfected at an MOI of 30 and various multiplicities of DIPs (MODIP), as indicated. Gray lines show RFP reporter expression trajectories from representative ...

DIPs reduce and delay viral gene expression in single cells.

Using the microwell platform, we used the measured RFP expression profiles from hundreds of isolated single cells, and for each cell we estimated four kinetic parameters. The distributions of these parameters were determined as a function of DIP level (MODIP) and cell environment (isolated or population), as shown in Fig. 5. Over the range of conditions tested, DIPs caused drops in RFP yield (up to 100-fold) and rise rate (up to 4-fold), indicating that DIPs act within cells by slowing and reducing the extent of viral gene expression (Fig. 5c). Moreover, coinfections with DIPs correlated with increases in latent time (up to 3-fold) and rise time (up to 50%), indicating that DIPs delay the production of detectable RFP and reduce the duration of RFP production. The variance of each distribution followed the same trends as the mean values, with smaller variances associated with drops in the yield and rise rate, and larger variances observed with increases in the latent time and rise time (Fig. 5c). To test how the presence of other cells may affect the inhibitory effects of DIPs on viral gene expression, we may compare RFP expression profiles from individual cells within a monolayer population containing millions of cells with profiles from single cells isolated in a microwell array. Infection parameters for cells in a population followed trends similar to those of cells in isolation; at higher DIP levels, yield and rise rates dropped while latent times and rise times increased (Fig. 5c, black bars versus gray bars). However, cells within a population appeared less sensitive to DIPs than cells in isolation; at an MODIP of 10, isolated cells exhibited 10-fold-lower mean values for RFP yields and 2-fold-longer latent times than cells in a population.

Effects of DIPs on distributions of infection parameters from single cells. (a and b) Estimated parameters for isolated single cells (n = 458–1,064) (a) and individual cells within a population (n = 2,452–6,056) (b). All cells were coinfected ...

Virus production is more sensitive to inhibition by DIPs than viral gene expression.

We assessed the productivity of coinfected cells using two measures, the intracellular production of detected RFP expression and the extracellular production of viable virus, as detected by plaque formation. Most coinfected cells within a population readily express detectable RFP, with productive cells representing 98.8% of the total population at an MODIP of 10 (Fig. 6). In contrast, the production of viable virus progeny, as detected by plaque formation, was quite sensitive to DIPs. If one assumes that DIPs adsorb to cells following a Poisson distribution, then MODIP of 0.1 and 1 would represent 9 and 37% of the population adsorbing a single DIP, respectively. If these singly adsorbed DIPs completely inhibited virus production, then one would expect the remaining 91 and 63% of the cells to produce plaques, values which are comparable to the observed plaque producers of 89 and 68% of the population at MODIP of 0.1 and 1, respectively. Further, at the highest DIP levels, coinfected cells that failed to produce plaques made up 87.5% of the population, providing additional evidence that high DIP levels are unable to completely suppress viral gene expression and viable virus production. At the highest DIP levels, the numbers of productive cells based on their ability to form plaques were less than the numbers of productive cells based on RFP expression, consistent with our previous finding that the single-cell production of viable virus is more sensitive to inhibition by DIPs than viral gene expression (Fig. 3).

Dependence of producer cells on DIP level. Cells were coinfected with RFP reporter virus (MOI of 30) and different levels of DIPs. Productive cells were quantified based on the presence of detectable RFP from single cells within a population or based ...

Mathematical modeling of DIP effects on viral gene expression kinetics.

To enable a more quantitative assessment of how DIP dose affects viral gene expression as reported and quantified through RFP, we sought to develop a mathematical model that expressed how the mean and variance of each RFP kinetic parameter, determined for hundreds of cells, depended on DIP dose. We assumed that all cells were highly infected with a fixed level of viable virus (MOI of 30) and that DIPs adsorbed to cells following a Poisson distribution (Fig. 7). The subpopulation of cells that received no DIPs had for each kinetic parameter a mean value (μ0) and variance (σ0), which could be estimated from cells infected solely with viable virus (MODIP of 0), while the subpopulation that received one DIP had a different mean (μ1) and variance (σ1), reflecting the behavior of the approximately 10% of total cells at an MODIP of 0.1 that adsorbed a single DIP. The mean and variance of each kinetic parameter shifted from the reference case (MODIP of 0) depending on the DIP level. This dependence was modeled using a gamma distribution, which has precedence as a flexible distribution for describing various biological measures (61, 62). The mean and variance of these distributions were defined as functions of the mean and variance of the reference condition (MODIP of 0) and DIP input level into a single cell. These functions were chosen from several options and based on relative likelihood ratios (Table 1). Depending on the observed changes in the mean and variance of the distributions (Fig. 5c), these functions either decreased with increasing DIP levels (degree of change in the mean [ka] < 0 and degree of change in the variance [kb] < 0), such as for infection yields and exponential rise rates, or they increased with increasing DIP levels (ka > 0 and kb > 0), as seen for latent and rise times. By fitting our model to the observed distributions, we determined the degree of change in the mean and variance with the addition of DIPs in isolated single cells (Fig. 8, insets, gray bars) and individual cells within a population (Fig. 8, insets, black bars). At optimum ka and kb values (Table 2), our model (Fig. 8, circles, and Fig. 9 and and10,10, red curves) captured well the observed single-cell distributions. A cross validation analysis indicated that the optimum model parameters do not change significantly when the results from only a subset of the experimental conditions were used to train the model, suggesting that the model can be generalized to other MODIP conditions.

Comparison of different models
Dependence of infection kinetic parameters on DIP levels. The observed changes (bars) and fit values (circles) in the mean (left) and variance (right) of relative yield, latent time, rise time, and rise rate of RFP (ordinate) in isolated single cells ...
Optimum model parameters with likelihood intervalsa
Measured and fit distributions of infection kinetic parameters for isolated single cells. Measured probability densities (ordinate) of relative yield, latent time, rise time, and rise rate of RFP (abscissa) are shown as black bars. Red curves represent ...
FIG 10
Measured and fit distributions of infection kinetic parameters for single cells in a population. Measured probability densities (ordinate) of relative yield, latent time, rise time, and rise rate of RFP (abscissa) are given as black bars. Red curves represent ...

Population context affects the degree of interference.

The fit of our model to our data indicates that DIP coinfections caused greater change in the behavior of isolated single cells than in individual cells within a population, as reflected in the magnitude of ka and kb values (Fig. 8, insets). The degrees of change in the mean and variance of the RFP yield, latent time, and rise rate for single isolated cells were about 2-fold higher than for single cells in a population. The differences between single infected cells in isolation versus in cell populations, even in the absence of DIPs, suggests that the presence of cell neighbors affects the physiological state of an infected cell, particularly with respect to its ability to express foreign genes introduced by the infecting virus. To test whether this effect extends to virus production, virus yields from infected cells were determined on a per cell basis as a function of plated infected-cell density. Average cell yields of viable virus from infected cells were found to increase 4-fold as infected-cell densities were increased 1,000-fold (Fig. 11), indicating a quantifiable effect of cells on virus yield per cell.

FIG 11
Effect of cell density on infectious virus yield. Infectious virus production on a per cell basis increases proportionally with the local density of cells. Virus yields at each cell density were measured by plaque assay (n = 10). Error bars represent ...

Correlations between infection kinetic parameters.

To see how DIPs affect the pairwise relationships between kinetic parameters associated with the reporter expression profiles, one may plot one parameter versus another, such as latent time versus relative yield for isolated single cells (Fig. 12, top left panel). When no DIPs are present (MODIP of 0), the relative yields are highest and the latent times are shortest, indicating conditions that promote overall early and high-level viral gene expression and virus particle production. However, as DIPs are added, shifting MODIP from 0.1 to 10, the relative yield drops, while the latent time distribution significantly broadens, indicating overall greater delay and lower levels of viral gene expression and particle production associated with higher DIP levels. In general, for individual cells within a population, the correlation patterns are similar to those found for isolated cells (Fig. 13). However, distributions appear less defined in shape, reflecting in part the larger number of cells (i.e., 2,500 to 6,000 cells versus 500 to 1,000 cells). Moreover, minimal rise times for individual cells in a population are longer than for isolated single cells (Fig. 12 and and13,13, bottom panels), indicating that viral gene expression, once detected, persists longer when cells are in a population than when they are in isolation. To be quantitative, we determined correlation coefficients for the six possible parameter pairings across the full range of DIP levels. We found significant correlation between pairings of relative yield, latent time, and rise rate for both isolated and populated cell environments (Table 3). Rise time exhibited relatively weak correlations with other parameters, suggesting that it is more variable than the other kinetic parameters, a feature that was reflected in the large variance of the rise time distributions (Fig. 5).

FIG 12
Correlations between infection kinetic parameters for isolated single cells. Pairs of infection kinetics parameters measured in isolated single cells were compared in each plot. Each data point represents a single cell coinfected at an MOI of 30 with ...
FIG 13
Correlations between infection kinetic parameters for single cells in a population. Pairs of infection parameters measured in individual cells in a population were compared in each plot. Each data point represents a single cell coinfected at an MOI of ...
Correlations between kinetic parameters of viral gene expression


Our results highlight a facet of a phenomenon that is important to keep in mind as single-cell methods are advanced for a variety of uses—cells that are members of a population differ from cells in isolation. Even in the absence of DIPs, average yields of viable virus from infected cells increased 4-fold as infected-cell densities in a population were increased 1,000-fold (Fig. 11), suggesting the presence of proviral factors in cell populations that are not available to isolated cells or the enrichment of antiviral factors by isolated cells. Common antiviral factors include interferons (IFNs), and IFN-mediated responses can be triggered by DIPs (63, 64); however, the host BHK cells used here lack such responses (65, 66). Alternatively, confinement of cells to microwells can reduce the capacity of cells to grow, as reflected by longer G1 and shorter G2/M phases of the cell cycle (67), conditions that are consistent with lower average productivities of VSV from BHK cells (51). Thus, the lower virus yields associated with isolated coinfected cells (Fig. 2) likely reflect antiproliferative conditions associated with cell isolation, combined with interfering effects of DIPs on virus production.

Further differences between coinfected cells in different environments were revealed by the kinetics of the viral reporter gene expression. Cells in a population subjected to high DIP levels (MODIP of 10) exhibited a relatively small shift in latent time from 4 to 6 h, while for cells in isolation, the corresponding shift was from 4 to 12 h (Fig. 5c). Consistent with this observation is the relatively strong correlation between longer latent times and smaller yields (top left in Fig. 12 and and13;13; Table 3) and the 10-fold-lower RFP yields in isolated cells versus cells in a population when coinfected with high DIP levels (Fig. 5c). It is known that the population context can affect the viral infection kinetics, particularly with respect to the efficiency of virus adsorption to its host (68); however, we excluded such differences by adsorbing virus to cells under the same solution conditions, thereby limiting the potential source of differences to postadsorption steps. During the postadsorption steps, biosynthesis and cell growth increase the rate and capacity for virus growth (69). However, the aforementioned antiproliferative conditions in microwells may limit stimulation of biosynthesis and thereby aggravate the host resource limitations exerted by DIPs. While single cells in microwells may appear to be artificially isolated from the reality that cells in nature generally exist as members of a population, it is noteworthy that the cells in these very different environments exhibited similar trends with respect to the dependence of their reporter expression on DIP levels. Moreover, the microwell technology offers a means to better control the single-cell microenvironment in ways that would be harder to achieve in a cell population.

Our fluorescent-protein reporter of VSV infection showed gene expression and virus particle release kinetics similar to those of the wild-type (Indiana serotype) virus. Classical measures of protein synthesis by VSV-infected cells, which used pulse-chase incorporation of radiolabeled amino acids, protein separation by polyacrylamide gel electrophoresis, and quantification of bands by scintillation counting, showed initiation of viral protein synthesis at 2 to 3 hpi, with continuing synthesis through at least 6 hpi (70). On the basis of the 5th position of our reporter protein, distal to the 3′ VSV promoter, we would expect its initial synthesis to resemble that of the natural 5th position encoded L protein, which was detected by Wagner et al. at 3 hpi. Our corresponding latent time to detection of the fluorescent reporter was 4 hpi (Fig. 5c), consistent with a 1-h delay for the synthesized RFP to mature and become detectable (71). Early studies found that progeny virus particles were initially released with latent times of 2 to 4 hpi and with titers peaking 8 to 10 hpi (72). We have observed similar but broader ranges from our single-cell kinetic studies of VSV particle release, with latent times ranging from 3.5 to 6.5 hpi and titers peaking 8 to 14 hpi (47). Moreover, expression of the reporter protein was found to have negligible effects on the kinetics and level of virus production from BHK cells (56), making the reporter strain a suitable tool to study how DIPs affect viral gene expression.

Beyond the latent time period, the effect of DIPs on gene expression from intact VSV was also consistent with the scant previous observations. For coinfections of intact virus (MOI of 10) with DIPs (MODIP of 10), it was previously found that viral protein synthesis rates peaked at about 5 hpi (not 3 or 7 hpi) and then shifted to 7 hpi at a higher DIP level (MODIP of 20), and at the highest DIP level (MODIP of 80), the viral protein synthesis window was prolonged, without decline, to at least 15 hpi (37). Although our coinfection conditions differ, we observed qualitatively similar results, where a peak in protein synthesis rate would occur in the period between the start of detectable protein synthesis (mean latent time) and the end of protein synthesis (mean latent time plus rise time). Based on measures of individual cells in a population at an MODIP of 10 (Fig. 5c), this corresponds to a window from 6 to 14 hpi, which falls within the range of 5 ± 2 hpi found by Von Laer et al. (37). We also observed at higher DIP levels a lengthening of mean latent and rise times (Fig. 5c), consistent with processes that would prolong the period of viral protein synthesis previously observed.

The effect of DIPs on the production of viable virus was consistent with previous observations of complete shutdown of virus production by low levels (1 to 3) DIPs per cell, but with incomplete shutdown of virus production at higher DIP levels. Specifically, the classic study of single cells infected by virus at an MOI of 8.5 and DIPs at multiplicities of 0, 1, and 16 found nonproducers in 2, 48, and 44% of total cells (49). At an MODIP of 1, assuming that DIPs follow a Poisson distribution, 37% of cells receive one DIP and 18% receive two DIPs, so 55% of the cells receive one or two DIPs, and the observed 48% of nonproducers is consistent with one or two DIPs being sufficient to shut down production of detectable virus. We observed values slightly below this range with 32% of the coinfected cells unable to make plaques (Fig. 6). For high DIP levels (MODIP of ≥10), cells that receive three or more DIPs make up 99.7% of the population, yet nonproducers were only 44% (49) or 88% of the population based on plaque production, supporting the observation that cells coinfected with higher DIP loads cannot completely shut down virus production, consistent with a mechanism where DIPs in the same cell interfere with their ability to shut down virus production, commonly referred to as “interference with interference” (49).

The extent of change of viral gene expression in single cells, extracellular infectious virus production, and infection spread provided hints on the infection steps affected by DIPs. The increase in latent time by the addition of DIPs suggests that interference starts early in infection. Processes that affect the latent time include primary transcription, which entails using the viral RNA polymerase associated with the entering viral genome to make viral mRNA, synthesis of viral proteins, replication of viral genomes, secondary transcription, and further viral protein synthesis. Recent measures and kinetic modeling of VSV early infection suggest that cellular resources for primary transcription and protein synthesis are not limiting (42), making these initial steps unlikely targets for interference by defective genomes. However, during secondary transcription and translation, as essential components for genome replication are synthesized, diversion of these resources to DIP genome replication, rather than viral genomic replication, reduces available levels of genomic template for production of viral mRNA and subsequent synthesis of viral proteins (6, 7, 37, 73). Ultimately, these processes contribute to reducing the rise rate and delaying the production of the detectable RFP expression that defines the latent time.

In addition to early effects of DIPs on viral gene expression, differences between the magnitude of inhibition of viral gene expression in cells, extracellular infectious virus production, and infection spread suggest that DIP-induced interference also acts during late phases of infection. At an MODIP of 10, viral gene expression ceased in 1.2% of cells within a population (Fig. 6), while infectious virus particle (PFU) production was stopped in ~16% of single cells, indicating that a significant portion of the RFP-positive cells failed to produce infectious virus particles at detectable levels (~15%). This observation was supported by 10- to 20-fold-higher reduction in PFU yield than in RFP yield in both isolated single cells (Fig. 3) and cell populations, and it suggests that infectious virus resources are diverted to the encapsulation of DI genomes and assembly of DIPs, reducing infectious virus assembly efficiency. Moreover, despite the relatively rare occurrence of complete inhibition of infectious virus production (~16%), infection spread was stopped in 87.5% of cells (Fig. 6), suggesting that even in the absence of complete infection inhibition in an initially infected cell, the reduction in virus yield and the potential enhancement of DIP production from the initial cell can completely halt subsequent infection spread.

By examining and quantifying viral gene expression and the yield of virus progeny from many individual cells, this work underscores the potentially diverse behaviors of individual cells within a population. An additional facet of the virus-cell interaction under investigation is the kinetics of innate immune activation in response to infection, a process that may be monitored at the single-cell level by employing two fluorescent reporters, one linked to viral gene expression, as in this work, along with a second that is driven by an interferon-sensitive promoter (56; Warrick et al., submitted). It is not yet known what role DIPs play in the growth, spread, and persistence of viruses in their natural host reservoirs or in the emergence of infectious diseases. However, we have seen here how DIPs can broaden the distribution of infection phenotypes, particularly in parameters of interaction where the variance increases with DIP dose. Specifically, the rise in the mean and variance of the latent time for the reporter of viral gene expression with higher DIP dose (Fig. 5), especially for isolated cells, may correlate at the population level with a broader window of infectious particle production. Further studies will be needed to elucidate how such effects play out over multiple infection cycles.


To characterize the effects of defective interfering particles on the intracellular kinetics of virus growth, we have employed recombinant viruses that use a fluorescent protein to report the viral gene expression in near real time. We found that the DIPs overall delayed and attenuated viral gene expression and virus particle production. Moreover, we developed a mathematical model to describe the mean and standard deviation of DIP loading effects on characteristic parameters of the kinetics of gene expression. We anticipate that similar approaches to modeling kinetic profiles from high-throughput studies will find broad application in a diversity of single-cell infection studies.


We thank Jay Warrick for helpful discussions and technical support for JEX, Ankur Gupta for help with maximum likelihood analysis, and Sean Palecek for his review of an early draft of this work.

This work was supported by the National Institutes of Health (grants AI091646 and AI104317), and A.T. was supported by an NIH Virology Training Program Fellowship (T32 AI078985).

We declare that we have no conflicts of interest.


1. Von Magnus P. 1954. Incomplete forms of influenza virus. Adv Virus Res 2:59–79. doi:.10.1016/S0065-3527(08)60529-1 [PubMed] [Cross Ref]
2. Marcus PI, Ngunjiri JM, Sekellick MJ 2009. Dynamics of biologically active subpopulations of influenza virus: plaque-forming, noninfectious cell-killing, and defective interfering particles. J Virol 83:8122–8130. doi:.10.1128/JVI.02680-08 [PMC free article] [PubMed] [Cross Ref]
3. Carpenter JE, Henderson EP, Grose C 2009. Enumeration of an extremely high particle-to-PFU ratio for varicella-zoster virus. J Virol 83:6917–6921. doi:.10.1128/JVI.00081-09 [PMC free article] [PubMed] [Cross Ref]
4. Huang AS, Baltimore D 1970. Defective viral particles and viral disease processes. Nature 226:325–327. doi:.10.1038/226325a0 [PubMed] [Cross Ref]
5. Cooper PD, Bellett AJD 1959. A transmissible interfering component of vesicular stomatitis virus preparations. J Gen Microbiol 21:485–497. doi:.10.1099/00221287-21-3-485 [PubMed] [Cross Ref]
6. Stampfer M, Baltimore D, Huang AS 1969. Ribonucleic acid synthesis of vesicular stomatitis virus. I. Species of ribonucleic acid found in Chinese hamster ovary cells infected with plaque-forming and defective particles. J Virol 4:154–161. [PMC free article] [PubMed]
7. Huang AS, Manders EK 1972. Ribonucleic acid synthesis of vesicular stomatitis virus. IV. Transcription by standard virus in the presence of defective interfering particles. J Virol 9:909–916. [PMC free article] [PubMed]
8. Lazzarini RA, Keene JD, Schubert M 1981. The origins of defective interfering particles of the negative-strand RNA viruses. Cell 26:145–154. doi:.10.1016/0092-8674(81)90298-1 [PubMed] [Cross Ref]
9. Saira K, Lin X, DePasse JV, Halpin R, Twaddle A, Stockwell T, Angus B, Cozzi-Lepri A, Delfino M, Dugan V, Dwyer DE, Freiberg M, Horban A, Losso M, Lynfield R, Wentworth DN, Holmes EC, Davey R, Wentworth DE, Ghedin E 2013. Sequence analysis of in vivo defective interfering-like RNA of influenza A H1N1 pandemic virus. J Virol 87:8064–8074. doi:.10.1128/JVI.00240-13 [PMC free article] [PubMed] [Cross Ref]
10. Aaskov J, Buzacott K, Thu HM, Lowry K, Holmes EC 2006. Long-term transmission of defective RNA viruses in humans and Aedes mosquitoes. Science 311:236–238. doi:.10.1126/science.1115030 [PubMed] [Cross Ref]
11. Li D, Lott WB, Lowry K, Jones A, Thu HM, Aaskov J 2011. Defective interfering viral particles in acute dengue infections. PLoS One 6:e19447. doi:.10.1371/journal.pone.0019447 [PMC free article] [PubMed] [Cross Ref]
12. Pesko KN, Fitzpatrick KA, Ryan EM, Shi P-Y, Zhang B, Lennon NJ, Newman RM, Henn MR, Ebel GD 2012. Internally deleted WNV genomes isolated from exotic birds in New Mexico: function in cells, mosquitoes, and mice. Virology 427:10–17. doi:.10.1016/j.virol.2012.01.028 [PMC free article] [PubMed] [Cross Ref]
13. Sugiyama K, Suzuki K, Nakazawa T, Funami K, Hishiki T, Ogawa K, Saito S, Shimotohno KW, Suzuki T, Shimizu Y, Tobita R, Hijikata M, Takaku H, Shimotohno K 2009. Genetic analysis of hepatitis C virus with defective genome and its infectivity in vitro. J Virol 83:6922–6928. doi:.10.1128/JVI.02674-08 [PMC free article] [PubMed] [Cross Ref]
14. Myhre MR, Olsen G-H, Gosert R, Hirsch HH, Rinaldo CH 2010. Clinical polyomavirus BK variants with agnogene deletion are non-functional but rescued by trans-complementation. Virology 398:12–20. doi:.10.1016/j.virol.2009.11.029 [PubMed] [Cross Ref]
15. Zhang D, Dong P, Zhang K, Deng L, Bach C, Chen W, Li F, Protzer U, Ding H, Zeng C 2012. Whole genome HBV deletion profiles and the accumulation of preS deletion mutant during antiviral treatment. BMC Microbiol 12:307. doi:.10.1186/1471-2180-12-307 [PMC free article] [PubMed] [Cross Ref]
16. Kukimoto I, Maehama T, Sekizuka T, Ogasawara Y, Kondo K, Kusumoto-Matsuo R, Mori S, Ishii Y, Takeuchi T, Yamaji T, Takeuchi F, Hanada K, Kuroda M 2013. Genetic variation of human papillomavirus type 16 in individual clinical specimens revealed by deep sequencing. PLoS One 8:e80583. doi:.10.1371/journal.pone.0080583 [PMC free article] [PubMed] [Cross Ref]
17. Chateigner A, Bézier A, Labrousse C, Jiolle D, Barbe V, Herniou EA 2015. Ultra deep sequencing of a baculovirus population reveals widespread genomic variations. Viruses 7:3625–3646. doi:.10.3390/v7072788 [PMC free article] [PubMed] [Cross Ref]
18. Calain P, Monroe MC, Nichol ST 1999. Ebola virus defective interfering particles and persistent infection. Virology 262:114–128. doi:.10.1006/viro.1999.9915 [PubMed] [Cross Ref]
19. Kennedy JC, Macdonald RD 1982. Persistent infection with infectious pancreatic necrosis virus mediated by defective-interfering (DI) virus particles in a cell line showing strong interference but little DI replication. J Gen Virol 58:361–371. doi:.10.1099/0022-1317-58-2-361 [PubMed] [Cross Ref]
20. Yount JS, Kraus TA, Horvath CM, Moran TM, Lopez CB 2006. A novel role for viral-defective interfering particles in enhancing dendritic cell maturation. J Immunol 177:4503–4513. doi:.10.4049/jimmunol.177.7.4503 [PubMed] [Cross Ref]
21. Baum A, Sachidanandam R, García-Sastre A 2010. Preference of RIG-I for short viral RNA molecules in infected cells revealed by next-generation sequencing. Proc Natl Acad Sci U S A 107:16303–16308. doi:.10.1073/pnas.1005077107 [PubMed] [Cross Ref]
22. Scott PD, Meng B, Marriott AC, Easton AJ, Dimmock NJ 2011. Defective interfering virus protects elderly mice from influenza. Virol J 8:212. doi:.10.1186/1743-422X-8-212 [PMC free article] [PubMed] [Cross Ref]
23. Bean WJ, Kawaoka Y, Wood JM, Pearson JE, Webster RG 1985. Characterization of virulent and avirulent A/chicken/Pennsylvania/83 influenza A viruses: potential role of defective interfering RNAs in nature. J Virol 54:151–160. [PMC free article] [PubMed]
24. Doyle M, Holland JJ 1973. Prophylaxis and immunization in mice by use of virus-free defective T particles to protect against intracerebral infection by vesicular stomatitis virus. Proc Natl Acad Sci U S A 70:2105–2108. doi:.10.1073/pnas.70.7.2105 [PubMed] [Cross Ref]
25. Killip MJ, Young DF, Precious BL, Goodbourn S, Randall RE 2012. Activation of the beta interferon promoter by paramyxoviruses in the absence of virus protein synthesis. J Gen Virol 93:299–307. doi:.10.1099/vir.0.037531-0 [PMC free article] [PubMed] [Cross Ref]
26. Dimmock NJ, Easton AJ 2014. Defective interfering influenza virus RNAs: time to reevaluate their clinical potential as broad-spectrum antivirals? J Virol 88:5217–5227. doi:.10.1128/JVI.03193-13 [PMC free article] [PubMed] [Cross Ref]
27. Notton T, Sardanyés J, Weinberger AD, Weinberger LS 2014. The case for transmissible antivirals to control population-wide infectious disease. Trends Biotechnol 32:400–405. doi:.10.1016/j.tibtech.2014.06.006 [PubMed] [Cross Ref]
28. Šantak M, Markušić M, Balija ML, Kopač SK, Jug R, Örvell C, Tomac J, Forčić D 2015. Accumulation of defective interfering viral particles in only a few passages in Vero cells attenuates mumps virus neurovirulence. Microbes Infect 17:228–236. doi:.10.1016/j.micinf.2014.11.006 [PubMed] [Cross Ref]
29. Thompson KAS, Yin J 2010. Population dynamics of an RNA virus and its defective interfering particles in passage cultures. Virol J 7:257. doi:.10.1186/1743-422X-7-257 [PMC free article] [PubMed] [Cross Ref]
30. Timm C, Akpinar F, Yin J 2014. Quantitative characterization of defective virus emergence by deep sequencing. J Virol 88:2623–2632. doi:.10.1128/JVI.02675-13 [PMC free article] [PubMed] [Cross Ref]
31. Frensing T, Pflugmacher A, Bachmann M, Peschel B, Reichl U 2014. Impact of defective interfering particles on virus replication and antiviral host response in cell culture-based influenza vaccine production. Appl Microbiol Biotechnol 98:8999–9008. doi:.10.1007/s00253-014-5933-y [PubMed] [Cross Ref]
32. Bellett AJD, Cooper PD 1959. Some properties of the transmissible interfering component of vesicular stomatitis virus preparations. J Gen Microbiol 21:498–509. doi:.10.1099/00221287-21-3-498 [PubMed] [Cross Ref]
33. Marcus PI, Sekellick MJ 1974. Cell killing by viruses. I. Comparison of cell-killing, plaque-forming, and defective-interfering particles of vesicular stomatitis virus. Virology 57:321–338. [PubMed]
34. Khan S, Lazzarini R 1977. The relationship between autointerference and the replication of a defective interfering particle. Virology 77:189–201. doi:.10.1016/0042-6822(77)90417-2 [PubMed] [Cross Ref]
35. DePolo NJ, Giachetti C, Holland JJ 1987. Continuing coevolution of virus and defective interfering particles and of viral genome sequences during undiluted passages: virus mutants exhibiting nearly complete resistance to formerly dominant defective interfering particles. J Virol 61:454–464. [PMC free article] [PubMed]
36. Palma EL, Huang A 1974. Cyclic production of vesicular stomatitis virus caused by defective interfering particles. J Infect Dis 129:402–410. doi:.10.1093/infdis/129.4.402 [PubMed] [Cross Ref]
37. Von Laer DM, Mack D, Kruppa J 1988. Delayed formation of defective interfering particles in vesicular stomatitis virus-infected cells: kinetic studies of viral protein and RNA synthesis during autointerference. J Virol 62:1323–1329. [PMC free article] [PubMed]
38. Holland JJ, Villarreal LP, Breindl M 1976. Factors involved in the generation and replication of rhabdovirus defective T particles. J Virol 17:805–815. [PMC free article] [PubMed]
39. Huang AS. 1973. Defective interfering viruses. Annu Rev Microbiol 27:101–117. doi:.10.1146/annurev.mi.27.100173.000533 [PubMed] [Cross Ref]
40. Lim K-I, Lang T, Lam V, Yin J 2006. Model-based design of growth-attenuated viruses. PLoS Comput Biol 2:e116. doi:.10.1371/journal.pcbi.0020116 [PubMed] [Cross Ref]
41. Hensel SC, Rawlings JB, Yin J 2009. Stochastic kinetic modeling of vesicular stomatitis virus intracellular growth. Bull Math Biol 71:1671–1692. doi:.10.1007/s11538-009-9419-5 [PMC free article] [PubMed] [Cross Ref]
42. Timm C, Gupta A, Yin J 2015. Robust kinetics of an RNA virus: transcription rates are set by genome levels. Biotechnol Bioeng 112:1655–1662. doi:.10.1002/bit.25578 [PubMed] [Cross Ref]
43. Blumberg BM, Kolakofsky D 1983. An analytical review of defective infections of vesicular stomatitis virus. J Gen Virol 64:1839–1847. doi:.10.1099/0022-1317-64-9-1839 [PubMed] [Cross Ref]
44. Stauffer Thompson K, Rempala G, Yin J 2009. Multiple-hit inhibition of infection by defective interfering particles. J Gen Virol 90:888–899. doi:.10.1099/vir.0.005249-0 [PMC free article] [PubMed] [Cross Ref]
45. Stamminger GM, Lazzarini RA 1977. RNA synthesis in standard and autointerfered vesicular stomatitis virus infections. Virology 77:202–211. doi:.10.1016/0042-6822(77)90418-4 [PubMed] [Cross Ref]
46. McLain L, Armstrong SJ, Dimmock NJ 1988. One defective interfering particle per cell prevents influenza virus-mediated cytopathology: an efficient assay system. J Gen Virol 69:1415–1419. doi:.10.1099/0022-1317-69-6-1415 [PubMed] [Cross Ref]
47. Timm A, Yin J 2012. Kinetics of virus production from single cells. Virology 424:11–17. doi:.10.1016/j.virol.2011.12.005 [PMC free article] [PubMed] [Cross Ref]
48. Hu J, Nudelman G, Shimoni Y, Kumar M, Ding Y, López C, Hayot F, Wetmur JG, Sealfon SC 2011. Role of cell-to-cell variability in activating a positive feedback antiviral response in human dendritic cells. PLoS One 6:e16614. doi:.10.1371/journal.pone.0016614 [PMC free article] [PubMed] [Cross Ref]
49. Sekellick MJ, Marcus PI 1980. Viral interference by defective particles of vesicular stomatitis virus measured in individual cells. Virology 104:247–252. doi:.10.1016/0042-6822(80)90385-2 [PubMed] [Cross Ref]
50. Schulte MB, Andino R 2014. Single-cell analysis uncovers extensive biological noise in poliovirus replication. J Virol 88:6205–6212. doi:.10.1128/JVI.03539-13 [PMC free article] [PubMed] [Cross Ref]
51. Zhu Y, Yongky A, Yin J 2009. Growth of an RNA virus in single cells reveals a broad fitness distribution. Virology 385:39–46. doi:.10.1016/j.virol.2008.10.031 [PMC free article] [PubMed] [Cross Ref]
52. Rand U, Hillebrand U, Sievers S, Willenberg S, Köster M, Hauser H, Wirth D 2014. Uncoupling of the dynamics of host-pathogen interaction uncovers new mechanisms of viral interferon antagonism at the single-cell level. Nucleic Acids Res 42:e109. doi:.10.1093/nar/gku492 [PMC free article] [PubMed] [Cross Ref]
53. Downey MJ, Jeziorska DM, Ott S, Tamai TK, Koentges G, Vance KW, Bretschneider T 2011. Extracting fluorescent reporter time courses of cell lineages from high-throughput microscopy at low temporal resolution. PLoS One 6:e27886. doi:.10.1371/journal.pone.0027886 [PMC free article] [PubMed] [Cross Ref]
54. Sáez-Cirión A, Nicola M-A, Pancino G, Shorte SL 2006. Quantitative real-time analysis of HIV-1 gene expression dynamics in single living primary cells. Biotechnol J 1:682–689. doi:.10.1002/biot.200600045 [PubMed] [Cross Ref]
55. Knapp B, Rebhan I, Kumar A, Matula P, Kiani NA, Binder M, Erfle H, Rohr K, Eils R, Bartenschlager R, Kaderali L 2011. Normalizing for individual cell population context in the analysis of high-content cellular screens. BMC Bioinformatics 12:485. doi:.10.1186/1471-2105-12-485 [PMC free article] [PubMed] [Cross Ref]
56. Swick A, Baltes A, Yin J 2014. Visualizing infection spread: dual-color fluorescent reporting of virus-host interactions. Biotechnol Bioeng 111:1200–1209. doi:.10.1002/bit.25170 [PMC free article] [PubMed] [Cross Ref]
57. Wertz GW, Perepelitsa VP, Ball LA 1998. Gene rearrangement attenuates expression and lethality of a nonsegmented negative strand RNA virus. Proc Natl Acad Sci U S A 95:3501–3506. doi:.10.1073/pnas.95.7.3501 [PubMed] [Cross Ref]
58. Timm A. 2013. Development of single-cell techniques to investigate heterogeneity in virus-cell interactions. Ph.D. dissertation. University of Wisconsin—Madison, Madison, WI.
59. Waters JC. 2009. Accuracy and precision in quantitative fluorescence microscopy. J Cell Biol 185:1135–1148. doi:.10.1083/jcb.200903097 [PMC free article] [PubMed] [Cross Ref]
60. Hudson DJ. 1971. Interval estimation from the likelihood function. J R Stat Soc Ser B 33(2):256–262.
61. Mittler JE, Sulzer B, Neumann AU, Perelson AS 1998. Influence of delayed viral production on viral dynamics in HIV-1 infected patients. Math Biosci 152:143–163. doi:.10.1016/S0025-5564(98)10027-5 [PubMed] [Cross Ref]
62. Taniguchi Y, Choi PJ, Li G-W, Chen H, Babu M, Hearn J, Emili A, Xie XS 2010. Quantifying E. coli proteome and transcriptome with single-molecule sensitivity in single cells. Science 329:533–538. doi:.10.1126/science.1188308 [PMC free article] [PubMed] [Cross Ref]
63. Sekellick MJ, Marcus PI 1978. Persistent infection I. Interferon-inducing defective-interfering particles as mediators of cell sparing: Possible role in persistent infection by vesicular stomatitis virus. Virology 85:175–186. [PubMed]
64. Baum A, García-Sastre A 2011. Differential recognition of viral RNA by RIG-I. Virulence 2:166–169. doi:.10.4161/viru.2.2.15481 [PMC free article] [PubMed] [Cross Ref]
65. Otsuki K, Maeda J, Yamamoto H, Tsubokura M 1979. Studies on avian infectious bronchitis virus (IBV). Arch Virol 60:249–255. doi:.10.1007/BF01317496 [PubMed] [Cross Ref]
66. Lam V, Duca KA, Yin J 2005. Arrested spread of vesicular stomatitis virus infections in vitro depends on interferon-mediated antiviral activity. Biotechnol Bioeng 90:793–804. doi:.10.1002/bit.20467 [PubMed] [Cross Ref]
67. Azarin SM, Larson EA, Almodóvar-Cruz JM, de Pablo JJ, Palecek SP 2012. Effects of 3D microwell culture on growth kinetics and metabolism of human embryonic stem cells. Biotechnol Appl Biochem 59:88–96. doi:.10.1002/bab.1003 [PMC free article] [PubMed] [Cross Ref]
68. Snijder B, Sacher R, Rämö P, Damm E-M, Liberali P, Pelkmans L 2009. Population context determines cell-to-cell variability in endocytosis and virus infection. Nature 461:520–523. doi:.10.1038/nature08282 [PubMed] [Cross Ref]
69. You L, Suthers PF, Yin J 2002. Effects of Escherichia coli physiology on growth of phage T7 in vivo and in silico. J Bacteriol 184:1888–1894. doi:.10.1128/JB.184.7.1888-1894.2002 [PMC free article] [PubMed] [Cross Ref]
70. Wagner RR, Snyder RM, Yamazaki S 1970. Proteins of vesicular stomatitis virus: kinetics and cellular sites of synthesis. J Virol 5:548–558. [PMC free article] [PubMed]
71. Bevis BJ, Glick BS 2002. Rapidly maturing variants of the Discosoma red fluorescent protein (DsRed). Nat Biotechnol 20:83–87. doi:.10.1038/nbt0102-83 [PubMed] [Cross Ref]
72. Wagner RR, Levy AH, Snyder RM, Ratcliff GA, Hyatt DF 1963. Biologic properties of two plaque variants of vesicular stomatitis virus (Indiana serotype). J Immunol 91:112–122. [PubMed]
73. Giachetti C, Holland JJ 1989. Vesicular stomatitis virus and its defective interfering particles exhibit in vitro transcriptional and replicative competition for purified L-NS polymerase molecules. Virology 170:264–267. doi:.10.1016/0042-6822(89)90375-9 [PubMed] [Cross Ref]

Articles from Journal of Virology are provided here courtesy of American Society for Microbiology (ASM)