Mutagenesis-mediated virus extinction: virus-dependent effect of viral load on sensitivity to lethal defection.

Abstract

Lethal mutagenesis is a transition towards virus extinction mediated by enhanced mutation rates during viral genome replication, and it is currently under investigation as a potential new antiviral strategy. Viral load and virus fitness are known to influence virus extinction. Here we examine the effect or the multiplicity of infection (MOI) on progeny production of several RNA viruses under enhanced mutagenesis.
The effect of the mutagenic base analogue 5-fluorouracil (FU) on the replication of the arenavirus lymphocytic choriomeningitis virus (LCMV) can result either in inhibition of progeny production and virus extinction in infections carried out at low multiplicity of infection (MOI), or in a moderate titer decrease without extinction at high MOI. The effect of the MOI is similar for LCMV and vesicular stomatitis virus (VSV), but minimal or absent for the picornaviruses foot-and-mouth disease virus (FMDV) and encephalomyocarditis virus (EMCV). The increase in mutation frequency and Shannon entropy (mutant spectrum complexity) as a result of virus passage in the presence of FU was more accentuated at low MOI for LCMV and VSV, and at high MOI for FMDV and EMCV. We present an extension of the lethal defection model that agrees with the experimental results.
(i) Low infecting load favoured the extinction of negative strand viruses, LCMV or VSV, with an increase of mutant spectrum complexity. (ii) This behaviour is not observed in RNA positive strand viruses, FMDV or EMCV. (iii) The accumulation of defector genomes may underlie the MOI-dependent behaviour. (iv) LCMV coinfections are allowed but superinfection is strongly restricted in BHK-21 cells. (v) The dissimilar effects of the MOI on the efficiency of mutagenic-based extinction of different RNA viruses can have implications for the design of antiviral protocols based on lethal mutagenesis, presently under development.

Full text

Page 1

Mutagenesis-Mediated Virus Extinction: Virus-Dependent
Effect of Viral Load on Sensitivity to Lethal Defection
He ´ctor Moreno1, He ´ctor Tejero1,2, Juan Carlos de la Torre3, Esteban Domingo1,4*, Vero ´nica Martı ´n1,5
1Centro de Biologı ´a Molecular ‘‘Severo Ochoa’’ (CSIC-UAM), Cantoblanco, Madrid, Spain, 2Dpto. de Bioquı ´mica y Biologı ´a Molecular I. Universidad Complutense de
Madrid, Madrid, Spain, 3Department of Neuropharmacology, The Scripps Research Institute, La Jolla, California, United States of America, 4Centro de Investigacio ´n
Biome ´dica en Red de Enfermedades Hepa ´ticas y Digestivas (CIBERehd), Barcelona, Spain, 5Centro de Investigacio ´n en Sanidad Animal (CISA-INIA) Instituto Nacional de
Investigacio ´n Agraria y Alimentaria, Valdeolmos, Madrid, Spain
Abstract
Background: Lethal mutagenesis is a transition towards virus extinction mediated by enhanced mutation rates during viral
genome replication, and it is currently under investigation as a potential new antiviral strategy. Viral load and virus fitness
are known to influence virus extinction. Here we examine the effect or the multiplicity of infection (MOI) on progeny
production of several RNA viruses under enhanced mutagenesis.
Results: The effect of the mutagenic base analogue 5-fluorouracil (FU) on the replication of the arenavirus lymphocytic
choriomeningitis virus (LCMV) can result either in inhibition of progeny production and virus extinction in infections carried
out at low multiplicity of infection (MOI), or in a moderate titer decrease without extinction at high MOI. The effect of the
MOI is similar for LCMV and vesicular stomatitis virus (VSV), but minimal or absent for the picornaviruses foot-and-mouth
disease virus (FMDV) and encephalomyocarditis virus (EMCV). The increase in mutation frequency and Shannon entropy
(mutant spectrum complexity) as a result of virus passage in the presence of FU was more accentuated at low MOI for LCMV
and VSV, and at high MOI for FMDV and EMCV. We present an extension of the lethal defection model that agrees with the
experimental results.
Conclusions: (i) Low infecting load favoured the extinction of negative strand viruses, LCMV or VSV, with an increase of
mutant spectrum complexity. (ii) This behaviour is not observed in RNA positive strand viruses, FMDV or EMCV. (iii) The
accumulation of defector genomes may underlie the MOI-dependent behaviour. (iv) LCMV coinfections are allowed but
superinfection is strongly restricted in BHK-21 cells. (v) The dissimilar effects of the MOI on the efficiency of mutagenic-
based extinction of different RNA viruses can have implications for the design of antiviral protocols based on lethal
mutagenesis, presently under development.
Citation: Moreno H, Tejero H, de la Torre JC, Domingo E, Martı ´n V (2012) Mutagenesis-MediatedVirusExtinction:Virus-DependentEffectofViral Load onSensitivity
to Lethal Defection. PLoS ONE 7(3): e32550. doi:10.1371/journal.pone.0032550
Editor: Jean-Pierre Vartanian, Institut Pasteur, France
Received November 17, 2011; Accepted February 1, 2012; Published March 19, 2012
Copyright: ? 2012 Moreno et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Work at Centro de Biologı ´a Molecular Severo Ochoa is supported by grants BFU2008-02816/BMC from Ministerio de Ciencia e Innovacio ´n (MICINN) and
Fundacio ´n Ramon Areces. CIBERehd (Centro de Investigacio ´n Biome ´dica en Red de Enfermedades Hepa ´ticas y Digestivas) is funded by Instituto de Salud Carlos III.
HM is supported by FI08/00775 from Instituto de Salud Carlos III. HT is supported by AP2006-01044, from MEC (Spain) and Grant no. BFU2009-12895-C02-02 from
MEC (Spain). VM is supported by a contract RYC-2010-06516 from MICINN (Spain). The funders had no role in study design, data collection and analysis, decision
to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: edomingo@cbm.uam.es
Introduction
RNA viruses display high mutation rates and frequencies that
lead to the generation of dynamic mutant spectra termed viral
quasispecies [1–8]. Quasispecies dynamics represents a serious
impairment for the efficacy of antiviral treatments because of the
rapid selection of viral mutants resistant to one or several antiviral
agents. This problem has encouraged the search for alternative
approaches to control viral infections. A new antiviral strategy
termed lethal mutagenesis aims at increasing the viral mutation rate
beyond a biologically tolerable threshold, resulting in reduced viral
fitness and, eventually, virus extinction [9–14]. Arenaviruses are
attractive pathogens to investigate lethal mutagenesis because of the
very limited number of effective antiviral inhibitors available, and
the impact of arenavirus diseases for human health [15–18].
Lymphocytic choriomeningitis virus (LCMV), the prototype
arenavirus, is an enveloped virus with a bi-segmented negative
strand RNA genome. Each genome segment uses an ambisense
coding strategy to direct the synthesis of two viral polypeptides in
opposite directions and separated by a non-coding intergenic
region (IGR) [19,20]. The L segment encodes the Z protein, a
small RING protein that has matrix-like functions including
budding activity and regulation of RNA synthesis [21,22], and the
RNA-dependent RNA polymerase or L protein [23–26]. The S
segment encodes the glycoprotein precursor (GP-C) and the
nucleoprotein (NP) [17]. Mutation frequencies within mutant
spectra of molecular clones of LCMV replicated in BHK-21 cells
are in the range of 1.061024to 2.761024substitutions per
nucleotide [27], values which are comparable to those quantitated
for other RNA viruses [28,29].
PLoS ONE | www.plosone.org1 March 2012 | Volume 7 | Issue 3 | e32550

Page 2

In cell culture, the pyrimidine analogue 5-fluorouracil (FU) is
mutagenic for LCMV, as evidenced by increases of the average
mutation frequency of mutant spectra, which is associated with a
decreased specific infectivity (PFUs per LCMV RNA molecule)
[27,30–34]. FU mutagenesis led to LCMV extinction both, during
serial passages of the virus in BHK-21 cells, and upon extended,
persistent infection in BHK-21 cells (infected cells maintained for a
maximum of 100 h; no virus passage is involved [33,30]). In both
cases extinction occurred without any modification of the consensus
nucleotide sequence of the population [31,30,27,33,32,34]. LCMV
decreased its intracellular and extracellular infectivity to a larger
extent than the viral RNA levels [30]. This result, together with in
silico simulations of the LCMV population dynamics, led to the
formulation of the lethal defection model of virus extinction. This
model proposes the participation of a class of altered RNA genomes
termed defectors in the loss of infectivity [33,30]. Defectors are
RNA replication-competent genomes that encode altered versions
of viral proteins that interfere with replication of the standard virus
[30,35,36]. It must be underlined that, as defined, defectors differ
from standard defective-interfering (DI) particles in that DI
RNAs are dependent on standard virus for completion of their
infectious cycle [37], whereas defectors may not depend on the
standard virus.
DIs can be viewed as standard virus-dependent defectors. DIs of
LCMV are rapidly generated during replication, and underlie
homologous interference (also termed autointerference) in stan-
dard virus production [38–43]. In the description of our results we
refer to defectors to mean any type of genome (dependent or not
on the standard virus for replication) that has the potential to
interfere with replication of the standard virus. The lethal
defection model is congruent with data that indicate that
interactions within mutant spectra may influence adaptation or
de-adaptation of genome subpopulations [44–46,36,35,47–49].
Studies with the picornavirus foot-and-mouth disease virus
(FMDV) showed that low viral load and low fitness favored
mutagen-induced virus extinction [50,51]. In the present report,
we have examined the effect of the initial viral dose on LCMV
extinction, and found that following infection at low MOI, FU
inhibited production of infectious progeny of LCMV, as well as of
the rhabdovirus vesicular stomatitis virus (VSV), to a much larger
extent than FMDV or encephalomyocarditis virus (EMCV). We
propose that this different behaviour might relate to a favored
generation of defector genomes during LCMV or VSV replica-
tion, as compared to the picornaviruses FMDV or EMCV. This
interpretation is reinforced by the observation that passaged
EMCV acquires an enhanced FU inhibition at low MOI. Given
the increasingly recognized influence of the entire mutant
spectrum in virus behavior [7], we present an extended lethal
defection model that considers the effect that intracellular
interactions among standard, mutant, and defector viruses can
have on the extracellular virus dynamics. The model provides
good support to the experimental results. Implications for antiviral
therapy of the disparate effects of the initial viral load on the
response of viruses to mutagenesis are discussed.
Results
Effect of 5-fluorouracil and the multiplicity of infection on
lymphocytic choriomeningitis virus progeny production
Infections of BHK-21 cells with LCMV ARM 53b at MOIs of
0.001, 0.01, 0.1, 1, 3 and 10 PFU/cell, in the absence or presence
of FU (20, 35, and 50 mg/ml), documented a dose-dependent
inhibitory effect of FU at all MOIs tested, a decrease of progeny
production as a function of the MOI in the passages in the absence
of FU, and a decrease of inhibition by FU as a function of the
MOI (Figure 1A). A general linear model analysis of the logarithm
of the progeny production as a function of the logarithm of MOI
and the inhibition by FU shows a significant interaction between
the MOI and the inhibition by FU (p,0.05). Post Hoc multiple
pairwise comparisons were carried out to compare the inhibition
exerted by FU at the lowest MOI with that exerted at each of the
other MOI for all the FU concentrations tested (p,0.05; pairwise
t-test with Bonferroni Correction). This analysis revealed a
significant decrease in the inhibition at the higher MOI for all
the pairs tested. No significant differences were observed when
comparing the differences in inhibition between MOI 3 PFU/cell
and MOI 10 PFU/cell for all the concentrations of FU tested
(p.0.05; pairwise t-test with Bonferroni correction).
The negative correlation between the inhibition of LCMV
production by FU and the MOI was observed when the virus titer
was measured at 24 h and 48 h p.i., but not at 8 h p.i. (Figure 1B).
Also, the decrease of infectious LCMV progeny production in the
absence of FU when the MOI increased was observed at 48 h p.i.
but not at 8 h or 24 h p.i. (Figure 1B). The effect of MOI on
progeny production of LCMV RNA was less pronounced than the
effect on infectivity, and the differences between RNA progeny at
different MOIs did not reach statistical significance (Figs. 2B and
2C). Thus, both production of infectious LCMV progeny and the
effect of FU on such a production depend on the MOI and the
time p.i. at which the amount of progeny is determined.
Effect of 5-fluorouracil and the multiplicity of infection on
the replication of other RNA viruses
To investigate whether the decreased inhibition of LCMV
infectious progeny production by FU at high MOI could be a
more general phenomenon relevant also to other RNA viruses, we
carried out experiments with VSV, FMDV and EMCV (Figure 2).
The range of FU concentrations used in the experiments with
FMDV and EMCV was increased relative to the concentration
used with LCMV or VSV because of the lower sensitivity of these
picornaviruses to the inhibitory activity of FU [50,52,53]. As with
LCMV, the inhibition of VSV progeny production by FU
decreased with the MOI (Figure 2A). A general linear model
analysis evidenced a significant interaction between the logarithm
of the MOI of VSV and the inhibition by FU (p,0.005). Post Hoc
multiple pairwise comparisons showed that the difference in the
inhibition, relative to the inhibition at MOI of 0.001 PFU/cell,
was always statistically significant for all MOI and FU concentra-
tions tested (p,0.05; pairwise t-test with Bonferroni Correction).
As in the case of LCMV, no significant differences in inhibition
were observed when comparing the inhibitions at MOI 1 PFU/
cell and MOI 10 PFU/cell for all FU concentrations tested
(p.0.05; pairwise t-test with Bonferroni correction).
In contrast to LCMV and VSV, FMDV and EMCV did not
display any global trend towards decreased inhibition by FU as a
function of the logarithm of the MOI over the range of MOI
tested (p=0.325 and p=0.517, respectively; General Lineal
Model) (Fig. 3B and 3C). However, for FMDV a significant
difference between the inhibition exerted by FU in the infections
at MOI of 10 and MOI 0.001 PFU/cell was observed; for EMCV
the difference was significant when comparing the MOI of 1 PFU/
cell with 0.001 PFU/cell (p,0.05 in both cases; pairwise t-test with
Bonferroni correction; Fig. 3B and 3C). Thus, a negative
correlation between the inhibition of viral progeny production
by FU and the MOI is not a general occurrence for RNA viruses,
since it is clearly detected for LCMV and VSV, but not for FMDV
and EMCV (Figs. 2 and 3).
Lethal Defection in Viruses
PLoS ONE | www.plosone.org2March 2012 | Volume 7 | Issue 3 | e32550

Page 3

Lethal Defection in Viruses
PLoS ONE | www.plosone.org3March 2012 | Volume 7 | Issue 3 | e32550

Page 4

Genetic analysis of the 5-fluorouracil-treated viral
populations obtained at different multiplicities of infection
Extended viral replication often results in an increase of the
complexity of mutant spectra of viral populations [4,7]. To
investigate whether the amount of progeny virus correlated with
genetic heterogeneity, we determined the complexity of the
mutant spectra of LCMV populations, as well as values of progeny
infectivity and progeny RNA and specific infectivity at 48 h p.i.,
following infection in the absence or presence of FU, at a MOI
0.01 PFU/cell and 10 PFU/cell (Table 1). The specific infectivity
of progeny virus was about 10-fold lower in infections carried out
at high MOI relative to low MOI, in the absence of FU. In
contrast, the specific infectivity was about 30-fold higher at high
MOI in the presence of FU. For all comparisons, the mutation
frequency was significantly higher in the presence of FU than in its
absence (p,0,05; x2test), as expected from the mutagenic activity
of FU on LCMV [30,31,27,32–34]. Interestingly, the mutation
frequency of populations passaged in the presence of FU was 1.8-
to 2.4-fold higher at a MOI of 0.01 PFU/cell than at a MOI of 10
PFU/cell, a difference that was statistically significant (p=0.047;
x2test). The increase of Shannon entropy in the populations
passaged in the presence of FU was at least 1.6-to 6.4-fold higher
at low MOI than at high MOI, mirroring the effect of the MOI on
the mutation frequency (Table 1). In all cases, a decrease in
specific infectivity was accompanied by an increase of mutant
spectrum complexity (compare columns 5, 6 and 7 in Table 1).
Unexpectedly, the bias in mutation types associated with FU
mutagenesis differed between the L- and Z-genomic regions
analyzed. ARG transitions were the most frequent mutation types
in L, while URC transitions were the most frequent types in Z
(Table 1), and the difference was statistically significant (p,0.001;
x2test). This genome segment-dependent mutational bias is under
investigation.
The variation of mutant spectrum complexity as a result of FU
mutagenesis in the course of infections with VSV, FMDV and
EMCV at different MOIs showed dissimilar patterns. VSV
displayed a behaviour similar to LCMV, but none of the
differences reached statistical significance (Table 2). Under the
conditions of MOI in which LCMV and VSV manifested a higher
sensitivity to FU inhibition, its corresponding mutant spectrum
displayed a higher complexity. The predominant mutation types
induced by FU in VSV were ARG transitions. FMDV and
EMCV showed the opposite trend than LCMV or VSV regarding
the effect of FU or the MOI on mutant spectrum complexity, but
none of the differences reached statistical significance (Table 2).
Thus, these two picornaviruses deviate from the behaviour of
LCMV regarding the effect of FU and the MOI on mutant
spectrum complexity.
An extended lethal defection model
For a better understanding of the viral dynamics involved in
lethal mutagenesis of LCMV, and to help in the interpretation of
the experimental results, an extension of the lethal defection model
[30,54,46,35] has been developed. The extracellular virus
dynamics follows the basic model of viral dynamics [55]. It is
assumed that the cell can be in three different states: susceptible
(S), exposed (E), and infected (I) (Figure 3A). One virus particle can
infect an S cell, which becomes an E cell. The maximum number
of particles that can enter a cell is determined by the parameter
maxVirusxCell (see Table S1 for a list of parameters and their
numerical values). Thus, the model takes into account different
degrees of cell coinfection. Viruses are assumed to have the same
internalization rate, b, for S and E cells due to an excess of
receptor molecules on the cell surface [56]. Cells are in the E state
during early times after virus penetration, prior to becoming I cells
with a rate constant ke. New virus particles can enter E cells but not
I cells. I cells release b infectious progeny with a rate constant ki.
Since LCMV induces a persistent, non-cytolytic infection, the cell
population (confluent monolayer) is taken as constant during the
simulation, i.e. no cell death or cell growth are considered. Viruses
are degraded according to a constant rate u (Figure 3A). Modeling
of the intracellular dynamics has been based on our current
understanding of the basic steps of arenavirus multiplication
(Figure 3B). Three kinds of virus have been considered: (i) viable
viruses that can complete an infectious cycle and produce
infectious progeny by themselves; (ii) DI particles that cannot
multiply in the absence of a viable virus, but that have a replicative
advantage over the viable viruses when both coexist in the same
cell [37,57–59]. (iii) Defector viruses that are RNA replication-
competent and can interfere with replication or production of the
standard virus [30].
For simplicity, viral genomes are represented by four ‘‘func-
tional genosets’’. A functional genoset includes the genomic
positions involved in a given viral function, irrespective of their
location in the viral genome (Figure 3B). A given nucleotide can
belong to more than one genoset but, as a first approach, this fact
is not taken into account in this model. The effect of a genoset in a
given viral function is expressed by a numerical value that depends
on the number of mutations (relative to a given reference
sequence) at any position within the genoset. Therefore, a viral
genome is represented by a vector whose components are the
mutations at each genoset.
Genoset S includes the residues that encode the exclusively cis-
acting viral functions needed for the genome to be replicated.
Genoset R involves the genome positions that participate in any
viral function (needed for genome replication) that acts both in cis
and trans. Genoset P includes the nucleotide positions involved in
the production and release of viral particles. Genoset D
encompasses any residue involved in the generation of DI
particles. When the D genoset has zero mutations the genome is
viable, and when the D genoset has at least one mutation the
genome is a DI. Genosets R and P serve also to define two classes
of defector genomes: the R genoset-associated defectors are those
that can be replicated but do not contribute to the viral functions
needed for genome replication; the P genoset-associated defectors
are those that can be encapsidated but do not contribute to the
production and release of viral particles. The mathematical
Figure 1. Effect of the multiplicity of infection (MOI) on the inhibition of LCMV production by 5-fluorouracil (FU). BHK-21 cells were
infected with LCMV ARM 53b at the indicated MOI in the absence (white bars) or presence of different concentrations of FU (increasingly dark bars),
as indicated in each panel. The asterisks above the bars indicate those inhibitions that are significantly different from those obtained in the infection
at a MOI of 0.001 PFU/cell. A. Cell culture supernatants were harvested at 48 h p.i. and titrated. Viral titers are the average of at least four
determinations and standard deviations are given. All pairwise comparisons with the inhibition at MOI 0.001 PFU/cell were significantly different. The
inset visualizes the variation of the percentage of inhibition by 35 mg/ml FU as a function of the MOI. B. Similar to A, but with values for supernatants
harvested at 8, 24 and 48 h p.i. Insets are as in panel A. C. Number of molecules of L genomic viral RNA in the cell culture supernatants harvested at 8,
24 and 48 h p.i. (same samples as those analyzed in B). RNA measures are the average of three determinations and standard deviations are given (low
and not visible). Procedures are described in Materials and Methods.
doi:10.1371/journal.pone.0032550.g001
Lethal Defection in Viruses
PLoS ONE | www.plosone.org4March 2012 | Volume 7 | Issue 3 | e32550

Page 5

Figure 2. Effect of the multiplicity of infection (MOI) on the inhibition of vesicular stomatitis virus (VSV), foot-and-mouth disease
virus (FMDV), and encephalomyocarditis virus (EMCV) by 5-fluorouracil (FU). BHK-21 cells were infected with VSV (A), FMDV (B) or EMCV
(C) at the indicated MOI in the absence (white bars) or presence (increasingly dark bars) of different concentrations of FU, as indicated in each panel.
Cell culture supernatants were harvested at 24 h p.i. and titrated as described in Materials and Methods. Viral titers are the average of at least four
determinations and standard deviations are given. The asterisks above the bars indicate those inhibitions that are significantly different from those
obtained in the infection at a MOI of 0.001 PFU/cell. No asterisks were added for VSV because all the pairwise comparisons with the inhibition at MOI
0.001 PFU/cell were significantly different. The panels on the right visualize the variation of the percentage of inhibition (by 35 mg/ml FU for VSV,
200 mg/ml FU for FMDV and EMCV) as a function of the MOI.
doi:10.1371/journal.pone.0032550.g002
Lethal Defection in Viruses
PLoS ONE | www.plosone.org5March 2012 | Volume 7 | Issue 3 | e32550

Page 6

functions relating the mutations in a given genoset, and their
associated numerical functions are given in the Text S1.
Based on the four genosets, equations for the replication rate
and burst size have been formulated. The model considers a single
replication cycle. Replication of genome i produces a number of
RNA copies (wi) per round of replication, which is given by the
nearest integer number to wi~w0:si:(cisR:riztransR:? r r)(di),
where w0is the number of genomes produced by the standard
LCMV genome, siand riare the numerical values associated to the
S and R genosets, respectively, and cisR and transR are two
parameters that permit assigning more weight to either cis or trans
interactions in the functions associated with the R genoset; cisR and
transR must fulfill the condition that cisR+transR=1. The
intracellular interactions among different viral genomes is
introduced by means of the average replicative ability of the viral
population, ? r r~P
same cell. RepAdv is a parameter that expresses the replicative
advantage of DIs relative to standard and defector viruses. Thus,
i
ri, which is the average of the numerical values
associated with all the R genosets of the genomes present in the
the replication ability wiof a genome depends on the intracellular
viral genome population as a whole. During genome replication,
each genoset mutates according to a Poisson distribution whose
mean is the mutation rate per genoset (US, UR, UP, and UD). The
mutation rates per genoset are the product of the mutation rate
per site, m, and the genoset length (see Text S1 and Table S1). A
mutation rate per genome, U, is defined as the sum of the
mutation rates per genoset.
Each cell can produce a maximum number of viral particles
per unit time (maxBsize), which depends on the metabolic
and physiological state of the cell. The total number of virus
particles, infectious or not, b, produced by a cell, is given by
b~min(maxBsize,Ngen):? p p, where min(maxBsize, Ngen) equals
maxBsize when Ngen (which is the number of genomes inside the
cell) is larger than maxBsize size; min(maxBsize, Ngen) equals Ngen
when the latter is lower than maxBsize. To obtain the number of
virus released from the cell, min(maxBsize, Ngen) is multiplied by the
average efficacy of viral production in a cell, given by? p p~P
i
pi(a
term equivalent to the average numerical values for all the P
Figure 3. Basic features of the genosets model of lethal defection. A. Scheme of the infection of susceptible cells (S) by LCMV (V), stages of
the infection process, and main parameters involved in the model (further detailed in the text and in Text S1). B. Scheme of the intracellular dynamics
(excluding transcription) with formation of mutant genomes (dots on wavy lines) that may or may not be encapsidated. The boxes depict the
genoset organization, which does not fit the physical map of the LCMV genome (see text), and which forms the conceptual basis of the extended
lethal defection model presented here.
doi:10.1371/journal.pone.0032550.g003
Lethal Defection in Viruses
PLoS ONE | www.plosone.org6 March 2012 | Volume 7 | Issue 3 | e32550

Page 7

Table 1. Quasispecies analysis of FU-treated LCMV populations.
5-Fluorouracil
concentration
(mg/ml)a
Gene
analyzedb
MOI
(PFU/cell)a
Progeny
infectiviy
(PFU/ml)c
Progeny RNA
(molec/ml)d
Specific
Infectivity
(PFU/RNA
molec)e
Mutation
frequency
(sn/nt)f
Shannon
Entropyg
Sequenced
nucleotides
(nts)
Mutations Types
Transitions
Transversionsh
ARG
GRA
CRU
URC
CRA
0
L
0.01
(5.661.7)6108
(2.060.01)61012
2.761024
1.861024
0.13
16,800
0
1
1
1
0
0
L
10
(2.560.6)6105
(1.060.01)61011
2.361025
3.861024
0.27
15,680
4
0
0
0
1
0
Z
0.01
(5.661.7)6108
(2.060.01)61012
2.761024
,1.261024
0
7,800
0
0
0
0
0
0
Z
10
(2.560.6)6105
(1.060.01)61011
2.361025
2.561024
0.10
7,800
0
0
0
1
1
35
L
0.01
(1.762.1)6102
(3.760.01)6109
4.561028
2.261023
0.83
16,144
28
0
0
5
0
35
L
10
(8.762.5)6104
(6.660.01)61010
1.361026
9.061024
0.44
14,325
12
1
0
1
0
35
Z
0.01
(1.762.1)6102
(3.760.01)6109
4.561028
1.861023
0.46
7,800
2
1
0
11
0
35
Z
10
(8.762.5)6104
(6.660.01)61010
1.361026
1.061023
0.25
7,800
0
0
0
7
1
aBHK-21 cell monolayers were infected at the indicated MOI in the absence or presence of 5-fluorouracil, as detailed in Materials and Methods.
bThe genomic regions sequenced were the entire Z-coding region and residues 3654 to 4260 of the L-coding region.
cTiter of LCMV progeny determined at 48 h p.i.
dQuantification by real time RT-PCR of LCMV RNA in progeny, extracellular virus at 48 h p.i. in molecules/ml.
eRatio of virus titer and number of viral RNA molecules per ml of cell culture supernatant determined at 48 h p.i.
fAverage number of mutations per nucleotide, counted relative to the corresponding consensus nucleotide sequence.
gProportion of different genomic sequences in the mutant spectrum of the quasispecies, calculated as described in Materials and Methods.
hThe omitted transversion types were not represented in the sequences analyzed.
doi:10.1371/journal.pone.0032550.t001
Lethal Defection in Viruses
PLoS ONE | www.plosone.org7 March 2012 | Volume 7 | Issue 3 | e32550

Page 8

Table 2. Quasispecies analysis of FU-treated VSV, FMDV and EMCV populations.
5-Fluorouracil
concentration
(mg/ml)a
Virus
analyzedb
MOI
(PFU/cell)a
Progeny infectiviy
(PFU/ml)c
Mutation
frequency
(sn/nt)d
Shannon
Entropye
Sequenced
nucleotides
(nts)
Mutations Types
Transitions
Transversionsf
ARG
GRA
CRU
URC
CRA
ARC
URG
GRU
GRC
0
VSV
0.01
(2.764.1)61010
6.661025
0.05
30,060
0
1
0
0
0
0
1
0
0
0
VSV
10
(2.664.1)61010
6.261025
0.05
32,064
1
0
0
1
0
0
0
0
0
0
FMDV
0.001
(8.564.8)6104
1.361024
0.10
23,715
1
0
0
2
0
0
0
0
1
0
FMDV
1
(3.360.9)6106
3.961025
0.02
25,296
1
0
0
0
0
0
0
0
0
0
EMCV
0.001
(4.561.4)6108
1.961024
0.15
30,720
2
0
2
2
0
0
0
0
0
0
EMCV
1
(2.862.0)6108
3.461025
0.02
29,440
0
1
0
0
0
0
0
0
0
35
VSV
0.01
(1.360.1)6107
4.761024
0.41
32,064
9
3
2
1
0
1
0
0
0
35
VSV
10
(9.963.9)6108
2.561024
0.21
31,396
5
1
1
1
0
0
0
0
0
200
FMDV
0.001
(2.762.7)6102
1.961024
0.13
25,296
2
1
2
0
0
0
0
0
0
200
FMDV
1
(9.266.2)6103
4.561024
0.29
22,134
1
0
1
7
0
0
0
1
0
200
EMCV
0.001
(2.560.2)6107
1.961024
0.15
30,720
1
0
3
1
1
0
0
0
0
200
EMCV
1
(5.460.8)6107
2.361024
0.18
30,080
1
1
1
3
0
0
0
0
0
aBHK-21 cell monolayers were infected with VSV, FMDV or EMCV at the indicated MOI in the absence or presence of 5-fluorouracil, as detailed in Materials and Methods.
bThe genomic regions sequenced were nucleotides 5902 to 6569 of the L-coding region of VSV, 6800 to 7350 of the 3D (polymerase)-coding region of FMDV, and 6718 to 7403 pf the 3D (polymerase)-coding region of EMCV.
cTiter of VSV, FMDV and EMCV progeny determined at 24 h p.i.
dAverage number of mutations per nucleotide, counted relative to the corresponding consensus nucleotide sequence. The mutation frequency of VSV increased in the presence of FU, and was 1.9-fold higher at low than at high
MOI, but the differences were not statistically significant (p=0.33; x2test).
eProportion of different genomic sequences in the mutant spectrum of the quasispecies, calculated as described in Materials and Methods. Variations in Shannon entropy followed, the same trend as variations in mutation
frequency but, again, the differences were not statistically significant (see footnote d).
fThe omitted transversion types were not represented in the sequences analyzed.
doi:10.1371/journal.pone.0032550.t002
Lethal Defection in Viruses
PLoS ONE | www.plosone.org8March 2012 | Volume 7 | Issue 3 | e32550

Page 9

genosets present in the cell under consideration). Then, b genomes
are selected at random and released from the cell. When trans
interactions are not allowed, ? p p is taken as 1, and genomes are
encapsidated according to the value of their P genoset.
An initial virtual virus population was obtained in a preliminary
simulation consisting of a clone passaged ten times at a mutation
rate (U) of 0.15 mutations per genome and replication round, a
value which is used as the reference mutation rate throughout this
study, based on average mutation rates determined for RNA
viruses [60–62]. The initial population reached a mutation-
selection balance [63] and a constant fraction of DI particles. A
simulation begins with c0initial S cells that are infected at a given
MOI (arbitrarily chosen between 0.01 and 3 PFU/cell) by the
initial virus population, according to a Poisson distribution whose
mean is the MOI. The infection continued until either the virus
was exhausted or all the cells received the maximum number of
virus particles allowed (maxVirusxCell). The S cells became E cells,
and the simulation proceeded according to the parameters listed in
the Text S1 and Table S1. An agent-based computational
implementationof themodel
MATLABH. Details of the software will be published elsewhere
and provided upon request.
hasbeendevelopedusing
Computational simulations using the genosets model,
and comparison with the experimental results
The computational model (described in Materials and Methods,
with the invariant model parameters given in the Text S1 and
Table S1) predicts that when the mutation rate U is 0.15 mutations
per genome and replication round (m/g/r), infectious progeny
production should decrease as the MOI increases because of
homologous interference (autointerference) exerted by DI particles
(Figure 4A). In contrast, the converse is predicted when U is
increased to values of 4.5 to 7.5 m/g/r, (p,0.001, General Linear
Model). These higher U values necessitate enhanced mutagenesis
over the basal levels displayed by RNA viruses [62,60,61]
(Figure 4A). When an arbitrary time factor spanning values of 1
to 2.5 was considered, the decrease in infectious progeny
production with the MOI was only predicted at late times
(t=2.5) when U was 0.15 m/g/r. However, with a U of 6 m/g/r
the infectious yield increased with the MOI at all times tested
(t=1, 2, 2.5) (Figure 4B). When the total extracellular progeny
(infectious and non-infectious, taken as the total amount of viral
RNA) was considered, an increase in total progeny production
with the MOI was consistently observed taking a U of 6 m/g/r,
but not a U of 0.15 m/g/r (Figure 4C). The model predictions
regarding the effect of MOI and time of infection on progeny
production (Fig. 5B and 5C) are consistent with the experimental
results (Fig. 2B and 2C).
With a mutation rate U of 0.15 m/g/r, the model anticipates no
significant differences among the mutation frequencies at the three
MOI tested (p=0.063; one-way ANOVA test), which is the result
found with VSV in the absence of FU (Table 2). However, at U of
4.5 or 7.5 m/g/r the model predicts significant differences in
mutation frequency at the three MOI tested (p,0.001, for both
U=4.5 m/g/r and U=7.5 m/g/r; one-way ANOVA test)
(Table 3). Post Hoc analysis shows that at both U of 4.5 m/g/r
and U of 7.5 m/g/r, the mutation frequency at a MOI of 3 PFU/
cell is significantly lower than the mutation frequencies at a MOI
of 1 PFU/cell and 0.1 PFU/cell, (p,0.001 in all cases, Post-Hoc
Bonferroni analysis); also, the mutation frequency at a MOI of 1
PFU/cell is significantly greater than the mutation frequency at a
MOI of 0.1 PFU/cell (p,0.001 in all cases, Post-Hoc Bonferroni
analysis), as found experimentally (Table 1). The theoretical model
also predicted that the higher the MOI, the earlier the initiation of
viral production (Figure 5A), and this was verified experimentally
(Figure 5B).
To explore to what extent the effect of MOI on infectious
progeny production is affected by cis-trans interactions, simulations
were carried out with the computational model either with no trans
interactions allowed (cisR=1, transR=0), or with cis and trans
interactions allowed (cisR=0.25; transR=0.75). A significant
interaction between the MOI and the degree of cis-trans
interactions, cisR, was observed both for U=4.5 m/g/r and
U=7.5 m/g/r, (p,0,005 in both cases, General Linear Model)
indicating that the lower cisR, the higher the effect of the MOI on
infectious progeny production.
Influence of prior mutagenesis of encephalomyocarditis
virus on the multiplicity of infection-dependent
inhibition by 5-fluorouracil
The different influence of the MOI on the inhibition of
infectious progeny production by FU observed for different RNA
viruses was intriguing (Tables 2 and 3). We considered that the
different behavior of the viruses tested could be related to an easier
production of DIs by LCMV and VSV than by FMDV and
EMCV. Although DIs and other defective RNAs have been
described for FMDV and EMCV [64–69], their interfering
activity is generally less pronounced than that exerted by DIs of
negative strand RNA viruses [70,37,59]. Since passages of virus
increases the complexity of mutant spectra [4,7], we investigated
whether passage of EMCV could render the inhibition by FU
MOI-dependent. The virus passaged 20 times in the absence or
presence of ribavirin (R), a drug that is mutagenic for
picornaviruses [71–73], showed a statistically significant increase
of the inhibition by FU at low MOI relative to high MOI
(p=0.012, p=0.047,respectively;
(Figure 6). The result implies also that the MOI dependence of
the FU inhibition of progeny production can be influenced by a
population context provided by the passage history of a virus.
General linearmodel)
LCMV coinfections and reinfections in BHK-21 cells
Interference can be the result of mutated, defective genomes
that act on infectious genomes contained in the same cell, or
mutated genomes that are released from the cells where they are
produced and then can reinfect other cells together or sequentially
with standard infectious virus. To test whether a cell infected with
LCMV could be reinfected by another LCMV (that is, that there is
no superinfection exclusion that would preclude interference by
externally added mutants), we used trisegmented versions of
recombinant LCM viruses (r3LCMV) [74] expressing either RFP
(r3LCMV/RFP) of GFP (r3LCMV/GFP). Cells were first infected
with r3LCMV/RFP at MOIs of 0.001, 0.01, 0.1, 1 or 10 PFU/
cell, and the infection monitored based on intracellular RFP
expression. The first red cells could be detected at 24 h p.i. in the
infections carried out at MOIs of 0.001, 0.01 or 0.1 PFU/cell, and
at 8 h p.i. at a MOI of 1 or 10 PFU/cells (Figs. 7A and S1). At 0,
4, 8, 12 and 24 h after the first infection, the cells were infected
with r3LCMV/GFP at a MOI of 0.1 PFU/cell. Replication of the
second virus (green cells) was not observed in any case
(independent on the time between both infections) in cells that
had been infected with r3LCMV/RFP at high MOI (Figs. 7A and
S1). In cells infected with r3LCMV/GFP at low MOI, replication
of the second virus (r3LCMV/GFP) was observed when the time
between infections was 12 h or less (Figs. 7B and S1). Thus,
superinfections were heavily restricted in BHK-21 cells under the
infection conditions used. This has been introduced in the
extended lethal defection model by allowing new virus particles
Lethal Defection in Viruses
PLoS ONE | www.plosone.org9March 2012 | Volume 7 | Issue 3 | e32550

Page 10

Lethal Defection in Viruses
PLoS ONE | www.plosone.org 10March 2012 | Volume 7 | Issue 3 | e32550

Page 11

Figure 4. Viral production as a function of the multiplicity of infection (MOI) using the computational model developed in the
present study (described in Materials and Methods and the Text S1). The asterisks above the bars indicate those inhibitions that are
significantly different from those obtained at the lowest MOI in each simulation. A. Production of infectious virus (here termed PFU in the ordinate) at
low (U=0.15), medium (U=4.5) and high (U=6 and U=7.5) genomic mutation rate (mutations per genome and round of copying) for different MOIs.
B. Time dynamics of infectious virus production at different MOI and U=0.15 and U=6. Time is expressed in arbitrary units. C. Same as B but with
infectious and noninfectious virus, equivalent to total virus progeny. The parameters used are listed in Text S1 and Table S1, but in the simulations
performed in B and C the number of cells was reduced to 500. The results are the average of five simulations and the error bars are the standard
deviations of those five simulations.
doi:10.1371/journal.pone.0032550.g004
Figure 5. Kinetics of infectious progeny production as a function of the MOI, as predicted the theoretical model, and determined
experimentally. A. Computational simulation of infections at MOI of 0.1, 1 and 10 PFU/cell. The parameters used are listed in Text S1 and Table S1.
B. Experimental results: BHK-21 cells were infected with LCMV Arm53 at MOIs of 0.01, 1 and 10 PFU/cell, and supernatants were harvested at the
indicated times postinfection and titrated. Viral titers are the average of at least four determinations and standard deviations are given. Procedures
are described in Materials and Methods.
doi:10.1371/journal.pone.0032550.g005
Lethal Defection in Viruses
PLoS ONE | www.plosone.org11March 2012 | Volume 7 | Issue 3 | e32550

Page 12

to enter E cells but not I cells. The inability of defective interfering
particles to superinfect a cell was previously documented for
poliovirus [75]. Coinfection of the same cell by the two LCMV
versions was allowed, as suggested by the cells that included both
the red and green markers (yellow cells, color merge), detected at
time 0 between infections (Figures 7B and S1). Therefore,
interference by genomes introduced by external LCMV particles
requires the simultaneous infection by standard and interfering
virus because superinfections appear to be highly restricted. The
absence of reinfections implies that interference in a cell will be
exerted essentially by interfering genomes generated in the same
cell.
Discussion
The continuous presence of a dynamic mutant spectrum in
replicating RNA virus populations is a fundamental feature of RNA
viruses that exerts a profound influence in their biology, and can
jeopardize strategies to control viral disease [7,8,4]. Mutagenic
nucleoside analogues are currently being investigated as antiviral
agents because they can promote RNA virus extinction through
lethal mutagenesis [[52,50,76,51,30,33,77,78,53,73,79–81]; reviewed
in [11,12,9]]. Examination of the events that accompany FU-
mediated extinction of LCMV [31,33,27,34,30,32] led to the
proposal of lethal defection as a mechanism of virus extinction
[30]. The basic concept behind the lethal defection model is that
some functionally altered genomes or gene products that are
generated upon replication under enhanced mutagenesis interfere
with replication of the more competent genomes, thus contributing to
the replicative collapse of the entire viral population. Since its
formulation, the lethal defection model has been supported by
additionalexperimentalresultsobtainedwithLCMVandotherRNA
viruses, as well as by theoretical studies [76,36,47,33,45,30,35,49,82].
The present investigation was undertaken to explore with LCMV, a
system with a high endogenous production of defective genomes, the
effect of the infecting viral dose on mutagen-induced virus extinction.
The inhibitory activity of FU on infectious progeny production of
LCMV was more pronounced in infections carried out at low than at
high MOI, accompanied of a larger increase of mutant spectrum
complexity in the progenypopulations. While VSV, another negative
strand RNA virus, behaved similarly to LCMV, a low MOI did not
enhance either the inhibition by FU or the mutant spectrum
complexity of FMDV and EMCV (compare Figures 1, 2 and
Tables 1, 2). The results with VSV are in agreement with early
quantifications of the effect of FU mutagenesis with VSV by Holand
and collegues [77].The difference observed may be influenced bythe
lower tendency of the positive strand RNApicornaviruses to establish
interactions in trans as compared with negative strand RNA viruses
[83,84]. This is suggested by modest MOI-dependent progeny
production acquired by EMCV upon passage in BHK-21 cells
(Figure 6). Thus, passage history, related to production of, defective
genomes, might be the relevant factor irrespective of the virus being
of positive or negative sense RNA. The most abundant mutation
types expected from FU-mediated mutagenesis are ARG and URC
transitions [52,53,80,85,50], and these are also those observed in the
present experiments (Tables 1 and 2). In the case of LCMV, the most
frequent transitionswere ARG forgene L andURC forZ (Table 1).
An approach to validate the experimental results using totally
different tools has been to test whether the results are consistent
with theoretical predictions. To this aim, we have developed an
extended model inspired in previous versions of the lethal
defection model developed by Manrubia and her colleagues
[30,46,35]. The main features of this extended model are: i) cell
coinfection is included; ii) an adjustable degree of trans-interaction
is explicitly taken into account; iii) DI’s are considered as a subclass
of standard virus-dependent defectors; and iv) both replication and
release of viral progeny from the cell can be affected by the
mutation frequency. The model reproduces correctly the kinetics
Table 3. Computer simulations of the mutation frequency and percentage of inhibition of viral infectious progeny production, as
a function of the multiplicity of infection (MOI) and of the cis- trans interactions.
MOI (PFU/cell)Mutation frequencya
% Inhibition progeny
production Mutation frequencya
% Inhibition progeny
production
(mut/genome//mut/nt)relative to U=0.15b
(mut/genome//mut/nt)relative to U=0.15b
cisR=0.25 cisR=1
Average number of mutations per genome, U=0.15
0,11.0960.52//7.363.561024
-1.8860.92//1.360.661023
-
11.4160.65//9.464.361024
-1.5560.34//1.060.261023
-
31.6360.09//1.160.161023
- 1.6860.16//1.160.161023
-
Average number of mutations per genome, U=4.5
0,18.9360.48//6.060.361023
5464 9.1261.06//6.160.761023
4164
17.3860.54//4.960.461023
4469 6.9060.76//4.660.561023
3165
3 5.7960.17//3.960.161023
28695.3160.14//3.560.0961023
2364
Average number of mutations per genome, U=7.5
0,112.5760.59//8.460.461023
7762 11.2160.94//7.560.661023
6065
110.5960.21//7.160.161023
68619.2660.47//6.260.361023
5063
3 8.5260.11//5.6860.0761023
5364 7.5960.09//5.160.661023
4362
aThe mutation frequency is the average of ten simulations and is expressed as the number of mutations per genome (mut/genome) or mutations per nucleotide (mut/
nt).
bThe inhibition of infectious virus progeny production is calculated according to I(U)~(1{
(the equivalent of viral titer) when the mutation rate is U=0.15 mutations/genome/replication round, and V(U) is the number of viable progeny viruses when the
mutation rate is U=4.5 or U=7.5. Results are the average of ten simulations.
doi:10.1371/journal.pone.0032550.t003
V(U)
V(U~0:15)):100, where, V(U=0.15) is the number of viable progeny viruses
Lethal Defection in Viruses
PLoS ONE | www.plosone.org12March 2012 | Volume 7 | Issue 3 | e32550

Page 13

Figure 6. Effect of the multiplicity of infection (MOI) on the inhibition of mutagenized and non-mutagenized EMCV progeny
production by 5-fluorouracil (FU). A. BHK-21 cells were infected with EMCV passaged 20 times in absence of ribavirin at the indicated MOI, in the
absence (white bars) or presence of different concentrations of FU (increasingly dark bars), as indicated in each panel. Cell culture supernatants were
harvested at 24 h p.i. and titered. Viral titers are the average of at least three determinations and standard deviations are given. The inset visualizes
the variation of the percentage of inhibition (by 200 mg/ml FU) as a function of the MOI. B. Same experiment, but using a EMCV passaged 20 times in
the presence of 800 mM ribavirin. The asterisks above the bars indicate the FU-inhibition values that are significantly different from those obtained
with the corresponding lowest MOI tested. The inset is as in panel A. Procedures are described in Materials and Methods.
doi:10.1371/journal.pone.0032550.g006
Lethal Defection in Viruses
PLoS ONE | www.plosone.org13 March 2012 | Volume 7 | Issue 3 | e32550

Page 14

of progeny production and autointerference of LCMV observed at
high MOI [86,42,87] (Figures 4 and 5). The model also
reproduces the negative correlation between the inhibition of
infectious virus production by FU and the MOI when the
mutation rate is increased. In infections carried out at low initial
MOI a larger number of infection rounds are needed to infect all
the cells and, therefore, viruses can acquire a larger number of
mutations per genome. In contrast, high MOI increases the
number of initial infections, and reduces the number of replication
rounds needed to reach the plateau of virus production. As a
consequence, the viruses accumulate a lower average number of
mutations, as reflected in the decrease in the mutation frequency
Figure 7. LCMV coinfections and reinfections of BHK-21 cells. Cells were infected with RFPrLCMV at the indicated MOI. At 0, 4, 8, 12 or 24 h
p.i. cells were re-infected with GFPrLCMV at a MOI of 0.1 PFU/cell. The different panels show representative images of infected cells at 0 (A) and 48 (B)
h p.i. with GFPrLCMV. Magnification is 20-fold. Nuclei were stained with DAPI. The origin of the marked RFPrLCMV and GFPrLCMV and procedures for
the infections are described in Materials and Methods. Additional images of coinfections and reinfections at different MOI, and controls are depicted
in Figures S1, S2 and S3.
doi:10.1371/journal.pone.0032550.g007
Lethal Defection in Viruses
PLoS ONE | www.plosone.org 14 March 2012 | Volume 7 | Issue 3 | e32550

Page 15

(Table 3). An increase of the mutation frequency at low MOI
favors the production of mutant and defector viruses, which in
turn decrease the infectious progeny production. The lower
progeny production of LCMV at high MOI may be due to
interfering genomes and subsequent autointerference [86,87]. The
effect was noticed at late times presumably because interfering
genomes had accumulated (Figure 1).
Results of infections with differentially labeled LCMVs
indicated that while coinfections are allowed, superinfection is
strongly restricted in BHK-21 cells (Figure 7). Restriction of
superinfection was also observed with Junin virus in persistently
infected BHK and Vero cells [88,89]. Thus, interference and
autointerference depend on the coinfection of cells by standard
and defective genomes, and on the intracellular generation of
mutant genomes, favored by FU mutagenesis.
Our experimental findings, supported by the extended lethal
defection model, are relevant to the planning of lethal mutagenesis
protocols because lower mutagenic intensities may be sufficient to
achieve the extinction of viruses that are prone to produce defective
genomes during their natural infectious cycle. It will be of interest to
compare the effect of sequential and combination treatments
involving a mutagenic agent and a non-mutagenic inhibitor
[82,85,90,76] on viruses prone to generate and tolerate a higher
basal level of defective genomes (without an added mutagen) and on
viruses restricted in defective genome generation and tolerance.
Materials and Methods
Cells, virus and infections
Growth of BHK-21 and Vero cells and virus infections were
carried out as previously described [31,91,50,27,30,33,32]. LCMV
ARM 53b is a triple plaque-purified clone from ARM CA 1371,
passaged four times in BHK-21 cells. An LCMV virus stock (termed
p0) was prepared by infecting BHK-21 monolayers (36106cells in
100 mm-diammeter dishes)with 0.01 PFU of LCMVARM53b per
cell. The p0 virus preparation was used for all the experiments.
Other viruses employed are the Mudd-Summers strain of the
Indiana serotype of VSV [92], clone C-S8c1of FMDV [93], and the
Rueckert strain of EMCV [94]. Procedures for LCMV infection
have been described [33,27,32,31,30]. Briefly, semiconfluent
monolayers of BHK-21 cells (36106cells in 100 mm-diammeter
dishes) were infected with LCMV ARM 53b p0 at different MOIs.
After a 90 min adsorption period at 37uC and 7% CO2, the inocula
were removed and the monolayers washed with Dulbecco modified
Eagle medium (DMEM). The infected cultures were maintained in
10 mlofDMEMsupplementedwith10%fetalcalfserum(FCS),2%
L-glutamine, 0.52% glucose, 50 mg/ml gentamicin, either in the
absence or the presence of FU (concentrations indicated in each
experiment) at 37uC and 7% CO2. Supernatants of the infected cell
culturesweresampledatdifferenttimespostinfection(p.i.)andstored
at 280uC. The infectivity of extracellular virus was determined by
plaque assay on Vero cell monolayers as follows: 106cells per well in
six-well dishes were infected by applying 300 ml of a dilution of the
culture medium to be tested; after 90 minutes of adsorption at 37uC
and 7% CO2, the monolayers were washed with DMEM and
overlaid with 4 ml of DMEM 1% FCS, 1% DEAE-Dextran, 2% L-
glutamine, 50 mg/ml gentamicin and 0,3% agar. After 7 days, cells
were stained with 2% crystal violet in 2% formaldehyde, and viral
plaques were counted. Supernatants from mock-infected cultures
maintained in parallel were titrated to monitor the absence of viral
contamination. No contamination (cytopathology or plaques) was
detected in the control cultures throughout the experiments.
Infections of BHK-21 cells with FMDV, EMCV and VSV
(66106cells in 100 mm-diameter dishes) at different MOI were
done as described [95,93,96]. The infected cultures were harvested
24 hours (h) p.i. and stored at 280uC. Virus infectivity was
determined by plaque assay on BHK-21 cell monolayers as
follows: 106cells in each of 35 mm-diammeter dishes were
infected by applying 100 ml of serial dilutions of the culture
medium; after 60 minutes of adsorption at 37uC and 7% CO2, the
monolayers were washed with DMEM and overlaid with 4 ml of
DMEM 1% FCS, 1% DEAE-Dextran, 50 mg/ml gentamicin and
0,5% agar. After 24 h, cells were stained with 2% crystal violet in
2% formaldehyde, and viral plaques were counted. Titers shown
are the mean of at least three determinations. Again, mock-
infected cultures were maintained in parallel to ascertain the
absence of undesired viral infections.
To study the permissivity of BHK-21 cells to superinfection by
LCMV, we used trisegmented versions of LCMV [74] engineered to
express either the red fluorescent protein (RFP) (r3LCMV/RFP) or
the green fluorescent protein (GFP) (r3LCMV/GFP). Semiconfluent
monolayers of BHK-21 cells (16106cells in 16 mm-diameter dishes)
were infected with RFPrLCMV at different MOIs. Then the cell
monolayers were washed, and the cells reinfected with GFPrLCMV
at a MOI of 0.1 PFU/cell, at different times after the first infection.
At 0 and 48 h after the second infection, the cells werefixed with 4%
formaldehyde, washed two times with PBS, and cell nuclei were
counterstained using DAPI for 10 minutes, and then maintained in
PBS (100 ml/well) [74]. An inverted Axiovert S100 Zeiss fluores-
cence microscope was used to visualize the samples.
Treatment with 5-fluorouracil
Preparation of FU stock solutions, determination of FU toxicity
for BHK-21 cells, and procedures for viral infections in the
presence of FU have been described [30,50].
RNA Extraction, reverse transcription–PCR, nucleotide
sequencing, and determination of mutant spectrum
complexity
RNA was extracted from supernatants of infected cultures using
Trizol (Invitrogen), following the manufacturer’s protocol. Trip-
licate RNA samples were amplified by reverse transcription–PCR
(RT-PCR) using RT transcriptor (Roche) and PFU DNA polymerase
(Promega). To ascertain that an excess of template was amplified
and that the amount of template was not limiting during the RT-
PCR amplification, a 1:10 and 1:100 dilution of the RNA was
subjected to RT-PCR in parallel. Positive amplifications with the
diluted template ensured that there was no limitation in the
amount of viral RNA molecules as template for the RT-PCR
amplification carried out with undiluted RNA [72]. cDNAs were
purified with a Wizard PCR purification kit (Promega), pGEM-T
Easy Vector (Promega), and cloned in Escherichia coli DH5a. cDNA
from individual bacterial colonies was amplified with Templiphi
(GE Healthcare). The oligonucleotides used for amplification of
two LCMV genomic regions were: L3654 (forward 59- AGT TTA
AGA ACC CTT CCC GC - 39) and L4260 (reverse 59 CGA GAC
ACC TTG GGA GTT GTG C - 39) for the polymerase region;
L7 (forward 59- GGG GAT CCT AGG CGT TTA GT- 39) and
L402 (reverse 59- GGA ACC GCA CGT CGC CCA ACG CAC -
39) for the Z gene. The number in the primer designation
corresponds to the 59 nucleotide position, and refers to the
consensus genomic RNA sequence determined previously [31]
[GenBank accession numbers AY847351(L) and AY847350(S)].
The oligonucleotides used for amplification of a fragment of the
polymerase gene of VSV, FMDV and EMCV were L5902F
(forward 59- GCAAGTGATTTAGCTCGGATT - 39; residues
5903-5923) andL6569R(reverse59-GGTGGTTATTC-
Lethal Defection in Viruses
PLoS ONE | www.plosone.org15 March 2012 | Volume 7 | Issue 3 | e32550