Intracellular bottlenecking permits no more than three tomato yellow leaf curl virus genomes to initiate replication in a single cell

Abstract

Viruses are constantly subject to natural selection to enrich beneficial mutations and weed out deleterious ones. However, it remains unresolved as to how the phenotypic gains or losses brought about by these mutations cause the viral genomes carrying the very mutations to become more or less numerous. Previous investigations by us and others suggest that viruses with plus strand (+) RNA genomes may compel such selection by bottlenecking the replicating genome copies in each cell to low single digits. Nevertheless, it is unclear if similarly stringent reproductive bottlenecks also occur in cells invaded by DNA viruses. Here we investigated whether tomato yellow leaf curl virus (TYLCV), a small virus with a single-stranded DNA genome, underwent population bottlenecking in cells of its host plants. We engineered a TYLCV genome to produce two replicons that express green fluorescent protein and mCherry, respectively, in a replication-dependent manner. We found that among the cells entered by both replicons, less than 65% replicated both, whereas at least 35% replicated either of them alone. Further probability computation concluded that replication in an average cell was unlikely to have been initiated with more than three replicon genome copies. Furthermore, sequential inoculations unveiled strong mutual exclusions of these two replicons at the intracellular level. In conclusion, the intracellular population of the small DNA virus TYLCV is actively bottlenecked, and such bottlenecking may be a virus-encoded, evolutionarily conserved trait that assures timely selection of new mutations emerging through error-prone replication.

Full text

RESEARCH ARTICLE
Intracellular bottlenecking permits no more
than three tomato yellow leaf curl virus
genomes to initiate replication in a single cell
Ruifan Ren
1,2,3☯
, Limin Zheng
2☯
, Junping Han
2
, Camila Perdoncini Carvalho
2
,
Shuhei Miyashita
4
, Deyong Zhang
1,3
*, Feng QuID
2
*
1Longping Branch, College of Biology, Hunan University, Changsha, China, 2Department of Plant
Pathology, The Ohio State University, Wooster, Ohio, United States of America, 3Hunan Plant Protection
Institute, Changsha, China, 4Graduate School of Agricultural Science, Tohoku University, Sendai, Japan
☯These authors contributed equally to this work.
*dyzhang78@163.com (DZ); qu.28@osu.edu (FQ)
Abstract
Viruses are constantly subject to natural selection to enrich beneficial mutations and weed
out deleterious ones. However, it remains unresolved as to how the phenotypic gains or
losses brought about by these mutations cause the viral genomes carrying the very muta-
tions to become more or less numerous. Previous investigations by us and others suggest
that viruses with plus strand (+) RNA genomes may compel such selection by bottlenecking
the replicating genome copies in each cell to low single digits. Nevertheless, it is unclear if
similarly stringent reproductive bottlenecks also occur in cells invaded by DNA viruses. Here
we investigated whether tomato yellow leaf curl virus (TYLCV), a small virus with a single-
stranded DNA genome, underwent population bottlenecking in cells of its host plants. We
engineered a TYLCV genome to produce two replicons that express green fluorescent pro-
tein and mCherry, respectively, in a replication-dependent manner. We found that among
the cells entered by both replicons, less than 65% replicated both, whereas at least 35% rep-
licated either of them alone. Further probability computation concluded that replication in an
average cell was unlikely to have been initiated with more than three replicon genome cop-
ies. Furthermore, sequential inoculations unveiled strong mutual exclusions of these two
replicons at the intracellular level. In conclusion, the intracellular population of the small
DNA virus TYLCV is actively bottlenecked, and such bottlenecking may be a virus-encoded,
evolutionarily conserved trait that assures timely selection of new mutations emerging
through error-prone replication.
Author summary
An important unresolved issue in virus life cycles is how natural selection acts on individ-
ual virus copies occupying the same cell. Unlike cellular organisms in which a chromo-
some harboring an advantageous or deleterious mutation usually shares the hosting cells
with no more than one homologous sister chromosome, viruses could potentially
PLOS PATHOGENS
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 1 / 21
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Ren R, Zheng L, Han J, Perdoncini
Carvalho C, Miyashita S, Zhang D, et al. (2023)
Intracellular bottlenecking permits no more than
three tomato yellow leaf curl virus genomes to
initiate replication in a single cell. PLoS Pathog
19(5): e1011365. https://doi.org/10.1371/journal.
ppat.1011365
Editor: Aiming Wang, Agriculture and Agri-Food
Canada, CANADA
Received: December 31, 2022
Accepted: April 17, 2023
Published: May 1, 2023
Copyright: ©2023 Ren 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.
Data Availability Statement: The data collected in
this study has been provided in the Supporting
Information (S1 and S2 Datasets).
Funding: This study is supported in part by the
NSF grant 1758912 awarded to FQ. RFR is
supported in part by a scholarship from China
Scholarship Council, and the Postgraduate
Scientific Research Innovation Project of Hunan
Province (QL20210118). LMZ is supported in part
by a grant from the National Key R&D Program of

reproduce with numerous genome copies per cell, permitting sharing of protein products,
thereby greatly diminishing phenotypic impacts of otherwise eventful mutations. Previous
investigations suggest that (+) RNA viruses solve this problem by bottlenecking the num-
ber of replicating genome copies in each cell to low single digits. The current study reveals
strikingly similar intracellular population bottlenecks for a small DNA virus. Further
mechanistic interrogations could avail the virus-encoded bottleneck-enforcing apparatus
as targets for antiviral therapy and prevention.
Introduction
The replication processes of many viruses are highly error-prone, causing most descendant
viral genome copies to differ from their parents by a minimum of one mutation [1–5]. While
these mutations are mostly phenotypically neutral or near neutral, lethal mutations do occur
and can reach high numbers, given that millions of new viral genome copies are frequently
produced in each cell. Loss-of-function mutations in viral proteins essential for the replication
of viral genomes, if not promptly isolated and purged, pose serious threat to the survival of the
cognate virus population. This is because the mutation-containing “cheater” genomes could
steal the corresponding mutation-free proteins produced by sister genome copies in the same
cell to support their own replication. More ominously, absent of diligent surveillance, similar
cheater mutants arise ceaselessly as the replication reiterates. Together these mutant genomes
could quickly reproduce themselves to dominance, and steadily dilute out the genome copies
still producing mutation-free proteins, ultimately obliterating the cognate viral population
[1,2,6,7].
The same challenge also applies to positive selection responsible for enriching beneficial
viral mutations. By definition, positive selection feeds the phenotypic gain arisen from a bene-
ficial mutation back to the very genome copy carrying this mutation, causing it to have more
surviving descendants than its sister copies. However, such exclusive phenotype-to-genotype
feedback is difficult to achieve if mutation-endowed phenotypic gains, most frequently in the
form a fitter protein, must be shared among numerous genome copies in the same cell, with
the mutation-carrying copy being just one of them. This challenge is further exacerbated by
the fact that most beneficial mutations confer minimal phenotypic gains, which could be easily
lost if not promptly selected. Therefore, the Darwinian law of natural selection predicts that
successful viruses must constrain the number of the reproductive genome copies per cell in
order to minimize phenotype cheating or sharing. However, exactly how this is accomplished
remains poorly understood.
Recent observations by us and others prompted a new model that explains how plus-strand
(+) RNA viruses achieve reproductive isolation [8–11]. Briefly, this Bottleneck, Isolate,
Amplify, Select (BIAS) model postulates that multiple genome copies of the same virus, upon
penetrating the same cell, cooperate to erect intracellular population bottlenecks from which
very few copies (as few as one) of the viral genome could escape to initiate replication. Due to
the stochastic nature of the bottlenecks, a viral genome copy with a lethal mutation has the
same chance as other sister copies to escape the bottlenecks and use replication proteins pro-
duced by the latter to initiate replication. However, should this occur, its replication would
lead to the amplification of descendants that all contain the same lethal mutation. Upon invad-
ing a fresh cell collectively, such a mutant genome lineage would then be forced to bear the
consequence of the lethal mutation, because the mutation-complementing sister genomes
would now be absent (or inadequate if the bottleneck size is >1), leading to the elimination or
drastic suppression of viral genome copies bearing lethal mutations.
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 2 / 21
China (2018YFE0112600). SM is supported in part
by Japan Society for the Promotion of Science
(JSPS) KAKENHI grant 21K05591. The funders
played no roles 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.

Obviously, the same BIAS arrangement also assures positive selection that proliferates ben-
eficial mutations as soon as they emerge. Nevertheless, while the BIAS model is amply sup-
ported by evidence derived from (+) RNA viruses, it is not known whether it also applies to
DNA viruses. To address this knowledge gap, here we examined a small DNA virus to deter-
mine whether its populations also encountered narrow reproductive bottlenecks in infected
cells. We adopted tomato yellow leaf curl virus (TYLCV) as the model virus for this study.
TYLCV is a member of the genus Begomovirus, family Geminiviridae, with a single-stranded,
circular DNA genome of approximately 2,800 nucleotides (nt), encoding at least six proteins
[12–14]. The four genes on the complementary strand of the genome are early expressing,
encoding proteins (C1-C4) that participate in various aspects of viral genome replication, tran-
scriptional activation, and host defense neutralization (Fig 1A) [13,15]. In particular, C1 is
absolutely required for the rolling circle replication of TYLCV genome [16–18]. On the other
hand, the two genes on the viral strand are both late expressing, dependent on successful
genome replication. They encode V1 and V2, which are capsid protein (CP) and suppressor of
RNA silencing, respectively (Fig 1A). V1 (CP) is not essential for intracellular replication of
TYLCV [19], but needed for the intra- and inter-cellular movement of the virus [20]. In the
current report, the genome of the TYLCV SH2 isolate was modified to encode nuclearly local-
ized fluorescent proteins in place of V1, permitting convenient tracking of viral replication in
single cells. We found that TYLCV intracellular populations were constrained by stringent
bottlenecks that limit replicating genome copies to no more than three in each cell. Further-
more, TYLCV genomes that entered cells early exerted strong superinfection exclusion (SIE)
against those entering merely 24 hours later. Together these data suggest that small DNA
viruses like TYLCV may also utilize a strategy similar to BIAS to enable timely natural selec-
tion of new mutations.
Results
TYLCV replicons readily replicate in cells of Nicotiana benthamiana leaves
The SH2 isolate of TYLCV has a single-stranded, circular DNA genome of 2,781 nt (Fig 1A).
Previous studies by others established that expression of TYLCV V1 gene, encoding viral CP,
occurs in a replication-dependent manner [21–23]. We hence reasoned that its replication
could be monitored microscopically by replacing part of the CP coding sequence with that of
green fluorescence protein (GFP) or the red fluorescent mCherry. Such replacements gave rise
to two TYLCV replicons designated as LM4-G and LM4-R, respectively (Fig 1A). Note that
the V1 and V2 genes partially overlap. As a result, the V2-overlapping portion of V1 was
retained in LM4-G and LM4-R in the form of N-terminal fusions to GFP and mCherry. This
65 amino acid (aa) N-terminal fusion harbors a nuclear localization signal (NLS) that routes
most of the GFP and mCherry signals to the cell nuclei, permitting convenient quantification
of TYLCV-replicating cells. Also worth noting is that CP replacement by GFP/mCherry abol-
ished viral cell-to-cell movement, thus restricting the replication of LM4-G and LM4-R in sin-
gle cells.
To test their replicability, the LM4-G and LM4-R replicons were cloned into the pAI101
binary vector [24]. The resulting constructs, still referred to as LM4-G and LM4-R for simplic-
ity, were transformed into Agrobacterium tumefaciens (strain C58C1; simplified as agro hereaf-
ter), suspensions of which were pressure-infiltrated into leaves of N.benthamiana plants
(agro-infiltration). Both replicons produced abundant, strongly fluorescent signals that were
predominantly located in the nuclei of epidermal cells, though modest levels of fluorescence
were also detected in the cytoplasm of some cells (Fig 1B, panels 1 and 2). This result demon-
strated that both LM4-G and LM4-R replicated robustly in epidermal cells of N.benthamiana.
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 3 / 21

Fig 1. TYLCV-derived replicons LM4-G and LM4-R replicate to produce fluorescent signals that are
predominantly localized to nuclei of N.benthamiana cells. A. Schematic depiction of the 2,781-nt TYLCV genome
(isolate SH2) in linearized, double-stranded form inserted in a binary shuttle plasmid (pAI. See Materials and Methods
for details). Note that nt positions 2566–2781 (256 nt) and 1–158 (158 nt) were duplicated at 5’ and 3’ ends,
respectively. The large arrows depict various TYLCV-encoded proteins. The V1 ORF, which was modified to
accommodate GFP and mCherry insertions in LM4-G and LM4-R, is highlighted as solid black. The 65-aa N-terminal
nuclear localization signal (NLS) of V1 was retained as N-terminal fusions of GFP and mCherry. The C1 ORF was also
modified in one of the constructs (LM4-C1fs-G) and is depicted as a gray arrow. Also note the small blue arrows on the
top, denoting two primers (SH2-495R and SH2-2255F) used for specific detection of circularized form of the
replication-generated viral DNA. B. Replication-dependent expression of GFP and mCherry from LM4-G and LM4-R,
respectively. The LM4-C1fs-G mutant failed to express GFP as a result of C1 disruption. Bar = 200 μm. C.
Complementation of LM4-C1fs-G by co-delivered LM4-R. D. Verification of replication of LM4-G and LM4-R, as well
as replicational complementation of LM4-C1fs-G by LM4-R, with PCR (top) coupled with MfeI digestion. A 436-bp
ribosomal DNA (rDNA) fragment was simultaneously amplified (18 PCR cycles) to serve as the control to ensure the
samples contained similar amounts of template DNA. Sequences of the rDNAprimers are available upon request. The
MfeI site present in wildtype (wt) C1 coding sequence was abolished in LM4-C1fs-G by the one-ntinsertion within the
site that also disrupted the C1 ORF. E. Western blotting detection of GFP and mCherry proteins expressed by the
LM4-G and LM4-R replicons.
https://doi.org/10.1371/journal.ppat.1011365.g001
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 4 / 21

To further confirm that expression of fluorescent proteins from LM4-G and LM4-R was
dependent on viral replication, we produced the LM4-C1fs-G mutant by inserting one nt early
in the C1 open reading frame (ORF) of LM4-G, causing the C1 frame to shift and stop prema-
turely (Fig 1A, bottom right). This mutant no longer produced any GFP fluorescence (Fig 1B,
panel 3). Crucially, the C1 defect of LM4-C1fs-G was successfully complemented by wildtype
C1 proteins provided in trans from the co-delivered LM4-R replicon, causing most of treated
cells to express both GFP and mCherry (Fig 1C).
Replication of LM4-G and LM4-R, as well as replicational complementation of LM4-C1fs-
G by LM4-R, were further verified with two additional approaches. First, 15 cycles of polymer-
ase chain reaction (PCR) were carried out with a pair of primers (SH2-2255F and SH2-495R;
Fig 1A) that could only produce a fragment from replication-generated circular viral genomes.
As illustrated in Fig 1A, these two primers annealed to positions outside of the duplicated ter-
mini (nt positions 2,566–2,781 at the left terminus, 1–158 at the right terminus) of LM4, and
oriented away from each other in the linear genome inserted in the replicon constructs, hence
should not generate a fragment from the non-replicating replicon DNA. As expected, 15 cycles
of PCR were sufficient to produce a 1,022 base pair (bp) fragment from DNA samples
extracted from tissues receiving LM4-G, LM4-R, and LM4-C1fs-G plus LM4-R (Fig 1D, top
panel, lanes 1, 2, 4), but not LM4-C1fs-G alone (lane 3). To ensure similar amounts of template
DNA were present in all samples, an 18-cycle PCR was simultaneously performed to amplify a
436-bp ribosomal DNA (rDNA) fragment (Fig 1D, middle panel).
The one-nt insertion in LM4-C1fs-G also disrupted an MfeI site (CAATTG). We thus used
MfeI digestion of the PCR products to verify the replicational complementation of LM4-C1fs-
G by LM4-R. As shown in Fig 1D, bottom panel, the LM4-G and LM4-R-derived PCR prod-
ucts, upon MfeI digestion, were split into two smaller fragments of 714 bp and 308 bp, respec-
tively (lanes 1 and 2). By contrast, the majority of PCR product derived from the LM4-C1fs-G
plus LM4-R sample remained undigested (lane 4), thus verifying successful replication of
LM4-C1fs-G in the presence of LM4-R.
We next verified the replication-dependent expression of GFP and mCherry proteins with
Western blotting, using antibodies that specifically reacted to either one, but not both. As
shown in Fig 1E, GFP protein was detected in samples with LM4-G, LM4-C1fs-G plus LM4-R
(top panel, lanes 1 and 4), but not in those with LM4-R or LM4-C1fs-G alone (lanes 2 and 3).
Similarly, mCherry protein was detectable in samples containing LM4-R and LM4-C1fs-G
plus LM4-R (Fig 1E, middle panel, lanes 2 and 4). Together these experiments established
LM4-G and LM4-R as robust TYLCV replicons suitable for examining the intracellular bottle-
necking of viral populations.
Agro-infiltration enables five or more agro cells to deliver transfer DNA
(T-DNA) into a single N.benthamiana cell
To adopt LM4-G and LM4-R for investigating the intracellular population dynamics of
TYLCV, we first must determine whether these two replicons, when delivered via agro-infiltra-
tion, could co-enter a sufficiently high percentage of N.benthamiana cells. Note that successful
agro-infiltration-mediated delivery of two or more constructs into the same plant cells has
been widely reported, most notably for bi-fluorescence complementation experiments (BIFC;
e.g. ref 25); but also for studying plant viruses with multi-partite genomes [26,27]. However,
these earlier studies did not carefully evaluate the frequency of cells internalizing all co-deliv-
ered constructs, and conditions for maximizing same-cell internalization.
We began our assessment using 35S-GFP and 35S-mCherry, two non-replicating constructs
that drove transient expression of GFP and mCherry with the strong 35S promoter of
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 5 / 21

cauliflower mosaic virus (CaMV). A third construct expressing the p19 silencing suppressor of
tomato bushy stunt virus (TBSV) was also included to counteract the silencing of GFP and
mCherry transcripts. Suspensions of agro cells harboring 35S-GFP, 35S-mCherry, and p19
were first diluted to OD
600
= 0.1, and the diluted agro suspensions were mixed at a volume
ratio of 1:1:1. As a result, the final concentration of each agro line was OD 0.033 (Fig 2A). As
shown in Fig 2A, both GFP and mCherry were detected in close to 100% of the cells, with
more than 95% cells expressing both. This high level of co-expression suggested that each of
the N.benthamiana cells must have received 35S-GFP and/or 35S-mCherry constructs translo-
cated from 5 or more independent agro cells. (We recognize that only the T-DNA portion of a
binary construct was mobilized into plant cells by agros; thus, the somewhat inaccurate use of
the term “construct” here is purely for simplicity)
Fig 2. A.tumefaciens (agro) harboring two different constructs deliver both constructs into same N.benthamiana
cells at very high efficiencies even at the low OD
600
values of 0.033–0.05. A. Both 35-GFP and 35-mCherry were
non-replicating constructs that expressed GFP and mCherry transiently upon entering cells. Band C. Agros harboring
LM4-R and LM4-G replicons were mixed with those containing 35S-GFP and 35S-mCherry, respectively. Prior to
mixing, all agro suspensions were diluted to OD 0.1; and all mixtures contained an equal volume of all constituents.
https://doi.org/10.1371/journal.ppat.1011365.g002
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 6 / 21

To explain the rationale for the above conclusion, we first imagine just one agro cell suc-
ceeded in engaging every N.benthamiana cell. Given the inoculum contained an equal amount
of 35S-GFP and 35S-mCherry-carrying agro cells, 50% of plant cells would have received
35S-GFP, the remaining 50% 35S-mCherry. Put it differently, any given single plant cell should
have a 50% chance to receive either 35S-GFP or 35S-mCherry, but never both. Using the sym-
bols “g” and “r” to represent 35S-GFP and 35S-mCherry, respectively, these chances could be
expressed with the following equation (Table 1, row 1):
50%gþ50%r¼0:5gþ0:5r
We next consider the scenario of two agro cells engaging each plant cell. Given that cells of
N.benthamiana vastly outsizes that of agro cells (roughly 150 μm versus 3 μm in lengths, thus
50: 1 length-wise, >50,000: 1 volume-wise), engagement of a plant cell by two agro cells most
likely constituted two independent events. Thus, the chances of a cell of receiving two copies
of 35S-GFP (g
2
), two copies of 35S-mCherry (r
2
), or one of each (gr) could be expressed as
(Table 1, row 2):
ð0:5gþ0:5rÞX50%gþ ð0:5gþ0:5rÞX50%r¼ ð0:5gþ0:5rÞXð0:5gþ0:5rÞ
¼0:25g2þ0:5gr þ0:25r2
Therefore, successful engagement of a single plant cell by two separate agro cells would per-
mit just 50% of plant cells to express both GFP and mCherry, whereas those expressing GFP or
mCherry alone would each amount to 25%. Repeating this thought experiment for situations
of 3, 4, and 5 agro cells engaging each plant cell, we arrived at three additional equations, listed
as rows 3–5 of Table 1. According to these equations, 5 different agro cells penetrating each
plant cell would be expected to permit 93.75% of plant cells to express both GFP and mCherry.
Comparing this predicted value with the observed value of >95%, we concluded that at the
concentration of OD 0.033 per agro line (OD 0.066 combined), at least 5 agro cells succeeded
in translocating their binary constructs into an average N.benthamiana cell. Note that the
actual number was likely even higher because these calculations left out agro cells carrying the
p19 construct. Additionally, Oltmanns and colleagues [28] found that 4–20 copies of the same
binary construct routinely resided in each agro cell. Even assuming just 3 of these 4–20 copies
were mobilized into plant cells by every agro cell, a minimum of 5 agro cells engaging each
plant cell would cause an average plant cell to receive 3 X 5 = 15 copies of binary constructs.
The LM4-G and LM4-R-carrying agro cells were expected to behave similarly to those har-
boring 35S-GFP and 35S-mCherry, because these binary constructs had the same backbone
plasmid (pAI101), thus the same essential elements required for T-DNA translocation (e.g. the
Table 1. Estimating the number of agro cells translocating binary constructs to each plant cell.
Agro cell # (g = 35S-GFP; r = 35S-mCherry) Plant cells fluorescing:
green both red
1 50%g + 50%r = 0.5g + 0.5r 50% 0 50%
2 (0.5g + 0.5r) X (0.5g + 0.5r) = (0.5g + 0.5r)
2
= 0.25g
2
+ 0.5gr + 0.25r
2
25% 50% 25%
3 (0.5g + 0.5r)
3
= 0.125g
3
+ 0.375(g
2
r + gr
2
) + 0.125r
3
12.5% 75% 12.5%
4 (0.5g + 0.5r)
4
= 0.0625g
4
+ 0.25(g
3
r + gr
3
) + 0.375g
2
r
2
+ 0.0625r
4
6.25% 87.5% 6.25%
5 (0.5g + 0.5r)
5
= 0.03125g
5
+ 0.15625(g
4
r + gr
4
) + 0.3125 (g
3
r
2
+ g
2
r
3
) + 0.03125r
5
3.125% 93.75% 3.125%
https://doi.org/10.1371/journal.ppat.1011365.t001
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 7 / 21

right and left border motifs); and were transformed into the same agro strain (C58C1). Never-
theless, we next tested the mixed agro-infiltrations for the 35S-GFP & LM4-R pair (Fig 2B), as
well as the LM4-G & 35S-mCherry pair (Fig 2C). Note that even though the mCherry encoded
by LM4-R was fused with the V1 NLS, cytoplasmic mCherry fluorescence was detected (e.g.
Fig 2B, middle), probably due to its high expression levels in the context of viral replication.
Also note that the p19-expressing construct was no longer necessary because replicating
LM4-G and LM4-R provided the TYLCV-encoded V2 silencing suppressor. Fig 2B and 2C
illustrated that these two pairs of agros likewise resulted in co-expression of GFP and mCherry
in more than 95% of N.benthamiana cells, suggesting that at the concentration of OD 0.05
each (OD 0.1 combined), the replicon-carrying agro cells were competent at delivering con-
structs into plant cells at a minimum ratio of 5 agro cells per plant cell.
We wish to warn against confusing the above conclusions with the bottleneck size estima-
tions to be described below. Although similar probabilistic laws applied to both sets of experi-
ments, the experiments so far served to ensure two or more co-delivered constructs entered the
same agro-infiltrated cells at sufficiently high proportions. In these experiments the LM4-G and
LM4-R replicons were paired with the non-replicating 35S-mCherry and 35S-GFP, respectively,
but not with each other. As a result, potential intracellular bottlenecks exerted by the replicons
would exclusively limit the number of replication-initiating LM4-G or LM4-R copies, without
affecting expressions launched by 35S-GFP or 35S-mCherry. By contrast, subsequent experi-
ments aimed to resolve whether the successfully internalized LM4-G and LM4-R replicon cop-
ies were then bottlenecked inside the cells, blocking most of them from active replication. It is
obvious that this latter question could not be convincingly addressed unless efficient same-cell
internalization by multiple replicon copies was established first.
LM4-G and LM4-R co-replicate in less than 65% of cells that receive both
constructs
Results of the previous section predicted that a mix of LM4-G and LM4-R-containing agros,
each at the concentration of OD 0.05, would likely enable at least 5 individual agro cells to
mobilize their binary constructs into each N.benthamiana cell, furnishing the latter with 15 or
more replicon construct copies. Moreover, increasing the agro titers by 10 folds to OD 0.5
would likely further increase the number of replicon genome copies entering each N.
benthamiana cell. Establishment of same-cell co-existence of multiple copies of two replicon
constructs equipped us with the ideal system to address whether replicon genome copies in the
same plant cell would be subject to intracellular reproductive bottlenecks that prevent all but a
few of them from initiating replication.
Assuming different replicon copies initiate replication independently of each other, we pre-
dicted that if the number of replication-initiating copies was 4 or fewer in each cell, the chance
of all of them being either LM4-G or LM4-R alone would be sufficiently high (>6% each) to
permit detection, as these cells would express only GFP or mCherry, but not both. We based
this prediction on the same probability equations invoked earlier (Table 1), though this time
focusing exclusively on replication-initiating (thus bottleneck-escaping) replicon copies. To
test this prediction, we prepared two agro mixes that contained an equal amount of LM4-G
and LM4-R-containing agro cells, with the concentration of each agro line adjusted to OD
0.05 in the first mix, and OD 0.5 in the second. The infiltrations were carried out on at least six
N.benthamiana leaves, with each leaf divided into two halves: one half receiving the OD 0.05
mix, the other the OD 0.5 mix. Four days after infiltration, 10 2.4-mm
2
(1.55 X 1.55) viewing
fields were randomly chosen for quantification of cells that expressed GFP, mCherry, or both.
The entire procedure was repeated three times to ensure reproducibility. Fig 3A shows four
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 8 / 21

Fig 3. Co-delivery of LM4-G and LM4-R replicons reveals narrow intracellular population bottlenecks of TYLCV.
A. Images of a typical 2.4 mm
2
viewing field taken from a leaf section that received both LM4-G and LM4-R, showing
nuclei that emit green (GFP) or red (mCherry) fluorescence, or both (Merged), one of each highlighted with
arrowheads of different colors. The gray scale image serves as reference for cellsizes, shapes, and boundaries. B.
Quantification of percentages of cells replicating LM4-G, LM4-R, or both (LM4-G & LM4-R). Shown are box plots
derived from numeration data of six experimental groups (three repeats, each with two different agro concentrations,
10 data points per group).
https://doi.org/10.1371/journal.ppat.1011365.g003
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 9 / 21

confocal images of one viewing field, collected under different light channels. Three of the
nuclei were highlighted with green, red, and yellow arrow heads (Fig 3A), representing cells
that replicated LM4-G, LM4-R, and both, respectively. Comparing images of different chan-
nels (GFP, mCherry, Merge) permitted straightforward differentiation of these cells. Six sets of
raw counts of green-, red-, and yellow-fluorescent cells, collected from 10 different viewing
fields per set, are provided in the S1 Dataset. For each of the six repeats, the combined cell
numbers were then calculated for each of the three categories (green, red, or both), and com-
piled in Table 2. Their respective percentages were also calculated (Table 2), and used to gener-
ate the box plots shown in Fig 3B. As shown in Table 2 and Fig 3B, across the six repeats (three
at OD 0.05 and three at OD 0.5), the percentages for cells replicating solely LM4-G ranged
from 15.70% to 20.09%, whereas those replicating only LM4-R ranged from 22.56 to 30.14%.
Finally, cells replicating both replicons were between 49.77 and 60.87%. These data unveiled a
consistent trend of stringent intracellular population bottlenecking that limited the number of
replicon genome copies replicating in each cell to low single digits (more in next section).
The size of TYLCV intracellular population bottlenecks is no larger than
three
We next set out to estimate the size of intracellular bottlenecks encountered by the LM4-G and
LM4-R replicons. As noted earlier, the equations in Table 1 assumed the two co-delivered con-
structs (35S-GFP and 35S-mCherry) being similarly competent. However, the data in Table 2
and Fig 3B indicated that in LM4-G & LM4-R mixed infections, LM4-G replicated in a slightly
smaller number of cells than LM4-R. Thus, the earlier equations must be adjusted before being
adopted for the current purpose. To this end, we calculated a relative LM4-G/LM4-R ratio (g:
r) for each of the six repeats (Table 2, right column). These repeat-specific ratios were then
averaged to obtain a mean ratio of 47: 53 for cells replicating LM4-G and LM4-R, respectively.
The probability equations adjusted using this ratio are listed in Table 3.
According to Table 3, if just 3 of the replicon copies succeeded in initiating replication in
the same cell, meaning an intracellular population bottleneck size of 3, we would detect 75%
cells fluorescing yellow, 10% green, and 15% red. Higher percentages of green or red-only cells
would indicate a smaller bottleneck size. By comparison, the observed percentages depicted in
Table 2 and Fig 3B were 49.77–60.87% yellow, 15.70–20.09% green, and 22.56–30.14% red.
These value ranges, with fewer yellow cells but more green and red cells, indicated that the
actual bottleneck size must be smaller than 3. On the other hand, an intracellular population
bottleneck size of 2 would predict 50% cells being yellow, 22% green, and 28% red. These pre-
dicted values were modestly lower than observed for yellow cells, higher for green cells, but
nearly as predicted for red cells. Taken together, our data fit the prediction of a bottleneck size
Table 2. Combined counts of green-, red-, and yellow (g &r) fluorescent cells for each of the six repeats, and their respective percentages, obtained with mixed deliv-
ery of two independent constructs (LM4-G and LM4-R).
Repeat Number and fraction of fluorescent cells g: r ratio
Green (g) Red (r) g & r Total
# % # % # % # %
1
st
, OD 0.05 200 15.73 373 29.45 701 54.82 1274 100 45.7: 54.3
1
st
, OD 0.5 231 18.28 356 29.35 628 52.37 1215 100 46.5: 53.5
2
nd
, OD 0.05 336 20.09 498 30.14 834 49.77 1668 100 46.8: 53.2
2
nd
, OD 0.5 293 19.12 344 22.56 876 58.32 1513 100 49: 51
3
rd
, OD 0.05 175 19.94 231 26.5 468 53.56 874 100 47.9: 52.1
3
rd
, OD 0.5 149 15.7 220 23.43 579 60.87 948 100 47.6: 52.4
https://doi.org/10.1371/journal.ppat.1011365.t002
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 10 / 21

between 2 and 3, meaning no more than 3 replicon copies initiated replication in an average
N.benthamiana cell.
To obtain more precise estimates, we assumed that the bottleneck sizes varied among indi-
vidual cells according to the Poisson distribution, with an average size of λ. We then computed
λand its standard deviation using a maximum-likelihood method first developed by Miyashita
and colleagues [8]. Meanwhile, since LM4-G alone replicated in slightly fewer cells than
LM4-R alone, we simultaneously estimated the ratio of g relative to (g + r), designated r
g
in the
equations below, for each repeat to account for the effect of replication bias. Specifically, using
the variables λand r
g
, the expected proportion of LM4-G only cells (P
g
), LM4-R only cells (P
r
),
and co-infected cells (P
g+r
), can be expressed as
Pg¼1
1p0X1
k¼1pk�rg
k;
Pr¼1
1p0X1
k¼1pk� ð1rgÞk;and
Pgþr¼1PgPr;
respectively, where p
k
is the probability to have the bottleneck size of kfollowing the Poisson
distribution with an average size of λ:
pk¼lkel
k!:
The likelihood L(in other words, fitness to the observed data) can be calculated using the
observed numbers of LM4-G only cells (c
g
), LM4-R only cells (c
r
), and co-infected cells (c
g+r
)
together with their expected proportions P
g
,P
r
, and P
g+r
as
L¼ctotal!
cg!cr!cgþr!Pg
cgPr
crPgþr
cgþr:
Then, the most likely estimates for λand r
g
were obtained by searching for λand r
g
that
maximize the likelihood L(see Materials and Methods and SI Text for code availability infor-
mation). We allowed different r
g
values for different experimental repeats, because the relative
activities of mixed agro suspensions varied slightly. As a result of maximum-likelihood com-
putations, we obtained λvalues of 2.38 ±0.05 and 2.60 ±0.05 for OD 0.05 and 0.5, respectively
(Table 4), with the g/(g + r) ratio r
g
ranging from 0.42 to 0.48. This suggests that the size of
intracellular TYLCV population bottlenecks was no more than three, and was largely unaf-
fected by a 10-fold increase of the inoculum dose.
Table 3. Equations for estimating bottleneck sizes with g: r = 47: 53.
Bottleneck size Equation Predicted outcomes
green both red
1 (0.47g + 0.53r)
1
= 0.47g + 0.53r
47% 0 53%
2 (0.47g + 0.53r)
2
= 0.22g
2
+ 0.5gr + 0.28r
2
22% 50% 28%
3 (0.47g + 0.53r)
3
= 0.10g
3
+ 0.35g
2
r + 0.40gr
2
+ 0.15r
3
10% 75% 15%
4(0.47g + 0.53r)
4
= 0.049g
4
+ 0.22g
3
r + 0.372g
2
r
2
+ 0.28gr
3
+ 0.079r
4
4.9% 87.2% 7.9%
https://doi.org/10.1371/journal.ppat.1011365.t003
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 11 / 21

TYLCV intracellular population bottlenecks remain stringent when the
LM4-G and LM4-R replicons are combined in a single construct
Earlier experiments indicated that co-delivering LM4-G and LM4-R in the form of mixed agro
suspensions, at OD 0.05 each, permitted >95% of N.benthamiana cells to receive both repli-
cons. Nevertheless, there was still a small chance that a few N.benthamiana cells received mul-
tiple copies of a single replicon construct (meaning LM4-G or LM4-R only), thus distorting
the results of bottleneck size estimations. To address this concern, we next assembled a new
construct, designated LM4-G_LM4-R, by combining both replicons in the same DNA mole-
cule. The LM4-G_LM4-R construct was then delivered into N.benthamiana cells at two agro
concentrations (OD 0.05 and 0.5), and cell nuclei expressing GFP only, mCherry only, or both
were counted for three repeat experiments. As shown in Fig 4,Table 5, and S2 Dataset, this
new construct further compromised the relative competitiveness of LM4-G. This notion was
supported by maximum-likelihood estimations, which showed that the g/(g + r) ratio r
g
was
significantly lower than 0.5 (ranging from 0.13 ±0.01 to 0.32 ±0.01; Table 6). Nevertheless,
the estimated bottleneck sizes, 2.37 ±0.07 and 2.20 ±0.08 for OD 0.05 and 0.5, respectively,
were not substantially different from the estimates obtained with two separate constructs
(Table 4). Thus, guaranteeing the same cell co-existence of both LM4-G and LM4-R by com-
bining them in the same construct provided further corroboration of earlier estimation of an
intracellular bottleneck size of no more than 3.
LM4-G and LM4-R exert strong intracellular SIE to each other
We reported earlier that in (+) RNA virus infections SIE manifested intracellular bottlenecking
of viral populations [8–11,29,30]. To determine whether the TYLCV replicons also exhibited
intracellular SIE, we next used LM4-G and LM4-R to infect N.benthamiana leaf cells in a
sequential manner. As controls, N.benthamiana leaf cells pre-infiltrated with either infiltration
buffer or the 35S-GFP transient expression construct still permitted the replication of superin-
fecting LM4-R (Fig 5A and 5B). By contrast, a pre-introduced LM4-G exerted a near complete
SIE against LM4-R, blocking the latter from replicating in nearly all cells (Fig 5C). Indeed,
among more than 1,000 cells inspected in multiple repeat experiments, just one cell was found
to replicate the superinfecting LM4-R (Fig 5C, yellow arrowhead). Similarly, a pre-introduced
LM4-R completely blocked the replication of LM4-G in the same cells (Fig 5D and 5E).
The mutual SIE between LM4-G and LM4-R could have been caused by exhaustion of cel-
lular resources by the pre-introduced replicon, given the 24-hour time lag between two agro-
infiltrations. To address this concern, we tested whether turnip crinkle virus (TCV), a (+)
RNA virus, could exert SIE to TYLCV replicons, as TCV replication would be expected to con-
sume large amounts of cellular resources crucial for TYLCV replication as well, such as those
needed for transcribing viral mRNAs and translating viral proteins. To this end, we adopted
TCVdMP_sg2R, a TCV replicon that expressed mCherry in a replication-dependent manner
and, due to a deletion within its movement protein (MP) ORFs, did so exclusively in cells
receiving this construct [10]. As shown in Fig 5F and 5G, TCVdMP_sg2R replicated in fewer
cells than LM4-G both when pre-introduced and when superinfecting. Nevertheless, in both
Table 4. Maximum-likelihood estimates of λand g/(g + r) with LM4-G and LM4-R as separate constructs.
λg/(g + r)
Repeat 1 Repeat 2 Repeat 3
Estimates (OD = 0.05) 2.38 ±0.05 0.42 ±0.01 0.44 ±0.01 0.46 ±0.01
Estimates (OD = 0.5) 2.60 ±0.05 0.43 ±0.01 0.48 ±0.01 0.45 ±0.01
https://doi.org/10.1371/journal.ppat.1011365.t004
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 12 / 21

Fig 4. Guaranteeing same-cell internalization of both replicons using a single construct to host both LM4-G and
LM4-R. A. Images of a typical 2.4 mm
2
viewing field taken from a leaf section that received the single construct
LM4-G_LM4-R, showing nuclei that emit green (GFP) or red (mCherry) fluorescence, or both (Merged), one of each
highlighted with arrowheads of different colors. The gray scale image serves as reference for cell sizes, shapes, and
boundaries. The intracellular population bottlenecks were again manifest, as indicatedby fractions of cell nuclei
replicating LM4-G, LM4-R, or both (LM4-G & LM4-R). B. Box plots derived from numeration data of six
experimental groups (three repeats, each with two different agro concentrations, 10 data points per group).
https://doi.org/10.1371/journal.ppat.1011365.g004
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 13 / 21

cases nearly all cells that replicated TCVdMP_sg2R also replicated LM4-G. Therefore, mutual
SIE between LM4-G and LM4-R was most likely due to self-exclusion.
We further investigated whether the pre-introduced replicon must replicate itself in order
to exert SIE against the superinfecting replicon. To this end, we pre-introduced the defective
replicon LM4-C1fs-G, and superinfected the same cells with LM4-R. As shown in Fig 5H and
5I, the defective LM4-C1fs-G was actually complemented by the superinfecting LM4-R,
enabling the former to replicate in most cells to produce green-fluorescent nuclei. Intriguingly,
some of the cells expressed GFP but not mCherry, suggesting that in these cells, even though
the LM4-R replicon construct provided the C1 protein to complement the replication of
LM4-C1fs-G, its own replication was blocked by the intracellular reproductive population bot-
tlenecks. Note that in these cells the LM4-R replicon could not be said to be blocked by SIE,
because LM4-C1fs-G, despite being pre-introduced, could not replicate until after LM4-R
introduction, thus had no chance to exert SIE against the latter.
Consistent with this idea, no complementation occurred when the defective replicon was
introduced as a superinfector (Fig 5J). This result, when considered together with the results in
Fig 5C and 5E, indicated that by this time (24 hours after introduction of the primary repli-
con), the intracellular bottlenecks were so firmly established that newly internalized replicons
of the same virus no longer had the chance to initiate their own replication. Overall these
results suggested that replication of the primary replicon, and/or pre-accumulation of the C1
protein, were needed for robust SIE between TYLCV replicons.
Discussion
In the current study, we sought to resolve whether multiple copies of the single-stranded, cir-
cular genome of a small DNA virus, upon entering the same cell collectively, could all initiate
replication or, whether they were subject to intracellular reproductive bottlenecking in a man-
ner similar to (+) RNA viruses [9,10]. To this end, two TYLCV replicons, LM4-G and LM4-R,
were created that expressed two different fluorescent proteins (GFP and mCherry) in a replica-
tion-dependent manner. Note that these replicons lacked the ability to move cell-to-cell due to
the disruption of V1 (CP) ORF. These two replicons were then introduced into leaves of N.
Table 5. Combined counts of green-, red-, and yellow (g &r) fluorescent cells for each of the six repeats, and their respective percentages, obtained with a single con-
struct combining LM4-G and LM4-R.
Repeat Number and fraction of fluorescent cells
Green (g) Red (r) g & r Total
# % # % # % # %
1
st
, OD 0.05 50 8.08 293 47.9 278 44.02 621 100
1
st
, OD 0.5 38 4.97 557 71.65 185 23.38 780 100
2
nd
, OD 0.05 40 5.81 418 56.32 291 37.86 749 100
2
nd
, OD 0.5 68 7.86 490 57.78 293 34.36 851 100
3
rd
, OD 0.05 130 14.2 415 41.87 428 43.93 973 100
3
rd
, OD 0.5 72 8.43 430 51.73 340 39.84 842 100
https://doi.org/10.1371/journal.ppat.1011365.t005
Table 6. Maximum-likelihood estimates of λand g/(g + r) with LM4-G and LM4-R combined in a single construct.
λg/(g + r)
Repeat 1 Repeat 2 Repeat 3
Estimates (OD = 0.05) 2.37 ±0.07 0.27 ±0.01 0.21 ±0.01 0.32 ±0.01
Estimates (OD = 0.5) 2.20 ±0.08 0.13 ±0.01 0.22 ±0.01 0.26 ±0.01
https://doi.org/10.1371/journal.ppat.1011365.t006
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 14 / 21

benthamiana plants in the form of mixed agro suspensions to ensure both entered the same
cells. Additionally, they were also combined in a single plasmid construct (LM4-G_LM4-R) to
guarantee their simultaneous penetration of the same cells. Importantly, GFP and mCherry
expressed from LM4-G and LM4-R replicons translocated to the cell nuclei, permitting
straightforward numeration of cells that replicated one or both replicons. Counting of tens of
thousands of fluorescent cell nuclei, followed by probability computation, led to the conclusion
that TYLCV populations were severely bottlenecked intracellularly, permitting no more than
three genome copies to commence replication in each cell. Finally, if a molecular mechanism
similar to BIAS was responsible for actively enforcing these intracellular bottlenecks, it would
likely also explain the cellular level SIE observed between the two TYLCV replicons (Fig 5).
The question of exactly how many genome copies of a DNA virus initiate productive infec-
tions in each infected cell have been investigated for several larger DNA viruses with double-
stranded DNA genomes [31–33]. In one study, the Autographa californica nuclear polyhedro-
sis virus (AcMNPV) with a 134 kb dsDNA genome was estimated to found cellular infections
with approximately 4.3 genome copies per cell [31]. In another, the plant-infecting CaMV
with an 8 kb dsDNA genome (though replicating via an RNA intermediate) were estimated to
initiate effective cellular infections with 2–13 copies of viral genome, depending on leaf posi-
tions and infection stages [32]. It should be noted that the per-cell genome copy numbers
established in these earlier studies, designated multiplicity of infection or MOI by the authors,
were computed indirectly from results of serial passaging of multiple host individuals [31], or
from serial sampling of the systemically infected plant leaves [32]. The authors thus had to
make assumptions about the number of cells passed through by the viral genomes. The third
study adopted a greatly improved system in which the replication of a recombinant genome of
the pseudorabies virus (PRV263) produced three different types of progeny viruses that each
expressed a different fluorescent protein [33]. By following the fates of different progeny types
in a verifiable number of successive cells, the authors estimated that PRV263 initiated cellular
infections with fewer than 7 copies of the genome. However, this study was carried out in cell
cultures, which could differ from infected host individuals.
Fig 5. LM4-G and LM4-R exhibit mutual exclusion at the cellular level when delivered sequentially. The primary and superinfecting agros
were administered to the same leaf areas with a 24 hour (hr) interval.
https://doi.org/10.1371/journal.ppat.1011365.g005
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 15 / 21

Another more recent study appeared to suggest that faba bean necrotic stunt nanovirus
(FBNSV), which divides its genome in 8 different small circular ssDNA segments, may enlist
intercellularly trafficked protein products to complete replication cycles of different segments
in different cells, hence requiring <1 genome copy per cell on average [27]. However, given
that different genome segments of multipartite or segmented viruses could replicate to vastly
different levels in different cells [34,35], and the possibility that the in situ hybridization proce-
dure used by the authors may not be sufficiently sensitive to detect low level replication, these
findings await further corroborations.
Compared with these earlier studies, our experiments were carried out in cells of intact host
plants, where the population dynamics of viral replicons were followed at the single cell level,
yielding direct conclusions drawn from primarily infected cells of host plants. Moreover, since
the observable output were fluorescent signals emitted by GFP and mCherry, which should
have been amplified by replication-dependent transcription and translation, the sensitivity of
our approach should far surpass other methods that directly monitored genomic DNA copies,
thus generating highly reproducible observations.
Exactly how many genome copies of a virus initiate replication in a single cell is a critical
question because it is becoming increasingly clear that many viruses invade cells with not one,
but many virions, hence many copies of viral genomes [36]. Consistent with this view, copies
of the (+) sense poliovirus genomic RNA were found to direct translation of viral proteins at
multiple intracellular sites before commencing replication [37]. On the other hand, even if just
one copy of viral genome entered a cell initially and it succeeded in launching replication, we
are still faced with the question of how many of the progeny genome copies repeat the replica-
tion cycle in the cell of their parent.
The need to address this question is even more acute considering the high error rate of the
replication processes of many viruses, estimated to be approximately 10
−4
for every nt incorpo-
rated [7,38,39]. Depending on the size of viral genomes, this error rate translates into approxi-
mately one mutation for every new genome copy synthesized. Given the fact that viruses often
replicate millions of progeny genome copies in each infected cell, it is inevitable that some of
the progeny genome copies contain loss-of-function mutations in essential protein-coding
genes. If even 100 of the progeny genomes were permitted to re-replicate in the cell of their
own genesis, those with lethal errors would be able to hitchhike on the allelic proteins pro-
duced by sister genomes, thus could not be efficiently purged from the virus population.
While it is critically important to weed out lethal mutations in virus populations, similarly
important is the need to positively select beneficial mutations. Absent of bottleneck-mediated
isolation, a viral genome copy with a beneficial mutation in a protein-coding gene must share
the intracellular environment with numerous sister genome copies, thus is unlikely to be the
exclusive beneficiary of the mutation-endowed beneficial phenotype. Such a genome copy
would then replicate to similar levels as its sister copies, despite the phenotypic advantage it
encodes. This scenario is contrary to the real world observations of rapid enrichment of benefi-
cial mutations by viruses, most notably SARS-CoV-2 [40], suggesting intracellular population
bottlenecking as a virus-encoded trait to assure both positive and purifying selections [41–44].
Besides serving as a DNA virus model for the BIAS hypothesis, the TYLCV system has addi-
tional advantages. To address the challenge of constant emergence of viral mutations, earlier
researchers proposed that some of the replication proteins encoded by (+) RNA viruses serve
exclusively the very RNA from which they are translated, hence are cis-acting [26,45]. The cis-
acting arrangement could explain certain aspects of replicating (+) RNA viruses for which
both translation and replication occur in the cytoplasm. However, it would be untenable for
DNA viruses like TYLCV. This is because the TYLCV-encoded replication protein (C1) must
be translated in the cytoplasm of the infected cells, and then routed back to cell nuclei to
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 16 / 21

orchestrate the rolling circle replication. It is then impossible to differentiate between C1 pro-
teins originated from different TYLCV genome copies.
We hasten to note that this TYLCV-based experimental system is not without limitations.
For example, replacing part of the V1 coding sequence with GFP or mCherry interrupted the
natural infection process by preventing the intra- and intercellular spread of the viral genomes.
Furthermore, as evidenced by the unequal replication of LM4-G and LM4-R, insertion of
DNA fragments with different sequences in TYLCV genome likely affected viral replication
efficiency to varying extents. Finally, despite our efforts ensuring the same-cell penetration by
multiple replicons, it remains to be established exactly how many copies of TYLCV genome
are simultaneously translocated into an uninfected cell during natural infections. In short,
many gaps remain to be filled with additional investigations.
In summary, with the current study we have demonstrated that populations of the small
DNA virus TYLCV are intracellularly bottlenecked in a manner similar to (+) RNA viruses.
This finding provides exciting leads for follow-up studies aimed at elucidating the mechanism
(s) of the bottlenecking, and examining its potential role in facilitating natural selection in
viruses. Outcomes of these follow-up studies will likely avail the bottlenecking machinery as
the target for preventive as well as therapeutic interventions of virus diseases of plants, animals,
and humans.
Materials and methods
Constructs
The original TYLCV infectious clone (isolate SH2. The Genbank accession number is
AM282874.1) was kindly provided by Dr. Xueping Zhou of China Institute of Plant Protection
[46]. The full-length, double-stranded form of TYLCV genome, plus a 216-bp duplication at
the 5’ end, and a 158-bp duplication at the 3’ end, was subcloned into pAI101, a E.coli-A.tume-
faciens shuttle vector modified from pCambia1300 in our lab [24,47], leading to a new TYLCV
infectious clone we call LM4. To create LM4-G and LM4-R, two KpnI sites were introduced
into the V1 gene, at positions 307/308 and 881/882 (numbering relative to the full length
genome sequence), respectively. The coding sequences of uvGFP and mCherry were then PCR
amplified and cloned between the KpnI site using the NEBuilder kit (New England Biolabs).
The sequences of LM4, LM4-G and LM4-R encompassing the entire TYLCV DNA (and its
modified forms) were verified with Sanger sequencing. Other constructs (p19, 35-GFP, 35S-
mCherry) were described previously [30,48,49].
Agrobacterium infiltration
(agro-infiltration). All DNA constructs destined for testing in N.benthamiana plants were
transformed into electrocompetent A. tumefaciens strain C58C1 via electroporation using the
AGR setting on the Bio-Rad Micropulser Electroporator. Briefly, 5 μl of the plasmid DNA was
mixed with 40 μl of agro cells and maintained on ice until electroporation. After electropora-
tion, 900 μl of SOB media was added and the suspension was incubated at 28˚C for one hour.
Selection was carried out on solid Terrific Broth (TB) media containing rifampicin, gentamy-
cin, and kanamycin. Successful introduction of the plasmid was confirmed using colony PCR.
A single colony confirmed to have the desired plasmid was used to inoculate 3 ml TB liquid
media with the same antibiotics, and incubated overnight at 28˚C. The culture was diluted
1:100 with fresh TB liquid media and incubated under the same conditions for another night.
The second culture was centrifuged at 4,000 rpm for 20 min, and resuspended in agroinfiltra-
tion buffer (10 mM MgCl2, 10 mM MES, and 100 μM acetosyringone). All suspensions were
diluted to OD600 = 1 and incubated at 28˚C for 3 hours. Agrobacterium suspensions were
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 17 / 21

then mixed and introduced into leaves of young N.bethamiana plants via a small wound,
using a needleless syringe.
Confocal microscopy
Four days after agro-infiltration, leaf discs were collected from the plants. Confocal micros-
copy was performed at the Molecular and Cellular Imaging Center (MCIC), the Ohio Agricul-
tural Research and Development Center, using a Leica DMI6000 laser confocal scanning
microscope. To detect GFP and mCherry fluorescence, sequential excitation at 488 nm and
587 nm was provided by argon and helium-neon 543 lasers, respectively.
Numeration of the fluorescent nuclei
To count the cells that replicate LM4-G, LM4-R, or both, images of 2.4 mm
2
(1.55 mm X 1.55
mm) were collected using a 10X lens, from randomly selected leaf sections receiving different
combinations of agro suspensions. For samples treated with the mixture of LM4-G and
LM4-R, or the combined LM4G_LM4-R construct, images of three separate channels were col-
lected for every selected viewing field to allow for the separate counting of different colored
spots (GFP, mCherry, or all spots). For each treatment group, at least 10 images were collected
from six different leaves. The fluorescent spots representing nuclei of infected cells were
counted using the ImageJ program. The number of nuclei that simultaneously replicated both
LM4-G and LM4-R was calculated by subtracting the number of all spots from the sum of
green and red spots.
Statistics and bottleneck size computation
The calculation of sums, means, percentages, standard deviations, were mostly carried out
with various tools available through Excel. The box plots were generated with the JMP Pro
16.0.0 software package. Bottleneck size estimations were performed using a maximum likeli-
hood algorithm described in our previous studies [8,9,41], based on the R software package
ver. 4.1.3 [42]. The script is available as S1 Text in SI Appendix, and downloadable at GitHub
(https://github.com/ShuheiMiyashita/Ren_et_al_2022).
Supporting information
S1 Dataset. Raw counts of green (G), red (R), and G&R fluorescent cells for Table 2 and
Fig 3B.
(XLSX)
S2 Dataset. Raw counts of green (G), red (R), and G&R fluorescent cells for Table 5 and
Fig 4B.
(XLSX)
S1 Text. an R script for λand g/(g + r) computation.
(DOCX)
Acknowledgments
We thank Dr. Xueping Zhou of China Institute for Plant Protection for generously sharing the
SH2 infectious clone. We are indebted to Dr. K. Andrew White for critically reading the man-
uscript, and making insightful suggestions. Members of the Qu lab are greatly appreciated for
discussions and technical assistances.
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 18 / 21

Author Contributions
Conceptualization: Feng Qu.
Data curation: Ruifan Ren, Limin Zheng, Shuhei Miyashita.
Formal analysis: Shuhei Miyashita, Feng Qu.
Funding acquisition: Limin Zheng, Feng Qu.
Investigation: Ruifan Ren, Limin Zheng, Junping Han, Camila Perdoncini Carvalho.
Methodology: Junping Han.
Project administration: Feng Qu.
Resources: Junping Han, Camila Perdoncini Carvalho.
Software: Shuhei Miyashita.
Supervision: Deyong Zhang, Feng Qu.
Visualization: Feng Qu.
Writing – original draft: Feng Qu.
Writing – review & editing: Feng Qu.
References
1. Eigen M. Error catastrophe and antiviral strategy. Proc Natl Acad Sci USA. 2002; 99:13374–6. https://
doi.org/10.1073/pnas.212514799 PMID: 12370416
2. Bull JJ, Sanjua
´n R, Wilke CO. Theory of Lethal Mutagenesis for Viruses. J Virol. 2007; 81:2930–9.
https://doi.org/10.1128/JVI.01624-06 PMID: 17202214
3. Sanjua
´n R, Agudelo-Romero P, Elena SF. Upper-limit mutation rate estimation for a plant RNA virus.
Biol Lett. 2009; 5:394–6. https://doi.org/10.1098/rsbl.2008.0762 PMID: 19324646
4. Malpica JM, Fraile A, Moreno I, Obies CI, Drake JW, Garcı
´a-Arenal F. The Rate and Character of Spon-
taneous Mutation in an RNA Virus. Genetics. 2002; 162:1505. https://doi.org/10.1093/genetics/162.4.
1505 PMID: 12524327
5. French R, Stenger DC. Evolution of wheat streak mosaic virus: Dynamics of population growth within
plants may explain limited variation. Annu Rev Phytopathol. 2003; 41:199–214. https://doi.org/10.1146/
annurev.phyto.41.052002.095559 PMID: 12730393
6. Domingo E, Perales C. Viral quasispecies. PLoS Genet. 2019 Oct 17; 15(10):e1008271. https://doi.org/
10.1371/journal.pgen.1008271 PMID: 31622336
7. Duffy S. Why are RNA virus mutation rates so damn high? PLoS Biol. 2018; 16:e3000003–e3000003.
https://doi.org/10.1371/journal.pbio.3000003 PMID: 30102691
8. Miyashita S, Kishino H. Estimation of the size of genetic bottlenecks in cell-to-cell movement of soil-
borne wheat mosaic virus and the possible role of the bottlenecks in speeding up selection of variations
in trans-acting genes or elements. Journal of virology. 2010 Feb; 84(4):1828–37. https://doi.org/10.
1128/JVI.01890-09 PMID: 19955302
9. Miyashita S, Ishibashi K, Kishino H, Ishikawa M. Viruses Roll the Dice: The Stochastic Behavior of Viral
Genome Molecules Accelerates Viral Adaptation at the Cell and Tissue Levels. PLOS Biol. 2015; 13:
e1002094. https://doi.org/10.1371/journal.pbio.1002094 PMID: 25781391
10. Zhang XF, Sun R, Guo Q, Zhang S, Meulia T, Halfmann R, et al. A self-perpetuating repressive state of
a viral replication protein blocks superinfection by the same virus. PLOS Pathog. 2017; 13:e1006253.
https://doi.org/10.1371/journal.ppat.1006253 PMID: 28267773
11. Qu F, Zheng L, Zhang S, Sun R, Slot J, Miyashita S. Bottleneck, Isolate, Amplify, Select (BIAS) as a
mechanistic framework for intracellular population dynamics of positive-sense RNA viruses. Virus Evol
[Internet]. 2020;6:veaa086. Available from: https://doi.org/10.1093/ve/veaa086 PMID: 33343926
12. Prasad A, Sharma N, Hari-Gowthem G, Muthamilarasan M, Prasad M. Tomato Yellow Leaf Curl Virus:
Impact, Challenges, and Management. Trends Plant Sci. 2020; 25:897–911. https://doi.org/10.1016/j.
tplants.2020.03.015 PMID: 32371058
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 19 / 21

13. Hanley-Bowdoin L, Bejarano ER, Robertson D, Mansoor S. Geminiviruses: masters at redirecting and
reprogramming plant processes. Nat Rev Microbiol. 2013; 11:777–88. https://doi.org/10.1038/
nrmicro3117 PMID: 24100361
14. Zhou X. Advances in Understanding Begomovirus Satellites. Annu Rev Phytopathol. 2013; 51:357–81.
https://doi.org/10.1146/annurev-phyto-082712-102234 PMID: 23915133
15. Fondong VN. Geminivirus protein structure and function: Geminivirus proteins. Mol Plant Pathol. 2013;
14:635–49.
16. Orozco BM, Hanley-Bowdoin L. Conserved Sequence and Structural Motifs Contribute to the DNA
Binding and Cleavage Activities of a Geminivirus Replication Protein. J Biol Chem. 1998; 273:24448–
56. https://doi.org/10.1074/jbc.273.38.24448 PMID: 9733736
17. Orozco BM, Miller AB, Settlage SB, Hanley-Bowdoin L. Functional Domains of a Geminivirus Replica-
tion Protein. J Biol Chem. 1997; 272:9840–6. https://doi.org/10.1074/jbc.272.15.9840 PMID: 9092519
18. Orozco BM, Kong LJ, Batts LA, Elledge S, Hanley-Bowdoin L. The Multifunctional Character of a Gemi-
nivirus Replication Protein Is Reflected by Its Complex Oligomerization Properties. J Biol Chem. 2000;
275:6114–22. https://doi.org/10.1074/jbc.275.9.6114 PMID: 10692401
19. Hanley-Bowdoin L, Elmer J S, Rogers S G. Expression of functional replication protein from tomato
golden mosaic virus in transgenic tobacco plants. Proc Natl Acad Sci USA. 1990; 87:1446–50. https://
doi.org/10.1073/pnas.87.4.1446 PMID: 11607065
20. Zhao W, Wu S, Barton E, Fan Y, Ji Y, Wang X, et al. Tomato Yellow Leaf Curl Virus V2 Protein Plays a
Critical Role in the Nuclear Export of V1 Protein and Viral Systemic Infection. Front Microbiol. 2020;
11:1243. https://doi.org/10.3389/fmicb.2020.01243 PMID: 32587585
21. Sunter G, Bisaro DM. Transactivation of geminivirus AR1 and BR1 gene expression by the viral AL2
gene product occurs at the level of transcription. Plant Cell. 1992; 4(10):1321. https://doi.org/10.1105/
tpc.4.10.1321 PMID: 1446172
22. Sunter G, Bisaro DM. Transactivation in a geminivirus: AL2 gene product is needed for coat protein
expression. Virology. 1991; 180:416–9. https://doi.org/10.1016/0042-6822(91)90049-h PMID: 1984661
23. Sunter G, Bisaro DM. Regulation of a Geminivirus Coat Protein Promoter by AL2 Protein (TrAP): Evi-
dence for Activation and Derepression Mechanisms. Virology. 1997; 232:269–80. https://doi.org/10.
1006/viro.1997.8549 PMID: 9191840
24. Lin J, Guo J, Finer J, Dorrance AE, Redinbaugh MG, Qu F. The Bean Pod Mottle Virus RNA2-Encoded
58-Kilodalton Protein P58 Is Required in cis for RNA2 Accumulation. J Virol. 2014; 88:3213–22.
25. Zhang X, Zhang X, Singh J, Li D, Qu F. Temperature-Dependent Survival of Turnip Crinkle Virus-
Infected Arabidopsis Plants Relies on an RNA Silencing-Based Defense That Requires DCL2, AGO2,
and HEN1. J Virol. 2012; 86:6847–54. https://doi.org/10.1128/JVI.00497-12 PMID: 22496240
26. Yi G, Gopinath K, Kao CC. Selective Repression of Translation by the Brome Mosaic Virus 1a RNA
Replication Protein. J Virol. 2007; 81:1601. https://doi.org/10.1128/JVI.01991-06 PMID: 17108036
27. Sicard A, Pirolles E, Gallet R, Vernerey MS, Yvon M, Urbino C, et al. A multicellular way of life for a mul-
tipartite virus. eLife. 2019; 8:e43599. https://doi.org/10.7554/eLife.43599 PMID: 30857590
28. Oltmanns H, Frame B, Lee LY, Johnson S, Li B, Wang K, et al. Generation of Backbone-Free, Low
Transgene Copy Plants by Launching T-DNA from the Agrobacterium Chromosome. Plant Physiol.
2010; 152:1158–66.
29. Zhang XF, Guo J, Zhang X, Meulia T, Paul P, Madden LV, et al. Random Plant Viral Variants Attain
Temporal Advantages During Systemic Infections and in Turn Resist other Variants of the Same Virus.
Sci Rep [Internet]. 2015;5. Available from: http://www.nature.com/articles/srep15346
30. Guo Q, Zhang S, Sun R, Yao X, Zhang XF, Tatineni S, et al. Superinfection Exclusion by p28 of Turnip
Crinkle Virus Is Separable from Its Replication Function. Mol Plant-Microbe Interact. 2020; 33:364–75.
https://doi.org/10.1094/MPMI-09-19-0258-R PMID: 31880982
31. Bull JC, Godfray HCJ, O’Reilly DR. Persistence of an Occlusion-Negative Recombinant Nucleopolyhe-
drovirus in Trichoplusia ni Indicates High Multiplicity of Cellular Infection. Appl Environ Microbiol. 2001;
67:5204–9.
32. Gutie
´rrez S, Yvon M, The
´baud G, Monsion B, Michalakis Y, Blanc S. Dynamics of the Multiplicity of Cel-
lular Infection in a Plant Virus. Roossinck MJ,. PLoS Pathog. 2010; 6:e1001113.
33. Kobiler O, Lipman Y, Therkelsen K, Daubechies I, Enquist LW. Herpesviruses carrying a Brainbow cas-
sette reveal replication and expression of limited numbers of incoming genomes. Nat Comm. 2010;
1:146. https://doi.org/10.1038/ncomms1145 PMID: 21266996
34. Russell AB, Trapnell C, Bloom JD. Extreme heterogeneity of influenza virus infection in single cells.
eLife. 7:e32303. https://doi.org/10.7554/eLife.32303 PMID: 29451492
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 20 / 21

35. Bermu
´dez-Me
´ndez E, Katrukha EA, Spruit CM, Kortekaas J, Wichgers Schreur PJ. Visualizing the ribo-
nucleoprotein content of single bunyavirus virions reveals more efficient genome packaging in the
arthropod host. Commun Biol. 2021; 4:345. https://doi.org/10.1038/s42003-021-01821-y PMID:
33753850
36. Sanjua
´n R. Collective Infectious Units in Viruses. Trends Microbiol. 2017; 25:402–12. https://doi.org/10.
1016/j.tim.2017.02.003 PMID: 28262512
37. Egger D. Intracellular location and translocation of silent and active poliovirus replication complexes. J
Gen Virol. 2005; 86:707–18. https://doi.org/10.1099/vir.0.80442-0 PMID: 15722531
38. Elena SF, Sanjua
´n R. Adaptive value of high mutation rates of RNA viruses: separating causes from
consequences. J Virol. 2005; 79:11555–8. https://doi.org/10.1128/JVI.79.18.11555-11558.2005 PMID:
16140732
39. Duffy S, Holmes EC. Phylogenetic Evidence for Rapid Rates of Molecular Evolution in the Single-
Stranded DNA Begomovirus Tomato Yellow Leaf Curl Virus. J Virol. 2008; 82:957–65. https://doi.org/
10.1128/JVI.01929-07 PMID: 17977971
40. Obermeyer F, Jankowiak M, Barkas N, Schaffner SF, Pyle JD, Yurkovetskiy L, et al. Analysis of 6.4 mil-
lion SARS-CoV-2 genomes identifies mutations associated with fitness. Science. 2022; 376:1327–32.
https://doi.org/10.1126/science.abm1208 PMID: 35608456
41. McCrone JT, Woods RJ, Martin ET, Malosh RE, Monto AS, Lauring AS. Stochastic processes constrain
the within and between host evolution of influenza virus. eLife. 2018; 7:e35962. https://doi.org/10.7554/
eLife.35962 PMID: 29683424
42. Pauly MD, Procario MC, Lauring AS. A novel twelve class fluctuation test reveals higher than expected
mutation rates for influenza A viruses. eLife. 2017; 6:e26437. https://doi.org/10.7554/eLife.26437
PMID: 28598328
43. Grubaugh ND, Weger-Lucarelli J, Murrieta RA, Fauver JR, Garcia-Luna SM, Prasad AN, et al. Genetic
Drift during Systemic Arbovirus Infection of Mosquito Vectors Leads to Decreased Relative Fitness dur-
ing Host Switching. Cell Host & Microbe. 2016; 19:481–92. https://doi.org/10.1016/j.chom.2016.03.002
PMID: 27049584
44. Grubaugh ND. Translating virus evolution into epidemiology. Cell Host & Microbe. 2022; 30:444–8.
https://doi.org/10.1016/j.chom.2022.03.006 PMID: 35421340
45. Kawamura-Nagaya K, Ishibashi K, Huang YP, Miyashita S, Ishikawa M. Replication protein of tobacco
mosaic virus cotranslationally binds the 5’ untranslated region of genomic RNA to enable viral replica-
tion. Proc Natl Acad Sci USA. 2014; 111:E1620–8. https://doi.org/10.1073/pnas.1321660111 PMID:
24711385
46. Zhang H, Gong H, Zhou X. Molecular characterization and pathogenicity of tomato yellow leaf curl virus
in China. Virus Genes. 2009; 39:249–55. https://doi.org/10.1007/s11262-009-0384-8 PMID: 19590945
47. Zhang S, Sun R, Guo Q, Zhang XF, Qu F. Repression of turnip crinkle virus replication by its replication
protein p88. Virology. 2019; 526:165–72. https://doi.org/10.1016/j.virol.2018.10.024 PMID: 30391806
48. Qu F, Ren T, Morris TJ. The Coat Protein of Turnip Crinkle Virus Suppresses Posttranscriptional Gene
Silencing at an Early Initiation Step. J Virol. 2003; 77:511–22. https://doi.org/10.1128/jvi.77.1.511-522.
2003 PMID: 12477856
49. Sun R, Zhang S, Zheng L, Qu F. Translation-Independent Roles of RNA Secondary Structures within
the Replication Protein Coding Region of Turnip Crinkle Virus. Viruses. 2020; 12:350. https://doi.org/10.
3390/v12030350 PMID: 32235750
PLOS PATHOGENS
Reproductive bottlenecks of a small DNA virus
PLOS Pathogens | https://doi.org/10.1371/journal.ppat.1011365 May 1, 2023 21 / 21