ARTICLE

https://doi.org/10.1038/s41467-022-31422-0 OPEN

A plasmid system with tunable copy number

Miles V. Rouches 1, Yasu Xu1, Louis Brian Georges Cortes 2 & Guillaume Lambert 2✉

Plasmids are one of the most commonly used platforms for genetic engineering and
recombinant gene expression in bacteria. The range of available copy numbers for cloning
vectors is largely restricted to the handful of Origins of Replication (ORIs) that have been
1234567890():,;

isolated from plasmids found in nature. Here, we introduce two systems that allow for the
continuous, finely-tuned control of plasmid copy number between 1 and 800 copies per cell: a
plasmid with an anhydrotetracycline-controlled copy number, and a parallelized assay that is
used to generate a continuous spectrum of 1194 ColE1-based copy number variants. Using
these systems, we investigate the effects of plasmid copy number on cellular growth rates,
gene expression, biosynthesis, and genetic circuit performance. We perform single-cell
timelapse measurements to characterize plasmid loss, runaway plasmid replication, and
quantify the impact of plasmid copy number on the variability of gene expression. Using our
assay, we find that each plasmid imposes a 0.063% linear metabolic burden on their hosts,
hinting at a simple relationship between metabolic burdens and plasmid DNA synthesis. Our
systems enable the precise control of gene expression, and our results highlight the
importance of tuning plasmid copy number as a powerful tool for the optimization of syn-
thetic biological systems.

1 Field of Biophysics, Cornell University, Ithaca, NY 14853, USA. 2 School of Applied and Engineering Physics, Cornell University, Ithaca, NY 14853, USA.
✉email: lambert@cornell.edu

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

P
lasmids first became known for their role in spreading genetic inverter, and control the expression of complex biosyn-
antibiotic resistance in bacteria as extrachromosomal thetic pathways to optimize the production of the pigment Vio-
genetic elements capable of interspecies transfer1. Today, lacein. Our work thus provides a straightforward method for
plasmids are an invaluable tool in biotechnology and genetic generating copy number variants from a specific plasmid back-
engineering due to their small size and the ease with which they bone, which we anticipate will expand the synthetic biology
can be engineered for biotechnological applications. The plasmid toolkit and constitute the basis for more nuanced attention to
copy number is regulated by the replication origin, and number plasmid copy number in synthetic gene circuits, protein bio-
of plasmid per cell is influenced by several factors including synthesis, and other biotechnological applications.
growth conditions2–4, the host strain5,6, temperature and other
extracellular stresses7. Results
The introduction of engineered genetic systems inside bacteria, Model of ColE1 plasmid replication and copy number. The
however, often results in negative metabolic effects and added molecular mechanisms that regulate plasmid copy number con-
fitness costs. Indeed, recent studies have probed the effect of trol have been well characterized26. In the ColE1 origin of
plasmid copy number on host cell growth8,9 and the role of replication, used extensively in this work, two RNAs control
plasmids on central metabolic processes10. Recombinant gene replication (Fig. 1A): the priming RNA (RNA-p), which acts in cis
expression have also been shown to add significant metabolic as a primer near the origin of replication, and the inhibitory RNA
burdens on bacterial cells11–13, which may interfere with normal (RNA-i) which is transcribed antisense to RNA-p and acts in
cellular processes and even lead to cessation of growth. The trans to inhibit replication through binding to the priming RNA
control of plasmid copy number can also drastically alter the prior to hybridization near the origin (Fig. 1B)27.
behavior of simple repression systems14,15 which may sig- A simple macroscopic-scale model developed by Paulsson et al.
nificantly impact engineered gene circuits and should be impor- recapitulates these observations and predicts the functional
tant considerations for plasmid-based biotechnology and dependence of plasmid copy number on the molecular details
bioengineering applications. of this process28 (Fig. 1C). This process is classified as inhibitor-
Studies have also demonstrated the utility of modulating the dilution copy number control29, owing to the fact that the copy
plasmid copy number for biotechnological applications and to number is controlled through a combination of replication
enhance the behavior of engineered genetic systems. For example, inhibition and dilution of plasmids and regulatory components
runaway replication has been used as a method for increasing due to cell division.
cellular protein production16,17. Recent works have also leveraged Specifically, as the ratio of RNA-p transcription rate kp to
plasmid copy numbers to amplify a signal through a large RNA-i transcription rate ki is increased, plasmid copy number Np
increase in copy number18,19 and to increase the cooperativity will also increase according to
and robustness of synthetic gene circuits20. Single cell measure- "
 1 #
ments have also been used to accurately quantify the prevalence n ðϵ þ rÞ ρkp n
Np ¼ 1 ð1Þ
of plasmid loss within a population21. ki  kp r
Controlling plasmid copy number should allow for the reliable
scaling of biosynthesis or biological function while limiting where ϵ is the RNA degradation rate, ρ is the RNA-p priming
metabolic effects associated with engineered genetic systems. probability, r is the cellular growth rate, and n is the number of
While the first inducible plasmid copy number system was rate-limiting steps in the inhibition window28. The n → 1 and
developed in 198422 in order to show that the priming RNA of n → ∞ limits represent hyperbolic and exponential inhibition,
the ColE1 origin of replication controlled plasmid copy number, respectively (Fig. 1C).
this system had uncontrolled replication due to a removal of the As the priming-to-inhibitory RNA production ratio kp/ki
inhibitory RNA. More recently, inducible plasmid copy number approaches 1, the number of inhibitory RNA transcripts will
have been developed and are commercially available23,24, though not be present in high enough number to sequester priming RNA
these aren’t well-characterized and seem to act more as on-off transcripts, thereby failing to limit plasmids from duplicating in a
switches than truly tunable copy number systems. process called runaway replication (Fig. 1C). In addition,
Further, apart from a few well-documented point mutations and increasing the plasmid copy number may increase the metabolic
deletions that drastically increase or decrease the copy numbers of burden associated with plasmid expression and reduce the growth
specific replication origins25, the current toolkit for biologists to rate r, which in turn increases the plasmid copy number further
exert control over plasmid copy numbers is limited. Additionally, as r → 0.
we still lack a quantitative description of the relationship between On the other hand, as the RNA production ratio kp/ki
plasmid copy number and host metabolic burden. decreases, the number of plasmids inside each cell will approach
In this work, we use two approaches to actively control the 1 and small variations in the number of plasmids apportioned to
copy number of pUC19 and other ColE1-based plasmids to study each daughter cell during cell divisions may lead to the
the impact of copy number on cellular growth and metabolism. catastrophic consequence of plasmid loss. These two stresses,
Specifically, we first demonstrate a strategy for finely tuning plasmid loss and metabolic burden, necessitate the evolution of
plasmid copy number in a way that is dependent on the con- copy number control systems to ensure that neither stress results
centration of an exogenous inducer molecule. Then, we develop a in the loss of the plasmid30. The experimentally measured RNA
method that uses a parallelized assay for rapid screening of the production ratio kp/ki is closer to 0.3331, suggesting that the
dependence of a system on plasmid copy number. ColE1 ORI is naturally in a state that balances plasmid loss and
We further characterize our systems at both the macroscopic metabolic burdens.
and single cell level to uncover a relationship between gene copy
number, protein production, and cellular growth rates. Our Tuning priming and inhibitory RNA transcription to control
approach also leads to insights into the variation in plasmid copy ColE1 copy number. The Paulsson model of ColE1-type repli-
number and the phenomena of plasmid loss and runaway repli- cation predicts that the RNA-p transcription initiation rate
cation at the single-cell level. We finally demonstrate that should determine the plasmid copy number without drastically
manipulations of plasmid copy number can be used to tune the altering the sensitivity of copy number control29. Thus, in order
behavior of synthetic gene circuits such as a simple CRISPRi to only control the activity of the promoter upstream of the

2 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0 ARTICLE

Fig. 1 Replication of ColE1 origin plasmids. A Map of the pUC19 plasmid. The origin of replication consists of a 555 bp region containing the priming RNA
(RNA-p), the inhibitory RNA (RNA-i), the inhibition window, and the site of replication initiation (ORI). B Diagram of the steps leading to ColE1 replication.
In the ColE1 Origin, reversible binding of the inhibitory RNA-i to RNA-p exercises control over the copy number by inhibiting replication (left branch).
Production of RNA-p transcript triggers plasmid replication through the hybridization of RNA-p with the ORI (right branch). C Predicted plasmid copy
number for hyperbolic (upper) and exponential (lower) initiation models28 plotted as a function of RNA-p transcription initiation rate. Inset shows the
predicted dependence of plasmid copy number on host cell division time. Model parameters: ϵ = 0.545 min, ρ = 0.65, r = 0.01 min, ki = 0.95 min.
D Plasmid copy number measurements made by digital droplet PCR of the aTc-inducible pUC-pTet plasmid at increasing aTc concentrations; n = 3, error
bars derived from Poisson distribution of ddPCR counts. E Growth rates of cells hosting the pUC-pTet plasmid at increasing plasmid copy numbers. We see
a marginal effect of plasmid copy number on cellular growth rates, inset shows individual growth curves colored by aTc concentration. Data points
are means across n = 6 biological replicates, error bars = std. dev. F Plasmid copy number measurements of a plasmid that contains an additional copy of
the inhibitory RNA under the control of an IPTG-inducible promoter; n = 3, error bars derived from Poisson distribution of ddPCR counts. G Growth rate vs
plasmid copy number of the IPTG-inducible plasmid shows a strong metabolic burden at high levels of RNA-i expression. Data points are means across n =
6 biological replicates, error bars = std. dev.

priming RNA, we first replaced the native priming promoter in When no IPTG was added to the cells we measured ~ 270
the pUC19 plasmid with an anhydrotetracycline (aTc) inducible plasmids/genome, a number that agrees with what we have
promoter (Fig. 1D). Liquid cultures of TetR- and LacI- measured for the standard pUC19 plasmid, indicating that our
overexpressing E. coli carrying this plasmid were grown in aTc system efficiently represses the additional inhibitory RNA
concentrations spanning 4 orders of magnitude and the plasmid transcript and does not affect the copy number in the absence
copy number was determined via digital droplet PCR (ddPCR) of induction. Induction of the inhibitory RNA greatly reduced
from total DNA isolates by measuring the ratio of the bla gene on plasmid copy number to approximately 30 plasmids/genome at
the plasmid to the genomic dxs gene32. full RNA-i induction, demonstrating that tuning RNA-i produc-
This plasmid showed a robust control of copy number with tion rate can be used to control the plasmid copy number.
aTc induction (Fig. 1D). Copy numbers ranged from 1.4 copies It is interesting to note that we observed paradoxical effects on
per cell in the complete absence of aTc to roughly 50 copies per the growth rate from cells harboring this plasmid - at high
cell at full induction (100 ng mL1 aTc). plasmid copy numbers we see a faster growth rate and at lower
We subsequently measured the growth rate of cells harboring copy numbers we see a markedly slower growth rate (Fig. 1G).
the pUC-pTet plasmid as a function of the aTc concentration in a These observations are markedly different from what we see in
microplate reader (Fig. 1E). Though the absolute difference in the pUC-pTet plasmids, where increased plasmid copy number
growth rates is small, a cell with a few copies of the plasmid grows has a relatively modest effect on the growth rate of host cells.
roughly 10% faster than a cell with 50 plasmids, we can clearly see Instead, the reduction in the maximum cell density reached by
that increasing the plasmid copy number has a weak, though the IPTG-induced populations (Fig. 1G, inset) suggests that
consistent, effect on the host cell growth rate. overproduction of the inhibitory RNA may completely abolish
To complement the work above, we sought to demonstrate that plasmid replication in some cells, preventing them from dividing
the transcription level of inhibitory RNA can be used to control further.
ColE1 copy number as well. Seeking to do this with minimal
perturbation to the native pUC19 origin of replication, we
inserted a second copy of the inhibitory RNA downstream of the Gene expression scales with plasmid copy number. Under-
IPTG-inducible Lac promoter on the pUC19 plasmid. This gives standing the interplay between plasmid copy number, host cell
us control over the inhibition of replication through the addition growth rate, and gene expression is key to optimizing protein
of exogenous IPTG (Fig. 1F). production in bacterial cell cultures. If protein production is too

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

data and ddPCR measurements at the endpoints (0 and

100 ng/μL) we predicted the PCN of our pUC-pTet-sfGFP
plasmid in different media compositions (Fig. 2D). From this
data we observe that as the richness of the media is decreased we
see a net increase in the plasmid copy number at every aTc
concentration. While this results in a shift in the induction curve
we nonetheless see a broad range of induction across different
media conditions. We further observed that there was no
consistent effect of increased plasmid copy number on host cell
growth rate in the two tested minimal media compositions (Sup.),
perhaps indicating that other cellular processes are the dominant
limiting factor on host cell growth in these conditions.
Our pUC-pTet plasmid provides a simple way to control gene
expression without the need to insert an inducible promoter
upstream of the gene of interest.

Metabolic costs of the pUC-pTet plasmid at the single-cell level.

Fig. 2 Control of gene expression through induction of plasmid copy While macroscopic measurements show a clear interdependence
number. A Schematic of the effects of adding aTc to the pUC-pTet-sfGFP between gene expression and copy number, single cell measure-
plasmid. As the concentration is increased the number of plasmids and ments can provide a more powerful path towards understanding
sfGFP genes increases as well as the metabolic cost of the plasmid. the relationship between plasmid copy number and gene
B Growth rates of cells hosting the pUC-pTet-sfGFP plasmid as a function expression. In particular, single-cell studies allow us to determine
of aTc concentration. C sfGFP Production of the pUC-pTet-sfGFP plasmid the distributions of gene expression of single cells at varying
grown at different aTc concentrations. D Estimated plasmid copy numbers plasmid copy numbers and directly observe the phenomena of
from sfGFP measurements of our plasmids. We see that the copy number plasmid loss and runaway replication.
induction curve shifts towards lower aTc concentration as with decreasing To first corroborate protein production measurements with
media richness. For all panels, Data points are means over biological plasmid copy number, we observed the pUC-pTet-sfGFP system
replicates, error bars = std. dev. (n = 6). for up to 150 generations in a microfluidic device33,34.
In agreement with population-wide measurements (Fig. 2C),
individual cells show an increase in the mean fluorescence
low, one may have to process massive amounts of cells to obtain a with plasmid copy number (Fig. 3A). We also observe a
sufficient amount of their target protein. If too much protein is marked increase in the fluorescence of some cells at high aTc
produced, on the other hand, cells may stop growing or be out- concentrations. Interestingly, cells with an increased fluorescence
competed by cells harboring plasmids that lack the correct insert. level are found to be elongated (Fig. 3A and Supplementary
Understanding the interplay between plasmid regulation and Fig. 1). The intensity distribution (Fig. 3B) undergoes a profound
protein expression is key to solve protein production maximiza- change as the plasmid copy number increases: the distribution is
tion problems. peaked at a very low intensity at low plasmid copy number but, as
To directly measure the effects of plasmid and gene copy the copy number increases, the distribution widens and a high-
number on gene expression and host cell growth rate, we first intensity tail emerges.
inserted sfGFP downstream of a constitutive promoter on our The most drastic shift in the distribution occurs between 0.3 ng
pUC-pTet plasmid, whose copy number is controlled through mL−1 and 1 ng mL−1, where the average cell intensity (Fig. 3C) is
exogenous aTc induction. This gives us a straightforward way of in good agreement with population wide measurements (Fig. 2C),
monitoring the effects of plasmid copy number on gene but each population sees a steady increase in the number of
expression. Cell cultures with this tunable copy number plasmid individual cells with increased sfGFP levels (Fig. 3A, red arrows).
were grown in a microplate reader where simultaneous measure- As the plasmid copy number increases further, a growing number
ments of fluorescence and growth rate could be made. of cells reach an “activated” state where they show fluorescence
We first observe clear differences in growth rates as a function levels as high as those full pTet induction. Eventually, all cells for
of both plasmid copy number and aTc concentration (Fig. 2B) in aTc concentrations above 3 ng mL−1 reach this activated, fully-
H media. We see a roughly 15% decrease in host growth rate at induced state (Fig. 3C).
maximal induction of copy number (~50 copies per cell). This As the plasmid copy number sharply increases between 0.1 ng
metabolic burden is significantly higher than the pUC-pTet mL−1 and 1 ng mL−1 of aTc, we observe a clear drop in average
plasmid (Fig. 1E), indicating that the majority of the reduction in growth rate (Fig. 3D and Supplementary Fig. 2) similar to bulk
growth rate comes from the overexpression of sfGFP and not measurements (Fig. 2B). The distribution of cellular growth rates
from the increased plasmid copy number. In this sense, addition (Fig. 3D) show that this drop is largely due to a fraction of cells
of sfGFP to the plasmid provides a stronger coupling between the that suffer from reduced growth rate. This is likely due to the
host growth rate and plasmid copy number by taking up metabolic burden of plasmid maintenance and sfGFP expression
additional cellular resources. severely limiting the growth of a subset of the cellular population.
Despite the fitness costs associated with increasing the plasmid Above 1 ng mL−1 aTc, the growth rate of healthy cells seems to
copy number, we nevertheless observe an almost perfect sfGFP stabilize (Supplementary Fig. 2). However, we notice an
induction curve when compared to the aTc concentration increasing fraction of elongated cells (Fig. 3E), many of which
(Fig. 2C). We performed identical measurements in 3 different have lost the ability to grow and turned extremely bright
media conditions: H Media, M9 minimal media + 0.2% glycerol (Supplementary Movie 1). The inability of a fraction of cells to
+ 0.2% casamino acids, and M9 minimal media + 0.2% glucose. grow may explain why growth rate continually decreases even
In less optimal media we saw that the sfGFP induction curve was though plasmid copy number saturates at high aTc levels in bulk
shifted towards lower aTc values (Fig. 2C). From our fluorescence measurements (Fig. 2B).

4 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0 ARTICLE

Fig. 3 Microscopic Analysis of sfGFP Expressed by the pUC-pTet. A Representative snapshots of microfluidic chambers at various aTc concentrations.
Slow-growing cells at high copy number are indicated by orange arrows and cells at 0.3 ng/mL of aTc in an “activated” state are indicated by purple arrows.
Experiments in 100 ng/mL and 0.1 ng/mL conditions were repeated twice with similar results, all other experiments were repeated once. B Distribution of
cellular fluorescence intensities at increasing aTc concentrations. C Normalized mean fluorescence as a function aTc concentration, log scale. Inset:
Normalized mean fluorescence vs aTc concentration on a linear scale. N = number of cells listed in corresponding condition in Panel B. Data points are
averages over all cells, error bars = std. dev. of distributions in panel B. D Growth rate distributions of cells grown at 0.1, 0.3, and 1 ng/mL aTc. E Fraction of
slow-growing, elongated cells over time at several aTc concentrations. This measurement is indicative of the extent to which cells are burdened by the high
plasmid copy numbers. F Proportion of viable cells as a function of time at various aTc concentrations. The sharp decrease in the number of viable cells at
low aTc concentrations provides insight into the extent to which plasmid loss occurs at low copy numbers.

At low plasmid copy numbers, we see a decreased number of For each recovered mutant, the abundance of each construct at
cells in each microfluidic chamber. In fact, nearly every cell each time point was used to infer the growth rate of each
undergoes lysis within the first 1500 min of tracking for aTc individual construct. By coupling this information with the fold
concentrations lower than 1 ng mL−1, suggesting that cells lost change in sequencing counts from the library prior to
resistance to carbenicillin due to plasmid loss (Fig. 3F). transformation, we determined the relative copy numbers of
These observations highlight the delicate balance between every plasmid in this library (Fig. 4B, C) and their growth rates
plasmid loss at low copy number and heightened metabolic (Fig. 4D). Our method was able to generate 1,194 plasmids, each
burdens at high copy numbers. Our results demonstrate that both with a distinct plasmid copy number ranging from less than 1
phenomena occur stochastically and on a per-cell basis. copy per cell to close to 800 copies per cell (Supplementary
Data 1).
To make sure our sequencing-based method yields accurate
Massively parallelized assay determines copy numbers of plasmid copy number measurements, we next sought out to
ColE1-derived plasmids. Though we were able to obtain robust calibrate our sequencing-based numbers to absolute copy
control over plasmid copy number with our inducible plasmid, numbers made using digital droplet PCR measurements.
we next sought to exert control over plasmid copy number Individual constructs of 9 promoters were selected (light blue
without the need for external inducer molecules. In order to datapoints in Fig. 4B) and their absolute copy number were
survey the landscape of possible plasmid copy numbers that arise determined via digital PCR by the ratio of dxs to bla genes32.
from changes of the transcription rates of the RNAs controlling There is a good agreement between fold change in sequencing
ColE1 replication, we simultaneously constructed 1024 different abundance and the plasmid copy numbers as measured by digital
variants of the pUC19 plasmid through site directed mutagenesis droplet PCR (Fig. 4D, RMS error 0.994), indicating that the fold
of the −35 and −10 boxes of the promoter controlling the change in abundance of sequencing counts with a modest
priming RNA and at the +1 site of the priming RNA transcript correction applied to correct for differences in growth rates is a
(Fig. 4A, Pp library). The same procedure was performed on the sound measure of plasmid copy number.
promoter controlling the inhibitory RNA (Fig. 1A, Pi library), The simultaneous measurement of plasmid copy numbers and
giving us 2 separate libraries of plasmids: one with a variable the growth rates of their host cells provides us with a powerful
priming RNA transcription rate kp and another with a variable measurement of the interdependence of copy number and
inhibitory RNA transcription rate ki. metabolic burden. We were not able to determine any key
These libraries were constructed using multiplexed PCR and sequence motifs that could be used to predict plasmid copy
KLD assembly35 and electroporated into competent E. Coli cells number (Supplementary Fig. 3). While a high plasmid copy
(Fig. 4A, step 2). Cell cultures were grown in selection conditions number should reduce host cell growth, to what extent that
until they reached an OD600 of 1.0 and subsequently diluted by a occurs is not known. Our measurements of plasmid copy
factor of 4 and grown again to an OD600 of 1.0 (Fig. 4A, step 3). numbers and growth rates for the 830 plasmids in the priming
Excess culture from each step of dilution was set aside for promoter library, all with nearly identical origins of replications,
sequencing (Fig. 4A, step 4). provide a powerful way to deduce this relationship.

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

Fig. 4 Massively parallelized copy number assay. A Procedure for massively parallelized copy number assay. 1) The promoter upstream of the priming
RNA was randomized with site directed mutagenesis. 2) Plasmids were ligated and cells were transformed with this library of plasmids with high efficiency
and grown until OD600 of 1.0. 3) At which point a fraction of the culture was diluted and regrown and the remaining fraction was saved for plasmid
extraction. 4) Sequencing libraries were generated from isolated plasmids and sequenced using next generation sequencing. Growth rates and Plasmid
Copy Numbers (PCN) were then calculated from the frequencies of sequencing counts present at each time point. B Ranked list of the measured fold-
change for plasmids generated through mutations to the promoter driving the priming RNA in this work (n = 830). Promoters with a fold-change below
zero will result in plasmid loss (PCN < 1). (lower) The distribution of copy numbers arising from mutations to the priming RNA promoter. C Ranked list of
the measured fold-change for plasmids generated through mutations to the promoter driving the inhibitory RNA in this work (n = 365). D Agreement
between plasmid copy number as measured by sequencing counts and digital droplet PCR. Data points are averages over n = 3 biological replicates, error
bars = std. dev. E The relationship between plasmid copy number and growth rate as measured across 830 variants of the pUC19 plasmid. The size of each
data point is inversely proportional to the error in the measurement, with the top, middle two and lowest quartiles representing an error of less than 5%,
between 5% and 25%, and more than 25%, respectively. Data from Fig. 1E is plotted (red to blue gradient points) A linear metabolic burden seems to
account for the observed relation reduction in growth rate with increasing copy number. Data points represent average over n = 3 measurements.

In Fig. 4E, we observe that, as the plasmid copy number plasmid size (pUC19 is 2686 bp in size, which is approximately
increases, the growth rate of host cells decreases. A simple equal to 0.058% the length of the E. coli genome).
relationship in which the metabolic burden of plasmid expression To determine the extent to which antibiotic supplementation
is linear with the copy number largely agrees with these findings affected the growth rates and plasmid copy numbers36 of cells
–i.e., cost = a ⋅ b ⋅ x + b, where x is the plasmid copy number, hosting these plasmids we inserted sfGFP into our library and
with a = 0.065% ± 0.006% and b = 0.019 min−1 ± 0.003 min−1. measured the growth rates and fluorescence of eight constructs
Where errors were calculated from the variance estimates of the on a microplate reader. We observed no noticeable change in the
coefficients. This relationship can arise from a simple resource mid-exponential or endpoint fluorescence between samples
allocation model in which each plasmid takes up a small fraction treated with and without antibiotics. (Supplementary Fig. 4) This
of the resources necessary for cellular growth proportional to the indicates that there was no measurable change in the mean copy

6 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0 ARTICLE

Fig. 5 Optimization of violacein production by tuning plasmid copy number. A The action of 5 enzymes on two tryptophan molecules yields violacein (B)
The complete VioABCDE pathway was inserted in the pUC-pTet plasmid. C Violacein production scales with aTc concentration when the pathway is
inserted into the pUC-pTet plasmid. D A library of variable copy number plasmids containing the VioABCDE pathway was created by performing
mutagenesis on the priming promoter of a ColE1-based plasmid. E Violacein production is can be scaled up or down by plasmid copy number. At very high
copy number production is low due to increased metabolic burden. Inset: Lag time is drastically increased in constructs with high violacein production. Error
bars = std. dev. (n = 4). F A two plasmid system was created in which the VioABC enzymes were placed on a single plasmid of fixed copy number and the
VioDE were placed in our library of variable copy number plasmids. G The use of a two plasmid system outperforms a the single plasmid system in raw
violacein yields. H Growth rate is dramatically improved with a two plasmid system. Our randomized approach leads to the creation of constructs with fast
growth rates and high amounts of violacein production.

number due to antibiotic exposure. We observed a constant assembly39 (Fig. 5B). We then grew cells with the pUC-pTet-Vio
decrease in the growth rate when samples were treated with plasmid at increasing aTc concentrations and observed a clear
antibiotics that was independent of the plasmid copy number. increase in the violacein production (as measured by the
Together these indicate that our supplementation with antibiotics absorbance at 650 nm of methanol extractions from cells) with
did not have an observable effect on plasmid copy numbers or increasing aTc concentrations (Fig. 5C). This indicates that
measured growth rates in our experiments. increasing plasmid copy number is a viable way to scale up the
This results opens up the possibility for improved models of product of a biosynthetic reaction.
plasmid evolution and replication, as they indicate that plasmids Additionally, we inserted the VioABCDE pathway into a pUC
impose a metabolic burden that is linear with their copy number plasmid under the control of a moderate-strength promoter, and
on their hosts. Knowledge of this relationship between host we used site directed mutagenesis of the origin of replication in
growth and plasmid copy number may shed light on the the manner described above to generate a plasmid copy number
evolution of cis and trans acting regulatory elements as well as library (Fig. 5D). We plated cells hosting this library and picked
the stringency of plasmid copy number control mechanisms. individual colonies to measure their vioalcein production. We
sequenced the promoter controlling the priming RNA of these
plasmids to infer their copy number from the results of our
Plasmid copy number controls violacein biosynthesis. We parallelized assay and clearly saw that up to a certain point,
sought to demonstrate the potential for using plasmid copy number increases in plasmid copy number lead to an increased vioalcein
to control complex biosynthetic processes comprising several pro- yield (Fig. 5E). This further demonstrates that plasmid copy
teins and enzymes. We chose to focus on Violacein (Fig. 5A), a numbers are a reliable way to scale the production of a
purple pigment formed by the condensation of two tryptophan biosynthetic compound.
molecules37 which is commonly used to demonstrate biosynthesis We then reasoned that we could tune the stoichiometric ratio
optimization38. The violacein biosynthetic pathway contains five of the enzymes in this pathway to optimize violacein biosynthesis
enzymes, VioABCDE, each under the control of a different pro- by placing some enzymes in our tunable plasmid copy number
moter and spanning approximately 8kb of DNA (Fig. 5B). Hence, system and the others in a plasmid of fixed copy number.
we reasoned that tuning the plasmid copy number of the whole Specifically, we placed VioABC on a plasmid with the p15A
Violacein production pathway may serve as an effective method to origin of replication (~20–40 copies/cell) and while VioD and
scale up the expression of VioABCDE while keeping the production VioE were placed on plasmids with priming promoter library
of each enzyme under the same stoichiometric ratio. (Fig. 5F). The violacein production level of 15 colonies hosting
To achieve this, we constructed a pTet inducible plasmid both plasmids was measured. We found that splitting the
harboring the Violacein biosynthesis pathway using Gibson pathway on to two plasmids improved vioalcein yields (Fig. 5G)

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

and increased the growth rates of host cells (Fig. 5H) relative to
cells hosting the single plasmid system. While the improvements
in violacein were substantial, more impressive was the several fold
increase in the growth rate of cells hosting this pathway even at
very high levels of violacein production. Interestingly, there was
no clear relationship between violacein production and host cell
growth rates, as our highest producing constructs had growth
rates in the upper half of those measured and some of the slowest
growth rates were observed in constructs that had among the
lowest levels of violacein production (Fig. 5H).
Together this indicates the tuning plasmid copy numbers is a
viable way to optimize biosynthesis both through increased raw
product yields and substantially increased growth rates of host
cells. The value of our randomized approach is accentuated in our
observation that plasmid copy number and growth rate were not
necessarily correlated with final vioalcein yields (Fig. 5H) in our
two plasmid system.
Fig. 6 Control of a CRISPRi inverter with competitor binding sites. A Top:
Diagram showing how production of a dCas12a CRISPR RNA (crRNA) is
Control of a CRISPRi system through sponge binding sites. controlled by an aTc-inducible promoter, and the dCas12a+crRNA duplex
Controlling the plasmid copy number may also be useful in can either bind to the active site Sa, which turns off production of sfGFP, or
synthetic biology and in the creation of robust gene to a decoy site Sd residing on a “sponge” plasmid. Bottom: Fluorescence
circuits18,20,40,41. To test this, we used a simple CRISPR-dCas12a intensity vs time for CRISPR-dCas12a systems with variable numbers of
inverter to demonstrate the effect of adding “sponge” binding sponge sites (Nc). B Induction curves for a CRISPRi system with sponge
sites to a simple genetic circuit through manipulation of plasmid sites on plasmid with varying copy numbers, n = 3 biological replicates for
copy numbers. Sponge sites are additional binding sites used to each data point. C ON-OFF levels for variable numbers of sponge sites. The
titrate out (“soak up”) transcription factors, and they have been ON and OFF fluorescence level of the inverter both increase as the number
used to redirect flux in metabolic pathways to optimize arginine of sponge sites varies from 0 to 270. The dynamic range decreases with
production42, to activate silent biosynthetic gene clusters43, to increasing sponge copy number. D Active site occupancy vs number of
tune gene expression timing44, and mitigate protein toxicity45. In sponge sites, as the number of sponge sites increases the active site
these studies, variable numbers of sponge sites are placed on high becomes less occupied. The gray area represents the numerical solution of
or low copy number plasmids. a fold-change model15,35 for dCas12a where the binding energy equal
Here, we added a sponge site to a few isolates from our tunable to − 13.25 kBT (top) and − 12.25 kBT (bottom).
copy number library to alter the behavior of a simple genetic
circuit. Specifically, we use a two plasmid system in which a low-
copy plasmid (pSC101, 2–5 copies per cell) contains an aTc Discussion
inducible CRISPR-Cas12a guide RNA which can bind to an active This work details the development of a system of tunable copy
site Sa and turn OFF a promoter driving sfGFP, while the second number plasmids and demonstrates their utility in protein
plasmid contains a decoy binding site Sd site that sequesters expression, biosynthetic pathway optimization, and synthetic
CRISPR-Cas12a molecules away from the active site Sa. The biology applications. Using a combination of high-throughput
sponge plasmids were picked out from our library of variable measurements and single-cell experiments, we deduce relations
copy number plasmids generated in Fig. 4B and span copy between plasmid copy number and host cell’s growth rate, protein
numbers from 30 to 270. expression, biosynthesis, and the performance of simple reg-
Time-series measurements of the fully-induced CRISPRi ulatory elements. This system could help resolve known issues
inverter show that the OFF state of the inverter has a higher currently plaguing the design and scaling up of synthetic gene
fluorescence level as the number of sponge sites is increased circuits such as leaky expression, narrow ranges of activation and
(Fig. 6A). This suggests that the addition of sponge sites reduces repression control, and a large metabolic burden due to compe-
the occupancy of the active site Sa and decreases the repression tition over a common pool of resources47–49.
efficiency of the CRISPRi inverter. Using a massively parallel directed mutagenesis approach, we
We also observe a reduction in the dynamic range of circuits as find that the most straightforward and reliable way to change the
the number of sponge sites is increased (Fig. 6B), where cells with plasmid copy number in the ColE1 Origin is through mutations
more than 150 sponge sites have at least a 40% reduction in to the promoter controlling the priming RNA. Changes to the
dynamic range compared to those with no sponge sites. In machinery controlling inhibitory RNA levels are much less likely
contrast, the addition of 30 to 50 sponge sites only reduces the to give stable constructs as mutations to the promoter controlling
dynamic range by approximately 6–8% (Fig. 6C). inhibition change the sequence of the priming transcript. Chan-
While a large dynamic range is often desirable in genetic ges to the sequence of the priming transcript can have drastic
devices, a reduction in dynamic range could be used to divert the effects on plasmid copy numbers through secondary structure
flux in a metabolic pathway, ensure that the concentration of changes50. Overall, these results suggest that the ColE1 replication
some protein stays within a given range, or to reduce the toxicity machinery is more robust to changes in the priming RNA tran-
of CRISPR-Cas proteins46 due to spurious off-target binding. scription initiation rate and much less robust to changes in the
Using a simple model of protein-DNA binding15, we can discover parameters detailing inhibition.
that the occupancy of the active binding site Sa (Fig. 6D) in the Our multiplexed method of generating a plasmid copy number
OFF state is reduced from nearly 100% with zero sponge sites to library allows one to rapidly screen the effects of plasmid copy
50% when 270 sponge sites are present. Overall, we demonstrate number on a given system and could be tailored to select for
that plasmid copy number can be used to accurately tune the level genetic devices that perform the best under a given set of con-
of transcription factor titration in a simple genetic inverter. ditions. A key finding of this facet of the work is that ColE1-type

8 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0 ARTICLE

plasmids impose a linear metabolic burden on their hosts pro- The PCR product was then circularized with electroligase (NEB) and
portional to the relative size of the plasmid, which may shed light transformed with high efficiency (≥1.5 million CFU/mL) into Endura
Electrocompetent cells. The transformation efficiency was determined by plating of
on the evolution of various regulatory elements in inhibitor- serial dilutions of recovered cells. After transformation liquid cultures were grown
dilution copy number control mechanisms. in Terrific Broth (TB, VWR) supplemented with 0.4% glycerol until reaching an
Our single cell measurements also show that plasmid loss and OD600 of 1.0, at which point cultures were diluted 1:4 and regrown to OD600 of 1.0,
runaway replication are quite common phenomena. These find- a procedure that was repeated 4 times. Excess culture from each time point was set
ings are in line with recent single cell measurements of plasmid aside for plasmid isolation. Plasmids were isolated by miniprep (Zymo) and a
243 bp region of the ori was amplified (pBR322-5prime-promoter1-rev 5′-
copy numbers21 and illustrate that very careful attention should TCTACACTCT TTCCCTACAC GACGCTCTTC CGaTcTCAGA
be given to the plasmids used to house genetic circuits in order to CCCCGTAGAA AAGaTcAAAG GaTcTTC-3′, pBR322-3prime-for 5′-
ensure reliability. Since we observed an increasingly large fraction GACTGGAGTT CAGACGTGTG CTCTTCCGAT CTCGAGGTAT
of non-growing cells as we increased the plasmid copy number, GTAGGCGGTG CTACA-3′) via PCR, at which point universal primers and
indices were attached to the segment using an additional PCR step. Prepared
our results further suggest that lower growth rate for populations libraries were then diluted to 50 pM and sequenced on the Illumina iSeq using
carrying a high-copy number plasmid may be explained by non- overlapping 2 × 150 bp overlapping reads.
growing cells as opposed to a universal reduction in cellular Only sequences with zero mismatches in the ori region were used for further
growth rates. While this work is restricted to the ColE1 origin of analysis. The distribution of the number of counts per construct can be seen in
Supplementary Fig. 5 and a full list of the promoter sequences and their respective
replication, similar principles can be used to tune the copy number of counts at each time points is provided in Supplementary Data 1, 2, and 3.
numbers of other inhibitor-dilution controlled plasmids. Speci-
fically, one could mutate the promoters controlling the priming Construction of inducible plasmids. The pUC-pTet plasmid, whose copy number
and inhibitory RNAs in any other antisense-RNA controlled is inducible in the presence of aTc was constructed using a PCR-based insertion as
origin of replication to generate a library of plasmid copy num- follows. Primers (pTet-tn10-for 5′-AAATAACTCT aTcAATGATA GAGTGT
bers from a template plasmid. CAAG AAGaTcCTTT GaTcTTTTCT ACGGGGTCTG A-3′, pTet-tn10-rev 5′-
TACCACTCCC TaTcAGTGAT AGAGATaTcT GCAAACAAAA AAACCA
The implications of plasmid copy number go beyond its effects CCGC TACCAGC-3′) overlapping with the pUC19 origin of replication and
on gene expression and host cell fitness that we’ve presented thus containing a pTet-inducible promoter were used to linearize and amplify the
far. As vehicles of horizontal gene transfer, the prevalence of pUC19 backbone. The linearized plasmid was then circularized using a KLD
which increases with plasmid copy number, plasmids play a key reaction (NEB) and transformed into competent cells via electroporation. During
the recovery period 1x aTc was added to the recovery media to facilitate replication
role in bacterial ecology and evolution51. On the other hand, a
of the plasmid.
consequence of a high copy number is a reduced retention and The pUC-pLac-RNAi plasmid, whose copy number can be reduced through
fixation of novel plasmid variants52. Together, these two issues the addition of IPTG, was constructed as follows. The inhibitory RNA (commonly
create a complicated image of the role of plasmid copy number in called RNAI) from the pUC19 plasmid was isolated via PCR. In the process 15 bp
bacterial diversity and evolution, it’s possible that the methods for overhangs were added to facilitate Gibson assembly. Another pUC19 plasmid was
the linearized downstream of the pLac promoter (primers) and the linearized
controlling plasmid copy number presented here can be used to plasmid together with the inhibitory RNA insert were assembled via Gibson
help shed light on the role of plasmid replication in bacterial assembly at 50 °C for 1 h and transformed into competent cells via
evolution. electroporation.
The evolution of the fitness cost associated with plasmid-
borne antimicrobial resistance genes is of central importance in Copy number and growth rate measurement via next-generation sequencing
preventing the spread of drug-resistant bacteria53. Our work reads. To determine the relative plasmid copy number from the abundance of
next generation sequencing reads we compare the fraction of the sequencing
presented here could provide a clear path for investigating the reads at several time points with that in the initial library. The ratio of the two,
role of plasmid copy number in the evolution of the cost of with a small correction applied to account for differences in the growth rates of
antimicrobial resistance. Further, the fitness cost of plasmid cells harboring different plasmids, is what we report as a relative plasmid copy
expression is highly species-dependent and dependent on the number. Unless otherwise noted, all experiments and cell cultures were per-
formed at 37 °C.
diversity of a bacterial community51 and recent findings have We measure the initial fraction of the reads in the ligation product (fi) and the
shown that plasmids can quickly achieve fixation in a population final fractions of the reads of the plasmids harvested at different time points
under non-selective conditions54, further understanding of (f1,f2,f3,f4). Differences in the doubling time can be measured from differences in
plasmid replication mechanisms and copy number effects might the fractions of reads from points separated in time. These relationships can be
described in the equations below. First, to determine the doubling time from
allow for a more nuanced understanding of these varied prop-
adjacent time points:
erties of plasmids.
Together, these findings demonstrate that plasmid copy f n ¼ f n1 2δtðττav Þ
1 1

number can have drastic effects on host cellular processes and The above line states that the fraction at a given time point is equal to the
provide straightforward ways of investigating these effects in fraction at the previous time point multiplied by the difference in the number of
other systems. Our results underscore the importance of tuning doublings from the bulk culture. Below we have rearranged the equation to solve
plasmid copy number for the optimization of biosynthetic path- for the growth rate from the variables we can either observe or control.
ways, gene regulatory circuits, and other synthetic biological  f 
Log 2 f n 1 1
systems. n1
þ ¼
δt τ av τ
Secondly, to determine the copy number from the growth rate and the
difference in sequencing counts we express the fraction at the first time point as the
Methods product of the fraction in the ligation, the relative copy number, and the number of
Next-generation sequencing library preparation. The multiplexed libraries of excess doubling times:
plasmids was generated by site-directed mutagenesis using the NEB Q5 High-  
fidelity polymerase. Primers (pBR322-promoter1-1024-rev 5′-GAKKKMAAGA
f f ¼ f i C2t ττav
1 1

AGaTcCTTTG aTcTTTTCTA CGGGGTCTG-3′,pBR322-promoter1-1024-for 5′-

CTTTTTTTCT GCGCGTWWRM TGCTGCTNGC AAACAAAAAA ACCA
CCG-3, pBR322-promoter1-1024-rev 5′-GTTAGGCCAC CAKKKMAAGA ACTC  
f n t τ τ av
1 1
TGTAGC ACCGCCTACA TACC-3′, pBR322-promoter1-1024-for 5′-TACGG 2 ¼C
fi
CTWWR MTAGAAGNAC AGTATTTGGT aTcTGCGCTC TGCTG-3′) con-
taining degenerate nucleotides in the −35, −10, and +1 sites of the promoters By inserting the above expression for the doubling time, we can have the relative
controlling transcription of both RNAs were used to amplify the standard pUC19 copy number solely in terms of the fractions of each sequencing reads and
plasmid (NEB). After PCR amplification samples were digested with DpnI (NEB) the ratio of the time between time points and the time between transformation
restriction enzyme to remove excess pUC19 template. and the first time point.

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

f f f n1  t Violacein production measurements. Cultures of single colonies hosting plas-

2 δt ¼ C mids containing the violacein biosynthesis pathway were grown overnight in 3 mL
fi fn
Terrific Broth with appropriate antibiotics. Overnight cultures were then shaken
for 24 h at 300 rpm at room temperature in 15 mL Falcon tubes. First, 200 μL of
This analysis hinges on the assumption that upon transformation each cell liquid culture was centrifuged and cells were resuspended in 50 μL of water. Then,
receives a single plasmid, that cells do not transfer plasmids between each other, 450 μL of methanol was added to the resuspended cells and they were shaken for
and that the number of transformants of a given plasmid is proportional to the 1 h at room temperature. Cell extracts were spun down and 200 μL of the super-
amount of that construct present in the initial electroligation product. Further, this natant was transferred to a 96-well plate where the absorbance at 650 nm was used
analysis assumes that cells harboring different constructs have identical lag times to quantify violacein production.
and that cells remain in the exponential growth phase until an an OD600 of 1.0
(Supplementary Fig. 6 provides evidence of this). Single cell microscopy. The single cell data was collected within a month over
8 subsequent experiments with varying aTc concentrations. Each experiment was
Absolute copy number determination via digital droplet PCR (ddPCR). Digital run for 1300–5000 min, corresponding to at least 20–80 generations. All experi-
Droplet PCR (ddPCR) was used to corroborate our sequencing measurements of ments were started from the same stock bacterial colony, initially grown to
plasmid copy number. Colonies hosting the appropriate constructs were picked off OD600 0.5 at 10 ng mL−1 aTc and kept refrigerated. In order to start the experi-
of plates after transformation and grown in 3 mL of TB or H media (10 g L−1 ment, 3 μL of the stock colony was dispersed in 3 mL of H media with plasmid
tryptone, 8 g L−1 NaCl) to an OD600 of 1.0. Cells were then serially diluted 8000- selecting antibiotic carbenicillin and aTc. The cells were grown to OD600 0.5 ± 0.1,
fold in nuclease free ddH2O. Cells were lysed by heating at 95 °C for 20 min. After concentrated by centrifugation and inserted in plasma cleaned microfluidic chips.
lysis, 1 μL of the lysate was used in each digital PCR reaction. Two digital PCR After entering the chip bacteria populated and grew in 60 μm long, 7.5 μm wide
reactions were performed per sample - one targeting the genomic dxs gene and and 1.05 μm thick chambers connected to a supply channel flowing fresh media
another targeting the beta lactamase bla gene on the plasmid (primers). This pressurized to 4 psi34. On top of H media, carbenicillin and aTc, 0.1 g L−1 of bovine
defines the plasmid copy number as the number of plasmids per genome. After serum albumin was added to the media to improve evacuation of cells from the
generating droplets on the QX200 droplet generator (Biorad), reactions were supply channel55.
thermal cycled at 95 °C for 10 min, followed by 35 cycles of 98 °C for 30 s, 58 °C for Single cells were resolved by epi-fluorescence microscopy of sfGFP with a 100 x,
30 s, and 72 °C for 1 min. Samples were held at 4 °C after the reaction was com- 1.4 NA apochromat Leica objective. Single cell detection was performed using
pleted and before droplets were read. Droplets were read with the QX-200 Droplet customized Python scripts and OpenCV software (https://opencv.org). Tracking
Reader (Biorad) and concentrations were determined using the Quantasoft analysis enabled measurement of cell dimensions, intensity and growth rate overtime. Cells
pro software (Biorad). can be classified in three categories: (1) non dividing dark cells (loss of antibiotic
resistance due to plasmid loss), (2) healthy cells and (3) non growing bright cells
(likely due to plasmid overload). Whereas in bulk pathological cells of type (1) or
PCN sponge effects on CRISPRi inverter. CRISPR system with nuclease activity (3) would quickly be overgrown by healthy cells, in chambers, we find that they can
deactivated Cas protein(dCas) can be used to interfere with transcription regulation get stuck at the back or even populate the whole chamber, rendering the chamber
(CRISPRi) by design crRNA targeting at the promoter for the gene of interest unusable when the cells eventually die. This is particularly true for non dividing
which acts a repressor of gene. However, dCas/crRNA complex can also bind to dark cells at low aTc concentration. By using 7.5 μm wide chambers, we allow
other sites that share the same DNA sequence as the target but does not directly many cells to grow side by side, which increases the chances of healthy cells to
regulate the gene (these are called competitor sites). When the second scenario repopulate the chambers. However, this is not enough to maintain healthy
happens, it will have an impact on the repression level of the gene at a given chambers for more than a few hours at low aTc concentration. For aTc
concentration of dCas/crRNA, i.e., a sponge effect for the gene of interest by concentrations of 1 ng mL1 and above a majority of chambers last for the whole
consuming dCas/crRNA from the same pool. To test the sponge effect at different experiment time of 5000 min. We base intensity measurements on growing cells
numbers of competitor sites, a dual-plasmid system was used. The main plas- only (growth rate larger than 1/3 of mean growth rate) to prevent over
mid(Kanamycin resistant) is a low copy(pSC101) pTet-inducible CRISPRi- representation of non growing cells that can reach up to 40% of the detected cells
dFnCas12a inverter, where the crRNA transcribes by pTet target at a promoter that (Fig. 3F).
drivers a green fluorescence protein(sfGFP). The second plasmid is edited from the Due to the rapid loss of chambers we are unable to measure a stabilized
above-mentioned pUC19-PCN library by inserting a single competitor via standard fluorescence level for aTc concentration below 1 ng mL−1. However, we can still
Site-Directed Mutagenesis(NEB, E0554). Two plasmids were co-transformed into observe that the aTc concentration promotes plasmid replication. Indeed, the
the GL002 strain that has dFnCas12a and TetR integrated in its genome with Mix chamber survives longer at higher aTc concentration (plasmid loss is less frequent)
& Go! E.coli Transformation Kit(T3001). Five different PCN-plasmids were tested and the average fluorescence increases with aTc. For aTc concentration of 1 ng mL−1
with the same main inverter plasmid. The expression of sfGFP from the dual- to 100 ng mL−1 time dependent fluorescence and growth rate curves (Supplementary
plasmid system with different PCN was individually measured at a gradient of 20 Fig. 5) stabilize after about 600 min (i.e., 12 generations) and are expected to
aTc concentrations on the BioTek (Synergy H1) plate reader in H medium. Three accurately represent PCN in exponential growth condition.
biological replicates were used for each system, i.e., three different colonies were
picked up and grown in 2 mL H media without aTc for approximately 18 h. Then,
100 μL of overnight cell culture was used and diluted into 2 mL fresh H media Reporting summary. Further information on research design is available in the Nature
before transferring to a 96 well plate where each of the middle of 10 x 6 wells Research Reporting Summary linked to this article.
contains 1 μL of diluted cell culture and 200 μL fresh H media with corresponding
aTc concentration.
Data availability
Growth rate measurements. Growth rates of individual constructs were measured The raw sequencing data for this project have been deposited in the NIH BioProject
on the BioTek Synergy H1 plate reader. All plate reader experiments were per- Database under accession code PRJNA835501.
formed in VWR 96 well plates with a plastic lid, wells on the outer edge of the plate
were not used to avoid evaporation effects. Overnight cultures of cells grown to
saturation were serially diluted by a factor of 100,000 into well-mixed 200 μL
aliquots of media containing the appropriate antibiotics and inducer concentra- Code availability
tions. Cells were grown at 37 °C for 36 h while to OD600 and fluorescence (if Data analysis scripts are available at https://github.com/lambert-lab/plasmid-system-
appropriate) was recorded every 3 min. Cells were continuously shaken orbitally at tunable-copy-number. Source data for figure generation are provided in this repository.
240 rpm. Growth rates were calculated from the resultant time series data by
averaging a window of ±6 min around the maximal growth rate, calculated as a
discrete difference of OD600 readings between time points. Received: 21 July 2021; Accepted: 17 June 2022;

Gene expression measurements. The expression of green fluorescent protein

(sfGFP) was used as a measure of gene expression in this work. Fluorescence
measurements were taken on cells prepared as described above as they grew. The
mean value of the fluorescence at the point of maximal growth rate divided by the
OD at which the maximal growth rate was achieved was the reported fluorescence. References
Six replicates of each inducer concentration were taken and the average and 1. Lederberg, J. Cell genetics and hereditary symbiosis. Physiological Rev. 32, 403
standard deviation of this measure of gene expression were reported. Plasmid copy (1952).
numbers were estimated from this data in the mid-exponential phase using the 2. Wegrzyn, G. Replication of plasmids during bacterial response to amino acid
relation P ¼ Gμk , where μ is the growth rate, G is the sfGFP signal, and k is a
starvation. Plasmid 41, 1 (1999).
constant estimated by fitting the data to digital PCR measurements at the 3. Friehs, K. Plasmid copy number and plasmid stability, in New Trends and
endpoints. Developments in Biochemical Engineering, Advances in Biochemical

10 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0 ARTICLE

Engineering, edited by T., Scheper (Springer, Berlin, Heidelberg, 2004) 47–82 34. Lambert, G. & Kussell, E. Memory and fitness optimization of bacteria under
https://doi.org/10.1007/b12440. fluctuating environments. PLoS Genet 10, e1004556 (2014).
4. Silva, F., Passarinha, L., Sousa, F., Queiroz, J. A. & Domingues, F. C. Influence 35. Specht, D. A., Xu, Y. & Lambert, G. Massively parallel crispri assays reveal
of growth conditions on plasmid DNA production. J. Microbiol. Biotechnol. concealed thermodynamic determinants of dcas12a binding. Proc. Natl Acad.
19, 1408 (2009). Sci. 117, 11274 (2020).
5. Taylor, R. G., Walker, D. C. & McInnes, R. R. E. coli host strains significantly 36. Kang, C. W. et al. Synthetic auxotrophs for stable and tunable maintenance of
affect the quality of small scale plasmid DNA preparations used for plasmid copy number. Metab. Eng. 48, 121 (2018).
sequencing. Nucleic Acids Res. 21, 1677 (1993). 37. Lichstein, H. C. & Van De Sand, V. F. Violacein, an antibiotic pigment
6. Gonçalves, G. A. L., Bower, D. M., Prazeres, D. M. F., Monteiro, G. A. & produced by chromobacterium violaceum. J. Infect. Dis. 76, 47 (1945).
Prather, K. L. J. Rational engineering of escherichia coli strains for plasmid 38. Kholany, M. et al. Extraction and purification of violacein from yarrowia
biopharmaceutical manufacturing. Biotechnol. J. 7, 251 (2012). lipolytica cells using aqueous solutions of surfactants. J. Chem. Technol.
7. Wegrzyn, G. & Wegrzyn, A. Stress responses and replication of plasmids in Biotechnol. 95, 1126 (2019).
bacterial cells. Microb. Cell Factories 1, 2 (2002). 39. Gibson, D. G. et al. Enzymatic assembly of dna molecules up to several
8. Diaz Ricci, J. C. & Hernández, M. E. Plasmid effects on escherichia coli hundred kilobases. Nat. Methods 6, 343 (2009).
metabolism. Crit. Rev. Biotechnol. 20, 79 (2000). 40. Atkinson, M. R., Savageau, M. A., Myers, J. T. & Ninfa, A. J. Development of
9. Karim, A. S., Curran, K. A. & Alper, H. S. Characterization of plasmid burden genetic circuitry exhibiting toggle switch or oscillatory behavior in escherichia
and copy number in saccharomyces cerevisiae for optimization of metabolic coli. Cell 113, 597 (2003).
engineering applications. FEMS Yeast Res. 13, 107 (2013). 41. Mileyko, Y., Joh, R. I. & Weitz, J. S. Small-scale copy number variation and
10. Ow, D. S.-W., Nissom, P. M., Philp, R., Oh, S. K.-W. & Yap, M. G.-S. Global large-scale changes in gene expression. Proc. Natl Acad. Sci. 105, 16659 (2008).
transcriptional analysis of metabolic burden due to plasmid maintenance in 42. Wang, T., Tague, N., Whelan, S. A. & Dunlop, M. J. Programmable gene
Escherichia coli dh5αduring batch fermentation. Enzym. Microb. Technol. 39, regulation for metabolic engineering using decoy transcription factor binding
391 (2006). sites. Nucleic Acids Res. 49, 1163 (2021).
11. Bentley, W. E., Mirjalili, N., Andersen, D. C., Davis, R. H. & Kompala, D. S. 43. Wang, B., Guo, F., Dong, S.-H. & Zhao, H. Activation of silent biosynthetic
Plasmid-encoded protein: The principal factor in the “metabolic burden” gene clusters using transcription factor decoys. Nat. Chem. Biol. 15, 111
associated with recombinant bacteria. Biotechnol. Bioeng. 35, 668 (1990). (2019).
12. Pasini, M. et al. Using promoter libraries to reduce metabolic burden due to 44. Ali, M.Z. & Brewster, R. C. Controlling gene expression timing through gene
plasmid-encoded proteins in recombinant escherichia coli. N. Biotechnol. 33, regulatory architecture, bioRxiv, 2021.04.09.439163 (2021) https://doi.org/10.
78 (2016). 1101/2021.04.09.439163
13. Rozkov, A. et al. Characterization of the metabolic burden on escherichia coli 45. Wan, X., Pinto, F., Yu, L. & Wang, B. Synthetic protein-binding DNA sponge
dh1 cells imposed by the presence of a plasmid containing a gene therapy as a tool to tune gene expression and mitigate protein toxicity. Nat. Commun.
sequence. Biotechnol. Bioeng. 88, 909 (2004). 11, 5961 (2020).
14. Weinert, F. M., Brewster, R. C., Rydenfelt, M., Phillips, R. & Kegel, W. K. 46. Zhang, S. & Voigt, C. A. Engineered dcas9 with reduced toxicity in bacteria:
Scaling of gene expression with transcription-factor fugacity. Phys. Rev. Lett. Implications for genetic circuit design. Nucleic Acids Res. 46, 11115 (2018).
113, 258101 (2014). 47. Tan, C., Marguet, P. & You, L. Emergent bistability by a growth-modulating
15. Brewster, R. C. et al. The transcription factor titration effect dictates level of positive feedback circuit. Nat. Chem. Biol. 5, 842 (2009).
gene expression. Cell 156, 1312 (2014). 48. Qian, Y., Huang, H.-H., Jiménez, J. I. & Del Vecchio, D. Resource competition
16. Togna, A. P., Shuler, M. L. & Wilson, D. B. Effects of plasmid copy number shapes the response of genetic circuits. ACS Synth. Biol. 6, 1263 (2017).
and runaway plasmid replication on overproduction and excretion of 49. Ceroni, F., Algar, R., Stan, G.-B. & Ellis, T. Quantifying cellular capacity
lactamase from escherichia coli. Biotechnol. Prog. 9, 31 (1993). identifies gene expression designs with reduced burden. Nat. Methods 12, 415
17. Nordström, K. & Uhlin, B. E. Runaway-replication plasmids as tools to (2015).
produce large quantities of proteins from cloned genes in bacteria. Bio/ 50. Tomizawa, J. & Itoh, T. Plasmid cole1 incompatibility determined by
Technol. (Nat. Publ. Co.) 10, 661 (1992). interaction of RNA i with primer transcript. Proc. Natl Acad. Sci. 78, 6096
18. Dwidar, M. & Yokobayashi, Y. Riboswitch signal amplification by controlling (1981).
plasmid copy number. ACS Synth. Biol. 8, 245 (2019). 51. Rodríguez-Beltrán, J., DelaFuente, J., León-Sampedro, R., MacLean, R. C. &
19. Sheth, R. U., Yim, S. S., Wu, F. L. & Wang, H. H. Multiplex recording San Millán, A. Beyond horizontal gene transfer: the role of plasmids in
of cellular events over time on crispr biological tape. Science 358, 1457 bacterial evolution. Nat. Rev. Microbiol. 19, 347 (2021).
(2017). 52. Ilhan, J. et al. Segregational drift and the interplay between plasmid copy
20. Potvin-Trottier, L., Lord, N. D., Vinnicombe, G. & Paulsson, J. Synchronous number and evolvability. Mol. Biol. Evolution 36, 472 (2019).
long-term oscillations in a synthetic gene circuit. Nature 538, 514 (2016). 53. Vogwill, T. & MacLean, R. C. The genetic basis of the fitness costs of
21. Shao, B. et al. Single-cell measurement of plasmid copy number and promoter antimicrobial resistance: A meta-analysis approach. Evolut. Appl. 8, 284
activity. Nat. Commun. 12, 1475 (2021). (2015).
22. Panayotatos, N. Dna replication regulated by the priming promoter. Nucleic 54. Wein, T., Hülter, N. F., Mizrahi, I. & Dagan, T. Emergence of plasmid stability
Acids Res. 12, 2641 (1984). under non-selective conditions maintains antibiotic resistance. Nat. Commun.
23. Sektas, M. & Szybalski, W. Novel single-copy pETcocoTM vector with dual 10, 2595 (2019).
controls for amplification and expression. InNovations 14, 10 (2003). 55. Yang, D., Jennings, A. D., Borrego, E., Retterer, S. T. & Männik, J. Analysis of
24. Shetty, R. P., Endy, D. & Knight, T. F. Jr. Engineering BioBrick vectors from factors limiting bacterial growth in pdms mother machine devices. Front.
BioBrick parts. J Biol Eng. 2, 1611 (2008). Microbiol. 9, 871 (2018).
25. Camps, M. Modulation of cole1-like plasmid replication for recombinant gene
expression. Recent Pat. DNA gene sequences 4, 58 (2010).
26. del Solar, G., Giraldo, R., Ruiz-Echevarría, M. J., Espinosa, M. & Díaz-Orejas,
R. Replication and control of circular bacterial plasmids. Microbiol. Mol. Biol. Acknowledgements
Rev. 62, 434 (1998). We wish to thank David Specht for helpful discussions and critical review of the
27. Tomizawa, J.-i. Control of cole1 plasmid replication: Initial interaction of rna i manuscript. We also thank members of the Lambert laboratory for providing feedback
and the primer transcript is reversible. Cell 40, 527 (1985). on the manuscript. This work was supported by NIH funding under 1R35 GM133759
28. Paulsson, J., Nordström, K. & Ehrenberg, M. Requirements for rapid plasmid Maximizing Investigators’ Research Award (MIRA) awarded to G.L.
cole1 copy number adjustments: A mathematical model of inhibition modes
and rna turnover rates. Plasmid 39, 215 (1998).
29. Paulsson, J. & Ehrenberg, M. Noise in a minimal regulatory network: Plasmid
copy number control. Q. Rev. Biophysics 34, 1 (2001). Author contributions
30. Paulsson, J. Multileveled selection on plasmid replication. Genetics 161, 1373 M.V.R. and G.L. designed and conceived the study. M.V.R. and G.L. wrote the initial
(2002). manuscript. All authors edited and reviewed the manuscript. M.V.R., L.B.G.C., and Y.X.
31. Lin-Chao, S. & Bremer, H. Effect of the bacterial growth rate on replication performed experiments. M.V.R., L.B.G.C., Y.X, and G.L. analyzed data. G.L. supervised
control of plasmid pbr322 in escherichia coli. Mol. Gen. Genet. MGG 203, 143 work and acquired funding.
(1986).
32. Plotka, M., Wozniak, M. & Kaczorowski, T. Quantification of plasmid copy
number with single colour droplet digital pcr, PLoS ONE, 12, https://doi.org/
10.1371/journal.pone.0169846 (2017). Competing interests
The authors declare no competing interests.
33. Wang, P. et al. Robust growth of escherichia coli. Curr. Biol. 20, 1099 (2010).

NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31422-0

Additional information Open Access This article is licensed under a Creative Commons
Supplementary information The online version contains supplementary material Attribution 4.0 International License, which permits use, sharing,
available at https://doi.org/10.1038/s41467-022-31422-0. adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Correspondence and requests for materials should be addressed to Guillaume Lambert. Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
Peer review information Nature Communications thanks the anonymous reviewer(s) for indicated otherwise in a credit line to the material. If material is not included in the
their contribution to the peer review of this work article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
Reprints and permission information is available at http://www.nature.com/reprints the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s) 2022

12 NATURE COMMUNICATIONS | (2022)13:3908 | https://doi.org/10.1038/s41467-022-31422-0 | www.nature.com/naturecommunications