Engineering genetic circuit interactions within and between
synthetic minimal cells
Katarzyna P. Adamala
¥,1
, Daniel A. Martin-Alarcon
¥,2
, Katriona R. Guthrie-Honea
1
, and
Edward S. Boyden
*,1,2,3
1
Media Lab, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
2
Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge,
Massachusetts, USA
3
McGovern Institute for Brain Research, Department of Brain and Cognitive Sciences,
Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
Abstract
Genetic circuits and reaction cascades are of great importance for synthetic biology, biochemistry,
and bioengineering. An open question is how to maximize the modularity of their design to enable
the integration of different reaction networks and to optimize their scalability and flexibility. One
option is encapsulation within liposomes which enables chemical reactions to proceed in well-
isolated environments. Here we adapt liposome encapsulation to enable the modular, controlled
compartmentalization of genetic circuits and cascades. We demonstrate that it is possible to
engineer genetic circuit-containing synthetic minimal cells (synells) to contain multiple-part
genetic cascades, and that these cascades can be controlled by external signals as well as inter-
liposomal communication without cross-talk. We also show that liposomes containing different
cascades can be fused in a controlled way so that the products of incompatible reactions can be
brought together. Synells thus enable more modular creation of synthetic biology cascades, an
essential step towards their ultimate programmability.
Introduction
Chemical systems capable of performing biochemical reactions in the absence of live cells
have been extensively used in research and industry to study and model biological
processes
1,2
, to produce small molecules
3,4
, to engineer proteins
5,6
, to characterize RNAs
7
,
as biosensors
8,9
and molecular diagnostic tools
10
, and to extend the sensing abilities of
natural cells
11
. Organisms from all three domains of life have been used to obtain
transcription/translation (aka TX/TL) extracts for cell-free production of biochemical
Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research,
subject always to the full Conditions of use: http://www.nature.com/authors/editorial_policies/license.html#terms
*
Correspondence and request for materials should be addressed to E.S.B. esb@media.mit.edu.
¥
These authors contributed equally to this work
Author contributions
KPA and DAM-A contributed equally to this work. KPA, DAM-A and KRG-H performed experiments. KPA, DAM-A and ESB
analyzed data and wrote the manuscript.
HHS Public Access
Author manuscript
Nat Chem
. Author manuscript; available in PMC 2017 May 14.
Published in final edited form as:
Nat Chem
. 2017 May ; 9(5): 431–439. doi:10.1038/nchem.2644.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
products from genetic codes
12
. Encapsulating cell-free TX/TL extracts into liposomes
creates bioreactors often referred to as synthetic minimal cells (SMCs or synells)
13–16
.
Although synells have been used to make functional proteins using encapsulated systems
reconstituted from recombinant cell-free translation factors
17–19
, as well as cell-free extracts
from bacterial
6,20
and eukaryotic cells
21
, work on liposomal synells has so far focused on
expression of single genes, with the goal of synthesizing a single gene product, and within a
homogenous population of liposomes.
Here, we confront a key issue in synthetic biology: the modularity of multi-component
genetic circuits and cascades. We show that by encapsulating genetic circuits and cascades
within synells (Figs. 1a and 1b) and orchestrating the synells to either operate in parallel
(Fig. 1c), communicate with one another (Fig. 1d), or fuse with one another in a controlled
way (Fig. 1e). We can create genetic cascades that take advantage of the modularity enabled
by liposomal compartmentalization. Thus, our strategy enables genetic cascades to proceed
in well-isolated environments while permitting the desired degree of control and
communication. We present design strategies for constructing and utilizing such synell
networks, thus expanding the utility of liposome technology and improving the modularity
of synthetic biology. Synell networks may support complex chemical reactions that would
benefit from both the high-fidelity isolation of multiple reactions from one another, as well
as controlled communication and regulatory signal exchange between those reactions. We
show, for example, the controlled fusion of two populations of synells that contain
mammalian transcriptional and mammalian translational machinery, respectively, which are
normally incompatible when combined in the same compartment.
Results
Confinement of genetic circuits in liposomes
Before exploring the control of, and communication with, synells containing genetic
cascades, we first characterized the basic structural and functional properties of individual
synells. To characterize the size and functionality of our liposomes, we labeled liposome
membranes with red dye (rhodamine functionalized with a lipid tail) and filled the liposomes
with cell-free transcription/translation (TX/TL) extract derived from HeLa cells
22–25
, as well
as DNA encoding either GFP or split GFP. Structured illumination microscopy (SIM)
images showed that GFP liposomes had a diameter between 100 nm and 1 μm (Fig. 2a), a
measurement that we confirmed with dynamic light scattering (Fig. S1). We used flow
cytometry to quantify the functional expression of genes by synells; 68.4% of the GFP
liposomes expressed fluorescence, along with 61.8% of those encapsulating split GFP (Figs.
2b – 2d; for control flow cytometry experiments, see Fig. S2). We characterized the
enzymatic activity of several reporters in our liposomes (Fig. S3) and used a Western blot to
provide an additional non-enzymatic characterization of luciferase expression (Fig. S4). We
compared the performance of mammalian (HeLa) and bacterial (
E. coli
) TX/TL systems in
our liposomes, finding the mammalian system to be slower and have a lower protein yield
(Fig. S5).
Having established that the liposomes were of proper size and functionality, we next sought
to verify that a well-known advantage of liposomal compartmentalization—facilitated
Adamala et al.
Page 2
Nat Chem
. Author manuscript; available in PMC 2017 May 14.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
reaction efficacy due to molecular confinement (since encapsulating reactants within a
liposome facilitates their interaction due to the small volume)
26–29
—can help support multi-
component genetic circuits as well as chemical reactions of higher order. We compared cell-
free transcription/translation (TX/TL) reactions that produce firefly luciferase (fLuc) from
one, two, or three protein components, testing them in bulk solution vs. synells. In this
experiment, we used HeLa cell extract constitutively expressing the Tet protein to mediate
small-molecule induction of transcription of the one, two, or three fLuc components, as well
as alpha-hemolysin (aHL), which serves as a pore to admit doxycycline (Dox) to trigger Tet
function
20,30,31
. The one-component luciferase was simply conventional monolithic fLuc
(Fig. 3a); the two-component system (i.e., to explore 2
nd
-order reactions) comprised the two
halves of split firefly luciferase, each attached to a coiled coil and a split intein fragment to
bring the halves together and covalently bridge them (Fig. 3b)
32
; and the three-component
system involved the halves of split firefly luciferase bearing coiled coils and split inteins,
with the coiled coils targeting a third protein, a scaffold (Fig. 3c)
32
.
For all three orders of luciferase-producing reactions, the effect of dilution on fLuc
expression was weaker for liposomes than for bulk solution (Figs. 3d – 3f; P < 0.0001 for
interaction between factors of encapsulation and dilution factor; ANOVA with factors of
encapsulation and dilution factor; see Tables S1 – S3 for full statistics and Fig. S6 for
corresponding experiments under the control of a constitutive P70 promoter). As expected,
fLuc expression was proportional to the concentration of Dox added to the external solution,
and depended on aHL (Figs. 3g – 3i show end-point expression after 3 h; see Fig. S7 for the
corresponding expression at a 1 h end-point, and Figs. S8 – S10 for the same reactions in
bulk solution). Liposomes produced lower amounts of fLuc than the same volume of TX/TL
extract in bulk solution—likely due to the well-known property of stochastic loading of
reagents into liposomes
27,28
(P < 0.0001 for factor of encapsulation in ANOVA with factors
of time, encapsulation, and order; see Table S4 for full statistics). For the third-order
reaction, we found that liposome encapsulation resulted in efficacy nearly equal to that of
bulk solution (P = 0.1324 for factor of encapsulation in ANOVA with factors of time and
encapsulation; Fig. 3l; see Table S7 for full statistics), whereas for the first-order and
second-order reactions the liposomes resulted in lower efficacy (P < 0.0001 for factor of
encapsulation in ANOVAs for both analyses, each with factors of time and encapsulation;
Figs. 3j and 3k; see Tables S5 and S6 for full statistics). Molecular confinement in liposomes
thus may help facilitate higher-order reactions that require multiple chemical building blocks
to be brought together, since the restricted movement of reagents increases the probability of
the requisite multi-way interactions.
Insulation of genetic circuits operating in parallel liposome populations
As a next step towards engineering sets of liposomes that can communicate with one
another, we set out to determine whether liposomes could be used to insulate multiple and
potentially incompatible genetic circuits from each other, so that they could operate in the
same bulk environment. This insulation would enable modular design; each circuit could be
optimized independently and deployed in the same environment as other circuits without
interference. These circuits could reuse the same parts (proteins, DNA) for different
purposes in different liposomes, thereby circumventing one limitation of genetic circuits
Adamala et al.
Page 3
Nat Chem
. Author manuscript; available in PMC 2017 May 14.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
designed to all operate within the same living cell (where one must assume that all circuit
elements might encounter each other and must therefore be inherently orthogonal). Different
liposome populations could also contain chemical micro-environments that are not mutually
compatible (e.g., bacterial and mammalian extracts, or mammalian transcriptional and
mammalian translational machinery)—there are numerous examples throughout chemistry
of reactions being run under specialized, and thus often isolated, reaction conditions
33
.
We first assessed whether multiple liposomal circuits could operate in parallel without
crosstalk. To do this, we created populations of liposomes that could respond differently to
the same external activator. We built two populations of liposomes carrying mammalian
TX/TL extract and the same amount of Dox-inducible luciferase DNA (either Renilla or
firefly luciferase), but varied the amount of alpha-hemolysin DNA to result in high-aHL and
low-aHL synell populations (Fig. 4a). High-aHL and low-aHL synells responded to the non-
membrane-permeable Dox in the external solution, doing so proportionally to their own aHL
concentration (Fig. 4b). We observed no evidence that doxycycline acting upon one
liposome population affected expression of luciferase in the other population: specifically,
there was no significant difference in fLuc expression in high-aHL fLuc liposomes when the
rLuc liposomes were high-aHL vs. low-aHL, and the same held for the other combinations
(Fig. 4b; Sidak’s multiple comparisons test after ANOVA with factors of luciferase type and
alpha-hemolysin combination; see Table S8 for full statistics, and Figs. S11 and S12 for
rLuc and fLuc expression data at different aHL plasmid concentrations, for two different
time points). That is, luciferase expression from each liposome population depended only on
the amount of aHL DNA present in that population, and not on that of the other population
(Figs. 4c – 4e). This experiment thus not only verifies the independent operation of multiple
non-interacting liposomes, but also verifies that multiple liposome populations can be
programmed in advance to have varying response levels to a given trigger, and subsequently
in the same internal solution, triggered to function simultaneously.
Communication between genetic circuits operating in multiple liposome populations
Having established that genetic circuits in separate populations of liposomes could operate
independently, we next sought to begin to create controlled communication pathways
between populations of synells. In this way we could create a compartmentalized genetic
circuit—which as noted above may need to be separated from others for reasons of control
fidelity, toxicity, or reagent tunability—and connect it to other compartmentalized circuits.
While previous works have emphasized the importance of modularity in genetic circuits
34
,
to our knowledge nobody has approached the problem by physically separating circuit
elements into different liposomes. We built two-component circuits by mixing together two
populations of liposomes, a “sensor” that senses an external small molecule cue and a
“reporter” that receives a message from the sensor population and produces an output; we
could vary the occupancy of each population to achieve a different overall ratio of the two
components (Fig. 5a; see Fig. S13 for additional characterization of the membrane-
permeable small molecules used throughout this figure, and Tables S9 – S10 for the
associated statistics). Our first version was built with bacterial TX/TL extract (Fig. 5b). The
sensor liposomes contained IPTG (a small, non-membrane-permeable activator that induces
the
lac
promoter) and the arabinose-inducible gene for aHL (arabinose, which unlike IPTG,
Adamala et al.
Page 4
Nat Chem
. Author manuscript; available in PMC 2017 May 14.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
is membrane-permeable); these liposomes thus sensed arabinose and released IPTG by
expressing aHL channels. We combined these with reporter liposomes containing
constitutively-expressed aHL, in which fLuc was under the control of the
lac
promoter—
either directly (fLuc under
lac
promoter) or indirectly (T7RNAP under the
lac
promoter and
fLuc under T7 promoter)—and found that multi-component compartmentalized genetic
circuits thus constructed were able to operate as coherent wholes. We tested both systems
with multiple dilutions of sensor and reporter liposomes, and found similar dose-response
curves from titration of either species of liposome (Figs. 5c and 5d; bars in these panels
represent final time points of 6 h; for the complete time series that includes the data in Fig.
5c, see Fig. S14; for the end-point expression of the circuit in Fig. 5c without arabinose
triggering, see Fig. S15; for the complete time series that includes the data in Fig. 5d, see
Fig. S16; for the end-point expression of the circuit in Fig. 5d without arabinose triggering,
see Fig. S17). Using this modular architecture, we constructed a genetic circuit that
combines both bacterial and mammalian components (Fig. 5e). The sensor liposome in this
case responded to theophylline (membrane-permeable) to release doxycycline (non-
membrane-permeable). Dox, in turn, activated fLuc expression in reporter liposomes built
with mammalian components. As before, we showed that the multi-compartment genetic
cascade could function as designed, with fLuc expression dose-response curves similar upon
titrating either sensor or reporter liposome concentration (Fig. 5f; bars in this panel represent
final time points of 6 h; for the complete time series that includes the data in Fig. 5f, see Fig.
S18; for the end-point expression of the circuit in Fig. 5f without theophylline triggering, see
Fig. S19). Thus, even multi-component genetic circuits with different chemical micro-
environments (e.g., made from bacterial vs. mammalian cell extracts) can be assembled into
coherent networks comprising multiple modules.
Fusion of complementary genetic circuits
Finally, having established that it is possible to maintain liposomes in high-integrity states
despite being mixed, we sought to engineer synells to fuse so that they could bring together
two genetic cascades into the same environment in a programmable fashion. Two precursors
might require synthesis in different milieus, but ultimately need to be reacted with one
another. One prominent example is that of mammalian transcription and translation. Mixed
mammalian transcription and translation cell-free extracts are not able to functionally result
in transcription of DNA into RNA and then the translation of RNA into protein, perhaps
because the micro-environments of the mammalian nucleus and cytoplasm are quite
different, making their cell-free extracts incompatible (Fig. S20). Rather than mixing the two
cell-free extracts into a single non-functioning mixture, it might be preferred to use synells
to compartmentalize the reactions. Once nuclear-extract synells have completed
transcription, it might be desirable to fuse them with cytoplasmic-extract synells for
translation to take place.
Thus, we sought to make liposomes capable of controlled fusion (Fig. 6a). Fusing liposomes
of opposite charge was previously demonstrated to activate gene expression in liposomes
35
.
Our system uses only one kind of membrane composition (POPC cholesterol membranes,
known to be a good environment for membrane channels like aHL), so to achieve fusion
between liposomes we used SNARE/coiled-coil hybrid proteins (here called SNAREs for
Adamala et al.
Page 5
Nat Chem
. Author manuscript; available in PMC 2017 May 14.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript