Research Article Cite This: ACS Synth. Biol. XXXX, XXX, XXX−XXX
pubs.acs.org/synthbio
Plug-and-Play Multicellular Circuits with Time-Dependent Dynamic Responses Arturo Urrios,†,§ Eva Gonzalez-Flo,‡,§ David Canadell,† Eulàlia de Nadal,† Javier Macia,*,‡ and Francesc Posas*,† †
Cell Signaling Research Group, ‡Synthetic Biology for Biomedical Applications Group, Departament de Ciencies Experimentals i de la Salut, Universitat Pompeu Fabra (UPF), E-08003 Barcelona, Spain S Supporting Information *
ABSTRACT: Synthetic biology studies aim to develop cellular devices for biomedical applications. These devices, based on living instead of electronic or electromechanic technology, might provide alternative treatments for a wide range of diseases. However, the feasibility of these devices depends, in many cases, on complex genetic circuits that must fulfill physiological requirements. In this work, we explored the potential of multicellular architectures to act as an alternative to complex circuits for implementation of new devices. As a proof of concept, we developed specific circuits for insulin or glucagon production in response to different glucose levels. Here, we show that fundamental features, such as circuit’s affinity or sensitivity, are dependent on the specific configuration of the multicellular consortia, providing a method for tuning these properties without genetic engineering. As an example, we have designed and built circuits with an incoherent feed-forward loop architecture (FFL) that can be easily adjusted to generate single pulse responses. Our results might serve as a blueprint for future development of cellular devices for glycemia regulation in diabetic patients. KEYWORDS: synthetic biology, biological computation, multicellular circuits, dynamic responses, diabetes
S
In this study, we explored the development of cellular devices with complex dynamic responses based on a multicellular architecture. The devices that we designed and assessed contained several cell types that were minimally engineered to respond in the presence of a single signal according to a simple logic: Identity (ID), where the cellular response was produced only in the presence of a trigger signal; or NOT, where the cell responded only in the absence of the signal. We hypothesized that these minimal multicellular devices would be simpler, and more adaptive and tunable than single cell systems. We explored how external modulation of the communications between cells modified circuit properties in terms of affinity and sensitivity, and how complex time-dependent dynamics could be implemented without introducing additional genetic changes into the cells. As a proof of concept, we used yeast as a model organism, because yeast is an excellent workbench in which to explore different theoretical designs. Nevertheless, it should be noted that most gene circuits that have been built in yeast are often fully functional in mammalian cells after minor refinements.16 This adaptability of genetic circuits across different organisms allows the design of circuits in silico, their characterization in yeast and finally their reimplementation in mammalian cells.17 Several synthetic circuits devoted to glucose homeostasis regulation based on genetically engineered cells have been recently developed with very promising results.18−20 Here, we developed devices based on multicellular consortia able to
ynthetic biology studies have the potential to create biologically engineered devices for biomedical applications. The use of biological devices opens the door to the development of new and more efficient alternative treatments for a wide range of health problems.1−4 In general, biological devices rely on a synthetic genetic circuit composed of basic heterologous control components that fine-tune transgene expression in response to particular exogenous or endogenous signals.5 However, medical applications require complex responses that occur with high precision and specific dynamics and that have a predictable logic in response to external or internal signals. As a consequence, the building of high-order networks, including sequential6 or feedback control7,8 systems, is still a major challenge in synthetic biology due to the emergence of additional constraints associated with the complexity of genetic circuits that compromise their viability.9 An alternative to the use of single cell circuits for biological computation is the use of multicellular consortia. In general, multicellular devices are composed of different cell types that sense input signals and produce biochemical outputs thanks to a cell−cell communication network. Multicellular circuits have been shown to be useful for implementing complex logic circuits10−12 or memory devices13 in addition to overcoming strong constraints such as the so-called wiring problem14 or component reusability. Multicellularity provides multiple advantages, for example, noise reduction or simplification of the genetic engineering needed in every single cell, thus minimizing interference with the physiology of the host cells.15 However, this approach has not been systematically applied either in analog or in time-dependent circuits. © XXXX American Chemical Society
Received: December 27, 2017 Published: March 27, 2018 A
DOI: 10.1021/acssynbio.7b00463 ACS Synth. Biol. XXXX, XXX, XXX−XXX
Research Article
ACS Synthetic Biology
Figure 1. Design and characterization of glucose sensor cells. (a) MATα W303 cells were modified to express αSc under the control of different HXT promoters: HXT1, HXT2, HXT3, HXT4 and HXT7. Cells were incubated together with αSc:GFP reporter cells for 4 h in three different glucose concentrations (0,5%, 2% or 5%). GFP fluorescence (a.u.) of αSc:GFP cells was analyzed using flow cytometry. Data are shown as means ± SD from three independent experiments. (b) Cells containing two copies of HXT1-αSc (HXT1:αSc cell) or HXT7-αSc (HXT7:αSc cell) were incubated together with αSc:GFP cells for 4 h in the presence of different concentrations of glucose (0.1%, 0.5%, 1%, 2%, 3% or 5%). GFP fluorescence was analyzed as in (a). Data are shown as means ± SD from three independent experiments. (c) HXT1:αSc and HXT7:αSc cells were incubated in different concentrations of glucose (0.1%, 0.5%, 1%, 2%, 3% or 5%) for 2 h. The cell supernatants were then collected and the cells were washed, rediluted in fresh media containing the above different glucose concentrations and incubated for a further 2 h, following which the cell supernatants were again collected. All supernatants were analyzed by incubation with αSc:GFP cells for 4 h and subsequent measurement of GFP using flow cytometry. Data are shown as mean from three independent experiments.
different cell types: (i) sensor cells that discriminate between high or low glucose levels and secrete a wiring molecule, (ii) effector cells that produce an output, i.e., insulin or glucagon, in response to the presence of the wiring molecule in the medium, and (iii) modulator cells that, in response to an external signal, modify the communication between sensor and effector cells by altering the levels of the wiring molecule (see Table S1 for details). In order to build a set of sensor cells that respond to a range of glucose levels, we exploited the natural capabilities of yeast to sense environmental glucose levels and to adjust the expression of genes encoding hexose transporters (HXT) to maximize glucose uptake.21,22 We made several genetic constructs in which the expression of the α-factor (pheromone) from Saccharomyces cerevisiae (αSc) was under the control of different HXT promoters (i.e., HXT1, HXT2, HXT3, HXT4 and HXT7). This αSc is the wiring molecule that permits communication between different cells in the consortia. We engineered different strains that carried different HXT promoters and that secreted αSc in response to different glucose concentrations. The αSc level was indirectly quantified
discriminate between different glucose levels in the extracellular environment and to secrete either insulin, at high glucose levels, or glucagon, at low glucose levels. Furthermore, we implemented circuits, using yeast as a model organism, that in response to a physiological signal respond synthetizing and secreting a molecule of interest according to two different dynamics: time-independent or single pulse response. It is worth mentioning that devices with this architecture could be potentially applied to different biomedical scenarios, such as the maintenance of glucose homeostasis in diabetic patients.
■
RESULTS The Use of Hexose Transporter Promoters Serves To Create Cells Able To Detect and Respond to Different Extracellular Glucose Levels. Glycemia regulation in healthy mammalian organisms relies on the balance between insulin and glucagon secretion in response to different glucose levels. In order to mimic this natural behavior in a multicellular synthetic device, we considered two minimal circuits, one responsible for insulin secretion and the other for glucagon secretion. We designed a library of cells containing three B
DOI: 10.1021/acssynbio.7b00463 ACS Synth. Biol. XXXX, XXX, XXX−XXX
Research Article
ACS Synthetic Biology
Figure 2. Characterization of αSc:GFP, αSc:INS and αSc:GCG cells. (a) αSc:GFP cells were incubated with different amounts of αSc (0.01, 0.1, 0.5, 1, 2, 5, 10, or 20 nM) for 4 h. GFP fluorescence (a.u.) was then analyzed using flow cytometry. Data are shown as means ± SD from three independent experiments. (b,c) αSc:INS cells (b) and αSc:GCG cells (c) were incubated with different amounts of αSc (0, 1, 2, 5, or 10 nM) for 1 h. Cell supernatants were then collected and insulin (b) or glucagon (c) levels were analyzed using ELISA. Data are shown as means ± SD from three independent experiments.
by means of a reporter cell that sensed αSc and subsequently expressed a green fluorescent protein (GFP) under the control of the pheromone inducible FUS1 promoter (FUS1::GFP). Cells containing different HXT promoters fused to αSc synthetic constructs were incubated together with FUS1::GFP cells in the presence of one of several extracellular glucose levels (0.5%, 2% or 5%) for 4 h to ensure a robust GFP signal. GFPfluorescence was then analyzed using flow cytometry. Cells containing the HXT1 construct increased the production of αSc in response to an increase in glucose levels, behaving as a high glucose sensor (ID logic). Cells containing HXT3 displayed αSc production that was independent of the extracellular glucose levels. In contrast, cells containing either HXT2, HXT4 or HXT7 acted as glucose repressible systems, since αSc production was repressed in response to an increase in the concentration of glucose (NOT logic) (Figure 1a). These behaviors are consistent with previously published studies.25 HXT1 and HXT7 were selected to build in vivo glucose sensors able to discriminate between high and low glucose levels. To improve the efficiency of the secreting cells (HXT1 and HXT7) two copies of an HXT:αSc construct were introduced into each cell to create HXT1:αSc or HXT7:αSc sensor cells. We then assessed secretion of the wiring molecule (referred to as the transfer function) by these new cells by incubating them in the presence of different glucose levels (0.1%, 0.5%, 1%, 2% or 5%) together with the FUS1::GFP reporter cells for 4 h (Figure 1b). Additionally, we tested whether the sensor cells retained memory of previous glucose states or not. HXT1:αSc or HXT7:αSc cells were grown in the presence of different glucose concentrations (0.1%, 0.5%, 1%, 2%, 3% or 5%) for 2 h.
The cell supernatants were then collected, and the cells were washed and incubated with the above different glucose levels again for an additional 2 h. αSc:GFP cells were incubated with the cell supernatants and GFP-fluorescence was then assessed using flow cytometry. Neither of the glucose sensors displayed a strong dependence on the previous glucose states although some minor differences were observed specially between extreme conditions. These results pointed out that αSc production by these cells mainly depends on the extracellular glucose concentration and not so much on their previous glucose state (Figure 1c, Figure S1). Additionally, we observed that glucose slightly altered the expression of the GFP from the FUS1::GFP reporter when stimulated with synthetic αSc, however these changes were attributed mainly to the effect of glucose on the fluorescence of the GFP (Figure S2a). Thus, the use of HXT promoters served to create sensor cells that responded to different glucose levels. Implementation of Different Minimal Circuits To Produce Hormones in a Glucose-Dependent Manner. To create a circuit able to produce different hormones in response to the presence of the wiring molecule from the sensor cells, we removed GFP from the reporter cell and replaced it with either insulin (αSc:INS) or glucagon (αSc:GCG) under the control of the FUS1 promoter (Figure 2a). The αSc:INS cell contained a construct (PFUS1-αINS) that expressed a modified version of an insulin analogue precursor (IAP) with a short C-chain (EWK) fused to the preproleader sequence of α-factor for efficient secretion in yeast,23 under the FUS1 promoter. The αSc:GCG cell expressed a modified version of glucagon fused to the preproleader sequence of αC
DOI: 10.1021/acssynbio.7b00463 ACS Synth. Biol. XXXX, XXX, XXX−XXX
Research Article
ACS Synthetic Biology
Figure 3. Design and implementation of multicellular consortia for glucose-responsive devices. (a) Minimal circuits for glucose-mediated insulin and glucagon production. One circuit consisted of HXT1:αSc cells incubated together with αSc:INS cells, using αSc as a wiring molecule. Similarly, a second circuit combined HXT7:αSc cells with αSc:GCG cells using αSc as a wiring molecule. Simultaneous use of both circuits required physical segregation to prevent crosstalk. (b) The circuits produced insulin and glucagon in response to glucose. Both circuits were exposed to different glucose levels (0.5%, 1%, 2%, 3% or 5%) for 1 h. Cell supernatants were collected, and secreted hormones were detected using ELISA. Data are shown as means ± SD from three independent experiments. (c) Both consortia were initially exposed to 5% glucose and cell supernatants were taken every 30 min for 120 min. The cells were then diluted, shifted to media containing 0.5% glucose, and supernatants were taken every 30 min over a further 120 min incubation. The cells were then diluted again, shifted back to medium containing 5% glucose for another 120 min incubation and subsequently shifted to medium containing 0.5% glucose for a further 120 min incubation. Supernatant samples were taken as above. Hormones in the supernatants were assessed using the corresponding ELISA kit. Data are shown as means ± SD from three independent experiments.
factor for efficient secretion,24 also under the FUS1 promoter. For their characterization, αSc:INS and αSc:GCG cells were incubated with different amounts of αSc for 1 h and insulin and glucagon levels in the supernatant were then respectively quantified using specific ELISA kits. In the absence of αSc, neither insulin nor glucagon was detected, whereas, upon addition of αSc, cells produced and secreted insulin or glucagon that accumulated in the media (Figure 2b,c). Moreover, when these cells were shifted from a media containing the wiring molecule into new media without αSc, they ceased to secrete insulin or glucagon unless αSc was readded to the medium. Hormone secretion by these cells therefore did not depend on previous exposure to extracellular αSc, but depended solely on the presence of currently available extracellular αSc. In this setup, glucagon and insulin levels were detectable after 30 min of induction (Figure 3). Moreover, we did not observe an effect of the extracellular glucose in the ability of αSc:INS to produce insulin when stimulated with synthetic αSc (Figure S2b). These results suggest that the αSc signal can be transformed into different biological outputs depending on the presence of different effector cells in the cell culture. To build a circuit that produces insulin in response to high glucose levels, HXT1:αSc and αSc:INS cells were cultured
together. Analogously, for glucagon production at low glucose levels, HXT7:αSc and αSc:GCG cells were combined in a second multicellular consortium (Figure 3a). Both consortia were incubated with different glucose levels for 1 h, and insulin and glucagon production were then measured. Insulin was produced at high glucose levels (>2%) with a response that resembled that of HXT1-induced αSc secretion (HXT1:αSc transfer function), while glucagon was produced at low glucose concentrations (
