A synthetic gene circuit for imaging-free detection of dynamic cell signaling

Cells employ intracellular signaling pathways to sense and respond to changes in their external environment. In recent years, live-cell biosensors have revealed complex pulsatile dynamics in many pathways, but studies of these signaling dynamics are limited by the necessity of live-cell imaging at high spatiotemporal resolution1. Here, we describe an approach to infer pulsatile signaling dynamics from just a single measurement in fixed cells using a pulse-detecting gene circuit. We computationally screened for circuit with pulse detecting capability, revealing an incoherent feedforward topology that robustly performs this computation. We then implemented the motif experimentally for the Erk signaling pathway using a single engineered transcription factor and fluorescent protein reporter. Our ‘recorder of Erk activity dynamics’ (READer) responds sensitively to both spontaneous and stimulus-driven Erk pulses. READer circuits thus open the door to permanently labeling transient, dynamic cell populations to elucidate the mechanistic underpinnings and biological consequences of signaling dynamics.


49
Many cell signaling pathways exhibit pulses, oscillations or even traveling waves of pathway 50 activity. Examples include the signaling pulses observed from the tumor suppressor p53, the 51 mitogen associated protein kinase (MAPK) Erk, and the immune signaling transcription factor 52 NF-κB 1-5 . Pulses of Erk activity have been observed in vivo in the early mouse embryo 6,7 and in 53 tumors 8 , and self-organize into propagating waves from sites of epithelial injury in both mouse 9 54 and zebrafish 10 . The breadth of biological systems exhibiting signaling dynamics suggests that 55 they may play important functional roles. Yet in nearly every context, dynamics are studied 56 exclusively using time-lapse microscopy in single living cells. This granularity of measurement 57 is crucial: to determine whether a cell has pulsed, one must perform at least three measurements 58 to observe a succession of low, high, and low signaling states. However, live imaging can be a 59 severe constraint, limiting the throughput of chemical and genetic screens and restricting in vivo 60 studies to tissues that are compatible with single-cell imaging. We thus set out to explore 61 whether we might be able to construct simple synthetic gene circuits to label pulsing cells 62 without live imaging ( Figure 1A). 63 64 Our first goal was thus to identify circuit topologies that might serve as pulse detectors, 65 selectively responding to dynamics while filtering out and ignoring constant high or low 66 signaling states. We focused our attention on feedforward loops (FFLs), a class of network 67 topologies that repeatedly arise in generating or processing dynamic information [11][12][13][14][15] . FFLs are 68 either coherent or incoherent based on whether the two paths connecting input and output have 69 the same or different signs ( Figure 1B). We devised a simple, modular 2-equation model to 70 represent all 8 FFLs with AND logic at the output node 12 ( Figure S1, Supplementary  71 Information), and in each case simulated 10,000 random parameter sets with 3 input conditions: 73 sustained on, sustained off and a pulse of activation. 74

75
We assessed the performance of each circuit by calculating integrated output over time in 76 response to each input. We then plotted the ratio of the pulse-induced response to both the 77 constant-on and constant-off cases ( Figure 1C). By definition, a pulse detector circuit should 78 show stronger induction in response to a pulse than either constant stimulus, leading to high 79 values of both ratios and enrichment in the upper-right quadrant of such a plot, whereas simple 80 activators (circuits that induce gene expression in proportion to the quantity of input signal) 81  Figure 1C). Analysis of all 8 FFL topologies revealed that only a single topology, 83 Circuit 7, was capable of performing pulse detection (Figure 1D, left). Pulse detection also 84 appeared to be a robust feature of the Circuit 7 FFL, with 96% of simulations showing a stronger 85 response to pulsed stimuli than either high or low constant inputs (Figure 1D, right). We also 86 tested all 8 FFL topologies with OR logic at the output node ( Figure S2). While none exhibited 87 pulse-specific activation, one OR-FFL circuit did exhibit pulse-specific repression and can be 88 understood as the logical inverse of our pulse-detecting Circuit 7 FFL (see Supplementary 89 Information; Figure S2B). 90

91
Examining simulation trajectories provided further insight into the operation of the Circuit 7 92 FFL ( Figure 1E, Figure S3). Application of a stimulus ("input") rapidly results in production of 93 an intermediate node (x1), but also blocks the ability for x1 to activate an output node (x2). Only 94 upon removal of the stimulus is repression relieved, enabling x1 to trigger output. Constant-on 95 inputs are unable to trigger a response because input permanently blocks output, whereas 96 constant-off inputs fail because the essential activator x1 is not produced. Overall, our 97 simulations reveal an intuitive and logical relationship between the Circuit 7 FFL topology and 98 pulse detection, demonstrating that pulse detection can arise quite generally out of this particular 99 FFL architecture. 100 101 We next set out to implement our pulse detector circuit in the context of a dynamic signaling 102 pathway in mammalian cells, the Erk pathway. Our implementation centered around a single 103 synthetic transcription factor that is regulated by Erk in two opposing ways (Figure 2A). For the 104 (2) under constant ON stimuli, KGV is expressed but exported from the nucleus, preventing GFP production; (3) under pulsed stimuli, KGV is first expressed and then imported into the nucleus, leading to high GFP expression. (c) Images of representative fields of NIH3T3 READer cells exposed to constant serum or a 1 h serum pulse. (d) Flow cytometry distribution of GFP levels in cells expressing READer (green) incubated in growth factor free media (constant OFF), 10% serum (constant ON) or a one-hour pulse of 10% serum; wild-type NIH3T3s are shown in gray. (e) Quantification of flow cytometry data shows % of GFP-high cells in all three conditions. (f-g) Mapping how pulse duration affects READer circuit output. Serum inputs of varying duration were applied to cells, which were fixed 3 h after the end of the pulse (schematic in f) and analyzed by flow cytometry for GFP induction (data in g). (h) A extended mathematical model of the READer circuit incorporating previously measured negative feedback on Erk target gene induction. An input u (gold) stimulates intermediate node x 1 (blue) , which produces a negative regulator x 3 (purple) that inhibits the production of x 1 . (i) Quantification of flow cytometry data in g reveals that pulses between 5-120 min result in potent GFP accumulation. Inset shows simulated results from the model from h, with (green) or without (grey) negative feedback. two-step transcriptional cascade: an Erk-responsive promoter to drive expression of a synthetic 107 Gal4-VP64 transcription factor, which then induces GFP expression from a Gal4-responsive 108 UAS promoter. To match the Circuit 7 FFL topology, our synthetic transcription factor must also 109 be rapidly and reversibly inhibited by Erk (so that the Erk input also directly inhibits x2 110 production). We realized that fusion with an Erk "kinase translocation reporter" (ErkKTR) 111 would be ideal for implementing this stimulus-dependent inhibition of the engineered 112 transcription factor 16 . Because the ErkKTR is exported from the nucleus in response to Erk 113 activity, an ErkKTR-transcription factor fusion protein would be precluded from encountering 114 DNA and expressing a target gene as long as the pathway remained active. 115

116
To realize this design experimentally we expressed a KTR-Gal4-VP64 synthetic transcription 117 factor (abbreviated throughout as KGV) downstream of the Erk-responsive FOS promoter (PFOS). 118 We then used a standard reporter construct, the Gal4-responsive UAS promoter driving 119 destabilized GFP, to record the circuit's output. Only in response to a pulse of Erk would KGV 120 be first expressed and then shuttled into the nucleus, resulting in GFP production ( Figure 2B). 121 We termed our circuit -comprising a dynamics-sensitive transcription factor and reporter gene -122 a Recorder of Erk Activity Dynamics, or READer. We transduced NIH3T3 cells with a lentiviral 123

PUAS-dGFP reporter and transfected them with a PiggyBAC transposase-integrable PFOS-KGV 124
Erk-responsive transcription factor, based on our prior data showing that the PiggyBAC system 125 can generate strongly Erk-responsive gene expression 17 , and sorted clonal cell lines harboring 126 both components (Figure S4; see Methods). 127 Just as in our simulations, cells expressing the READer circuit were able to discriminate 129 between pulsed and constant signaling inputs. We cultured cells overnight in media lacking 130 growth factors (GF-free media), and then monitored GFP induction by time-lapse microscopy 131 after addition of 10% serum (constant-on), or a 1 h pulse of serum followed by a return to GF-132 free media (pulsed) (Figure 2C; Figure S5; Movie S1). Performing confocal imaging for GFP 133 induction in each case revealed that a pulse of serum led to strong GFP induction within 4 hours, 134 whereas constant-on and constant-off stimuli each led to minimal GFP accumulation. 135

136
We reasoned that a pulse detection circuit should also enable inference of prior signaling 137 dynamics from a single measurement in fixed cells. We again exposed NIH3T3 READer cells to 138 constant-off, constant-on and pulsed serum inputs, fixed cells 4 h after the start of stimulation 139 and performed flow cytometry for GFP levels ( Figure 2D). We found that constant-on and 140 constant-off conditions failed to induce GFP in most READer cells, with a small tail of GFP-141 high cells that will be discussed in detail below. In contrast, a pulse of serum induced strong GFP 142 induction within 3-6 hours in approximately 50% of cells, while the remainder of the population 143 remained un-induced ( Figure 2E; Figure S6). Subsequent experiments revealed that this 144 bimodal response arose because only a subset of cells could transmit Erk activity to downstream 145 gene expression. Cells sorted from only the GFP-high or GFP-low populations generated the 146 same bimodal response upon a second stimulus challenge, indicating a non-genetic source of 147 response variability (Figure S7). Furthermore, the fraction of signaling-responsive cells could be 148 increased by pre-treatment with 10 ng/mL anisomycin (Figure S8), a treatment that we 149 previously observed to increase Erk-dependent transcription of endogenous immediate-early 150 genes 18,19 . We also tested whether Erk-triggered target gene induction depended on cell cycle phase, but found that GFP-high and GFP-low cells each exhibited similar DNA content 152 distributions, arguing against cell cycle control over IEG induction ( Figure S9). Together, our 153 data demonstrates that the READer system labels cells with pulsatile Erk activity and a 154 permissive transcriptional state for immediate-early gene induction. 155

156
The scalability of flow cytometry enabled us to rapidly scan additional stimulus conditions to 157 test how the READer circuit filtered a broad range of dynamic inputs. We first tested how 158 selective the circuit was to changes in the pulse duration. Endogenous Erk pulses are typically 159 observed to be less than 1 h in length, with sustained responses lasting for multiple hours 8,20-22 . 160 We applied pulses of different durations ranging from 5 min to 12 h, then incubated cells for an 161 additional 3 h prior to fixation to allow GFP to accumulate (Figure 2F). Although pulses from 5 162 min to 2 h resulted in similar profiles of GFP expression, longer pulses were filtered and ignored 163 by the circuit (Figure 2G). We also tested for GFP induction in response to a broad range of 164 dynamic Erk inputs delivered using our OptoSOS optogenetic system (see Methods) 23,24 . Trains 165 of multiple pulses also led to GFP accumulation, indicating that the READer circuit detects 166 persistent signaling oscillations as well as a single pulse ( Figure S10). Together, these data 167 reveal that the READer circuit responds broadly to pulsatile Erk stimuli while filtering out 168 constant high or low signaling states. 169 170 Our data revealed that the READer circuit ignores very long pulses greater than 2 h in length 171 ( Figure 2G). While this long pulse rejection was not a prediction from our original "Circuit 7" 172 model, it can be readily understood based on the prior observation that even a sustained Erk 173 stimulus only drives a transient, 30 min pulse of IEG expression, after which subsequent expression is suppressed 25-27 . Thus, after a long input pulse, KGV RNA/protein levels could 175 drop, leaving little protein to return to the nucleus to drive GFP expression (Figure 2H). We 176 verified that our transposase-integrated PFOS promoter indeed produced a transient pulse of 177 expression, in agreement with prior data on fos expression (Figure S11) 25  To directly compare endogenous Erk pulses to GFP accumulation, we next transduced our 202 READer clonal cell line with the ErkKTR-mScarlet fluorescent biosensor to monitor both 203 biosensors in the same live cells (Figure 3A). We incubated cells in serum-free media overnight 204 and switching to 'constant-on' growth media at the start of imaging, based on simulations which 205 indicated that such an input should prevent READer system activation during the constant-on 206 phase but elicit a sharp rise in GFP induction upon the spontaneous switch to pulsatile Erk 207 activity ( Figure 3B). Indeed, we found that serum stimulation first drove a constant-on Erk state 208 (leading to ErkKTR nuclear export), but after 15 hours Erk activity began pulsing in some cells 209 ( Figure 3C; Movie S3). The switch to a pulsing state was accompanied by a rapid increase in 210 GFP intensity, a phenomenon that was matched by our computational model when stimulated 211 with the experimentally-observed trajectory of Erk dynamics (Figure 3C, inset; see Figure S16 212 for additional cells and simulations). Taken together, these data demonstrate that the READer 213 circuit can indeed sense spontaneous, naturally occurring Erk pulses. 214

215
As a final test of the READer biosensor, we set out to compare its response to staining for 216 classic Erk target genes. To our knowledge, no endogenous Erk target genes have been identified 217 that specifically sense pulsatile stimuli, but the Fos immediate-early gene product has long been 218 used as a marker to identify cells exhibiting high levels of Erk pathway activity 30 (Figure 3D).  see Figure S17 for full joint READer/Fos distributions). We observed rapid, strong induction of 227 Fos at early time points regardless of stimulus duration, demonstrating that simply measuring 228 Fos cannot be used to discriminate pulsatile from sustained signaling. In contrast, the READer 229 circuit only triggered GFP expression in response to a pulse but not constant stimulation. These 230 results confirm that the dynamic information using READer cannot be obtained by staining for 231 classic Erk target genes like Fos. 232

233
Here we report the discovery and characterization of a simple gene network that can 234 selectively and robustly differentiate between pulsatile and constant signaling states. Our 235 network is based on an incoherent feedforward loop with slow activation and fast repression. 236 Incoherent feedforward loops have been studied extensively for their dynamic filtering 237 capabilities, including pulse generation and temporal ordering 11,12,15 ; our work adds highly-238 selective pulse detection to this list of capabilities. We also report a simple, flexible 239 implementation of this network architecture for mammalian signaling, centered on the pathway-240 regulated expression of a transcription factor that is fused to a kinase translocation reporter. 241 Although we have focused on Erk signaling in this work, we believe this report provides the 242 roadmap to the development of a suite of new reporters that capture the previous dynamic history 243 for many dynamic signaling pathways (i.e. p53, Wnt, NFkB, etc.