Synthetic protein circuits for programmable control of mammalian cell death

Abstract

Natural cell death pathways such as apoptosis and pyroptosis play dual roles: they eliminate harmful cells and modulate the immune system by dampening or stimulating inflammation. Synthetic protein circuits capable of triggering specific death programs in target cells could similarly remove harmful cells while appropriately modulating immune responses. However, cells actively influence their death modes in response to natural signals, making it challenging to control death modes. Here, we introduce naturally inspired "synpoptosis" circuits that proteolytically regulate engineered executioner proteins and mammalian cell death. These circuits direct cell death modes, respond to combinations of protease inputs, and selectively eliminate target cells. Furthermore, synpoptosis circuits can be transmitted intercellularly, offering a foundation for engineering synthetic killer cells that induce desired death programs in target cells without self-destruction. Together, these results lay the groundwork for programmable control of mammalian cell death.

Keywords: apoptosis; cell death; protein circuit; pyroptosis; synpoptosis; synthetic biology; synthetic circuit.

Figure 1.. Synpoptosis circuits control user-selectable cell death programs

(A) Synthetic cell death “synpoptosis” circuits steer the mode of cell death by operating orthogonally to cell-intrinsic death programs. (B) Molecular building blocks of synpoptosis circuits include caspase-3 subunits, gasdermin domains, viral proteases such as TEVP, degrons, maltose-binding protein, and leucine zippers. (C) Synthetic apoptosis modules use viral proteases, such as TEVP, to activate or repress engineered variants of caspase-3. We transiently transfected HEK cells with plasmid DNA encoding the synthetic modules and then quantified cell death by staining and flow cytometry. Throughout the figure, the gray window indicates the fractions established by negative and positive transient DNA transfection controls; dots represent biological replicates (distinct culture wells). (D) Synthetic pyroptosis modules similarly use TEVP to regulate engineered GSDMA. (E) Apoptosis and pyroptosis exhibit different staining patterns with Annexin and Sytox. Annexin stains both apoptotic and pyroptotic cells, while Sytox primarily stains pyroptotic cells. We calculated the fraction of cells that stained positive for each dye after gating on the transfected cells based on fluorescence of a co-transfected marker. Data represent three independent experiments. See also Figure S1.

Figure 2.. Synpoptosis circuits lead to canonical features of cell death

(A) Transient transfection of HEK cells with plasmid DNA encoding the TEVP-activated caspase-3 circuit triggered asynchronous apoptosis, shown by flow cytometry. Between 16 and 24 hours after transfection (gray window), Sytox remains low in apoptotic cells and therefore can reliably distinguish between apoptotic and pyroptotic cells. Throughout the paper, dots represent biological replicates (distinct culture wells). (B) Similarly, the TEVP-activated GSDMA circuits triggered asynchronous pyroptosis. (C) Q-VD-OPh suppressed circuit-induced apoptosis but not pyroptosis, shown by flow cytometry. The gray window indicates the fractions established by negative and positive transient DNA transfection controls. (D) TO-PRO-3 stains cells killed by the apoptosis and pyroptosis circuits, shown by flow cytometry. (E) The apoptosis and pyroptosis circuits differently triggered the release of IL-1β and IL-18 from engineered HEK cells that stably express these cytokines. (F) Transient transfection using in vitro transcribed mRNA transcripts of the synpoptosis circuits led to loss of ATP-based cell viability. (G) The mRNA version of the pyroptosis circuit triggered more LDH release than the apoptosis circuit. (H) The mRNA version of the pyroptosis circuit triggered more ATP release than the apoptosis circuit. The same wells were repeatedly captured across a time course. See also Figure S2.

Figure 3.. Synpoptosis circuits direct the mode of cell death

(A) Transient transfection of plasmid DNA encoding TEVP and engineered TEVP-activatable caspase-3 induced apoptosis in HEK cells, shown by flow cytometry. TEVP-activatable GSDMA overrode the apoptotic program, leading to pyroptosis. Throughout the figure, horizontal lines indicate the fractions established by negative and positive transient DNA transfection controls, established separately for Annexin and Sytox; dots represent biological replicates (distinct culture wells). (B) Cells transfected with plasmid DNA encoding wildtype GSDME, a natural substrate of caspase-3, underwent pyroptosis in response to TEVP-mediated caspase-3 activation. (C) GSDME N(mut) overcame the tendency of wildtype GSDME-expressing cells to undergo pyroptosis downstream of caspase-3 activation and promoted apoptosis. (D) The ratio between the two forms of cell death is tunable by adjusting the plasmid DNA amount of GSDME N(mut) relative to wildtype GSDME (low, 1:1; medium, 2:1; high, 4:1). The heights of the stacked bars indicate means of three biological replicates (distinct culture wells) measured by flow cytometry. (E) In vitro transcribed mRNA versions of synpoptosis circuits drove apoptosis and pyroptosis in Jurkat and THP-1 cells, which have more sophisticated endogenous cell death circuitry than HEK cells. See also Figure S3.

Figure 4.. Synpoptosis circuits perform combinatorial computation

(A) Synpoptosis circuits enabled context-dependent apoptosis by responding to logical combinations of protease inputs (TEVP as Input 1 and TVMVP as Input 2). In the first demonstration, transient transfection of plasmid DNA encoding both protease inputs must be present to activate the engineered caspase-3 and cause apoptosis, shown by flow cytometry. In the second demonstration, either input is sufficient to trigger apoptosis. In the third demonstration, apoptosis occurs only when the first input is present and the second input is absent. Throughout the figure, the gray window indicates the fractions established by negative and positive transient DNA transfection controls; dots represent biological replicates (distinct culture wells). (B) Similar design principles were applied to engineer the same gating functions for pyroptosis. (C) Using these synpoptosis gates, we can eliminate specific cells within mixed populations that exhibit distinct intracellular states defined by the expression profiles of input proteases and their fluorescence reporters, shown by flow cytometry. The cell states were established by stable protein expression using lentiviral transduction. Data represent three independent experiments. (D) Quantification of Annexin staining revealed the efficacy of the synpoptosis gates. Horizontal lines indicate quadrant-specific negative and positive controls of the Annexin signal. Bars indicate means of three biological replicates (distinct culture wells). See also Figure S4.

Figure 5.. Synpoptosis circuits selectively eliminate target cells

(A) Synpoptosis circuits can selectively kill target cells by incorporating a sensor module that identifies the target cells. We used a synthetic sensor of active Ras. The sensor is a split TEVP with each half – nTEVP and cTEVP – tethered to a Ras-binding domain (RBD). In wildtype (WT) HEK cells with inactive Ras, the sensor is catalytically inactive. In cells stably expressing active Ras (RasGC cells), TEVP is reconstituted by proximity. (B) Transient DNA transfection of a caspase-3-based synpoptosis circuit including the synthetic Ras sensor module enabled selective apoptosis of RasGC cells, but not WT cells. Throughout the figure, horizontal lines in the bar plots indicate the fractions established by negative and positive transient DNA transfection controls, separately for each dye and each cell line; dots represent biological replicates (distinct culture wells); bars indicate means; histograms represent three independent experiments. (C) Similarly, a Ras-sensing synpoptosis circuit using GSDMA as the output triggered selective pyroptosis of RasGC cells. See also Figure S5.

Figure 6.. Synpoptosis circuits support intercellular operations

(A) Synpoptosis circuits can be transmitted intercellularly by virus-like particles (VLPs). To address the key challenge of sender cell death, a simple strategy is to use a split-sender system, in which an inactive executioner and an activating protease are packaged separately. (B) A model VLP transiently delivered nucleic acid encoding Cherry from sender HEK cells to receiver HEK cells, where the cargo was expressed. Experiments were performed by supernatant transfer from sender to receiver cells after transient DNA transfection of sender cells. Line segments connect means at different time points. Dots represent biological replicates (distinct culture wells). (C) Two model cargoes, Cherry and Citrine, were separately packaged by two sender populations and co-delivered to the same receiver population, supporting the split-sender system. Scatter plots represent three independent experiments. (D) A synthetic pyroptosis circuit, consisting of inactive GSDMA and the activating protease TEVP, was delivered by VLP using the split-sender system. Throughout the figure, the gray window indicates the fractions established by negative and positive transient DNA transfection controls; bars indicate means; dots in the two-dimensional sender death-receiver death plots indicate means of three biological replicates (distinct culture wells). (E) Analogously, the split-sender system enabled VLP delivery of a synthetic apoptosis circuit, by separately packaging inactive caspase-3 and the activating protease TEVP. Appending CAAX tails to both circuit components enhanced apoptosis. (F) A more elegant, compact, single-sender system directly delivers an active executioner. This strategy requires sender-specific silencing of the active executioner. (G) A synthetic protein-level silencer (GSDMA Z-C) potently inhibited an engineered active executioner (GSDMA N-Z), likely by using leucine zippers (Z) to mask a loop critical for the executioner’s activity, shown by AlphaFold models. The inhibitory curve was obtained by transient transfection of HEK cells with plasmid DNA encoding the active executioner and the silencer at indicated plasmid mass ratios. (H) Senders transfected with plasmid DNA encoding the silencer, but not wildtype senders, could directly package VLPs that express the active executioner. See also Figure S6.