Results

In humans, glucocorticoids are sensed via glucocorticoid receptor (GR) encoded by the NR3C1 gene. In the absence of glucocorticoids, GRs are believed to form complexes with molecular chaperones like HSP90 in the cytosol. They translocate into the nucleus upon glucocorticoid stimulation and bind to Glucocorticoid responsive elements (GRE) in the promoter regions of the target genes (Weikum et al., 2017). Inspired by this endogenous signaling cascade, we designed a synthetic glucocorticoid-sensing circuit comprising GR and a chimeric promoter (P_GRE) consisting of 3 or 6 tandem repeats of GREs upstream of a miniCMV promoter. Glucocorticoid-mediated activation of GR triggers the GR-P_GRE binding and therefore activates downstream gene transcription (Figure 1a). To evaluate the glucocorticoid-sensing performance of this cascade, we co-transfected human embryonic kidney (HEK-293T) cells with the GR-expressing PNC046 plasmid and pNC047 (or pNC048) encoding P_GRE-driven secreted alkaline phosphatase (SEAP)-expressing cassette. As expected, we observed a mild upregulation of SEAP expression in cells treated with 100 nM glucocorticoid compared to the un-stimulated controls (1.92-fold for the P_GRE3 group and 1.55-fold for the P_GRE6 group). However, cells treated with either 10 nM or 50 nM of glucocorticoids showed no significant SEAP activation compared to the control cells (Figure 1b), suggesting that these circuits might not be sensitive enough for subcutaneous glucocorticoid levels (approximately 1-10 nM).

Glucocorticoid-triggered nucleus entry of GR is a key step for GR-mediated transcriptional activation. We then went on to ask if the overexpressed GRs were correctly translocated into the nucleus upon glucocorticoid stimulation in our host cells. For this purpose, we constructed a GR-EGFP fusion protein to monitor the GR localization changes with the green fluorescent signal. Intriguingly, we observed a robust nucleus-like localization of the GR-EGFP protein in pNC049 (P_CMV-GR-EGFP) transfected cells that were not even treated with glucocorticoids (Figure 1c, left panel). Confocal microscopy of HEK-293T cells co-transfected with actin-labeling pXQ100 (P_CMV-lifeact-mcherry) and pNC049 (P_CMV-GR-EGFP) also clearly showed that GR-EGFP proteins were majorly located in an actin-free, nucleus-alike location (Figure 1c, right panel) in both untreated control cells and glucocorticoid-treated cells. These findings surprisingly revealed that GRs might not be correctly located in HEK-293T cells, limiting its potential as a glucocorticoid sensor in this well-characterized host cell line.

Figure 1. Human glucocorticoid receptor (GR)-based genetic circuits for glucocorticoid detection in HEK-293T cells. (a) Schematic representation of the GR-based glucocorticoid sensing circuit. Overexpressed GR binds to glucocorticoid, moves into the nucleus, and binds to a synthetic GR responsive promoter (P_GRE), activating downstream gene expression. (b) P_GRE-mediated SEAP production in glucocorticoid-stimulated HEK-293T cells co-transfected with pNC046 (P_CMV-GR) and indicated reporter plasmids (pNC047 GRE₃-P_mini-SEAP or pNC048 GRE₆-P_mini-SEAP). SEAP production was measured 48 h post-DMEM or glucocorticoid stimulation; data shows mean±SD, n=3 independent experiments. (c) EGFP-labeled GR localization in HEK-293T cells w/wo glucocorticoid stimulation. For wide-field microscopy, cells were transfected with pNC049  (P_CMV-GR-EGFP). For confocal images, cells were co-transfected with F-actin labeling pXQ100 (P_CMV-lifeact-mcherry) and GR labeling pWY049 (P_CMV-GR-EGFP). Photos were taken 48 h post-DMEM or glucocorticoid- administration, scale bar: 100 µm for wide-field microscopy and 10 µm for confocal microscopy. Data are representative images of 3 independent experiments.

Human GRs share a similar domain structure consisting of an N-terminus domain (NTD), a zinc-finger DNA Binding Domain (DBD), a hinge region, and a C-terminus Ligand binding domain (LBD) (Figure 2a). The LBD domain is closely correlated to ligand binding, subcellular localization, and transcriptional activation. Since the full-length GR didn’t work well in HEK-293T cells, we then explored whether it was the LBD domain that got affected by the host cell. Hence, we constructed a fusion protein comprising the GR_LBD domain and EGFP (Figure 2a). Interestingly, both wide-field and confocal fluorescent imaging of pNC051 (P_CMV-GR_LBD-EGFP) transfected cells showed that GR_LBD-EGFP proteins were broadly spread all over the untreated control cells and were majorly nucleus-localized in glucocorticoid-stimulated cells (Figure 2b), suggesting the truncation of GR into GR_LBD might be a possible approach to improve the sensitivity of a synthetic glucocorticoid-sensing circuit.

Figure 2. Glucocorticoid-stimulated translocation of GR Ligand Binding Domain (LBD) in HEK-293T cells. (a) Schematic representation of the GR_LBD-EGFP construct. The LBD domain of human GR was fused to the N-Terminus of EGFP with a GGGSG linker. (b) EGFP-labeled GR_LBD localization in HEK-293T cells w/wo glucocorticoid stimulation. For wide-field microscopy, cells were transfected with pNC051 (P_CMV-GR_LBD-EGFP). For confocal images, cells were co-transfected with F-actin labeling pXQ100 (P_CMV-lifeact-mcherry) and GR_LBD labeling pNC051 (P_CMV-GR_LBD-EGFP). Photos were taken 48 h post-DMEM or glucocorticoid administration, scale bar: 100 µm for wide-field microscopy and 10 µm for confocal microscopy. Data are representative images of 3 independent experiments.

Building on this knowledge, we then designed Tet_stress, a novel synthetic transcriptional factor, on the bases of GR_LBD. Since we have abolished the DNA binding ability of GR_LBD during the truncation, we then supplemented tetR, a well-characterized DNA binding protein, on the C-terminus or the N-terminus of the GR_LBD to rewire the GR signaling to the well-established P_tetO7 promoter. Also, since the NTD domain carries a transcriptional activation function, we designed several variants adding the VP64 transcriptional activator as a replacement (Figure 3a). To evaluate the function of these variants, we co-transfected HEK-293T cells with pXQ164 (P_tetO7-SEAP) and a Tet_stress variant (pNC054 P_CMV-tetR-GR_LBD, pNC057  P_CMV-tetR-VP64-GR_LBD, pNC056 P_CMV-GR_LBD-tetR or pNC058 P_CMV-GR_LBD-tetR-VP64). Results demonstrated a significantly improved glucocorticoid response in both tetR-GR_LBD and GR_LBD-tetR configurations (5.21-fold for tetR-GR_LBD and 22.50-fold for GR_LBD-tetR compared to their respective non-treated controls). At the same time, their corresponding VP64-supplemented variants showed poor glucocorticoid response mainly due to the high background activation (Figure 3b). Further characterization of pNC054 (P_CMV-tetR-GR_LBD) demonstrated that the tetR-GR_LBD configuration resulted in a strong SEAP activation with a high basal expression, which significantly narrowed the dynamic range (approximately between 0-50 nM) of the circuit (Figure 3c). For the GR_LBD-tetR configuration, HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP) and pNC056 (P_CMV-GR_LBD-tetR) showed a much lower basal leakage compared to the tetR-GR_LBD configuration, resulting in a significantly wider dynamic range (wider than 0-100 nM) and a better dose dependence (Figure 3d).

Figure 3. Glucocorticoid responsive transcriptional activation mediated by the synthetic Tet_stress transcriptional factor (a) Schematic representation of the TetR-GR_LBD-based glucocorticoid responsive circuit. TetR was fused to GR_LBD on either its N-terminus or C-terminus. Optional VP64 domains were included to further improve transcriptional activation efficiency.  (b) Glucocorticoid-activated SEAP production in HEK-293T cells co-transfected with pXQ164 (P_TetO7-SEAP) and a Tet_stress variant (pNC054 P_cmv-TetR-GR_LBD, pNC057 P_cmv-TetR-VP64-GR_LBD, pNC056 P_cmv-GR_LBD-TetR or pNC058 P_cmv-GR_LBD-TetR-VP64). SEAP production was measured 48 h post-DMEM or glucocorticoid stimulation; data shows mean value, n=2 independent experiments. (c-d) Glucocorticoid-dose-dependent SEAP production in HEK-293T cells mediated by the optimal Tet_stress in (b). Cells were co-transfected with pXQ164 (P_TetO7-SEAP) and a TetR-GR_LBD variant (pNC054 P_cmv-TetR-GR_LBD for c, pNC056 P_cmv-GR_LBD-TetR for d). SEAP production was measured 24 or 48 h post-DMEM or glucocorticoid stimulation. Data shows mean±SD, n=3 independent experiments.

With the GR_LBD-tetR as the optimal configuration of the synthetic Tet_stress transcriptional factor, we then went on to develop the HEK_stress cells that generate fluorescent or colorimetric signals upon glucocorticoid stimulation, which can possibly serve as the basis of a biomedical tattoo. To enable fluorescent output, we constructed pNC059, a P_tetO7-tdTomato reporter plasmid (Figure 4a). By co-transfecting HEK-293T cells with pNC060 (P_CMV-GR_LBD-tetR) and pNC056, we noticed a strong red fluorescent signal upon 100 nM glucocorticoid stimulation compared to the control cells treated by low concentration (1 nM) of glucocorticoid (Figure 4b). Since fluorescent signals were not visible to the naked eye, we changed the reporter gene into tyrosinase, the rate-limiting enzyme in synthesizing the black melanin pigment using Tyrosine as the substrate (Figure 4c). To fully demonstrate the potential of our glucocorticoid-sensing circuit, we generated a stable HEK_stress cell line carrying both P_tetO7-TYR and P_CMV-GR_LBD-tetR expressing cassette with the sleeping beauty transpose system. Upon 96 hours of glucocorticoid stimulation, we observed dose-dependent enrichment of melanin pigment in HEK_stress cells. Phase contrast imaging showed that melanin pigment gathered into spherical droplets within living cells, providing a possibly visible output while keeping the cells alive (Figure 4d). To further validate whether the output signals were actually visible, we trypsinized the cells and obtained cell pellets by centrifuge. As shown in Figure 4e, we observed a visible, dose-dependent color change on the cell pellet, firmly proving that our circuit may produce a tattoo-like function that effectively senses glucocorticoid.

Figure 4. Glucocorticoid responsive fluorescent or colorimetric output with HEK_stress cell. (a) Schematic representation of HEK_stress cell with a red fluorescent output signal.   (b) Glucocorticoid-activated fluorescent signal in HEK-293T cells co-transfected with pNC059 (P_tetO7-tdTomato) and pNC060 (P_CMV-GR_LBD-tetR). Fluorescent images were taken 48 h post-glucocorticoid stimulation, scale bar: 100 µm. (c) Schematic representation of HEK_stress cell with a black colorimetric output signal. In the lysosome, tyrosinase (TYR) generated by GR_LBD-tetR activation transforms Tyrosine into melanin, thereby forming dark pigments. (d) Bright-field (ND) and Phase contrast (Ph) images of attached or trypsinized HEK_stress cells under glucocorticoid stimulation. HEK_stress cells stably express P_tetO7-TYR and P_CMV-GR_LBD-tetR cassette.Images were taken 96 h post-glucocorticoid stimulation, scale bar: 100µm. (e) Cell pellets of the corresponding groups in panel d. For panels b, d, and e, data represent three independent experiments.

With a solid proof-of-concept, we then went on to further characterize and optimize the performance of the Tet_stress-based circuit. To figure out how to improve the sensitivity, we discussed our project with Mirta Viviani, a Ph.D. student from Westlake University. She is experienced in engineering mammalian cell gene circuits (link to the human-practices page). She suggested several potential approaches like optimizing P_tetO7, changing linkers between tetR and GR_LBD, and further manipulating the subcellular localization of the Tet_stress transcriptional factor. Meanwhile, our modeling team performed a sensitivity screening among all the parameters that can be changed by circuit design. Their results showed that the circuit’s sensitivity largely depends on the efficiency of nuclear import of Tet_stress upon glucocorticoid stimulation (link to the modeling page).

Building on these suggestions and findings, we then tested two potential approaches to improve Tet_stress performance. On the one hand, we tried to manipulate the linker length between tetR and GR_LBD. Taking the tetR-GR_LBD configuration as an example, we replaced the original 1xGS linker comprising a GGGSG sequence in the pNC054 (P_CMV-tetR-GR_LBD) plasmid into a 3xGS ([GGGSG]₃) or a 5xGS ([GGGSG]₅) linker. Another configuration with no linker added between tetR and GR_LBD was also generated (Figure 5a, upper panel). Unexpectedly, while the removal of the linker totally abolished the transcriptional activation function of Tet_stress, a similar level of glucocorticoid-induced SEAP production was observed in HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP), and one of these variants carrying GS linkers (pNC054 P_CMV-tetR-GR_LBD, pNC065 P_CMV-tetR-[3xGS]-GR_LBD, or pNC066 P_CMV-tetR-[5xGS]-GR_LBD). We even observed an increasing basal SEAP production in configurations carrying longer GS linkers compared to the original 1xGS linker configuration (Figure 5a, lower panel).

On the other hand, we tried to include a nuclear export signal (NES) between tetR and GR_LBD to manipulate the subcellular localization of the Tet_stress (Figure 5b, upper panel). While NES integration showed no significant improvement in the tetR-GR_LBD configuration, a striking boost of glucocorticoid-stimulated SEAP production was observed in HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP) and pNC069 (P_CMV-GR_LBD-NES-tetR) compared to the cells transfected with the variant without NES (91.6x vs. 25x, compared to their untreated control). At the same time, the basal leakage SEAP expression remained similar between the two configurations (Figure 5b, lower panel). To explore the nature of such a boost, we performed fluorescent imaging on cells transfected with the pNC069 (P_CMV-GR_LBD-NES-EGFP) plasmid to track the subcellular localization changes of the NES-integrated GR_LBD. Interestingly, we observed a significantly improved cytosol localization and a reduced nuclear localization in the untreated cells (Figure 5c, upper row) compared to the GR_LBD-EGFP configuration shown in Figure 2. We also observed a clear nucleus localization of GR_LBD-NES-EGFP fusion protein in cells treated with 100 nM glucocorticoids (Figure 5c, lower row). These findings partly explained the boost of SEAP production in the NES-integrated Tet_stress variant.

Figure 5. Nuclear export signal (NES) allowed further improvement of Tet_stress performance. (a) Glucocorticoid-activated SEAP production in HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP) and a linker-optimized tetR-GR_LBD variant (pNC064 P_CMV-tetR-[no linker]-GR_LBD, pNC054 P_CMV-tetR-GR_LBD, pNC065 P_CMV-tetR-[3xGS]-GR_LBD, or pNC066 P_CMV-tetR-[5xGS]-GR_LBD, as schematically demonstrated in the upper panel). (b) Glucocorticoid-activated SEAP production in HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP) and a NES-integrating tetR-GR_LBD variant (pNC054 P_CMV-tetR-GR_LBD, pNC067  P_CMV-tetR-NES-GR_LBD, pNC056 P_CMV-GR_LBD-tetR or pNC069 P_CMV GR_LBD-NES-tetR, as schematically demonstrated in the upper panel). For panels a and b, SEAP production was measured 64 h post-DMEM or glucocorticoid stimulation. Data shows mean±SD, n=3 independent experiments. (c) EGFP-labeled NES-GR_LBD localization in HEK-293T cells w/wo glucocorticoid stimulation. For wide-field microscopy, cells were transfected with pNC062 (P_CMV-GR_LBD-NES-EGFP). For confocal images, cells were co-transfected with F-actin labeling pXQ100 (P_CMV-lifeact-mcherry) and GR_LBD labeling pNC063 (P_CMV-GR_LBD-NES-EGFP). Photos were taken 48 h post-DMEM or glucocorticoid- administration, scale bar: 100 µm for wide-field microscopy and 10 µm for confocal microscopy. Data are representative images of 3 independent experiments.

Then, additional experiments were performed to profile the transcriptional activation kinetics and the glucocorticoid-dose dependence of NES-integrated Tet_stress. As shown in Figure 6a, HEK-293T cells co-transfected with pXQ164 (P_tetO7-SEAP) and pNC069 (P_CMV-GR_LBD-NES-tetR) produced a detectable amount of SEAP (0.69±0.15 U/L) at the 6^th hour after 100 nM glucocorticoid stimulation, suggesting that short-term exposure to glucocorticoid might not be able to activate the Tet_stress-mediated transcription, which nicely met our requirement for chronic stress detection. Also, pXQ164 and pNC069 co-transfected cells exposed to a gradient amount of glucocorticoid demonstrated that NES-integrated Tet_stress would allow HEK_stress cells to distinguish a 5 nM glucocorticoid difference within the 2-15 nM range (Figure 6b).

Due to the time limitation, we could not perform a thorough functional validation with fluorescent or colorimetric output using stable cell lines. Instead, we only demonstrated that NES-integrated Tet_stress could effectively generate glucocorticoid-induced fluorescent or colorimetric responses even by transient transfection and under a lower glucocorticoid concentration (50 nM). As shown in Figure 6c, HEK-293T cells co-transfected with pNC069 (P_CMV-GR_LBD-NES-tetR) and pNC059 (P_tetO7-tdTomato) showed a strong red fluorescent signal upon 50 nM glucocorticoid stimulation. In contrast, no fluorescent signal could be observed in the untreated cells. Similarly, 50 nM glucocorticoid stimulation could trigger a visible melanin production and accumulation in cells co-transfected with pNC069 (P_CMV-GR_LBD-NES-tetR) and pNC060 (P_tetO7-TYR) (Figure 6d). Our future experiments will characterize the detection limits and dynamic range of the NES-integrated Tetstress.

Figure 6. Characterization of NES-integrated HEK_stress cells. (a) SEAP production kinetics of HEK_stress cells exposed to Glucocorticoid for a different amount of time. SEAP production was measured every 3 hours post-DMEM or glucocorticoid stimulation. Data shows mean±SD, n=3 independent experiments. (b) SEAP production in HEK_stress cells exposed with increasing Glucocorticoid concentration. SEAP production was measured 48 h post-glucocorticoid stimulation. Data shows mean±SD, n=3 independent experiments. For panels a and b, HEK_stress cells were generated by co-transfecting HEK-293T cells with pXQ164 (P_tetO7-SEAP) and pNC069 (P_CMV-GR_LBD-NES-tetR). (c) Glucocorticoid-activated fluorescent signal in HEK-293T cells co-transfected with pNC059 (P_tetO7-tdTomato) and pNC069 (P_CMV-GR_LBD-NES-tetR). (d) Columns 1 and 2: bright-field images of attached or trypsinized HEK-293T cells co-transfected with pNC060 (P_tetO7-TYR) and pNC069 (P_CMV-GR_LBD-NES-tetR). Column 3: Pellet of corresponding cells shown in the left columns. For panels c and d, images were taken 48 h post-glucocorticoid stimulation, scale bar: 100 µm. Data were representative of 3 independent experiments.