Drug discovery would greatly benefit from a high-throughput drug screening platform that allows simultaneous assessment of multiple drug candidates with resource-efficient and simple experiments. To identify potential melatonin receptor agonists more efficiently, we designed a melatonin-regulated reporter assay, which utilizes a synthetic excitation-transcription circuit that couples the activation of melatonin receptor 1a (MTNR1A) with the expression of secreted Nanoluc luciferase (Nluc).

In mammals, the activation of the melatonin receptor is known to increase intracellular calcium levels by activating phospholipase C (PLC), which catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP₂) into inositol trisphosphate (IP₃) and diacylglycerol (DAG). The resulting IP₃ stimulates the release of Ca²⁺ from the endoplasmic reticulum (ER) into the cytosol, increasing intracellular calcium concentrations ¹. Building on this well-characterized signaling pathway, we designed a synthetic melatonin-responsive circuit incorporating MTNR1A and a chimeric calcium-sensing promoter (P_NFAT). This promoter consists of multiple repeats of nuclear factor of activated T cells (NFAT) binding sites followed by a minimal CMV promoter. Elevating intracellular calcium levels would activate endogenous NFAT, leading to its binding at the NFAT sites and subsequent transcription of downstream proteins (Fig. 1A).

To assess the functionality of this circuit, we co-transfected human embryonic kidney (HEK-293T) cells with MTNR1A-expressing pNC099 plasmids, alongside either pNC102, pNC100, pNC104, or pNC103, which encode a P_NFAT-driven Nluc-expressing cassette. Unexpectedly, no significant Nluc expression was observed in cells treated with near-physiological concentrations of melatonin (1 nM) compared to untreated controls (Fig. 1B). In contrast, robust Nluc expression could be detected in cells treated with Thapsigargin (Fig. 1C), a known inducer of intracellular calcium release via ER depletion ². In addition, cells transfected with pNC107 (lacking NFAT binding sites) showed no significant response to Thapsigargin, confirming the specificity of the response. These findings suggest that MTNR1A may not effectively induce sufficient calcium signaling in HEK-293T cells to activate the NFAT-dependent calcium-responsive promoters.

Figure 1 NFAT-based calcium-sensing promoters failed to response to the activation of melatonin receptor. (A) Schematic diagram of the gene circuit design that response to the MTNR1A-induced elevation of calcium level. The activation of MATNR1A triggers the elevation of intracellular Ca²⁺ level via the PLC/IP₃ pathway; the elevated Ca²⁺ triggers the dephosphorylation of NFAT, which would then bind to the NFAT binding sites on the chimeric promoter and activate the downstream gene expression. (B) The P_NFAT promoters respond poorly to melatonin receptor activation. HEK293T cells were co-transfected with melatonin receptor plasmid pNC099(P_CMV-MTNR1A) and either pNC102(P_{NFAT_1}-IgK-Nluc), pNC100(P_{NFAT_5}-IgK-Nluc), pNC104(P_{NFAT_6}-IgK-Nluc) or pNC103(P_{NFAT_7}-IgK-Nluc). Cells were treated with either DMSO or melatonin 6 hours post transcription. Data represents mean±SD of nanoluc expression levels measured at 24 h after melatonin stimulation (n = 3 independent experiments). (C) The P_NFAT promoters respond robustly to the Thapsigargin-induced elevation of intracellular calcium level. HEK293T cells were transfected with either pNC102(P_{NFAT_1}-IgK-Nluc), pNC100(P_{NFAT_5}-IgK-Nluc), pNC104(P_{NFAT_6}-IgK-Nluc) or pNC103(P_{NFAT_7}-IgK-Nluc). Cells were treated with either DMSO or Thapsigargin 6 hours post transcription. Data are mean±SD of nanoluc expression levels measured 48 h after thapsigargin stimulation (n = 3 independent experiments).

To investigate why the calcium-based circuit did not perform as expected, we consulted Dr. Yiming Dong, a synthetic biologist, and Chief Scientific Officer of Atantares. Drawing from her experience with calcium-sensing promoters in mammalian cells, Dr. Dong suggested that we examine the calcium dynamics triggered by melatonin receptors. She noted that, in her experience, most calcium-responsive promoters required a sustained elevation of intracellular calcium levels for proper activation (refer to the Human Practices and Engineering section for more details).
To assess the calcium dynamics following MTNR1A activation, we co-transfected HEK-293T cells with the MTNR1A-expressing pNC099 plasmid and a calcium indicator plasmid encoding GCaMP6 ³ (P_CMV-GCaMP, courtesy of Mingqi Xie' s lab at Westlake University). Surprisingly, we observed no significant change in GCaMP fluorescence in cells treated with melatonin for 48 hours, either through fluorescence microscopy (Fig. 2A) or with a fluorescent plate reader (Fig. 2B), compared to untreated controls. However, real-time live-cell calcium imaging revealed a significant spike in intracellular calcium levels immediately after melatonin administration (Fig. 2C). Importantly, this calcium surge was short-lived, dissipating within 100 seconds of melatonin stimulation. In contrast, Thapsigargin induced a prolonged elevation of intracellular calcium levels (Fig. 2C).
These findings suggest that the transient nature of melatonin-induced calcium elevation is likely the primary reason for the lack of responsiveness in our P_NFAT-Nluc design. Developing a new gene switch that responds to such rapid calcium fluctuations would be time-consuming, prompting us to consider exploring alternative signaling pathways.

Figure 2. Melatonin-triggered alteration of intracellular Calcium dynamics in HEK-293T cells. (A and B) Long-term calcium response to melatonin stimulation in HEK-293T cells. HEK293T cells were co-transfected with the MTNR1A-expressing plasmid pNC099 (P_CMV-MTNR1A) and the plasmid expressing the calcium indicator GCaMP6 (P_CMV-GCaMP). Cells were treated with either DMSO or Melatonin 6 hours post transfection. For (A), Fluorescent images were taken 48 h after stimulation, scale bar: 100 µm. Data are representative images of 3 independent experiments. For (B), cells were subjected to fluorescence detection with a plate reader 48 h after stimulation. Data are mean±SD of relative fluorescence intensity (in RFU) measured 48 h after stimulation (n = 8 independent experiments). (C) Short-term calcium response to melatonin stimulation in HEK-293T cells. HEK293T cells were co-transfected with the MTNR1A-expressing plasmid pNC099 (P_CMV-MTNR1A) and the plasmid expressing the calcium indicator GCaMP6 (P_CMV-GCaMP) and treated with the indicated reagent at the 48 h post transfection. Cells were imaged through a fluorescent microscope every 5 seconds, and 20 cells were selected for fluorescent analysis in each group with ImageJ software. The cellular fluorescent levels of each group were normalized to the steady state level before chemical induction, Data shows Mean±SEM.

We then looked at other potential downstream signaling pathways of MTNR1A that we may possibly build synthetic gene circuits. Interestingly, we came across a very recent research that showed a robust increase of cAMP levels in melatonin-treated, MTNR1A-expressing HEK-293T cells ⁴. Additionally, other research has shown increased phosphorylation of cAMP-response element binding protein (CREB) in melatonin-stimulated cells ⁵. Following these leads, we proposed the possibility of rewiring the activation of MTNR1A to target gene transcription via the cAMP/PKA/CREB pathway. To accomplish this, we developed a new synthetic melatonin-responsive circuit, which includes MTNR1A and a chimeric cAMP-sensing promoter (P_CRE) containing multiple repeats of CREB binding sites followed by a miniCMV promoter. Activation of MTNR1A would lead to an increase in cAMP and the phosphorylation of the endogenous CREB protein. The phosphorylated CREB protein would then bind to the CREB site of the P_CRE promoter, activating the expression of the downstream genes (Fig. 3A).

To evaluate the function of this circuit, we co-transfected the HEK-293T cells with MTNR1A-expressing pNC099 plasmids and either pNC101, pNC105, or pNC106, which encodes P_CRE-driven Nluc-expressing cassette. As expected, cells expressing MTNR1A and P_CRE-Nluc variants showed significant activation of Nluc expression upon melatonin treatment compared to the untreated control. Among these P_CRE variants, the variant contains 4 tandem repeats of the CRE sequence showed the optimal performance (~8.62-fold activation compared to the untreated control, Fig. 3B). Also, fine-tuning the P_CRE-Nluc level could efficiently adjust the basal expression level and the dynamic range of the system (Fig. 3C).

To further characterize the melatonin-activating circuit, we generated a stable HEKMT cell line carrying both P_CRE4-Nluc and P_CMV- MTNR1A - expressing cassette with the sleeping beauty transpose system. As shown in Fig. 3D, HEKMT cells exposed to a gradient amount of melatonin could produce detectable amount of Nluc at around the 12th hour in a dose-dependent manner, with a EC50 of 0.4228 nM at the 24 h post melatonin stimulus (Fig. 3E). Additionally, we also showed that HEKMT cells are also capable of responding dose-dependently to Ramelteon, a commercially available melatonin receptor agonist (Fig. 3F). Together, these findings demonstrated the feasibility of the HEKMT cells as the basis of a drug-screening platform.

Figure 3. The construction and characterization of a cAMP/CREB-based melatonin-responsive gene circuit in HEK-293T cells. (A) Schematic diagram of the gene circuit design that responds to the MTNR1A-induced alteration of cAMP level. The activation of MATNR1A triggers the elevation of intracellular cAMP level, which triggers the activation of CREB, which would then bind to the CRE sites on the chimeric promoter and activate the downstream gene expression. (B) Co-expression of MTNR1A and P_CRE-promoter variants enables robust transcriptional activation upon melatonin stimulation. HEK293T cells were co-transfected with pNC099(P_CMV-MTNR1A) and either pNC101(P_{CRE_4}-IgK-Nluc), pNC105(P_{CRE_5}-IgK-Nluc), or pNC106(P_{CRE_6}-IgK-Nluc). Cells were treated with DMSO or melatonin at 6 hours post transfection; data are mean±SD of nanoluc expression levels measured 24 h after melatonin stimulation (n = 3 independent experiments). (C) Finetuning of P_CRE4-Nluc allows the tuning of system behavior. HEK293T cells were co-transfected with pNC099(P_CMV-MTNR1A) and pNC101(P_{CRE_4}-IgK-Nluc) under the indicated plasmid ratio. Cells were treated with DMSO or melatonin at 6 hours post transfection. Data are mean±SD of nanoluc expression levels measured at 24 h after melatonin stimulation (n = 3 independent experiments). (D) Step-response dynamics of HEKMT cells under melatonin treatment. HEKMT cell line carrying both P_CRE4-Nluc and P_CMV-MTNR1A expressing cassette was stimulated with the indicated amount of melatonin 16 hours post seeding. Nanoluc expression levels were measured at different time points after melatonin stimulation (Data are mean±SD, n = 3 independent experiments). (E) Dose-dependence curve of HEKMT cells under melatonin treatment. HEKMT cell line was stimulated with the indicated amount of melatonin 16 hours post seeding. Nanoluc expression levels were measured at 24 h after melatonin stimulation (Data are mean±SD, n = 3 independent experiments, EC50=0.4228 nM). (F) Ramelteon response of the HEKMT cells. HEKMT cells were treated with ramelteon 16 h post seeding. Nanoluc expression levels were measured at 24 h after ramelteon stimulation (Data are mean±SD, n = 3 independent experiments).

With a functioning melatonin receptor-responsive gene circuit, we then went on to build a drug screening platform to help screening potential MTNR1A agonists in a high throughput manner. On the basis of the HEKMT cells, we generated a control cell line stably expressing only the P_CRE4-Nluc cassette (HEKCTR cells) so we can distinguish the MTNR1A-independent activation of the P_CRE4-Nluc cassette from the real MTNR1A agonism reported by the HEKMT cells. While we were building the HEKCTR cells, we interviewed the front-line psychiatrists and engaged the potential user of the drug we might discover through our screening platform to gain some hints about what we can start testing our screening platform on. Interestingly, the term “aromatherapy” pops up as its been widely used around the globe for thousands of years to help falling asleep, and the general public feels more comfortable using these natural product as a daily supplement (See Human Practices and Engineering page for more information). Hence, we purchased 37 plant essential oils online and hope to see if we can get lucky finding some new agonists for MTNR1A.

Intriguingly, while most of the plant essential oils showed no significant alteration in the Nluc production of either the HEKMT cells or the HEKCTR cells, we surprisingly found that the HEKMT cells treated with Clove Pod oil showed a robust elevation in Nluc production compared to the DMSO-treated control (~5.25-fold). In comparison, Nluc production in HEKCTR cells showed no significant alteration, suggesting that Clove Pod oil activates Nluc production in a MTNR1A-dependent manner (Fig. 4a). Similar results were also observed on Caraway Oil, with a milder induction rate of Nluc production in HEKMT cells (~2.32-fold, Fig. 4a). Further analysis revealed that both Clove Pod oil and Caraway Oil could robustly induce MTNR1A-dependent Nluc production, but no clear dose-dependence could be observed on Caraway Oil (Fig. 4b and 4c). These findings not only showed the feasibility of our designer-cell-based platform for rapid, robust, and resource-efficient high-throughput drug screening but also identified two essential oil hits for agonist discovery.

Most essential oils are a mixture of multiple natural product molecules contributing differently to the overall therapeutic function. However, screening over all constituents in the essential oil hits to identify the functioning molecule can sometimes be difficult due to the availability of the constituents and the time- or financial-constraints. Similar concerns were also raised by Dr. Mingshun Li, the instructors of HZAU-CHINA, during our presentation in the Central China iGEMer’s Meetup in Wuhan, China. Dr. Li suggested us to screen over the essential oil constituents using in silico before moving towards the cellular experiments (refer to the Human Practice Page for more information). In addition, we also realized that by incorporating in silico screening against MTNR1A and MTNR1B, we may possibly find molecules that can selectively interact with MTNR1A instead of MTNR1B, therefore addressing the concerns raised by the clinical psychiatrists in our human practice efforts.

To identify the putative active constituents in the Clove Pod and Caraway essential oils that account for the activation of melatonin receptors, we generated an AutoDock Vina ⁶-based pipeline to predict the binding affinity of the essential oil constituents against the known structure of melatonin receptors (PDB IDs: 6ME2 for MTNR1A and 7VH0 for MTNR1B). The result suggest that CTL 01-05-B-A05 is a highly potent binder to both MTNR1A and MTNR1B (Fig. 5A, upper panel, also see the Modeling page for more information), outperforming other ligands in binding affinity. Notably, It displays a clear preference for MTNR1A over MTNR1B. In contrast, melatonin was found to be a relatively weak binder towards MTNR1A/MTNR1B receptors, consistent with its role in promoting sleep and being naturally degraded during the sleep cycle. These findings validate the feasibility of the docking parameters employed. Then, we performed in silico screening of all constituents of Clove Pod ⁷ and Caraway ⁸ essential oils by molecular docking. Interestingly, the major constituents of Clove Pod oil were found to bind to both MTNR1A/ MTNR1B receptors. Only Eugenol exhibited a preference towards MTNR1A (Fig. 5A). Structural analysis revealed that the hydroxy group of Eugenol may form a hydrogen bond with the main chain oxgen of Ala104 in MTNR1A (Fig. 5B). More importantly, in vitro assay showed a significantly increased Nluc production in HEKMT cells treated with eugenol, while no Nluc induction was observed in HEKCTR cells (Fig. 5C). Together, these results demonstrated the feasibility of our drug-screening pipeline that incorporates both designer cell-based in vitro screening and molecular docking-based in silico screening, and also suggested that eugenol might serve as a promising selective agonist for MTNR1A.