Research ArticleCELL ENGINEERING

Smartphone-controlled optogenetically engineered cells enable semiautomatic glucose homeostasis in diabetic mice

See allHide authors and affiliations

Science Translational Medicine  26 Apr 2017:
Vol. 9, Issue 387, eaal2298
DOI: 10.1126/scitranslmed.aal2298

Wireless diabetes control? There’s an app for that

In an elegant feat of synthetic biology, Shao et al. were able to remotely control release of glucose-lowering hormones by engineered cells implanted into diabetic mice. These designer cells were transfected with optogenetic circuits, which enabled them to produce the hormones in response to far-red light. A smartphone could adjust far-red light intensity and duration with the help from a control box. Implanting hydrogel capsules containing both engineered cells and light-emitting diode light sources provided a semiautonomous system that maintained glucose homeostasis over several weeks in the diabetic mice. This study illuminates the potential of cell-based therapies.

Abstract

With the increasingly dominant role of smartphones in our lives, mobile health care systems integrating advanced point-of-care technologies to manage chronic diseases are gaining attention. Using a multidisciplinary design principle coupling electrical engineering, software development, and synthetic biology, we have engineered a technological infrastructure enabling the smartphone-assisted semiautomatic treatment of diabetes in mice. A custom-designed home server SmartController was programmed to process wireless signals, enabling a smartphone to regulate hormone production by optically engineered cells implanted in diabetic mice via a far-red light (FRL)–responsive optogenetic interface. To develop this wireless controller network, we designed and implanted hydrogel capsules carrying both engineered cells and wirelessly powered FRL LEDs (light-emitting diodes). In vivo production of a short variant of human glucagon-like peptide 1 (shGLP-1) or mouse insulin by the engineered cells in the hydrogel could be remotely controlled by smartphone programs or a custom-engineered Bluetooth-active glucometer in a semiautomatic, glucose-dependent manner. By combining electronic device–generated digital signals with optogenetically engineered cells, this study provides a step toward translating cell-based therapies into the clinic.

INTRODUCTION

Continued progress in cell phone development has transformed smartphones into powerful mobile computers equipped with sophisticated cameras, ultrahigh-resolution crystal displays, powerful microprocessors, and multiple capacitive sensors (1). Life in the 21st century has become increasingly smartphone-dependent, with most users in modern civilizations storing large amounts of important personal information on their phone, including bank accounts, health data, and instant Global Position System (GPS) locations. With some data being continuously synchronized with cloud storage online, smartphones provide immediate and on-demand availability of selected information at any geographical place.

The use of smartphones for mobile health (mHealth)—defined as the practice of medicine supported by portable diagnostic devices to allow easy and accurate characterization of health and diseases—is shifting health care models toward increasingly patient-centric designs (2, 3). Initially exclusively based on high-resolution imaging and microscopy for field testing, diagnostics with smartphones have now changed focus to advanced mHealth devices, combining the data analysis capacity of smartphones with the rapid, simple, and robust testing procedures of point-of-care technologies (POCTs) (1, 2, 4). In a typical telemedicine scenario, a patient would use minimally invasive point-of-care diagnostics to quickly and accurately measure the expression of a target disease marker at home with one drop of blood, upload this information onto an mHealth-enabled smartphone, and share the data with doctors or other qualified medical caretakers to consult regarding further therapeutic and/or preventive interventions (46).

Diabetes mellitus is a complex and progressive disease characterized by chronically deregulated blood glucose concentrations affecting at least 415 million people worldwide (7) and generating a global market of at least $10 billion for blood glucose–monitoring POCTs (4). In diabetic patients, efficient control of blood glucose can only be achieved with a strict regimen of food control along with lifelong injections of insulin [for insulin-deficient type 1 diabetic (T1D) patients] or glucagon-like peptide-1 (GLP-1) analogs [for insulin-resistant type 2 diabetic (T2D) patients] within daily to weekly periods (811). Currently, although personal glucometers providing immediate, accurate, and cost-effective tracking of glycemia greatly enhance the precision of antidiabetic medication, total therapeutic efficacy is still limited by the lack of automation between diagnostics and therapy (12). Although recent breakthroughs in biomedical research have resulted in a variety of attractive cell-based theranostic (combined diagnostic and therapeutic) solutions including stem cell–derived pancreatic β cells (13) and pH-triggered designer human cells (14), biological cells are generally unable to achieve the digital sensing precision that is achieved by electronic sensors. Therefore, a technological platform enabling engineered human cells to be controlled by fully digital electronic signals would maximally exploit the biomedical potential of cell-based medicines by combining diagnostic and therapeutic precision.

Here, we used a multidisciplinary design principle coupling electrical engineering, software development, and synthetic biology to create a technological interface that allows wireless regulation of engineered cell activity with an Android-operating smartphone. Using a biocompatible far-red light (FRL) source as a carrier signal, we translated electronic commands into biological optogenetic responses by engineering two FRL-triggered engineered cell systems based on the bacterial light-activated cyclic diguanylate monophosphate (c-di-GMP) synthase BphS and the c-di-GMP–specific phosphodiesterase YhjH to regulate drug production in different mammalian cell types and diabetic mice. To remotely control engineered cells with conventional POCT diagnostic devices, we custom-engineered a Bluetooth-active glucometer to interface with an intelligent home server, automatically triggering FRL activity and insulin or shGLP-1 [short variant of human GLP-1 (8)] production in diabetic mice according to user-defined glycemic thresholds (Fig. 1A). The principles underlying the modular design of this semiautomatic platform can be applied to other metabolic diseases and could advance the progress of cell-based treatments toward the clinic.

Fig. 1. Schematic of a programmable smartphone-regulated electronic control system.

(A) Abstract diagram showing smartphone-controlled engineered cells enabling semiautomatic point of care for combating diabetes. (B) Connectivity map of sender, transmitter, and receiver devices. An intelligent device termed a SmartController (receiver unit) is programmed to receive signals sent by a smartphone (sender unit) through the GSM network and accordingly activates different electronic devices. The SmartController consists of four core modules: (i) an Android-based smartphone input module running a custom-designed app; (ii) a conventional internet communication base station signal transduction module; (iii) a programmable electric circuit within the SmartController containing an ac/dc converter, a wireless signal receiver, a 32-bit embedded MPU chip, and a customizable number of relay drivers and units that translate the smartphone signal into different electrical activities; and (iv) commercial electronics. (C) Detailed electric circuit diagram of the SmartController 1.0 version. An ①-② ac/dc converter enables power supply from different types of line currents and is capable of supporting ③ the wireless signal processor, ④ the embedded MPU chip, ⑤ the relay drivers, and ⑥ relay units, as well as different electronic devices used in biomedicine such as ⑦ a physiotherapy lamp or ⑧ light-emitting diode (LED) arrays.

RESULTS

SmartController 1.0—An intelligent electronic home server receiving and processing wireless input signals

Inspired by the Smart Home project creating completely wireless homes furnished with smartphone-regulated electrical appliances (15), we built an intelligent electronic home server box (SmartController) capable of regulating various electrical appliances according to smartphone-transmitted remote commands (Fig. 1B). The key constituents of the SmartController are a 32-bit embedded microprocessor unit (MPU) chip that automatically integrates the reception, processing, and execution of electronic signals in a programmable manner; an alternating current/direct current (ac/dc) converter to assure high-voltage ac current at the input and output ports but low-voltage dc current within the mainboard interior; and a receiver unit that translates wireless fidelity (Wi-Fi) signals from an internal antenna into MPU-compatible messages (Fig. 1C).

To enable remote control of SmartController activity with a smartphone, we developed the “ECNU-TeleMed” mobile application (app) using the Android Software Development Kit containing different pre-set algorithms to regulate the activity of SmartController-connected electronic devices (fig. S1). Whenever the smartphone is connected to the internet via the Global System for Mobile Communications (GSM) network, the ECNU-TeleMed app automatically generates a private internet protocol (IP) address that is identical to or compatible with that of the Wi-Fi–connected SmartController server, enabling the smartphone to remotely send specific executive commands to the MPU from any GSM-covered geographical location in the world.

Smartphone-regulated control of mammalian engineered cells using FRL

To transform smartphone-triggered digital electronic signals into well-defined biological responses, optogenetic systems are the tools of choice because light can be easily generated by electronic commands and synchronously trigger many unique biological processes including circadian rhythms (16), cardiac (17, 18), neural (1921), and motor (22) activities or gene expression (Fig. 2A) (8, 2325). Although several potent light-triggered transcriptional gene switches have been characterized in mammalian cells over the past 5 years (8, 2427), important parameters such as light source, illumination strength, and stimulation frequency must be carefully chosen in regard to biomedical applications involving living animals (8, 25). For example, continuous exposure of mammalian cells to blue light is often cytotoxic (26), reducing the gene expression capacity of transfected cells (fig. S2, A and B). In this regard, FRL is a biocompatible light source that has been exploited for several decades by physiotherapy infrared lamps because of its ability to deeply penetrate into tissues.

Fig. 2. Smartphone-regulated gene expression in engineered mammalian cells.

(A) Custom-designed smartphone app controlling the SmartController activities. The ECNU-TeleMed app is programmed to regulate the activity of a custom-designed mammalian cell culture–compatible far-red LED array through the SmartController 1.0 interface, enabling user-defined determination of LED brightness and illumination time. (B) FRL-inducible mammalian engineered cells version 1 (FRL-v1). FRL (~730 nm) activates the engineered bacterial photoreceptor BphS, which converts guanylate triphosphate (GTP) into c-di-GMP. c-di-GMP binds and activates the endogenous human STING pathway. Activated STING triggers a tank-binding kinase 1 (TBK1)–dependent signaling cascade, resulting in nuclear translocation of phosphorylated interferon regulatory factor 3 (IRF3) and activation of different cognate promoters engineered to contain different configurations of IRF3 binding sites (PFRL1.x). Intracellular c-di-GMP production can be modulated by varying the amount of c-di-GMP–specific phosphodiesterase (YhjH). Solid lines, signaling incorporated by the engineered gene circuit; dashed lines, endogenous signaling. (C) Illumination-dependent FRL-v1 activity. Human embryonic kidney (HEK) 293 cells were transfected with pWS46 (PhCMV-BphS-2A-YhjH-pA), pSTING (PhCMV-STING-pA), and pWS67 (PFRL1.6-SEAP-pA) in a 10:1:10 (w/w/w) ratio, illuminated with FRL for 4 hours at different light intensities (0 to 2000 μW/cm2), and SEAP expression in the culture supernatant was scored after 72 hours. (D) Exposure time–dependent FRL-v1 activity. HEK-293 cells were transfected with pWS46, pSTING, and pWS67 and illuminated with FRL (1 mW/cm2) for different time periods, and at 72 hours after illumination, SEAP expression in the culture supernatant was profiled. (E) FRL-v1–controlled SEAP expression in different mammalian cell lines. Different mammalian cell lines were transfected with pWS46, pSTING, and pWS67 and exposed to FRL (1 mW/cm2; 730-nm LED) for 4 hours every 24 hours, and SEAP expression in the culture supernatant was scored at 48 hours after the first illumination. (F) Chromatic specificity of FRL-v1–controlled SEAP expression. pWS46/pSTING/pWS67-cotransfected HEK-293 cells were illuminated (1 mW/cm2) for 2 hours with different wavelengths of light (400 to 730 nm), and SEAP expression in the culture supernatant was scored at 48 hours after illumination. (G) Reversibility of FRL-v1 triggered transgene expression. pWS46/pSTING/pWS67-cotransfected HEK-293 cells were either kept in the dark (OFF) or illuminated with FRL (1 mW/cm2) for 20 min (ON), and SEAP production was scored every 6 hours for 72 hours. The culture medium was exchanged every 24 hours with concomitant reversal of illumination status. (H) FRL-v1–compatible light sources. pWS46/pSTING/pWS67-cotransfected HEK-293 cells were illuminated (1 mW/cm2) with the LED array or a physiotherapy lamp for 4 hours every 24 hours, and SEAP expression in culture supernatant was scored every 24 hours. (I) FRL-inducible mammalian engineered cells version 2 (FRL-v2). FRL (~730 nm) activates the engineered bacterial photoreceptor BphS, which converts GTP into c-di-GMP. Different synthetic mammalian FRTAs were engineered by assembling BldD (a c-di-GMP–binding domain derived from sporulating actinomycete bacteria), p65 (the NF-κB–transactivating domain), and VP64 (a tetramer of the herpes simplex virus–derived VP16 activation domain) into different protein configurations. Increased cytosolic c-di-GMP production triggers FRTA dimerization and binding to different chimeric promoters PFRL2.x engineered to harbor different configurations of BldD-specific operator DNA sites. (J) Illumination-dependent FRL-v2 activity. HEK-293 cells were transfected with pWS46 (PhCMV-BphS-2A-YhjH-pA), pGY32 (PhCMV-FRTA3-pA; FRTA3, p65-VP64-NLS-BldD), and pXY34 (PFRL2.13a-SEAP-pA) in a 1:1:1 (w/w/w) ratio and illuminated with FRL for 4 hours at different light intensities (0 to 2000 μW/cm2), and SEAP expression in the culture supernatant was scored after 72 hours. (K) Exposure time–dependent FRL-v2 activity. HEK-293 cells were transfected with pWS46, pGY32, and pXY34 and illuminated with FRL (1 mW/cm2) for different time periods, and at 72 hours after illumination, SEAP expression in the culture supernatant was profiled. (L) FRL-v2–controlled SEAP expression in different mammalian cell lines. Different mammalian cell lines were transfected with pWS46, pGY32, and pXY34 and exposed to FRL (1 mW/cm2; 730-nm LED) for 4 hours every 24 hours, and SEAP expression in the culture supernatant was scored at 48 hours after the first illumination. (M) Chromatic specificity of FRL-v2–controlled SEAP expression. pWS46/pGY32/pXY34-cotransfected HEK-293 cells were illuminated (1 mW/cm2) for 2 hours with different wavelengths of light (400 to 730 nm), and SEAP expression in the culture supernatant was scored at 48 hours after illumination. (N) Reversibility of FRL-v2 triggered transgene expression. pWS46/pGY32/pXY34-cotransfected HEK-293 cells were either kept in the dark (OFF) or illuminated with FRL (1 mW/cm2) for 20 min (ON), and SEAP production was scored every 6 hours for 72 hours. The culture medium was exchanged every 24 hours with concomitant reversal of illumination status. (O) FRL-v2–compatible light sources. pWS46/pGY32/pXY34-cotransfected HEK-293 cells were illuminated (1 mW/cm2) with the LED array or a physiotherapy lamp for 4 hours every 24 hours, and SEAP expression in the culture supernatant was scored every 24 hours. All the data are means ± SD; n = 3 independent experiments. All individual-level data in table S2.

The BphS is an FRL-activated c-di-GMP synthase retaining full functionality when ectopically expressed in mammalian cells (2830). Capitalizing on the identification of human stimulator of interferon genes (STING) as an intracellular cyclic dinucleotide sensor that triggers endogenous interferon promoters to activate innate immune responses (31, 32), we engineered a variety of putative FRL-dependent STING-responsive synthetic promoters (PFRL1.x) containing different combinations of response elements for human interferon-β and interferon-stimulated genes (table S1). Initial experiments coexpressing BphS, STING, and pWS32 [PFRL1.1-SEAP-pA; PFRL1.1, (ISRE)5-Pmin] showed intact PFRL1.1 promoter activity in human cells, however, with saturating basal secreted embryonic alkaline phosphatase (SEAP) expression in the absence of FRL (fig. S3A). Coexpression of an additional c-di-GMP phosphodiesterase YhjH resulted in a reduction of basal intracellular c-di-GMP production into an optimal window (fig. S3, A and B) that enabled FRL-triggered SEAP expression (Fig. 2B).

Optimization of the STING-dependent FRL-triggered optogenetic system (FRL-v1)

To engineer an optimal FRL-triggered gene switch permitting minimal basal transgene expression in the absence of light and maximal induction ratios in the presence of light, we optimized the STING-dependent FRL-triggered optogenetic system (FRL-v1) by testing different constitutive promoters for driving BphS and YhjH expression (fig. S3C), different amounts of ectopic STING (fig. S3D), and different variants of STING-responsive promoters (fig. S3E). We found that a plasmid composed of pWS46 (PhCMV-BphS-P2A-YhjH-pA), pSTING (PhCMV-STING-pA), and pWS67 [PFRL1.6-SEAP-pA; PFRL1.6, (hIFN-RE)-(ISRE)3-Pmin-40 bp] in a 10:1:10 (w/w/w) ratio showed optimal FRL-dependent induction profiles (fig. S4, A and B). Control experiments of SEAP expression verified that neither prolonged FRL illumination nor the ectopic presence of FRL-v1 constituents influenced the overall gene expression capacity of the engineered human cells (fig. S5, A to C). The detailed performance of transgene expression kinetics of FRL-v1 was further studied. The data showed that the transgene expression triggered by FRL was illumination- and exposure time–dependent (Fig. 2, C and D). The FRL-v1 was functional in different mammalian cell lines except for a few cell lines (Fig. 2E), probably due to the compatibility of the FRL-v1 to host cells. The different colors of light-triggered transgene expression of FRL-v1 revealed that this optogenetic device exhibited the best SEAP induction folds under FRL (730 nm) illumination and the second best under red light (630 nm) illumination (Fig. 2F). Moreover, FRL-v1 also demonstrated fully reversible transgene activation kinetics (Fig. 2G) and was compatible to various FRL sources (Fig. 2H).

Design and construction of an orthogonal FRL-triggered optogenetic system (FRL-v2)

Although the gene expression profiles of the optimized FRL-v1 system met almost all criteria for an ideal mammalian optogenetic system (Fig. 2, C to H, and fig. S6), its strict dependence on the endogenous STING pathway renders the system vulnerable to potential cross-talk with important endogenous processes of human immunity, such as surges of cyclic guanosine monophosphate?"adenosine monophosphate (cGAMP) (fig. S7A) and/or temporal interferon-β activation (fig. S7B). To minimize the potential for non–FRL-mediated effects during in vivo applications, we developed a fully orthogonal (a configuration that avoids signaling cross-talk with endogenous pathways) FRL-triggered optogenetic system based on the Streptomyces coelicolor–derived transcription factor BldD (33, 34). In mammalian cells, ectopically expressed BldD dimerizes in the presence of c-di-GMP into a protein conformation capable of binding DNA operator sequences that contain cognate bldM or whiG motifs. To engineer an FRL-triggered transcriptional gene switch, we created different synthetic mammalian transactivators (FRTAs) by assembling BldD, p65 [65-kDa transactivator subunit of nuclear factor κB (NF-κB)], VP64 (tetrameric core of herpes simplex virus–derived transactivation domain), and HSF1 (heat shock factor 1) into various configurations, constructing multiple FRTA-specific chimeric promoter (PFRL2.x) variants (Fig. 2I, fig. S8, and table S1). We found that a combination of FRTA3 (pGY32; PhCMV-FRTA3-pA; FRTA3, p65-VP64-NLS-BldD) and PFRL2.13a [pXY34; PFRL2.13a-SEAP-pA; PFRL2.13a, pA-(whiG)3-PhCMVmin] showed the best FRL-triggered BphS/YhjH-mediated transcriptional responses (Fig. 2I). When compared to the STING-dependent FRL-v1 system, the orthogonal FRL-v2 system was unresponsive to cGAMP and/or interferon-β (fig. S9), showed higher FRL-triggered induction ratios (Fig. 2, J versus C, K versus D, and figs. S10 versus S6), was functional in a greater number of mammalian cell types (Fig. 2, L versus E), and was more specific to FRL (Fig. 2, M versus F), all while retaining similar spatiotemporal activation precision (fig. S11), the inertness of the system constituents (fig. S12), and fully reversible transgene activation kinetics (Fig. 2N). Collectively, its compatibility with many mammalian cell types (Fig. 2L) and various FRL sources (Fig. 2O) qualifies the FRL-v2 as a most suitable receiver device for smartphone-transmitted FRL signals from the SmartController.

Smartphone regulation of engineered cells implanted into diabetic mice

To validate the feasibility of an ECNU-TeleMed remote-controlled therapeutic scenario in vivo (Fig. 3A), we subcutaneously implanted microencapsulated FRL-v2 transgenic human cells (fig. S13, A to E) under the dorsum of wild-type mice and observed smartphone-triggered FRL-specific SEAP accumulation in the animals’ bloodstreams (Fig. 3B) with rapid expression kinetics (Fig. 3C). Continuous illumination of the mice at relatively mild exposure strengths [4 hours with FRL (25 mW/cm2)] was sufficient to trigger high induction ratios in vivo (Fig. 3, D and E).

Fig. 3. Smartphone-regulated transgene expression in diabetic mice.

(A) Experimental setting of SmartController 1.0–assisted engineered cell therapy. The ECNU-TeleMed app is programmed to remotely activate a SmartController-powered physiotherapy lamp with FRL-v2–activating parameters (0 to 25 mW/cm2; 730 nm), enabling smartphone-controlled illumination of mice that harbor different types of FRTA3/PFRL2.13a transgenic implants. (B to E) SEAP expression using alginate-poly-(l-lysine)-alginate microcapsule implants. A total of 2 × 106 microencapsulated pWS46/pGY32/pXY34 transgenic HEK-293 cells (200 cells per capsule) were subcutaneously implanted under the dorsum of wild-type mice and illuminated through the SmartController 1.0 for different durations (typically 4 hours per day) at different light intensities (0 to 25 mW/cm2). Unless otherwise stated, SEAP expression in the animals’ serum was profiled at 48 hours after implantation. (F to I) SEAP expression using hollow fiber implants. Pairs of 2.5-cm hollow fibers containing a total of 2 × 106 pWS46/pGY32/pXY34 transgenic HEK-293 cells were subcutaneously implanted under the dorsum of wild-type mice and illuminated through the SmartController 1.0 for different durations (typically 4 hours per day) at different light intensities (0 to 25 mW/cm2). Unless otherwise stated, SEAP expression in the animals’ serum was profiled at 48 hours after implantation. (J to M) SmartController 1.0–assisted engineered cell therapy in T1D mice. Pairs of 2.5-cm hollow fibers containing a total of 2 × 106 pWS46/pGY32/pWS213 (PFRL2.13a-EGFP-P2A-mINS-pA) transgenic HEK-293 cells were subcutaneously implanted under the dorsum of T1D mice and illuminated (25 mW/cm2) through the SmartController 1.0 for 4 hours per day. Control mice either received FRL but no implants (−, +), received implants but no FRL (+, −), or received no treatment (−, −). Blood insulin (J), blood glucose (K), and intraperitoneal glucose tolerance (L and M) were analyzed 48 hours after implantation. (M) Area under the curve (AUC) analysis of the intraperitoneal glucose tolerance test (IPGTT) performed in (L). (N to S) SmartController 1.0–assisted engineered cell therapy in T2D mice. Pairs of 2.5-cm hollow fibers containing a total of 2 × 106 pWS46/pGY32/pWS212 (PFRL2.13a-shGLP-1-pA) transgenic HEK-293 cells were subcutaneously implanted under the dorsum of T2D mice and illuminated (25 mW/cm2) through the SmartController 1.0 for 4 hours per day. Control mice either received FRL but no implants (− , +), received implants but no FRL (+, −), or received no treatment (−, −). Blood GLP-1 (N), blood glucose (O), intraperitoneal glucose tolerance (P and Q), insulin tolerance (R), and insulin resistance (S) were analyzed 48 hours after implantation. All the data (B to S) are means ± SEM; statistics by two-tailed t test, n = 6 mice (B to M) and n = 5 mice (N to S). P values were calculated by Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. P values in table S3 (L, P, and R). All individual-level data in table S2.

Limitations of the microcapsule implants include not only the lack of uniformity of FRL exposure because of their poor spatial mobilization and localization control after implantation but also relatively high basal SEAP expression observed in the engineered cells’ inactive state (Fig. 3, B to E). Therefore, we encapsulated FRL-v2 transgenic engineered cells into 2.5-cm hollow fiber implants (fig. S13, F to J) and implanted pairs of fibers subcutaneously in mice. Mice receiving these macroencapsulated engineered cells showed improved SEAP expression when treated with the same ECNU-TeleMed program (Fig. 3, F to I), confirming superior FRL-dependent regulation performance with this type of implant. T1D mice implanted with hollow fibers containing isogenic FRL-dependent, insulin-expressing engineered cells (fig. S14) showed a rapid restoration of homeostatic insulin (Fig. 3J) and blood glucose (Fig. 3K) as well as improved glucose tolerance (Fig. 3, L and M) with ECNU-TeleMed–assisted remote control. Similarly, analogously treated T2D mice implanted with hollow fibers containing FRL-dependent, shGLP-1–expressing engineered cells (Fig. 3N and fig. S15) also showed rapid restoration of homeostatic blood glucose (Fig. 3O), improved glucose (Fig. 3, P and Q), and insulin tolerance (Fig. 3R), as well as reduced insulin resistance (Fig. 3S). Control mice received either FRL without implants, to exclude unexpected side effects of the FRL, or implants without FRL, to eliminate pleiotropic induction in engineered cells carrying implants (Fig. 3, J, K, N, O, and S). In addition, in contrast to previous reports of metabolic side effects elicited by endogenous light-dependent processes (35), FRL did not influence glucose or insulin tolerance in our studies (fig. S16). Collectively, these experiments confirm the precise, robust, and tunable remote control of optogenetically engineered cells using the ECNU-TeleMed app, enabling rapid initiation of effective therapy in diabetic mice.

Therapeutic efficacy of smartphone-assisted optogenetic regulation of blood glucose

To study the therapeutic efficacy of the smartphone-assisted FRL-inducible optogenetic device for combating diabetes, we produced two different stable cell lines, HEKFRL-SEAP-P2A-mINS (Fig. 4A) and HEKFRL-shGLP-1-P2A-SEAP (Fig. 4B), which are stably transgenic for pYH88 (ITR-PhCMV-BphS-P2A-YhjH-P2A-P65-VP64-BldD-P2A-mCherry-pA:PmPGK-PurR-pA-ITR) and pWS251 (ITR-PFRL2.13a-SEAP-P2A-mINS-pA:PmPGK-ZeoR-P2A-EGFP-pA-ITR) or pYH88 and pWS252 (ITR-PFRL2.13a-shGLP-1-P2A-SEAP-pA:PmPGK-ZeoR-P2A-EGFP-pA-ITR). SEAP expression could be activated by FRL in the monoclonal HEKFRL-SEAP-P2A-mINS cell line (clone no. 14) for up to 15 days in vitro (Fig. 4C). When implanted into wild-type mice, FRL triggered SEAP expression in the bloodstream over 15 days (Fig. 4, D and E). To demonstrate antidiabetic efficacy, T1D and T2D mice were implanted with HEKFRL-SEAP-P2A-mINS and HEKFRL-shGLP-1-P2A-SEAP cells, respectively. FRL triggered a sustainable expression of insulin (Fig. 4F) and shGLP-1 (Fig. 4G) in the bloodstream within 15 days, which was sufficient for significant restoration of glucose homeostasis in each diabetic mouse model (Fig. 4, H and I). To confirm the time resolution of the optogenetic device for controlling blood glucose homeostasis in diabetic mice, insulin-deficient T1D mice were implanted with HEKFRL-SEAP-P2A-mINS cells and illuminated by FRL for up to 150 min. Nondiabetic blood glucose levels (<10 mM) could be stably reached within less than 2 hours after illumination without causing hypoglycemic side effects (Fig. 4J).

Fig. 4. The therapeutic efficacy of a smartphone-regulated optogenetic construct in diabetic mice.

(A and B) Design and construction of the stable cell lines. FRL-induced SEAP expression of different transgenic cell clones [HEKFRL-SEAP-P2A-mINS (A) and HEKFRL-shGLP-1-P2A-SEAP (B)]. HEK-293 cells were stably transfected with pYH88 and pWS251 in (A) or pYH88 and pWS252 in (B), and 60 randomly selected cell clones were profiled for their FRL-stimulated SEAP expression performance by the SmartController 1.0 (1 mW/cm2; 730-nm LED) for 4 hours per day. SEAP expression in the culture supernatant was scored 72 hours after the first illumination. The blue frame marks the best-in-class cell clone chosen for the following experiments. (C) Characterization of SEAP expression in HEKFRL-SEAP-P2A-mINS cells. HEKFRL-SEAP-P2A-mINS cells (1 × 105) were exposed to FRL (1 mW/cm2; 730-nm LED) for 4 hours each day, and the SEAP expression in the culture supernatant was scored for up to 15 days after the first illumination. Data (A to C) are means ± SD; n = 3 independent experiments. (D and E) Characterization of SEAP expression profiles with microcapsule (D) or hollow fiber (E) implants in wild-type (WT) mice. Microencapsulated HEKFRL-SEAP-P2A-mINS cells (4 × 106) (D) or two pairs of 2.5-cm hollow fibers containing 4 × 106 HEKFRL-SEAP-P2A-mINS cells (E) were subcutaneously implanted under the dorsum of wild-type mice and illuminated (25 mW/cm2) through the SmartController 1.0 for 4 hours per day, and the SEAP expression in the animals’ serum was profiled for up to 15 days after implantation. (F to I) SmartController 1.0–assisted engineered cell therapy in T1D and T2D mice. (F and G) FRL-induced insulin expression and glycemic control in T1D mice. Two pairs of 2.5-cm hollow fibers containing 4 × 106 HEKFRL-SEAP-P2A-mINS cells were subcutaneously implanted under the dorsum of T1D mice and illuminated (25 mW/cm2) through the SmartController 1.0 for 4 hours per day, and (F) insulin expression and (G) blood glucose concentration in the animals’ serum were profiled for up to 15 days after implantation. (H and I) FRL-induced shGLP-1 expression and glycemic control in T2D mice. Two pairs of 2.5-cm hollow fibers containing 4 × 106 HEKFRL-shGLP-1-P2A-SEAP cells were subcutaneously implanted under the dorsum of T2D mice and illuminated (25 mW/cm2) through the SmartController 1.0 for 4 hours per day, and (H) shGLP-1 expression and (I) glycemia in the animals’ serum were profiled for up to 15 days after implantation. (J) Time resolution of the optogenetic device for controlling blood glucose homeostasis in T1D mice. Microencapsulated HEKFRL-SEAP-P2A-mINS cells (4 × 106) were subcutaneously implanted under the dorsum of T1D mice and illuminated (25 mW/cm2) through the SmartController 1.0 for up to 150 min, and the blood glucose concentration was recorded every 15 min. All the data (E to J) are means ± SEM; statistics by two-tailed t test, n = 6 mice. P values were calculated by Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. P values in table S3. All individual-level data in table S2.

SmartController 2.0—Smartphone-assisted remote control of implanted HydrogeLED

Although the previous experiments validate the feasibility of smartphone-regulated engineered cell therapy, two major disadvantages restricting patient compliance are (i) its strict reliance on subcutaneous implants in the vicinity of shaved skin, risking pleiotropic activation by conventional “white” light (fig. S17) or sunlight, and (ii) the strict requirement for a fixed therapy location such as a physiotherapy lamp operating room, which restricts a patient’s mobility during treatment. To adapt the ECNU-TeleMed–assisted therapy to improve these features, we engineered an enhanced version of the SmartController device by replacing the physiotherapy lamp with an autonomous electromagnetic emission circuit (EEC) containing a low-dropout (LDO) linear regulator chip, a power amplifier circuit (0 to 24 V), and a homemade transmitting circular coil device capable of generating electromagnetic sine wave signals with 180 kHz (fig. S18). In this SmartController 2.0 system, the EEC functions as a wireless charger, using electromagnetic induction to power receiver coils containing electronic devices (Fig. 5, A and B, and fig. S19). To engineer customized receiver electronic devices compatible with a SmartController 2.0–dependent therapy, we tailor-designed an implant architecture named HydrogeLED (fig. S20A) by incorporating engineered cell–carrying alginate hydrogel and wirelessly powered FRL LEDs into 15-mm coiled-coil cylinders (fig. S20, B and C). HydrogeLED devices retained full functionality when immersed in liquid solutions (fig. S20D) and enabled mice implanted intraperitoneally with the devices to freely move within an area of constant wireless power supply, with engineered cell instant activity remotely regulated by ECNU-TeleMed–transmitted signals (Fig. 5B). After initial control experiments confirmed the rapid, precise, and tunable regulation of SEAP production from smartphone-regulated HydrogeLED implants both in vitro (fig. S21) and in vivo (Fig. 5, C to E), the immediate antidiabetic efficacy of FRL-dependent engineered cell therapy was similarly validated in T1D (Fig. 5, F to I) and T2D mice (Fig. 5, J to O) using the SmartController 2.0 system.

Fig. 5. Design and validation of the SmartController 2.0 version.

(A) Electric circuit diagram of SmartController 2.0. The key innovative feature of the 2.0 version is the integration of an ① electromagnetic field regulator containing ② a LDO linear regulator chip and ③ a transmitter coil driven by the SmartController ④ relay units. The electromagnetic field regulator containing transmitter coil enables wireless power supply for ⑤ custom-designed receiver coil–containing electronic devices. (B) Experimental setting of SmartController 2.0–assisted engineered cell therapy. The 2.0 version of the SmartController system allows the ECNU-TeleMed app to remotely control the activity of wireless electronic devices. In a therapeutic scenario, a smartphone can be used to regulate transgene activity in mice implanted with custom-designed alginate hydrogel implants containing both FRL-v2 transgenic engineered cells and a receiver coil connected to a resonance capacitor and far-red LEDs (HydrogeLED). (C to E) SmartController 2.0 regulating SEAP expression in vivo. Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pXY34 transgenic HEK-293 cells were intraperitoneally implanted into wild-type mice and illuminated through the SmartController 2.0 for different durations per day at different light intensities. Unless otherwise stated, SEAP expression in the animals’ serum was profiled at 48 hours after implantation. (F to I) SmartController 2.0–assisted engineered cell therapy in T1D mice. Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pWS213 transgenic HEK-293 cells were intraperitoneally implanted into T1D mice and illuminated (25 mW/cm2) through the SmartController 2.0 for 4 hours per day. Control mice either received FRL but no implants (−, +), received implants but no FRL (+, −), or received no treatment (−, −). Blood insulin (F), blood glucose (G), and intraperitoneal glucose tolerance (H and I) were analyzed at 48 hours after implantation. (I) AUC analysis of the IPGTT performed in (H). (J to O) SmartController 2.0–assisted engineered cell therapy in T2D mice. Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pWS212 transgenic HEK-293 cells were intraperitoneally implanted into T2D mice and illuminated (25 mW/cm2) through the SmartController 2.0 for 4 hours per day. Control mice either received FRL but no implants (−, +), received implants but no FRL (+, −), or received no treatment (−, −). Blood GLP-1 (J), blood glucose (K), intraperitoneal glucose tolerance (L and M), insulin tolerance (N), and insulin resistance (O) were analyzed at 48 hours after implantation. All the data (C to O) are means ± SEM; statistics by two-tailed t test, n = 6 mice. P values were calculated by Student’s t test. **P < 0.01, ***P < 0.001 versus control. P values in table S3 (H, L, and N). All individual-level data in table S2.

SmartController 3.0—Semiautomatic point-of-care digitalized glucose sensing

Point-of-care diagnostics enable rapid, accurate, and self-managed determination of disease metabolite concentrations using as little as one drop of blood at hospitalization-independent locations (4). Blood glucometers are typical point-of-care devices capable of measuring blood glucose concentration based on a simple and cell-free electrochemical reaction (36). To enable our SmartController-regulated antidiabetic engineered cells to sense blood glucose, we engineered a wireless glucometer by integrating a Bluetooth transmitter into the electric circuit of a commercial point-of-care glucometer (fig. S22), which allowed remote transmission of glycemic values to the enhanced SmartController 3.0 version engineered with a 16-bit MSP430 ultralow-power microcontroller chip (Fig. 6A). This microcontroller chip could be programmed to automatically translate different user-defined glycemic thresholds into different FRL illumination strengths (fig. S23), enabling semiautomatic but fully self-sufficient glycemia-dependent activation of antidiabetic engineered cells (Fig. 6A). With the all-wireless SmartController 3.0 system, digitalization of glucose sensing could be achieved to control SEAP (Fig. 6B), insulin (Fig. 6C), and shGLP-1 production (Fig. 6D) and could be threshold-dependently triggered with aqueous glucose solution (Fig. 6B) and whole murine and human blood samples (Fig. 6, C and D) applied to a glucometer in vitro or with blood samples from mice implanted with HydrogeLEDs (Fig. 6E), confirming the high precision achievable with this engineered cell–regulating electronic interface.

Fig. 6. SmartController 3.0 enables semiautomatic control of blood glucose in diabetic mice.

(A) Electric circuit diagram of SmartController 3.0. The 3.0 version of SmartController integrates ① a Bluetooth receiver and ② a liquid crystal display to enable simultaneous reception of different types of wireless signals. Thus, ③ the microcontroller unit can be programmed to interpret data transmitted from ④ a custom-designed glucometer containing ⑤ a Bluetooth transmitter to automatically trigger different illumination intensities of ⑥ remotely controlled receiver LEDs according to user-specified glycemic thresholds. All signal processing data are synchronized with the ECNU-TeleMed app on a smartphone to enable real-time surveillance and optional intervention by humans. (B to D) Digitalization of glucose sensing in vitro. The SmartController 3.0 software is programmed with four glycemic thresholds (GTs) (that is, GT1, <6.1 mM; GT2, 6.1 to 11.1 mM; GT3, 11.1 to 16.8 mM; and GT4, >16.8 mM), each triggering a different illumination intensity of the receiver LEDs (GT1, OFF; GT2, 0.2 mW/cm2; GT3, 1.0 mW/cm2; GT4, 5.0 mW/cm2) for typical illumination durations of 4 hours per day. (B) Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pXY34 transgenic HEK-293 cells were cultured in different tissue culture plates, and SEAP expression remotely triggered by aqueous solutions of different glucose concentrations was scored at 48 hours after glucometer-transmitted illumination. (C) Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pWS213 transgenic HEK-293 cells were cultured in different tissue culture plates, and insulin expression remotely triggered by different groups of blood samples collected from different healthy and diabetic mice was scored at 48 hours after glucometer-transmitted illumination. (D) Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pWS212 transgenic HEK-293 cells were cultured in different tissue culture plates, and shGLP-1 expression remotely triggered by different groups of blood samples collected from different healthy and diabetic human donors was scored at 48 hours after glucometer-transmitted illumination. All in vitro data in (B) to (D) are means ± SD; n = 3 independent experiments. P values were calculated by Student’s t test. (E) Digital glucose sensing in vivo. Custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pXY34 transgenic HEK-293 cells were intraperitoneally implanted into wild-type or T2D mice, and at 48 hours after transplantation, SEAP expression remotely triggered by each animal’s own blood was scored. Data are means ± SEM; n = 6 mice. P values were calculated by Student’s t test. (F to I) Semiautomatic regulation of blood glucose homeostasis in diabetic mice. (F) Monitoring schedule. Fasted T2D mice were implanted with custom-designed HydrogeLED implants containing 2 × 106 pWS46/pGY32/pWS212 transgenic HEK-293 cells, and each animal’s instant glycemia was analyzed by the custom-designed glucometer. The efficacy (G) of glucose-triggered shGLP-1 production (H) was profiled every 24 hours for 3 days. All the data in (G) and (H) are means ± SEM; statistics by two-tailed t test, n = 5 mice. P values were calculated by Student’s t test. (I) Corresponding LED intensity activating the engineered cells, as reported on the smartphone. All individual-level data in table S2.

In the theranostic scenario of automatically regulating blood glucose homeostasis using remotely controlled engineered cell implants, the possibility for a human practitioner to intervene at any time must be provided in case of unforeseeable errors. Bluetooth pairing enables fast and stable short-distance point-to-point connections, even during mobile network interruptions, and therefore, the SmartController 3.0 system was fine-tuned to allow instant Bluetooth-mediated synchronization of glycemic and FRL data with the ECNU-TeleMed app on a user’s smartphone, enabling real-time surveillance over engineered cell states (Fig. 6A and fig. S22). As an example, we set up a proof-of-principle experiment showing semiautomatic regulation of blood glucose in T2D mice that received HydrogeLED implants containing FRL-regulated, shGLP-1–expressing engineered cells. Specifically, the animals were treated with repeated daily cycles of (i) sampling one drop of blood and (ii) using this blood sample at the glucometer to automatically trigger SmartController 3.0–dependent shGLP-1 production (Fig. 6F). Self-sufficient production of shGLP-1 (Fig. 6G) resulted in a rapid approach to normoglycemia (Fig. 6H), which eventually reduced the SmartController 3.0–mediated shGLP-1 production in a feedback-controlled manner as reported to the smartphone (Fig. 6I).

DISCUSSION

Personalized medicine is committed to surveying, monitoring, and diagnosing health risks and providing individuals with information based on their unique medical background (37). In recent years, rapid advances in smartphone and communication network technologies have resulted in mHealth systems offering predictive, preventive, and participatory diagnostic tools (2, 37). For example, telemedicine capitalizes on the increasing number of “connected users” in developed countries sharing medical information via the internet to help patients actively participate in self-diagnosis and self-care (2). Today, although molecular diagnostics greatly benefit from advanced POCTs that rapidly provide accurate disease information using simple monitoring protocols, mHealth is limited by its therapeutic decisions, which still consist of classical drug prescription and intake. Therefore, technological advances enabling digitalized personal medication with synchronized interventions between diagnosis and therapy might set the stage for a global health care revolution.

In recent years, synthetic biology–inspired engineering approaches have given rise to a variety of rationally programmed engineered cells that behave in a user-defined and problem-oriented way (3843). Although cells engineered for precise theranostic correction of a variety of metabolic diseases have been validated in animal models and are expected to enter human clinical studies in the not-too-distant future (44), biosensing is generally analog and might, therefore, never achieve the digital precision of electronic sensors (45). Here, our SmartController 3.0 architecture combines the unique ability of electronic devices to read and generate digital signals with the maximal theranostic precision of biological cells, and future optimization into a (for example) more sophisticated implant design might boost the progression of cell-based precision medicines toward the clinic. The SmartController system engineered in this work couples the digitalized diagnostic accuracy of POCT devices with the inimitable efficiency of engineered cells to self-sufficiently produce and deliver drugs at optimal bioavailable standards, and it enables remote-controlled intervention into this closed-loop circuit at any user-defined time using conventional smartphones.

By combining telecommunication technology with optogenetics, we have provided the missing link enabling smartphone-controlled transgene expression in mammalian cells and mice. An ideal optogenetic device to interface with the smartphone would be orthogonal (provides minimal signaling interference with host endogenous pathways) and provide robust, fine-tunable, and reversible transgene expression profiles in response to light with deep-tissue penetration, low brightness, short illumination time, and negligible phototoxicity and without the need for xenogeneic chromophores. Continuous illumination of the mice at relatively mild exposure strengths [4 hours with FRL (25 mW/cm2)] was sufficient to trigger high induction ratios in vivo—a stimulation method that is better tolerated by animals (and cells) than exposure to repetitive short pulses over days, as in the case of many other optogenetic systems reported in the literature (8, 24, 25). The FRL-v2 optogenetic device developed in this study is biocompatible and meets the criteria for safe medical applications in humans (46).

However, a major limitation of the current SmartController version is the need to manually trigger the therapeutic response. To overcome this semiautomatic restriction, a future SmartController architecture could have the glucometer replaced by a continuous glucose monitor (47), which would also be implanted in the body to continuously monitor the glycemia dynamics for 24 hours and share the data via smartphones and, in the meanwhile, produce corresponding doses of insulin or shGLP-1 in a hyperglycemia-dependent manner. To further increase patient compliance and freedom of mobility during the therapy, the HydrogeLED could be powered by clinically approved batteries to avoid continuous exposure to electromagnetic radiation. To translate an eventual SmartController-driven concept of fully automated engineered cell theranostics into the clinic, the FRL-triggered genetic circuit would need to be validated in approved cell types or patient-derived autologous cells with long-term performance that genetically integrated with the optogenetic device. Subcutaneous implantation of engineered cell implants such as HydrogeLED are preferable to facilitate physical removal or exchange in case of therapy termination or alteration at regular intervals. Therefore, macroencapsulation systems might be preferred over microencapsulation technologies. Given that regulatory authorities such as the U.S. Food and Drug Administration and the European Medicines Agency have declared cells used to treat diseases to be “cellular medicaments” that must fulfil the same criteria as small molecules and biologics for clinical approval (48) and that macroencapsulated engineered cells have already been accepted for clinical studies (49), we believe that the SmartController concept could pave the way for a new era of personalized, digitalized, and globalized precision medicine.

MATERIALS AND METHODS

Study design

The aim of this study was to use a multidisciplinary design principle to create a platform that enables smartphones or Bluetooth-active point-of-care diagnostic devices to remotely control gene expression of engineered cells for diabetes therapy with a closed-loop telemedicine concept. We developed an Android-based smartphone app, ECNU-TeleMed, to regulate a custom-designed home server SmartController via the global GSM network. To translate this electronic controller circuit into biological responses, we further developed two FRL-responsive optogenetic systems as the interface to regulate human cell activities. To develop this wireless controller network toward therapeutic ends, we then tailor-designed a HydrogeLED implant with an engineered cell–carrying alginate hydrogel and wirelessly powered FRL LEDs. With diabetes mellitus as a model disease, in vivo expression of insulin or shGLP-1 from HydrogeLED implants in mice could be controlled by pre-set ECNU-TeleMed programs and also by a custom-engineered Bluetooth-active glucometer in a semiautomatic, glycemia-dependent, and self-sufficient manner. All in vitro experiments were done in triplicate, each containing six samples. For in vivo experiments, 12-week-old male db/db mice (BKS.Cg-Dock7m +/+ Leprdb/J, derived from C57BL/6J mice, Charles River Laboratory) were chosen as a disease model for T2D. A T1D mouse model was created for this study as well, which is described in the Supplementary Materials. In all experiments, mice were randomly assigned to the individual groups of five to six mice, and the experimenter was blinded to the analysis of all samples. Neither animals nor samples were excluded from the study. Comparisons among groups were performed using Student’s t test, and the results were presented as means ± SEM. All mice were sacrificed after the termination of the experiments.

SmartController design and fabrication

1.0 version. A custom-designed box containing a 32-bit embedded MPU chip (STM32F103, STMicroelectronics), a 443-MHz radio signal receiver and transmitter (Huawei Technologies Co. Ltd.), and a 24-V ac/dc power supply adapter (MeanWell Enterprises Co. Ltd.) (Fig. 1C) was driving either a Phillips physiotherapy lamp (E27, Phillips) or different 4 × 6 LED arrays that were custom-designed to fit the dimensions of a 24-well tissue culture plate (each LED centered above a single well; illumination, 0 to 5 mW/cm2; 400 to 730 nm) (figs. S24 and S25).

2.0 version. The same box used in the 1.0 version was set to drive an autonomous EEC containing an LDO linear regulator chip (TPS79733, Texas Instruments Inc.), a power amplifier circuit (0 to 24 V), and a homemade transmitting circular coil device capable of generating a 180-kHz electromagnetic sine wave signal (fig. S18) and capable of wirelessly powering custom-designed receiver LED integrating receiver coils, an SMD 0805 tantalum resonance capacitor (15 nF), and two far-red LEDs (SMD 3535, Epistar Corporation) (Fig. 4A and figs. S19 and S20).

3.0 version. To enable the reception of Bluetooth signals, the microcontroller chip of versions 1.0 and 2.0 was exchanged for a 16-bit MSP430 (MSP430F1611, Texas Instruments Inc.), and a QC12864B liquid crystal display was integrated for signal interpretation and analysis (Fig. 5A). The microcontroller chip was programmed with a defined algorithm to translate different glycemic thresholds into different power outputs (fig. S23). A commercial blood glucometer (5D-8B, Yicheng Co. Ltd.) was modified to harbor a Bluetooth transmitter (Nippon Electric Company) powered by a 3.7-V rechargeable lithium battery. Glycemia measured by the glucometer is transmitted to the SmartController via a universal asynchronous receiver transmitter module and optionally synchronized with the ECNU-TeleMed app (Fig. 5A and figs. S22 and S23).

FRL-controlled transgene expression in mammalian cells

Cells were seeded in a 24-well cell culture plate and cotransfected with corresponding plasmid mixtures. At 18 hours after transfection, the culture plate was placed below a custom-designed 4 × 6 LED array (fig. S25A) and illuminated for different time periods or light intensity as remotely set by the ECNU-TeleMed app (Fig. 2A).

SEAP expression assay

The production of human placental SEAP in cell culture medium was quantified as previously reported (8). Briefly, 120 μl of substrate solution [100 μl of 2× SEAP assay buffer containing 20 mM homoarginine, 1 mM MgCl2, 21% diethanolamine (pH 9.8), and 20 μl of substrate solution containing 120 mM p-nitrophenylphosphate] was added to 80 μl of heat-inactivated (65°C for 30 min) cell culture supernatant, and the light absorbance was recorded at 405 nm (37°C) from 10 to 30 min using a Synergy H1 hybrid multimode microplate reader (BioTek Instruments Inc.) using Gen5 software (version 2.04). The SEAP expression in mouse serum was quantified using a chemiluminescence-based assay kit (Roche Diagnostics GmbH, catalog no. 11779842001, lot. no. 10514400).

Mouse experiments

The backs of 12-week-old male wild-type C57BL/6J mice [East China Normal University (ECNU) Laboratory Animal Center], 12-week-old male db/db mice (BKS.Cg-Dock7m +/+ Leprdb/J, derived from C57BL/6J mice, Charles River Laboratory), or 12-week-old male T1D mice were shaved/unshaved and received different subcutaneous/intraperitoneal implants containing 2 × 106 transfected HEK-293 cells. One hour after implantation, the mice were exposed to different FRL sources controlled by the SmartController, and control mice were kept in the dark. Blood samples were collected via centrifugation (5000 rpm for 10 min) of clotted blood (37°C for 0.5 hour and then 4°C for 2 hours) at 24 or 48 hours after implantation for analytical assays, and physiological tests were performed at 48 hours after implantation.

Microcapsule implants. Transfected HEK-293 cells were encapsulated into coherent alginate-poly-(l-lysine)-alginate beads (400 μm; 200 cells per capsule) using a B-395 Pro encapsulator (BÜCHI Labortechnik AG) set to the following parameters: a 200-μm nozzle with a vibration frequency of 1300 Hz, a 25-ml syringe operated at a flow rate of 450 units, and 1.10-kV voltage for bead dispersion.

Hollow fiber implants. Transfected HEK-293 cells were seeded into 2.5-cm semipermeable KrosFlo hollow fiber membranes (Spectrum Laboratories Inc.), and both membrane ends were heat-sealed using a Webster smooth needle holder. Implants were subcutaneously implanted beneath the dorsal skin surface of the anesthetized mice (two or four 2.5-cm hollow fibers in each mouse).

HydrogeLED implants. Transfected HEK-293 cells were suspended in 1.5% (w/v) sodium alginate buffer (dissolved in Dulbecco’s modified Eagle’s medium) to a final concentration of 4 × 106 cells/ml. Five hundred microliters of this suspension was subsequently pipetted onto a receiver coil–containing LED (fig. S20) placed on the bottom of one well of a 24-well plate, solidified over 10 min by adding 500 μl of polymerization buffer [100 mM CaCl2 and 10 mM Morpholinopropanesulfonic acid (MOPS) (pH 7.2)], and incubated for another 10 min in 0.05% poly-l-lysine solution [0.05% poly-l-lysine (molecular weight, 15,000 to 30,000), 10 mM MOPS, and 0.85% NaCl (pH 7.2)]. After anesthetizing the mice by intraperitoneal injection of pentobarbital sodium salt (30 mg/kg), the HydrogeLED implants were implanted into the abdominal cavity of mice. The incised skin was sutured with skin staples, and the animals were transferred to the SmartController-driven wireless power generator at 1 hour after implantation (Fig. 5B).

Ethics. All experiments involving animals were performed according to the protocol approved by the ECNU Animal Care and Use Committee and in direct accordance with the Ministry of Science and Technology of the People’s Republic of China on Animal Care Guidelines. The protocol was approved by the ECNU Animal Care and Use Committee (protocol ID: m20150304). The blood samples from diabetic patients were provided by the Shanghai International Medical Center (SIMC) and approved by the SIMC Ethics Committee (approval number SIMC-20160107).

Statistical analysis

All in vitro data represent means ± SD of three independent experiments, each containing six samples. For the animal experiments, each treatment group consisted of randomly selected mice (n = 5 to 6). The blood sample analysis was blinded. Comparisons between groups were performed using Student’s t test, and the results are expressed as means ± SEM. Differences were considered statistically significant at P < 0.05. Prism 5 software (version 5.01, GraphPad Software Inc.) was used for statistical analysis. n and P values are reported in the figure legends. Individual-level data are reported in table S2, and P values are reported in table S3.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/387/eaal2298/DC1

Materials and Methods

Fig. S1. ECNU-TeleMed app remotely controlling the SmartController system.

Fig. S2. Blue light–triggered optogenetic devices.

Fig. S3. Optimization of the FRL-v1 system.

Fig. S4. FRL-triggered optogenetic devices.

Fig. S5. FRL-v1 control experiments.

Fig. S6. Fluorescence micrographs profiling EGFP expression kinetics of FRL-v1.

Fig. S7. The orthogonality of the FRL-v1 system.

Fig. S8. Optimization of the FRL-v2 system.

Fig. S9. The orthogonality of the FRL-v2 system.

Fig. S10. Fluorescence micrographs profiling EGFP expression kinetics of FRL-v2.

Fig. S11. Spatial control of FRL-dependent transgene expression.

Fig. S12. Impact of ectopic FRL-v2 constituents’ expression on the metabolic integrity of human cells.

Fig. S13. FRL-v2–dependent SEAP expression in different types of implants.

Fig. S14. FRL-v2 triggered insulin and EGFP expression in vitro.

Fig. S15. FRL-v2 triggered shGLP-1 expression in vitro.

Fig. S16. Impact of FRL on glucose and insulin metabolism of db/db mice.

Fig. S17. Unwanted activation of the FRL-v2 system by white light.

Fig. S18. Photographs of the custom-designed EEC.

Fig. S19. Implementation of the SmartController 2.0 system described in Fig. 5.

Fig. S20. Characteristics of HydrogeLED.

Fig. S21. FRL-v2–dependent SEAP expression in custom-designed HydrogeLED implants.

Fig. S22. Implementation of the SmartController 3.0 system described in Fig. 6.

Fig. S23. Different GTs programmed to translate into corresponding FRL illumination strengths.

Fig. S24. Schematic of the circuit design of a 4 × 6 FRL LED array.

Fig. S25. Photograph of the custom-designed FRL LED array and the physiotherapy lamp.

Table S1. Plasmids designed and used in this study.

Table S2. Individual-level data (provided in Excel).

Table S3. P values (provided in Excel).

References (5056)

REFERENCES AND NOTES

  1. Acknowledgments: We thank P. Wang (School of Life Sciences, ECNU) for providing plasmid pcGAS. We are grateful to all the laboratory members for their cooperation in this study. Funding: This work was financially supported by the grants from the National Natural Science Foundation of China (NSFC; no. 31522017) for outstanding young scientists, the National Key Research and Development Program of China, Stem Cell and Translational Research (no. 2016YFA0100300), the NSFC (nos. 31470834 and 31670869), the Science and Technology Commission of Shanghai Municipality (nos. 15QA1401500 and 14JC1401700), and the Thousand Youth Talents Plan of China to H.Y. This work was partially supported by the grant from ECNU for outstanding doctoral dissertation cultivation plan of action (no. YB2016024) to S.X. Author contributions: H.Y. conceived the project. J.S. and H.Y. designed the project, analyzed the results, and wrote the manuscript. J.S., S.X., G.Y., X.Y., S.Z., L.Y., J.Y., Y.W., and S.L. performed the experimental work. J.S. and S.X. performed the mouse experimental work and analyzed the results. J.S., Y.B., and H.Y. designed, constructed, and assembled the electronic components. Y.Y. optimized the optogenetic circuits and completed the drawing art. S.G. provided the human blood samples. M.X. and M.F. reviewed and revised the manuscript. All authors read and approved the manuscript. Competing interests: H.Y., J.S., S.X., G.Y., X.Y., Y.Y., S.Z., and Y.B. are inventors on patent application (Chinese patent application nos. 201610136156.1, 201610136489.4, 201610136478.6, 201610136465.9, 201610136462.5, and 201610136498.3) submitted by ECNU that covers an FRL-controlled transgene expression device for diabetes therapy, a remote-controlled gene circuit, an ultraremote-controlled gene circuit for diabetes therapy, a method for ultraremote control of transgene expression, an FRL-controlled transgene expression kit, and a protocol using FRL to control transgene expression. All other authors declare that they have no competing interests. Data and materials availability: All data pertaining to this study are present in the paper and/or in the Supplementary Materials. All genetic components related to this paper are available with a material transfer agreement and can be requested from H.Y. (hfye{at}bio.ecnu.edu.cn).

Stay Connected to Science Translational Medicine

Navigate This Article