Optogenetic operated probiotics to regulate host metabolism by mimicking enteroendocrine

The enteroendocrine system plays an important role in metabolism. The gut microbiome regulates enteroendocrine in an extensive way, arousing attention in biomedicine. However, conventional strategies of enteroendocrine regulation via gut microbiome are usually non-specific or imprecise. Here, an optogenetic operated probiotics system was developed combining synthetic biology and flexible electronics to achieve in situ controllable secretion to mimic enteroendocrine. Firstly, optogenetic engineered Lactococcus lactis (L. lactis) were administrated in the intestinal tract. A wearable optogenetic device was designed to control optical signals remotely. Then, L. lactis could secrete enteroendocrine hormone according to optical signals. As an example, optogenetic L. lactis could secrete glucagon-like peptide-1(GLP-1) under the control of the wearable optogenetic device. To improve the half-life of GLP-1 in vivo, the Fc domain from immunoglobulin was fused. Treated with this strategy, blood glucose, weight and other features were relatively well controlled in rats and mice models. Furthermore, up-conversion microcapsules were introduced to increase the excitation wavelength of the optogenetic system for better penetrability. This strategy has biomedical potential in metabolic diseases therapy by mimicking enteroendocrine.


Construction of wearable optical device
A wireless-controlled, flexible optical device with a weight of only 4.798 g was fabricated to trigger 17 the L.LGLP-1 using blue light (Fig. 3a, Fig. S4 and Supplementary Videos). A Bluetooth module 18 was used to implement wireless control through a smart-phone application. Based on the 19 physiological structure of the rat intestine, four LED light sources have been arranged squarely to 20 cover the entire length of the small intestine. Each source consisted of four mini-LEDs to ensure the 21 necessary light intensity ( Fig. 3b and 3c). The optical and electrical properties of the device were 22 also characterized. Mini-LEDs with a size of 0.8 × 0.3 mm 2 showed a forward current of 9 mA under 23 a voltage of 3.3 V (Fig. 3d). The emission wavelength of the device has been measured to be 462 24 nm, which complies with the activation wavelength of promoter pDawn (Fig. 3e). In order to avoid the high temperature caused by continuous stimulation, graphene membranes were modified to the 1 backside of the flexible printed circuit board to facilitate heat dissipation. Also, the device was 2 allowed to work in a pulse-mode with a peak working current of ~60 mA to regulate the temperature 3 of the LED chip within an acceptable range less than 40 ºC (Fig. 3f and 3g). Tissue penetration of 4 LED photons was also simulated according to optical absorbance and scattering parameters of the 5 470 nm wavelength in different tissues using a Monte Carlo simulation software described in details 6 in the supporting information. The simulation result showed that the intensity in the gut, after 7 penetrating the rat external tissues, was around 1.5 mW/mm 2 ( Fig. 3h and 3i), which is greater than 8 the minimum excitation intensity required to activate promoter pDawn (1 nW/mm 2 ) 18 . 9 The feasibility of the OOPS with wearable optical device for in vivo glucose regulation was 10 demonstrated. Rats were given L.LGFP by oral gavage for 4 days, followed by shaving hairs on the 11 belly before equipping the LED device to reduce the light loss ( Fig. 3j and Fig. S5). The intestines 12 of the rats were harvested after 0 h, 0.5 h and 1 h of blue light stimulation. Intestinal contents and 13 villi were scraped to collect bacteria for analysis. Microscope fluorescence micrographs and flow 14 cytometric analysis showed that the LED device successfully induced GFP expression of L.LGFP in 15 the intestinal tract (Fig. 3k).

Metabolism regulation via wearable optical device
Based on optogenetic L. lactis and optical device, the ability of OOPS to regulate glucose 6 metabolism in vivo was evaluated in the T2D rat model (Fig. 4a). 7-week-old rats were firstly fed 7 with a high-fat diet (HFD) for 2 months to produce insulin resistance (Fig. S6a). Each rat was then 8 injected with Streptozotocin (STZ-1) and finally developed symptoms of T2D indicated by a high 9 glucose levels of 20mmol/L on average compared with 5mM for normal rats and 7mM for only 10 HFD rats (Fig. S6b). Rats in Blue group (treatment group) then received L.LGLP-1 gavage for 4 days. 11 Rats were then equipped with wearable optical device and received blue light (30 mW/mm 2 , 5 Hz) 12 for 1 h per day. Blood samples were collected right after stimulation, and blood insulin and GLP-1 13 concentrations were both almost twice than control ( Fig. 4c and 4d). Glucose was found to be 14 reduced from 24.4 mmol/L to 20.9 mmol/L during stimulation, while the glucose levels increased 15 slightly from 24.9mmol/L to 27.2mmol/L in average for the control (Fig. 4e). Random blood glucose 16 of treated rats showed gradually decrease from 27.4mmol/L in day one to 15.1mmol/L in day 6 ( Fig.  17 4f). After daily treatment for one week, the random blood glucose of blue group reduced by 7 18 mmol/mL as compared with the control group (Fig. 4g). However, the intraperitoneal glucose 19 tolerance test didn't show significant difference in glucose tolerance ( Fig. 4h and 4i), which may 20 due to insufficient GLP-1 secretion and treatment period. In addition to long-term hyperglycemia, 21 the high-fat diet and STZ-1 induced characteristic hyperlipidemia in T2D rats (Fig. 4j(1)). Blood lipids remained normal in the blue light group, and began to rise after treatment was discontinued 1 ( Fig. 4j and 4l), indicating that GLP-1 also plays an important role in lowering blood lipids 29 . 2 After all the above measurements were completed, rats were sacrificed for tissue section analyses. 3 Islet fluorescence staining showed higher activity and proportion of islet cell β (Fig. 4m-4o and Fig.  4 S7). Hematoxylin and eosin (H&E, Fig. S8) stained slices of pancreas tissues show common 5 vacuolization of the cytoplasm (black arrow), rare punctate necrosis and nuclear fragmentation 6 (yellow arrow) of islet cells in the control group. That there is only a decrease of islet cell cytoplasm 7 in the blue light group indicates the protective effect that GLP-1 exerts on islet cells. HE staining 8 sections of other tissues showed no significant adverse effects (Fig. S8). Thus, we conclude that the 9 animal experiment demonstrates that the OOPS lowers blood glucose by 10%-20% and improves 10 the blood lipid metabolism, suggesting the potential of this strategy for glucose metabolism 11 regulation by mimicking GLP-1 enteroendocrine. 12 6 P values were calculated by Student's t-test (***p < 0.001, **p < 0.01, *p < 0.05 criteria). Sample size: 7 n= 7 (c), n=5 (d-l).

9
Optimizing the stability of GLP-1 Due to the short half-life of GLP-1, mimicking GLP-1 enteroendocrine with optogenetic probiotics 10 might be deficient in long-term effects, and thus limiting the biomedical translation prospects of this 11 strategy. To optimize the stability of GLP-1, we constructed another L. latics secreting clone (L.Lsh), 12 that expressing more stable human GLP-1 (shGLP-1) under blue light control (Fig. 5a, Fig. S9a). 13 L.Lsh exhibited continuous increase in shGLP-1 secretion even after 60 mins of blue light exposure. 14 In contrast, GLP-1 secretion reached the maximum value after 20 mins of light exposure with 15 L.LGLP-1 (Fig. 5b), indicating that introduction of L.Lsh can effectively extend the working periods 16 of the bacteria. To achieve more precise GLP-1 delivery control, we also explored the relationship 17 between the amount of L.Lsh gavage and GLP-1 secretion. When the amount of L.Lsh gavage varied 18 from 1 mL to 4 mL, the concentrations of shGLP-1 measured after light stimulation of 60 min were 19 between 7.9 and 8.5 pM, suggesting that increased L.Lsh would not significantly increase the 20 secretion of shGLP-1 (Fig. 5c). To compare the effectiveness of using light stimulation of L.Lsh to 21 regulate insulin levels, referred to as L.Lcons, another modified L. latics that can spontaneously and 22 constantly secret shGLP-1 without optogenetic control has been adopted (Fig. S9b). 23 1 8 after gavage, the levels of inflammatory factors (IL-2, IL-6, TNF-α) were measured by ELISA to 2 confirm the immune safety of gavage (Fig. S10). The L.Lsh gavage group was treated for 30 days 3 by receiving 1 h of blue light stimulation (30 mW/mm2, 5 Hz) per day. Moreover, shGLP-1 could 4 act on the brain and intestines, inhibits intestinal peristalsis and gastric emptying (Fig. S9c-f), food 5 intake was inhibited in the treated rats, and thus cause weight loss, which is beneficial to the 6 treatment of T2D ( Fig. 5d and 5l). During blue light stimulation, the serum GLP-1 concentration 7 significantly increased, and insulin secretion was promoted to reduce blood glucose by 6.05±2.45 8 mmol/L (Fig. 5e-5g). Compared with continuously secretion of shGLP-1, optogenetic control would 9 reduce the burden of L. latics, thus higher concentration of shGLP-1 was detected in blue group. 10 Random blood glucose test also showed reduced blood glucose in blue and cons group (Fig. 5h). 11 On the 20 th day of treatment, we performed an intraperitoneal glucose tolerance test. Blue light 12 treatment markedly reduced fasting glucose level (Fig. 5i) and exhibit better glucose tolerance than 13 L.Lcons (Fig. 5j and 5k). Because GLP-1 has multiple physiological functions, we also examined 14 the cardiovascular effects of the treatment. The treatment successfully restored TG and TC 15 concentration to normal level in the first days, no significant change was observed in heart rate and 16 blood pressure (Fig. S11). At last, to verify the effect of the treatment on the intestinal flora, a 17 microbiota sequencing experiment was performed, we did not find any adverse effects after a 18 month-long exposure to L. lactis on other intestinal flora (Fig. S12). We interpret this as evidence 19 of the biosafety of OOPS. 20 Subsequently, to evaluate the adaptability of other animal models, the system has been used on diet-21 induced obesity (DIO) mice. Each mouse was given 500μL L.Lsh for 4 days followed by continuous glucose concentration compared to the untreated mice ( Fig. 5n and 5o), while the insulin level and 3 glucose tolerance maintained (Fig. S13), indicating other ways of blood glucose handling beside the 4 insulinotropic effect of shGLP-1. However, the treated mice exhibited significant weight loss (Fig.  5 5p) after three weeks of treatment. The mice were sacrificed to acquire adipose tissue for H&E 6 staining analysis (Fig. 5q and 5r). The result indicated reduced adipocyte diameter due to the 7 SIRT1-dependent lipolytic and oxidative capacity effect of shGLP-1 30 . 8

Probiotics operation via up-conversion microcapsules
Continuous exposure to strong blue light might cause cytotoxic effect 31 . Besides, poor tissue 9 penetration of blue light further limits its clinical application. To side-step these issues, up-10 conversion materials were used for blood glucose metabolism management under near-infrared 11 control (Fig. 6a and 6b) 32,33 . Using up-conversion microrods as an energy transfer relay, near-12 infrared light (980 nm) with deeper penetration than blue light (Fig. S14) is converted to blue light 13 (475 nm) in vivo ( Fig. 6c and 6d, Fig. S15). Then the blue photons stimulated gene expression of 14 plasmid pDawn in L.LGLP-1. However, because delivering up-conversion microrods to the intestinal 15 tract is cytotoxic and leads to excessive dispersion of the microrods that would reduce the intensity 16 of the glow, the microrods were encapsulated in a biocompatible microcapsule 34 (Fig. S16) using 17 microfluidic technology (Fig. 6e and 6f). The microdroplets showed high stability, low permeability 18 at different pH over 20 hours (Fig. 6g and 6i). Then ICG (Indocyanine Green) encapsulated 19 microcapsules were delivered to the intestinal tracts of rats. These were used to monitor in vivo 20 distribution of the microdroplets by imaging the intestinal tract and the feces every 4 hours using an 21 in-vivo imaging system (Fig. 6h). The microcapsules were localized in the small intestine in the 22 first 4 hours and started to be expelled from the body at around 8 hours after gavage. 23 To validate the up-conversion control in vivo, L.
LGFP and UCMCs were delivered to the rats by 1 gavage. GFP was successfully expressed after 0.5 hour and 1 hour of exposure to NIR laser 2 stimulation (Fig. S17). Subsequently, the L.LGLP-1 and the UCMCs were delivered to the diseased 3 rats by gavage. One hour after stimulation under a near-infrared laser, more than 88.9% higher in 4 blood GLP-1 and 44.3% higher in insulin concentrations were detected in the treated group through 5 ELISA right after the stimulation (Fig. 6j and 6k). The blood glucose of rats in NIR group reduced 6 by 3.3mmol/L in comparison with almost unchanged blood glucose in the control group (Fig. 6l). 7 H&E-stained slices of liver, kidney and intestines of the treated rats revealed no conspicuous 8 damage (Fig. 6m), indicating that the up-conversion optogenetic nano system affords a biosafe 9 regulation of glucose metabolism in vivo.   16 17

Discussion
A system for metabolism regulation via optogenetic probiotics has been put forward, which 18 introduced synthetic biology to design optogenetic L. lactis and adopted flexible electronics 19 techniques and rare-earth nanotechnology to control L. lactis. Considering the flexibility of 20 synthetic biology design and the convenience of optogenetic control, this strategy theoretically has 21 universality in biomedical applications. Due to the intestinal function, the effects of engineered 22 probiotics administration can mimic the enteroendocrine pathway. Therefore, the strategy provides 23 a new intervention method for the metabolic regulation which the enteroendocrine system involved 24 in, and has prospects for the treatment of metabolic diseases.
Despite a complete solution has been given here, it is worth to be mentioned that several challenges 1 remain. Firstly, every segment of enteroendocrine and corresponding metabolism is complicated, 2 and may involve the interaction of multiple organs, cells, and signal pathways 2, 35 . As a metabolism 3 regulation strategy, the development of this strategy depends on the achievements of 4 enteroendocrine mechanism research. At the same time, we expect that the OOPS can help in the 5 mechanism research of enteroendocrine in the future. Otherwise, by secreting shGLP-1, the blood 6 glucose of T2D rats decreased by 6mM, but still far from the normal level. This may be caused by 7 resistance of the L. lactis to prolonged blue light stimulation, leading to limited concentration of 8 blood shGLP-1. We hypothesize that this limitation can be optimized to some extent by introducing 9 genetic circuits 36, 37 . Besides, L. lactis is a prokaryote, it can produce only a limited number of 10 hormones, and the products tend to accumulate in the cytoplasm limited by the secretion capacity. 11 The follow-up work needs to focus on synthetic biology to enrich product types 38 to better mimic 12

enteroendocrine. 13
In brief, we have proposed a new strategy that accurately and effectively regulates metabolism by 14 mimicking enteroendocrine. This strategy is based on optogenetic probiotics and involves two 15 technologies to assist in light control: wearable devices and up-conversion microcapsules. This 16 strategy is precise and controllable for metabolic intervention, and has potential universality. It 17 might bring new opportunities for the treatment of metabolic diseases, and is expected to inspire the 18 research on the relationship between gut microbes and the host. For this goal, in the future, we will 19 conduct more in-depth studies on the interaction between the engineered probiotics and the host. 20

Materials and Methods
Construction of strains. Plasmid pDawn was constructed according to the method of Ohlendorf et.
al. Gene sequence of signal peptide usp45, rat GLP-1 and Fc of IgG in rat was found in NCBI. The 1 recombinant plasmid was constructed by PCR using AceTaq® DNA Polymerase (),and finally 2 transformed to L. lactis NZ9000 competent cells by electro-electroporation. 3 Flow cytometry analysis. In-vitro induced bacteria were centrifuged at 8000 rpm for 5 mins and 4 washed twice with PBS. Rats were sacrificed, and the intestine was removed and segmented. The 5 intestinal mucosa was collected and resuspended in PBS at a volume ratio of 1:1. The precipitate 6 was removed by centrifugation at 2000 rpm for 5 mins, and the bacteria were collected by 7 centrifugation at 8000 rpm for 5 mins. Both sediments were fixed with 1% paraformaldehyde for 8 20 mins followed by PBS washing. The sediment was then resuspended in 1 mL PBS before flow 9 cytometry analysis. Flow cytometric analysis was performed using a FACSCalibur instrument 10 (Becton Dickinson, Sunnyvale, CA). Data were analyzed on Flowjo VX. 11 Cell experiment. Cellular cultures of ins-1 were incubated in RPMI-1640 medium in 12-well cell 12 culture plates to a proper density. L.LGLP-1 in the logarithmic phase was added in 100 μl aliquots to 13 ins-1 per well. Then, 100 μl bacterial medium and 1 ng/mL GLP-1 solution were added to other 14 wells as negative control groups. Co-culture lasted for 1 h under 25 mW/cm 2 blue light stimulation. 15 The supernatant was then collected and tested using an insulin ELISA kit. For cell proliferation 16 assay, ins-1 was incubated in a 96-well cell culture plate at a density of 2000 cells per well. L.LGLP-17 1 in the logarithmic phase was induced under 25 mW/cm 2 blue light for 1 h. The supernatant was 18 collected and filtered with a 0.2 μm filter membrane, then added as cell-free supernatant to ins-1 at 19 10 μl per well. Next, 10 μl bacterial medium and 1 ng/ml GLP-1 solution were added to other wells 20 as negative control groups. The culture plates were incubated in a cell incubator for 12 h. A CCK-8 21 cell viability kit was used to monitor cell growth by measuring the absorbance at OD450 using a microplate reader. Oxidative stress protection was detected by ROS-ID® Hypoxia/Oxidative stress 1 detection kit(ENZ-51042-0125, Enzo Life Sciences Inc., Switzerland). The supernatant of L.LGLP-1 2 after 1 hour blue light induction and incubated for 24 h in 12-well plates, followed by ROS induction, 3 confocal and flow cytometry detection as recommended in the manual. 4 Validation of intestinal L.LGLP-1. L.LGLP-1 was co-cultured for 30 mins with 50 µM CFSE at a 5 volume ratio of 10:1, then centrifuged for 3 mins at a speed of 8000 rpm. Unbound dye was removed 6 by PBS wash for 2 cycles. The CFSE dyed L.
LGLP-1 was delivered to rats by continuous oral gavage 7 for 4 days. On the fifth day, rats were sacrificed and harvested the intestine. Samples were soaked 8 in paraformaldehyde for 1 day and divided into segments identified as duodenum, jejunum, ileum 9 (into two parts), and colon. Each segment was divided into two sections: one section was 10 immobilized in the embedding fluid for use in frozen sections, the other was dissected and the 11 mucosa scraped to harvest cells for flow cytometric analysis. 12 Construction of the wearable optical device. Low footprint, packaged off-the-shelf components 13 were chosen as the basis of the circuit. High-brightness mini-LEDs (S-32CBMUD, Sanan 14 Optoelectronics) were mounted to a flexible printed circuit board using low-temperature Ag solder 15 paste. A low-power microcontroller (CC2541, Texas Instruments) with Bluetooth capability 16 controlled the working sequence of LEDs. Low-on-resistance (22 mΩ) metal oxide silicon field-17 effect transistors (CSD17585F5, Texas Instruments) and a low-dropout (120 mV), low-quiescent-18 current (12 µA) linear regulator (LP5907MFX-3.3, Texas Instruments) with a fixed output voltage 19 of 3.3 V were used to drive LEDs. Passive components with 0201 and 0402 packages further 20 minimized the layout. 21 using a source meter (Keithley 2400, Tektronix). The electroluminescent spectra were obtained 1 using a spectrometer (USB2000+, Ocean Optics). Thermal measurements were performed using a 2 thermal imaging camera (226s, FOTRIC) and a close-up lens, with a background temperature of 3

℃. 4
Animal experiments. Seven-week-old rats were given a high-fat diet for 2 months to produce 5 insulin resistance. Oral glucose tolerance (OGTT) was then tested. The model was considered 6 successful if the blood glucose did not drop below 6 mmol/ml within 2 h. Achieving this setpoint, 7 each rat was injected with 1% stz-1 solution in the tail vein at a dose of 28.5 mg/kg. This procedure 8 produced the rat model of T2D. All rats were fed normal food during the experiment. For DIO mice 9 experiment, after purchasing 16-week-old DIO mice and their control group (fed with normal diet), 10 they were fed with high-fat diet or normal diet for two weeks to adapt to the environment, and then 11 carried out experiments.  where ω is the initial photon weight, μa is the absorption coefficient of the medium.

15
The scattering direction can be estimated by the Henyey-Greenstein phase function, where ωi is the weight absorbed into the element i, N is the number of photon packets and Vi is the area 23 of the element.

24
Briefly, a length of 1 mm on the left boundary was set as the light source, with a cosine launching property.

25
The photon count was set to be 10 6 , and a square area of 1 × 5 mm 2 was set as the medium to mimic the