To prepare Sanghuang Tongxie Formula (SHTXF), ten herbs (Morus alba L. [Moraceae] and Coptis chinensis Franch. [Ranunculaceae], Magnolia officinalis Rehder & E. H. Wilson [Magnoliaceae], Pueraria montana var. Thomsoni (Benth.) M.R.Almeida [Fabaceae], Astragalus mongholicus Bunge [Fabaceae], Cornus officinalis Siebold & Zucc. [Cornaceae], Raphanus raphanistrum subsp. Sativus (L.). Domina [Brassicaceae], Anemarrhena asphodeloides Bunge [Asparagaceae], Polygonatum odoratum (Mill.) Druce [Asparagaceae], and Atractylodes lancea (Thunb.) DC. [Asteraceae]) (Figure 1) were purchased from Beijing Tongrentang Tangshan Chain Store Drug Store Co., Ltd., and authenticated by Prof. Chunyu Tian at the College of Traditional Chinese Medicine, North China University of Science and Technology. The manufacturer, batch number, and weight of each crude herb used for each dose are listed in Table 1.

Combined drugs of one dose (75 g) were weighed and soaked in 1.5 L of double-distilled water (ddH₂O) for 12 h (Wang et al., 2020), followed by heating for 2 h. Next, the extract was separated and centrifuged at 4,200 × g for 30 min at 4°C, and the residue was collected to repeat the abovementioned steps. The two supernatants were then mixed and filtered by vacuum filtration with a 0.45-µm filter. Finally, the pumped liquid was placed in a beaker with medium fire to obtain a stock extract at a concentration of 0.5 g/ml, when the volume of the liquid reached 150 ml (Supplementary Figure S1). The prepared extract was directly mixed in the regular food to obtain a final concentration of 0.0125, 0.025, 0.05, or 0.1 g/ml SHTXF medium. Water alone was used as a control food.

Standard chemicals (berberine, magnolol, and morroniside) were purchased from the National Institutes for Food and Drug Control (NIFDC). Briefly, the quality of SHTXF was determined using a high-performance liquid chromatography (HPLC) system equipped with a UV-visible detector (Kamal et al., 2015). The analysis was performed using a Shimadzu LC-20A instrument. All chromatographic separations were carried out on an Agilent Eclipse XDB-C18 column (4.6 mm × 250 mm, 5 μm) at 35°C. The mobile phase for berberine was acetonitrile (A) and 0.1% phosphoric acid (B) (A: B = 50:50) by adding 0.1 g of sodium dodecyl sulfonate per 100 ml. A gradient system consisting of methanol (C) and water (D) (C:D = 78:22) was used to elute the magnolol. The mobile phase of morroniside was acetonitrile (E) and 0.3% phosphoric acid solution (F), and the elution gradients were 0–20 min (E: 7%, F: 93%) and 20–50 min (E: 7%–20%, F: 93%–80%).

Flies were maintained on a diet of cornmeal and sucrose medium (100°ml medium preparation: 5 g of sucrose, 1 g of agar, 6 g of cornflour, 3 g of yeast, 0.6 ml of propionic acid, and 100 ml of ultra-pure water) at 25°C in a 12 h light-dark cycle incubator with 50%–60% relative humidity. The Drosophila stocks used included: w ¹¹¹⁸ (#3605), Cg-Gal4 (#7011), UAS-InR^K1409A (#8253), tGPH (#8164), and dFoxO-GFP (#37585) obtained from Bloomington Drosophila stock center (BDSC) in Indiana University. The crossing scheme for all the groups is shown in Supplementary Figure S2. For all fly crossing assays, healthy, unmated male and female parents were randomly assigned to different groups.

Early third-instar larvae were starved for 2 h under adverse food conditions (0.8% agar in PBS) and then transferred to fresh dye-containing food (0.5% Brilliant Blue FCF) for 20 min. After feeding, the larvae were washed with phosphate-buffered saline (PBS), dried on tissue paper, and homogenized in an Eppendorf (EP) tube containing 100 µl of lysis buffer using a micro high-speed homogenizer (HR-68; Shanghai Hutong Industrial Co., Ltd.). After boiling and centrifuging for 3 min, followed by clearing by a high-speed refrigerated centrifuge (5430R; Eppendorf), 2 µl of the supernatant was analyzed with a spectrophotometer at 630 nm. A small amount of dye-containing food was weighed, processed as described above, and used as the standard to calculate the amount of ingested food (Matsuda et al., 2015).

The larvae were collected, rinsed with phosphate-buffered saline (PBS), and dried on tissue paper. The cuticle was carefully torn to release hemolymph on the parafilm membrane. Two microliters of hemolymph were collected using a micropipette, diluted 20 times with cold PBS, centrifuged for 3 min, and cleared by a high-speed refrigerated centrifuge. The supernatant was boiled at 75°C for 10 min and cleared by centrifugation. Next, glucose was measured after 15 min of incubation at 37°C using HK reagent (G3293; Sigma). A plate reader (M200PRD; Tecan) was used to measure absorbance at 340 nm.

Trehalose was converted using porcine trehalase (T8778; Sigma) overnight at 37°C, and the total amount of glucose was measured in the same way. The hemolymph trehalose concentration was determined by subtracting the value of free glucose from the untreated sample.

Six third-instar larvae were collected and rinsed several times with PBS to remove traces of food. The frozen samples were homogenized using a pellet pestle in 200 µl of cold 0.05% PBST (PBS + 0.05% Tween-20) on ice, immediately heat-inactivated at 75°C for 10 min, and then cooled to room temperature (RT). This was followed by homogenization in an EP tube containing 100 µl of lysis buffer using a high-speed homogenizer. Next, 10 μl of homogenate were centrifuged for 3 min, cleared by a high-speed refrigerated centrifuge, and used for the measurement of free glycerol. Ten microliters of homogenate were used to determine the protein content using a bicinchoninic acid assay (BCA) protein assay kit (P0010; Beyotime).

Ten microliters of the homogenate were mixed with 10 µl of a triglyceride reagent (T2449; Sigma) including lipase for hydrolysis of triglycerides to glycerol, incubated at 37°C for 60 min, centrifuged for 3 min, and cleared by a high-speed refrigerated centrifuge. Ten microliters of the supernatant were used for the measurement of triglycerides (TAG) using free glycerol reagent (F6428; Sigma) at 540 nm. A triolein-equivalent glycerol standard (G7793; Sigma) was used as the standard. The TAG concentration for each sample was determined by subtracting the values of free glycerol from the untreated samples.

Five microliters of the sample were incubated with 10 µl of buffer A [5 mM Tris-HCl (pH 6.6), 137 mM NaCl, and 2.7 mM KCl] containing amyloglucosidase (10115-1 g; Sigma) at 50°C for 60 min to digest glycogen. Ten microliters of the sample were incubated with 5 µl of buffer A without enzymes in parallel for the determination of glucose levels. Glycogen was used as a standard (10901393001; Sigma). The glycogen concentration for each sample was determined by subtracting the values of free glucose in the untreated samples. The amounts of TAG and glycogen per fly were normalized to soluble protein levels per fly (Li and Tennessen, 2017).

For Nile red staining, larvae were quickly inverted and dissected in 0.3% PBST. The tissues were incubated for 5 min with 0.001% Nile Red (72485; Sigma) in 0.3% PBST and washed twice with PBS at room temperature. The stained samples were mounted in 75% glycerol for microscopic analysis.

The larval discs were dissected and fixed in 4% formaldehyde for 20 min at room temperature (RT). After several washes with 0.3% (v/v) PBST, the discs were stained with primary antibodies at 4°C overnight and then with secondary antibodies at RT for 2 h. The following antibodies were used: mouse anti-Dlg1 (1:100) (4F3 anti-discs large; Developmental Studies Hybridoma Bank), rabbit anti-Akt (1:400) [4,691; cell signaling technology (CST)], rabbit anti-p-Akt (1:400) (4,060; CST), goat anti-mouse-cyanine3 (Cy3) (1:1000) (A10521; Life Technologies), and goat anti-rabbit Alexa 488 (1:1000) (4,412; CST). Vectashield medium (H-1500; Vector Laboratories) with DAPI (4’,6-diamidino-2-phenylindole) was used for mounting.

Whole larvae were lysed in radioimmunoprecipitation assay (RIPA) buffer (P0013; Beyotime) containing phosphatase inhibitor cocktail A (P1082; Beyotime) and phenylmethanesulfonyl fluoride (PMSF). Equal amounts of protein (10–30 µg, measured by BCA assay) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to a polyvinylidene fluoride (PVDF) membrane, and subjected to the standard western blot protocol, as previously described (Roth et al., 2018). The antibodies used in this study were rabbit anti-Akt (1:1000) (4,691; CST), rabbit anti-p-Akt (1:1000) (4,060; CST), rabbit anti-α-tubulin (1:1000) (2,125; CST), and goat anti-rabbit IgG (H + L, HRP) (1:10000) (5220-0336; Sera Care).

TRIzol (Invitrogen) and PureLink^TM RNA Mini Kit (12183018A; Life Technologies) were used to isolate total RNA from third-instar larvae of the indicated groups, followed by qRT-PCR as previously described (Wang et al., 2003). The relative amounts of transcripts were calculated using the comparative C _T method with ribosomal protein 49 (rp49) as the reference gene. The following primers were used:

For rp49 Sense: 5’- TCC​TAC​CAG​TTC​AAG​ATG​AC—3’

All data were verified using at least three independent experiments. The results were processed using GraphPad Prism 8.0 software as bar graphs. One-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison test was employed for statistical analysis. The p value was set at p < 0.05, and the center value was considered the mean. The error bars indicate standard deviation. “ns” denotes “not significant” values (p ≥ 0.05); ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

The Sanghuang Tongxie Formula (SHTXF) is composed of 10 herbal medicines, including Morus alba L. [Moraceae] (Sangbaipi) and Coptis chinensis Franch. [Ranunculaceae] (Huanglian), Magnolia officinalis Rehder and E. H. Wilson [Magnoliaceae] (Houpo), Pueraria montana var. Thomsoni (Benth.) M.R.Almeida [Fabaceae] (Gegen), Astragalus mongholicus Bunge [Fabaceae] (Huangqi), Cornus officinalis Siebold & Zucc. [Cornaceae] (Shanzhuyu), and Raphanus raphanistrum subsp. Sativus (L.) Domin [Brassicaceae] (Laifuzi), Anemarrhena asphodeloides Bunge [Asparagaceae]; Zhimu, Polygonatum odoratum (Mill.) Druce [Asparagaceae] (Yuzhu), and Atractylodes lancea (Thunb.) DC. [Asteraceae] (Cangzhu) (Figure 1; Table 1). To examine the quality of SHTXF, three representative compounds were selected as the quality standards. When the SHTXF high-performance liquid chromatography (HPLC) chromatogram was compared with the standard control (Supplementary Figure S3), we observed distinguished test peaks of berberine, magnolol, and morroniside in the chemical fingerprints of the SHTXF aqueous extract (Figure 2). Additionally, the proportions of the three ingredients determined by HPLC were 1.4%, 0.34%, and 0.022%, respectively (Table 2). This suggests that the prepared SHTXF contains the necessary active components and is suitable for further study.

To better dissect the effect of drugs on IR, we employed the GAL4/upstream activation sequence (Gal4/UAS) system and generated an IR model in Drosophila third-instar larvae (Brand and Perrimon, 1993), which was achieved by expressing a dominant-negative (DN) form of Drosophila InR (carrying the amino acid replacement K1409A, InR^K1409A, FBtp0018596, http://flybase.org/reports/FBtp0018596) under the control of the fat body (equivalent to mammalian adipose and liver tissues)-specific Collagen (Cg) promoter (FBti0027802, https://flybase.org/reports/FBti0027802) (Wu et al., 2005; Hennig et al., 2006). Misexpression of the InR antimorphic allele (UAS-InR^K1409A)-driven Cg-Gal4 (Cg > InR^K1409A) interfered with the physiological function of endogenous InR and undermined the downstream response of the insulin signaling cascade. As a result, both the hemolymph glucose level and the hemolymph trehalose (a glucose disaccharide that constitutes the major form of circulating sugar in insects) were significantly increased by 67.57% and 40.00%, respectively, in the larval stage (Figures 3A,B, Supplementary Figure S2). This suggests that fat body impairment in insulin signaling triggers insulin resistance (IR) and diabetic phenotypes.

To determine whether there was an inhibitory effect, we prepared feeding media consisting of SHTXF aqueous extract and raised flies with genotype Cg > InR^K1409A from the egg stage to the third-instar larval stage. The results showed that the dominant-negative InR-induced elevated circulating glucose was significantly suppressed by SHTXF extract at the concentrations of 0.025, 0.05, and 0.1, but not 0.0125 g/ml (Figure 3A). For circulating trehalose, SHTXF at a concentration of 0.025 g/ml displayed a stronger inhibitory effect than 0.0125 and 0.05 g/ml, while the concentrations of 0.1 g/ml did not show an obvious inhibitory effect (Figure 3B). Next, to exclude the possibility that changes in feeding rate may interfere with the metabolic phenotypes, we evaluated the level of food intake and observed a normal ingestion rate of drug-addled groups or compared with that of the control group (Figure 3C).

The protein content of larvae in each group was similar, with no significant statistical difference (Figure 3D). In addition, the hyperglycemic diabetic phenotypes of Cg > InR^K1409A larvae were notably inhibited by treatment with metformin (10 mM, a positive control) (Shirazi et al., 2014), which is one of the most popular oral glucose-lowering medications and is widely considered to be the optimal initial therapy for patients with T2DM (Figures 3A,B) (Lv and Guo, 2020). Collectively, SHTXF displayed a dose-dependent effect on Cg > InR^K1409A-triggered diabetic high circulating sugar in the Drosophila larval stage.

Consistent with the widely investigated roles of the insulin pathway in promoting anabolism, the downregulation of insulin signal activity by expressing InR^K1409A in the fat body reduced the total body storage sugar glycogen level by 17.48% during development compared with the Cg-Gal4/+ genetically matched controls (Supplementary Figure S4). However, we found that neither SHTXF (0.0125 and 0.025 g/ml) nor metformin significantly altered the overall larval glycogen content in the Cg > InR^K1409A background. Intriguingly, SHTXF extract at a concentration of 0.05 g/ml or 0.1 g/ml even decreased the level of stored glycogen (Supplementary Figure S4).

To examine the effect of SHTXF on lipid metabolism, we first checked the whole body TAG content and found that fat body showed low activity of the insulin pathway (Cg > InR^K1409A) decreased the storage lipid TAG level by 15.67% compared with the control (Figure 4A). Moreover, the reduced TAG level in the IR model was restored by treatment with metformin or 0.025 g/ml SHTXF (Figure 4A). Taken together, these data indicate that SHTXF may contribute toward carbohydrate clearance from the circulatory system by converting it into TAG, not glycogen, for nutrient storage. Ultimately, a concentration of 0.025 g/ml for the SHTXF water extract was chosen, which showed a strong inhibitory effect, to further explore the mechanisms.

To further determine the role of SHTXF in modulating lipid storage, we analyzed the distribution and lipid content in the fat body visualized by Nile Red staining. Consistent with a previous study (Fischer et al., 2017; Lourido et al., 2021), we observed that Cg > InR^K1409A-promoted IR enlarged the fat body cell lipid droplet size and reduced the lipid droplet number, total area of lipid droplets, and cell size when compared with the control (Figures 4B,C,F–H; Supplementary Figure S5). SHTXF partially reversed the increase in the size of lipid droplets and the decrease in lipid droplet number, total area, and cell size (Figures 4E–H; Supplementary Figure S5). Except for lipid droplet size, metformin displayed a similar effect on IR-mediated lipid storage disorder as SHTXF (Figures 4D,F–H; Supplementary Figure S5). These results indicate that SHTXF reverses insulin resistance-induced lipid homeostasis.

As phosphoinositide 3-kinase (PI3K) plays a pivotal role in the insulin pathway, we introduced an in situ fluorescent reporter termed tGPH (a fusion protein containing a tubulin promoter, GFP, and PH domain) (Britton et al., 2002), which recognizes PI (3,4,5)P₃ and visualizes PI3K activation by observing cell membrane-associated GFP, to check whether insulin signaling was truly impaired in IR diabetic files and the effect of SHTXF. Consistent with our previous work, compared with the control, the fat body cells of the IR model showed decreased cell membrane localization, indicating reduced PI3K activity and the downstream signaling cascade (Figures 5A–A’’’,B–B’’’). While Cg > InR^K1409A flies were raised on metformin- or SHTXF-added medium, the GFP signal of fat body cells was recruited to the membrane, indicating that PI3K activity was restored (Figures 5C–C’’’,D–D’’’). These results indicate that SHTXF may improve PI3K activity to ameliorate insulin resistance.

To further identify the effect of SHTXF on insulin resistance, we monitored Akt phosphorylation using an in vivo immunostaining assay and a western blotting (WB) assay in vitro. As expected, the misexpression of InR^K1409A in the larval fat body (Cg > InR^K1409A) led to reduced Akt Ser505 (equivalent to Ser473 in human Akt1) phosphorylation (p-Akt) in this organ compared with the control (Figures 6A–A’’’,B–B’’’,E,F), while total Akt levels were unchanged (Figure 6E; Supplementary Figure 6A–A’’’,B–B’’’). The decreased p-Akt level in the Cg > InR^K1409A IR model was largely rescued by SHTXF or moderately inhibited by metformin (Figures 6C–C’’’,D–D’’’,E,F). However, the expression of total Akt protein was unaffected by SHTXF or metformin (Supplementary Figures S6C–C’’’,D–D’’’). Taken together, these data confirmed that SHTXF enhanced Akt phosphorylation.

To address whether SHTXF modulates the nuclear–cytoplasmic shuttling of Drosophila FoxO (dFoxO), we examined the subcellular localization of the dFoxO-GFP fusion protein in fat body cells (Wu et al., 2015). We observed that the nuclear localization of dFoxO-GFP was significantly increased when InR^K1409A was expressed in fat body cells by Cg-Gal4 (Figures 7A–A’’’,B–B’’’). SHTXF, similar to metformin, hindered the nuclear localization of dFoxO-GFP (Figures 7C–C’’’,D–D’’’). In addition, to monitor dFoxO activity directly, we detected the transcription of the well-characterized dFoxO target gene 4E-BP using a quantitative reverse transcription-polymerase chain reaction (qRT-PCR) assay. Compared to the control, the upregulated 4E-BP mRNA level in Cg > InR^K1409A larvae was appreciably blocked by SHTXF and metformin (Figure 7E), indicating that SHTXF promotes dFoxO cytoplasmic localization and inhibits its transcriptional activity.