3.1 TFA Decreased Blood Glucose and Energy Intake in T2DM Animals
To confirm quality of total flavonoids from Astragalus membranaceus (TFA), liquid chromatography was carried out; and we found main flavonoids were included in TFA (Figure 1A and 1B).
To induce T2DM, 8-week-old C57BL/6J mice were fed with high fat diet for 8 weeks, and then injected with streptozotocin (STZ, 50 mg/kg.d) for 5 days. Successful construction of T2DM was believed when the fasting blood glucose above 11.1 mmol/L or 120 min IPGTT ≥11.1 mmol/L. Diabetic mice were daily administrated with TFA or metformin as indicated for 16 weeks, and normal control mice (NC) were administrated with normal saline (Figure 1C). As shown in Figure 1D and E, TFA administration significantly decreased fasting blood glucose (Figure 1D); at Week 16, decrement of fasting blood glucose in TFA-L group is more significant than that of T2DM, TFA-M, TFA-H or positive control (metformin) groups (Figure 1E). More importantly, food intake was dramatically inhibited compared with T2DM group (Figure 1F, p<0.01). We did not find significant effect of TFA on reducing body weight in the present study (Figure 1G).
3.2 TFA Ameliorated Brain Impairment and Reduced Aβ Aggregation
Cognitive impairment is one of a common complication in T2DM. To firstly observe histological changes of the brain, H&E staining and immunohistochemistry were carried out. We found area of hippocampus as well as the whole volume of the brain in T2DM were decreased, the neurons in cortex and hippocampus in T2DM group were shown with loose structure and exhibited with cell body shrinking and vacuoles, and TFA administration ameliorated damage of the brain (Figure 2A-C). Aβ aggregation within brain tissue has been recognized as cause and representation for cognitive disorders. By immunohistochemistry and ELISA (Figure 2D and E), we found Aβ aggregation in the brain was significantly lowered compared with T2DM (p < 0.05), suggesting potential effects of TFA on ameliorating diabetic cognitive impairment.
3.3 TFA Improved Synapse Function and Promoted BDNF Expression
To further evaluate effects of TFA on diabetic cognitive impairment, expression of PSD95 and synapsin in hippocampus were detected by Western Blot to evaluate function of synapse. As observed in Figure 3A, expression of PSD95 and synapsin in T2DM mice were dramatically decreased by as much as 40% (p<0.05, vs NC), while TFA administration significantly up-regulated their expression (p<0.01, vs T2DM).
BDNF is a neuro-peptide which is secreted by the autocrine manner and exerts effects on neuron grow and differentiation. In the present study, secretion of BDNF in serum and brain tissues was determined by ELISA. We observed in T2DM mice that BDNF was decreased in both serum and brain tissues, and TFA or MET administration significantly increased expression of BDNF in the brain (Figure 3B and C). Further study demonstrated that the up-regulation of BDNF was via its upstream signaling pathway, CREB, as expression and activation of p-CREB was strengthened by TFA or MET in the hippocampus (Figure 3A).
Advanced glycation end products (AGEs) has been believed as an inducer and promoter in diabetic complication, and its aggregation in the brain has been found to play a role in cognitive dysfunction [27, 28]. In the present study, we also observed TFA or MET significantly reduced level of AGEs in the brain (Figure 3D). Latest research indicates that γ-aminobutyric acid (GABA) is associated with the development of neurodegenerative disease [29–31]. In our present study, we found that GABA was significantly increased in high fat and STZ induced diabetic mice brain, while was decreased to normal level after TFA treatment (Figure 3E).
3.4 TFA Promoted Mitochondrial Biogenesis in the Hippocampus
To explore potential targets and mechanism of TFA in its modulated brain function, proteomics analysis of mice hippocampus tissue was performed. In the present study, a total of 3327 proteins were detected, 85 proteins were found to be differentially expressed in T2DM compared with NC group, and 43 were found to be differentially expressed in T2DM+TFA compared with T2DM group (Figure 4A, p < 0.05). Among the 43 proteins, 31 was up-regulated while 12 were down-regulated. Through enriched GO terms analysis, we found the tight junction proteins were decreased in T2DM, while were increased by TFA administration (Figure 4A); moreover, the energy metabolism was dysregulated in T2DM, and this was ameliorated after TFA treatment. KEGG pathway enrichment analysis showed that the differentially expressed proteins in T2DM+TFA compared with T2DM group were mainly related with metabolic, mitochondria dysfunction and neurodegeneration disease in the hippocampus (Figure 4B). To further interpret the interaction among the 43 differentially expressed proteins, the protein-protein interaction (PPI) enrichment network analysis was performed by STRING. We found that the hub proteins were mainly enriched in oxidative phosphorylation, synaptic function and metabolic process (Figure 4C).
In order to clearly show changes in electron transport chain (ETC), differential proteins were indicated in ETC pathway. We can clearly see that T2DM reduced ATP production by inhibiting the activity of NADH dehydrogenase and cytochrome c oxidase (Figure 4D). While TFA treatment reserved this situation by increasing the activity of mitochondrial respiratory chain complex I (CxI) (Figure 4E). The representative proteins in each biological process were show in Figure 4F and G, which maybe the potential targets for TFA to against cognitive impairment in diabetes. Converging from results of proteomics and PPI analysis, we concluded that mitochondria might play a pivotal role in diabetic cognitive dysfunction. For this aim, level of mitochondrial DNA as well as the underlying mechanism was studied.
As the indicator of mitochondrial biogenesis, the mtDNA/nDNA ratio was firstly determined in hippocampus by Q-PCR. We found the copy number of mitochondrial DNA was significantly decreased in T2DM (p < 0.05, vs NC), and TFA administration significantly increased the copy number by over 2 times as compared with that in NC group (p <0.01, vs T2DM and NC) (Figure 4H).
To explore the underlying mechanism of TFA in mitochondrial biogenesis, expression and activation of intracellular signaling pathway proteins were determined by WB. As shown in Figure 4I, TFA significantly reversed reduction of PGC-1α and p-AMPK/AMPK in diabetic mice (p < 0.01, vs T2DM); moreover, expression of mitochondria unfolded proteins including LONP1, CLPP, HSP60 and HSP70 were significantly increased by TFA to the level that was comparable to that of metformin (MET), suggesting TFA protected neuron function by promoting mitochondrial biogenesis and maintaining energy metabolic balance. Further study found effects of TFA was realized through estrogen receptor beta pathway, as its expression was elevated on TFA administration (Figure 4I).
3.5 TFA Protected Blood Brain Barrier (BBB) and Gut Barrier from Diabetic Damage
Blood-brain barrier (BBB) plays a pivotal role in maintaining homeostasis of the brain. By proteomic analysis, we found tight-junction proteins in the brain were reduced in T2DM, while were elevated after TFA treatment (Figure 4A). To elucidate effects of TFA on BBB, expression of ZO-1 and claudin 5 in hippocampus was determined by WB. As shown in Figure 5A, TFA administration significantly increased expression of ZO-1 and claudin 5 in the hippocampus (p < 0.05, vs. T2DM). This was validated in the in vitro study in Bend.3 cells (Figure 5B-E, and H) that TFA increased both cell viability and intracellular tight junction proteins’ proteins. We also observed that TFA significantly increased the protein expression and activation of PGC-1α and p-AMPK/AMPK in Bend.3 cells (Figure 5F, G and I).
Another barrier that preserves homeostasis of internal environment is gut barrier. It has been widely recognized that disruption of the integrity of gut barrier is closely associated with diabetic complications [32–34], as this will make gut-sourced endotoxin easier to enter into the body and induce a so called sustained low-grade inflammation. To this aim, we firstly observed morphological changes before and after TFA administration. As shown in Figure 6A, the intestinal muscularis structure in NC group was tight, while this structure was strikingly changed in T2DM in that the villi density was sparse, muscularis of intestinal was relaxed, and the number of goblet cells was reduced; and TFA or MET administration improved the structure integrity. By immunohistochemistry and WB, we also assessed location as well as expression of tight junction proteins in the gut barrier. As observed in Figure 6B and C, TFA or MET administration significantly elevated expression of tight junction protein occludin (p < 0.05, vs T2DM). The intestine barrier protective effect of TFA was further confirmed in the in vitro study of CaCO2 cells. With the treatment of TFA, the cell viability was elevated, protein expression of ZO-1, occludin, and claudin 5 were significantly increased and the gut barrier leakage was significantly dropped (Figure 6D-M).
Tight junction of gut barrier relies on energy supply, thus mitochondrial function is important. To this end, expression of CLPP and HSP60, the core markers of mitochondrial unfold protein response (UPRmt), on gut lumen was determined by WB (Figure 6C). We found CLPP and HSP60 were significantly up-regulated in T2DM mice (p < 0.01, vs NC), suggesting that UPRmt was continuously activated in the gut of diabetic mice; and their expression was significantly down-regulated to the normal level after TFA or MET treatment.
To further evaluate functional integrity of gut barrier, levels of inflammatory cytokines in the serum were determined in diabetic mice. As shown in Figure 6N-Q, contents of IL-6, TNF-α, IL-1β, and LPS were strikingly elevated in diabetic mice compared with control (p < 0.05, vs. NC), and this trend was significantly reversed by TFA administration. Interestingly, we also found reduction of 5-HT, which may inhibit nerve activation and induce emotion as well as cognition changes, on application of TFA, but no statistical significance was observed (Figure 6R, TFA vs. T2DM). GLP-1, which is produced by L cells in intestinal mucosa, has been verified to possess effects on inhibiting the uptake of glucose [35], ameliorating T2DM, and improving memory performance [36, 37]. In the present study, we observed that GLP-1 was significantly elevated in intestine of T2DM mice, while was decreased after TFA administration (Figure 6S, TFA vs. T2DM). Moreover, TFA or MET administration significantly ameliorated reduction of GABA in gut of the diabetic mice (Figure 6S, TFA vs. T2DM).
It should be noted that LPS cannot be generated by the body itself, thus its decrement should attribute to (1) gut integrity enhancement, or (2) its production reduction within gut lumen. In this sense, influence of TFA on abundance as well as composition of gut microbiota was assessed in the following study.
3.6 TFA Re-constructed Composition of Gut Microbiota
Intestinal microbes have been recognized to play an important role in ameliorating diabetes as well as cognitive impairment. Studies indicated that oral drug administration may influence disease development via modulating composition and metabolites of gut microbiota. To this end, gut microbiota as well as its metabolites were investigated in animals after 16 weeks’ oral administration with TFA. As shown in Figure 7A, number of OTUs in diabetic mice was significantly decreased (p < 0.01, vs. NC), and TFA or MET treatment significantly increased abundance of OUTs to the normal levels. Analysis of observed species, ACE index and chao1 index suggested that administrating of TFA or MET was benefit for higher species richness in gut microbiota compared with that in T2DM group (Figure 7B-D). Mice in TFA or MET group shared more common species with that in NC group (Figure 7E-H). As depicted in Figure 7I and J, although TFA treatment did not totally overturn changes of gut microbiota in diabetic mice, abundance of probiotics, such as butyrate-producing p_Acidobacteria, p_Proteobacteria, p_Gemmatimonadetes, p_Deferribacteres, p_unidentified_Bacteria and p_Latescibacteria was dramatically increased, while microbiota including p_Bacteroidetes, p_Tenericutes, p_Melainabacteria and p_Chloroflexi was decreased (Figure 7J). Further study by genus species analysis (Figure K and L) showed that TFA administration significantly increased abundance of g_Aerococcus, g_Bifidobacterium, and g_Faecalibacterium et al. Converging evidence suggested that TFA application increased abundance as well as diversity of gut microbiota; more importantly, abundance of butyrate-producing bacteria was significantly elevated, suggesting metabolites of microbiota (eg. butyrate) may play an important role in it.
To demonstrate the above hypothesis, STZ-induced diabetic mice were orally administrated with sodium butyrate (NaB), and neuron changes was studied. As shown in Figure 8A-C, NaB oral administration slightly reduced body weight in diabetic mice, but no statistical significance was found; however, NaB significantly reduced fasting blood glucose level as compared with T2DM group. Brain pathology study by H&E staining indicated that degeneration such as neuron shrinkage (black arrow) and swelling (green arrow) were more common in diabetic brain (Figure 8D), while NaB reversed this damage.
To preliminary explore mechanism of NaB in diabetic brain injury, we determined level of BDNF in brain by ELISA and found its expression was significantly elevated by TFA treatment (Figure 8E); moreover, levels of AGEs and IL-1β were strikingly reduced (p < 0.01, vs. T2DM) (Figure 8F and G). Interestingly, we found GABA was decreased in the intestine tissue, while increased in brain tissue in T2DM mice; and after NaB treatment, the content of GABA was returned to normal level (Figure 8H-J). We also observed that the increased expression of GLP-1 within gut lumen was reduced by 16 weeks’ oral administration with NaB (Figure 8J).
3.7 TFA Ameliorated Mitochondrial Dysfunction in AGEs-induced HT22 Cells
AGEs has been widely recognized to be an inducer and promoter in diabetic complications [38]. To further investigate mechanism of TFA on protecting brain function, HT22 cell line was applied. We firstly incubated HT22 cells with different concentrations of TFA for 24 or 48 hours (Figure 9A), and found TFA at 5 µg/ml has most significant effects on increasing cell viability. While AGEs inhibited cell viability in a concentration-dependent manner (Figure 9A). Converging with published reports, we choose 200 µg/ml of AGEs to induce cell damage. We found when the cells were co-incubated with 200 µg/ml AGEs and 5 µg/ml TFA, the viability was highest (Figure 9C). In order to determine the optimal concentration of AGEs and TFA, the apoptosis test was performed by acridine orange (AO) / ethidium bromide (EB) staining. We observed that the cell apoptosis was significantly increased under 200 µg/ml AGEs incubation, while reserved to normal levels by 5 µg /ml or 10 µg /ml of TFA after 48 h treatment (p < 0.01, vs. AGEs) (Figure 9B and D).
The influence of TFA on the secret of BDNF was verified in HT22 cells by ELISA and Q-PCR. We observed that the level of BDNF was significantly down-regulated under AGEs administration, while increased by TFA; further studies suggested this was due to the increased expression and activation of CREB (Figure 9E-G). RAGE, as a receptor of AGEs, has been considered as a contributor to causes of cognitive decline. In the present study, the levels of RAGE were tested by Q-PCR and immunofluorescence. We found RAGE was significantly decreased by TFA (p < 0.01, vs. AGEs) (Figure 9H, I and K). Moreover, TFA increased the protein expression of PSD95 (p<0.05, vs. AGEs) (Figure 9J and K). Proteomic analysis in this study found that the pathogenesis of diabetic cognitive impairment is closely related to mitochondrial dysfunction (Figure 4A and D). Based on this finding, ROS as the indicator of mitochondrial dysfunction and repair was tested by Dihydroethidium staining. We found 5 µg /ml TFA decreased the content of AGEs-induced ROS in HT22 cells (p < 0.05, vs AGEs) (Figure 9L). The mitochondrial number (Figure 9M and N), the mitochondria membrane potential (Figure 9O and Q), as well as the activity of antioxidant enzyme (Table 2) were dramatically elevated on TFA application. Mechanism investigation suggested that TFA modulated mitochondrial function via modulating its biosynthesis and energy metabolism (Figure 9P, and R-W), and estrogen receptor beta may participate in this modulation (Figure 9X).