Type 2 diabetes mellitus (T2DM) is a chronic and common metabolic disease with a worldwide increase in prevalence, and is associated with severe syndrome and characterized with deficit in insulin production or sensitivity (Costa et al., 2004). Large-population investigations have revealed that diabetes mellitus is associated with Alzheimer’s disease (AD) (Gasparini et al., 2002), mild cognitive impairment (MCI) (Li W. et al., 2016), anxiety and depression (Anderson et al., 2001) in Central Nervous system. AD is even considered as type 3 diabetes mellitus due to insulin resistance in the brain similar to DM in peripheral lesions. Besides, T2DM and AD often co-exist in several individuals. Recently, several clinical observations have provided significant insights into certain injuries of brain structural and functional (Martinez-Tellez et al., 2005). These impairment in global memory, abstracted reasoning, attention and visual-motor arrangements frequently happened in the diabetic population (Ryan and Geckle, 2000a,b). Moreover, insulin action in the brain was selectively impaired in T2DM, especially in the hippocampus, hypothalamus and cortex (Bruehl et al., 2009). The lesions of insulin action in diverse regions could be attributed to the generation of some factors such as inflammatory mediators, oxidative stress, impaired insulin receptor signaling, Aβ (β-amyloid) plaque and hyper-phosphorylated tau protein.

The medial prefrontal cortex (mPFC), a critical structure in the brain, coordinates the activities of multiple cortical regions, exerts the effects of the ongoing supervisation of information, being critical to daily work (Krueger et al., 2007; Einarsson and Nader, 2012). Accordingly to a current study, memory-related brain circuits in mPFC were impaired in T2DM (Tagougui et al., 2015) and prefrontal atrophy who had been diagnosed (Lee et al., 2013; Moran et al., 2013). A longitudinal neuropsychological study has convincingly shown that patients with damage to PFC exhibit disorders in the higher cognitive functions (Cohen, 2000). Clinical research indicated that the terrible executive function and memory lesion in T2DM correlated predominantly with reduced glucose metabolism in the mPFC and orbital, temporal, and cerebellum regions (Garcia-Casares et al., 2014). Some studies suggested the memory dysfunction occurred in T2DM patients during acute hyperglycemia (Cox et al., 2005). The abnormal change in the insulin signaling pathway might contribute to damage the neuronal structure or function (Zhao and Alkon, 2001) in T2DM, and further disturbed the impulse conduction of the nerve, the transmitter release, information export and so on. Therefore, the mechanisms which triggered neuronal lesions in the mPFC of DM by metabolic syndrome were still a subject that scientists needed to investigate. What’s more important in DM encephalopathy to our studies, BBr whether alleviated the cognitive impairment through modulating insulin signaling transduction and anti-inflammatory pathway in mPFC still require more researches to clarify.

Berberine which derives from Traditional Chinese Medicine has shown its safety and efficiency in human and animals. It is an isoquinoline alkaloid and isolated from Coptis chinensis Franch (Teodoro et al., 2013). Recently, BBr has been confirmed to have versatile functions such as anti-inflammation, anti-bacterial, hypoglycemic, cholesterol-lowing, clearance of oxygen radical, anti-cancer and a well-documented effect against dysfunction in various brain nervous systems (Kumar et al., 2015). It has also been verified effective in cerebral ischemia, hypertension, hyperlipemia, mental depression, anxiety and AD (Abd El-Wahab et al., 2013; Kumar et al., 2015). Besides, it exhibited great effects in the face of intestinal mucosal barrier dysfunction in type 2 diabetic rats (Gong et al., 2017).

Fear-stimulating events can persist even over a lifetime and is widely considered to be a highly persistent and durable form of memory (Bergstrom, 2016). The mPFC modulates the expression of contextual fear conditioning (Lisboa et al., 2010). Increasing evidences indicated that contextual fear condition assay could be applied to appraise the lesion in PFC of AD (Moreira et al., 2012; Zelikowsky et al., 2013). However, there are no investigations concerning about whether BBr could suppress DM-induced fear-conditioned cognitive defect. In our study, the contextual fear condition assay was applied to detect the effects of BBr on the fear-conditioned cognitive function of diabetic rats. Moreover, we expounded the probable mechanisms that BBr improved in ordinate inflammation and insulin pathway in mPFC of DM as well.

The cognitive changes induced by diabetes have been acknowledged. Most clinical researches proved the connectivity around the mPFC was impaired in patients with T2DM (Cui et al., 2015). Recent results also suggest that DM could partly affect the dendritic morphology in the occipital cortex, prefrontal cortex and hippocampus (Martinez-Tellez et al., 2005). Aβ accumulation in the brain is the feature of aging and cognitive ability is decreased during the development of pathophysiology, especially in AD and MCI induced by DM (Ma et al., 2015). A number of studies had proved that inflammation promoted the Aβ peptide to aggregate rapidly (Philippens et al., 2016). Moreover, inflammation also disturbed the insulin signaling pathway; conversely, aberrant insulin signaling pathway exacerbated the inflammation production (Nandipati et al., 2016), and led to more Aβ peptide aggregation and impaired cognitive function (Philippens et al., 2016). This study indicated for the first time that monomeric BBr administration modulated aberrant insulin signaling pathway, reduced inflammatory mediators and improved the lesion of cognitive in DM rats.

Met is widely used as a first-line treatment for patients with T2DM. Previous studies demonstrated that Met could rapidly cross the blood-brain barrier (Labuzek et al., 2010) and had several beneficial effects in the brain, such as anti-inflammatory (Wang et al., 2016) and neuroprotective effects (Gupta et al., 2011). Meanwhile, Met could alleviate the impairment of cognitive function in db/db mice (Li et al., 2012; Chen et al., 2016). Many clinic trails were also shown that long-term treatment with Met could reduce the risk of cognitive decline (Guo et al., 2014; Ng et al., 2014). So Met is one of the best candidates for the positive control in the DM-induced cognitive impaired model.

Here, we used the contextual fear condition assay to test BBr on the improvement of the cognitive function. On the 1st day, we found the freezing percent in DM group was decreased (Figure 1B) when compared to NOR. BBr group seemingly quickly remembered the influence of stimulation and significantly increased the freezing percent (Figure 1B). After 1.5-h intervals, the percent freezing in DM groups was almost recovered to NOR level (Figure 1C). The same results also emerged in the drug groups in our study. This might indicate the retrieval of information in short-term memory, no matter DM group and the treatment groups, almost all recovered up to normal level. Besides, the results probably also suggested that BBr group dealt with problems more quickly when compared to DM group (Figures 1B,C). On the next day, the freezing percent in DM group was decreased once again than in the 1st day when detection was executed after 1.5 h, but no significant changes were found in BBr (Figure 1D). These observations indicated that BBr accelerated the reinforcement of the information. Meanwhile, BBr still significantly augmented the freezing percent when compared to DM after 24 h. The same results occurred in the 3rd day, and Met also ameliorated the fear memory defect. It may imply that long-term memory, rather than short-term memory is impaired in DM and BBr could effectively inhibit the cognitive impairment. Moreover, BBr seemed to reduce cognitive impairment which caused by DM earlier than Met (Figures 1D,E).

Desensitized insulin receptor signaling was involved in both AD and DM. In most of the previous literature, it had been demonstrated that venenous Aβ oligomer caused partial loss of insulin receptor in dendritic spine mainly in aged primary hippocampus neurons, with increased distribution of insulin receptor in the cell soma but no alternations in absolute IR subunit levels, rendering insulin receptor non-responsive in the AD brain (Zhao et al., 2008). In the present study, our results also showed that no distinctive changes of the IR protein levels in mPFC of the four groups (Figure 2A). Of course, this result needs more experiments to classify whether IR internalization was generated. Phospho-Ser307 in IRS-1 blocks the interaction between the IRS-1 PTB domain and insulin receptor, which might weaken the coupling between the two proteins (McEwen and Reagan, 2004) and lead to insulin resistance. Our results distinctly revealed the increased p-IRS-1^ser307 in DM group was significantly reduced by BBr. The condition of insulin resistance in the mPFC of DM rats was also inhibited by Met (Figure 2B). Additionally, the ROS generation which mediated by HG-induced also activated the aberrantly PI3K/Akt signaling pathway (Lee and Han, 2010). Chronically excessive activation of Akt also caused the development of Tau pathological and Aβ42 generation under insulin resistant situation in db/db and HSFD mouse brains (Kim et al., 2015). In our study, activated PI3K/ Akt signaling pathway of DM-induced was remarkably inhibited in BBr and Met groups (Figures 3A,B). Similarly, the excessive activation or overexpression of GSK3β could trigger a series of pathological alterations, most of which are hallmarks of T2DM and AD (Gao et al., 2011). Additionally, overproduction of APP also obviously activated the GSK3β in the hippocampus of APP-V717I amyloid transgenic mice (Terwel et al., 2008). In our research, the increased p-GSK3β^Ser9 protein was significantly rectified by BBr in the mPFC of DM as well (Figure 3C). In additionally, the activation of mTOR signaling pathway was involved in inflammation transduction (Xiao et al., 2017) and facilitation of neurodegeneration in AD (Wang et al., 2014). Obviously, the decreased mTOR substrate p-4E-BP1 (Thr37/46) in our experiment (Figure 3D) indicated that the PI3K/Akt/mTOR signaling was activated. This is coherent to other group (Karki et al., 2016). In summary, these results ascertained that activated PI3K/Akt/mTOR in DM group might accelerate the generation of Aβ (Figures 3A–D).

There is a close relationship between the impairment of cognitive function and the activation of inflammatory pathways (Li Y.B. et al., 2016). The transcription factor NF-κB was a key protein involved in cellular responses to stimulations such as oxidation stress, ultraviolet irradiation, bacterial, cytokines and so on. The release of the NF-κB is an initiation of the degradation of IκB proteins via activation of a kinase IKK. Our previous researches indicated that BBr attenuates the production of inflammatory mediators through suppression of TLR4 (toll-like receptor 4)/NF-κB signaling in intestine tissue of diabetic rats. The present work was performed to study the anti-inflammation effect of BBr in mPFC of DM rats. The activation of n-PKC also attributed to the upregulation of NF-κB. We speculated that BBr not only reduced the phosphorylation levels of IKKβ and NF-κB (Figures 4A,B), but also suppressed the upregulation of n-PKC (Figure 5C). Moreover, as showed in Figure 4C, the expression of FGAP in DM group was inhibited by BBr as well as Met. Meanwhile, the increased protein levels of pro-inflammation IL-18 (Figure 4D) in DM group were significantly decreased by BBr in the supernatant fluids of the tissue homogenate, but few effects on IL-1β and TNF-a (Figures 4E,F) were found. In additionally, the activation of MAPK signaling pathway was involved in inflammation transduction (Zhang et al., 2017) and facilitation of neurodegeneration in AD (Kirouac et al., 2017). Our results also ascertained that activated JNK (Figure 4H) was showed in DM group and BBr decreased the increment. The sustained neuroprotective effects of BBr observed in the present study is consistent with the research about protecting against neuronal damage via suppression of inflammation in traumatic brain injury (Chen et al., 2014). Taken together, our findings suggested that BBr developed its neuroprotective function against cognitive deficits in DM rats via inhibiting the production of inflammation in astrocyte.

Then we focus on the uptake and metabolism of glucose in the mPFC of DM brain. In vivo PET imaging is a common method to detect the uptake of glucose in brain. The impairment of glucose uptake in the brain of T2DM aggravated the pathological alteration. This phenomenon also emerges in our research. In Figure 5A, the shiny red color which represented the uptake of glucose in NOR was completely disappeared in the DM. The therapeutic drug BBr obviously recovered the uptake of glucose. Met had a slightly increment. Of note, the high density of GLUT3 in neurons was pivotal in the energy supplement and neuronal activities. Moreover, the impaired glucose uptake/metabolism was also inversely correlated to the densities of NFTs (Neurofibrillary tangles) (Liu et al., 2008). In our study, the decreased GLUT3 was presented in the mPFC of DM rats (Figure 5B). Traditional Chinese medicine BBr could largely enhance GLUT3 expression and reverse the lesion. Emerging evidence indicated that protein kinase C epsilon (PKC𝜀) participated in insulin-regulated trafficking of the glucose transporter GLUT4 in adipocytes (Tsuchiya et al., 2013). Then, we determined the expression level of PKC and found that these conclusions were more or less discrepant with our discoveries. As show in our results (Figure 5C), both of two novel Ca²⁺-insensitive isoforms of PKC were activated under insulin resistance. The distinct scene might be a negative feedback regulation in the mPFC, and BBr effectively blunt the increase extent of n-PKC (novel-PKC) in DM rats as well as Met. This was coherent to some researches (Brownlee, 2001; Giorgi et al., 2010) that more than one PKC isoform in DM were activated by hyperglycemia, since many PKC isoforms were DAG-dependent style. Nevertheless some reports indicated that PKC𝜀 activation was required for recognition memory in the rats’ hippocampus (Zisopoulou et al., 2013). Meanwhile, the translocation of NF-κB was also increased in neuron of mPFC of DM rats. This might indicate that the neuron in DM group had some capacities to resist the generation of damages. But, the exact molecular mechanism needs further investigation. Overall, related proteins were activated in mPFC of DM in the insulin signaling pathway except GLUT3 in neuron, this scene might involve in some other cells activation but not neuron, just as the astrocyte. These results indicate that BBr not only facilitates glucose metabolism in mPFC, but also promotes the uptake of glucose, even in the DM rats under cognitive impairment.

It is well known that Aβ42 is the primary pathogenic factor in AD. Growing evidences indicated that DM was one of the upper most risk factors for developing AD and accelerated Aβ42 pathological changes (Moran et al., 2015). APP is best known as the precursor molecule whose proteolysis generates Aβ. It was reasonable to observe the incremental levels of APP and BACE-1 in mPFC of DM (Figures 6A,B) because their increases also appeared in monkey with DM (Okabayashi et al., 2015). The administration of BBr can reduce the incremental levels of APP and BACE-1 in mPFC of DM (Figures 6A,B) and clean the oligomeric toxic protein Aβ42 (Figure 6C). However, it remains unknown how BBr promote the reduction of APP levels. In this regard, it still needs great efforts to examine the underlying mechanisms about the effect on transcriptional inhibition or translation degeneration. As the mentioned above, insulin sensitization agent Met reduced the APP level, but the enzyme BACE-1 level was largely augmented. This co-action might accelerate the Aβ42 production as in DM and this discovery had been reported before (Chen et al., 2009). There are maybe some reasons, such as rapid degradation or some other kinase activities were enhanced, ascribe to the weaker function in Met group.

Our present findings (Figure 7) suggest that BBr not only modulated the aberrant inflammation responses, but also regulated some in ordinate proteins of insulin signaling pathway in the mPFC of DM rats. Moreover, it promoted the uptake of glucose in the brain of DM rats and increased the metabolism of glucose in neuron. These results probably more clearly elucidate that BBr exhibits the importance of multi-targets in the mPFC of DM. Moreover, these apparently protective actions of BBr prevent diabetes-associated cognitive decline in DM rats.

The study protocol was in agreement with Animal Care and Use Committee affiliated to the Huazhong University of Science and Technology.

QC and RM performed most of the experiments. NW, XZ, and CS participated in data analyses. JG and KF contributed to the discussion of the manuscript. JL, DW, and DY helped to the preformation of the experiments and collected the samples. KW and JC participated in its design and coordination. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.