MiR‐132 down‐regulates high glucose‐induced β‐dystroglycan degradation through Matrix Metalloproteinases‐9 up‐regulation in primary neurons

Abstract Cognitive dysfunction is one of the complications of diabetes. Unfortunately, there is no effective methods to block its progression currently. One of the pathophysiological mechanisms is synaptic protein damage and neuronal signal disruption because of glucose metabolism disorder. Dystroglycan protein, located in the post‐synaptic membrane of neurons, links the intracellular cytoskeleton with extracellular matrix. Abnormal expression of dystroglycan protein affects neuronal biological functions and leads to cognitive impairment. However, there are no relevant studies to observe the changes of β‐dystroglycan protein in diabetes rat brain and in primary neurons under high glucose exposure. Our data demonstrated the alterations of cognitive abilities in the diabetic rats; β‐dystroglycan protein degradation occurred in hippocampal and cortical tissues in diabetic rat brain. We further explored the mechanisms underlying of this phenomenon. When neurons are exposed to high glucose environment in long‐term period, microRNA‐132 (miR‐132) would be down‐regulated in neurons. Matrix Metalloproteinases‐9 (MMP‐9) mRNA, as a target of miR‐132, could be up‐regulated; higher expression and overlay activity of MMP‐9 protein could increase β‐DG protein degradation. In this way, β‐DG degradation may affect structure and functions among the synapses, which related to cognition decline. It may provide some theoretical basis for elucidating the molecular mechanism of diabetes‐induced cognitive dysfunction.


Diabetes mellitus (DM) is concerned with cognitive dysfunction.
Most studies reported the aggregate outcome that the incidence of 'any dementia' in diabetic patients is higher than that in non-diabetic patients; many studies determined the intensity of the association between diabetes and dementia. 1 Cognitive changes in patients with type 2 diabetes mainly include impaired memory and learning ability. 2 Diabetes can affect the occurrence and development of dementia through a variety of pathophysiological mechanisms. Although the mechanisms are complex and diverse, it generally includes glucose toxicity, lipid toxicity, insulin resistance and so on. [2][3][4] Early studies have reported that high glucose toxicity can slowly cause progressive structural and functional abnormalities in the brain. 5 Several lines of evidence also suggested rodents suffered from chronic hyperglycaemia had abnormal synaptic plasticity and cognitive impairments. 6 Thus, we concluded that toxic effects of high glucose on synaptic proteins may aggravate cognitive impairments. However, there is still many room for the high glucose effects on synaptic proteins.
There exists a large quantity of research on the role of dystroglycan in the function of neuromuscular synapses, opposed to a relatively small number of studies that deal with dystroglycan at inter-neuronal synapses. Dystroglycan protein widely exists in cortex and hippocampus neurons in brain. 7 Dystroglycan protein is located in the post-synaptic membrane of neurons; the protein connects the intracellular cytoskeleton and extracellular matrix (ECM). 8 The dystroglycan propeptide is translated from a single mRNA, which is subsequently cleaved into two non-covalently associated subunits: extracellular-dystroglycan (α-DG) and transmembrane-dystroglycan (β-DG). 9 The α-DG is a kind of peripheral membrane protein, which can be connected to many extracellular matrix parts 10 and presynaptic neurexins. 11 The β-DG connects dystrophin glycoprotein complex (DGC) to F-actin cytoskeleton through proline-rich C terminus. 12,13 Dystroglycan may play a key role in synaptic plasticity and cognition. It has been proved that mutant rats that are deficient in dystroglycan protein selectively have cognitive impairments and insufficient long-term potentiation effect. 14,15 Dystroglycan protein may also be affected by hyperglycaemia. Recent studies reported that the β-DG protein expression significantly reduced in skeletal muscle of Goto Kakizaki rat model of type 2 diabetes. 16 However, there are no relevant studies to observe the changes of dystroglycan protein in neurons under the high glucose toxic.
MicroRNAs (miRNAs) are short non-coding RNAs, which are important regulators for controlling many biological processes and post-transcriptional gene expression. They usually bind specifically to the 3′ untranslated region (3′ UTR) of messenger RNA (mRNA) and collect RNA induced silence complex. [17][18][19][20] It has been found that miRNAs are particularly rich in brain and give scope to neuronal plasticity, including synaptic protein synthesis and senior functions of brain. [21][22][23][24] The miR-132 is abundant in human and rat brain, especially in the hippocampi or cortices. Abnormal miR-132 expression in a certain kind of cell or tissue may be involved in diseases processes. 25 However, the expression of miR-132 in neurons under the high glucose exposure has not been reported. Up to now, several targets of miR-132 have been mentioned, including P300, MeCP2, Foxp2 and Matrix Metalloproteinases-9 (MMP-9). [26][27][28][29] The miR-132 could pair to the 3' UTR of target mRNA thus promote mRNA degradation and repress translation.
Matrix Metalloproteinases (MMPs) are a family of proteolytic enzymes that are able to affect cell behaviour through cleaved structural components of various extracellular matrix molecules. 30 The MMPs subfamily of gelatinases, especially MMP-9 and Matrix Metalloproteinases-2 (MMP-2), is currently the widely studied. The expression level of MMP-9 in hippocampus of adult rodent is especially high and MMP-9 co-localizes with pre-synaptic and postsynaptic markers. 31 It is worth mentioning that dystroglycan and MMP-9 proteins jointly play an essential role in neuronal function.
On the one hand, the gene absence of dystroglycan gene (DAG1) in the brain leads to neuronal plasticity defect, similar to the inhibition of MMP-9. 32,33 On the other hand, there is evidence that MMP-9 may degrade the β-DG to set free a 30-kD product from the 43-kD full-length subunit. 34 Furthermore, the proteolysis of dystroglycan by MMP-9 results in the allosteric regulation of neurexins and the rest of dystroglycan, which in turn produces specific structural and functional changes in the pre-synaptic and post-synaptic cytoplasmic environments. 35 However, the expression level of MMP-9 and the regulated role of MMP-9 in dystroglycan degradation in neurons under the high glucose exposure has not been reported.
Based on the above evidence, we suggested that high glucose toxic may induce the β-DG degradation and aggravate cognitive impairments, such as learning and memory; miR-132 and MMP-9 might be involved in the mechanisms underlying of this phenomenon. In this report, we provided several lines of evidence supporting a notion that miR-132 down-regulates high glucose-induced β-DG degradation through MMP-9 down-regulation in primary neurons; ultimately, the changes in the pre-synaptic and post-synaptic cytoplasmic environments affect transmission of information among the synapses and cognition function. It might provide valuable insights for elucidating the molecular mechanism and finding therapeutic strategies for the treatment of diabetes-induced cognitive dysfunction.

| Animal model and treatment
Specific pathogen-free (SPF) SD rats aged 4 weeks were purchased from Shanghai Bikai Laboratory Animal Co. Ltd. Rats were housed in SPF environment and given food and water. A 12-h/12-h light/ dark cycle was maintained. After one month of adaptive feeding, the rats were randomly divided into two groups: (a) sham operated control rats (control group); (b) streptozocin (STZ) and high-fat diet (HFD) induced type 2 diabetes rats (DM group). The DM group rats were fed with HFD for 4 weeks and then injected once with STZ (in the tail vein at 50 mg/kg bodyweight) to induce partial insulin deficiency, followed by continued HFD feeding for an additional 4 weeks. Then, the rat presence of hyperglycaemia (fasting blood glucose level ≥11.1 mM and non-fasting glucose levels of ≥16.7 mM) were used for the next experiments. The control group rats were fed with control diet with 10 kcal% fat and injected same dose of saline. All procedures were conducted in accordance with the National

| Morris Water Maze
Spatial learning and memory were assessed by the MWM test.
The test consisted of 5-day training and a probe trial on day 6.
Rats were individually trained in a circular pool (diameter 120 cm, height 50 cm) containing 30-cm high water. The water was rendered opaque by the addition of black non-toxic ink and maintained at about 25℃. A submerged platform (9 cm in diameter, 2 cm below the surface of water) was placed in the centre of one quadrant of the pool, and its position was fixed throughout the training sessions. There were 5 rats in the control group and the same numbers in DM group. Rats were given four trials of 90 s per day for five consecutive days. Animals were guided to the platform if it was not located within 90 s. All the rats remained on the platform for 10 s before removal. The time (escape latency) that elapsed until the rat reaches the platform was noted each day. After the acquisition phase, rats were given one probe trial of 60 s for which the platform was removed from the pool on day 6. The number of times that the animals swim over the platform location and the pathways that elapsed were evaluated. Data of the escape latency, the number of platform location crossings and the pathways were collected by the video tracking equipment and processed by a computer equipped with an analysis-management system ('Dr Rat' Animal Behavior Analysis System, version 2.20).

| Primary neuronal isolation, culture and reagents
SPF E18 SD rats aged 10-12 weeks were purchased from Shanghai Bikai Laboratory Animal Co. Ltd. All procedures were conducted in accordance with the National Institutes of health guidelines for the care and use of laboratory animals. Make every effort to reduce the number of animals used and their suffering. During the experiment, ensure that the pregnant rats are healthy and no stillborn, otherwise they are excluded. The pregnant rats were anaesthetized by carbon dioxide. The methods referred to several articles about inhalation anaesthetics. [36][37][38] When the rat loses response to toe pinch, the cervical dislocation was carried out. Spray the dam's abdomen area with 75% ethanol. Then, make a incision along the midline of abdomen to expose the uterus. The uterus is then exposed along the midline of the abdomen. Remove the embryos and place individual ones in a petri dish with cold Hank's Balanced Salt Solution (Beyotime). Then, the embryonic brain is dissected and immediately placed in an ice vessel containing Dulbecco's modified Eagle's media (DMEM, Gibco). Hippocampi or cortices tissues were isolated from embryos by using a dissecting microscope. Tissues are digested with 0.125% trypsin for 30 minutes at 37℃, 5% carbon dioxide and shaken every 5 minutes; then, the digested tissues were terminated in DMEM containing 10% FBS (Gibco, BRL, USA).

| Transfection of primary neurons
The miR-132 mimics and its control, miR-132 inhibitors and its control, small interfering RNAs (siRNAs) targeting MMP-9 and the negative control siRNA were designed and constructed by RiboBio.

| Extraction of total RNA and microRNA
The neurons were gathered, and all RNA were extracted with TRIzol reagent (Invitrogen, Thermo Fisher Scientific) according to the manufacturer's protocol. For all RNA samples, the Nanodrop (Thermo Fisher Scientific) was used to detect the purity and quality.

| ELISA
After the treatment, the supernatant of the culture medium was collected. The concentrations of MMP-9 were determined using Rat MMP-9 ELISA Kits (YCBIO). Then, the following procedures were performed following the manufacturer's instructions.

| Gelatin Zymography
The neurons were treated, and the supernatant of the culture medium was collected. The activity of MMP-9 was determined using MMP gelatin zymography electrophoretic analysis technique. MMP gelatin zymography kit was purchased from GENMED SCIENTIFICS INC. Samples of the culture medium were mixed with APMA (GENMED SCIENTIFICS INC.) and incubated at 37 ℃ for an hour before electrophoresis. Then, the following procedures were performed following the manufacturer's instructions.

| Statistical analysis
All experimental procedures strictly follow the principle of randomization and double-blind. All experiments have been replicated at least 3 times. The Western blot images were quantified with ImageJ Software (version 1.51k, NIH). All of the data were presented as mean ± standard error mean (SEM). Student's t test was performed as appropriate using GraphPad Prism 6. A probability value (P) of less than .05 was considered to be statistically significant. All figures were put together in Adobe Photoshop CS6.

| Alterations of cognitive abilities and β-DG in the diabetic rats
Spatial learning and memory were evaluated on the basis of MWM performance in diabetes rats (DM group) and sham operated control rats (control group). The behavioural training protocol is presented in MWM method. The time (escape latency) that elapsed until the rat reaches the platform was noted during acquisition trials (days 1-5). The number of times that the animals swim over the platform location and the pathways that elapsed were evaluated during the probe test (day 6). We compared learning curves of two group rats.
As shown in Figure 1A, the average escape latency of the diabetes rats was longer than those of control rats every day during the learning phase. A considerable difference was also observed between the performance of diabetes and control rats on day 3 and 4. As depicted in Figure 1B, during the probe test, the diabetes rats crossed the platform less often than the control rats; control rats spent significantly more time in the platform quadrant (the green circle) whereas diabetes animals swam randomly. We suggested that the spatial learning and memory decline in diabetic rats was related to dystroglycan protein degradation. Thus, we took out hippocampal and cortical tissues from the two group rats to detect the changes of β-DG protein. The full-length forms of 43-kD and degradation forms of 30-kD β-DG were detected by Western blot. Higher levels of 30/43-kD β-DG (P < .05) were observed in the diabetic rats compared to the control group ( Figure 1C,D). Thus, we concluded that the toxic effects of hyperglycaemia could significantly induce β-DG degradation of in both hippocampal and cortical tissues in brain.

| High glucose induces β-DG degradation in primary neurons
To investigate the 'toxic' effects of hyperglycaemia on neurons, we cultured hippocampal and cortical neurons and the cultures were treated with high glucose for 48 h. The control group was treated with mannitol to adjust osmotic pressure. As shown in Figure 2B, there were no obvious changes in neuron morphology observed through a microscope. Cell viability of primary neurons with high glucose treatment was lower than that of the control group ( Figure 2C).
Based on the above hypothesis, we verified the high glucose effect on β-DG degradation. Western blot was used to detect the fulllength 43-kD β-DG and the degradation 30-kD β-DG, in high glucose and normal glucose groups. In Figure 2D,E, higher levels of 30/43-kD β-DG were observed in the high glucose group compared to the control normal glucose group. As a result, we concluded that high glucose could significantly induce the degradation of β-DG in both primary hippocampal and cortical neurons.

| High glucose decreases miR-132 expression in primary neurons and miR-132 down-regulates high glucose-induced β-DG degradation in primary neurons
Pre-miRNA has two arms, 5p and 3p, respectively. In many cases, only one arm is processed as a mature miRNA. However, in some

| High glucose increase the MMP-9 expression via decreasing the levels of miR-132
As mentioned above, miRNAs mostly promote mRNA degradation and translation by binding to the sequences of target mRNA. To explore the regulatory functions of miR-132 in the high glucoseinduced β-DG degradation, we screened authoritative databases of predicting target genes of miRNAs to identify target genes of the miR-132. Thus, we looked through the following databases: PicTar, GSEC, microRNA and miRSponge. Useful information from four databases was summarized to draw the Venn diagram ( Figure 4A).
The results showed that miR-132 may directly bind to the family of MMP. In addition, it has reported that miR-132 can directly bind to the 3' untranslated region (UTR) of the MMP-9 mRNA in neurons.
The binding site is shown in Figure 4B. 29 Thus, we suggested that there were regulatory functions of MMP-9 in the high glucose-induced β-DG degradation. It has been reported that MMP-9 is produced at the synapse and released into the extracellular space in a precursor form, and then, it is activated by proteolytic cleavage. 34 It was demonstrated that only secreted and activated MMP-9 causes the degradation of its extracellular substrates. Therefore, we used ELISA, Western blot and gelatin zymography to detect the expression and activity levels of MMP-9 protein in cell culture. The results in Figure 4C-H and Figure S1A,B showed that high glucose could significantly up-regulated MMP-9 mRNA and protein expression, as well as protein activity in primary neurons. In detail, the qRT-PCR analysis showed MMP-9 mRNA expression in hippocampal ( Figure 4C) and cortical neurons ( Figure 4D) under high glucose exposure was significantly higher than those of control one. ELISA and Western blot analysis showed the MMP-9 protein expression in hippocampal ( Figure 4E, Figure S1A) and cortical neurons ( Figure 4F, Figure S1B) under high glucose treatment was higher than those of control group. Gelatin zymography analysis showed the MMP-9 activity in hippocampal ( Figure 4G) and cortical neurons ( Figure 4H) exposed under high glucose was higher than those of control one.
In order to validate that MMP-9 is miR-132 target gene in neurons under high glucose exposure, we transfected miR-132 mimics or miR-132 inhibitor into hippocampal and cortical neurons and then followed by stimulation with high glucose. The qRT-PCR ( Figure 4I,J), Western blot ( Figure S1C,D) and ELISA data ( Figure 4K,L) indicated that the miR-132 mimics down-regulated MMP-9 mRNA and protein expression compared to the mimic negative control under the same glucose concentration, whereas the miR-132 inhibitor up-regulated MMP-9. Therefore, data from previous studies and our findings indicated that MMP-9 mRNA was a target of miR-132 in primary neurons exposed by high glucose. Through the above experiments, we concluded that high glucose could increase the expression of MMP-9 via decreasing the levels of miR-132.

| Suppression of MMP-9 inhibits high glucoseinduced β-DG degradation in primary neurons
It has reported that MMP-9 may digest the β-DG to release a 30-

| The miR-132 inhibits the degradation of β-DG via down-regulation of the expression level of MMP-9
Follow on, we examined whether β-DG degradation through miR-132 inhibitor was attenuated by the knockdown of MMP-9. Co-transfect primary neurons with MMP-9 siRNA and miR-132 inhibitor for 48h.
Then, the β-DG expression in neurons was examined by Western blot. The graphs in Figure 6 showed the ratio of 30/43-kD β-DG F I G U R E 2 The primary neurons were cultured in vitro, and the effects of high glucose exposure on them were detected. (A) Images of cultured primary neurons. Images of cultured hippocampal and cortical neurons at different stages during culture-DIV 1, 3, 5 and 7. Note how neuronal morphology and arborization develop over time. Scale bar = 100 μm (B, C) Primary neurons were treated with high glucose. The control group was treated with mannitol to adjust osmotic pressure. No obvious changes in neuron morphology observed through a microscope. Scale bar = 100 μm (B). Cell viability of neurons with high glucose treatment was lower than that of the control group (C). (D, E) Western blot was used to detect the full-length 43-kD β-DG and the degradation 30-kD β-DG. Figures showed higher levels of 30/43-kD β-DG in the high glucose group compared to the control normal glucose group. Data represent mean ± SEM. *P < .05, **P < .01 detected by Western blot. As shown in Figure 6A

| Schematic diagram of the proposed model shows how high glucose-induced β-DG degradation
Above all of our findings, we proposed the model of high glucose inducing β-DG degradation. The schematic diagram was showed in Figure 7. When neurons are exposed to high glucose environment in long-term period, miR-132 would be down-regulated in neurons.
MMP-9 mRNA, as a target of miR-132, could be up-regulated; higher expression and overlay activity of MMP-9 protein could increase β-DG protein degradation. In this way, β-DG degradation may affect structure and functions among the synapses, which related to cognition decline. (I-L) Transfected miR-132 mimics or miR-132 inhibitor into neurons and then followed by stimulation with high glucose. Levels of MMP-9 mRNA were measured by qRT-PCR in hippocampal neurons (I) and cortical neurons (J). Levels of MMP-9 protein were measured by ELISA in hippocampal neurons (K) and cortical neurons (L). Data represent mean ± SEM. *P < .05, **P < .01, ***P < .001, ****P < .0001 compared with corresponding control group It has also proved that dystroglycan might have a post-synaptic role in cognitive impairment, such as learning and memory. Moreover, deficient in dystroglycan protein has insufficient long-term potentiation effect and synaptic plasticity in the brain. 14,15 Subsequently, neuronal activity decreases and cognitive impairment occurs.

| D ISCUSS I ON
MicroRNA is essential for the normal development of brain and its function. 40 The miR-132 in neurons has been widely reported to play an important role in the development of nervous system. 41 In our report, high glucose down-regulated the expression of miR-132. This is consistent with previous studies that miR-132 is downregulated in AD patient brain. 25  has not been reported. We guess the reason for this phenomenon is the difference of steady-state regulation ability. Cell homeostasis is a kind of dynamic balance. When we treated neurons with high glucose, cell homeostasis would be destroyed and the cells would initiate protection through a series of complex mechanisms. In cortical neurons, the miR-132 content under 48 hours of high glucose treatment is higher than that under 24 hours. But the miR-132 content is opposite in hippocampal neurons. We speculate that cortical neurons have a stronger ability to regulate homeostatic imbalance compared hippocampal neurons under 48 hours of treatment. In this way, miR-132 content could be compensatory increased, but it was still lower than that of the control group.
The miR-132 can also affect the expression of some synapseassociated proteins. In this study, we found that the miR-132 decreased the glucose-induced β-DG degradation. These experiment results imply that miR-132 is involved in glucose-induced β-DG degradation and serves as a negative feedback regulator. Previous studies also demonstrated that overexpression of miR-132 could protect the neurons. This protection improves learning and memory. 42 It has been known that MMP-9 is involved in synaptic plasticity in cognitive processes. The abnormal MMP-9 activity may lead to cognitive dysfunction. 32 Our study found that high glucose significantly increased the expression and activity levels of MMP-9 in primary neurons. It is not inconsistent with those previous reports that up-regulation of MMP-9 can aggravate many neurological diseases, including stroke, PD and AD. 30,43 Moreover, previous studies and our findings indicated that MMP-9 mRNA was the target of miR-132 in neurons. Thus, we further verified that high glucose increased the expression of MMP-9 via inhibiting the levels of miR-132. Moreover, our data indicated that miR-132 regulated β-DG degradation through down-regulation of MMP-9. Other studies also had shown that the specific metalloproteinase activity of β-DG was associated with MMP-9.

F I G U R E 7
The schematic diagram showed the effect of miR-132 and MMP-9 on β-DG degradation under high glucose exposure. At the steady state (A), dystroglycan protein extends from the post-synaptic side into the ECM to binds neurexins, which located in the pre-synaptic sides. When neurons are exposed to high glucose environment in long-term period (B), miR-132 would be down-regulated in neurons. MMP-9 mRNA, as a target of miR-132, could be up-regulated; higher expression and overlay activity of MMP-9 protein could increase β-DG protein degradation. In this way, high glucose could induce the β-DG degradation, consequently might affect synaptic plasticity and neuronal signalling, as well as cognitive impairment, such as learning and memory. Ultimately, β-DG degradation may affect structure and functions among the synapses, which related to cognition decline Through the above experiments, our results showed that high glucose significantly increased the degradation of β-DG in primary neurons. We further explored the mechanisms underlying of this phenomenon. In summary, we concluded that miR-132 could downregulate high glucose-induced β-DG degradation through MMP-9 down-regulation in neurons; ultimately, β-DG degradation might affect structure and functions among the synapses, which related to cognition decline. It may provide some theoretical basis for elucidating the molecular mechanism of diabetes-induced cognitive dysfunction. At the same time, it also provides new ideas for clinical treatment of diabetes-induced cognitive dysfunction.
Although we believe the experiment is adequate and the argument is sufficient, there are still some shortcomings and disputes.
Firstly, in the process of culturing neurons, the primary neurons gradually mature after 5 days of culture. [44][45][46] We found that the neurons became stable and mature around day 7. In most of the literatures with similar purposes, the authors chose to treat neurons cultured for about 7 days. 47,48 Thus, we chose to intervene at this time. Although we believed that the cells were intervened during the optimal processing period, there were indeed limitations. Our results were validated on the 10-day-old neuronal cultures, but not on more mature or immature neurons. In addition, the dystroglycan protein is composed of two non-covalently associated subunits: α-DG and β-DG. 8 However, we only explored the β-DG degradation under the high glucose but ignored the changes of α-DG. Current research suggests that the functions of two subunits are consistent. 10,12 Thus, it can be assumed that if β-DG subunit is damaged, the structure and function of the dystroglycan protein could be damaged. Yet it is warranted to examine the new theories about dystroglycan in the inter-neuronal junction and to subsequently check for possible α-DG changes under high glucose toxic in the future. Besides, a small increase in cell death was detected in neurons exposed to high glucose in our experiment, which is consistent with many previous studies where apoptosis was found in neural cells exposed to high glucose [47][48][49][50][51] and in the hippocampus of diabetic animals. 52,53 But it has been reported in the literature that 50mmol/L HG Neurobasal Medium does not cause cell death. 54 Thus, the methods of high glucose treatment in each literature were not exactly the same, such as the high glucose treatment time and the purity of neurons, and the results might be different. As long as it is handled properly, the results are reliable.

ACK N OWLED G EM ENTS
The authors appreciate the support provided by the National

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the first author upon reasonable request.