Improved Promotion of M2 Microglial Polarization by Zuogui Jiangtang Jieyu Formulation in Diabetes-Related Depression


 Background Zuogui Jiangtang Jieyu formulation (ZGJTJY) is a Chinese polyherbal prescription for diabetes-related depression (DD). The mechanism underlying hippocampal M1/M2 polarization in DD and the ZGJTJY treatment effects remain unclear. This study aimed to investigate M1/M2 microglial polarization in the hippocampus of DD rats and HAPI (highly aggressively proliferating immortalized) cells simulating the DD state, as well as to examine the ZGJTJY intervention effects, both in vivo and in vitro. Methods We subjected Sprague Dawley rats to a high-fat diet, streptozotocin, and unpredictable chronic mild stress; subsequently, we orally administered ZGJTJY. HAPI cells were induced using high glucose and corticosterone; subsequently, ZGJTJY-containing serum was added to examine changes in M1/M2 microglial polarization. Moreover, metformin combined with fluoxetine (DMGB/F) was used as a positive drug for evaluating the ZGJTJY intervention. Laser confocal scanning was used to examine the microglial morphology. Further, real-time PCR was used to determine M1 markers (MHCII, iNOS, MCP-1, CD11b), M2 markers (Arg1, Mrc1, Ym1), pro-inflammatory cytokines (IL-1β, IL-6, TNF-α), and anti-inflammatory cytokines (IL-4, IL-10). Additionally, an enzyme-linked immunosorbent assay was used to examine inflammatory cytokines. Results There was significant activation of M1 polarization in the hippocampus of DD rats and HAPI cells induced using high glucose and corticosterone. Compared with DMGB/F, ZGJTJY inhibited and promoted M1 and M2 polarization, respectively; moreover, it decreased the M1-to-M2 polarization ratio both in vivo and in vitro. Conclusions The study indicated that hippocampal M1 polarization is crucially involved in DD pathogenesis; moreover, there is a need for further research on the neuroprotective effect of Chinese medicine associated with M2-polarized microglia.


Animal preparation and drug administration
We purchased 24 male Sprague-Dawley rats (200-220 g) from Slack Scene of Laboratory Animal Company (Hunan, China) and kept them in the SPF Laboratory Animal Center in Hunan Chinese Medicine University. All procedures were approved by the Ethics Committee of Hunan University of Chinese Medicine (No: ZYFT20171206) and conducted according to the guidelines for the care and use of laboratory animals from the National Institutes of Health. Fig. 1 presents the flow chart of the study protocol.
The animals were housed in an SPF room (22 °C ± 3 °C, 50% ± 5% humidity, and a 12/12-h circadian rhythm) with free access to water and diet. After 1 week of adaptive feeding, the rats were randomly divided into two groups. One group received a normal diet (containing 13.68% fat, 64.44% carbohydrate, and 21.88% protein) and citrate buffer once (i.v., the same volume as STZ). The other group received a HFD (p.o., 10% cholesterol, 0.2% propylthiouracil, 20% lard oil, 20% Tween 80, and 20% propylene glycol to still water, 10 ml/kg/d) for 14 days and STZ once (0.1 mol/L in citrate buffer [pH 4.5], 38 mg/kg i.v., fasting overnight before injection) at day 16. After three days, rats with a fasting plasma glucose level ≥ 16 mmol/L were selected and randomized into three groups: vehicle, ZGJTJY, and DMGB/F groups (n = 6 per group). Subsequently, the rats were subjected to 28 days of unpredictable chronic mild stress (UCMS), the Morris water maze test, open field test, and sucrose preference test for the successful establishment of the CUMS model as previously reported 2015;Liu et al., 2019;. Simultaneously, 10.26 g/kg/day ZGJTJY or a combination of 1.8 and 10.8 mg/kg/day DMGB and fluoxetine, respectively, were administered. The control group received an equal normal saline volume. A previous study showed that 10.26 g/kg/day ZGJTJY was the most effective dose for improving DD model rats and their hippocampal pathological changes . DMGB and fluoxetine doses were calculated as human-to-rat equivalent doses based on the body surface areas.
On day 48, rats were anesthetized using pentobarbital sodium (30 mg/kg, i.p.); subsequently, some whole brains were obtained after perfusion with saline and 4% paraformaldehyde (PFA) while the other brains were obtained to collect the hippocampus.

Plasma glucose detection
A single touch glucometer (One Touch Ultra 2; LifeScan, High Wycombe, UK) was used to determine glucose levels in plasma collected from the tail vein.

Morris water maze test
The Morris water maze was composed of a circular fiberglass pool (200 cm in diameter) filled with water (25 ± 1 °C) and made opaque using black non-toxic paint. The pool was surrounded by light blue curtains fixed with three distal visual cues. Four floor light sources with equal power provided uniform illumination in the pool and testing room. A charge-coupled device camera (kl-9511zh, Konlan Company, Shuozhou, China) was suspended above the pool center to record the swim paths of the animals; further, the video output was digitized using an EthoVision XT tracking system (Noldus Information Technology, Inc., Leesburg, VA, USA).
Four trials in each quadrant were conducted once a day for five days. The video analysis system tracked, recorded, and analyzed the swimming speed and time taken to locate the platform for each animal. Each trial lasted until the rat located the platform or for 60 s, with this time being recorded as the escape latency time (ELT); moreover, the learning outcome was the mean ELT for the last four days. On the final day, the platform was removed for a 60-s probe trial, with the time spent swimming in the platform quadrant being recorded as the space exploration time (SET).

Open field test
The open-field device was an 80 × 80 cm square box divided into 25 equilateral squares. The rat was placed in the central square; subsequently, we measured the number of squares crossed by the rat (only squares entered with all feet were included in the horizontal activity score) and the duration spent on hind limbs (the vertical activity score). Each rat underwent a 5-min test, which was scored by two observers with their average value being recorded. The sum of the horizontal and vertical activity scores was considered as indicative of the locomotor activity (LMA).

Sucrose preference test
Two bottles (1% sucrose solution vs. pure water) were individually presented to the rats for 24 h. After adaptation, the rats fasted for 12 h. Subsequently, both 1% sucrose solution and pure water were presented to rats for 15 h, with measurement of the pre-and post-test sucrose intake volume. Sucrose preference was defined as the ratio of sucrose to the total weight (sucrose + water).

Drug-containing serum preparation
ZGJTJY and DMGB/F doses were calculated as thrice the effective dose in 2.3. Male Sprague-Dawley rats were randomly divided into the ZGJTJY (oral ZGJTJY administration at 30.78 g/kg/day) and DMGB/F-treated groups (oral administration of metformin [5.4 mg/kg/day] and fluoxetine [32.4 mg/kg/day]). Each treatment was administered twice daily for three days. One hour after the last intragastric administration, all the rats were anesthetized using an intraperitoneal injection of 10% chloral hydrate; moreover, blood was collected through the abdominal aorta under sterile conditions. Subsequently, the serum obtained was sterilized through filtering through a 0.22 μm filter and stored at -80 °C until use.

Cell model preparation and intervention
HAPI cells were cultured in high glucose DMEM (10% fetal bovine serum, 5% CO2, 37 °C) followed by treatment with high glucose (150 mM) and corticosterone (200 μM) for in vitro simulation of the DD state, as previously reported . The control group was administered PBS. The ZGJTJY and DMGB/F groups were further administered with 10% volume of drug-containing serum while the control group received the corresponding serum. After fractionation, all the groups were tested for 24 h.

Immunofluorescence
Brains were collected and fixed in 4% PFA for 6-8 h, followed by paraffin embedding and slicing into sections. After blocking using 10% normal goat serum for 1 h at room temperature, the brain slices were incubated with primary antibodies: goat anti-Iba1 (1:500) in 1% BSA at 4 °C overnight. After washing with PBS for 3 × 5 min, the slices were exposed to secondary antibodies: donkey anti-goat IgG H&L (FITC) (1:1000) and DAPI (1: 1000) for 1 h at room temperature. Fluorescent images were captured using a confocal microscope (ZeissLSM800, Jena, Germany).
Cultured cells were fixed with 4% PFA; subsequently, they were incubated with primary antibodies, secondary antibodies, and DAPI; moreover, immunofluorescence was detected through high-content analysis (PerkinElmer Operetta, Waltham MA, USA).

Quantitative Real-Time PCR
Total RNA was extracted from the hippocampus or cultured HAPI cells and isolated using TRIzol. We used a Transcriptor First Strand cDNA Synthesis Kit for cDNA synthesis. Real-time PCR was performed using the StepOne TM Real-Time PCR System (StepOne, Foster City, USA) with NoVoStart SYBR qPCR SuperMix plus. Table 1 presents the used primer sequences. The cycling conditions as follows: 95 °C for 5 min and 40 cycles of 95 °C for 10 s, followed by 60 °C for 30 s. Gene expression data were normalized using β-actin; moreover, relative gene expression levels were calculated using the 2 -ΔΔCT method.

Statistical Analysis
All statistical analyses were performed using SPSS 16.0 software (version 16.0, SPSS, Chicago, IL, USA). Results were presented as mean ± standard error of the mean. Comparisons were performed using one-way analysis of variance, followed by a least significant difference test. Statistical significance was set as P ˂ 0.05.

Results
3.1 The DD model was successfully established using the combination of HFD, STZ, and UCMS, with ZGJTJY effectively improving the plasma glucose disorder and depression symptoms in DD model rats.
Compared with the control group, the vehicle group showed significantly higher plasma glucose levels (P < 0.05). In the Morris water maze test, compared with the control group, the vehicle group showed significantly longer ELTs on days 2, 3, and 4, as well as significantly shorter SETs on day 5 (P ˂ 0.05, P ˂ 0.01). In the open field test and sucrose preference test, compared with the control group, the vehicle group showed lower LMA scores and sucrose preference (Tab.2) Compared with the vehicle group, both DMGB/F and ZGJTJY reduced blood glucose, shortened EL times, increased total LMA scores, and increased sucrose intake. (Tab.2) Tab. 2 Capability of learning and memory, locomotor activity, sucrose preference, and plasma glucose level in each group * P ˂ 0.05 and ** P ˂ 0.01, significantly different from control. # P ˂ 0.05 and ## P ˂ 0.01, significantly different from the vehicle group, n = 6.

3.2
The combination of HFD, STZ, and UCMS induced significant changes in microglial morphology and quantity in the hippocampal CA1 and DG areas. Unlike DMGB/F, which only acted on the CA1 area, ZGJTJY reversed these abnormalities in the CA1 and DG areas.
In the CA1 area of the vehicle group, there was cell body expansion, shortening and thickening of synapses, and a decreased microglial number. Both ZGJTJY and DMGB/F reversed these changes by ameliorating the cell body swelling, as well as promoting the increase, lengthening, and thinning of synaptic branches. Compared with DMGB/F, ZGJTJY appeared to have a more obvious effect on promoting cell body retraction and microglial proliferation. (Fig. 2). The vehicle group showed similar microglial changes in the DG area, which were alleviated by ZGJTJY, but not DMGB/F, administration (Fig. 2).

High glucose and corticosterone induced microglial activation similar to the DD state, which was attenuated by ZGJTJY-containing serum in vitro.
Further, high glucose combined with corticosterone induced microglial cell expansion, increased branching thickness, and decreased cell number (Fig. 6).

Discussion
This study confirmed that the combination of HFD, STZ, and CUMS could induce M1 microglial polarization and increase hippocampal pro-inflammatory cytokines. Moreover, we found that the combination of high glucose and corticosterone induced microglial polarization similar to the DD state in vitro. Additionally, we found that ZGJTJY administration reversed these abnormalities by inhibiting and promoting M1 and M2 polarization, respectively. Compared with DMGB/F, ZGJTJY had no significant advantage in downregulating M1 polarization; however, it allowed better promotion of M2 polarization. These findings may help provide insight for research on the mechanism underlying TCM treatment of neuropsychiatric diseases.
We have previously established the rat model of DD (Yang, 2013;Wang et al., 2014;2015). The combination of HFD, STZ, and CUMS could induce high blood glucose levels, increase insulin sensitivity, and depressive-like behaviors, which mimicked the DD clinical features. The model establishment methods were reliable and stable. da Silva  established a DD model by injecting STZ into overnight fasted rats. Compared with this method, the chronic mild stress used in our research can better simulate the daily-life living pressure.
In vitro, LPS and IL-4 are commonly used to induce M1 and M2 microglial polarization, which is related to depression, diabetes mellitus (Guo et al., 2019;, and other neuroinflammation-related diseases (Jung et al., 2016;. Our previous study  and other studies (Ding et al., 2018;Lebedeva, et al., 2017;Jimeno et al., 2018) have shown that high corticosterone levels induced by a handling and restraint stressor could increase blood glucose levels, which causes depressive behaviors. Moreover, hippocampal corticosterone accumulation contributes to hippocampal damage in DD. Hsieh et al. (2019) reported that acute glucose fluctuation induces stress, which alters microglial polarization. Furthermore, we used a combination of high glucose and corticosterone to induce primary hippocampal neuron damage, similar to the DD state (Zhang et al., 2014). We recently established a hippocampal NVU system comprised of a triple-cell co-culture system (brain microvascular endothelial cells, astrocytes, and neurons) in vitro using a combination of high glucose and corticosterone . In this study, M1 microglial polarization was induced by high glucose and corticosterone levels. This yielded a cell model of microglial polarization that could better reflect the physiological characteristics of DD; moreover, it confirmed that M1 microglial polarization caused by high corticosterone levels is closely associated with hippocampal damage.
Previous studies have reported that the monoaminergic system, hypothalamic-pituitaryadrenal axis, neuro-inflammation, and hippocampal damage are involved in DD pathogenesis da Silva Dias et al., 2016;Wang et al., 2015). However, the biological mechanism underlying this relationship remains unclear. Several studies have reported that M1 microglial polarization can directly inhibit hippocampal nerve regeneration (Michelle, et al., 2003), weaken the hippocampal neuroprotective effect (Patterson, 2015), and induce neurotoxicity (Ekdahl, et al., 2009) through inflammatory cytokine release. Meanwhile, microglia can be repeatedly activated, which causes multiple neuronal trauma by acting with astrocytes and microvascular cells (Steelman et al., 2014;Takaki et al., 2012;da Fonseca et al., 2014). Further, it can reduce 5-HT synthesis by activating the indoleamine 2,3-dioxygenase enzyme, which is the first rate-limiting enzyme in 5-HT precursor tryptophan metabolism (Xie et al., 2014); moreover, it promotes Glu release from astrocytes, which results in excitatory poisoning of the central nervous system by binding to NMDA receptors (Steiner et al., 2011). Moreover, the released neurotoxic media can act as microglia, deepen microglial polarization, aggravate neuronal damage, and form a vicious circle (Barger et al., 2007;). Therefore, we speculated that under the continuous hyperglycemic stress, continuous activation of hippocampal M1 microglia can cause severe hippocampal damage. This suggests that M1 polarization arising from hippocampal corticosterone accumulation can underlie DD pathogenesis.
ZGJTJY has been shown to reduce hippocampal corticosterone expression and increase glucocorticoid receptor (GR) expression in DD rats . GR has been shown to have a negative regulatory effect on M1 polarization (Réus et al., 2015). It is involved in microglial differentiation, proliferation, and motility; moreover, it is related to TLR4 (Carrillo-de Sauvage et al., 2013). TLR4 signaling pathway activation is involved in promoting M2 microglia polarization Tian et al., 2019). Therefore, we suggest that the interaction of ZGJTJY with microglial polarization is dependent on corticosterone/GR/TLR4 signaling. M1/M2 microglial polarization is crucially involved in the balance between inflammation promotion and suppression (Tao et al., 2016). Transition to the M1 and M2 phenotypes causes neurotoxicity and neuroprotective effects, respectively (Cherry et al., 2014). The ratio between M1 and M2 polarization reflects the transition from M1 to M2 activation. In this study, ZGJTJY administration reduced the ratio of MCP-1/Mrc1 and IL-1β/IL-4, which induced M1-to-M2 transition.
This study has several limitations. First, we only assessed the hippocampal CA1 and DG areas, which cannot represent the global hippocampal microglia polarization. However, we selected these two areas because a previous study reported a predominant increase in IL-1β immunoreactivity in the hippocampal CA1 and DG areas (Hwang et al., 2014). Consistently, neuron proliferating disorder has been observed in the DG and CA1 areas of type-2 diabetes (Hwang et al., 2008). These findings implicate the hippocampal CA1 and DG areas in the interaction between diabetes and depression. Second, we only detected hippocampal microglial polarization and did not examine nerve apoptosis and regeneration in the CA1 and DG areas. Third, we did not screen the active ingredients of the ZGJTJY formulation that acted on microglial polarization.

Conclusion
Our findings suggest that DD is associated with an enhanced pro-inflammatory M1/M2 microglial ratio both in vivo and in vitro. ZGJTJY administration inhibited microglial polarization to a pro-inflammatory state and promoted an anti-inflammatory state by decreasing the M1/M2 microglial ratio. Future interventions on neuroinflammation-related neuropsychiatric diseases should target both prevention and promotion of M1 and M2 microglial polarization, respectively.

Ethics approval and consent to participate
Not applicable