Liraglutide attenuate central nervous inflammation and demyelination through AMPK and pyroptosis‐related NLRP3 pathway

Abstract Aims Multiple sclerosis (MS) still maintains increasing prevalence and poor prognosis, while glucagon‐like peptide‐1 receptor (GLP‐1R) agonists show excellent neuroprotective capacities recently. Thus, we aim to evaluate whether the GLP‐1R agonist liraglutide (Lira) could ameliorate central nervous system demyelination and inflammation. Methods The therapeutic effect of Lira was tested on experimental autoimmune encephalitis (EAE) in vivo and a microglia cell line BV2 in vitro. Results Lira administration could ameliorate the disease score of EAE mice, delay the disease onset, ameliorate pathological demyelination and inflammation score in lumbar spinal cord, reduce pathogenic T helper cell transcription in spleen, restore phosphorylated adenosine monophosphate‐activated protein kinase (pAMPK) level, autophagy level, and inhibit pyroptosis‐related NLR family, pyrin domain‐containing protein 3 (NLRP3) pathway in lumbar spinal cord. Additionally, cell viability test, lactate dehydrogenase release test, and dead/live cell staining test for BV2 cells showed Lira could not salvage BV2 from nigericin‐induced pyroptosis significantly. Conclusion Lira has anti‐inflammation and anti‐demyelination effect on EAE mice, and the protective effect of Lira in the EAE model may be related to regulation of pAMPK pathway, autophagy, and NLRP3 pathway. However, Lira treatment cannot significantly inhibit pyroptosis of BV2 cells in vitro. Our study provides Lira as a potential candidate for Multiple Sclerosis treatment.


| INTRODUC TI ON
As one of the most studied central nervous system demyelinating and autoimmune degeneration diseases, MS still maintains increasing prevalence and poor prognosis. 1  Glucagon-like peptide-1 (GLP-1) is a kind of endogenous incretin, first discovered in the early 1980s, produced from both gut and brain and exerts its role by binding to GLP-1R, a G protein-coupled receptor, then activating its downstream signal transduction cascades. In the digestive system, GLP-1 is secreted from intestinal L cells by the stimulation of food intake and enhances insulin secretion from pancreatic islets, thus exerting its glucose-dependent hypoglycemic effect, while in the nerve system, GLP-1 is mainly secreted by preproglucagon-expressing neurons from the nucleus tractus solitarius and can act as neurotransmitters, then regulate the activity of vagal nerve system and limbic nerve system to affect varieties of biological processes including eating behavior, weight control, emotions, energy metabolism homeostasis, cognitive abilities, and cardiovascular functions. 4,5 It has been reported that GLP-1R exists in neurons and microglia and is widely distributed in the brain and spinal cord, especially expressed in cognitive-related areas such as the hippocampus, giving GLP-1 the potential for its neuroprotective effects. [6][7][8][9] However, endogenous GLP-1 will be quickly inactivated by dipeptidyl peptidase-4 (DDP-4) after release to the extracellular space, 10 so the pharmacological usage of which is limited.
In the 1990s, after the discovery of long half-life DPP-4 resistant human GLP-1 analog exendin-4 from the venom of Heloderma lizard, many long-acting GLP-1R agonists were developed based on exendin-4 structure and approved clinically for its hypoglycemic or bodyweight lowering usage. 11 Among them, Lira is synthesized by Novo Nordisk with Lys34Arg amino acid substitution to enhance the resistance for DDP-4 degradation and fatty acid side chains to reduce renal filtration 12 and was the second-licensed GLP-1R agonist by Food and Drug Administration (FDA) in January 2010 for type 2 diabetes treatment.
Moreover, during 10 years of GLP-1R activating therapies, the neuroprotective therapeutic efficacy of Lira and other GLP-1R agonists has been revealed and yielded widespread attention. Not only clinical trials but also animal experiments indicate GLP-1R agonists could improve neurodegenerative diseases, such as Parkinson's disease 13,14 and Alzheimer's disease, 15,16 while GLP-1R also exerts therapeutic efficacy on brain ischemia, 17,18 traumatic brain injury, 19 and psychiatric disorders [20][21][22] in animal models. Besides, various kinds of DPP-4 inhibitors, which can increase the level of endogenous GLP-1 level, exhibit neuroprotective and cognitive protective effects either. [23][24][25] GLP-1R agonists also play a role in neuroimmune processes.
In terms of MS, GLP-1R agonists and DPP-4 inhibitors have been reported to have therapeutic efficacy on neuroinflammation and demyelination on EAE model, 26,27  When GLP-1R is activated by agonists, the G protein dissociates subsequently, activates adenylate cyclase and then generates cyclic adenosine monophosphate to trigger acute cell response. Meanwhile, activated G protein will stimulate a broad range of downstream molecules such as phosphoinositide 3-kinase (PI3K), protein kinase B (Akt), and mitogen associated protein kinase to cause long-term biological effects, including facilitation of insulin signaling, neurotrophy, anti-oxidative and anti-inflammation roles. 31,32 Recently, it has been demonstrated that Lira exerts therapeutic roles through AMPK, autophagy, and NLRP3 inflammasome in various kinds of animal or cell disease models. 33-37 AMPK is a major sensor of cell metabolic status and is closely related to inflammation regulation. 38 Phosphorylation of AMPK is related to the disease severity and treatment outcome in EAE models. 39,40 Autophagy, which is a lysosome-dependent degradation pathway to clear potential cell toxic molecules and organelles, is in close relationship with MS pathogenesis. 41 Impaired autophagy influx in nerve tissue may worsen EAE, 42 and restoring autophagy levels in nerve tissue may have therapeutic effects on EAE. 43 Similarly, NLRP3 inflammasome is a potent inflammatory signals sensor and involved in the pathogenesis of MS and EAE, 44 which can recruit PYD and CARD domain-containing (ASC), then activates caspase 1, then splices substrates including interleukin-1β (IL-1β), interleukin 18 (IL-18), and Gasdermin D (GSDMD), and even in some situation causes pyroptosis, a newly discovered programmed cell death, which leads to cell perforation and a large amount of proinflammatory cytokines release, thus may exacerbate MS/EAE. 45 In our research, to the best of our knowledge, we tested the antiinflammation and anti-demyelination effect of Lira on mice EAE model for the first time; then, we tested whether Lira intervention could salvage the microglia from pyroptosis in vitro, therefore to expand the data for pharmacological effects of Lira on demyelinating disease.

| MATERIAL S AND ME THODS
Materials and methods can be found in Appendix S1, Tables S1 and S2.

| RE SULTS
3.1 | Lira ameliorated the disease score and delayed the disease onset of EAE mice After 6 batches of experiments in search for the optimal and safe dosage of Lira for EAE mice, mice equivalent dosages of Lira for its clinically used human dosage (HD) exerting hyperglycemia effect were found to cause large numbers of unexpected death in EAE mice but not in healthy control (Ctrl) mice (Appendix S2, Figures S1-S2, Tables S3-S4). The dosage of 10 μg/kg.d (mice equivalent dosage for 1/10 of the minimum clinically used HD for its hypoglycemic activity) one time in 2 days (qod) starting from 8 days postimmunisation (dpi) was finally decided for observing its anti-inflammation and anti-demyelination therapeutic effect, which showed slightly improved disease score and accumulated disease score compared with EAE group after disease onset ( Figure 1A,B).

| EAE versus Lira in blood glucose and insulin levels
Random blood glucose was monitored every 3-4 days after immunization with the dosage of Lira as 62.5 μg/kg d (half of the minimum HD for its hypoglycemic activity) once a day (qd) after 4 dpi ( Figure 1D) in batches 3 and 4. At each time point, there was no significant difference between Ctrl and Ctrl + Lira group or EAE and EAE + Lira group, suggesting that this dosage of Lira did not affect random blood glucose significantly. However, the EAE group and EAE + Lira group both had an approximate 30-50% reduction in random blood glucose level compared with the Ctrl group and F I G U R E 1 Liraglutide (Lira) administration ameliorated the disease score of experimental autoimmune encephalitis (EAE) mice and delayed the disease onset. EAE induction could reduce random blood glucose level, while Lira administration did not have an extra hypoglycemic activity. Certain dosages of Lira had insulin lowering effect compared with EAE group. Lira was administered subcutaneously (i.c.) daily (qd) or one time in 2 days (qod) after certain time points days postimmunisation (dpi). Data were shown in the form of mean ± SEM. * represents p < 0.05, ** represents p < 0.01, *** represents p < 0.001. (A-C) shows disease score, accumulated disease score and disease onset of EAE + Lira group and EAE group, n = 13-28 per group. The rating scale is 1 (paralyzed tail); 2 (posterior limb paresis); 3 (posterior limb paraplegia); 4 (posterior limb paraplegia with forelimb weakness or paralysis); 5 (moribund, or dead). Mice judged between grades received intermediate scores (±0.5). (D) shows random blood glucose level among different groups in serial time points, n = 5-15 per group. (E) shows random blood insulin level when mice were sacrificed, n = 5-10 per group. In addition, two unexpected died mice after Lira administration were tested Ctrl + Lira group (p < 0.05), suggesting that EAE induction itself could cause a decrease in random blood glucose level.
Moreover, the blood serum insulin level was monitored by enzyme-linked immunosorbent assay (ELISA) when the animal was sacrificed or in its moribund state ( Figure 1E) in batches 5 and 6, and an unsignificant increasing trend of insulin was observed in EAE group compared with Ctrl group (49.61±21.48 ng/ml versus 37.39 ± 22.06 ng/ml, p = 0.095). Meanwhile, the EAE + Lira group with the Lira dosage of 20-60 μg/kg d qod after 8 dpi had significantly decreased insulin level compared with the EAE group (29.11 ± 7.40 ng/ml, p = 0.017). Interestingly, blood serum of 2 unexpected dead mice after Lira administration was luckily obtained in their moribund state, and the values were high (82.12 ng/ml and 51.70 ng/ml, respectively).

| Lira ameliorated demyelination as well as inflammation and regulated Th cell transcription
The protective therapeutic effect of Lira was semi-quantitatively confirmed by Luxol fast blue (LFB) staining ( and Interleukin-6 (IL-6). In the EAE situation, those cytokines mRNA expressions were all upregulated 5-23-fold compared with the Ctrl group, while Lira intervention downregulated them (among them, TNFα did not reach statistical significance).
Th cells in the spleen are highly related to EAE pathogenesis, and Th cell-related transcription was also found regulated by Lira intervention (Figure 3D-F). Th17 cell-related mRNA, RAR-related orphan receptor gamma (RORγt), and Th1 cell-related mRNA T-box 21 (Tbet) were upregulated by twofold in the EAE group compared with Ctrl group, which is widely considered detrimental to EAE progress, while Lira reduces these mRNA expression levels. In addition, regulatory T (Treg) (which is considered to retard the autoimmune reactions and alleviate EAE severity) cell-related mRNA transforming growth factorβ (TGFβ) unsiginificantly decreased in EAE situation and was restored after Lira administration.

| The molecular protective mechanism of Lira in the EAE model
Through literature review, some pathways were focused on to investigate whether they were enrolled in the protective effect of Lira on EAE. First, as it is reported that EAE could downregulate GLP-1R expression, the GLP-1R mRNA expression level was tested and found downregulated to 28.81 ± 5.41% of Ctrl group after EAE induction.
However, Lira intervention could unsignificantly elevate the GLP-1R expression to 78.71 ± 60.50% of the Ctrl group ( Figure 3G). Next, we (P and Q) shows the semi-quantitative demyelination score (based on LFB staining) and inflammation score (based on HE staining). N = 4 per group. * represents p < 0.05, ** represents p < 0.01. Data were shown in the form of mean ± SEM. The black scatters represent average scores of samples, and the red scatters represent score given by the first observer. (R) Schematic figure to illustrate the process of semi-quantitative evaluation for LFB and HE staining slices. Six of 10 fields (squares in the picture) were selected randomly by two observers blinded to the grouping information, and the average score was used to represent the extent of demyelination and inflammation for one slice. The inflammation score scale is 0 (normal); 1 (lymphocyte infiltration around meninges and blood vessels); 2 (1-10 lymphocytes in a field); 3 (11-100 lymphocytes in a field); 4 (over 100 lymphocytes in a field). And the demyelination score scale is 0 (normal); 1(small regions of sporadic myelin sheath loss); 2 (a few areas of myelin sheath loss); 3 (massive myelin sheath loss)

| Lira treatment in pyroptosis of BV2 cells in vitro
As the animal experiment earlier indicated, there was an inhibition effect of Lira on the NLRP3 pathway, and the previous study shows  Figure 6D) showed a flattened shape with pores on the cell body, which was characterized for pyroptosis, while LPS-primed cells ( Figure 6E,F) showed extended projections and coarse surface of soma compared with unprimed negative cell ( Figure 6G,H), indicating its activated state. However, a large proportion of ATP-challenged cells showed apoptosis-like characters such as cell shrinkage and membrane blebbing (data not shown); thus, nigericin was chosen for subsequent experiments.
Next, gradient concentrations of Lira ranging from 50 nM to 2000 nM were used to incubate the BV2 cell for 24 h as a preventive treatment, while in the LPS group and pyroptosis (Pyro) group, F I G U R E 4 Autophagy was inhibited after experimental autoimmune encephalitis (EAE) induction, and Liraglutide intervention could partly restore the autophagy level (n ≥ 4 per group). * represents p < 0.05, ** represents p < 0.01, *** represents p < 0.001 and **** represents p < 0.0001. Data were shown in the form of To sum up, Lira could not significantly stop the pyroptosis of BV2 cells in vitro but had a trend to ameliorate it.

| DISCUSS ION
In this study, Lira, to our knowledge, for the first time to be demonstrated having anti-inflammation and anti-demyelination roles in the mice EAE model. Similar to the effect of Lira on the rat EAE model, 30 Lira significantly delays the disease onset and alleviates the disease severity. Moreover, as a supplement to a previous study on rats, the neuroprotective effect was also semi-quantitatively confirmed pathologically, with less inflammatory cell infiltration and white matter demyelination in EAE model.

F I G U R E 5
Liraglutide administration inhibited pyroptosis-related NLR Family, pyrin domain-containing protein 3 (NLRP3) signaling pathway (n ≥ 3 per group). * represents p < 0.05, ** represents p < 0.01, *** represents p < 0.001. Data were shown in the form of mean ± SEM. (A-D) shows result of key pyroptosis indicators mRNA expression level. (E-H) shows result of key pyroptosis indicators protein expression level F I G U R E 6 An in vitro microglia pyroptosis model showed that Liraglutide (Lira) could not significantly salvage the cell from pyroptosis. * represents p < 0.05, ** represents p < 0.01, *** represents p < 0.001 and **** represents p < 0.0001. Data were shown in the form of mean ± SEM. (A) shows Cell Counting Kit 8 (CCK8) results after 4 h of gradient concentration lipopolysaccharide (LPS) stimulation (OD450 values were compared with Ctrl group), 1.0 μg/ml LPS manifested most profound stimulation effect and was selected for subsequent experiments. Thus, in (B and C), gradient nigericin and adenosine triphosphate (ATP) were used to incubate with LPS-primed BV2 cells for 4 h and 2 h, respectively, to induce pyroptosis, and CCK8 test was conducted (OD450 values were compared with LPS-primed group).
(D-H) shows representative scanning electron microscope pictures of differentially treated cells. D shows a pyroptosis BV2 cell induced by nigericin, which was flattened and pored, E and F shows LPS-primed cell, which had long projections and rugged activated soma, G and H shows a normal BV2 cell which is not primed by LPS and exhibits shorter projections and soma with smoother surface (scale bar = 5 μm). neuroinflammation, improve memory function, and reduce amyloidβ deposition and oxidative injury in Alzheimer's disease animal model, [53][54][55]  Interestingly, the dosage of Lira we tested on EAE is only 1/10 of the minimum HD for its hypoglycemic effect. According to our calculation, Lira was tested safe for its threefold minimum hypoglycemic HD on rat EAE model, 30 and another GLP-1R agonist dulaglutide was tested safe for its minimum hypoglycemic HD on mice EAE model. 26 On the one hand, our results implicated that Lira may exert its anti-inflammatory and anti-demyelinating effects independent of its hypoglycemic effect; on the other hand, possible reasons for unexpected death for EAE mice after Lira administration remain to be investigated. A possible conjecture we made is the EAE mice would lose weight, and Lira administration might cause extra body weight loss and probably some digestive system symptoms such as nausea and anorexia, then cause general weakness and possible hypoglycemia, and finally cause unexpected death. In this regard, we monitored the random blood glucose level and insulin level accordingly.
Intriguingly, EAE itself caused significant blood glucose reduction, but Lira did not have an extra hypoglycemic effect. Moreover, there was an unsignificant upregulating trend of blood serum insulin level after EAE induction. Through literature review, we do not acquire much information about the relationship between EAE and insulin excretion, which should be investigated more in the future.

| LIMITATI ON S AND FUTURE PER S PEC TIVE S
The reason for unexpected death caused by Lira administration in the mice EAE model could be investigated, and serial dosage of Lira as either prophylactic or therapeutic interventions could be tried on the EAE model to find the optimal dosage. Moreover, the mechanism of the neuroprotective effect of Lira should be investigated in-depth in the future.

| CON CLUS ION
Liraglutide administration could ameliorate the disease score of EAE mice and delay the disease onset, ameliorate demyelination and inflammation in nerve tissue, and regulate Th cell transcription in the spleen of EAE mice. The protective effect of liraglutide in the EAE model may be related to regulation of AMPK pathway and autophagy, as well as inhibition of pyroptosis-related NLRP3 pathway, but liraglutide treatment could not significantly inhibit pyroptosis of BV2 cells in vitro. Our study provides liraglutide as a potential therapeutic candidate for MS treatment.

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

AUTH O R CO NTR I B UTI O N S
All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

CO N S ENT TO PA RTI CI PATE
This article does not contain any studies with human participants performed by any of the authors.

DATA AVA I L A B I L I T Y S TAT E M E N T
The supplementary material for this article can be found online. All processed data used in this study can be obtained from the corresponding author on reasonable request.