Ischemic stroke comprises one of the most serious macrovascular complications and one of the major causes of death in type 2 diabetic patients. The risk of stroke in type 2 diabetes mellitus (DM) is twice as large as in the general population and increases with DM duration [1].

Moreover, almost one in two people with type 2 diabetes mellitus (DM) has chronic brain dyscirculation. This condition can be associated with more or less pronounced cognitive deficits in the absence of focal neurological symptoms and can lead to a dramatic reduction in quality of life and even life expectancy [2]. Chronic brain dyscirculation or chronic brain damage can be characterized by cerebral atrophy, decrease in grey matter volume, dilatation of the ventricles and subarachnoid space, decreased density of the space surrounding the lateral ventricles - leukoariosis, “periventricular space glow” [3]. Importantly, in chronic brain damage small focal lacunar infarctions in the white substantive and subcortical ganglia can be observed, so-called microstrokes or silent strokes. Due to their volume, they do not cause the classic symptoms of unilateral motor and sensory dysfunction, but their pathogenesis is very similar to that of clinically manifest stroke [4].

Nowadays, the concept of the so-called glucose-centric model of type 2 DM treatment has changed to a disease-modifying model, in which the organ-protective properties of glucose-lowering drugs are given priority. Thus, according to existing guidelines [5, 6], glucagon-like peptide-1 receptor agonists (GLP-1RAs) and sodium-glucose co-transporter type 2 inhibitors (SGLT-2i) are the two leading groups of glucose-lowering drugs with the widest spectrum of pleotropic organ-protective effects, including cardio- and nephroprotective properties. Nevertheless, data concerning neurotropic effects of the above-mentioned drug classes remain limited. Thus, according to the meta-analyses summarizing randomized controlled trials (RCTs) results, only two drugs can certainly decrease ischemic stroke risk in type 2 diabetic patients—the long-acting GLP-1RAs dulaglutide and injectable semaglutide [7]. However, accumulating evidence suggests that SGLT-2i inhibitors may favorably influence the course of atrial fibrillation and thus protect against the cardioembolic ischaemic stroke subtype [8, 9], and the SGLT-2i canagliflozin may reduce the risk of hemorrhagic stroke [10]. Data on the impact of GLP-1RAs and SGLT-2i on chronic brain dyscirculation are even more limited.

A number of experimental studies have demonstrated the possibility of GLP-1RA liraglutide to decrease the brain damage volume and neurological deficit in animals both with DM and without it [11, 12, 13, 14]. At the same time neuroprotective effect of long-acting GLP-1RA dulaglutide in diabetic animals undergoing acute ischemic-reperfusion injury is not fully elucidated, as well as experimental data concerning potential influence of SGLT-2i on stroke are very limited. To our knowledge, no experimental studies have directly compared the neurotropic potential of GLP-1RA and SGLT-2i in diabetic animals with stroke.

Furthermore, despite the growing interest in the potential of novel glucose-lowering drugs to protect the central nervous system (CNS), the potential mechanisms of this effect remain not fully elucidated, both in the conditions of acute and chronic brain damage. One of the pathogenetic pathways of neuroprotection, in addition to direct effects on neurons, may be the influence on microglial activation and microglial damage [15].

Therefore, the aim of our study was to compare the neuroprotective properties of GLP-1RAs with different duration of action, liraglutide and dulaglutide, and SGLT-2i canagliflozin in acute and chronic brain injury in diabetic Wistar rats and to evaluate potential mechanisms, including microglial involvement. We chose the low-selective SGLT-2i canagliflozin for our study because it affects not only the second type of sodium-glucose cotransporter, but also the first type expressed in the CNS [16], and therefore this drug may have a potential neuroprotective property.

We succeeded in modeling DM in 57 out of 60 rats, while DM criteria were not reached in the 3 animals: their BGL during OGTT reached the maximum values of 10.1, 10.7 and 10.9 mmol/L respectively, therefore they were excluded from the further experiment. For this reason, we had to include 3 more animals to maintain the stated sample size. None of the rats developed clinical manifestations typical for absolute insulin insufficiency throughout the experiment.

3.1 Blood Glucose Level Dynamics

The BGL dynamics are shown in Fig. 2. For ease of analysis, we plotted blood glucose levels for every other week. All the rats in the diabetic groups had similarly elevated glycemic levels in the beginning. The start of therapy with canagliflozin, liraglutide and dulaglutide was characterized by BGL decrease and the glycemic profile remained normal in all the treatment groups throughout the rest of the experiment, with no differences between groups. In addition, BGL in “DM+CANA”, “DM+LIRA” and “DM+DULA” groups did not differ from that in “Control” group.

Fig. 2.

Blood glucose level dynamics. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”. Glycemic level in all diabetic groups was higher than in “Control” group. All the study drugs (canagliflozin, liraglutide and dulaglutide) caused glycemic profile normalization; BGL in “DM+Canagliflozin”, “DM+Liraglutide” and “DM+Dulaglutide” groups did not differ between each other and with “Control” group. Strept., streptozotocin; DM, diabetes mellitus; BGL, blood glucose level.

3.2 Canagliflozin, Liraglutide and Dulaglutide Influence on Functional, Morphological and Biochemical Parameters in Acute Brain Injury

3.2.1 Neurological Deficit and Brain Damage Volume

As mentioned above, neurological status 48 hours after acute cerebral ischemia was evaluated using the J.H. Garcia scale, that includes motor and sensory parameters and were the healthy animal receives 18 points and the most severe neurological deficit is described as 3 points.

Animals in “DM” group had more prominent neurological deficit than in “Control” (8.0 [4.0; 9.0] and 12.0 [10.0; 15.0] points, respectively, p = 0.02). All study treatments resulted in a significant improvement in neurological status. Thus, the neurological deficit in “DM+CANA”, “DM+LIRA” and “DM+DULA” was smaller than in “DM” (p = 0.000, p = 0.005 and p = 0.000, respectively) and even in “Control” group (except for “DM+DULA” where neurological deficit tended to be smaller than in “Control”, p = 0.077). Importantly, there were no between-group differences between treatment groups (Fig. 3).

Fig. 3.

Neurological deficit by Julio H. Garcia scale 48 hours after acute brain ischemia. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”. Neurological deficit was harder in “DM” in comparison with “Control”. Neurological status was significantly better in “DM+Canagliflozin”, “DM+Liraglutide” and “DM+Dulaglutide” comparing with “DM” and with “Control” (except for “DM+DULA”). No differences were observed between the study drugs. DM, diabetes mellitus.

Brain damage volume was similarly big in “Control” and “DM” groups (17.53 [14.23; 26.58] % and 15.87 [13.40; 22.68] % of total brain volume, respectively, p = 0.631). All the study drugs diminished brain damage volume comparing with “DM” and “Control”. Importantly, brain necrosis volume in “DM+LIRA” was smaller than in “DM+CANA” (2.9 [1.83; 4.71] % and 6.17 [3.88; 8.88] %, respectively, p = 0.015) and did not significantly differ from that in “DM+DULA” (2.9 [1.83; 4.71] % and 4.57 [3.27; 7.90] %, respectively, p = 0.165) (Fig. 4A,B).

Fig. 4.

Brain damage caused by ischemia-reperfusion injury. (A) Brain damage volume measurement results presented as dot plots with median values. (B) Representative images of brain slices stained with triphenyltetrazolium chloride. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”, §p 0.05 comparing with “DM+Canagliflozin”. Brain damage volume in “DM” was as large as in “Control”. All the study drugs reduced the damage volume comparing with “DM” and “Control”, while necrosis volume in “DM+Liraglutide” was smaller than in “DM+CANA” and not significantly different from that in “DM+Dulaglutide”. DM, diabetes mellitus.

3.2.2 Biochemical Neuroglial Damage Markers

NLC level was similarly high in “Control” and “DM” groups (p = 0.963). Both the SGLT-2i canagliflozin and the long-acting GLP-1RA dulaglutide were associated with a decrease in this parameter (p = 0.029 comparing “DM+CANA” and “DM”, p = 0.026 comparing “DM+DULA” and “Control”). However, liraglutide had no effect on NLC concentration (Fig. 5A).

Fig. 5.

Neurofilament light chains and S100BB protein concentrations 48 hours after ischemia-reperfusion injury. (A) Neurofilament light chains (ng/mL) level, (B) S100BB protein (pg/mL) level. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”, ¶ p 0.05 between “DM+Liraglutide” and “DM+Dulaglutide”. Neurofilament light chains level did not differ in “DM” and “Control”. Dulaglutide and canagliflozin administration caused prominent NLC decrease while liraglutide did not influence this parameter. Protein S100BB level was similar in “DM” and “Control” groups. Both liraglutide and dulaglutide administration was associated with S100BB decrease whereas liraglutide caused the largest S100BB decrease, on the other hand canagliflozin did not significantly influence it. NLC, neurofilament light chains; DM, diabetes mellitus. The green * is the statistic outlier.

Similarly, S100BB levels were high in both “Control” and “DM” with no differences between the groups (p = 0.113). Both GLP-1RAs caused a decrease in S100BB compared to “Control”, although the effect of liraglutide was significantly more pronounced than that of dulaglutide (p = 0.035). On the other hand, canagliflozin did not influence S100BB concentration (Fig. 5B).

3.3 Canagliflozin, Liraglutide and Dulaglutide Influence on Neurons Structure and Microglia Activation in Chronic Brain Injury, Immunohistochemical Assay Results

3.3.1 Nissl Staining

A study of rat brain paraffin sections stained with toluidine blue using the Nissl method showed that in “Control” group neurons in the CA1 hippocampal zone were arranged densely in several rows, forming one layer of cells called the stratum pyramidale. These cells had a light, large nucleus and 1–2 metachromatically colored nucleoli. A uniformly distributed chromatophilic substance was clearly visible in the cytoplasm of the neuronal cell bodies. The initial segments of neuronal dendrites were often visible, extending into the stratum radiatum (Fig. 6A).

Fig. 6.

Neurons of Cornu Ammonis (CA)1 zone of rat hippocampus. Nissle staining. “Control” group (A); diabetes mellitus, “DM” (B,C); diabetes mellitus + canagliflozin, “DM+CANA” (D,D1), diabetes mellitus + liraglutide, “DM+LIRA” (E,E1); diabetes mellitus + dulaglutide, “DM+DULA” (F) (D,E,F — lower magnification, D1,E1,F1 — higher magnification). Arrows point at expanded cells (B,D1), shriveled neurons (C,F1), hyperchromatosis (D,F,E1), neurons outside the pyramidal layer (E). Arrowheads point at hypochromatosis (C,D1) and expanded cell (E1). Asterisk indicates shadow cells (C) and empty area between neurons (D1). Scale bar 20 µm (A–F) and 5 µm (D1–F1). In the “Control” group neurons in the CA1 hippocampus zone were arranged densely in several rows, forming one layer of cells called the stratum pyramidale. These cells had a light, large nucleus and 1–2 metachromatically colored nucleoli. In the “DM” group there were hypo- and hyperchromatic, swollen, wrinkled neurons; in the pyramidal layer of CA1 zone neurons number was decreased, neurons located in several rows alternated with neurons located in 1–2 rows, areas were observed where neurons were absent and the pyramidal layer was interrupted. In the “DM+CANA” group, there were hypo- and hyperchromatic nuclei throughout the entire studied area, wrinkled neurons were less common than in “DM” group. In the “DM+LIRA” most neurons had normal morphology but there were cells with signs of hyperchromatosis, often with polygonal shape. In “DM+DULA” group most cells did not have pathological changes. There were single wrinkled neurons and cells with signs of hyperchromatosis or swelling. SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatum.

In the “DM” group hypo- and hyperchromatic neurons, as well as swollen (or edematous) neurons, were identified (Fig. 6B,C). Throughout the CA1 zone a large number of wrinkled neurons were observed – cells with pronounced signs of hyperchromatosis, uneven borders and a decrease in the cytoplasm and cell nuclear volume (Fig. 6C). There were also cells with cleared cytoplasm and the absence of chromatophilic substance (Fig. 6C, arrowhead). Rarely, shadow cells were found - cells in which toluidine blue staining was almost completely absent both in the cytoplasm and in the nucleus (Fig. 6C, asterisk). In the “DM” group, a decrease in the number of neurons was often observed in the pyramidal layer of the CA1 zone - neurons located in several rows could alternate with neurons located in 1–2 rows, while areas were observed where neurons were absent and the pyramidal layer was interrupted.

In the “DM+CANA” group, similar to the “DM” group, neurons were characterized by the presence of hypo- and hyperchromatic nuclei throughout the entire examined area (Fig. 6D). In the “DM+CANA” group wrinkled neurons were less common than in the “DM” group. Swollen cells and shadow cells were also found (Fig. 6D1).

In the “DM+LIRA” group most neurons had normal morphology. However, there were cells with signs of hyperchromatosis, often these cells had a polygonal shape. Individual cells showed signs of swelling. Rarely individual neurons or groups of neurons were located outside the pyramidal layer (Fig. 6E,E1).

In both the “DM+DULA” and the “DM+LIRA” groups, the vast majority of cells did not show pathological changes. There were individual wrinkled neurons and cells with signs of hyperchromatosis or swelling (Fig. 6F,F1).

Statistical analysis showed that the number of neurons with normal unaltered structure was lowest in the “DM” group. Administration of all the study drugs was associated with an increased number of normal cells while higher level was seen in the “DM+LIRA” and “DM+DULA” groups with no difference between them (Fig. 7A). The number of pathological cells was mostly similar in all the groups (Fig. 7B). We also calculated the ratio of normal to pathological cells and found that animals in the “DM” group had the lowest ratio, while all study drugs increased this ratio and the “DM+DULA” group was even characterized by a normal/pathological cell ratio similar to that in the “Control” group (Fig. 7C).

Fig. 7.

Number of normal, pathological cells and normal/pathological cells ratio of CA1 zone of rat hippocampus, Nissl staining. (A) number of normal cells/mm${}^2$, (B) number of pathological cells/mm${}^2$, (C) normal cells/pathological cells ratio. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”, § p 0.05 comparing with “DM+CANA”, ¶p 0.05 between “DM+LIRA” and “DM+DULA”. Number of pathological cells was mostly similar in all groups. “DM” was characterized by the lowest value of normal cells and the lowest normal/pathological cells ratio. Canagliflozin, liraglutide and dulaglutide administration led let to increase in both these parameters while GLP-1RA liraglutide and dulaglutide had more prominent effect.

3.3.2 Microglial Activation, Immunohistochemical Reaction to Iba-1

An immunohistochemical study of microglial cells in the CA1 hippocampus zone in the stratum radiatum layer revealed that in the “Control” group microglia were represented by the ramified type (Fig. 8, Control). These cells had a small cell body (and therefore cell bodies were rarely visible) and long, thin branching processes. In “DM” group microglia were characterized by a large cell body and often short and thick processes suggesting that microglia here were represented by activated forms (Fig. 8, DM). The number of microgliocytes also increased dramatically. In the “DM+CANA” group microglia were characterized by transitional forms. Cell bodies were small, but larger in size compared to “Control” group. The processes were long and strongly branched, slightly thicker compared to the “Control” (Fig. 8, DM+CANA). The number of microgliocytes was comparable to “DM” group. In the “DM+LIRA” and “DM+DULA” groups microglia were of the ramified type (Fig. 8, DM+LIRA, DM+DULA). The number of cells in the “DM+DULA” group was comparable to the “Control” group.

Fig. 8.

Microglia of stratum radiatum of rat hippocampal CA1 zone. Immunohistochemical reaction to Iba-1. Effects of DM (diabetes mellitus), canagliflozin (DM+CANA), liraglutide (DM+LIRA) and dulaglutide (DM+DULA) on microglia activation. Arrows point at Iba-1+-cells. Scale bar 20 µm. In the “Control” group microglia were represented by the ramified type characterized by a small cell body and long, thin branching processes. In the “DM” group microglia were characterized by a bigger number of cells, larger cell body, short and thick processes typical for activated forms. In the “DM+CANA” group microglia were characterized by transitional forms. In the “DM+LIRA” and “DM+DULA” groups microglia were of the ramified type.

Statistical analysis confirmed the data described above: DM was characterized by an increased number of immunopositive cells. Canagliflozin treatment did not decrease the number of immunopositive cells compared to the “DM” group. On the other hand, both GLP-1RAs reduced the number of activated microgliocytes compared to the “DM” group, with dulaglutide being more effective than liraglutide (Fig. 9).

Fig. 9.

Number of immunopositive cells (activated microgliocytes) in stratum radiatum of rat hippocampal CA1 zone, immunohistochemical reaction to Iba-1. *p 0.05 comparing with “Control”, # p 0.05 comparing with “DM”, §p 0.05 comparing with “DM+Canagliflozin”, ¶p 0.05 between “DM+Liraglutide” and “DM+Dulaglutide”. DM was characterized by increased number of immunopositive cells — activated microgliocytes. Canagliflozin treatment did not decrease immunopositive cells number in comparison with “DM” group, while both GLP-1RAs decreased activated microgliocytes number comparing with the “DM” group, with dulaglutide being more effective than liraglutide. DM, diabetes mellitus.

Our study was aimed to investigate the influence of diabetes mellitus on the severity of brain damage in conditions of acute and chronic brain ischemia as well as to hold the direct comparison of GLP-1RAs and SGLT-2i neuroprotective properties under these conditions.

In clinical practice, DM is known to be a major cause of acute cardiovascular accidents such as myocardial infarction and stroke, leading to significant disability or even death much more frequently than in the general population [24]. Surprisingly, the experimental data are not so unambiguous. Thus, it has been shown that the infarction area in diabetic animals was even smaller than in nondiabetic animals [25]. The explanation for such a phenomenon could be some kind of metabolic preconditioning that takes place in hyperglycemic conditions. Conversely, there are a number of experimental studies showing an outstanding negative impact of hyperglycemia in organ damage volume, including myocardial infarction and stroke, and poor outcome [26, 27, 28]. In our experiment brain damage volume was similar in untreated diabetic and non-diabetic animals, whereas the neurological deficit was more pronounced in diabetic animals, implying a worse recovery after stroke and therefore potentially a worse outcome.

To compare the neuroprotective effects of glucose-lowering drugs, we chose liraglutide and dulaglutide as two GLP-1RAs with different durations of action. According to RCTs, liraglutide has demonstrated the potential to reduce all-cause mortality, cardiovascular mortality and the incidence of myocardial infarction, without a certain impact on the risk of stroke. In contrast, dulaglutide is one of the long-acting GLP-1RAs that can reduce the risk of non-fatal stroke [29]. Interestingly, there is a large body of experimental work focusing on the neuroprotective effect of liraglutide in both diabetic and non-diabetic rats, used before or after stroke [11, 12, 14, 30]. Thus, liraglutide has demonstrated anti-ischemic properties in experimental conditions. On the contrary, we have not found any data on the neuroprotective potential of dulaglutide in stroke conditions in diabetic rats. Part of the rats in our experiment received canagliflozin as a non-selective SGLT-2i, which means that this drug suppresses not only the second type of sodium-glucose cotransporter, which is mainly found in the kidneys, but also the first type, which is widely expressed in the CNS and vascular endothelium [16]. In this context, as mentioned above, canagliflozin could be a perspective drug to realize neuroprotective properties. Surprisingly, although there are some publications on the neuroprotective properties of other SGLT-2i, for example the highly selective empagliflozin [31, 32], we could not find any work on the effect of canagliflozin in transient brain ischemia.

To our knowledge, our study is the first direct comparative investigation of the neuroprotective properties of GLP-1RAs and SGLT-2i in acute brain ischemia in diabetic rats.

Previously, our group performed a similar direct comparative study of SGLT-2i and GLP-1RA neuroprotective effects in stroke, but in non-diabetic conditions [33]. Such a model of non-diabetic rats undergoing stroke helps to phenomenologically confirm the presence of neuroprotective potential of some substances - in this situation, glucose-lowering drugs.

The current study held in diabetic rats helps to mimic the conditions close to the real-life ones. We succeeded in reaching similarly satisfactory glycemic profile in all treatment groups so that glycemic level did not differ significantly between the canagliflozin, liraglutide and dulaglutide treatment groups. We believe this allows us to analyze pleotropic neuroprotective effects of these drugs without strong association with glycemic control.

We found out that canagliflozin, liraglutide and dulaglutide all reduced the brain damage volume and neurological deficit in transient ischemia-reperfusion injury. Although all three drugs had a similar beneficial effect on neurological status, the infarct-limiting property of liraglutide was more pronounced than that of canagliflozin and did not differ from that of dulaglutide, suggesting that GLP-1RAs may be more protective than SGLT-2i in acute ischemic stroke. However, these data need to be investigated further to understand whether this is a drug-effect or a class-effect.

A possible explanation for the more pronounced infarct-limiting property of GLP-1RAs may be related to the broader expression of GLP-1 receptors in the CNS than that of SGLT-2. For example, GLP-1 receptors are found in the frontal cortex, thalamus, hypothalamus, hippocampus, cerebellum and substantia nigra–regions known to be involved in the regulation of energy homeostasis and autonomic functions [34]. Moreover, GLP-1 receptors are also present in the dorsal vagus complex, particularly in the nucleus of the solitary tract, and to a lesser extent in the subfornical organ and area postrema [35]. GLP-1RA are known to realize their protective effect in ischemic conditions through several ways including activation of anti-apoptotic genes and suppression of proapoptotic ones [36]. Anti-ischemic properties of SGLT-2i are much less studied and the specific mechanisms require further investigation. Nevertheless, low-selective SGLT-2i might by more promising regarding ischemic conditions as this is SGLT-1 activation during stroke that realizes proinflammatory action, increases ischemia-reperfusion injury and decreases neuronal survival [16].

Ischemia-reperfusion injury in our study was associated with increase in neurofilament light chains independently on glycemic profile which resulted in similar increase of this parameter in both control group and untreated diabetic group. NLC are the proteins comprising predominantly large-caliber myelinated axons and are the part of a complex called neurofilament [37]. Increased levels of NLC can be observed in biological fluids, including blood, in various conditions associated with nervous system damage, such as neurodegenerative diseases and traumatic brain injury [38]. It has been reported that serum NLC level correlates with the infarct volume in patients with acute ischemic stroke [39] and can be a biochemical predictor of unfavorable outcome and secondary neurodegeneration development after stroke [40]. In our study NLC level was lower in animals treated with canagliflozin and long-acting GLP-1RA dulaglutide. Surprisingly, liraglutide did not have similar action. The positive effect of both SGLT-2i and long-acting GLP-1RAs on NLC levels suggests that these drugs have a direct effect on neuronal tissue survival.

Another potential mechanism involved in ischemic stroke may be related to microglial damage. Microglia are resident CNS immune cells that act as critical sensors, effectors and regulators of CNS homeostasis under normal and pathological conditions [41]. In acute damage such as ischemic stroke microglia play a crucial role in a number of pathophysiological pathways leading to the activation of complement system and reactive oxygen species release. During the course of stroke, microglia undergo severe morphological changes and participate in neuroinflammation and at the same time in the regeneration process, thus realizing both alternating and beneficial effects [42].

To assess the microglial component of the damage, we measured serum levels of S100BB. S100 are the brain glial proteins produced mainly by astrocytes. Cerebral S100 is a combination of two closely related family proteins: S100A1 (S100$\alpha$) and S100BB (S100$\beta$). Due to their ability to regulate a number of proteins activity, S100A1 and S100BB are involved in the transduction of signals that control energy metabolism enzymes activity in brain cells, calcium homeostasis, cell cycle, cytoskeletal functions, transcription, proliferation and cell differentiation, their motility, secretory processes, structural organization of biomembranes [43]. The biological role of S100BB secreted by astrocytes is thought to be different: at physiological (nanomolar) concentrations the neurotrophic effect predominates, which occurs for example during CNS development or neural regeneration, whereas at high (micromolar) concentrations the possible neurotoxic effects dominate [44]. During brain injury, including ischemic stroke, S100 readily crosses the damaged blood-brain barrier and can be found elevated in the bloodstream [45]. Recent studies have demonstrated the role of S100BB protein in acute ischemic stroke. It has been reported that serum S100BB elevation is positively correlated with the extent of brain damage and worse outcome [46]. In our study, S100BB was elevated after ischemia-reperfusion injury in both untreated groups, control and diabetic, suggesting a microglial damage component in the pathogenesis of acute ischemic stroke. Both GLP-1RAs, liraglutide and dulaglutide, reduced S100BB, with liraglutide having the most pronounced effect. At the same time, canagliflozin did not affect this parameter. We can therefore assume that GLP-1RAs reduce astroglial damage, whereas SGLT-2i do not have such an effect. Taking into account that liraglutide demonstrated an outstanding infarct-limiting effect and caused S100BB but not NLC decrease, we can assume that this drug exerts its neuroprotective effect in acute stroke mainly by affecting microglia rather than neurons themselves. However, this idea needs further confirmation.

We also investigated the CNS damage in the conditions of chronic brain ischemia using DM model itself assuming that long-standing diabetes is characterized by the kind of CNS lesion by the mechanism of “silent” strokes as described earlier. Immunohistochemical assay approved that DM is characterized by pathological changes in the neurons. All the study drugs administration was associated with increased number of cells with unchanged structure, while GLP-1RAs had more pronounced effect, without differences between liraglutide and dulaglutide. This finding suggests direct positive influence of SGLT-2i and especially GLP-1RAs on neurons structure. Besides, we revealed pathological microglial activation in DM animals which was diminished by both GLP-1RAs with dulaglutide being more effective, whereas SGLT-2i canagliflozin did not realize this effect.

Thus, it can be assumed that microglial damage and pathological hyperactivation are involved in the pathogenesis of both acute and chronic brain damage. GLP-1RA can diminish microglial reaction while protective effect of SGLT-2i lies aside from the influence on microglia.

We can conclude that ischemic stroke course might be more severe in the presence of DM which represents in worse neurological outcome. Brain ischemia independently on the presence of DM is accompanied by both neuronal and glial damage. In acute ischemia-reperfusion injury both GLP-1RAs and SGLT-2i are neuroprotective while GLP-1RAs pleotropic effect might be more prominent as this group of drugs has more pronounced infarct-limiting effect. Microglia damage and pathological activation are parts of acute and chronic brain damage pathogenesis. We can suggest that both GLP-1RAs and SGLT-2i in acute stroke and in chronic brain damage have direct positive influence on neuronal structure and survival while only GLP-1RAs also decrease microglial damage and pathological hyperactivation.

All datasets on which the conclusions of a manuscript depend are included in the article in the form of graphic material including individual values or boxplots. Additional information is available from the corresponding author on reasonable request.

DK, TV and TK designed the research study. AS, OF, NT, AI, DS and OK performed the research. AS, OF, DS and OK analyzed the data. AS and DS wrote the manuscript. All authors contributed to editorial changes in the manuscript, read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

All experimental procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH publication No. 85–23, revised 1996) and the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes. The study protocol was approved by Institutional Animal Care and Use Committee of Almazov National Medical Research Centre (Protocol Number PZ_22_2, Feb 16 2022). All efforts were performed to protect the laboratory animals and minimize their suffering throughout the study. The experiments complied with the ARRIVE guidelines (https://arriveguidelines.org/).

Not applicable.

This research was funded by the Ministry of Health of Russian Federation, agreement no. 124021600064-8.

The authors declare no conflict of interest.