Background
Dipeptidyl-peptidase 4 (DPP-4) inhibitors (gliptins) are oral antidiabetic drugs used to treat type 2 diabetes mellitus (T2D). Gliptins mediate their anti-diabetic effects by primarily inhibiting degradation of endogenous glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic peptide (GIP), resulting in prolongation of postprandial insulin secretion [
1]. Recent research has shown that gliptins can also reduce stroke-induced brain damage in animal models in presence or absence of diabetes [
2]. Furthermore, several reports have shown that gliptins mediate positive pleiotropic effects in animal models of Alzheimer’s disease (AD) [
3‐
7] and in diabetic patients with AD [
8]. Translation of these positive functional results to diabetic (and non-diabetic) individuals affected by stroke remains to be demonstrated since large clinical studies have not yet evaluated the potential of gliptins in improving functional stroke outcomes [
9]. Instead, these studies assessed gliptins’ efficacy to prevent cardiovascular events (including stroke) and to decrease mortality in people with diabetes with basically neutral results [
2,
10,
11].
The molecular mechanisms underlying gliptin-mediated effects in brains are also largely unknown. GLP-1 and GIP are regarded as main DPP-4 substrates. However, we recently showed that Linagliptin can improve stroke outcome independently from glycemia regulation [
12] and GLP-1R [
13]. These data indicate that one or more additional DPP-4 substrates with direct or indirect neuroprotective properties may be involved in gliptin-mediated brain effects. GIP can play a role in neuroprotection after stroke [
14]. However, DPP-4 also cleaves other peptides, of which many exhibit direct actions on the cardiovascular system [
15‐
17]. Among these, a promising candidate is the C-X-C motif chemokine 12 (CXCL12) [stromal cell-derived factor 1 alpha (SDF-1α)], which has been demonstrated to be fundamentally involved in brain homeostasis [
18]. SDF-1α is a small cytokine mediating mobilization and homing of bone marrow-derived stem and progenitor cells in vascular injury [
19], lymphopoiesis, myelopoiesis and germ cell mobilization [
20]. To exert its actions, SDF-1α activates two receptors, CXCR4 and CXCR7 [
19]. SDF-1α and its receptor CXCR4 are abundant and ubiquitously expressed in the developing and mature central nervous system, playing a role in neurogenesis and contributing to the neuronal development [
19,
21]. Furthermore, the levels of SDF-1α and expression of CXCR4 in plasma and cerebrospinal fluid were decreased in clinical and preclinical studies of AD and negatively correlated to changes in cognitive functions [
22]. The role of SDF-1α in cerebral ischemic injury is complex since some studies have shown positive effects of SDF-1α in the acute phase after stroke [
23,
24] whereas other studies have demonstrated positive effects by blocking the SDF-1α/CXCR4 pathway in the recovery phase after stroke [
25,
26].
In the present study, we investigated whether protective effects of linagliptin after stroke are mediated via SDF-1α by blocking CXCR4 with the selective CXCR4 antagonist AMD3100. Additionally, by using tandem mass spectrometry, we identified effectors putatively involved in gliptin-mediated effects.
Discussion
The primary objective of this study was to determine whether the improved outcome after stroke following gliptin treatment is SDF-1α/CXCR4-dependent. We showed that linagliptin improves functional stroke outcome in a SDF-1α/CXCR4-dependent manner. Secondarily, we demonstrated that linagliptin after stroke decreased the presence of peptides derived from NEUG and MBP.
Different research groups have shown that gliptins reduce brain damage and improve functional parameters after stroke in various animal models independently from a T2D background (reviewed in [
2,
41,
42]). A few large clinical studies with gliptins in diabetic patients have investigated the potential of these drugs to decrease cardiovascular incidence (including stroke) and death with neutral results (reviewed by Nauck et al. [
10]). However, since the efficacy measures in these clinical studies (stroke incidence and death) did not address functional outcomes after stroke, further clinical studies are needed to evaluate the potential of these drugs to improve functional stroke outcome [
9]. Interestingly, the ongoing CARMELINA study with linagliptin (Clinicaltrials.gov; NCT01897532) contains a post-stroke functional sub-study using the modified Rankin scale to assess stroke-induced disability approximately 1 week following stroke and at ~ 3 months after stroke-onset. Preclinical data indicating that gliptins can improve stroke outcome in the post-stroke recovery phase have been recently shown by Ma et al. in a model of transient cerebral ischemia induced by bilateral common carotid artery occlusion. The study showed that sustained linagliptin treatment after cerebral ischemia counteracted cognitive impairment and brain atrophy, independently from the regulation of glycemia [
43]. This study is remarkable because their model allows extending the observation period for several weeks after artery occlusion thus evaluating effects of sustained gliptin treatment in the post-stroke recovery phase. Our results confirm that sustained linagliptin treatment after stroke is necessary to improve stroke outcome while a single acute bolus administration of linagliptin at stroke time was ineffective (previously published in [
13]).
Clinical data suggesting that gliptins can exert beneficial effects in the damaged brain do also exist. Isik et al. recently showed that a treatment for 6 months with sitagliptin was associated with improvement of cognitive function in elderly diabetic patients with and without Alzheimer’s disease [
8]. To support the translation of preclinical functional outcome stroke studies with gliptins to clinical settings, it is helpful to identify mechanisms of action of this class of drugs in the brain.
The antidiabetic effects of gliptins are mediated via GIP and GLP-1 regulation, but other incretin-independent mechanisms may also be involved [
15]. We recently showed that mice lacking the GLP-1-receptor exhibit improved stroke outcome after linagliptin treatment [
13]. Recent research by Han et al. has shown that a dual agonist targeting both GLP-1 and GIP receptors promoted stronger neuroprotection against stroke than the GLP-1 analogue Val(8)-GLP-1(glu-PAL) alone thus suggesting a mechanism mediated by GIPR activation [
14]. Further studies employing mice lacking GIP and/or GIPR are needed to investigate this hypothesis. However, we showed in this study that brain GIP levels were unaffected by linagliptin treatment. Furthermore our previous study using mice lacking the GLP-1R [
13] showed that linagliptin can improve stroke outcome independently from GLP-1R. Although we cannot rule out peripheral effects mediated by GIP, this suggests that the positive effect of gliptins on stroke outcome may not be necessarily related to incretins.
The DPP-4 substrate SDF-1α plays a pivotal role in the brain, as it regulates neurovascular remodeling after stroke [
23‐
26]. The beneficial effects of SDF-1α were also shown in rats after traumatic brain injury [
44] and in an AD animal model, where SDF-1α treatment decreased beta-amyloid deposition [
45]. Of relevance for our study, recent research in myocardial infarction (MI) has shown that increased SDF-1α by gliptins mediates protective effects against MI through anti-apoptotic effects [
46,
47].
Our results show that sustained linagliptin treatment increases active SDF-1α in brain parenchyma. Importantly, by sustained blocking of the SDF-1α/CXCR4 pathway, linagliptin-mediated effects on functional and histological outcomes after stroke were diminished. This indicates that improved stroke outcome by linagliptin occurs via the activation of the SDF-1α/CXCR4 pathway. DPP-4 activity was similarly inhibited in linagliptin, and linagliptin/AMD 3100-treated animals (data not shown). Therefore, the inhibitory effect of AMD 3100 over linagliptin on stroke outcome could not be linked to altered DPP-4 activity between the groups. The pro
versus adverse effects of SDF-1α after gliptins treatments in diabetic complications has been recently deeply discussed [
48,
49]. Our results suggest that, at least when it comes to post-stroke treatment, the activation of the SDF-1α/CXCR4 pathway promotes beneficial effects. Two weaknesses of this set of results that need to be addressed in the future are: (1) the fact that the study was performed in naïve mice and that a diabetic background could have affected the outcome; (2) it is unclear why the SDF-1α/CXCR4 pathway seems to be more involved in the functional (Fig.
3a) rather than in the structural recovery (Fig.
3d). Nevertheless, previous studies show that SDF-1α is involved in axonal path finding, outgrowth and branching; all functions involved in functional recovery [
19].
The effects of linagliptin to reduce the injury after stroke could involve the neuroprotective, non-neurogenic rapid effects of neural progenitor cells (NPCs) [
50] since CXCR4 inhibition in NPCs leads to failure of newborn neurons to localize to the ischemic brain tissue [
19]. The linagliptin effects via SDF-1α/CXCR4 to reduce the brain injury after stroke could also be mediated by the regulation of neovascularization through endothelial progenitor cells (EPCs) [
51] since gliptins increase ischemic angiogenesis by preserving EPCs function [
52]. Moreover, SDF-1α is known to be involved in the recruitment to the injury region of EPCs [
19]. This action of SDF-1α on EPCs could contribute to explain the stronger efficacy of linagliptin to improve stroke outcome after 3 weeks versus 3 days of treatment as it has been shown that EPCs recruitment occurs 2 weeks after ischemic injury [
53]. New studies should be performed in the future to demonstrate this hypothesis.
Finally, the SDF-1α/CXCR4 pathway could play a role in functional regulation of the brain vasculature, since linagliptin improves endothelium-dependent relaxation independently of glucose regulation [
28]. Furthermore, SDF-1α/CXCR4 signaling activates endothelial nitric oxide synthase [
32] which is a key enzyme maintaining homeostasis by inducing vasodilatation and whose impairment is implicated in the pathogenesis of stroke [
54]. These results support the possibility that the positive effects of gliptins on stroke outcome could also occur via increased blood perfusion in collateral vessels in the
penumbra region of ischemic brains thus mitigating the degenerative effects of MCAO.
To further elucidate effects of linagliptin after stroke, we analyzed brain tissue samples by mass spectrometry. We did not detect linagliptin under our experimental conditions in agreement with previous published data demonstrating that linagliptin does not cross the blood brain barrier under physiological conditions [
55]. However due to MALDI-TOF/TOF mass spectrometry conditions, we cannot rule out that traces of linagliptin could enter the brain under stroke conditions.
Peptides that are generated by proteolytic processing of larger precursor proteins can be regarded as surrogate markers for expression levels of proteins or peptidase activity [
56]. In samples from animals subjected to MCAO, peptides derived from G3P were observed. We speculate that these peptides reflect a direct effect of ischemia causing release of substances from the cytosol due to cell membrane instability.
The analysis further revealed that NEUG peptides exhibit lower signal intensities in stroke tissue samples from animals treated with linagliptin (Fig.
4 and Figure S2 in Additional file
1). These results might suggest lowered expression of NEUG or a reduced susceptibility to proteolytic processing of free NEUG not bound to Calmodulin.
The protein level or proteolytic cleavage pattern of an isoform of MBP appeared to be also altered (Fig.
5). The observed pattern of MBP peptides (Figure S3 in Additional file
1) shows a very good fit to Calpain processing based on the substrate and observed cleavage sites according to MEROPS, a database of proteolytic enzymes [
57]. We presume that the observed MBP peptides do not represent breakdown products due to stroke since no significant difference between VS and LC/VC samples was present. Rather the low intensities or even diminished presence of MBP peptides in samples from mice subjected to stroke and treated with linagliptin, mirror altered proteolytic activities or insusceptibility to processing of MBP due to e.g. Calmodulin binding [
40] as a second-tier effect (see below) analog to NEUG. These peptides are not affected in samples from linagliptin-treated control mice, which is in line with the observation that linagliptin does not cross the blood brain barrier under physiological conditions [
55].
These data suggest that the proteolytic processing products of two Calmodulin-binding proteins exhibit significantly lower signal intensities in brain samples after stroke under linagliptin treatment either due to altered expression, insusceptibility to proteolytic processing because of Calmodulin binding and/or altered activity of proteases like Calpain. Since presence of free NEUG and MBP and/or altered Calpain activity are all dependent on intracellular Ca
2+ concentration [
38,
58], we hypothesize that linagliptin treatment, presumably through SDF-1α, could affect Ca
2+ homeostasis. Indeed, a study by Nicolai et al. [
59], showed that SDF-1α selectively inhibits the expression of NR2B, a regulatory subunit of NMDA receptor, altering NMDA-induced Ca
2+ responses associated with neuronal death, while promoting pro-survival pathways. However further studies (intracellular Ca
2+ measurement, Calpain activity) are necessary to verify or falsify the proposed mechanism.
Authors’ contributions
FC performed immunohistochemistry studies and stereology analysis; acquired and processed images and figures; contributed to discussion; and wrote the manuscript. HT performed the mass spectrometry experiments, contributed to discussion and edited the manuscript. HP and GL contributed to discussion and helped with the immunohistochemistry experiments. MC provided expertise and resources and edited the manuscript. TN provided expertise and resources, contributed to discussion and edited the manuscript. TK conceived the research plan, provided expertise in DPP-4 inhibitors, coordinated ELISA studies, contributed to discussion, and edited the manuscript. VD designed, conceived and performed the stroke experiments, contributed to discussion and wrote the manuscript. CP conceived, designed, and coordinated the research plan, contributed to discussion and wrote the manuscript. All authors read and approved the final manuscript.