Obesity-induced type 2 diabetes impairs neurological recovery after stroke in correlation with decreased neurogenesis and persistent atrophy of parvalbumin-positive interneurons

Stroke is the number one cause of permanent disability and a major social and economic burden [1–3]. Although a large proportion of the stroke survivors regain neurological function after rehabilitation therapy, approximately one-third of all stroke patients remain dependent on supportive care in activities of daily living (ADL) [4]. Type 2 diabetes (T2D), which is one major risk factor for stroke [5], also dramatically hampers neurological recovery in the surviving patients [6–8] and is a strong predictor of persistent ADL dependency [4]. Experimental stroke studies using toxin-induced hyperglycemia have shown that, similar to clinical observations, the neurological recovery in hyperglycemic animals is significantly attenuated [9,10]. Similar findings have been shown in genetically obese T2D animals (reviewed in [11]). The ones employing more physiological and clinically relevant obese/T2D models (induced by high-fat diet (HFD)) have mostly focused on the acute outcome (infarct size and neurological impairment) after stroke and, in the cases with more long-term outcome measures, the increase in the stroke injury was likely a determinant factor for decreased neurological recovery (reviewed in [11]). At present, it is therefore undetermined whether T2D induced by a chronic obesogenic diet affects long-term stroke recovery and if so, whether this effect is related to the impairment of brain’s self-repair mechanisms.

Impaired neurological recovery in T2D could not be explained solely by more severe gray or white matter infarctions [9,12,13] and several mechanisms including impaired vascular restoration [14] and increased inflammation [13] have been proposed. Recently, diminished cortical plasticity [9] and impaired initial stages of stroke-induced neurogenesis [10] have also been suggested. However, these studies have employed genetic and toxin-induced models of T2D, and the results need to be verified in more clinically relevant T2D models.

Stroke-induced neurogenesis in striatum is a potential contributor to stroke recovery [15]. After stroke, a portion of neural stem cells (NSCs) from the subventricular zone (SVZ) of the lateral ventricle migrates to the adjacent damaged brain areas, i.e. striatum, where they differentiate into neurons integrating in the brain circuits [16,17]. The ablation of this process impairs stroke recovery [18,19]. Moreover, in their immature, neuroblast state, these cells exhibit homeostatic, non-neurogenic functions by providing neurotrophic support to surviving neurons [20]. A recent study has shown that hyperglycemia impairs the first stages of striatal stroke-induced neurogenesis, i.e. NSCs proliferation and neuroblast formation [10]. However, it is unknown whether obesity-induced T2D produces the same effects and importantly, whether it leads to the actual reduction in terminal neuronal differentiation and striatal integration.

Experimental stroke after transient middle cerebral artery occlusion (tMCAO) can also affect hippocampus and induce long-term cognitive deficits [21], which have been shown to be more severe and longer lasting with underlying diabetes [22]. Similarly, hippocampal stroke in humans is associated with pronounced cognitive derangements [23]. Hippocampal neurogenesis is essential for the normal functioning of hippocampus [24,25] and since this process is increased after stroke, it could be important for cognitive recovery [26–28]. Therefore, it is important to determine whether reduced cognitive recovery in T2D after stroke is linked to decreased stroke-induced hippocampal neurogenesis. Few studies have in part addressed this question by showing that early stages of stroke-induced hippocampal neurogenesis are impaired by T2D [10,29,30]. However, it remains unknown if terminal neuronal differentiation is also affected by T2D.

The impairment of neuroplasticity mechanisms by hyperglycemia has been recently reported in the cerebral cortex of mice after toxin-induced hyperglycemia [9,10], as well as in human T2D stroke patients [31]. Neuroplasticity is an important contributor to stroke recovery and GABAergic inhibitory interneurons play a major role in this process [32]. Parvalbumin+ (PV) interneurons are a subtype of GABAergic interneurons that have been shown to modulate brain plasticity [33]. For instance, the regulation of PV expression in these interneurons has been recently shown to correlate with improved neurological recovery, suggesting that the expression of PV could normalize aberrant neuronal activities [34]. This is also supported by a study showing the vulnerability of these cells to stroke [35]. Moreover, degeneration and reduction in size (as a measure of cell atrophy) of parvalbumin-positive (PV+) interneurons has been reported in Huntington’s disease [36], suggesting that the regulation of these parameters can contribute to neuroplasticity after injury. Whether T2D-induced impairment of the normal functioning of PV+ interneurons could have negative implications on stroke recovery is unknown.

The goal of the present study was to determine whether obesity-induced T2D in the middle aged mice impairs the sensorimotor function during the stroke recovery phase in a mild stroke model mainly affecting basal ganglia. We also investigated whether the obese/T2D effect is linked to impaired stroke-induced neurogenesis and pathological alterations of striatal and cortical GABAergic inhibitory PV+ interneurons.

Fifty-six male, 8-week-old C57/BL6j mice (Charles River Laboratories, Germany) were used. All applicable international, national and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards of the Karolinska Institutet where all the studies were conducted. The ethical approval number is: S7-13 (Karolinska Institutet). All mice were housed in environmentally controlled conditions (25 ± 0.5°C, 12/12-h light/dark cycle with ad libitum access to food and water). From the age of 2 months, mice were fed with either standard laboratory diet (SD) (n=28) or HFD (60% energy from fat) (n=28) for 12 months. Body weight (BW), fasting blood glucose and oral glucose tolerance (OGT) were measured to verify induced-obesity/T2D (the data are summarized in Supplementary Figure S1). Once HFD-induced impaired glucose tolerance, obesity and fasted hyperglycemia were established, mice were subjected to experimental stroke (SD, n=20; HFD, n=20) or sham surgery (SD, n=8 HFD, n=8). Four mice (one SD, three HFD) died after stroke and the remaining mice were then randomly allocated to two studies: Study 1 (see below), to assess the recovery of forelimb sensorimotor function during 6 weeks following tMCAO and histological outcome at 6 weeks and Study 2 (see below) to assess the histological outcome at 2 weeks after tMCAO. Since the main goal of the study was to evaluate stroke recovery and neurogenesis at 6 weeks after MCAO, we prioritized Study 1 over Study 2 that was performed only for the histological assessment of early neurogenic response. Thus, more animals were allocated to Study 1 (see below).

The mice after tMCAO or sham surgery, (SD stroke, n=14; HFD stroke, n=12; SD sham n=4, HFD sham n=4) were tested for forelimb sensori-motor function by employing the grip strength test [37–41] (see also below). The test was performed before surgery and at 1 and 6 weeks after tMCAO/sham surgery. The mice were killed at 6 weeks and the histological analyses of neurogenesis/neuroplasticity were then performed (see below). Four mice from the HFD stroke group displayed lack of neurological deficits and had no visible brain damage (evaluated histologically at 6 weeks) after tMCAO. Therefore, they were removed from the study.

The mice (SD stroke n=5, HFD stroke, n=5; SD sham n=4, HFD sham n=4) were killed in week 2 after tMCAO/sham surgery for additional histological analysis at this earlier time point.

In both the Studies, to mark newly born cells, all mice received daily intraperitoneal injections of the thymidine analog bromodeoxyuridine (BrdU; 50 mg/kg of BW) for 2 weeks following tMCAO/sham surgery.

The Experimental Design of the study is summarized in Supplementary Figure S2.

tMCAO was used to model stroke by the intraluminal filament technique [42]. To induce striatal infarct mainly affecting basal ganglia and with minimal cortical and hippocampal damage, MCA was occluded for 30 min. Briefly, mice were anesthetized by 3% isoflurane, then maintained by 1.5% isoflurane through a snout-mask throughout the surgery. Body temperature was maintained at 37–38°C using a heated pad. Through midline incision, left common, external and internal carotid arteries were exposed. Through an incision in an external carotid artery, a 15-mm long, 7-0 silicone-coated monofilament (total diameter 0.17 –0.18 mm) was inserted into the internal carotid artery until it could not be advanced any further and at least 8–10 mm of filament length has passed the carotid bifurcation thus blocking the origin of the middle cerebral artery. Then the wounds were temporarily closed, blood glucose levels were measured (Supplementary Figure S3) and mice were allowed to wake up. After 25 min, mice were re-anesthetized, wound reopened and the occluding filament removed (total tMCAO time: 30 min). Additionally, the success of the stroke induction was evaluated by scoring the severity of neurological deficits 1 h after reperfusion and the next day after MCAO surgery using a 3-point system: 1 point for circling toward the paretic side, 1 point for right forelimb sensory deficit (left hanging from the table edge) and 1 point for hind limb sensory deficit. All animals used in the study reached minimum 2 points.

The forelimb sensorimotor function was measured by forepaw grip strength [37–41] using grip strength meter (Harvard apparatus, MA, U.S.A.) before and at 1 and 6 weeks after tMCAO. Briefly, mice were firmly held by the body and allowed to grasp the grid with individual forepaws. Mice were gently dragged backward until the grip was released. Ten trials were performed and the highest value was recorded as described previously [41]. Grip strength of left and right forepaws was measured separately, and motor asymmetry was determined by the right to left forepaws strength ratio.

Brain preparation was performed as described previously [43]. Mice were deeply anesthetized and transcardially perfused with 4% ice-cold paraformaldehyde, then the brains were removed and after overnight post-fixation submerged in phosphate-buffered saline (PBS) with 20% sucrose until they sank. The brains were cut in 30-μm-thick coronal sections using a sliding microtome.

Immunofluorescence staining was performed using free-floating method. Following primary antibodies were used; rabbit anti-Ki67 (1:300 dilution; #ab15580; Abcam), a marker of cell proliferation; goat anti-DCX (doublecortin) (1:300 dilution; #sc-8066; Santa Cruz Biotechnology), a marker for migrating neuroblasts; mouse anti-NeuN (1:200 dilution; #MAB377; Millipore), a neuronal marker; rabbit anti-DARPP32 (1:500 dilution; #ab40801; Abcam), a marker of medium-size spiny neuron; rat anti-BrdU (1:500 dilution; #ab6326; Abcam), a marker of cell proliferation; and rabbit anti-parvalbumin (PV) (1:1000 dilution; #ab11427; Abcam), a marker of PV-expressing interneuron. A combination of rat anti-BrdU with anti-NeuN and anti-DARPP32 were employed to assess neurogenesis, respectively. Sections were incubated with primary antibodies overnight at 4°C in a phosphate buffer containing 3% appropriate serum and 0.25% Triton X-100. Primary antibodies were detected by Alexa 488- or Alexa 594-conjugated (Vector) or Cyanine3-conjugated (Thermo Fisher Scientific) secondary antibodies (1:200 dilution). Sections were incubated with secondary antibodies for 2 h at room temperature (approximately 21°C) in phosphate buffer containing 3% of the appropriate serum and 0.25% Triton X-100. For antigen retrieval 1 mM EDTA for 30 min at 64°C (for PV immunostaining) or 1 M HCl for 20 min at 64°C (for double staining with BrdU).

Brain sections from bregma 0.98 mm, −1.06 mm and 2.06 mm were stained with Luxol Fast Blue (LFB) for quantitation of neuronal myelin fiber bundles/contents [44]. Brain sections were mounted on the gelatin-coated slides. Sections were immersed in 95% ethanol for 3 min, then incubated with 0.1% LFB solution containing the Solvent blue 38 (Sigma) at 58 C for 15 h. Sections were rinsed with 95% ethanol and distilled water, then differentiated by lithium carbonate solution for 5 min followed by immersion in 70% ethanol and rinsed in water. The differentiation process was repeated for six times.

Stroke volume was evaluated based on NeuN staining. NeuN staining is a consistent method for quantitating neural damage, since it exclusively stains neurons and can be reliably used to evaluate neuronal loss even several weeks after stroke, unlike ubiquitous cell markers like TTC (3,5-triphenyltetrazolium chloride) or Hematoxylin and Eosin that are accurate only within few days after the injury, due to later inflammatory cell infiltration and glial scar formation at injury site. Stroke volume was determined as described previously [41,45–48]. Briefly, NeuN-labeled brain sections that contained stroke damage were displayed live on a computer monitor and the areas of the whole contralateral hemisphere and the intact region of the ipsilateral hemisphere were measured by using NewCast Software (Visiopharm). The measured area was multiplied by the distance between the sections to estimate the volume. The stroke volume was calculated by subtracting the volume of intact ipsilateral hemisphere from the volume of the whole contralateral hemisphere.

The Olympus BX51 epi-fluorescent/light microscope (Olympus) connected with computerized setup for stereology (NewCast Software, Visiopharm) were used for cell counting. Three consecutive brain sections spaced at 300 μm containing striatum (from Bregma 1 to 0.5 mm) were used. The first section was chosen based on an anatomical location along the rostra-caudal axis (approximately 1 mm from Bregma). The second and the third sections were 300 and 600 μm caudal from the first section respectively. Similarly, three evenly spaced sections were chosen for analyses of hippocampal neurogenesis starting at −1.8 mm from Bregma. The number of Ki67-positive (Ki67+) cells was counted in the SVZ and the subgranular zone (SGZ) of dentate gyrus of hippocampus. The number of DCX-positive (DCX+) cells, NeuN/BrdU and DARPP32/BrdU double positive cells were counted in striatum and the granule cell layer (GCL) of dentate gyrus (NeuN/BrdU). Double-staining was later verified using confocal microscopy. For counting of PV+ interneurons the NewCast software (VisioPharm, Denmark) was used. Briefly, striatum or cortex on three coronal sections (described above) was delineated and the counting frame was systematically moved at preset intervals starting from random position, so that the representative fraction of the region of interest (ROI) was sampled. The total cell number from three sections was estimated using the following formula: Total number of PV+ interneurons = Counted number × (Step area/Counting frame area) [49]. Mean PV cell body volume (a measure of potential cellular atrophy) was estimated with the nucleator technique using 40–50 random (experimenter-independent, software-determined random positions within the area of interest) cells from three sampled sections [49]. Myelin determination includes the thickness of corpus callosum, the striatal number of neuronal myelin fiber bundles and the striatal amount of myelin, based on optical density. The thickness of the corpus callosum was measured approximately 2 mm from the midline of both hemisphere via using the NewCast Software (Visiopharm). ImageJ software was used for quantitations of the optical density and of the number of neuronal myelin fiber bundles. A rectangular ROI was manually chosen in the center of ipsilateral striatum and corresponding location in contralateral striatum. Since we observed the deformation in ipsilateral hemisphere, the ROI in the infract striatum was adjusted by deformation value. The deformation value was calculated by subtracting the area of ipsilateral hemisphere from the area of the contralateral hemisphere. The number of neuronal myelin fiber bundles were counted in ROI. LFB optical density was quantitated by measuring the intensity from ROI on RGB image with background correction as previously described [44,50]. All procedures were performed by experimenter blinded to experimental groups.

For grip strength test analyses, two-way repeated measures ANOVA was used followed by Tukey’s test to compare grip strength between time-points within each group and Bonferroni’s test to compare grip strength between the groups at each time-point. Ordinary Two-way ANOVA followed by Tukey’s test was employed for the majority of the immunohistochemistry (IHC) studies when comparing the differences between the groups in relation to two parameters. When comparing the differences between two groups in relation to one single parameter, the unpaired t test with Welch’s correction was used. The comparison versus the respective sham group in the IHC experiments was performed using the ordinary one-way ANOVA test with Dunnett’s multiple comparisons. Data are expressed as mean ± SD. P-value less than 0.05 was considered statistically significant. All data were analyzed by using GraphPad Prism 7.

Sham surgery did not affect the grip strength neither in SD- nor in HFD-fed mice (Figure 1A). No differences were detected in upper-limb grip strength between SD-fed mice and HFD-fed mice before stroke and at 1 week after stroke, although stroke significantly decreased upper-limb grip strength (before stroke vs. 1 week) in both SD and HFD groups (P<0.0001 for both groups) (Figure 1A). However, at 6 weeks after stroke, SD-fed mice significantly recovered upper-limb grip strength (1 vs. 6 weeks, P<0.0001), while HFD-fed mice did not and remained significantly different from before stroke and HFD sham (Figure 1A). The recovery in the SD group was significant at 6 weeks compared with 1 week after stroke, and also showed no difference to before stroke (Figure 1A). Additionally, a head to head comparison of grip strength at 6 weeks between SD and HFD mice revealed that SD mice significantly outperformed HFD mice in grip strength test (P<0.0001) (Figure 1A).

The ischemic brain damage was present within three-fourths of the striatum and mainly localized in dorsolateral striatum without significant differences between SD- and HFD-fed mice (Figure 1B–D). Ipsilateral cortical damage was minimal or not present.

We then determined the potential effects of obesity/T2D on the number of striatal myelinated fiber bundles after stroke. The number of myelinated fiber bundles decreased similarly in the stroke-damaged striatum of both SD and HFD-fed mice (as compared with uninjured, contralateral striatum; Supplementary Figure S4A–C). Furthermore, the thickness of the corpus callosum in the dorso-ventral axis and the amount of myelin in the remaining ipsilateral and contralateral striatal myelinated fiber bundles were not different between the two groups or hemispheres (data not shown). This indicates that stroke decreased the number of neuronal myelinated fiber bundles, with no effect on the content of myelin within the remaining myelinated fiber bundles. These effects were not modified by T2D.

To study stroke-induced neurogenesis in striatum, we assessed NSCs proliferation and neuroblast formation by quantitating Ki67+ and DCX+ cells in the SVZ and striatum, respectively and neurogenesis by quantitating BrdU/NeuN and BrdU/Darpp32 positive cells in striatum.

The results in Figure 2A–F show that at 2 weeks after stroke, SD-fed mice significantly increased the number of Ki67+ and DCX+ cells in the ipsilateral versus the contralateral hemisphere (P=0.0126 and P<0.0001, respectively), while in HFD-fed mice only a non-significant trend toward the increase in DCX+ neuroblasts was noted (P=0.1), indicating that T2D impairs stroke-induced NSCs proliferation and neuroblast formation during the first 2 weeks after stroke.

At 6 weeks after stroke, the number of Ki67+ cells was still higher in ipsilateral versus contralateral hemispheres in SD-fed mice (P=0.0184; Figure 3A,C,D), but not in HFD-fed mice. Interestingly, the number of DCX+ cells was significantly increased in the ipsilateral versus the contralateral hemispheres of both SD and HFD-fed mice (P=0.0041 and P=0.0036, respectively), see Figure 3B,E,F. In consideration of the results at 2 weeks after stroke (Figure 2B) where the HFD mice show no increase in DCX+ positive cells in the ipsilateral hemisphere, the results at 6 weeks indicate a delayed activation of neuroblast formation in the HFD group.

The number of Ki67+ and DCX+ cells were similar in SD- and HFD-fed, sham-operated mice at both 2 and 6 weeks after the surgery (Supplementary Figure S5A–D).

To assess neurogenesis, we quantitated the number of NeuN/BrdU+ and DARPP32/BrdU neurons in ipsilateral to stroke striatum at 6 weeks after stroke. The results show no significant difference in stroke-induced neurogenesis between SD-fed mice and HFD-fed mice (Figure 4A–D).

Activation of hippocampal neurogenesis after stroke even without detectable ischemic damage to hippocampus has been shown in several rodent stroke models [18,51] allowing to study this process without local tissue damage being a confounding factor. To evaluate stroke-induced NSCs proliferation and neuroblast formation in hippocampus, we quantitated the number of Ki67+ cells and DCX+ cells in the SGZ and GCL, respectively, at 2 and 6 weeks after stroke.

The number of Ki67+ and DCX+ cells were similar in SD- and HFD-fed, sham operated mice at both 2 and 6 weeks after the surgery (Supplementary Figure S6A–D).

The results in Figure 5A–D show that at 2 weeks after stroke, SD-fed mice increased the number of Ki67+ and DCX+ cells in the ipsilateral versus the contralateral hemisphere (P=0.0256 and P=0.0001, respectively), while HFD-fed mice did not. This indicates that obesity/T2D impairs stroke-induced NSCs proliferation and neuroblast formation in hippocampus during the first 2 weeks after stroke.

At 6 weeks after stroke, we observed no difference in the number of Ki67+ and DCX+ cells between the ipsilateral and contralateral hemispheres in either experimental group (Figure 6A–D).

To assess neurogenesis, we quantitated the number of NeuN/BrdU+ neurons in GCL at 6 weeks after stroke. The result in Figure 6E,F show a significantly higher number of newly generated mature neurons in the ipsilateral versus the contralateral hemisphere of SD-fed mice (P=0.0328). On the contrary, the results in HFD-fed mice show no difference between the two hemispheres (Figure 6E,F), demonstrating the impairment of stroke-induced hippocampal neurogenesis by obesity/T2D.

We quantitated the number of PV+ interneurons and measured their soma volume, as a measure of potential atrophy, in striatum and overlaying cortex in both ipsilateral and contralateral to stroke hemispheres. The cortical assessment was mainly aimed to determine potential effects of T2D on PV+ cell atrophy rather than cell number, since cortex was not damaged by stroke after 30 min of tMCAO. The results show that the number of PV+ interneurons and PV cell volume in either cortex or striatum were not different between SD-fed mice versus HFD-fed mice in the sham groups (Supplementary Figure 7A–D). This suggests that T2D did not affect the number or soma volume of these neurons under un-injured (sham) conditions. Moreover, we found no differences in the number of PV+ interneurons and PV cell volume in cortex of stroke-subjected animals versus the sham in either diet group (Figure 7A–C), indicating, as expected, no influence of stroke on cortical PV+ interneurons at 2 weeks after tMCAO.

In striatum, the number and soma volume of PV+ interneurons were significantly and similarly reduced in the ipsilateral hemisphere of both SD-fed mice and HFD-fed mice after stroke compared with sham mice (P=0.0001 and P=0.0001; Figure 7D,E). The soma volume of PV+ interneurons was measured separately within infarct and peri-infarct areas. The analyses revealed that the stroke-induced reduction in PV+ cell volume affected both infarct (P=0.0001 and P=0.0001 in SD and HFD, respectively) and peri-infarct areas (P=0.0023 and P=0.042 in SD and HFD, respectively) and was significantly greater within the infarcted region, independently from diabetes (Figure 7E,F). Altogether, our results demonstrate striatal PV+ interneuron loss, as well as cellular atrophy of surviving PV+ interneurons at 2 weeks after stroke. These effects were not influenced by obesity/T2D.

At 6 weeks after surgery, the sham groups, similar to stroke groups at 2 weeks, were not different regarding the number of PV+ interneurons and their average cell volume in either cortex or striatum (Supplementary Figure S8A–D). No difference in the number of PV+ interneurons versus the sham was found in cortex of stroke-subjected animals in both groups (Figure 8A,C). However, the PV+ cell volume of SD-fed mice was increased in both ipsi- and contralateral hemispheres in comparison with the sham groups (P=0.0213 and P=0.0401, respectively; Figure 8B,C). This effect was blunted in HFD-fed mice (Figure 8B,C), indicating the modifying effect of obesity/T2D on the increase of PV cell volume in cortex in response to stroke. As expected, no difference in PV+ cell number was detected in the cortex.

In striatum, the number of PV+ interneurons in the ipsilateral hemisphere in both SD- and HFD-fed mice were similarly reduced by stroke (P=0.0001 and P=0.0001, respectively; Figure 8D,F). As at 2 weeks, the average volume of PV+ cells was measured separately in the infarct and peri-infarct areas. The results show that the PV+ cells in the peri-infarct area of SD-fed mice recovered to original soma size, while in HFD-fed mice the cell bodies remained shrunken at 6 weeks after stroke (P=0.0269; Figure 8E,F).

These results show that 6 weeks after stroke SD-fed mice increase the size of PV+ interneurons in cortex and that this neuroplasticity-like effect is impaired by T2D (Figure 8B,C). The results also show that 6 weeks after stroke, striatal PV+ interneurons in the peri-infarct striatum of SD-fed mice respond to the brain damage by recovering (Figure 8E,F) from the stroke-induced soma atrophy observed 2 weeks after stroke (Figure 7E,F). This neuroplasticity-like response in striatum is hampered by obesity/T2D.

The primary goal of the present study was to determine whether obesity-induced T2D impairs the recovery of forelimb sensorimotor function after stroke. We show that obesity/T2D induced by 12 months of HFD feeding dramatically hampers recovery without exacerbating neuronal or white matter injury. Secondarily, we demonstrate that the obesity/T2D-induced impairment of the recovery of forelimb sensorimotor function after stroke correlates with transient changes in the turnover of NSCs and neuroblasts in striatum and with a dramatic decrease in hippocampal neurogenesis. Finally, we show that stroke induces the atrophy of surviving PV+ interneurons early after the insult and that these cells gradually recover to their original size. However, obesity/T2D dramatically impairs such recovery effect.

The effect of genetic, toxin-induced T2D, or short-term ‘high fat, high sugar’ diets to worsen brain injury and exacerbate neurological deficits after stroke has been reported by a number of preclinical studies [9–11]. In these studies, the difference in the severity of brain injury after stroke between normal and T2D animals not only complicates the clear-cut determination of whether T2D also influences post-acute neurological recovery, but also the understanding of the underlying mechanisms, since the initial differences in injury severity influences both these parameters [52]. In agreement with our results, an interesting study by Zhang et al. [10] showed that hyperglycemia induced by streptozocin (STZ) administration impaired neurological recovery in rats without affecting the severity of the neuronal damage. However, differently from our study, Zhang et al. [10] also recorded the exacerbation of white matter injury, which could play a role in the observed reduction in long-term neurological recovery under hyperglycemia. Moreover, as interesting as these results are, this model of diabetes might not fully represent the clinical reality of T2D patients, as it induces fast changes in glycemic state and establishes sustained hyperglycemia, as opposed to slow development of metabolic syndrome in typical, lifestyle-induced T2D [53]. Therefore, the likelihood of such acute stress to modify stroke outcome, as well as to alter neurological recovery cannot be discounted. The goal of the present study was to determine whether obesity/T2D impairs the recovery of forelimb sensorimotor function after stroke by employing an animal model that resembles the clinical situation of progressive T2D in the middle age. Indeed, the incidence of T2D increases with age [54]. Additionally, as explained above, the absence of differences in brain injury was an important requirement to study the cellular processes behind the impaired motor-recovery. The strength of our model is that unlike genetic/toxin-induced diabetes, or short term ‘high fat, high sugar’ diet-exposure, the chronic (12 months) feeding with HFD resulted in a condition similar to severe clinical obesity accompanied with mild T2D in the middle age. Indeed, models of obesity based on HFD feeding are believed to mimic better the state of common obesity in humans than most of the genetically modified models [55,56] that often do not model the disease etiology of the majority of patients by representing mutations leading to very rare cases of obesity in humans (e.g. of the leptin pathway) [57] and limiting reliable and translatable insight into human T2D [58]. Our model based on chronic HFD feeding did not exacerbate neuronal injury after stroke (compared with age-matched healthy mice), but strongly impaired the recovery of forelimb sensorimotor function. These results are in line with a clinical study by Megherbi et al. [8] who showed that 3 months after stroke T2D patients present increased handicap and disability compared with non-diabetics. This model also did not exacerbate white matter injury after stroke. The disruption of white matter integrity in the uninjured brain of T2D patients has been recently shown [59]. However, whether it is exacerbated after stroke in T2D patients is still largely undetermined [60,61]. A few preclinical studies show increased white matter injury in T2D rodents, although they have been conducted in hyperglycemic models induced by STZ [10] or by employing genetic animal models of T2D, i.e. db/db mice (already hyperglycemic and hyperinsulinemic a few weeks after birth) [62–64]. To our knowledge, the present study is the first to report no major difference in the white matter damage after stroke after a prolonged obesogenic diet. Altogether, the lack of both white and gray matter damage exacerbation in our obese/T2D model allowed the study of the recovery of forelimb sensorimotor function, stroke-induced neurogenesis and neuroplasticity without the increased brain injury as a confounding factor.

The deficiency in the capacity to recover from stroke-induced neurological impairments in T2D patients has long been established [6–8]. To understand the mechanisms underlying this phenomenon, animal models that accurately reflect this clinical reality are needed. The challenge of modeling long-term motor deficits in animals is hampered by the ability of small rodents to shortly recover from stroke-induced motor deficits, unless the brain injury is considerably severe [65–67]. A potential positive aspect of our study was the use of a motor assessment test that was sensitive enough to detect neurological impairments for several weeks after stroke. The forepaw grip strength as a measure of forelimb sensorimotor function has been used previously for the evaluation of neurological recovery in experimental stroke [37–41]. However, the grip strength has been usually measured in mice grasping the instrument with both forepaws simultaneously, while the stroke-induced motor impairment is unilaterally localized. These studies report quick recovery (within a couple of weeks) to baseline. In our study, we measured the grip strength of individual forepaws that we believe has increased the sensitivity of the test. Additionally, it has been reported that the grip strength test can detect ageing-associated decline in skeletal muscle function in the mouse [68] suggesting that when using this test, the age of the animals needs to be taken under consideration. Indeed, we have recently demonstrated full recovery of grip strength within 3 weeks in young mice after focal ischemia [41], while in this study in middle-aged mice, even at 6 weeks after tMCAO the grip strength did not fully recover. Taken together, these data argue that individual forepaw grip strength measurement is sufficiently sensitive for evaluating motor recovery for several weeks after experimental stroke in middle-aged and obese/T2D mice. On the other hand, the neurological recovery assessed by solely measuring the forelimb sensorimotor function with the grip strength test also represents a weakness of the present study and the results will have to be confirmed by employing additional behavioral tests. Another limitation of the present study is also represented by the fact that only one animal model of stroke and obesity has been employed and therefore additional studies in the field will be needed to confirm our results.

We hypothesized that the impairment of motor recovery by obesity/T2D could be linked to suppression of brain’s self-repair mechanisms such as stroke-induced neurogenesis and neuroplasticity. Besides the potential direct effect of T2D on NSCs and neurons [69], T2D-induced adverse microvascular changes could also potentially be of importance for explaining decreased recovery after stroke and several recent studies have investigated this matter [70–73]. Moreover, the role of vasculature on stroke-induced neurogenesis has been previously demonstrated [74].

The effect of diabetes on basal adult neurogenesis has been previously described in rodent in vivo (for review, see [75]) and in vitro models [69,76]. Different models as well as differences in the stages of the diabetes development at the time of investigation, resulted in discrepancies ranging from decrease to no effect or even increase in the proliferative activity of NSCs and neural progenitor cells [75]. Although the negative effect of toxin-induced hyperglycemia on the initial stages of striatal neurogenesis after stroke has been reported by Zhang et al. [10], the effect of obesity-induced T2D on the generation of new, mature neurons in striatum has not been previously determined. In the present study, we observed the suppression of the initial stages of stroke-induced striatal neurogenesis such as SVZ cell proliferation and neuroblast formation by T2D early (2 weeks) after stroke. This effect was not detectable at later time-point (6 weeks). These observations are different from those reported by Zhang et al. [10], where the reduction in both SVZ stem cell proliferation and neuroblast formation was sustained up to 5 weeks post stroke. Model differences between toxin-induced and diet-induced diabetes or the rodent species could potentially account for this discrepancy. Importantly, when we examined the end-result of stroke-induced striatal neurogenesis (the formation of mature neurons) by quantitating the cells expressing BrdU and the mature neuronal markers NeuN and Darpp32, we did not observe differences between healthy and obese/T2D mice. These results indicate that obesity/T2D does not affect the differentiation of neuroblasts into mature striatal neurons while it affects the initial stages of the stroke-induced neurogenesis.

The data from Arvidsson et al. [16] show that more than 80% of newly formed neural precursor cells in striatum die before neuronal differentiation and thus before final integration in the neuronal circuits. Our results confirm this observation by showing that despite large numbers of DCX+ neuroblasts at 2 weeks, very few newly formed, mature striatal neurons (BrdU/NeuN+, BrdU/DARPP32+) were present in either group at 6 weeks after tMCAO. On the other hand, non-neurogenic, trophic effects of neural precursor cells after brain injury have been previously reported (for review, see [77]). Although speculative, it is likely, that the initial increase (at 2 weeks) of NSCs (Ki67+) and neuroblasts (DCX+) in striatum of healthy mice can contribute to improved functional recovery and that this effect is hampered by obesity/T2D. Due to increased neural precursor turnover, these cells could die more in healthy versus obese/T2D mice thus explaining the lack of difference between healthy and obese/T2D mice in the amount of newly formed, mature striatal neurons.

As in striatum, the initial stages of hippocampal neurogenesis after stroke (by quantitating the proliferative marker Ki67 and the neuroblast marker DCX 2 weeks after stroke) were also impaired by obesity/T2D. These results are in line with previous studies [10,29,30,75]. Interestingly, in contrast with striatum, the generation of hippocampal mature neurons 6 weeks after stroke (by detecting double positive BrdU/NeuN mature neurons) was also severely suppressed by obesity/T2D. To our knowledge this is the first study showing that obesity/T2D impairs stroke-induced hippocampal neurogenesis. The functional role of hippocampal neurogenesis in the overall recovery of impaired hippocampal function after stroke is largely unknown, although studies suggest it is limited or even harmful, mainly due to the limited ability of new neurons to integrate in neuronal networks properly [51,78]. However, recent animal studies have shown that T2D decreases cognitive functions after stroke [10,79], also by employing HFD feeding [22,80]. Whether the impairment of hippocampal neurogenesis by T2D has functional implications is an exciting question that needs to be addressed in future studies.

Functional recovery after brain injury heavily relies on neuroplasticity (for review see [81]) and an impairment of this process has been reported in rodents after toxin-induced hyperglycemia [9]. GABAergic, PV+ striatal interneurons are important contributors to neuroplasticity after injury [32,33]. Interestingly, it has been recently reported that toxin-induced hyperglycemia reduces PV+ interneurons survival after focal ischemia [82], although it is unclear whether slow development of metabolic syndrome (after HFD exposure) would lead to similar results. To determine the effect of chronic HFD on PV+ interneurons in the post-stroke recovery phase, we quantitated the number of surviving PV+ interneurons and evaluated the magnitude of short- and long-term soma atrophy [36] after tMCAO.

A significant number of PV+ interneurons were lost in striatum after stroke and a strong atrophy was recorded in surviving PV+ interneurons at 2 weeks after stroke. The atrophy of PV+ interneurons has been observed in neurodegenerative conditions [36]. We did not record any additional effect by obesity/T2D on these parameters at this time point. However, 6 weeks after stroke, PV+ interneurons in non-diabetic mice regained their original soma volume in peri-infarct striatum. In addition, in the uninjured cerebral cortex of these mice PV+ interneurons grew by ≈10% in size in both contralateral and ipsilateral hemispheres (Figure 8B,C). Larger neuronal soma size has been shown to be associated with increased neuronal activity and consequently elevated metabolic demands [83,84]. Therefore, we hypothesize that an increase of cortical PV+ interneuron soma size after stroke could indicate increased neuronal activity, which could in turn be interpreted as a compensatory response to the loss of striatal neurons. These effects were blunted in the obese/T2D mice suggesting an impairment of these potential compensatory mechanisms. The results suggest a previously uncharacterized form of neuroplasticity in the post-stroke recovery phase where PV+ interneurons can regulate their activity based on size changes, both in cortex and in the peri-infarct ischemic brain regions. Whether this mechanism plays a determinant role in stroke recovery remains to be addressed in future studies. If so, our data showing that obesity/T2D impairs this process will have an important role for understanding the mechanisms behind impaired stroke recovery in T2D.

In conclusion, we show impaired long-term recovery of forelimb sensorimotor function in middle-aged obese/T2D mice without substantial effects on stroke-induced brain injury. We also show for the first time that stroke-induced neurogenesis in striatum (initial stages) and hippocampus (terminal neuronal differentiation) are severely impaired by obesity/T2D. In this regard, future studies will have to determine whether the modulation of non-neurogenic trophic effects by NSCs and/or neural precursor cells in striatum and of newly formed neurons in the hippocampus could affect motor recovery and cognition, respectively. Finally, we identified a potentially new form of PV+ interneuron-mediated neuroplasticity in the long-term recovery phase after stroke which is characterized by changes in neuronal cell size and that was severely impaired by obesity/T2D.

Overall, the results provide new knowledge on the effects of obesity/T2D on the recovery phase after stroke that could be exploited to develop new strategies for stroke rehabilitation.

• The present study was undertaken to determine whether obesity-induced T2D affects long-term neurological recovery after stroke and if so, through which cellular mechanisms.

• We show in a preclinical setting of clinical relevance that obesity-induced T2D dramatically impairs long-term motor recovery after stroke. This detrimental effect was not associated with stroke-induced gray or white matter damage, but with impaired neurogenesis and atrophy of surviving PV+ interneurons.

• The results of thepresent study could be clinically relevant to understand the stroke pathology in T2D. Moreover they could contribute to identify the cellular mechanisms behind impaired stroke recovery in T2D thus facilitating the development of new strategies for stroke rehabilitation in the obese/T2D population.

We thank Dr. Fuad Bahram (Södersjukhuset) for technical assistance and Dr. Hans Pettersson for advice on statistical analyses.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the European Foundation for the Study of Diabetes (EFSD)/Sanofi European Diabetes Research Programme in Macrovascular Complications, the Swedish Research Council [grant number 2018-02483]; the Swedish Heart-Lung Foundation [grant number 20160511]; the Svensk Förening för Diabetologi; Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse; the Novo Nordisk Foundation; the EFSD Albert Renold Travel Fellowship Programme (to F.C.); the Karolinska Institutet (Foundation for Geriatric Diseases (to H.P.) and the KI Stiftelser och Fonder (to G.L.); the Stohnes Stiftelse, O. E. och Edla Johanssons Stiftelse, Magnus Bergvalls Stiftelse, STROKE Riksförbundet, Gamla Tjänarinnor Stiftelse and by the Regional Agreement on Medical Training and Clinical Research (ALF) between Stockholm County Council and the Karolinska Institutet; in part by the Boehringer Ingelheim; unrestricted grants from the Astrazenca (to T.N.); Consultancy Fees from Boehringer Ingelheim (to T.N.); the Eli Lilly (to T.N.); the Novo Nordisk (to T.N.); the Merck (to T.N.); and the Sanofi-Aventis (to T.N.).

H.P. performed IHC studies and stereology analysis; acquired and processed images and figures; contributed to discussion; and wrote the manuscript. F.C. and G.L contributed to discussion and helped with the immunohistochemistry experiments. I.L.A. participated in the analysis of additional behavior data and edited the manuscript. T.N. provided expertise and resources, contributed to discussion and edited the manuscript. C.P. conceived, designed, and coordinated the research plan, contributed to discussion and edited the manuscript. V.D. conceived and designed the study, performed the stroke experiments, contributed to discussion and edited the manuscript.

ADL

activities of daily living

BrdU

bromodeoxyuridine

DCX

doublecortin

GCL

granule cell layer

HFD

high-fat diet

IHC

immunohistochemistry

Ki67+

Ki67-positive

LFB

Luxol Fast Blue

NSC

neural stem cell

PV+

parvalbumin-positive

PBS

phosphate-buffered saline

ROI

region of interest

SD

standard laboratory diet

STZ

streptozocin

SVZ

subventricular zone

T2D

type 2 diabetes

tMCAO

transient middle cerebral artery occlusion