Stroke leads to cognitive impairment partly due to the disruption of hippocampal function. Recent studies provided evidence that stroke causes hippocampal structural changes and dysregulation of hippocampal networks, leading to memory decline in humans [1
]. Interestingly, cerebral ischemia has been shown to stimulate the proliferation of endogenous neural progenitor cells in animal models [6
] and stroke patients [7
]. Previous studies have suggested that ischemic stroke triggers increased proliferation of neural progenitor cells in the hippocampal subgranular zone [8
], and the migration of these cells to the damaged areas [15
]. However, these potential self-repair mechanisms are thought to operate only for a restricted time period after stroke, with the number of newborn neurons being insufficient for full tissue repair and their existence being transitory [19
]. Recent findings suggest that a significant portion of newborn neurons generated after stroke reveal aberrant morphology and fail to correctly integrate into pre-existing networks [20
]. Given that the capacity of this regenerative mechanism is limited, a therapeutic intervention that can stimulate neurogenesis and promote the integration of newborn neurons would be highly desirable to promote post-stroke recovery of cognition.
Interestingly, the pro-cognitive effects of growth hormone (GH) have been demonstrated in multiple neurological conditions, including traumatic brain injury [22
]. GH has been shown to stimulate the genesis of neurons and endothelial cells, as well as to promote myelination and synaptogenesis [23
]. Importantly, we have recently shown that GH treatment after photothrombotic vascular occlusion of the somatosensory and motor cortex promoted a marked improvement in the performance of associative memory cognitive domain as measured by the paired-associate learning task [24
]. We additionally found an enhancement of neurorestorative processes, such as increased cell proliferation, neurogenesis, increased synaptic plasticity, myelination and angiogenesis within the peri-infarct region, most likely contributing to the improvement in motor outcomes [24
]. However, we did not investigate the pro-cognitive effects of GH treatment on other cognitive domains and possible neurorestorative changes in the hippocampus. Early studies by Pathipati at al. demonstrated that intracerebroventricular delivery of GH after experimental stroke improved spatial memory using the Morris water maze test; however, they did not investigate changes in the hippocampus [29
]. Thus, there are limited studies documenting the role of GH on hippocampal plasticity in the context of stroke recovery.
Therefore, an important question that we wanted to address in the current study was how GH treatment could enhance neurorestorative processes within the hippocampus remote to a cortical stroke, thus leading to an improvement in overall post-stroke cognition. Specifically, we examined changes in the density, distribution and protein expression of markers of neural progenitor cells in the hippocampal formation, using bromodeoxyuridine (BrdU) tagging and doublecortin (DCX). We also examined changes in the AMPA receptor subunit glutamate receptor 1 (GluR1), which plays an important role in synaptic plasticity. In addition, we examined changes in cerebrovascular density, using tomato lectin, a blood vessel stain. We assessed the effect of GH treatment on learning and memory using a mouse touchscreen platform for the hippocampal-dependent visual discrimination (VD) task [30
We demonstrated that GH treatment after experimental cortical stroke stimulates proliferation of neural progenitor cells and increases synaptic plasticity within the hippocampus, a remote region from the initial cortical injury, in parallel with improvements in hippocampal-dependent VD task. While long-term GH treatment after experimental stroke has been previously shown to improve cognitive function [24
], this is the first study to our knowledge to comprehensively show both functional and hippocampal plasticity of GH treatment. Thus, the present study extends the understanding of how GH treatment acts on hippocampal plasticity to promote post-stroke cognitive recovery.
Our group recently established that cortical photothrombotic stroke impairs the ability of mice to discriminate between stimuli with a high degree of similarity [31
]. Although our cortical stroke model does not cause a direct damage to the hippocampus, we observed an indirect significant decrease in the number of neurons within the hippocampus at 2 weeks post-stroke [31
]. It should be noted that a cortical stroke could also induce a dysfunction in the hippocampal-thalamic network [32
]. While cerebral ischemia causes persistent neuronal loss in remote regions that are functionally connected to the primary infarction site [33
], interestingly, it also triggers a transient proliferation of endogenous neural progenitor cells in the DG of the hippocampus [6
]. Importantly, we previously demonstrated that GH treatment after experimental stroke promotes the associative memory cognitive domain as measured by the paired-associate learning task [24
]. However, we only investigated the neurorestorative effects of GH treatment within the peri-infarct regions [24
]. The observed enhancement of recovery-promoting mechanisms within the peri-infarct area is most likely contributing to the improvement in motor outcomes [25
], whereas the promotion of hippocampal neurogenesis and increased synaptic plasticity reported here are likely to be of importance for improved cognition.
The present study reports novel evidence that GH treatment promotes hippocampal neurogenesis, remotely from the cortical injury, and enhances the performance of hippocampal-dependent VD tasks. Our findings further suggest that higher levels of plasma IGF-1 are associated with better performance in hippocampal-dependent VD task, which is consistent with prior findings [24
]. GH can exert its effects directly or indirectly via its mediator insulin-like growth factor 1 (IGF-1) on hippocampal plasticity [37
]. Firstly, GH receptor expression is widespread in the brain and there are studies supporting a passage of GH over the blood–brain barrier; therefore, GH may have direct effects on the brain [38
]. Secondly, it is possible for GH to increase production and secretion of local IGF-1 in the brain [39
]. Thirdly, GH induces the secretion of IGF-1 in the circulatory system, and peripheral IGF-1 can penetrate the blood–brain barrier to exert effects on the brain [41
]. However, it is not possible for our experimental design to specifically elucidate which mechanisms are involved. Nevertheless, we would suggest that it is most likely a combination of these pathways. We used BrdU tagging, to assess whether there is an increase in neural progenitor cells. We also evaluated the expression of DCX, a marker of immature neurons [43
]. Our results show an increased number of BrdU-positive cells and expression of DCX in the DG subregion in GH-treated stroke mice. Together, these findings of increased neurogenesis are critical, as recovery of the cognitive function following stroke depends on cellular plasticity in the hippocampal formation [14
We also investigated other key neurorestorative processes in the hippocampal formation or in its vicinity. Firstly, we assessed synaptic plasticity by measuring the AMPA receptor subunit GluR1, which plays an important role in synapse formation, stabilization, and plasticity [45
]. We found that GluR1-positive structures and protein levels were increased in the hippocampus in GH-treated mice. While the results might be an indication of higher synapse numbers or higher synaptic activity, it should be noted that GluR1 may also be found extrasynaptically [48
]. Future comprehensive investigation on synaptic changes using co-immunolabelling of synaptic structures and high-resolution microscopy is warranted. Secondly, given that stroke-induced remote secondary neurodegeneration is a disruption of connections between the cortex and hippocampus [31
], we evaluated white matter tract disturbances. The results from Sudan black staining suggest that GH treatment provides restorative effects on white matter structural loss. Together, these results are particularly interesting when considered in conjunction with the cognitive function findings. Thirdly, we explored the effect of GH treatment on cerebrovascular remodeling. In our previous studies, we showed that GH has the ability to enhance angiogenesis, and increases the density and area coverage of both CD31 and Collagen IV–positive cells in the peri-infarct region [24
]. In the current study, we analyzed vessel density in the hippocampus area using immunofluorescence labelled tomato lectin. However, GH treatment did not affect the % area covered by tomato lectin or the number of BrdU-lectin-positive cells in the hippocampal formation, as we have previously shown in the peri-infarct area. These results suggest that some effects of GH appear to be global in the brain, while others are more brain region-specific and cell type-specific.
There are two important points regarding the experimental design of the current study that require consideration. Firstly, we used r-hGH for practical purposes and from the translational point of view. r-hGH is much more accessible than mouse GH, and has been used in a wide setting of mouse experimental models with expected GH-related results. However, there some differences in the actions of r-hGH and mouse GH. Although r-hGH stimulates the GH receptor and an increase in serum IGF-I, there is also a potential minor stimulation of prolactin receptors [49
]. Nonetheless, it appears that prolactin is not able to induce local IGF-I, at least not in the skeleton [51
]. Therefore, this does not compromise the results, but gives an alternative mechanism. Secondly, we administered r-hGH subcutaneously via mini-osmotic pumps for 28 days after experimental stroke, as we accounted for the impact of stress due to repeated injections. However, continuous infusion of GH may disrupt the endogenous GH pulsatile secretion, and different modes of GH administration can have minor different effects on the brain [52
]. This warrants further investigation using programmable pumps.
5. Materials and Methods
All animal experiments were approved by the University of Newcastle Animal Care and Ethics Committee (A-2014-432, 23 November 2015) and undertaken in accordance to the Animal Research: Reporting of In Vivo Experiments guidelines. C57BL/6 mice (male, 10 weeks old, n = 48) were obtained from the Animal Services Unit at the University of Newcastle. Mice were maintained in a 12:12h reverse light–dark cycle (lights on 19:00h). All procedures conducted in the dark phase.
5.2. Sample Size Calculation
Sample size was estimated using G*Power 3.1 software. Using previous and preliminary data [24
], we obtained an effect size of d
= 1.6. Allowing a type 1 error of 5%, α
= 0.05, with the power of 80%, β
= 0.2, we calculated a samples size of 8 animals per group.
5.3. Experimental Design
The experimental design of this study was as described in Figure 1
]. A total of 48 mice were used in this study. All the experimental groups are randomized, and all outcome analyses were performed in a blinded manner.
The first cohort of mice (n = 24) was used to assess cognitive function, and brains were collected for histological analysis. At day 0 (D0), all mice were subjected to photothrombotic occlusion. At D2, mice were randomly allocated to receive either r-hGH or saline at 1.4 mg/kg body weight per day subcutaneously via mini-osmotic pumps for 28 days (stroke + saline, n = 12, and stroke + r-hGH, n = 12). At D3, mice were injected with BrdU (50 mg/kg body weight) for 5 consecutive days. To evaluate cognitive function, mice were subjected to mouse touchscreen platform for visual discrimination (VD) task for 18 days (days 9–26 post-stroke). At D30, mice were euthanized and their brains were collected. Mice were excluded from the study if we histologically identified that the stroke had not occurred or malfunction of the mini-osmotic pumps. Within the saline group, one mouse had to be excluded due to post-surgery death and another one due to no stroke. Within the r-hGH group, we had to exclude one mouse due to pump malfunction and two mice due to a problem with the perfusion procedure.
The second cohort of mice (n = 24) was used for biochemical analysis. At D0, mice were subjected to photothrombotic occlusion or sham surgery (stroke, n = 18, and sham, n = 6). At D2, stroke mice received r-hGH or saline as described above (sham + saline, n = 6, stroke + saline, n = 8, and stroke + r-hGH, n = 10). At D30, mice were euthanized and their brains were collected for Western blotting. Within this second cohort, one mouse from the stroke + r-hGH group had to be excluded for not having a stroke.
5.4. Photothrombotic Occlusion
Photothrombotic occlusion was performed as described previously [31
]. Briefly, mice were anesthetized by 2% isoflurane during surgical procedure on a temperature controlled (37 ± 1 °C) stereotaxic frame. Rose Bengal (200 μL, 10 mg/mL solution in sterile saline, Sigma-Aldrich, St. Louis, MO, USA) was injected intraperitoneally. At 8 min post-injection, the skull was exposed and illuminated for 15 min by a cold light source positioned at 2.2 mm left lateral of Bregma 0.0 mm. For the sham group, the Rose Bengal injection was substituted with 200 μL of sterile saline (0.9% NaCl, Pfizer, Sydney, NSW, Australia).
5.5. Mini-Osmotic Pump Placement
Mini-osmotic pump (Model 2004, Alzet, Cupertino, CA, USA) placement was performed as preciously described [24
]. At D2, the mini-osmotic pumps were implanted. An incision was made in the skin between the scapulae where the mini-osmotic pump was inserted. The pumps were filled with 200 μL of either recombinant human growth hormone (r-hGH, Somatropin 10 mg/1.5 mL, SciTropin A, SciGen, Belrose, NSW, Australia) or sterile saline. The delivery rate of the pumps was 0.25 μL/h for 28 days (0.04 mg r-hGH per day).
5.6. Visual Discrimination (VD) Task
Mouse touchscreen operant chambers were used in the cognitive testing as described [30
], and were conducted in a blinded and randomized manner. Briefly, mice were calorie restricted overnight before cognitive testing. Strawberry milkshake was used as a reward to motivate the performance of the mice. First, mice were trained to learn to associate a nose poke of the touchscreen and the delivery of the reward. Following training, mice underwent photothrombotic occlusion surgery. At D9, mice were subjected to the VD task. This task entailed simultaneous presentation of two stimuli: one correct (S+) and one incorrect (S−). The S+ stimulus appeared pseudo-randomly either in the right or left part of the touchscreen. When the mouse selected the correct choice, S+, a tone was triggered, the reward tray was illuminated, and strawberry milkshake was delivered to the tray. If the mouse selected the incorrect image, S−, there was no reward delivery, no tone, the house light was turned on for 5 s, and a correction trial was initiated. In each VD session, the testing ended once a mouse successfully completed 30 trials or reached a 60 min time limit, whichever occurred first. All mice were subjected to a total of 18 sessions.
5.7. Tissue Processing
For the first cohort, mice were anaesthetized with sodium pentobarbital. Mice were perfused via the ascending aorta with ice-cold 0.9% saline followed by ice-cold 4% paraformaldehyde (pH 7.4). Brains were dissected and post-fixed for 4 h in the same fixative then transferred to a 12.5% sucrose solution in 0.1M phosphate buffered saline (PBS) for storage and cryoprotection. Serial coronal sections were sliced on a freezing microtome (Leica, North Ryde, NSW, Australia) at a thickness of 30 μm.
For the second cohort, mice were anaesthetized with sodium pentobarbital, and transcardially perfused with ice cold 0.9% saline. Brains were dissected and rapidly frozen in −80 °C isopentane. Brain sections were sliced using a cryostat (−20 °C) at a thickness of 200 μm. The hippocampal formation (Bregma −1.2 to −2.5mm) samples were dissected and stored frozen in −80 °C until further analysis.
Free-floating fixed sections were co-immunostained as previously described [60
]. For BrdU staining, antigen retrieval was performed before the blocking step as follows: 10 min HCl (1M) incubation on ice, 10 min HCl (2M) incubation at room temperature, 20 min HCl (2M) incubation at 37 °C, 10 min borate buffer (0.1M) incubation at room temperature and three washes in PBS + 0.1% Triton X. Sections were blocked using 3% bovine serum albumin, then incubated with appropriate primary antibody (DCX, GluR1, BrdU, NeuN) overnight at 4 °C, and followed by 2 h incubation in corresponding secondary antibodies at 25 °C (see Table 1
for antibodies concentration). Tomato lectin staining was performed together with the secondary antibody incubation. NeuroTrace was used to counterstain sections after immunocytochemical detection of DCX or GluR1. Brain sections were washed with PBS in between each incubation step.
5.9. Sudan Black Staining
Sudan black staining was performed as previously described [62
]. Briefly, sections were mounted and rinsed with 70% ethanol followed by 15 min incubation with Sudan Black B solution (Sigma-Aldrich, USA). After staining sections were rinsed with 70% ethanol and water and 5 min counterstained with nuclear fast red solution (Sigma-Aldrich, USA).
5.10. Image Acquisition and Analysis
Immunofluorescence high-resolution confocal images were taken on a Leica TCS SP8 confocal microscope with a Leica HC PLC APO 10 × 0.40 objective. For each region of interest, 30 μm z-stacks with a step size of 1 μm were taken. Imaging parameters (laser power, resolution and gain) were held constant throughout all imaging sessions. Exhaustive automated BrdU cell counts were performed using ImageJ software. For the analysis of NeuN, DCX and GluR1 labelling, we performed thresholding analyses and chose the optimal pixel intensity that clearly reflected the immunofluorescence signal. To measure vessel coverage (% lectin+ area), the emission channels were split and the lectin emission image was uniformly thresholded at a high stringency. The area of vessel coverage was expressed as a percentage of the overall field of view (ImageJ Software, National Institutes of Health, Bethesda, MD, USA). For BrdU/NeuN and BrdU/lectin co-labelling, we used the plugin ‘colocalization’ for ImageJ. This plugin highlights the colocalizated points of two 8-bits images. The colocalizated points appear black by default.
Sudan black images were acquired at 20× using Aperio AT2 (Leica, Wetzlar, Germany). Specifically, we focus on the corpus callosum and we assessed white matter loss as a difference between the contralateral and ipsilateral hemispheric area (mm2). The quantitative analysis was undertaken specifically in two sections (Bregma 0.0 and −2.0 mm).
5.11. Protein Extraction and Western Blotting
Protein extraction and Western blotting were performed as previously described [63
]. Hippocampus samples were sonicated in 300 μL lysis buffer (50 mM tris(hydroxymethyl)aminomethane buffer pH 7.4 containing 80 μM ammoniummolybdate, 5 mM β
-glycerolphosphate, 1 mM dithiothreitol, 1 mM ethylenediaminetetraacetic acid, 1% sodium dodecyl sulfate, 1 mM sodium pyrophosphate, 1 mM sodium vanadate, 1 cOmplete™ protease inhibitor cocktail tablet, and 1 PhosSTOP™ phosphatase inhibitor cocktail tablet) and centrifuged at 14000 g
for 20 min at 4 °C. Supernatants were collected and protein concentrations were estimated by Pierce BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Sample buffer (2% sodium dodecyl sulfate, 50 mM tris(hydroxymethyl)aminomethane, 10% glycerol, 1% dithiothreitol, 0.1% bromophenol blue, pH 6.8) was added into the supernatants. To determine the protein levels of NeuN, DCX, GluR1 and Collagen IV, the protein homogenates from sham + saline (n
= 6), stroke + saline (n
= 8) and stroke + r-hGH (n
= 9) were analyzed together on 26-well gels. Then, 15 μg of protein lysate was electrophoresed into Biorad Criterion TGX Stain-Free 4–20% gels and transferred to polyvinylidene fluoride membranes. The membranes were blocked with 5% skim milk in TBST for 1 h at room temperature and incubated overnight at 4 °C with the appropriate primary antibody: NeuN, DCX, GluR1 and Collagen IV (see Table 1
for antibodies concentration). The next day, membranes were incubated with secondary antibody for 1 h at 25 °C. In between each incubation step, membranes were washed in TBST. Chemiluminescence signals were detected on an Amersham Imager 600 using Luminata Classico western blotting detection reagent (Merck Millipore, Burlington, MA, USA). The density of the bands was measured using Amersham Imager 600 analysis software (GE Healthcare Life Sciences, Pittsburgh, PA, USA). The protein markers were normalized to β
-actin (as loading control), and were expressed as a fold change of mean ± SD for each group relative to the mean of the sham + saline group.
5.12. Statistical Analyses
All data were expressed as mean ± SD and were analyzed using GraphPad Prism v7.02 (GraphPad Software, San Diego, CA, USA). The primary outcome measurement was differences between stroke + saline and stroke + r-hGH. Data from Western blotting, immunofluorescence labelling and white matter loss were analyzed using the 2-tailed t-test. VD task (18 sessions temporal analysis) was analyzed using 2-way ANOVA, followed by Sidak multiple comparisons. A p value < 0.05 was considered statistically significant.