Growth differentiation factor-11 supplementation improves survival and promotes recovery after ischemic stroke in aged mice

Brain GDF11 declines with age in mice and humans

To determine whole-brain GDF11 levels, young (8-12 weeks) and aged (20-22 months) male mice were euthanized and brains were collected for western blot and ELISA. Brain GDF11 levels were significantly lower in older mice (p<0.01, Figure 1A) as assessed by ELISA. Brain GDF11/8 (12.5 kDa) expression was also assessed by Western blot and was significantly lower in old (p<0.05, Figure 1B) versus young mice. GDF11/8 positive cells were significantly decreased (p<0.05, Figure 1C) in individuals above 75 years as compared to those <60 years of age.

Figure 1. GDF11/8 levels decline with age in both mice and humans. (A) ELISA shows GDF11 levels decline with age in the brain of aged mice (20-22-month-old), (B) reduced brain GDF11/8 expression in old mice was confirmed by Western blotting, (C) brain GDF11/8 (brown) positive cells decline with age in older humans. Hematoxylin (blue) counterstain. Mouse studies, n=5, human. Magnification 20X, n=8-9, *p<0.05, **p<0.01.

Recombinant GDF11 treatment improved early neurological recovery and reduced brain inflammation after transient ischemic stroke

It is unknown if the age-related loss of GDF11 contributes to the poor post-stroke repair seen in aged animals. To determine if GDF11 altered early neurological outcomes in aged male animals after stroke, we administered recombinant growth differentiation factor-11 (rGDF11) five days after middle cerebral artery occlusion (MCAo) for 5 days. At day 10 post-stroke, a significant decrease (p<0.05, Figure 2A) in the neurological deficit score was seen in the GDF11 treated mice suggesting an earlier recovery of neurological deficits. GDF11 acts by phosphorylation and activation of Smad2 and Smad3 [11, 30]. A slight increase in the brain pSmad2 and pSmad3 expression was seen in the sham mice treated with rGDF11 compared to sham vehicle, this did not reach statistical difference. An increase in brain pSmad2 and pSmad3 expression was seen between sham and MCAo vehicle-treated groups (Figure 2B, 2C). No difference between sham rGDF11 and MCAo rGDF11 treated mice was seen in brain pSmad2 and pSmad3 expression. A decrease in brain IL-18 levels with exogenous GDF11 was seen (p<0.05, Figure 2D). Furthermore, an interaction between surgery and treatment was seen in brain IL-15 levels (p<0.05). A significant decrease (p<0.05) in brain IL-15 levels was seen in the MCAo rGDF11 treated mice at day 10 post-stroke (Figure 2E) suggesting that rGDF11 reduced brain inflammation.

Figure 2. Recombinant GDF11 treatment improved early neurological recovery by reducing brain inflammation after transient ischemic stroke. (A) a decrease in NDS at day 10 post MCAo, (B) pSmad2 and (C) psmad3 expression in the vehicle and rGDF11 treated mice at day 10 post-stroke, (D) decrease in brain IL-18 and (E) IL-15 levels in the MCAorGDF11 treated mice at day 10 post-stroke n=3-7,*p<0.05.

Exogenous GDF11 administration led to decreased mortality, reduced brain tissue loss, and increased motor activity at 30 days after experimental stroke

There was a decline in the body weights in all stroke cohorts at day 7 post-stroke compared to respective sham-treated mice (p<0.05) which gradually recovered by day 30 (Supplementary Figure 1). We observed a significant increase in the survival probability between the MCAo rGDF11 group as compared to the MCAo vehicle group (p<0.05, Figure 3A). The mortality in the MCAo cohort treated with vehicle was as follows; day 5 (1 mouse), day 6 (1 mouse), day 8 (2 mice), day 10 (1 mouse), day 13 (1 mouse), day 18 (1 mouse) day 19 (1 mouse) and day 20 (1 mouse) after stroke. In the MCAo cohort treated with rGDF11, mortality was observed on day 11 and 19. In total nine mice in the vehicle group and two mice in GDF11 died respectively. There was no morality in sham mice.

Figure 3. Reduced tissue loss and mortality after exogenous GDF11 treatment in old MCAo mice. (A) survival probabilities, GDF11 treatment led to (B) less brain tissue loss, (C) decrease in cerebrospinal fluid (CSF) area, and preserved (D) corpus callosum (CC) area. An increase in (E) NeuN count and (F) intensity with rGDF11 treatment was seen at 30 days post-stroke. Magnification 20X. Scale bar 20 μm. n=5-8,*p<0.05, **p<0.01.

Cresyl violet staining showed a significant reduction in brain tissue loss in the rGDF11 group at 30 days post-stroke compared to vehicle mice (p<0.01, Figure 3B). This was confirmed by MRI. MRI showed a decrease in the CSF area (p<0.05, Figure 3C) and an increase in the CC area (p<0.05, Figure 3D) in the MCAo rGDF11 group as compared to the MCAo vehicle-treated mice. Thirty days post-MCAo, the number of NeuN⁺ cells and intensity was higher in the MCAo rGDF11 group as compared to MCAo vehicle (p<0.05, Figure 3E, 3F) suggesting a protective role of GDF11 during stroke recovery.

At day 14 post-stroke, an increase (p<0.05, Figure 4A) in the ability to walk, evaluated by digigait testing, was seen in rGDF11-treated mice as compared to vehicle-treated animals. An increase (p<0.05, Figure 4B) in the total distance moved in open field task was also seen in the rGDF11 treated cohort at day 30 post-stroke, demonstrating an improvement in overall locomotor activity. Mice in the rGDF11 group were more mobile during the tail suspension task compared to the stroke vehicle group suggesting a role of GDF11 in depressive behavior (Figure 4C). Nest building ability, a measure of sensorimotor and cognitive function [31], was not significantly different between the groups (p=0.05); although the nest building score was higher in the rGDF11 treated mice as compared to vehicle-treated mice (Figure 4D). No difference between sham vehicle and sham rGDF11 mice was seen in nest building, the ability to walk on digigait and distance moved on an open field (Supplementary Figure 2). There was a significant decrease in the time immobile on tail suspension task seen in the sham rGDF11 treated mice as compared to sham vehicle (Supplementary Figure 2D) suggesting GDF11 might reduce depressive phenotypes.

Figure 4. Improvement in motor function on day 30 with rGDF11 treatment. (A) improved gait function as assessed by the digigait at day 14, (B) increase in distance moved in the open field at day 30 post-MCAo, (C) reduction in time immobile, (D) nest-building score at day 30 post MCAo in aged mice. n=6-12 *p<0.05, ***p<0.001.

Exogenous GDF11 reduces gliosis and increases angiogenesis after stroke

The immunoreactivity of glial fibrillary acidic protein (GFAP), ionized calcium-binding adaptor protein (Iba-1), and CD31 was assessed 30 days after stroke to investigate if rGDF11 had effects on gliosis, or angiogenesis respectively. No difference between the average number of GFAP, Iba-1, and CD31⁺ endothelial cells was seen between sham vehicle and sham rGDF11 mice (Supplementary Figure 3). There was a significant decrease (p<0.05, Figure 5A) in GFAP⁺ cells in the peri-infarct area in the rGDF11 treated mice compared to vehicle-treated stroke mice, reflecting reduced astrogliosis after exogenous rGDF11 treatment. A reduced (p<0.05, Figure 5B) number of Iba-1⁺ cells was seen with exogenous rGDF11 treatment, suggesting a decrease in microglial activation in treated mice. Treatment with rGDF11 increased CD31⁺ endothelial cells (p<0.05, Figure 5C) as compared to vehicle-treated mice post-stroke. We observed an increase in the vessel percentage area (p<0.05, Figure 6B), the total number of vessel branch points (p<0.05, Figure 6C) and total vessel length (p<0.05, Figure 6D) in the rGDF11-treated mice as compared to the vehicle-treated group. Furthermore, an increase (p<0.05, Figure 6E) in BrdU/lectin⁺ cells was observed in the GDF11 group as compared to the vehicle group at day 30 post-stroke, reflecting an increase in angiogenesis. An increase (p<0.05, Figure 5F) in the number of BDNF/lectin⁺ cells was seen in the MCAo rGDF11 mice as compared to the MCAo vehicle group. There was an increase in BrdU⁺ cells after rGDF11 treatment (p<0.05) and stroke (p<0.05) in the subventricular zone (SVZ) (Supplementary Figure 4). An increase in the BrdU⁺ cells after and stroke (p<0.05) was seen in the hippocampus. There was no difference in DCX⁺ BrdU⁺ cells in the SVZ or hippocampus (Supplementary Figure 4) between rGDF11 and vehicle-treated groups at 30 days post-stroke.

Figure 5. Exogenous GDF11 reduced gliosis and increased CD31⁺ endothelial cells. (A) decline in GFAP ⁺ and (B) Iba-1 ⁺ cells in MCAo GDF11 group at day 30 post-stroke, (C) increase in CD31⁺ endothelial cells in the MCAo GDF11 cohort at day 30 after stroke. Magnification 20X. Scale bar 20μm. n=5. *p<0.05.

Figure 6. Exogenous GDF11 increased angiogenesis in older mice after stroke. (A) representative images showing lectin stain and inset shows quantitatively measured region using Angio tool software (Magnification 20X), (B) vessel percentage area, (C) total number of junctions, (D) total vessel length was increased in the MCAo rGDF11 treated older mice, (E) representative images showing BrdU and lectin stain in the treatment groups, BrdU/lectin⁺ cells and BDNF/lectin⁺ cells in the treatment groups, (F) increase in BDNF/lectin⁺ cells in the rGDF11MCAo group. Magnification 20X (Lectin/BrdU⁺ cells) magnification 63X (Lectin/BDNF⁺ cells). Scale bar 20μm. n= 5-6, *p<0.05.

GDF11 treatment reduces white matter injury and restores presynaptic plasticity

We investigated the role of GDF11 on synaptic plasticity, axonal damage, and demyelination in the corpus callosum and peri-infarct regions. Synaptophysin is a reliable marker for axonal damage, presynaptic plasticity, and synaptogenesis [32] and MBP is a marker of white matter damage [33]. Recombinant GDF11 treatment increased MBP intensity in the CC and peri-infarct striatum (STR) (p<0.05, Figure 7A). The expression of synaptophysin was increased in the rGDF11 treated group compared to vehicle-treated stroke mice (p<0.05, Figure 7B). Additionally, a decrease (p<0.05, Figure 7C) in the GFAP intensity and percentage area GFAP in the CC was observed in the rGDF11 treated group compared to vehicle-treated stroke mice.

Figure 7. GDF11 treatment restores white matter integrity and synaptic plasticity at day 30 post-MCAo. (A) increase in MBP intensity in the CC and peri-infarct striatum, (B) increase in synaptophysin immunostaining, (C) decrease in GFAP⁺ cells in CC at day 30 post MCAo in aged mice. Magnification 20X. Scale bar 50μm. n=5-7. *p<0.05, **p<0.01.

Study subjects

C57BL/6J young (8-12 weeks) and old (20-22 months) male mice were pair-housed in a specific pathogen-free facility (light cycle 12/12 h light/dark). Food and water were provided ad libitum. All animal procedures were performed in accordance with NIH guidelines for the care and use of laboratory animals and approved by the Animal Care Committee of the University of Texas Health Science Center at Houston, McGovern Medical School.

All human tissue samples were obtained from the University of Pittsburgh neurodegenerative brain bank with appropriate ethics committee approval (Committee for Oversight of Research and Clinical Training Involving Decedents). provides the demographic details for the human controls.

Experimental groups

To investigate baseline changes in GDF11 levels with age, young and old mice were euthanized and serum and brain GDF11 levels were assessed. For the long-term recovery study (30 days) mice underwent MCAo surgery and were randomly divided into four groups, sham vehicle, sham rGDF11, MCAo vehicle, and MCAo rGDF11. To investigate the role of rGDF11 in early stroke recovery, mice were randomly divided in the sham vehicle, sham rGDF11, MCAo vehicle, and MCAo rGDF11 and euthanized at 10-day post MCAo

Middle cerebral artery occlusion (MCAo) and treatment

Aged mice underwent transient focal ischemia under isoflurane anesthesia for 1 hour by occlusion of the right middle cerebral artery (MCA) [52] details of MCAo procedure are provided in Supplementary methods.

The body temperature was measured with a rectal probe and maintained at ≥ 36.5 ^oC using a feedback-controlled heating pad. Cerebral blood flow (CBF) was measured by laser Doppler flowmeter; animals that had ≥ 80% drop in CBF were included in the study. All mice were given subcutaneous injections of 0.9% sodium chloride twice a day for 7 days and were provided with wet mash in their cages. Body weight was recorded daily for the duration of the experiments.

To evaluate the effect of rGDF11 (PEPROTECH, 0.1 mg/kg, i.p.) in old mice, rGDF11 dissolved in phosphate-buffered saline was given 5 days after ischemia and continued for 5 days (group 3, and 4). For the vehicle-treated group, human IgG (R&D systems, 100 ug/kg, i.p.) was injected as described above. The dose of rGDF11 was chosen from earlier studies [7, 12]. BrdU (Sigma, 75 mg/kg, i.p.) was administered for 5 days starting at day 5 post-ischemia at an intraperitoneal site different from GDF11 injections.

The animals were randomized to treatment groups by block randomization. Investigators remained blinded to the treatment and group allocations during the experiments and data analysis. Only animals that reached the experimental endpoint (30 days) were included in the study. Animals that died before initiation of treatments were excluded (n=6). The reason for death was a hemorrhagic transformation as reported earlier [53]. Three mice died on day 1 post-MCAo, one mouse died on day 2 and two mice died on day 3.

Behavioral assessment

Neurological deficit score (NDS)

The NDS is a 5-point stroke severity scale [52]. NDS is scored from 0 to 4 with four being the highest level of deficit and 0 being no deficit detected.

Digigait testing

Digigait testing was performed and analyzed using the digigait ventral plane videography system (Mouse Specifics, Inc. Framingham, MA). The digigait ventral plane videography system consists of a transparent treadmill with a camera mounted beneath. Mice were placed on the treadmill briefly for 5-seconds before the treadmill was started at a speed of 1 cm/ second at an incline of zero degrees. The incline was kept consistent at zero degrees but the speed of the treadmill was increased by 1 cm/ second every 5-seconds until a final speed of 5 cm/ second was reached. The video camera mounted below recorded both the ability of the mouse to walk at the given speed and its gait. Once the test was completed, mice were placed in a recovery cage until all mice in their cage completed the task and then were returned to their home cage.

Tail suspension test

The tail suspension was performed as described in [54]. After acclimatization to the testing room for one-hour, the mice were suspended 60 cm above a table by their tail and recorded for 6 minutes using a digital camera. After the recording was completed, the mice were carefully removed and placed into a recovery cage away from its cage-mates until all had undergone testing after which they were returned to their home cages. The videos were then analyzed by a trained observer who was blinded to surgical and treatment groups who recorded the total time the mouse spent immobile. The testing apparatus was cleaned with 70% ethanol between mice.

Nest building score

The nest-building test was performed and analyzed as described earlier [31]. Mice were singly housed overnight in a cage on corncob bedding with one Nestlet, which is a cotton square, 2 inches X 2 inches (Lab Supply, Fort Worth, TX). Twelve hours later pictures were taken of the nests and scored using a 5-point scale. 1. The cotton square is untouched with over 90% still intact. 2. The cotton square is partially torn up but over 50% is still intact. 3. The cotton square is almost or completely torn up but no identifiable nest site is present (pieces scattered throughout the cage). 4. The cotton square is almost or completely torn up and there is an identifiable nest site but the sides of the nest are flat on more than 50% of its circumference. 5. The cotton square is almost or completely torn up and there is an identifiable nest site where the walls of the nest are higher than the body of the mouse on all sides.

Open field-testing

Locomotor activity was assessed as described previously at day 30 after MCAo [55]. Mice were placed in a brightly lit box, 50cm wide x 50cm length x 38 height and allowed to explore freely for 10 minutes. The mice were filmed from above throughout the test. These videos were then analyzed using Noldus Ethovision behavior software (Leesburg, VA). Between mice, the boxes were cleaned with 70% ethanol.

MRI testing

Mice were scanned on a Bruker 9.4T/Avance, 20 cm bore MRI system outfitted with microgradients and Paravision 5.1 software. Brain images were collected using a TURBORARE imaging sequence with a TR = 2500, a TE = 36 and NA = 2. The field of view was 3 cm and the slice thickness was 500 m and the matrix was set at 256 x 256. We utilized OsiriX^® MD.

Western blot analysis

Mice were euthanized, transcardially perfused with 60mL cold sterile phosphate-buffered saline (PBS) and the brains were harvested. The olfactory bulb, brainstem, and cerebellum were dissected as described previously [56] and the cortex was removed for further process. The mouse brain sections were homogenized using lysis buffer (1%NP-40, 1mM phenylmethylsulfonyl fluoride, protease inhibitors; complete and miniphosStop tablets) and the supernatant was resolved on 4-20% gradient SDS –PAGE and transferred to polyvinylidene difluoride membrane. Protein concentration was confirmed by BCA Protein Assay Kit (Thermo Fisher Scientific Inc, Rockford IL). Blots were blocked with 5% bovine albumin serum for an hour and incubated with GDF11/8 (1:1000, Ab124721), and pSmad 2/3 (1:1000, D27F4) overnight at 4⁰C. Beta-actin was used as a loading control. Secondary antibodies (1:10,000) were diluted and the blots were incubated for 1 hour, and an ECL detection kit was used for signal detection.

ELISA and multiplex

The brains from young and older mice were collected, and ELISA was performed following the manufacturer’s instructions (MyBioSource; MBS2021645) to determine GDF11 levels. This assay kit has high sensitivity and specificity for GDF11. To assess cytokines in the brain, a multiplex (Bio-Plex Pro^TM mouse cytokine 9-Plex Assay #MD000000EL) was performed following the manufacturer’s instructions.

Assessment of brain tissue loss

After transcardial PBS and paraformaldehyde 4% perfusion, the brains were harvested from mice in group 3. The olfactory bulb, brainstem, and cerebellum were removed. The brain was post-fixed overnight in 4% PFA and placed in 30% sucrose solution for 48 hours before processing. The brains were then cut into 30-μm sections on a freezing microtome and every eighth slice was stained by cresyl violet to visualize tissue loss [22562356]. The slices were photographed, tissue atrophy was analyzed using computer software (Sigma scan Pro5) as previously described [57]. Tissue atrophy percentage was calculated by using the following formula: percentage tissue atrophy = (total ipsilateral tissue/total contralateral tissue) × 100.

Assessment of CSF

After transcardial PBS and paraformaldehyde 4% perfusion, the brains were harvested from mice. The head was removed, skin, muscle, ears, nose tip and lower jaw were removed to expose the skull. The head was fixed in 4% PFA at 4⁰C. The head was then transferred to 40 ml of 0.01% sodium azide in PBS and rocked for 7 days at 4⁰C. The head was then transferred to a solution of 5mM gadopentetate dimeglumine (Bayer HealthCare Pharmaceuticals Inc., Wayne, NJ) and 0.01% sodium azide in PBS and rocked for 21 days at 4⁰C. We utilized OsiriX^® MD software to determine the areas of the ventricles and cerebrospinal fluid (CSF) of the mouse brains. Briefly, DICOM files of the acquired MRI data were imported into OsiriX. The whole brain was segmented as well as the regions of CSF and the ventricle areas ipsilateral and contralateral to the injury were identified and traced to obtain the CSF area.

Corpus callosum area analysis

Three consecutive sections were taken from the MRI images (+0.48, +0.72+ 1.92 mm from bregma). The corpus callosum (CC) was identified and quantification was performed by an investigator blinded to treatment groups. The area from contralateral and ipsilateral mouse brain was quantified using ImageJ. The CC area was normalized to contralateral CC. The entire analysis was repeated twice to ensure reproducibility of the results.

Immunohistochemistry (IHC) analysis

For estimation of gliosis, angiogenesis, and neurogenesis, mice in group 3 (n=5-6) were used. For the mouse brain, a standard procedure was utilized for IHC staining on 30-μm sections mounted on Fischer Scientific Superfrost Plus charged slides. Briefly, the tissue sections were rinsed in 0.1 M phosphate-buffered saline (PBS) pH 7.4. Antigen retrieval was performed by heating the tissue in a 10 mM sodium citrate buffer at pH 6.0. Tissue sections were incubated for 1 hour in blocking solution (0.1% Triton-X, 10% normal goat serum in 1X PBS) and then incubated overnight in primary antibody at 4°C (GDF11/8 1:250; IBA-1(Fujifilm Wako pure chemical corporation, NCNP24) 1:300; GFAP-Cy3 (Sigma-Aldrich C9205) 1:300; CD 31 (ab28364) 1: 100), DyLight 594 labelled Lycopersicon Esculentum (tomato) lectin (Vector Laboratories)1:100, BDNF (1:100), MBP (1:100, cell signaling technologies, 78896S), synaptophysin (1:300, ab14692).

BrdU staining

Following three washes with 0.1M PBS, pH 7.4 for 10 minutes, sections were placed in ice-cold 1N hydrochloric acid (HCl) for 10 minutes, followed by incubation with 2N HCl for 30 minutes at 37⁰C. After 30 minutes, the sections were rinsed three times with 0.1M 1X PBS and then sections were incubated with 0.1M boric acid for 10 minutes. The sections were washed three times and blocked with blocking solution and incubated with primary antibodies (Doublecortin (Cell signaling Technology, 4604S) 1:200 and anti-BrdU (ab6326) 1:25) overnight at 4⁰C as described previously. Following overnight incubation with primary antibodies, tissue sections were washed with PBST (0.1% tween 20 in 1X PBS) and then incubated with the appropriate conjugated secondary antibody (1:500, Molecular Probes, Invitrogen) for 2 hours at room temperature except for GFAP treated tissue. After washing in PBST, sections were coverslipped with mounting media (VECTASHIELD Antifade Mounting Medium with DAPI, Vector Laboratories).

Three coronal brain sections per mouse, taken 0.02, 0.45, and 0.98 mm from bregma, were stained and visualized for quantification at 20X magnification at the core/penumbra junction to quantify positive cells. Additionally, the expression of synaptophysin was quantified as described earlier [58]. Furthermore, immunoreactivity for MBP and GFAP was analyzed in CC and peri-infarct striatum [59]. The average numbers of cells visualized from three adjacent regions at the core/penumbra junction were recorded for each mouse and quantification was done using Image J software. For the quantification of vessel percentage area, a total number of junctions and total vessel length Angio Tool was used. To assess the number of DCX and BrdU positive neurons, 5 hippocampal sections, every fifth section (from bregma-1.46 mm to bregma-2.30 mm) were used. Three images were taken from each mouse brain and values were totaled. Similarly, three sections of SVZ per mouse brain were imaged and the values were totaled.

IHC for human tissue

All brain sections were from the mid-frontal cortex from the middle cerebral artery territory as determined by expert neuropathological assessments of H&E stained sections. The sections were from young controls [< 60 years=9, females (2) and males (7)], aged control [> 75 years, n=8, females (4) and males (4)]. Immunofluorescence staining was performed on the paraffin-embedded human brain tissue as described previously [60]. Immunohistochemically staining was performed by conventional deparaffinization of the human brains. Deparaffinized brain sections were incubated with 3% hydrogen peroxide for 15 minutes to reduce the background staining. Sections were washed three times with wash buffer (0.1% Triton-X in 1X PBS). The sections were blocked with blocking buffer (0.1% Triton-X, 10% fetal calf serum in 1X PBS) for an hour and incubated overnight with GDF11/8 primary antibody (1:200, Abcam) at 4⁰C. Samples were washed and subsequently incubated with Histofine (anti-rabbit secondary antibody fab fragment from Nichirei, Tokyo, Japan) for 30 min at RT. Immunoperoxidase staining was performed with the diaminobenzidine substrate kit (Nichirei, Tokyo, Japan) and counterstained with hematoxylin. For quantification, six images (20X and 40X) from the cortex of each human brain were taken on the Leica microscope using bright field and GDF11/8 immunoreactivity was quantified by scoring. Scoring was performed as follows: positively stained cells per section (40X) were scored as described, less than two cells = zero; three to five cells = 1; six to eight cells = 2; nine to twelve cells = 3; thirteen and above cells = 4. Data were analyzed by a blinded investigator.

Statistical analysis

Data are presented as mean ±SEM except for NDS, which was presented as median (interquartile range) and analyzed with the Mann-Whitney test. Grubbs test was used to analyses outliers. Two-group comparisons were analyzed by the Student t-test. Mortality was analyzed by generating Kaplan-Meier survival curves and performing a Mantel-Cox log-rank test. Statistical significance was set at p<0.05.