Thrombospondins 1 and 2 are Necessary for Synaptic Plasticity and Functional Recovery after Stroke

Abstract

Thrombospondins 1 and 2 (TSP-1/2) belong to a family of extracellular glycoproteins with angiostatic and synaptogenic properties. Although TSP-1/2 have been postulated to drive the resolution of postischemic angiogenesis, their role in synaptic and functional recovery is unknown. We investigated whether TSP-1/2 are necessary for synaptic and motor recovery after stroke. Focal ischemia was induced in 8- to 12-week-old wild-type (WT) and TSP-1/2 knockout (KO) mice by unilateral occlusion of the distal middle cerebral artery and the common carotid artery (CCA). Thrombospondins 1 and 2 increased after stroke, with both TSP-1 and TSP-2 colocalizing mostly to astrocytes. Wild-type and TSP-1/2 KO mice were compared in angiogenesis, synaptic density, axonal sprouting, infarct size, and functional recovery at different time points after stroke. Using the tongue protrusion test of motor function, we observed that TSP-1/2 KO mice exhibited significant deficit in their ability to recover function (P < 0.05) compared with WT mice. No differences were found in infarct size and blood vessel density between the two groups after stroke. However, TSP-1/2 KO mice exhibited significant synaptic density and axonal sprouting deficits. Deficiency of TSP-1/2 leads to impaired recovery after stroke mainly due to the role of these proteins in synapse formation and axonal outgrowth.

Introduction

New or recurrent stroke is a devastating disease affecting 700,000 people in North America alone every year (Thom et al, 2006). After a stroke, clinical deficits, such as motor paralysis, weakness, sensory numbness, and incoordination, are often observed in the upper and lower extremities contralateral to the injury. While the impaired limb could be permanently affected, significant recovery takes place during the first several weeks after the injury and lasts for at least 6 months, with the ultimate outcome, mainly a function of initial severity (Duncan, Lai and Keighley, 2000). Although part of the recovery process most likely involves resolution of pathophysiologic events caused by ischemic brain injury, many studies have provided substantial evidence that regenerative processes contribute toward the restitution of function (Nudo, 2007). Such processes include dendritic arborization among neurons spared by ischemic injury, the reorganization of existing synaptic structures, and regeneration of synapses.

Thrombospondins (TSPs) are large extracellular matrix proteins that mediate cell-cell and cell-matrix interactions by communicating with an array of membrane receptors, other extracellular matrix proteins, and cytokines (Adams and Lawler, 2004). There are 5 known TSPs, each encoded by a separate gene. Thrombospondin-1 and 2 have been shown to be involved in platelet aggregation, inflammation, and angiogenesis (Adams and Lawler, 2004). After ischemic injury in rats, TSP-1 and TSP-2 are upregulated in the affected area, and this robust expression was postulated to drive the resolution of postischemic angiogenesis through their antiangiogenic properties (Lin et al, 2003). Recently, however, TSP-1 and TSP-2 have been found to promote synaptogenesis in vitro and in vivo, and their deficiency results in reduced synaptic density during development (Christopherson et al, 2005). Therefore, the observations that TSPs are involved in cell—cell interactions and synaptogenesis raise the possibility that an increase in TSP expression after ischemic injury could relate to the subsequent neuroanatomical remodeling. In this study, we report that TSP-1 and TSP-2 contribute to the synaptic plasticity and axonal sprouting that underlie the recovery of motor function after ischemic injury.

Materials and methods

Surgical Methods

Experimental protocols were approved by the Stanford University Administrative Panel on Laboratory Animal Care. Eight to 12–week-old FVB wild-type (WT) and FVB mice with homozygous deletion of the TSP-1 and TSP-2 genes (TSP-1/2 KO; Agah et al, 2002) were used. To detect TSP-2 expression by proxy, a transgenic GENESAT line (Gong et al, 2003) that expresses green fluorescent protein from the TSP-2 promoter was used. The distal middle cerebral artery (dMCA) was exposed and cauterized above the rhinal fissure, and the ipsilateral common carotid artery (CCA) was permanently occluded with suture. Sham-operated animals only had the dMCA and the CCA exposed. Animals were given randomized identification numbers to perform surgeries in a blind fashion and it was impossible to visually distinguish the two groups apart. Triphenyltetrazolium chloride (TTC) staining was used to quantify infarct size 24 h after stroke (Joshi, Jain and Murthy, 2004).

RNA Extraction and qPCR Analysis

After permanent unilateral occlusion of the dMCA and the CCA, WT FVB mice were killed at days 1, 7, 14, and 28 after stroke (n = 4 per time point). Total RNA was extracted from the core of the ischemic area and from the contralesional homotopic cortex following a standard protocol (RNeasy Mini Kit, Qiagen). On-column DNase digestion was performed to remove genomic DNA contamination (RNase-Free-DNase Set, Qiagen, Valencia, CA, USA). A total of 500 ng RNA was reverse-transcribed with oligo-dT primers to generate cDNA (SuperScript III Reverse Transcriptase, Invitrogen, Carlsbad, CA, USA). To generate concentration standards, a small amount of cDNA from all samples was pooled and serially diluted from 1:10 to 1:100,000. Each individual sample was diluted at 1:100 with RNase/DNase-free ultrapure water and used as template in subsequent PCR reactions.

Primers for quantitative real-time PCR (qPCR) was obtained from PrimerBank (Wang and Seed, 2003) and synthesized (Invitrogen, Carlsbad, CA, USA). They were screened for the generation of a single amplification product using gel electrophoresis, for PCR efficiency using standard curve analysis, and for the absence of primer—dimer formation using melt-curve analysis (Halstenberg et al, 2005). The primers for TSP-1 (GenBank accession number: NM 011580) are as follows: 5‘-GAAG CAACAAGTGGTGTCAGT-3’ and 5‘-ACAGTCTATGTA GAGTTGAGCCC-3’, which generate a 106-bp amplicon (PrimerBank ID: 7110725a3). The primers for TSP-2 (GenBank accession number: NM 011581) are as follows: 5‘-CCTCAACTACTGGGTAGAAGGC-3’ and 5‘- TGACAC TGTCGATAAGATCGCA-3’, which generate a 146-bp amplicon (PrimerBank ID: 6755779a3). The primers for β-actin (GenBank accession Number NM007393) are as follows: 5‘-AGTGTGACGTTGACATCCGTA-3’ and 5‘-GC CAGAGCAGTAATCTCCTTCT-3’, which generate a 112-bp amplicon (PrimerBank ID: 6671509a3). β-Actin was used to control for sample variability in total cDNA content and reverse transcription efficiency (Halstenberg et al, 2005).

Quantitative polymerase chain reactions were run in the Mx3000P QPCR System (Stratagene, La Jolla, CA, USA) using SYBR Green chemistry (Brilliant SYBR Green QPCR Master Mix, Stratagene). All reactions were performed in triplicate using the following conditions: 10 mins at 95°C; then 40 cycles of 30 secs at 95°C, 1 min at 60°C, and 30 secs at 72°C. This was followed by melting curve analysis using a temperature range from 55°C to 95°C.

Mx3000P software was used to calculate the threshold cycle for each reaction, and the average threshold cycle (C_T) value was determined from the triplicate of each sample. Determination of the slope of the serially diluted standard curve provided the value of the PCR efficiency according to the equation: E = 10^−1/slope (Pfaffl, 2001). The TSP-1 and TSP-2 expression value relative to β-actin were determined for each sample based on the following equation: X = R*E(β-actin)^[C_T(β-actin)]/E[TSP)^ [C_T(TSP)]*100R (Pfaffl, 2001; Halstenberg et al, 2005), which corrects for the different lengths of the amplicon where R indicates length (amplicon of β-actin)/length (amplicon of TSP).

Tissue Preparation

On days 1, 7, 14 (WT, n = 4 per time point), and 28 (WT, n = 10; TSP-1/2 KO, n = 8) after induction of ischemia, mice were anesthetized and perfused transcardially with 300 mL of ice cold phosphate-buffered saline (PBS, pH 7.4), followed by 300 mL of 4% paraformaldehyde in PBS (pH 7.4). The brains were removed, postfixed by immersion in 4% paraformaldehyde in PBS (1 day, 4°C), and then with 20% sucrose, 4% paraformaldehyde in PBS (1 day, 4°C). Sham brains were also prepared as described above (WT, n = 2 per time point and TSP-1/2 KO, n = 3 at day 28). For immunohistochemical analysis, coronal sections of 25 μm thickness were sectioned with a cryostat and collected in antifreeze solutions containing 30% ethylene glycol and 30% glycerol.

Immunohistochemistry

Free-floating slices of 25 μm thick were washed three times in PBS, and blocked in PBS buffer containing 3% donkey serum (Jackson ImmunoResearch, West Grove, PA, USA) and 0.1% Triton X-100 (Sigma, St Louis, MO, USA) for 1 h. For synaptic staining, primary antibody solution was added consisting of mouse anti-synaptophysin (1:500, Sigma), rabbit anti-PSD-95 (1:250, Zymed Laboratories, South San Francisco, CA, USA), and anti-glial fibrillary acidic protein (GFAP) (1:500, Advanced Immunochemical Inc., Long Beach, CA, USA). Slices were incubated in primary antibody for three nights at 4°C, washed three times in PBS and 0.6% Triton X-100, and incubated with Cy3-conjugated donkey anti-mouse (1:500), FITC-conjugated donkey anti-rabbit (1:500), and Cy5 donkey anti-guinea pig (1:500) (Jackson ImmunoResearch, West Grove, PA, USA). After incubation in secondary antibodies for 3 h, sections were washed, incubated with 4′-6-diamidino-2-phenylindole for 5 mins, and mounted onto slides. To localize TSP expression, sections were stained with either a mouse monoclonal TSP-1 antibody (sc-59887, 1:200; Santa Cruz Biotechnology, Santa Cruz, CA, USA) or a mouse monoclonal TSP-2 antibody (sc-12313, 1:200; Santa Cruz Biotechnology). To verify TSP-2 antibody specificity, TSP-2 expression was also examined in a GENESAT transgenic mouse line that expresses green fluorescent protein under the TSP-2 promoter (Gong et al, 2003). Colocalization experiments for TSP expression were carried out with S100 (1:200, Novus Biologicals, Littleton, CO, USA) and GFAP (1:200, Advanced Immunochemical, Inc.).

Confocal Analysis of Synapse Number

Images of immunostained brains were collected on a Leica LSM 510 Meta confocal microscope. Optical sections were line-averaged and collected at 0.28 μm intervals from the penumbral and homotopic contralateral cortices. Gain thresholds and amplitude offsets were kept constant between imaging of the penumbral and homotopic contralateral areas for each section. Stacks of 10 optical sections were quantified for synapse number by projecting a series of 5 optical sections, and counting the number of synapses in each projection volume. Synaptic puncta were automatically counted using the Image J puncta analyzer program in TSP-1/2 KO mice (n = 8) and WT mice (n = 10) 4 weeks after unilateral occlusion of the dMCA and the CCA. On average, two stacks were obtained per ipsilesional and contralesional hemisphere for each section. For each mouse, 4 to 5 sections, which are 400 μm apart on average, were sampled in the ischemic area. For time course experiments, brains from each time point (days 1, 7, 14, 21, and 28) were sectioned and scanned to image the penumbral and homotopic contralateral cortical regions. To compare synaptic density between WT and TSP-1/2 KO mice, we obtained the ratio of synaptic puncta on the ipsilesional side to that on the contralesional side. This ratio corrects for differences in synaptic density between TSP-1/2 KO and WT mice.

Blood Vessel Density Quantification

Four weeks after unilateral occlusion of the dMCA and the CCA, WT (n = 6) and TSP-1/2 KO animals (n = 5) were perfused and 25 μm coronal tissue sections were stained with tomato lectin conjugated to streptavidin and followed with 3,3’-diaminobenzidine immunohistochemistry (Vector Labs, Burlingame, CA, USA). The number of vessel branch points was counted using the optical fractionator and branch point density was determined (Stereo Investigator, Williston, ME, USA). Branch points were counted in the penumbra area of the ischemic cortex and in the homotopic area of the contralateral cortex at × 40 magnification. Comparisons between different animals were made by dividing the vessel density in the ipsilesional cortex with the contralesional cortex to obtain a ratio. This ratio was used to correct for any endogenous differences in branch point density between WT and TSP-1/2 KO animals.

Biotinylated Dextran Amine Tracing of Axonal Sprouting

Four weeks after unilateral occlusion of the dMCA and the CCA, WT (n = 5) and TSP-1/2 KO animals (n = 3) were administered intracerebral injections of biotinylated dextran amine (BDA) near the contralesional motor cortex to investigate axonal sprouting. Each animal received 0.4 μL of a 10% solution of BDA-10000 (Invitrogen) in the motor and somatosensory cortex of adult mice, using stereotaxic coordinates relative to bregma: (site 1) AP + 1.1 ML + 1.2 DV −0.7, (site 2) AP + 1.3 ML + 1.3 DV −0.7, and (site 3) AP + 1.5 ML + 1.4 DV −0.7 (Chen, Schaevitz and McConnell, 2005). Sham mice (WT and TSP-1/2 KO, n = 2 per group) were also injected in identical coordinates. Seven days later, the animals were perfused and 25 μm coronal tissue sections were stained with Texas Red Streptavidin (Jackson ImmunoResearch). Although BDA is a bidirectional label, we only quantified the density of labeled axons in the cortico—cortico and cortico—striatal projections (Carmichael and Chesselet, 2002). Immunofluorescence density was quantified in the ipsilesional penumbra, striatal, and the central commissure of the corpus callosum region using the Image J Analyzer program. Density of labeled fibers in the penumbra and striatal regions was normalized to that in the corpus callosum to control for the extent of BDA labeling in each brain.

Behavioral Assay

To compare the rate of functional recovery after unilateral occlusion of the dMCA and the CCA between WT (n = 10) and TSP-1/2 KO (n = 8) animals, we used four different behavioral tests, namely accelerating rotarod, tongue protrusion (Gulyaeva et al, 2003), corner test, and cylinder test (Li et al, 2004). For all four behavioral tests, baseline performance before stroke was recorded. For the tongue protrusion, corner test, and cylinder test, animals were assessed for 3 consecutive days before stroke, and baseline performance was the average of these 3 days. All experiments were performed in a blind fashion. After stroke, performance was assessed in both groups at postischemia days 1, 7, 14, 21, and 28. For the rotarod test, animals were trained for three trials over 9 consecutive days with baseline performance established over the final 3 days before stroke. For testing, the rotarod was set to accelerate from 5 to 40 r.p.m. over the course of 5 mins, and the latency to complete two consecutive rolls, or to drop off the rod (performance) was recorded. For the tongue protrusion test, animals were placed in a rectangular plexiglass animal cage with a tongue protrusion apparatus consisting of a plexiglass tube (inner diameter 2 mm, length 15 mm) filled to the top with peanut butter and mounted upright on a petri dish. Animals were placed with the tongues protrusion apparatus for 12 h in a noise-free isolated environment. Tongue protrusion was measured as the distance between the open edge of the plexiglass tube and the lowest level of remaining peanut butter.

Statistical Analysis

Data analysis was performed with the statistic software package PRISM (GraphPad, San Diego, CA, USA). All means are presented ± s.e.m. All statistical analyses were performed with the use of analysis of variance followed by post hoc t-tests with the Bonferroni correction. A value of P < 0.05 was considered significant.

Results

Thrombospondin-1/2 are Upregulated in the Ischemic Penumbra

Previous studies have shown that TSP-1/2 are upregulated after stroke in rats (Lin et al, 2003). To verify whether a similar upregulation was present in our mouse stroke model, we investigated the levels of TSP gene expression with qPCR. Polymerase chain reaction amplification of TSP-1 or TSP-2 each produced a single cDNA fragment when visualized with ethidium bromide on agarose gel and qPCR melt-curve analysis resulted in a single melting temperature, thus confirming the specificity of the primer pairs used. Compared with the expression level of sham-operated animals (n = 3), TSP-1 expression increased at day 1 by 5.3-fold in the ischemic cortex (P < 0.05, n = 4) and declined thereafter (Figure 1A). Thrombospondin-2 expression increased at day 7 by 4.1-fold in the ischemic cortex (P < 0.05, n = 4) and remained elevated at day 14 by 2.7-fold (P < 0.05, n = 4) (Figure 1B). Immunohistochemical staining of TSP-1 showed an increase in TSP-1 protein expression in the core and penumbra of the infarct at day 1 after unilateral occlusion of the dMCA and the CCA (Figure 1C). A majority of cells expressing TSP-1 also coexpressed the astrocytic marker S-100. Thrombospondin-2 immunoreactivity and TSP-2 expression in the transgenic mouse line, which expresses green fluorescent protein under the TSP-2 promotor (Gong et al, 2003) could be visualizable in the ischemic core and penumbra at day 7 after ischemia. A large proportion of cells expressing TSP-2 also coexpressed the astrocytic marker GFAP at 7 days after unilateral occlusion of the dMCA and the CCA (Figure 1D). These findings validate that TSP levels increase after stroke in a mouse model of ischemia, and are consistent with previous reports of TSP-1/2 expression after ischemia in rats (Lin et al, 2003).

Figure 1

Thrombospondin-1 and 2 mRNA and protein expression are increased after stroke. (A and B) qPCR analysis shows the upregulation of mRNA expression of TSP-1 at day 1 and TSP-2 at 1 and 2 weeks after focal ischemia in the ipsilesional ischemic cortex (n = 4 for each time point). *P < 0.05 versus sham-operated controls. No significant changes were noted on the contralateral side. (C) Representative panel showing immunohistochemical staining for TSP-1 (green) colocalized with the astrocytic marker S100 (red) in the ischemic cortex at day 1 after unilateral occlusion of the dMCA and the CCA. (D) Representative panel showing immunohistochemical staining for TSP-2 (green) colocalized with the astrocytic marker GFAP (red) in the ischemic cortex at day 7 after unilateral occlusion of the dMCA and the CCA. DNA = blue. Scale bar = 20 μm.

Loss of Thrombospondin-1/2 does not Affect Blood Vessel Density after Cerebral Ischemia

Given the TSP's role as angiostatic factors, we investigated whether microvascular density after unilateral occlusion of the dMCA and the CCA would be affected by the loss of TSP-1/2. We compared microvascular densities at day 28 between TSP-1/2 KO and WT mice. The ratio of blood vessel density in the ischemic penumbra relative to that in the contralateral cortex in TSP-1/2 KO mice did not differ significantly from that found in WT mice. The average normalized blood vessel density was found to be 0.8149 ± 0.06 for WT and 0.8357 ± 0.03 for TSP-1/2 KO (P = 0.93) (Figures 2A and 2B).

Figure 2

Thrombospondin-1/2 KO mice do not show deficit in blood vessel density after stroke. (A) Representative images of lectin-stained blood vessels in the ischemic penumbra of wild-type (WT) (left panel) and TSP-1/2 KO (right panel) mice. (B) Blood vessel counts in ischemic penumbra normalized to counts in homotopic contralateral cortex 28 days after unilateral occlusion of the dMCA and the CCA of WT (n = 6) and KO (n = 5) mice. Quantitative analysis indicates no significant differences between penumbral blood vessels density in WT and TSP-1/2 KO mice (P = 0.93). Scale bar = 60 μm.

Synaptic Density Recovers Gradually in the Ischemic Penumbra after Stroke

To determine whether there are synaptic changes in WT mice following our model of stroke, synaptic density in the ischemic penumbra was examined at 1, 7, 14, 21, and 28 days after stroke in WT mice. Regions with interspersed GFAP-positive reactive astrocytes bordering a GFAP-negative core were identified as ischemic penumbra. Synaptic puncta, defined by a colocalization of presynaptic synaptophysin and postsynaptic PSD-95, were quantified. The ratio of synapses in the ischemic penumbra and the corresponding nonischemic contralateral cortex was calculated to correct for any endogenous differences. We defined this ratio as the synaptic density in the ischemic penumbra. After unilateral occlusion of the dMCA and the CCA, the synaptic density at day 7 was 68%, which is significantly lower than baseline at nearly 100% (P < 0.05). Synaptic density then began to increase and by day 28 reached 95.2%, which did not differ from baseline (P = 0.8) (Figure 3A). These data suggest that after focal ischemia, there is an initial period of synaptic disruption followed by a gradual recovery of synapse number.

Figure 3

Thrombospondin-1/2 KO mice show impaired synaptic recovery after stroke. (A) Synaptic puncta in the ischemic penumbra normalized to the homotopiccontralesional cortex quantified at days 1, 7, 14, 21, and 28 after unilateral occlusion of the dMCA and the CCA of wild-type (WT) mice show initial loss of synapse followed by gradual recovery (n = 4 to 6). (B) Synaptic puncta comparison at 28 days after unilateral occlusion of the dMCA and the CCA in WT (n = 10) and TSP-1/2 KO per time-point (n = 8) mice show lower synaptic density in KO mice. Wild-type mice have a mean of 95.2 ± 8.2% of synapses whereas TSP-1/2 KO mice have a mean of 71.1 ± 6.2%. *P<0.05. (C) A schematic diagram showing the regions where synaptic puncta were quantified. Gray area indicates the ischemic core. Region 1 indicates the ischemic penumbra whereas region 2 indicates the contralesional homotopic cortex. The ratio of synaptic puncta in region 1 to region 2 in the same section provides the final measured value. (D) Representative images showing synaptic puncta in the ischemic penumbra and contralesional cortex of WT and TSP-1/2 KO mice. SYN indicates the presynaptic protein synaptophysin (red) and PSD-95 is the postsynaptic protein (green), whereas yellow color shows colocalization of the pre and postsynaptic protein. Scale bar = 20 μm.

Loss of Thrombospondin-1/2 Impairs Synaptic Recovery after Stroke

Because TSP-1/2 have been shown to mediate synaptogenesis during development, we investigated the effect of a TSP-1/2 deficiency on the recovery of synapses after stroke. We quantified synapses in the ischemic penumbra and compared synaptic density in the ischemic penumbra between WT and TSP-1/2 KO mice 4 weeks after stroke. Because TSP-1/2 KO mice had been shown to have fewer synapses compared with WT mice (Christopherson et al, 2005), we corrected for this difference by defining a normalized synaptic density as the ratio of synaptic density in the ischemic penumbra to that in a homotopic area of the contralesional cortex. In TSP-1/2 KO mice, synaptic density in the ischemic penumbra was significantly reduced compared with that in WT mice at day 28 after stroke. Wild-type mice had a mean of 95.2 ± 8.2% of synapses whereas TSP-1/2 KO mice had a mean of 71.1 ± 6.2% (P < 0.05) (Figure 3B). To assess whether synaptic density changes in the non-infarcted cortex could affect the normalized synaptic density ratio after stroke, we compared synaptic counts obtained in the non-infarcted cortex with shams in both groups. No significant differences in synaptic density were seen on the non-infarcted cortex at day 28 as compared to sham in both WT and TSP-1/2 KO mice (data not shown). This reduction in synaptic density of TSP-1/2 KO mice suggests that there is a reduction or a delayed recovery of synapses in the ischemic penumbra of TSP-1/2 KO mice. Thrombospondin-1/2, therefore, may play a crucial role in mediating synaptic recovery after ischemic damage to the brain.

Loss of Thrombospondin-1/2 Impairs Axonal Sprouting after Stroke

It had been previously shown that following stroke, there is an increase in axonal sprouting from the contralesional cortex toward the ischemic penumbra and striatum on the other side of the brain (Carmichael and Chesselet, 2002). To determine whether TSP-1/2 KO mice exhibited a deficit in axonal sprouting, we compared the density of axonal projections from the contralesional cortex found in the ischemic penumbra at 5 weeks after stroke. We injected BDA, an anterograde axonal tracer, into a homotopic area in the contralesional cortex and quantified the density of BDA-labeled axons in the ipsilesional ischemic penumbra and striatum. Although the infarct produced by this model did not extend into the striatum, we wanted to examine axonal sprouting in the striatum because a large portion of contralateral input projects in both the ipsilesional penumbra and striatum (Carmichael and Chesselet, 2002). To control for the efficiency of BDA uptake, we normalized the density of BDA-labeled axons in the ipsilesional penumbra and striatum to that of decussating axons in the corpus callosum (Figure 4C). This was defined as the axonal sprouting density. Compared with WT mice (n = 5), the axonal sprouting density was significantly lower in the ischemic cortical penumbra, but not the striatal region, of TSP-1/2 KO mice (n = 3) at 5 weeks after stroke. In the ischemic penumbra, the axonal sprouting density was 0.53 in WT and 0.26 in TSP-1/2 KO mice; in the ipsilesional striatum, the axonal sprouting density was 0.47 in WT and 0.27 in TSP-1/2 KO mice (Figure 4A). These observations suggest that TSP-1/2 may play a role in promoting axonal sprouting in the ischemic penumbra from inputs originating in the contralateral cortex.

Figure 4

Thrombospondin-1/2 KO mice show impaired axonal sprouting from the contralesional cortex toward the ischemic penumbra after stroke. (A) Greater BDA-labeled axonal sprouting from contralateral inputs was quantified in the cortical penumbra and striatum of wild-type (WT) mice compared with TSP-1/2 KO mice. Axonal sprouting density is defined as the ratio of the density of BDA-labeled axons in the ischemic penumbra or striatum normalized to that of decussating axons in the corpus callosum. *P < 0.05 (B) A schematic diagram showing the regions where axonal density was quantified. Gray area indicates the ischemic core. Red oval indicates the site of BDA injection. Region 1 indicates the ischemic cortical penumbra and region 2 indicates the ischemic striatum. Region 3 indicates the central commissure of the corpus callosum. The ratio of axonal density in region 1 to region 3 and in region 2 to region 3 provides the final measured value for the cortical penumbra and striatum, respectively. (C) Representative images of BDA-labeled axons in the cortical penumbra and striatum of WT and TSP-1/2 KO mice. Scale bars = 50 μm.

Loss of Thrombospondin-1/2 Impairs Behavioral Recovery after Stroke

Previous studies have postulated that functional recovery may involve a gradual regeneration and reorganization of synaptic structures (Ito et al, 2006; Nudo, 2007). To determine whether TSP-1/2 play a role in functional recovery, we compared performance in various behavioral tests between WT (n = 10) and TSP-1/2 KO (n = 8) mice at days 1, 7, 14, 21, and 28 after stroke. No functional deficits compared with baseline performance could be found in limb-use asymmetry (data not shown), corner turning test (data not shown), or rotarod performance (Figure 5B) after stroke. However, acute functional deficit compared with baseline performance was observed in both WT and TSP-1/2 KO groups at 1 day after stroke in the tongue protrusion test (WT = 69.6%, KO = 68.7%, P < 0.01) (Figure 5A). Additionally, WT performance recovered to baseline at day 28 whereas performance of the TSP-1/2 KO group failed to reach baseline performance by day 28. Although both groups had similar baseline tongue protrusion ability before stroke, and exhibited similar deficits on days 1 and 7 after ischemia, tongue protrusion significantly differed between WT and TSP-1/2 KO animals on day 21 (P < 0.05) and day 28 (P < 0.05). These results indicate that the WT group exhibited significantly greater functional recovery than the TSP-1/2 KO group and suggest that the loss of TSP-1/2 interfered with full motor recovery of tongue function.

Figure 5

Thrombospondin-1/2 KO mice exhibit deficits in functional recovery after stroke in the tongue protrusion test. Performance on the tongue protrusion test (A) and accelerated rotarod (B) over 28 days after cerebral focal ischemia in wild-type (WT) (n = 10) and TSP-1/2 KO (n = 8) mice. *Indicates significant differences (P < 0.05) between performance of WT and KO groups. (C) A plot of infarct size measured by TTC staining (WT, n = 5; KO, n = 4) shows no significant difference between the two groups. (D) Representative sections stained with TTC to show infarct size (white area) from WT and TSP-1/2 KO groups.

It is possible that the differences in behavioral recovery could be attributed to differences in the severity of tissue infarction after unilateral occlusion of the dMCA and the CCA. To investigate this possibility, we compared the infarct volumes 2 days after unilateral occlusion of the dMCA and the CCA in both groups of mice. No differences were found in infarct volume between WT mice (n = 4) and TSP-1/2 KO mice (n = 3, P = 0.20) (Figure 5C) by TTC staining (Figure 5D). In addition to TTC staining, we performed cresyl violet staining on sections from WT- and TSP-1/2 KO-stroked brains and did not detect subcortical changes; similar results were obtained with staining for the neuronal marker NeuN; damage was restricted to the cortex (data not shown). On average, infarct size was 32.9% of the hemisphere for WT mice and 35.4% for TSP-1/2 KO mice. Moreover, overall appetite as measured by changes in weight relative to presurgical values did not differ between the two groups (data not shown). Taken together, these results suggest that TSP-1/2 play an important role in motor recovery after stroke.

Discussion

In the present study, we showed that the rate of synaptic recovery and axonal sprouting was significantly decreased in the ischemic penumbra of TSP-1/2-deficient mice, and this deficiency in structural plasticity correlated with the recovery of tongue protrusion, a test of behavioral recovery. This correspondence in synaptic reorganization and axonal sprouting with behavioral recovery suggests that there exists a TSP-1/2-mediated process of structural plasticity, which is required for postischemic motor recovery. Although other studies have shown the role of TSP in wound healing (Agah et al, 2002), nerve outgrowth (Hoffman and O'Shea, 1999), and synaptogenesis during central nervous system development (Christopherson et al, 2005), this is the first study to show that the loss of TSP impairs motor recovery after focal cerebral ischemia in the adult animal. Consistent with the observations of Christopherson et al (2005) during development, we did not observe a difference between WT and TSP-1/2 KO mice in the levels of two synaptic proteins, synaptophysin and PSD-95, by western blots 4 weeks after stroke (data not shown). This supports the hypothesis that TSPs are needed for inducing and maintaining synaptic alignment and adherence of pre and postsynaptic specializations during synaptogenesis rather than enhancing the expression of synaptic proteins.

Cortical injury that occurs after stroke has been found to impair the execution of muscular contraction and affect fine manipulative abilities, such as lingual coordination. In the weeks and months after injury, a gradual recovery of some motor abilities is observed and the extent of this process is determined by both the severity of the injury and the endogenous plasticity that facilitates the process (Nudo, 2007). To assess the extent of functional recovery in a focal model of ischemia in mice, we used an extensive battery of behavioral tasks sensitive to motor disturbance after ischemia. To facilitate the quantification of synapses in the ischemic penumbra, we used a focal model of ischemia that results in a distinct border between ischemic and nonischemic regions. This strategy also prevents the complication of contralesional hypoxia, which would affect axonal sprouting from the contralesional cortex. The drawback of such a strategy is that the infarct area produced by experimental stroke is relatively small, which may fail to result in observable behavioral deficits. In our model, the dMCA and the CCA were permanently occluded. It is possible that reperfusion injury may play a role in causing more observable behavioral deficits. Indeed, in three of the four behavioral tests we used, including the accelerating rotarod test, the corner test, and the cylinder test, we could not detect (as a group) a deficit in performance at either 1 day or longer after stroke. However, using the tongue protrusion test, we observed a sustained deficit lasting several weeks. We attribute the success of the tongue protrusion test to a relatively small lesion necessary to induce a deficit in tongue function, as compared with the larger lesion required to affect gross motor function in the rotarod, corner, and cylinder tests. In an ischemia model of rat, the tongue protrusion test has been shown to correlate with infarct volume and neurologic scores (Gulyaeva et al, 2003). In addition, lingual discoordination such as tongue palsy and dysarthria has often been observed following stroke in human patients (Daniels, Brailey and Foundas, 1999). Importantly, we were able to show that TSP-1 and TSP-2 KO mice exhibited a delayed rate of recovery relative to WT mice, suggesting that TSP-1/2 may mediate molecular processes required for the recovery of motor function.

Spontaneous functional recovery in stroke patients has been associated with the remapping of cortical function and the redistribution of neuronal activity (Nudo, 2007). This cortical plasticity is attributable to various mechanisms, such as enhanced dendritic growth, synaptogenesis, axonal sprouting, and the increased production of growth factors. Here we show that the loss of TSP-1/2 results in a significant impairment of synaptogenesis and axonal sprouting in the ischemic penumbra after stroke. During development, it has been shown that TSP-1/2 are secreted by astrocytes and the absence of TSP-1/2 results in the reduction of synapses in the brain (Christopherson et al, 2005). After ischemia, TSP-1/2 protein expression has been shown to colocalize with markers for astrocytes and endothelial cells (Lin et al, 2003). We found that TSPs are, in part, secreted by astrocytes after stroke, as evident by the colocalization of TSP-1 and TSP-2 with S100 and GFAP, respectively, and thus may play a role in regenerative processes. The temporal pattern of TSP-1/2 mRNA upregulation after ischemia also correlates with the gradual pattern of nerve growth-promoting gene expression seen after stroke, which stabilizes around week 4 (Carmichael et al, 2005). In our model, TSP-1 mRNA expression increased at day 1 after stroke, whereas TSP-2 mRNA expression increased at weeks 1 and 2. This pattern of gene expression also correlates with the temporal profile of synaptic recovery that we observed, namely a decrease in synaptic density at days 1 and 7 after ischemia, followed by gradual recovery over the next several weeks. This result is consistent with electron microscope synaptic profiling performed after stroke in gerbil brains (Ito et al, 2006).

In addition, TSP expression has been observed to dramatically increase along the path of regenerating axons after optic nerve crush in the goldfish (Hoffman and O'Shea, 1999). Thrombospondin has been shown to promote neurite outgrowth (Osterhout, Frazier and Higgins, 1992) and may help stabilize the extracellular matrix by interacting with other neurite-promoting factors, such as laminin and fibronectin (Lahav, Lawler and Gimbrone, 1984; Mumby, Raugi and Bornstein, 1984). In cerebral ischemia, the axonal sprouting response from the contralesional cortex to the ischemic penumbra and ipsilesional striatum is characterized by spontaneous, synchronized waves of neuronal discharges in the peri-infarct and contralateral cortical areas and by the upregulation of a number of axonal growth-promoting genes (Carmichael, 2003, 2005). Because we have shown that TSP-1/2 KO animals are deficient in axonal sprouting and have a reduction in synapses, it is conceivable that one of the mechanisms contributing to impaired synaptic density is a reduction of axonal sprouting from contralesional inputs. Thrombospondin-1/2 may interact with matrix proteins and facilitate axonal sprouting, which may eventually result in synaptic and behavioral recovery.

It is well known that angiogenesis often takes place in the ischemic penumbra after cerebral ischemia (Hayashi et al, 2003), and the number of new vessels and thus the extent of angiogenesis are correlated with longer survival in patients with stroke (Krupinski et al, 1994). Both TSP-1 and TSP-2 have been shown to exert antiangiogenic activities (Streit et al, 1999a, 1999b), and their upregulation after ischemia in rats has been postulated to regulate angiogenesis (Lin et al, 2003). The present study shows that there was no significant difference in blood vessel density between WT and TSP-1/2 KO mice at day 28 after ischemia. However, it is still possible that TSP-1/2, by serving as angiostatic factors, may affect the initial rate of postischemic angiogenesis but may not affect the final extent of angiogenic regrowth. Nevertheless, in a skin excision wound model of injury, it was found that while TSP-2 knockout mice showed an increased vascular density, neither TSP-1 knockout nor TSP-1 and TSP-2 double knockout mice showed a difference in vessel density compared with WT mice at days 7 and 14 after injury (Agah et al, 2002). Our finding that there was no significant difference in blood vessel density at 4 weeks after cerebral ischemic injury parallels the findings in the skin excision wound model. Moreover, there has been no direct experimental evidence supporting the idea that angiogenesis and its sequelae lessen tissue damage (Hayashi et al, 2003). This finding lends further credence to the importance of neural plasticity in mediating postischemic motor recovery.

Many agents that induce cortical plasticity have been shown to promote functional recovery after ischemia. For example, rats subjected to permanent ischemia and treated with d-amphetamine showed a significantly higher number of synaptic density and improved recovery of forelimb use (Stroemer, Kent and Hulsebosch, 1998). Moreover, administration of a neutralizing antibody to Nogo, a potent inhibitor of neurite outgrowth, enhanced motor cortico-efferent plasticity and recovery of function in rats (Seymour et al, 2005). Furthermore, administration of nicotine resulted in improved functional recovery and increased dendritic length and branching in pyramidal cells of the forelimb and cingulate areas after stroke (Gonzalez, Gharbawie and Kolb, 2006). Our findings that the loss of TSP-1/2 results in impaired synaptogenesis, axonal sprouting, and motor recovery further support the importance of cortical plasticity in functional recovery after stroke. The results here thus suggest that therapies aimed at enhancing the rewiring of the nervous system and promoting synaptogenesis may prove beneficial in restoring function in stroke patients.