Manganese-Enhanced MRI of Brain Plasticity in Relation to Functional Recovery after Experimental Stroke

Animals

All animal procedures were approved by the local ethical committee of Utrecht University and met governmental guidelines. A total of 28 male Wistar rats (weighing 250 to 350 g) were included in this study. Animals were divided into four experimental groups. In stroke groups, in vivo tract tracing using MEMRI was performed at 2 (Group S-2w; n=7), 4 (Group S-4w; n=7), or 10 (Group S-10w; n=7) weeks after experimental stroke. In a control group, MEMRI was performed at 4 weeks after sham operation (Group C-4w; n=7). Four animals of each group were also subjected to conventional tract tracing using WGA-HRP immunohistochemistry.

Stroke Model

Animals were anesthetized by subcutaneous injection of a mixture of 0.315 mg/mL fentanyl citrate and 10 mg/mL fluanisone (0.55 mg/kg), and 0.55 mg/kg midazolam. Blood oxygen saturation and heart rate were continuously monitored during surgical procedures. Body temperature was maintained at 37.0°C±0.5°C. Transient focal cerebral ischemia was induced by 90-min occlusion of the right middle cerebral artery with an intraluminal filament (Longa et al, 1989). In brief, a 4.0 silicon-coated polypropylene suture (Ethicon, Piscataway, NJ, USA) was introduced into the right external carotid artery and advanced through the internal carotid artery until a resistance was felt, indicating that the middle cerebral artery bifurcation was reached. Sham-operated animals were subjected to the same surgical procedures, except that the suture was advanced for only 1 mm, thereby not occluding the middle cerebral artery. After 90 mins, the filament was withdrawn from the internal carotid artery to allow reperfusion. After surgery, rats received a subcutaneous injection of 0.3 mg/kg buprenorphine (Schering-Plough, Utrecht, The Netherlands) for postsurgical pain relief and 5 mL saline to compensate for loss of water and minerals.

Functional Examination

Animals were subjected to two behavioral tests to assess sensorimotor function. First, we applied a battery of motor, sensory, and tactile tests, which provided a neurologic score on a scale of 0 to 20 points, with 20 as maximal deficit score (see Table 1) (modified from Reglodi et al, 2003). Second, an adhesive removal test was performed (Schallert et al, 2000). A small circular sticky tape was attached to the distal—radial region of the wrist of either the left or right forelimb (Schallert et al, 2000), and the sticky tape removal time was measured for each forelimb with a maximally allowed removal time of 60 secs.

OPEN IN VIEWER

Table 1

Neurologic score grading scale for rats

Sign	Description	Score
Motility, spontaneous activity	Normal or slightly reduced exploratory behavior	0–1
	Moving limbs without proceeding	2
	Moving only to stimuli	3
	Unresponsive to stimuli, normal muscle tone	4
	Severely reduced muscle tone, premortal signs	5
Gait disturbances	Straight walking	0
	Walking toward contralateral side	1
	Alternate circling and walking straight	2
	Alternate circling and walking towards paretic side	3
	Circling and/or other gait disturbances	4
	Constant circling towards paretic side	5
Postural signs
Forelimb flexion	Degree of limb flexion when held by tail	0–2
Thorax twisting	Degree of body rotation when held by tail	0–2
Lateral resistance	Degree of resistance against lateral push	0–2
Limb placing
Ipsilesional forelimb	Normal, weak, or no placing	0–2
Contralateral forelimb	Normal, weak, or no placing	0–2
Total score		0–20

Animals were trained at 4 and 3 days before stroke. Subsequently, behavioural examination was performed on days 0, 4, 7, and 10 after stroke, and every week thereafter.

Tracer Injection

Neuronal tract tracer was injected 4 days before MEMRI and subsequent removal of the brain for WGA-HRP immunohistochemistry. This time point was chosen to allow optimal combined detection of the two tracers.

Animals were anesthetized by subcutaneous injection of a mixture of 0.315 mg/mL fentanyl citrate and 10 mg/mL fluanisone (0.55 mg/kg), and 0.55 mg/kg midazolam and placed in a stereotactic holder. Blood oxygen saturation and heart rate were continuously monitored. Body temperature was maintained at 37.0°C±0.5°C.

Using a 2.0-μL Hamilton syringe, 0.2 μL of a solution containing either 1 mol/L MnCl₂ (n=3 for all groups) or a combination of 1 mol/L MnCl₂ and 5% WGA-HRP (Sigma Aldrich, Munich, Germany) (n=4 for all groups) was injected at a rate of 0.05 μL/min into the sensorimotor cortex through a burr hole in the skull (0.5 mm anterior, 1.5 to 3.0 mm lateral, and at a depth of 1.5 mm relative to bregma; Paxinos and Watson, 1998). After injection, the needle was left in place for 3 mins to prevent leakage along the injection track. Lateral coordinates were adjusted based on the lesion location determined by T₂-weighted MRI before tracer injection, to arrive at an injection site at approximately 0.5 mm from the lesion border. In the stroke groups, tracer was invariably injected into intact sensorimotor cortex, outside the T₂-defined lesion area. Injection sites for sham-operated rats were adjusted correspondingly. Mean lateral coordinates were 2.4±0.7, 2.2±0.6, 2.3±0.6, and 2.6±0.5 mm for Groups S-2w, S-4w, S-10w, and C-4w, respectively. As shown in our previous study, such small variation in lateral position of the injection site does not result in significant differences in the pattern of manganese enhancement in subcortical regions of the sensorimotor network (van der Zijden et al, 2007).

Magnetic Resonance Imaging

Magnetic resonance imaging measurements were performed on a 4.7 T horizontal bore MR spectrometer (Varian Instruments, Palo Alto, CA, USA) using a Helmholtz volume coil (90-mm diameter) and an inductively coupled surface coil (25-mm diameter) for signal excitation and detection, respectively.

Before MRI, rats were anesthetized with 4% isoflurane for endotracheal intubation, followed by mechanical ventilation with 2.5% isoflurane in N₂O/O₂ (70:30). Rats were placed in a home-built plastic holder and immobilized with earplugs and a tooth holder. Blood oxygen saturation and heart rate were monitored during MRI measurements. Body temperature was maintained at 37.0°C±0.5°C.

Multiecho, multislice T₂-weighted MRI (repetition time (TR)/echo spacing=3,000/17.5 ms; echo train length=8; acquisition matrix=128 × 128; voxel dimensions=0.2 mm³ × 0.2 mm³ × 1.2 mm³; 15 coronal slices; number of averages=2) was performed at (i) 1 week after stroke, (ii) 2 days before tracer injection, and (iii) 4 days after tracer injection. Quantitative T₂ maps were calculated on a voxelwise basis by weighted linear least squares fit of the logarithm of the signal intensity versus echo time (TE).

Saturation recovery T₁-weighted gradient-echo MRI with seven TRs (TR/TE=55 to 3,000/18 ms; acquisition matrix=128 × 128; voxel dimensions=0.2 mm³ × 0.2 mm³ × 1.2 mm³; 15 coronal slices; number of averages=2) was performed 2 days before tracer injection and 4 days after tracer injection. Quantitative pre- and postinjection R₁ (=1/T₁) maps were calculated on a voxelwise basis by performing a nonlinear least squares fit using the Levenberg—Marquardt method (Press and Vetterling, 1992).

Immunohistochemistry

At 4 days after tracer injection, rats that had received the mixture of MnCl₂ and WGA-HRP were deeply anesthetized by intraperitoneal injection of pentobarbital (120 mg/kg), and immediately transcardially perfused with saline followed by 4% paraformaldehyde in 0.1 mol/L phosphate-buffered saline (pH 7). Brains were removed and postfixed in 4% paraformaldehyde in 0.1 mol/L phosphate-buffered saline for an additional 2 h and subsequently stored in 20% sucrose in 0.1 mol/L phosphate-buffered saline at 4°C. Brains were cut in 40-μm-thick coronal sections on a freezing microtome, and every fifth section was collected in five containers consecutively and stored in 20% sucrose in 0.1 mol/L phosphate-buffered saline. Thus, each container consisted of 10 to 15 slices covering the whole brain. Free-floating sections were processed for immunohistochemical detection of WGA-HRP using goat anti-WGA (Vector Laboratories, Burlingame, CA, USA) as a primary antibody and development using a diaminobenzidine peroxidase substrate kit (Vector Laboratories) as described before (van der Zijden et al, 2007).

Data Analysis

Magnetic resonance imaging

Lesion volumes were determined from 11 adjacent slices of the T₂ dataset acquired at 1 week after stroke and defined as ipsilesional tissue greater than the mean+2 s.d. of T₂ in contralateral tissue. The edema-corrected hemispheric lesion volume (%HLV^e) was calculated as described by Gerriets et al (2004):

% H L V e = ( H V c − ( H V i − L V u ) ) / H V c × 100 % ,

where HV_c and HV_i are the contralesional and ipsilesional hemispheric volumes, respectively; and LV^u is the uncorrected lesion volume.

Longitudinal relaxation rate R₁ (=1/T₁) maps were coregistered to an averaged T₂-weighted brain MRI data set from six control rats using semiautomated image registration software (MNI Autoreg; Collins et al, 1994). Erroneous registration as a result of edema-induced brain distortion was corrected by a second registration procedure after manually selecting specific landmarks on brains with a stroke lesion. Coregistered pre- and postinjection R₁ maps were subtracted to generate ΔR₁ maps. As manganese-induced changes in R₁ are proportional to the local manganese concentration (Silva et al, 2004), manganese accumulation was measured from the difference between pre- and postcontrast R₁ (ΔR₁). Mean ΔR₁ values were determined in ipsi- and contralateral subcortical regions that are connected to the sensorimotor cortex, that is, striatum (St), thalamus (Th) and substantia nigra (SN), and corpus callosum (CC). Nonspecific distribution of manganese was verified in the ipsilateral visual cortex (VCX). Regions of interest (ROIs), shown in Figures 2 and 3A (top row), were identical for all animals and selected to be invariably outside the T₂-defined lesion area.

To measure the total volume of manganese enhancement in the ipsi- and contralateral hemisphere, we performed cluster-based segmentation of manganese-enhanced brain areas on ΔR₁ maps. Image intensity scale was standardized to correct for intensity variations between rats due to small differences in MnCl₂ concentration at the injection site. The average multislice ΔR₁ map of the control group was used to calculate a ‘standard’ intensity histogram, to which intensity histograms of ΔR₁ maps of individual rats were transformed (Nyul and Udupa, 1999). Subsequently, segmentation was performed using a fuzzy c-means method based on a clustering algorithm in which four classes (not enhanced; not likely enhanced; likely enhanced; very likely enhanced) were defined (Pham and Prince, 1999). Pixels segmented into classes 1 and 2 were considered nonenhanced, and pixels segmented into classes 3 and 4 were considered as manganese-enhanced. The volume of manganese enhancement was determined from the total number of manganese-enhanced pixels in the ipsi- and contralesional hemisphere of six adjacent brain slices posterior to the injection site.

Immunohistochemistry

WGA-HRP-stained brain slices were studied with a light microscope (Zeiss Axiophot with Sony 3 CCD Color Video Camera) under bright- and dark-field illumination. WGA-HRP-labeled neuronal profiles were counted on single brain sections that were at the same anterior—posterior position as the MRI slices used for ROI analysis. Counting of labeled profiles was performed in the anatomic area that best matched the corresponding ROI used for MRI analysis, where WGA-HRP staining was most pronounced. The number of labeled neurons was calculated using the following formula:

N = n T / ( T + h ) ,

where N is the number of neurons; n the number of neuronal profiles; T the slice thickness; and h the mean neuronal profile diameter (see Abercrombie and Johnson, 1946). For each rat, mean neuronal profile diameter (h) was determined from 10 random-labeled somata per ROI. The number of labeled neurons per 40 μm brain section was then multiplied by 30 to estimate the number of labeled neurons per 1.2-mm slice (i.e., the MRI slice thickness).

Functional Status

Figure 1 shows the changes in neurologic score (Figure 1A) and adhesive removal time (Figure 1B) as a function of time after stroke. All rats showed substantial functional deficits at day 4 after stroke. Functional recovery was reflected by a significant improvement of neurologic score on day 17 and adhesive removal time on day 10 as compared to day 4 after stroke. Plateau functional recovery, defined as the time point after which behavioral scores did not significantly change, occurred for both functional tests on day 17 after stroke.

OPEN IN VIEWER

Figure 1

Neurologic score±s.d. (A) and adhesive removal time±s.d. (B) as a function of time after stroke. x, Group C-4w (controls, 4 weeks after sham operation); ▪, Group S-2w (2 weeks after stroke); ♦, Group S-4w (4 weeks after stroke); ▴, Group S-10w (10 weeks after stroke). ∗P<0.05 versus score at first time point after stroke (i.e., day 4).

Ischemic Damage

Unilateral ischemic lesions, characterized by a significant increase in T₂, included part of the somatosensory cortex and the lateral striatum (Figure 2). Mean %HLV^e at 1 week after stroke was 17.1%±10.1%, 21.0%±8.3%, and 16.6%±8.1% for Groups S-2w, S-4w, and S-10w, respectively. There was no significant difference in lesion volumes between groups.

OPEN IN VIEWER

Figure 2

T₂ maps of four adjacent coronal brain slices at 4 weeks after stroke. The lesion is characterized by an increased T₂. Overlaid on the T₂ maps are ipsilateral regions of interest (ROIs): striatum (St), corpus callosum (CC), thalamus (Th), substantia nigra (SN), and visual cortex (VCX).

Manganese-Enhanced Magnetic Resonance Imaging

In all groups, 4 days after manganese injection into the sensorimotor cortex, R₁ maps clearly visualized manganese-enhanced regions that are connected to the site of injection, that is, striatum, thalamus, substantia nigra, and corpus callosum (Figure 3).

OPEN IN VIEWER

Figure 3

R₁ maps of six adjacent coronal brain slices. (A) Before (top row), and at 4 days after MnCl₂ injection in the ipsilateral sensorimotor cortex in a rat after sham operation (bottom row). (B to D) Postmanganese R₁ maps at 2, 4, and 10 weeks after stroke, respectively. Injection site (IN) and regions of interest (ROIs) are displayed on the precontrast control R₁ maps in the ipsilateral hemisphere (A, top row): striatum (St), corpus callosum (CC), thalamus (Th), substantia nigra (SN), and visual cortex (VCX). Matching ROIs were positioned in the contralateral hemisphere. (B to D) The stroke lesion is characterized by reduced R₁.

Preinjection R₁ values in striatum, thalamus, corpus callosum, and visual cortex were not significantly different between groups. A small but statistically significant R₁ increase in the ipsilesional substantia nigra was found between Groups C-4w and S-10w (0.95±0.02 and 1.06±0.04 sec⁻¹, respectively). However, this difference was minor compared to the postinjection R₁ values in this ROI (1.60±0.25 and 1.71±0.24 sec⁻¹ for Groups C-4w and S-10w, respectively).

In the ipsilateral hemisphere of control rats, R₁ values were significantly increased from baseline in all subcortical ROIs of the sensorimotor network after manganese injection (P<0.05). There was no significant R₁ increase in the visual cortex, which lies outside the sensorimotor network. Postmanganese R₁ values in subcortical sensorimotor regions were also significantly elevated in all stroke groups (P<0.05), except for the ipsilateral substantia nigra at 2 weeks after stroke. In the contralateral ROIs, manganese-induced R₁ changes were much smaller than in their ipsilateral counterparts. Nevertheless, a significant R₁ increase was detected in contralateral striatum in controls and at 2 weeks after stroke. At 4 and 10 weeks after stroke, R₁ was significantly increased in all sensorimotor ROIs of the contralateral hemisphere.

Figure 4 shows manganese-induced ΔR₁ values in ipsi- and contralateral ROIs. At 2 weeks after stroke, ΔR₁ was significantly reduced in the ipsilateral substantia nigra as compared to control rats. Thereafter, ΔR₁ values increased and were again similar to control values at 10 weeks after stroke. A significant increase in manganese-induced ΔR₁ as compared to controls was found at 10 weeks after stroke in the contralateral striatum, thalamus, and substantia nigra, as well as ipsi- and contralateral corpus callosum at 10 weeks after stroke.

OPEN IN VIEWER

Figure 4

Manganese-induced ΔR₁ (sec⁻¹) in ipsi- (A) and contralateral ROIs (B) at 4 days after MnCl₂ injection in the ipsilateral sensorimotor cortex of rats after sham operation, and at 2, 4, and 10 weeks after stroke. St, striatum; Th, thalamus; SN, substantia nigra; CC, corpus callosum; VCX, visual cortex. ∗P<0.05 versus sham-operated group.

With a fuzzy c-means algorithm we segmented ΔR₁ maps. For each of the four segmentation classes, mean nonnormalized ΔR₁ values were not significantly different between groups (Table 2). Classes 3 and 4 were considered to represent manganese-enhanced areas and were used for calculation of ipsi- and contralateral volume of manganese enhancement. Examples of segmented rat brain ΔR₁ maps of controls and at 4 weeks after stroke are shown in Figure 5. We found no significant differences in total volume of manganese enhancement in the ipsilateral hemispheres between control and all stroke groups (Figure 6). However, in the contralateral hemisphere, total volume of manganese enhancement was significantly larger at 4 weeks after stroke as compared to controls.

OPEN IN VIEWER

Figure 5

Fuzzy c-means-based segmentation of ΔR₁ maps of six adjacent coronal brain slices at 4 weeks after sham operation (A) or stroke induction (B). Four segmentation classes were discerned: ▪, not enhanced; , not likely enhanced; , likely enhanced; □, very likely enhanced.

OPEN IN VIEWER

Figure 6

Ipsi- and contralateral brain volumes of manganese enhancement as calculated from ΔR₁ segmentation maps. ∗P<0.05 versus sham-operated group.

OPEN IN VIEWER

Table 2

Mean ΔR₁±s.d. (sec⁻¹) of segmentation classes in control and stroke groups

Class	Group
	C-4w	S-2w	S-4w	S-10w
1	0.07±0.09	0.02±0.05	0.11±0.12	0.10±0.13
2	0.26±0.12	0.22±0.06	0.33±0.16	0.41±0.18
3	0.58±0.18	0.52±0.12	0.57±0.14	0.67±0.24
4	1.36±0.34	1.36±0.29	1.28±0.40	1.37±0.43

Immunohistochemistry

In correspondence with our previous study (van der Zijden et al, 2007), the spatial distribution of WGA-HRP staining in the ipsilateral hemisphere was largely similar to the pattern of manganese enhancement. In agreement with the above-described MEMRI findings, the degree of WGA-HRP staining in ipsilateral subcortical ROIs was reduced at 2 weeks after stroke, which was statistically significant in the striatum, and returned to control values after 4 and 10 weeks (Figure 7). However, we found no WGA-HRP-labeled neurons in contralateral ROIs.

OPEN IN VIEWER

Figure 7

Number of labeled neurons (× 10²) in ipsilateral subcortical ROIs of the sensorimotor network at 4 days after WGA-HRP injection in the ipsilateral sensorimotor cortex in rats after sham operation, and at 2, 4, and 10 weeks after stroke. St, striatum; Th, thalamus; SN, substantia nigra. ∗P<0.05 versus sham-operated group.

Restoration of Ipsilateral Connectivity

In agreement with our previous study (van der Zijden et al, 2007), reduced MEMRI and WGA-HRP staining in subcortical ROIs showed loss of connectivity between the perilesional sensorimotor cortex and ipsilateral subcortical sensorimotor network regions at 2 weeks after stroke. Numerous afferent and efferent cortical projections pass through or end in the infarcted zone, which included part of the somatosensory cortex and lateral striatum. After unilateral middle cerebral artery occlusion in rats, prominent axonal and terminal degeneration has been shown in cortical, striatal, thalamic, and nigral areas adjacent to the infarct (Iizuka et al, 1989; Kataoka et al, 1989). Furthermore, Yam et al (1998) have reported breakdown of axonal cytoskeleton and disruption of axonal transport inside and around the infarct after cerebral ischemia in rats. Clearly, these alterations give explanation to the disrupted tracer transport.

Significant ipsilateral loss of manganese enhancement and WGA-HRP labeling in subcortical ROIs at 2 weeks after stroke, however, was not present at later time points. These results suggest a reestablishment of connectivity between the perilesional cortex and subcortical regions, which may be explained by recovery of existing neuronal pathways and/or structural alterations such as regenerative growth of axons, collateral sprouting, and formation of new synaptic connections. Yet, it may also be possible that increased ipsilesional tracer accumulation represents enhanced take-up and/or transport of manganese or WGA-HRP as compared to earlier time points after stroke. Nevertheless, various histologic studies have provided evidence for structural plasticity in the lesion border zone. An increase of the growth cone marker, growth-associated protein 43 has been detected in perilesional tissue from the first days up to 4 weeks after unilateral stroke in rats (Stroemer et al, 1995). This was followed by an increase in synaptophysin, a marker of mature synapses, between days 14 and 60 (Stroemer et al, 1995). Ito et al (2006) recently reported elevated density of synapses and axon terminals in ischemic border zone from 1 to 12 weeks after stroke, subsequent to initial degeneration of neurites in the first week. Moreover, during the same period, dendritic spine formation is also increased in this area (Brown et al, 2007). Actual formation of new connections with the perilesional cortex through axonal sprouting within the ipsilateral hemisphere has been shown by Carmichael et al (2001). They injected the neuroanatomical tracer biotinylated dextran amine into the perilesional cortex and observed new intracortical projections at 3 weeks after focal ischemia in rats. Similarly, using the same tracer, Dancause et al (2005) detected corticocortical sprouting near a chronic ischemic lesion in monkey brain. Our data are in correspondence with these findings and further suggest that perilesional cortical connections with subcortical regions are also restored after a few weeks after stroke.

Enhanced Interhemispheric Connectivity

In addition to reestablishment of manganese enhancement in ipsilateral regions, we detected a significant increase in the total contralateral volume of manganese enhancement at 4 weeks after stroke, and significantly increased manganese accumulation in the contralateral sensorimotor ROIs at 10 weeks. Our results correspond with earlier studies that have provided evidence for structural remodeling in the contralesional hemisphere after unilateral ischemic damage. For example, Jones et al (1996) and Jones and Schallert (1992) have reported increased dendritic arborization and a larger number of synapses per neuron within the contralesional sensorimotor cortex between 18 and 30 days after lesion of the ipsilateral sensorimotor cortex in rats. In addition, Stroemer et al (1995) detected significantly elevated levels of synaptophysin in contralesional sensorimotor cortex at 14, 30, and 60 days after unilateral cortical infarction. Studies with an anterograde neuroanatomical tracer have shown axonal sprouting from the contralesional cortex to the ipsilesional striatum and to the perilesional cortex (Carmichael, 2003; Napieralski et al, 1996). As manganese is transported in both anterograde and retrograde directions (Pautler et al, 2003) and can be transferred transsynaptically (Pautler et al, 1998; Saleem et al, 2002), MEMRI allows visualization of wide-ranging neuronal networks, and not sole afferent or efferent projections. This explains why we have found increased manganese enhancement in multiple contralateral regions. The enhanced transhemispheric manganese transfer may be explained by retrograde transport by newly formed or unmasked corticocortical (Allegrini and Wiessner, 2003; Carmichael, 2003), corticostriatal (Napieralski et al, 1996), and corticothalamic projections (Yu et al, 1995). Evidence for increased manganese transport in crossing white matter fibers was found in the corpus callosum. This is in correspondence with findings by Allegrini and Wiessner (2003), who detected manganese enhancement in transhemispheric callosal fibres after manganese injection in the sensorimotor cortex contralateral to a cortical photothrombotic lesion. Also, increased manganese transport to the contralateral olfactory cortex has recently been detected at 4 weeks after a lesion in the lateral olfactory tract (Cross et al, 2006).

Despite the clear MRI-based detection of manganese in contralateral regions, marked WGA-HRP staining was not evident in the contralateral hemisphere in our study. The tracer survival time of 4 days, which was optimized for MEMRI, may have been too short for WGA-HRP transport to distant contralateral areas. In addition, WGA-HRP staining is not easily discernible because of punctate and incomplete filling of neuronal processes with this tracer (Ferguson et al, 2001) and may be a less sensitive method to detect tracer accumulation at more remote sites.

Correlation between Brain Reorganization and Functional Recovery

The time at which we detected increase in intra- and interhemispheric neuroanatomical connectivity corresponded with the stage of plateau recovery of sensorimotor function, that is, more than 2 weeks after stroke. Although our study does not provide direct evidence of a relationship between functional recovery and neuroanatomical reorganization, previous studies have shown that the temporal pattern of functional recovery correlates with neuroanatomical alterations, such as peri- and contralesional synaptogenesis (Stroemer et al, 1995) and contralesional dendritic sprouting (Jones and Schallert, 1992) after cortical brain injury. Furthermore, pharmacological stimulation of neuronal sprouting has been shown to improve behavioral performance (Chen et al, 2002; Kawamata et al, 1997; Stroemer et al, 1998). The lack of a direct correlation between early functional recovery and neuroanatomical changes in our study may be explained by insensitivity of the neuroanatomical tracer techniques to detect initial changes in projection patterns. Nonetheless, our MEMRI results suggest that plateau functional recovery is upheld by effective (re-)establishment of connections. Moreover, the maturation of a newly formed network ties in with recovery of cortical function adjacent to a unilateral infarct. With functional MRI, we have shown that neuronal activation responses in perilesional sensorimotor cortex, which are largely lost in the first days after stroke, are restored after 2 weeks (Dijkhuizen et al, 2001).

In conclusion, in line with studies on the bird song control system (Van der Linden et al, 2004) and mossy fiber sprouting in a rat epilepsy model (Nairismagi et al, 2006), our results show the unique potential of MEMRI for in vivo detection of structural plasticity in the brain. Our MEMRI data on spatiotemporal changes in neuroanatomical connectivity after stroke suggest that remodeling of intra- and interhemispheric neuronal networks effectively contributes to poststroke functional recovery.