Cryoanesthetized P1–3 rat pups were injected with a total amount of 1 μl of AAV1-hSynap-eGFP-WPRE-bGH (titer 10¹¹ copies/ml; Hutson et al., 2012) into four sites of right sensorimotor cortex (0.25 μl/site; coordinates from Bregma: 1.5 mm ML: 0.7 mm AP; 1 mm ML: 0.3 mm AP; 1 mm ML: 0.0 mm AP; 2 mm ML: 0.5 mm AP). Briefly, animals were fixed on a stereotaxic apparatus (Cunningham, Stoelting Co., Wood Dale, IL, USA) previously cooled with dry ice and viral injections were performed by using a pneumatic ejection pump (PDES-02TX, npi) attached with a pulled glass micropipette (30 μm tip). Pups were then placed under a heating lamp until they were fully awake and transferred to the home cage with their mothers.

P14 or P21 rats were administered intracortical injections of ET-1 (Sigma; 120 pmol/μL in sterile saline) as previously described (Gherardini et al., 2015). Briefly, animals were anesthetized by intraperitoneal injection of avertin (200 mg/Kg). Local injection of lidocaine (100 mg/Kg) under the scalp was provided. Body temperature was monitored and maintained at 37°C with a homothermic blanket (Harvard Apparatus Ltd, UK). Injections were performed at two sites of the left hemisphere corresponding to the primary motor area with respect to Bregma (1.5 mm ML: 0.0 mm AP; 1.5 mm ML: 0.5 mm AP; depth of 0.5 mm for P14 and 0.7 mm for P21 animals) by giving a total amount of 2 μl (240 pmol/μl) of ET-1 solution through a pulled glass micropipette connected to a manual pump at a rate of 0.5 μL/min with a 5-min interval before retracting the micropipette from the tissue. Sterile saline was injected in control groups with the same procedure. Immediately after surgery animals were given local antibiotic (aureomycin 3% cream) and analgesic (paracetamol, 10 mg/kg) administered in water. Animals were kept in a warming incubator until they recovered from anesthesia. P14 animals returned to their mothers in the home cage. All procedures were performed in accordance with the Italian Ministry of Health guidelines for care and maintenance of laboratory animals (law 116/92) and in strict compliance with the European Communities Council Directive 86/609/EEC. All the experiments were carried out in accordance with the directives of the European Community Council (2011/63/EU) and approved by the Italian Ministry of Health. Long Evans rats (n = 90) from P1 until P70–80 of both sexes were used.

For the tracing of cortico-rubral connections, ET1 and control animals not previously transfected with AAV-GFP tracer were injected at P40 with the anterograde dye Biotinylated Dextran Amine (BDA). Rats were deeply anesthetized with avertin and placed in a stereotaxic frame. Animals were pressure-injected with 2 μl of a solution of 10% BDA (10,000 MW) in 0.01 M phosphate buffer, pH 7.2 into the forelimb mapping area of the contra-lesional hemisphere, 1 mm depth from pial surface at four sites (0.5 μl each point; stereotaxic coordinates: A-P 0, M-L 3 mm; A-P 0; M-L 2.5 mm; A-P 2 mm, M-L 2.5 mm; A-P 2.5 mm, M-L 2.5 mm).

The effect of lesion-timing on long-term motor outcome was evaluated by general and fine motor behavioral tests performed in adult age, starting from P59 (see Supplementary Methods). An operator blind to assigned experimental group carried out all behavioral tests. Gait analysis to evaluate locomotion was performed by using a modified protocol from de Medinaceli et al. (1982), Metz and Schwab (2004) and Patterson et al. (2010). Sensorimotor coordination was assessed at vertical ladder according to Metz and Whishaw (2009), whereas muscle strength was measured by grip test (Doeppner et al., 2014). Reaching skills were daily tested from P60 by Montoya Staircase Test (MST) for 1 week (Montoya et al., 1991), skilled motor impairments were scored according to Smith et al. (2007) and Soleman et al. (2012) and expressed as percentage values.

To investigate the effects of an early activity-dependent modulation of CST plasticity after developmental lesion, a group of animals treated with ET-1 at P14 were trained twice/day in the MST apparatus till they reached the plateau level of skilled motor learning, starting from P21. The effect of this training on long-term motor outcome was assessed at P59 as described before and results were compared with those of Saline or ET1 P14 injected animals.

After 3 or 7 days after lesion induction, after behavioral assessment, or after 3 weeks following BDA injections (P60), animals were lethally anesthetized and transcardially perfused with 4% (PFA) 0.1 M phosphate buffer. Both brain and spinal cord were extracted and post-fixed overnight at 4°C, then immersed in a dehydrating solution of sucrose 30% and sodium azide 0.05% for at least 3 days. Dehydrated tissues were embedded with Tissue-Tek® OCT™, snap-frozen in −80°C cooled isopentane, and 50 μm coronal sections of brain and cervical spinal cord were cut with a cryostat (Leica).

Coronal sections of entire motor cortex were collected one out of two slices and then stained with propidium iodide (5 μg/μl) Sigma-Aldrich® and mounted and cover slipped. Mosaic images (2557 × 1917 pixels) were acquired with a fluorescence microscope (Axio Imager.Z2, Zeiss) using 5× EC-PLAN-NEOFLUAR objective (N.A. 0.25). Lesion area was measured using Image-J software. The volume was computed by multiplying individual lesion areas by the interslice distance.

For brains derived from animals injected with BDA, one every three slice were collected in the area of the tracer injection site, whereas all consecutive sections including the red nucleus were collected and then serially mounted on gelatine-coated slides. Slides were washed three times for 15 min in TBS-TX 0.3% pH 8.0 before incubation overnight with an avidin–biotin–peroxidase complex diluted in TBS-TX 0.3% (ABC elite; Vector Labs, Burlingame, CA, USA). The following day the slides were washed three times and rinsed with TBS-TX 0.3% for 30 min, and then for 5 min with 0.05 M Tris-HCl pH 8.0. Finally the tissue was stained using nickel-enhanced diaminobenzidine (DAB) protocol on slides (Vector Labs, Burlingame, CA, USA). Images were acquired with transmitted light microscope (Leica) and the area of interest (ROI), identified using the stereotaxic coordinates of the atlas of Paxinos and Watson (1998), was analyzed with the software Stereo Investigator (MBF Bioscience). A grid (2.5 mm² each square) was superimposed on each of section and cortico-rubral projections to the ipsilateral and contralateral red nucleus were analyzed quantitatively in three squares (7.5 mm²) on all sections. Two parameters were taken into account: the “innervation index” and the “midline crossing fibers index” (Z’Graggen et al., 2000). Innervation index was calculated as ratio of fiber density of the ipsilesional (denervated) and contralesional (innervated) red nucleus. Midline crossing fibers index was quantified by counting BDA-positive fibers crossing the midline on each section and corrected for the inter animal tracing variability by dividing the number of midline crossing fibers per the total CTS axons fibers quantified into a ROI of cerebral peduncle (7.5 mm²).

One every 10 slices of medullar pyramids (at the level of caudal cerebellum) were collected for quantification of the viral tracer labeled axons amount. Cervical spinal cord was sectioned from C6 to C8 levels for analysis of CST sprouting; a total number of four sections/level were obtained, collecting one every six-eight consecutive slices.

To quantify the bulk of the labeled dorsal CST (dCST) and ventral CST (vCST), 10× magnification (EC-PLAN-NEOFLUAR objective, N.A. 0.3) images were acquired at fluorescence microscope (Axio Imager.Z2, Zeiss) and integrated fluorescence density of the labeled CST was computed by ImageJ software. For each animal, at least three sections per level were analyzed. Appropriate exposure parameters were chosen in order to avoid saturation of the highly intense signals coming from the funiculi and to minimize such variability in signal detection. To correct for inter animal tracing variability integrated density of both dCST and vCST were corrected by the number of GFP⁺ fibers counted in at least three section of the medullar pyramids (see Supplementary Material).

GFP⁺ axonal sprouting from labeled dCST and vCST was analyzed by a fluorescence microscope (Leica), magnification 20×, using the software Stereo Investigator (MBF Bioscience). At least three sections per level were measured in each rat. For all slices, GFP signal was detected without immunohistochemistry procedures. GFP⁺ axonal fibers originating from labeled dCST and reaching contralateral denervated spinal cord were counted as fibers crossing the spinal cord midline (M) and extending through four different distance lines (D1–4), with each line being 150 μm apart and spanning across all gray matter. GFP⁺ axonal fibers arising from labeled vCST and reaching ipsilateral denervated spinal cord were counted as fibers crossing the boundary between white-gray matter of the medial part of ventral horn (Wahl et al., 2014). Axonal sprouting was normalized on the total number of GFP labeled corticospinal axons counted on at least three coronal sections of medullar pyramids (see Supplementary Material).

Z-stack images of three C6–C8 spinal cord sections per animal were scanned using fluorescence microscope (Axio Imager.Z2, Zeiss) and by applying structured illumination provided by Apotome.2 (Zeiss). 10X mosaic images (EC-PLAN-NEOFLUAR objective, NA 0.3) were acquired both in bright field and in green channel fluorescence, in order to recognize anatomical landmarks of spinal cord gray matter. Maximum intensity projections (MIPs) were generated from the group of ≅10 consecutive stacks yielding the higher mean pixel intensity. Semi-automated analysis for GFP⁺ axons detection and characterization was performed using a custom-made MATLAB® algorithm. Briefly, to sample all spinal cord laminae, 15 square ROIs (129 × 129 μm) per section covering all the dorso-ventral axes of spinal gray matter were selected using bright field cues (Supplementary Figure S2). GFP⁺ axons signal was therefore detected contemporary in the green channel. ROIs sampling was carried out in a standard fashion for all analyzed images. ROIs 1–5 identify dorsal sensory laminae (1–6), ROIs 6–10 the intermediate lamina 7 and ROIs 11–15 the ventral, motor laminae 8 and 9. Once selected, each ROI was processed for axons detection and measurement of their laminar pattern distribution. This parameter named “axonal complexity index” (ACI) refers to: Axonal Complexity Index = Na * Ca

where Na is the mean number of detected axons per laminae (normalized for the number of GFP⁺ axons in medullar pyramids) and Ca is the mean axonal complexity per laminae (sum of branch points and endpoints per object). For further information about the algorithm see Supplementary Material.

All data are presented as mean ± standard error of the mean (SEM). n denotes number of sampling in each group. Significant level was settled as 0.05. Lesion volume measures, general motor behavior, analysis of sprouting at red nucleus level, fluorescence integrated density of CST, sprouting from vCST and laminar distribution were analyzed using t-test, Two-Way ANOVA or One-Way ANOVA, when appropriate. Normality assumption was validated using Shapiro-Wilk test; when rejected, data were transformed on ranks and then analyzed with the appropriate statistic test. Number of GFP⁺ axons on medullar pyramids was analyzed through non-parametric Kruskal-Wallis test. Behavioral data of MST and axonal sprouting from dCST were analyzed using Two-Way repeated measure ANOVA, with respectively number of days and sequential distances as within subject factors. Post hoc multiple comparison using Holm-Sidak test was performed, when appropriate. SigmaPlot®12.0 was used to perform the analysis. Size of experimental groups is shown in Table 1.

To induce a focal ischemic stroke, we performed intracortical injections of ET-1 (Gherardini et al., 2015) targeting the primary forelimb motor area (fM1) corresponding to the caudal forelimb area (CFA). We first assessed whether the stroke induction protocol adopted (Figure 1A) caused the same tissue damage when performed at P14 or P21. Nuclear staining with propidium iodide showed the presence of consistent tissue loss extending from the pia to the white matter (Figure 1B) in lesioned animals once adults. Quantitation in Figure 1C revealed that no difference in lesion volume was present between ET1-P14 (n = 7) and ET1-P21 (n = 6) rats (Two-Way ANOVA, factor age p = 0.588). As expected, lesion was almost absent in control rats injected with saline (Figure 1C, factor lesion p < 0.001). The lesion size was not different between ET1-P14 and ET1-P21 rats also at shorter times after lesion (3 and 7 days after lesion, Supplementary Figure S2). Thus, our protocol of stroke induction caused similar tissue damage at P14 and at P21.

We then evaluated the effect of lesion timing on motor outcome. Both general and skilled motor behavior was measured when animals reached young-adult age (P59; Figure 1A). Both at P14 and at P21, ET1 lesion affected motor function and muscle strength, indeed ET-1 animals performed worse than control animals in ladder climbing (Figure 2A, factor age p = 0.924, factor lesion p < 0.001) and grip test (Figure 2B, factor age p = 0.070, factor lesion p < 0.001).

Intriguingly, analysis of locomotor behavior and skilled reaching ability revealed more pronounced deficits for rats lesioned at P14. Indeed, ET1-P14 animals had worse fore-hindlimb coordination with their affected contralesional side with respect to SAL-P14 and ET1-P21 animals (Figure 2D, right, factor lesion × age p = 0.002; ET1-P14 vs. SAL-P14, p < 0.001, ET1-P14 vs. ET1-P21 p < 0.001). By contrast forelimb rotation analysis revealed a significant abduction of the lesioned (contralesional) forelimb vs. controls (Figure 2C, right, factor lesion p < 0.001), but without any effect of lesion timing (Figure 2C, right, factor age p = 0.114). No alteration was observed on the ipsilesional forelimb both for limb rotation (Figure 2C, left, interaction lesion × age p = 0.304) and for interlimb coordination (Figure 2D, left, interaction lesion × age p = 0.735).

Assessing skilled reaching ability using the MST revealed a deleterious effect of lesion age. Grasping deficits were more prominent in ET1-P14 animals than in ET1-P21 (Figure 2E, factor lesion × age p = 0.02, p-values in figure legend). Moreover, motor task learning was also affected in an age-dependent manner. In fact, whilst ET1-P21 and control groups begin to show significant task learning starting respectively from 2nd to 3rd day of testing (Figure 2E, factor day p < 0.001, 2nd vs. 1st day within SAL-P14 p = 0.001; 3rd vs. 1st day within SAL-P21 p < 0.001; 3rd vs. 1st day within ET1-P21 p = 0.001), ET1-P14 animals showed a delayed learning of the task, with significant learning only present from day 6th (factor day p < 0.001, 6th vs. 1st day within ET1-P14 p = 0.02).

A comparable trend was observed when reaching and extension abilities of the affected forelimb were quantified by counting the number of steps reached for pellet retrieval. Although lesion induced an overall worsening with respect to controls (Figure 2F, factor group × age p < 0.001), P14 lesioned animals had a stronger long-term reaching impairment than P21 injured rats (factor lesion × age p = 0.041, ET1-P21 vs. ET1-P14 p < 0.05). Furthermore, learning curve comparisons confirmed the worse ET1-P14 extension ability. Indeed, animals started to learn the reaching motor task from the 6th day of the test whereas the other groups were able to learn from 2nd to 3rd day of test (factor day p < 0.001, 6th vs. 1st day within ET1-P14, p < 0.001; 2nd vs. 1st day within SAL-P14 p < 0.001; 3rd vs. 1st day within SAL-P21 p < 0.001; 3rd vs. 1st day within ET-1 P21 p = 0.004). No statistical difference was observed between SAL-P14 and SAL-P21 controls either in grasping (SAL-P14 vs. SAL-P21, p = 0.81) or in reaching performance (SAL-P14 vs. SAL-P21, p = 0.62).

To compare the structural consequences of P14 and P21 focal cortical stroke, we first analyzed sprouting of cortico-rubral fibers originating from the contralesional hemisphere in the red nucleus that is involved in the fine control of reaching movement (Z’Graggen et al., 2000; Williams et al., 2014). Axons were labeled with BDA at P40 and analyzed at P60 in P14 and P21 ET1 lesioned and control unlesioned rats (Figure 3A). We first counted fibers crossing the midline: ET1-P14 lesioned rats showed a significant increase in the number of BDA-labeled axons crossing the midline with respect to SAL-P14 (Figures 3B,C1, factor treatment × age p < 0.05, ET1-P14 vs. SAL-P14, p = 0.024) and ET1-P21 lesioned animals (ET1-P14 vs. ET1-P21, p < 0.05). We also computed an innervation index that represents the labeling density ratio between the ipsilesional (denervated) and the contralesional (innervated) nucleus. Also for this index we observed a significant increase in ET1-P14 lesioned animals vs. SAL-P14 and ET1-P21 group (Figures 3B,C2, factor treatment × age p = 0.004; ET1-P14 vs. SAL-P14 p = 0.003, ET1-P14 vs. ET1-P21 p = 0.001). These data strongly suggest that an early focal ischemia induce a contralesional reorganization of cortico-rubral pathway, that is not observed when the lesion occurs 1 week later.

We then decided to analyze CST sprouting at cervical spinal cord level. In this analysis, we evaluated not only the amount of sprouting, but also the precise localization of the newly formed fibers, a crucial aspect for correlation with behavioral outcome (Wahl et al., 2014). Thus, we decided to use a viral labeling technique (Okada et al., 1999; Spergel et al., 2001; Harvey et al., 2009; Alves et al., 2014) that had more temporal stability and did not require another post-lesion surgery session.

We used AAV1-GFP viral vector injected into the fM1 to anterograde trace axons originating from pyramidal neurons (Figure 4A and Supplementary Figure S3) of the uninjured cortex. GFP⁺ axons were easily detected at the level of pyramidal medulla, although we also distinguished the presence of spared cortico-reticular and cortico-vestibular projections (Figure 4B). Moreover, dCST on the opposite side, and vCST on the same side of AAV-GFP cortical injections, were clearly observed in coronal sections of spinal cord (Figure 4C). We quantified fluorescence (GFP⁺) integrated density of dCST and vCST funiculi at C6–C8 spinal cord level to verify whether timing of developmental lesion might differently interfere with normal maturation of contralesional CST (Figure 5A). No significant difference in dCST fluorescence integrated density was present between all groups (Figure 5B, factor treatment × age p = 0.385), confirming that labeling occurred similarly in all AAV injected animals. On the contrary, vCST fluorescence integrated density of ET1-P14 and ET1-P21 was significantly higher with respect to control (SAL-P14 and SAL-P21) rats (Figure 5C, factor treatment p = 0.003), but no differences were detected between rats lesioned at different ages. These data indicate that pruning of vCST that normally occurs during development is prevented by lesioning the contralateral cortex, and are in agreement with the presence of a competition for connecting with spinal cord targets between axons coming from ipsilateral and contralateral hemispheres. These results parallel clinical evidence obtained in hemiplegic cerebral palsy (CP) patients showing that axonal diameter of ipsilateral, but not contralateral, corticospinal axons was increased with respect to healthy controls (Eyre, 2007).

We further assessed sprouting from dCST and vCST into the denervated C6–C8 spinal cord gray matter. GFP⁺ axons sprouting from uninjured dCST were counted from midline (M) to different distances (up to 600 μm medio-lateral in four steps 150 μm apart) towards denervated cervical gray matter (Figure 6A, left panels). Interestingly, ET1-P14 rats had a significant greater number of GFP⁺ sprouted axons than SAL-P14 and ET1-P21 groups, with elongation significantly extending beyond 300–450 μm from midline (Figure 6B, factor group × distance p = 0.006, post hoc p < 0.05). The number of GFP⁺ fibers originating from vCST and crossing the border between white and gray matter in the ventral horn of spinal cord was also counted (Figure 6A, right panels). Similarly to dCST, ischemic lesion induced spontaneous structural plasticity also from vCST (Figure 6C, factor treatment p < 0.001). Intriguingly, in ET1-P14 rats vCST sprouting was significantly more prominent than in ET1-P21 animals (Figure 6C, factor treatment × age p < 0.05, post hoc p = 0.006).

Since ET1-P14 lesioned rats displayed worse motor outcome but widespread axonal sprouting, we analyzed whether lesion induced sprouting targeted the correct laminae in the denervated spinal cord. The distribution of axonal sprouts on different spinal cord laminae was measured using a semi-automated algorithm (Supplementary Figure S1, see “Material and Methods” Section and Supplementary Material) computing, for each position in the spinal cord, an index designed “ACI” combining the number of axons detected and their branching measured by summing the number of branch points and endpoints.

In normal rats the intermediate laminae received the great majority of CST axons. However, ET1-P14 rats showed enhanced sprouting with respect to P21 ET1 rats only in dorsal and ventral laminae but not in the intermediate laminae (Figures 7A–B3). Indeed, ET1-P14 rats exhibited a significant increase in ACI compared to ET1-P21 animals in dorsal (Figure 7A left panels and B1, factor treatment × age p = 0.006, post hoc between ET1 groups p < 0.05), and in ventral laminae (Figure 7A right panel and B3, factor treatment × age p = 0.009, post hoc between ET1 groups p < 0.05). Interestingly, there was no significant difference of ACI in intermediate laminae between ET1-P14 and ET1-P21 animals (Figure 7A middle panels and B2, factor treatment × age p = 0.049, post hoc between ET1 groups p = 0.146) indicating that the lamina normally targeted by CST axons does not receive more sprouts in P14 lesioned than in P21 lesioned rats. Furthermore, whilst ACI did not change across laminae between ET1-P21 and SAL-P21 animals (Figures 7B1–B3, ET1-P21 vs. SAL-P21 p > 0.05 for all laminae), there was a significant increase of ACI between ET1-P14 and SAL-P14 in all laminae (Figures 7B1–B3, ET1-P14 vs. SAL-P14 p ≤ 0.01 for all laminae). These observations support the idea that cortical stroke at P14 causes a widespread sprouting of CST invading territories that are not normally involved in CST control of spinal circuits.

We then asked whether the aberrant CST projections associated with poor behavioral outcome induced by P14 lesion could be recovered by early training. ET1-P14 rats were trained from P21 to learn a skilled reaching task by MST twice/day till the plateau level of learning was achieved (Figure 8A1). To investigate whether this early learning protocol could induce activity-dependent rearrangements of CST sprouting correcting the maladaptive contralesional rewiring observed after P14 lesion, we performed the same anatomical analysis described above.

We found that neither lesion size (Figure 8A2, t-test, p = 0.914) or bulk fluorescence density of dCST and vCST funiculi (Supplementary Figure S4A) were significantly affected by training (Supplementary Figures S4B,C, One-Way ANOVA, p = 0.101 for dCST and p = 0.151 for vCST). In contrast, early training significantly reduced the aberrant spontaneous sprouting present in ET1-P14 rats from either the dCST (Figure 8B, factor group × distance p < 0.001, ET1-P14 TRAINING vs. ET1-P14 p < 0.001, SAL-P14 vs. ET1-P14 TRAINING p = 0.643) and the vCST (Figure 8C, p < 0.001, ET1-P14 TRAINING vs. ET1-P14 p < 0.001, SAL-P14 vs. ET1-P14 TRAINING p = 0.702) restoring values of controls. Furthermore, a significant drop of complexity in arborization of sprouted axons (Figure 8D) in early trained ET1-P14 animals was observed across all spinal cord laminae (Figures 8E1–E3): dorsal (Figure 8E1, p < 0.001, ET1-P14 TRAINING vs. ET1-P14 p < 0.001, SAL-P14 vs. ET1-P14 TRAINING p = 0.839), intermediate (Figure 8E2, p = 0.004, ET1-P14 TRAINING vs. ET1-P14 p = 0.012, SAL-P14 vs. ET1-P14 TRAINING p = 0.610) and ventral (Figure 8E3, p = 0.01, ET1-P14 TRAINING vs. ET1-P14 p = 0.006, SAL-P14 vs. ET1-P14 TRAINING p = 0.610). Thus, early training is able to fully revert the mistargeted CST sprouting induced by P14 stroke.

To verify whether training induced rescue of axonal sprouting was associated with a better motor outcome of P14 stroke, we performed a behavioral assessment in early trained P14 stroke rats starting from P59. The same battery of tests previously described was used (Figure 8A1). We compared ET1-P14 rats with or without early training, with non-trained control groups. We found that early skilled motor training promoted a partial amelioration of sensorimotor coordination at vertical ladder climbing (Figure 9A, p < 0.001, ET1-P14 vs. ET1-P14 TRAINING, p = 0.003; ET1-P14 vs. SAL-P14, p < 0.001; ET1-P14 TRAINING vs. SAL-P14, p = 0.002), a complete recovery of both grip strength (Figure 9B, p < 0.001, ET1-P14 vs. ET1-P14 TRAINING, p < 0.001; ET1-P14 vs. SAL-P14, p < 0.001; ET1-P14 TRAINING vs. SAL-P14, p = 0.2) and of interlimb coordination (Figure 9D, p < 0.001, ET1-P14 vs. ET1-P14 TRAINING, p = 0.002; ET1-P14 vs. SAL-P14, p < 0.001; ET1-P14 TRAINING vs. SAL-P14, p = 0.98). No significant effect of motor training was detected in abduction of affected forelimb induced by lesion (Figure 9C, p < 0.001, ET1-P14 vs. ET1-P14 TRAINING, p = 0.55; ET1-P14 vs. SAL-P14, p = 0.002; ET1-P14 TRAINING vs. SAL-P14, p = 0.002). Interestingly, early motor training partially ameliorated grasping (Figure 9E, p < 0.001) and reaching (Figure 9F, p < 0.001) fine motor abilities. Indeed, ET1-P14 trained animals, similarly to controls, learned both grasping and reaching task significantly better than untrained ET1-P14 group (Figures 9E,F, factor treatment × days p < 0.001).