Lateral Cerebellar Nucleus Stimulation has Selective Effects on Glutamatergic and GABAergic Perilesional Neurogenesis After Cortical Ischemia in the Rodent Model

Abstract

BACKGROUND

Chronic deep brain stimulation of the rodent lateral cerebellar nucleus (LCN) has been demonstrated to enhance motor recovery following cortical ischemia. This effect is concurrent with synaptogenesis and expression of long-term potentiation markers in the perilesional cerebral cortex.

OBJECTIVE

To further investigate the cellular changes associated with chronic LCN stimulation in the ischemic rodent by examining neurogenesis along the cerebellothalamocortical pathway.

METHODS

Rats were trained on the pasta matrix task, followed by induction of cortical ischemia and electrode implantation in the contralesional LCN. Electrical stimulation was initiated 6 wk after stroke induction and continued for 4 wk prior to sacrifice. Neurogenesis was examined using immunohistochemistry.

RESULTS

Treated animals showed enhanced performance on the pasta matrix task relative to sham controls. Increased cell proliferation colabeled with 5’-Bromo-2’-deoxyuridine and neurogenic markers (doublecortin) was observed in the perilesional cortex as well as bilateral mediodorsal and ventrolateral thalamic subnuclei in treated vs untreated animals. The neurogenic effect at the level of motor cortex was selective, with stimulation-treated animals showing greater glutamatergic neurogenesis but significantly less GABAergic neurogenesis.

CONCLUSION

These findings suggest that LCN deep brain stimulation modulates postinjury neurogenesis, providing a possible mechanistic foundation for the associated enhancement in poststroke motor recovery.

ABBREVIATIONS

ANOVA

analysis of variance

AP

anterior-posterior

COI

conflict of interest

CTC

cerebellothalamocortical

DBS

deep brain stimulation

DCX

doublecortin

DV

dorsal-ventral

FITC

fluorescein isothiocyanate

ICH

immunohistochemistry

LCN

lateral cerebellar nucleus

MD

mediodorsal

ML

medial-latera

NIH

National Institute of Health

NPCs

neural progenitor cells

PBS

phosphate-buffered saline

SEM

standard error mean

SVZ

subventricular zone

VL

ventrolateral

Deep brain stimulation (DBS) is routinely used to suppress motor signs in essential tremor, Parkinson's disease, and dystonia¹ and is being explored as a treatment option for obsessive–compulsive disorder, major depression, epilepsy, chronic pain, and stroke.^2-6 Although its therapeutic mechanisms continue to be debated,⁷ it has been shown to modulate neural activity not only within the deep brain target but also across interconnected brain regions.⁷ Our group is interested in the application of DBS to facilitate motor recovery in chronic, poststroke patients,^2-3,5,7 and we have shown previously that DBS of the cerebellothalamocortical (CTC) pathway combined with motor rehabilitation, promotes motor recovery after both large² and focal⁷ cerebral cortical ischemia.

Using rodent models, we have shown previously that DBS of the lateral cerebellar nucleus (LCN) modulates cerebral cortical excitability, with the magnitude and sustainability of the effect stimulus frequency dependent.⁸ In subsequent experiments, therapeutically effective 30 Hz DBS was found to be associated with increased motor representation of the affected forelimb, greater expression of long-term potentiation markers in the perilesional area, and greater perilesional synaptogenesis.⁹ Here, we extend those findings and assess whether motor gains from chronic LCN DBS are also associated with changes in perilesional cytoarchitecture. In particular, we are interested in whether LCN DBS is associated with changes in neurogenesis and cytoarchitecture when applied in the chronic phase following ischemia, at a time when lesions have already become well established and encephalomalacia has occurred. As DBS is already a proven, safe technique in humans, there is significant translational potential in using this tool to gain control over cell proliferation in the management of poststroke hemiparesis and other acquired brain injuries.¹⁰ Here, we examine the effects of DBS on markers of neurogenesis in the perilesional cortex and in thalamic nuclei that form the CTC pathway.

METHODS

Animals

Twenty-three male Long Evans rats (200-224 g) were individually housed on a 12-h light/dark cycle and mildly food restricted (12 g of food per day/water ad libitum). Customized caging allowed behavioral testing and chronic stimulation delivery without manipulation or disconnection of the animal. All behavioral testing was performed during the dark phase of the light cycle under controlled red lighting. Animal work was approved by the Institutional Animal Care and Use Committee.

Surgical Procedure

LCN electrode implantation and focal ischemia were conducted as detailed previously.⁹ Briefly, rats were anesthetized and fixed in a stereotaxic frame (David Kopf Instruments, Tujunga, California), and a craniotomy was performed over the sensorimotor cortex contralateral to the dominant paw. Ischemia was induced by 6 intracortical injections of 800 pmol/2 μl endothelin 1 (ET-1, Millipore, Billerica, Massachusetts) at sites in relation to bregma (mm): (1) anterior-posterior (AP): –1.0, medial-lateral (ML): ±2.5, Dorsal-ventral (DV): –2.3; (2) AP: +1.0, ML: ±2.5, DV: –2.3; (3) AP: +3.0, ML: ±2. 5, DV: –2.3; (4) AP: –1.0, ML: ±3.5, DV: –2.3; (5) AP: +1.0, ML: ±3.5, DV: –2.3; and (6) AP: +3.0, ML: ±3.5, DV: –2.3. Under the same anesthesia, a bipolar macroelectrode (MS306; Plastics One, Roanoke, Virginia) was implanted in the LCN contralateral to the ET-1 injections at AP: –11.0 mm, ML: ±3.6 mm, and DV: –6.3 mm. The electrode was secured with dental acrylic and screws. Animals recovered with food and water ad libitum for 1 wk and buprenorphine was used for analgesia.

Experimental Design

Figure 1A provides an overview of the experimental design. Twenty-three animals underwent 3 wk of prestroke training on the pasta matrix task (10 min per day, 5 d per week) as described previously.⁷ Pasta was presented initially on both sides of the matrix during week 1, but restricted to the side opposite each animal's dominant forepaw once preference was established. Week 3 performance was averaged to establish prestroke baseline, and animals whose performance was more than 2 standard deviations outside of the cohort mean were withdrawn.

FIGURE 1.

A, Experimental schedule. Surgeries of stroke induction and electrode implantation were performed at week 0. Before that, as prestroke weeks, rats receive training on pasta matrix task. Poststroke baseline of pasta matrix performed was recorded at poststroke week 1 and 5. BrdU injection and electric stimulation were started at poststroke week 6. Stimulation was continued for 4 wk, namely W1-4. Afterward, rats were terminated and brain tissue was collected for IHC analysis. B, Locations for quantification of immunohistochemistry. BrdU/Nissl and BrdU/DCX-colabeled cells in the fixed area (highlighted with boxes) at the ventral perilesional site to stroke core (highlighted in red), MD and VL thalami were counted. BrdU/VGluT1 and BrdU/GAD colabeled cells in the fixed area at the medial perilesional site to stroke core were counted. © 2015 CCF. Used with permission.

Animals then underwent surgical stroke induction and electrode implantation followed by a 1-wk recovery. Pasta matrix performance was assessed for an additional week to confirm a functional deficit. One animal was withdrawn for failing to exhibit decreased performance compared to its prestroke baseline and a second due to the excessive severity of its deficit. The remaining 21 animals were assigned to 1 of 2 treatment groups: no stimulation (STIM⁻; n = 10) and stimulation (STIM⁺; n = 11). Assignments were made pseudorandomly (severity matching) based on mean poststroke pasta matrix performance. Thereafter, animals were provided free access to food and water for an additional 3 wk without stimulation or assessment. Animals resumed pasta matrix testing starting the fifth week poststroke and performance during that and the first poststroke week were used to index poststroke baseline severity.

Motor thresholds for cerebellar stimulation were determined for STIM⁺ animals during the fifth poststroke baseline week as previously reported.⁷ Briefly, stimulus amplitude was increased in a stepwise fashion until a reproducible motor response, typically in the ipsilateral forepaw, was noted (ie, motor threshold). Stimulation was then initiated in STIM⁺ animals beginning the sixth week poststroke and consisted of intermittent high-frequency (100 Hz) trains of square-wave pulses (burst) superimposed on a 30 Hz nonisochronous baseline, mimicking a Hebbian facilitation paradigm.^7,9,11 Each burst was 100 ms in duration, with the interburst interval randomly derived from a Gaussian distribution of 900 ± 71 ms (mean ± standard deviation). Stimulus amplitude for both the 30 Hz and burst stimuli was set individually to 80% of the motor threshold (group mean: 100 ± 30 μA; pulse width: 400 μs) and delivered for 12 h per day (dark phase) for 4 wk. Animals assigned to the STIM^– group were similarly tethered, but received no electrical stimulation. All animals were injected with 5-bromo-2’-dexoyuridine (50 mg/kg, i.p. [BrdU, MP Biomedical, Santa Ana, California]) daily for a total of 5 d starting on the first day of stimulation in order to label newly divided cells, as we speculated that neurogenesis would occur predominantly in the earlier phases of stroke recovery. At the conclusion of the treatment phase, an electrolytic lesion was created at the electrode tip by DC stimulation (+100 μA, 30 s) for histological verification.

Preparation of Brain Sections

Rats were deeply sedated (100 mg/kg Nembutal) and decapitated. Brains were removed and fixed in 4% phosphate-buffered paraformaldehyde (EMS, PA) for 10 d. After cryoprotection in 30% sucrose/phosphate-buffered saline (PBS), brain blocks were snap-frozen and stored at –80°C until slicing with cryostat. Thirty micrometer sections of cerebral cortex and cerebellum were mounted on polysine-coated slides.

Verification of Electrode Implantation in the LCN

LCN electrode localization was confirmed by Perls staining. Sections were incubated with a mixture of 4% potassium ferrocyanide and 0.1% HCl (Sigma, St. Louis, Missouri) for 30 min. The iron deposit along the electrode tract was visualized by incubating in 3,3΄-diaminobenzidine (Vector laboratories, Burlingame, California) followed by Nissl counter-staining (1% cresyl violet; Sigma), and the tips of the LCN lead track were located and coregistered to the rat brain atlas.¹²

Measurement of Lesion Volume

Every fourth coronal section containing the ischemic lesion was Nissl stained, and stroke volume was calculated as described previously.^12-13 Briefly, the lesion core of the cerebral sections was highlighted at 1×, the area of lesion was quantified using SLICE software,¹³ and volume was computed according to the intersection distance (160 μm). Thirty to 35 sections of each animals were included in the analysis.

Immunohistochemical Analysis

In order to identify and distinguish newly divided cells in the perilesional cortex, mediodorsal (MD), and ventrolateral (VL) thalamic nuclei, dual fluorescent immunohistochemistry (IHC) for specific marker proteins was performed. Newly divided neurons were characterized by BrdU and the expression of Nissl bodies, a marker for neuronal cells. To characterize neural progenitor cells (NPCs), colocalization of doublecortin (DCX) and BrdU was examined. Briefly, sections were treated initially with 1N HCl to expose BrdU antigens in the genomic chromatin. To ensure antigens were not hindered due to overfixation, they were retrieved by treating sections with hot (90°C-100°C) citrate buffer (10 mM sodium citrate and 0.05% Triton X-100, pH 6.0) for 30 min. Following blockage of nonspecific binding sites by 5% normal goat serum (Millipore) in 0.5% Triton X-100 containing PBS, sections were incubated with mouse anti-BrdU (1:500; Millipore) and primary antibodies of NPCs (guinea pig anti-DCX; 1:500; Millipore), glutamatergic (rabbit anti-vesicular glutamate transporter 1/VGluT1; 1:1000; Abcam, Cambridge, United Kingdom), and GABAergic (rabbit anti-glutamate decarboxylase/GAD; 1:1000; Millipore) markers overnight. The next day, sections were incubated with fluorescein isothiocyanate (FITC) anti-mouse IgG (1:500; for BrdU, Millipore) and Alexa Fluor633 anti-rabbit IgG (1:500; for VGluT1 and GAD, Life technologies, NY) for 2 h at room temperature. For double staining with antibodies of BrdU and DCX, Alexa Fluor610 antimouse IgG for BrdU (1:500; Life Technologies, NY) and Alexa Fluor488 anti-guinea pig IgG (1:500; Life Technologies, Carlsbad, California) for DCX were utilized. Fluorescent Nissl staining (1:300; Life Technologies) was also performed after the single immunostaining with anti-BrdU. Fluorescent stained sections were mounted with antifading agents (Vector Laboratories) for confocal microscopic analysis (Leica SP5, Leica, Wetzler, Germany).

Image Analysis

Experimenters were blinded to the identities of the slides in order to prevent bias in the manual counting process. The ipsilesional motor (–0.5 through +3.0 mm anterior/posterior) cortex was visualized at 100×. To avoid counting nonspecific content inside the ischemic core, we visually defined the stroke border on the affected side and counted cells at the most ventral perilesional region (0.09 × 0.09 mm²), which was approximately 0.05 mm down from the ventral border of the lesion (Figure 1B). In these areas, double-labeled cells with BrdU/DCX or BrdU/Nissl were manually counted. Every eighth section (320 μm) was sampled, with 15 to 20 sections analyzed per animal. Images at 630× magnification were used to confirm colocalization of BrdU and protein markers. Three-dimensional analysis was performed by z-stacking with a focal plane of 0.2 μm to reduce confounds from adjacent cells.

We further evaluated the effects of DBS on the expression of glutamatergic (VGluT1) vs GABAergic (GAD) markers across the individual laminar layers of perilesional motor cortex. Cortical layers were defined as thicknesses of I: 210 μm, II: 230 μm, III: 180 μm, IV: 180 μm, V: 520 μm, and VI: 550 μm.¹⁴ BrdU/VGluT1+ or BrdU/GAD+ cells in layers II, III/IV, and V/VI were counted in an area (0.09 × 0.09 mm²) 0.1 mm away from the lesion border of each section (Figure 1B). Since VGluT1 and GAD signals were detected in both neuronal bodies and processes, only the cell bodies with colocalization of either VGluT1 or GAD and BrdU were counted. Distribution of BrdU/VGluT1+ and BrdU/GAD+ cells in different laminar layers was quantified at 100×.

Because both MD and VL thalamic nuclei receive mostly contralateral projections from the LCN and project ipsilaterally to the cortex, we also examined neurogenesis in these thalamic nuclei. Each thalamic area was defined at a fixed size of 0.09 × 0.09 mm² and examined at 100×. Colocalization was visualized at 630×.

Statistics

BrdU/Nissl and BrdU/DCX counts were computed as densities (cells/mm² ± standard error mean (SEM)) and analyzed using Student's t-test. Glutamatergic and GABAergic neurogenesis were computed similarly and analyzed by 2-way analysis of variances (ANOVAs) with Bonferroni post hoc test in Prism software (GraphPad Prism, GraphPad, La Jolla, CA). Behavioral data were analyzed with repeated 2-way ANOVAs with Bonferroni post hoc tests. Differences were considered as significant at P < .05, P < .01, or P < .0001 as annotated in each figure.

RESULTS

Verification of LCN Electrode Placement and Lesion Volume Estimates

Eighteen animals were found to have the electrode tip well situated within the dorsal part of the LCN (Figure 2A), with only 2 animals excluded for improper placement. Stroke volume was 11.61 ± 0.49 mm³ in the STIM⁺ group and 11.49 ± 0.98 mm³ in the STIM⁻ group (P = .69, ns Figure 2B). One additional animal was excluded due to infection. In total, data from 8 and 10 animals were analyzed from the STIM⁻ and STIM⁺ groups, respectively.

FIGURE 2.

Verification of electrode implantation, endothelin 1-induced lesion volume and effect of chronic LCN DBS on the motor recovery. A, The locations of the implanted electrodes in the dorsal LCN. B, The lesion volumes (mm³ ± SEM) in both STIM⁻ and STIM⁺ groups were comparable. C, Performance on the pasta matrix task. Stimulation was initiated at week 1, 6 wk poststroke, and continued over 4 wk. Data are presented as pasta pieces retrieved ± SEM. *P < .05 and ***P < .0001 vs poststroke baseline.

Effects of Chronic LCN DBS on Motor Recovery

Both groups demonstrated a significant decline in poststroke pasta matrix performance (Figure 2C; STIM⁻: P < .05; STIM⁺: P < .0001), consistent with our prior findings.^7-8 After 4 wk of treatment (or sham), STIM⁺ animals improved by an average of 41.4% while STIM⁻ animals improved by 27.4%. The 2-way ANOVA revealed a significant difference between groups (Figure 2C; F = 2.20, Df = 1, P < .05) and across the treatment time (Figure 2C; F = 7.16, Df = 6, P < .00001). Moreover, an interaction effect was identified (Figure 2C; F = 5.41, Df = 6, P < .05); however subsequent post hoc testing did not reveal specific differences across each of the 4-wk periods (W1-4; W4: P = .068), likely indicating that the study was simply underpowered from a behavioral standpoint. Overall, the magnitude of the observed effect and its direction are consistent with our previous publications.^7-8

Effects of Chronic LCN DBS on Cortical Cell Proliferation

Cortical Neurogenesis

A significant increase in both BrdU/Nissl+ (P < .0001; Figures 3A and 3B) and BrdU/DCX+ (P < .0001; Figures 3C-3D) cells was found at the perilesional area in STIM⁺ compared to STIM⁻ rats, suggesting enhanced neurogenesis in treated animals.

FIGURE 3.

Neurogenic effect of LCN DBS in the perilesional motor cortex. Chronic LCN DBS was associated with increased density of A, BrdU/Nissl+ and C, BrdU/DCX+ cells in the perilesional motor cortex in STIM⁺. Scale bar = 100 μm (100×) and 16 μm (630×). Data were expressed as density of B, BrdU/Nissl+ and D, BrdU/DCX+ cells/mm² ± SEM. ***P < .0001 vs STIM⁻.

Cortical Perilesional Cytoarchitecture: Glutamatergic and GABAergic

We observed a significant effect of LCN DBS on the proliferation of BrdU/VGluT1+ cells at layers II, III/IV, and V/VI in the perilesional motor cortex (F = 157.8, Df = 1; P < .0001; II: P < .0001, III/IV: P < .0001, and V/VI: P < .01; Figures 4A and 4B), suggesting that DBS induced glutamatergic neurogenesis in the perilesional zone. Notably, a significant decrease in BrdU/GAD+ cells at layers II, III/IV, and V/VI was found in STIM⁺ animals compared to STIM⁻ (F = 225.7, P < .0001; II: P < .0001, III/IV: P < .0001, and V/VI: P < .01; Figures 4C and 4D).

FIGURE 4.

Effect of LCN DBS-induced glutamatergic and GABAergic arrangement at different cortical layers of the perilesional area. Chronic LCN DBS was associated with A, an increase of BrdU/VGluT1+ and C, a decrease of BrdU/GAD+ density in layers II, II/IV, and V/VI of the perilesional zone. Scale bar = 100 μm (100×). Data were expressed as density of B, BrdU/VGluT1+ or D, BrdU/GAD+ cells/mm² ± SEM. **P < .01 and ***P < .0001 vs STIM⁻.

Effects of Chronic LCN DBS on Thalamic Neurogenesis

Consistent with the cortical findings, a significant treatment-related increase in BrdU/Nissl+ (P < .0001; Figures 5A and 5B) and BrdU/DCX+ (P < .0001; Figures 5C and 5D) cells was found in the ipsilesional VL thalamic nuclei. Similarly, there was a significant increase in BrdU/Nissl+ (P < .0001; Figures 6A and 6B) and BrdU/DCX+ (P < .0001; Figures 6C and 6D) cells in the ipsilesional MD thalamus. These findings suggest that LCN DBS promoted thalamic neurogenesis.

FIGURE 5.

Neurogenic effect of LCN DBS on VL thalamic nucleus. Chronic LCN DBS increased the density of A, BrdU/Nissl+ and C, BrdU/DCX+ cells at the ipsilesional VL thalamic nucleus. Scale bar = 100 μm (100×) and 16 μm (630×). Data were expressed as density of B, BrdU/Nissl+ and D, BrdU/DCX+ cells/mm² ± SEM. ***P < .0001 vs STIM⁻.

FIGURE 6.

Neurogenic effect of LCN DBS on MD thalamic nucleus. Chronic LCN DBS increased the density of A, BrdU/Nissl+ and C, BrdU/DCX+ cells at the ipsilesional MD thalamic nucleus. Scale bar = 100 μm (100×) and 16 μm (630×). Data were expressed as density of B, BrdU/Nissl+ and D, BrdU/DCX+ cells/mm² ± SEM. ***P < .0001 vs STIM⁻.

DISCUSSION

We have previously shown that LCN DBS that is effect at enhancing motor recovery promotes significant trans-synaptic enhancement of excitability and synaptogenesis in the motor cortex.^7-9 In light of those findings, we hypothesized that the rehabilitative effects of stimulation might similarly be associated with neurogenesis across the CTC pathway. The CTC pathway is a di-synaptic excitatory pathway projecting from the LCN to the motor, premotor, and parietal cortices via specific and nonspecific thalamic subnuclei.^15-16 In particular, the LCN projects to VL thalamus, which has extensive and well-established projections to the motor cortex,^17-19 and to MD thalamus, which projects more diffusely across ipsilateral cortex.¹⁵ Our current results support our hypothesis, and we have shown, for the first time, that LCN DBS promotes poststroke neurogenesis not only in the perilesional cortex but also in VL and MD thalamic nuclei. Our findings suggest that LCN DBS has significant effects along the CTC pathway that parallel its facilitative effect on poststroke motor recovery.

Neurogenesis can occur spontaneously after stroke, as part of the perilesional reorganization associated with injury and apoptosis.²⁰ BrdU-labeled cells can be detected up to 3 to 4 wk following injection,^21,22 consistent with our detection of BrdU-labeled neurons 4 wk after BrdU injection. DCX-labeled cells are not typically found in the adult neocortex²³ but have been reported in the perilesional cortex during the first 2 to 5 wk after stroke.^24,25 In this study, animals received BrdU injection 6 wk after stroke, at a time when the acute effects of injury are likely resolved. Interestingly, a significant increase in both BrdU/DCX and BrdU/Nissl-labeled cells was found in the perilesional cortex of stimulated compared to control animals. The data suggest that LCN DBS promotes renewed plasticity of the perilesional region in the chronic phase, possibly extending or re-establishing the endogenous timeline of neuronal proliferation after ischemia. Our finding of neurogenesis in the adult thalamic nuclei is interesting, albeit somewhat controversial, given that the diencephalon is not a habitual site of neurogenesis.

Other groups have previously reported possible neurogenic effects, including an increase in NPCs in the subventricular zone (SVZ), associated with STN DBS in patients with Parkinson's disease.²⁶ Furthermore, the application of local electric fields has been shown to promote differentiation and migration of NPCs in Vitro.²⁷ Here, we have shown that chronic stimulation not only can promote neurogenesis after a stroke but can do so in later phases of recovery. Furthermore, we found that neurogenesis can be promoted trans-synaptically, at a distance from the LCN, possibly in response to chronic activation of the CTC pathway.

The design of the current study was limited, and functional contribution of the identified cells in regards to promoting neural transmission along the CTC pathway or behavioral recovery has yet to be established. Furthermore, we did not detect BrdU-labeled NPCs migrating from SVZ or subgranular zone to the perilesional motor cortex 4 wk after stimulation or 10 wk after stroke induction. A future study design that includes additional sampling time points during the stimulation period might further support the causal nature of the observed neurogenesis in relation to motor recovery. It is also unknown when newly divided neurons or NPCs begin to express glutamatergic or GABAergic markers. As the finding of GAD and DCX co-expression in spinal cord neurons of mice²⁸ indicates that immature neurons can express a GABAergic marker, we speculate that the BrdU/GAD+ neurons illustrated in this study may be a mixture of mature and immature GABAergic neurons. We also believe that the BrdU/VGlut1+ neurons identified in this study may be a mixture of mature and immature glutamatergic neurons. In order to clarify these remaining questions, a time course study should be performed to allow a longitudinal observation of the process of differentiation/maturation of NPCs following poststroke LCN stimulation. Furthermore, the functional contribution of the limited number of neurogenic cells could be investigated along with other mechanisms of motor recovery previously studied by our group, including synaptogenesis and reorganization of cortical representation. Another future opportunity is to evaluate the effects of DBS on gliogenesis, as it is well established that BrdU may also label other dividing cell types, such as astrocytes, oligodendrocytes, or microglia.^29-31

A striking observation within our study is the differential effect of LCN DBS on glutamatergic and GABAergic neurons. As noted above, neurogenesis is detected in the perilesional cortex at the early stages of ischemic stroke;^24-26 however, this natural postinjury neurogenesis is not well balanced across all neuronal types. Rather, it favors the formation of GABAergic neurons in the postischemic striatum³² and perilesional cortex following ischemia,³³ changes that are not seemingly conducive to successful rehabilitative outcomes. In fact, activation of GABAergic neurons in the sensorimotor cortex has been linked to poor motor recovery after stroke in rodents,³⁴ while inhibition of GABAergic outputs is thought to improve practice-dependent recovery.³⁵ Our results are interesting in that LCN DBS was associated not only with a reduction of GABAergic neurogenesis compared to controls but also enhanced glutamatergic neurogenesis. This leads us to speculate that chronic stimulation of the CTC pathway may promote a reparative rebalancing of the glutamatergic/GABAergic cytoarchitecture in the perilesional cortex. Although the current findings are not sufficient to establish causality between the modulation of poststroke neurogenesis and motor recovery, they suggest a strong association between stimulation-related motor gains and selective glutamatergic and GABAergic neurogenesis promotion and suppression, respectively.

CONCLUSION

In this study, it is demonstrated chronic LCN DBS is associated with changes in neurogenesis and cytoarchitecture. As DBS is already a proven, safe technique in humans, there is significant translational potential in using it to gain control over neurogenesis in the management of stroke.

Disclosures

This research was funded by The Charles and Christine Carroll Family Endowed Chair in Functional Neurosurgery, and National Institute of Health (NIH) (R01 HD061363). Dr Machado and Dr Baker have potential financial conflict of interest with this research related to intellectual property and distribution rights in Enspire, ATI, and Cardionomics. Dr Machado is a consultant to St Jude and Functional Neuromodulation and receives fellowship support from Medtronic. The Cleveland Clinic Conflict of Interest (COI) committee has approved a plan for managing these conflicts of interest. The authors have adhered to the management plan in the conduct and reporting of research findings. None of these entities had any role in the research or preparation of the manuscript. The other authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.

COMMENTS

Motor deficits from ischemic stroke frequently exhibit some recovery over time, so neuromodulation approaches that target these mechanisms may represent a therapeutic option. In this study, the authors applied DBS to the lateral cerebellar nucleus (8 rats, intermittent 100 Hz trains superimposed on continuous 30 Hz stimulation, 80% motor threshold, 12 hours per day for 4 weeks starting 6 weeks after stroke) and compared this with sham stimulation (10 rats). They documented 50% improvement in motor recovery that correlated with increased neuronal proliferation in perilesional motor cortex as well as bilateral mediodorsal and ventrolateral thalamic nuclei. Significantly more glutamatergic and less GABAergic cell proliferation was seen in the group exposed to stimulation. The authors conclude that post-stroke lateral cerebellar nucleus stimulation is associated with chronic changes in neurogenesis and cytoarchitecture and this may represent a novel therapeutic approach.

Simulation of deep cerebellar nuclei after stroke has been shown to alter cortical excitability, synaptogenesis, and long-term potentiation, but previous studies have largely focused on immediate stimulation effects rather than long-term changes. The observation that chronic stimulation leads to neurogenesis even after the acute injury has resolved suggests that this therapy may work by multiple mechanisms. Notably, this study evaluated only limited areas of the brain, so it is not clear that the observed changes are specific to motor pathways. Nevertheless, the results provide support for the concept that neuromodulation might represent a treatment option for stroke.

Jonathan P. Miller

Cleveland, Ohio

The authors state that they have “shown, for the first time, that LCN DBS promotes post-stroke neurogenesis not only in perilesional cortex but also in VL and MD thalamic nuclei.” I respectfully disagree with this conclusion, on the premise that basic standards of fluorescent immunohistochemistry related to the study of neurogenesis have not been met. Single-time-point experiments that employ only BrdU and 1 other marker simply are not sufficient for demonstrating that a cell is both neuronal and newborn. The absence of figure panels showing separate, single-channel fluorescent images also precludes the reader from adequately assessing the data. A more careful interpretation of these results is warranted, given how remarkable it would be if LCN DBS did promote neurogenesis in remote, typically non-neurogenic brain regions.

Robert Mark Richardson

Pittsburgh, Pennsylvania