Effects of vagus nerve stimulation are mediated in part by TrkB in a Parkinson’s disease model

Abstract

Vagus nerve stimulation (VNS) is being explored as a potential therapeutic for Parkinson’s disease (PD). VNS is less invasive than other surgical treatments and has beneficial effects on behavior and brain pathology. It has been suggested that VNS exerts these effects by increasing brain-derived neurotrophic factor (BDNF) to enhance pro-survival mechanisms of its receptor, tropomyosin receptor kinase-B (TrkB). We have previously shown that striatal BDNF is increased after VNS in a lesion model of PD. By chronically administering ANA-12, a TrkB-specific antagonist, we aimed to determine TrkB’s role in beneficial VNS effects for a PD model. In this study, we administered a noradrenergic neurotoxin, DSP-4, intraperitoneally and one week later administered a bilateral intrastriatal dopaminergic neurotoxin, 6-OHDA. At this time, the left vagus nerve was cuffed for stimulation. Eleven days later, rats received VNS twice per day for ten days, with daily locomotor assessment. Daily ANA-12 injections were given one hour prior to the afternoon stimulation and concurrent locomotor session. Following the final VNS session, rats were euthanized, and left striatum, bilateral substantia nigra and locus coeruleus were sectioned for immunohistochemical detection of neurons, α-synuclein, astrocytes, and microglia. While ANA-12 did not avert behavioral improvements of VNS, and only partially prevented VNS-induced attenuation of neuronal loss in the locus coeruleus, it did stop neuronal and anti-inflammatory effects of VNS in the nigrostriatal system, indicating a role for TrkB in mediating VNS efficacy. However, our data also suggest that BDNF-TrkB is not the sole mechanism of action for VNS in PD.

1. INTRODUCTION

Parkinson’s disease (PD), afflicts 7-10 million people worldwide [1] and is clinically characterized by motor, as well as non-motor symptoms. Neuropathologically, PD leads to early degeneration of locus coeruleus noradrenergic neurons (LC-NE neurons), followed by degeneration of substantia nigra dopaminergic neurons (SN-DA neurons). Another pathological hallmark of PD is accumulation of α-synuclein-positive aggregates known as Lewy bodies within remaining neurons [2]. This degeneration in the SN-DA system leads to development and progression of motor symptoms [3]. Several key mechanisms have been implicated in PD-related neurodegeneration, including α-synuclein aggregation, neuroinflammation, and decreased neurotrophic factor signaling [4,5,6].

Brain-derived neurotrophic factor (BDNF) is a neurotrophic factor important for the survival and maintenance of neurons, including DA and NE cell populations [7,8]. BDNF binds to its receptor tropomyosin receptor kinase-B (TrkB), resulting in auto-phosphorylation of several sites on the intracellular tyrosine kinase domain of the receptor. The phosphorylated TrkB receptor complex then induces neuroprotection by activating pro-survival signaling via the mitogen-activated protein kinase (MAPK), phosphoinositol-3-kinase/protein kinase B (PI3K/Akt), and phospholipase-C-γ pathways [9].

Postmortem human PD brains show reduced BDNF protein in the SN [10], and some evidence suggests that genetic polymorphisms resulting in decreased BDNF may increase the risk of developing PD [11,12]. Additionally, administration of TrkB agonists and infusion of BDNF can protect against 6-hydroxydopamine (6-OHDA)-induced nigrostriatal lesion in rodents, further highlighting the importance of BDNF for the integrity of the nigrostriatal DA system [13,14]. Furthermore, a genetic reduction of TrkB protein in mice increases α-synuclein aggregation in SN-DA neurons, likely as a result of impaired mitochondrial energy production, thereby showing a potential role for BDNF-TrkB in α-synuclein aggregation [15]. In addition, intraventricular BDNF pretreatment reduces pro-inflammatory cytokine production in rat models of stroke and meningitis, suggesting that BDNF has the ability to reduce neuroinflammation [16,17]. Administration of fingolimod, an immunosuppressive drug used to treat multiple sclerosis, increases BDNF in mice, reduces aggregation of α-synuclein in a PD model [18], and reduces neuroinflammation in a multiple sclerosis model [19]. Combined, results from these studies suggest a role of BDNF-TrkB in α-synuclein aggregation and neuroinflammation, mechanisms that are relevant in PD-related neurodegeneration [5].

Although there are no clinical treatments that slow the progression of PD, a preclinical study from our lab has demonstrated that vagus nerve stimulation (VNS) may slow PD-related neurodegeneration in a double lesion rat model, as evidenced by improved locomotor activity and beneficial effects on LC and SN neuronal survival, with reduced neuroinflammation in both of these regions [20]. While the precise mechanism of action for VNS remains unknown, BDNF seems to play a key role. Concentrations of BDNF are increased in the frontal cortex and hippocampus (targets of the LC) of normal rats after 1 month of VNS compared to non-stimulated rats [21]. Further, BDNF signaling is increased in the hippocampus after 14 days of VNS in a rat model of depression compared to non-stimulated counterparts, thereby demonstrating that VNS activates these BDNF-related neuroprotective pathways [22,23]. In a double lesion model of PD, we have reported that 10 days of twice daily VNS results in higher BDNF levels in both the dorsal striatum and the frontal cortex [24]. Therefore, the focus of the current study is to investigate whether VNS utilizes BDNF-TrkB to produce beneficial behavioral and neuronal effects for PD.

In this study, we inhibit BDNF binding to its receptor, TrkB, to test BDNF-TrkB as a potential mechanism of action for VNS. We systemically inject N-[2-[(2-oxoazepan-3-yl)carbamoyl]phenyl]-1-benzothiophene-2-carboxamide (ANA-12), a TrkB-specific antagonist that readily crosses the blood brain barrier [25], prior to afternoon VNS and locomotor sessions to determine whether TrkB inhibition can prevent the beneficial effects of VNS on behavior, neurons, and neuroinflammation previously observed in our PD model.

2. MATERIALS AND METHODS

2.1. Animals

For these experiments, all procedures were approved by the Institutional Animal Care and Use Committee at the Medical University of South Carolina (MUSC). Adult male Long Evans rats from Charles River Laboratories (200-225g) were housed at MUSC, with free access to food and water, on a 12 hr normal light:dark cycle. The animal facility was kept at 20-22°C and is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Rats for all experiments were divided into the following groups: saline (non-lesion) nonVNS vehicle (n=5), saline nonVNS ANA-12 (n=6), saline +VNS vehicle (n=5), saline +VNS ANA-12 (n=8), lesion nonVNS vehicle (n=5), lesion nonVNS ANA-12 (n=6), lesion +VNS vehicle (n=5), and lesion +VNS ANA-12 (n=6).

2.2. Surgical procedures

In order to mimic the pathology of PD in the LC and SN, we utilized a double lesion model as previously described (Figure 1) [20]. Briefly, a single injection of either sterile saline (0.9% sodium chloride, Hospira) or N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine (DSP-4, 50 mg/kg, intraperitoneal, Sigma) was administered to induce an LC-NE lesion. Seven days later, rats were anesthetized with isoflurane (5% for induction, plane of anesthesia maintained at 2-3%, Piramal Healthcare), and bupivacaine (0.1 mg/kg, subcutaneous, Hospira) was injected at skull and cervical incision sites. VNS cuff electrodes (<10 kOhms impedance) were implanted around the isolated left cervical vagus nerve for VNS rats and leads tunneled subcutaneously to the skull incision [26,27]. All rats then received bilateral intrastriatal 0.9% sodium chloride or 6-OHDA (6 μg/μL, 2 μL/site, made with 0.02% ascorbate, Sigma) via Hamilton syringe (SGE) using a stereotaxic frame (Stoelting) to induce nigrostriatal lesion at the following coordinates: Hole 1: AP +1.6, ML ±2.4, DV −4.2, Hole 2: AP +0.2, ML ±3.7, DV −5.0 [28]. The syringe was left in place for 5 min after each injection to maximize absorption into the tissue before slowly retracting. The leads from the VNS cuff were connected to a two-channel headcap that was secured to the skull with 4 bone screws and dental cement (Lang Dental).

Fig. 1. Experimental design timeline.

Rats received either saline (0.9% sodium chloride) or DSP-4 (50 mg/kg, intraperitoneal) to induce noradrenergic lesion. After 7 days, rats had surgery to implant the vagus cuff and headcap and to induce striatal dopaminergic lesion with 6-OHDA (6 μg/μL, 2 μL/site). After allowing 11 days for lesion development, rats began two VNS sessions per day for 10 days. Additionally, rats received either vehicle (17% DMSO in saline) or ANA-12 (0.5 mg/kg, intraperitoneal) 1 h prior to the afternoon VNS session. Locomotor activity (LA) of all rats was assessed for the first 10 min of the PM VNS session each day for all 10 days of stimulation. After the locomotor assessment on day 10, rats were euthanized so that brains could be processed for immunohistochemistry (IHC).

2.3. VNS protocol and drug administration

Beginning 11 days post-surgery, rats underwent 30 min VNS sessions twice per day for 10 days, as previously described [20], with daily sessions separated by 4 hr. Rats were plugged into the stimulator (A-M Systems) via the headcap connector and were allowed to move freely in a locomotor box (Omnitech) throughout the session. Each session consisted of 500-ms pulse trains delivered every 30 sec throughout the 30 min. Each pulse train was given at 30 Hz and was comprised of 15 biphasic pulses lasting 100 μs at a magnitude of 0.8 mA. These parameters have been previously shown to stimulate the whole cervical nerve bundle inside the vagus cuff [26] and have therapeutic effects in this double lesion PD model [20]. NonVNS rats were also placed in the locomotor box twice per day but did not receive stimulation. To assess whether VNS effects are dependent on binding of BDNF to TrkB, rats received daily injections of either ANA-12 (0.5 mg/kg in 17% dimethyl sulfoxide (DMSO) in sterile saline, ip, Sigma) or vehicle (17% DMSO in sterile saline, ip) 1 hr prior to the afternoon VNS and locomotor activity session. This dose of ANA-12 has been used acutely to sufficiently inhibit BDNF binding to TrkB without causing neuronal toxicity [25].

2.4. Locomotor activity

Locomotor activity (total distance traveled, in cm) was measured in Digiscan photobeam chambers (Omnitech) during the rats’ light cycle according to previous protocols [20,29]. Locomotor activity was recorded for the first 10 min of the 30 min afternoon VNS session. Data are reported both as a group average of cm traveled per day for each of the 10 stimulation days, as well as a collapsed mean distance traveled per group across the 10 days.

2.5. Immunohistochemistry

Bilateral SN/LC and left striatal tissue blocks were extracted from each animal and post-fixed in 4% paraformaldehyde for 48 hr, followed by 30% sucrose for at least another 48 hr. Tissue blocks were sectioned for immunostaining as previously described at 45 μm [20]. Serial sections for immunohistochemistry from the left striatum (every 12^th) used a rabbit polyclonal antibody for tyrosine hydroxylase (TH) to stain DA terminals (1:1000, Pel-freez). Serial sections from the SN (every 6^th) used rabbit polyclonal antibodies for either TH to stain DA neurons (1:1000, Pel-freez), glial fibrillary acidic protein (GFAP) to stain astrocytes (1:2000, Dako), or ionized calcium-binding adaptor molecule-1 (Iba-1) to stain microglia (1:1000, Wako) according to our standard protocol [20,29]. Briefly, tissue was washed in a 10% peroxide, 20% methanol in 0.01 M tris-buffered saline (TBS) solution to stop endogenous peroxidases, rinsed in TBS, then treated with sodium meta periodate in TBS to enhance permeability of the tissue. Tissue was then rinsed in TBS with 0.1% Triton-X-100 (TBST) and incubated in 10% normal goat serum made in TBST for 1 hr to block nonspecific binding prior to incubation in primary antibody overnight at the above dilutions. Tissue was rinsed in TBS with 3% normal goat serum the following day and incubated in biotinylated secondary antibody (goat anti-rabbit 1:200, Vector) for 1 hr, then rinsed in TBS and incubated in avidin-biotin complex solution (ABC kit, Vector) for 1 hr. After another rinsing step using TBS, a color reaction was developed using a VIP peroxidase substrate kit (Vector), which was stopped using TBS. Sections were mounted on glass slides coated with gelatin to enhance tissue binding to the slide, then dehydrated and coverslipped using Permount (Sigma).

2.6. Immunofluorescent staining

Immunofluorescent staining was conducted on serial sections in the LC (every 3^rd) using a sheep polyclonal antibody for TH (1:1000, Chemicon). For the striatum, serial sections (every 12^th) were stained using rabbit polyclonal antibodies for Iba-1 (1:1000, Wako) or GFAP (1:2000, Dako). Serial sections from the SN (every 6^th) were co-stained with sheep TH (1:2500, Chemicon) and rabbit α-synuclein (1:250, Cell Signaling) polyclonal antibodies as previously described [20]. Briefly, tissue was blocked in 10% normal donkey serum made in 0.1 M phosphate-buffered saline (PBS) + 0.3% Triton-X-100 for 1 hr to prevent nonspecific binding. Sections were then incubated in primary antibody overnight. The following day, tissue was rinsed in PBS + 0.05% Tween-20, then incubated in secondary antibody for 1 hr (tetramethylrhodamine (TRITC)-conjugated donkey anti-sheep for TH; fluorescein isothiocyanate (FITC)-conjugated donkey anti-rabbit for α-synuclein, Iba-1, and GFAP; both secondaries diluted at 1:200; Jackson ImmunoResearch). Sections were rinsed in PBS before mounting on gelatin-coated slides and coverslipped with ProLong Gold (Invitrogen).

2.7. Semiquantitation for immunoreactivity (ir) of TH, GFAP, and Iba-1

In the striatum, TH-ir (ROI outline: Figure 4A, images taken at 4x magnification) was conducted to assess whether ANA-12 could prevent survival of DA terminals after VNS. In the striatum and SN, Iba-1-ir (ROI outline: Figure 6A,J, images taken at 20x and 10x magnifications, respectively) and GFAP-ir (ROI outline: Figure 7A,J, images taken at 20x and 10x magnifications, respectively) assessed microgliosis and astrogliosis, respectively, to determine the effects of ANA-12 on neuroinflammation in the nigrostriatal system. All of these images were analyzed using a QImage R3 camera system (QImaging) with Image J software (NIH) [20,30]. Density values were measured in 16-bit images on a scale of 0-65535, where 0=white and 65535=black for brightfield staining (striatal TH, SN GFAP, and SN Iba-1), and 0=black and 65535=green for fluorescent staining (striatal GFAP and striatal Iba-1). Background measures from each section were subtracted from mean staining intensity. Corrected values from each section (4-6 sections per animal) were averaged to obtain a mean intensity per animal.

Fig. 4. Attenuation of nigrostriatal loss by VNS was prevented with ANA-12.

Measurement outlines are shown for the striatum (A) and SN (J). Photomicrographs of striatum are shown in A-H (scale = 500 μm, quantified in I). Saline treated rats had comparative TH-ir across treatment groups (A-B, E-F). TH-ir was significantly lower in lesion nonVNS vehicle rats compared to saline nonVNS vehicle rats (A, C, ****p < 0.0001). Lesion +VNS vehicle rats had greater TH-ir than lesion nonVNS vehicle rats (C-D, *p < 0.05), though this was only a partial increase compared to saline +VNS vehicle rats (B, D, **p < 0.01). This VNS effect was prevented with ANA-12 in the lesion +VNS group (D, H, *p < 0.05). Photomicrographs of SN are shown in J-Q (scale = 250 μm, quantified in R). Lesion nonVNS vehicle rats had significantly fewer TH-positive cells than saline nonVNS vehicle rats (J, L, ****p < 0.0001). Lesion +VNS vehicle rats had increased TH-positive cells compared to lesion nonVNS vehicle rats (L-M, **p < 0.01). TH-positive cells in the SN of lesion +VNS vehicle rats remained lower than saline +VNS vehicle rats (K, M, *p < 0.05). ANA-12 administration resulted in fewer TH-positive cells for lesion +VNS rats only (M, Q, ****p < 0.0001).

Fig. 6. Effects of VNS on microglia are prevented with ANA-12 in the striatum and SN.

White circles denote ROI measurement for the striatum (A) and SN (J). Photomicrographs of dorsal striatum are shown in A-H (scale = 100 μm, quantified in I). Lesion nonVNS vehicle rats had increased striatal Iba-1-ir compared to saline nonVNS vehicle rats (A, C, ****p < 0.0001). Striatal Iba-1-ir was similar in saline +VNS vehicle compared to saline nonVNS vehicle rats (A–B), while striatal Iba-1-ir was lower in lesion +VNS vehicle rats compared to lesion nonVNS vehicle rats (C-D, ***p < 0.001). Striatal Iba-1-ir was increased after ANA-12 in lesion +VNS rats only (E–H, *p < 0.05). Photomicrographs of the SN are shown in J-Q (scale = 250 μm, insets taken at 60x, quantified in R). Lesion nonVNS vehicle rats had increased Iba-1-ir in the SN compared to saline nonVNS vehicle rats (J, L, ****p < 0.01). SN Iba-1-ir was similar after VNS in saline rats (J-K) while Iba-1-ir was lower in lesion +VNS vehicle rats compared to lesion nonVNS vehicle rats (L-M, ****p < 0.0001), although lesion +VNS vehicle rats still had increased Iba-1-ir compared to saline +VNS vehicle rats (****p < 0.0001). ANA-12 prevented the SN Iba-1-ir decrease after VNS in lesion rats (M, Q, ****p < 0.0001), and ANA-12 did not alter Iba-1-ir in any other treatment groups (J–L, N–O). Insets demonstrate primarily resting state microglia in saline groups and the lesion +VNS vehicle group (J–K, M–O insets); however, in lesion nonVNS and lesion +VNS ANA-12 rats, most microglia appear to be in an activated state (L, P–Q insets).

Fig. 7. ANA-12 prevented effects of VNS on astrocytes in striatum and SN.

Measurement outlines are shown for the striatum (A) and SN (J). Photomicrographs of dorsal striatum are shown in A-H (scale = 100 μm, quantified in I). Lesion nonVNS vehicle rats had increased striatal GFAP-ir compared to saline nonVNS vehicle rats (A, C, ****p < 0.0001). Striatal GFAP-ir was similar in saline +VNS vehicle compared to saline nonVNS vehicle rats (A–B), while striatal GFAP-ir was lower in lesion +VNS vehicle rats compared to lesion nonVNS vehicle rats (C–D, ****p < 0.0001). Striatal GFAP-ir was increased after ANA-12 in lesion +VNS rats only (E–H, ****p < 0.0001). Photomicrographs of the SN are shown in J–Q (scale = 250 μm, quantified in R). Lesion nonVNS vehicle rats had increased GFAP-ir in the SN compared to saline nonVNS vehicle rats (J, L, ****p < 0.01). SN GFAP-ir was similar after VNS in saline rats (J-K) while GFAP-ir was lower in lesion +VNS vehicle rats compared to lesion nonVNS vehicle rats (L-M, ****p < 0.0001), although lesion +VNS vehicle rats still had increased GFAP-ir compared to saline +VNS vehicle rats (****p < 0.0001). ANA-12 prevented the effects of VNS in the lesion rats for the SN (M, Q, ****p < 0.0001), and ANA-12 did not alter GFAP-ir in any other treatment groups (J–L, N–P).

2.8. Semiquantitation of intrasomal α-synuclein-ir

Density of α-synuclein within TH-positive neurons in the SN was assessed to determine the effects of ANA-12 on intrasomal α-synuclein accumulation, as intrasomal accumulation has been previously observed in both post-mortem PD brains and in animal models of PD [20,31,32]. For this analysis we took Z-stack images from 3-4 sections per animal using a 40x magnification on a BX-61 confocal microscope (Olympus) with Fluoview software (Olympus). Forty-five TH-positive cells per animal were outlined using the TRITC channel, which labeled TH-positive neurons, and corresponding neurons in the FITC channel, which labeled α-synuclein-ir, were identified using ImageJ’s ROI manager function. A mean density measurement of intrasomal α-synuclein-ir was taken within each cell, and background measures from each section were subtracted from the mean density values for each cell. Corrected values from 45 cells per animal were averaged to get a mean α-synuclein density measurement per animal.

2.9. Stereological cell counting

To measure the effects of ANA-12 on TH-positive neuron survival, we calculated the number of TH-positive neurons in the LC (ROI outline: Figure 3A) and SN (ROI outline: Figure 4J) using unbiased, stereological cell counting [20,29]. Cell counts for the LC included the neurons on the lateral edges of the fourth ventricle using the laterodorsal tegmentum as the anterior border and the parvicellular portion of the medial vestibular nucleus as the posterior border. For the SN, counts excluded the SN pars reticulata and ventral tegmental area, so that only the SN pars compacta was counted. Images were taken using Stereoinvestigator software (MicroBrightField) and a BX-61 microscope (Olympus) fitted with an automated headstage for the optical fractionator counting method. A low magnification lense (10x) was used to outline the LC (every 3^rd section) and SN (every 6^th section). Cells were counted in a rostral to caudal manner at 20x with a systematic random sampling of dissector frames (100×100 μm) from 4-6 sections per animal. Average cells per counting frame were 3.50 cells per frame for the LC, and 1.65 cells per frame for the SN. The Stereoinvestigator software then computed an estimated population of cells for each animal.

Fig. 3. ANA-12 partially prevented VNS-induced attenuation of TH-positive LC loss.

Photomicrographs of LC (A-H, scale = 100 μm, quantified in I), measurement outline shown in A. TH-positive cells were significantly lower in the LC of lesion nonVNS vehicle rats compared to saline nonVNS vehicle rats (A, C, ****p < 0.0001). Saline vehicle treated rats had increased TH-positive cells after VNS (A-B, **p < 0.01), and lesion +VNS vehicle rats had greater TH-positive cells than lesion nonVNS vehicle rats (C–D, ****p < 0.0001). Administration of ANA-12 resulted in fewer TH-positive neurons for the lesion +VNS group (D, H, ****p < 0.0001), although numbers were still higher for the lesion +VNS ANA-12 rats compared to the lesion nonVNS ANA-12 rats (G-H, **p < 0.01). ANA-12 did not alter LC TH-positive neurons for the saline nonVNS, saline +VNS, or lesion nonVNS groups (A–C, E–G).

2.10. Statistical analysis

Data are displayed as mean ± standard error of the mean (SEM), and graphs were assembled in GraphPad Prism (GraphPad Software). Locomotor activity per day was analyzed using a repeated measures 2(Lesion) × 2(Stimulation) × 2(Drug) × 10(Day) analysis of variance (ANOVA) in SPSS (IBM). All histological data, as well as the collapsed mean locomotor activity across days were analyzed in SPSS (IBM) using a 2(Lesion) × 2(Stimulation) × 2(Drug) ANOVA followed by group-wise difference determination using Bonferroni’s correction for multiple comparisons at a significance threshold of p<0.05.

3. RESULTS

3.1. ANA-12 did not block VNS-induced locomotor activity increase

The effects of ANA-12 on locomotor activity (total distance traveled) were assessed during the second daily stimulation session. Across all 10 days, while no significant main effects or interactions of Lesion, Stimulation, or Drug existed using a repeated measures ANOVA, saline nonVNS vehicle rats consistently had the highest locomotor activity, and lesion nonVNS rats had the lowest locomotor activity. Locomotor activity of lesion +VNS rats was consistently higher than lesion nonVNS rats during the stimulation period, and by day 10 of VNS, was similar to saline nonVNS vehicle rats (Figure 2A). Locomotor activity was averaged across all ten days of VNS (Figure 2B) to assess overall locomotor differences, and a 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA revealed a significant effect of Lesion only (F(1,39)=4.99, p=0.031) with the only significant interaction occurring between Lesion x Stimulation (F(1,39)=6.12, p=0.018). Pairwise comparisons using the Bonferroni method revealed a significant reduction of locomotor activity in lesion nonVNS vehicle compared to saline nonVNS vehicle rats (p<0.01). A trend towards increased locomotor activity was observed in lesion +VNS vehicle rats compared to lesion nonVNS rats (p=0.052). Although there were no main effects of ANA-12 on behavior, locomotor activity was significantly reduced with ANA-12 administration in saline nonVNS rats only compared to vehicle-treated counterparts (p<0.05). These results demonstrate that while locomotor activity was reduced after lesion, this behavior was increased after VNS in lesion animals. Although ANA-12 reduced locomotion of saline nonVNS rats, it had no effect on VNS in lesion rats. This indicates that TrkB does not contribute to improvements in locomotor activity by VNS in our model.

Fig. 2. ANA-12 did not block VNS-induced locomotor activity increase.

Total distance was measured in cm across all 10 days of VNS, and locomotor activity was consistently lower in lesion nonVNS rats compared to saline nonVNS rats (A). Total distances were averaged for each animal to give an average distance traveled (B). Lesion nonVNS vehicle rats move significantly less than saline nonVNS vehicle rats (**p < 0.01), and there was a trend toward increase in lesion +VNS rats compared to lesion nonVNS (p = 0.052). While ANA-12 reduced locomotor activity in saline nonVNS rats (*p < 0.05), other groups’ locomotion was unaffected by ANA-12 administration.

3.2. ANA-12 partially prevented VNS-induced attenuation of TH-positive LC loss

Studies have demonstrated that VNS effects are mediated through the LC [33], and that the LC degenerates early in PD progression [34]. Therefore, stereological counts of TH-positive neurons were conducted in the LC to determine whether ANA-12 prevents our previously observed increases in TH-positive LC-NE neurons after VNS [20]. A 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA revealed significant main effects of all three ANOVA factors: Lesion, Stimulation, and Drug (Table 1), with significant interactions of Lesion x Stimulation, Lesion x Drug, and Stimulation x Drug, although no significant 3-way interaction existed (Table 2). Bonferroni’s adjustment method showed lesion nonVNS vehicle rats had significantly fewer TH-positive neurons in the LC than saline nonVNS vehicle rats (p<0.0001, Figure 3A,C,I). TH-positive cells were greater in both saline +VNS vehicle rats (p<0.01) and lesion +VNS vehicle rats (p<0.0001) compared to their nonVNS counterparts (Figure 3A–D,I). ANA-12 reduced TH-positive cells in lesion +VNS rats compared to the lesion +VNS vehicle group (p<0.0001), though this was only a partial reduction with greater TH-positive cells in lesion +VNS ANA-12 rats compared to lesion nonVNS ANA-12 rats (p<0.01, Figure 3D,G–I). Overall, our LC TH results show that while lesion reduces LC-NE neurons, these numbers are greater after VNS regardless of lesion, and ANA-12 partially prevents this VNS effect in lesion rats. Together, these data demonstrate that TrkB contributes in part to the ability of VNS to attenuate loss of LC-NE neurons.

Table 1.

3-way ANOVA main factor results for immunohistochemical data.

Region	Lesion		Stimulation		Drug
	F(1,39)	p-value	F(1,39)	p-value	F(1,39)	p-value
TH
LC	104.29	<0.0001****	55.08	<0.0001****	4.78	0.0356*
Striatum	147.32	<0.0001****	1.83	0.1835	0.04	0.8389
SN	134.35	<0.0001****	2.28	0.1399	12.82	0.0010***
Intrasomal α-synuclein
SN	38.42	<0.0001****	2.55	0.1189	36.61	<0.0001****
Iba-1
Striatum	19.64	0.0001***	4.26	0.0461*	0.53	0.4698
SN	490.06	<0.0001****	18.96	0.0001***	30.85	<0.0001****
GFAP
Striatum	51.58	<0.0001****	8.75	0.0054**	5.17	0.0289*
SN	2468.10	<0.0001****	133.65	<0.0001****	62.13	<0.0001****

Significant p-values:

^*p<0.05

^**p<0.01

^***p<0.001

^****p<0.0001

Table 2.

3-way ANOVA interaction results for immunohistochemical data.

Region	Lesion × Stimulation$		Lesion × Drug$		Stimulation × Drug$		3-way Interaction$
	F(1,39)	p-value	F(1,39)	p-value	F(1,39)	p-value	F(1,39)	p-value
TH
LC	11.09	0.0021**	4.24	0.0471*	12.89	0.0010**	0.44	0.5106
Striatum	1.71	0.1982	4.82	0.0342*	0.23	0.6288	3.25	0.0793
SN	9.87	0.0033**	1.06	0.3106	8.85	0.0051**	6.11	0.0181*
Intrasomal α-synuclein
SN	3.74	0.0607	0.32	0.5772	8.14	0.0070**	6.55	0.0147*
Iba-1
Striatum	6.29	0.0167*	0.062	0.8055	5.13	0.0295*	3.696	0.0623
SN	21.52	<0.0001****	26.16	<0.0001****	14.99	0.0005***	18.50	0.0001***
GFAP
Striatum	15.95	0.0003***	0.67	0.4186	18.53	0.0001***	9.36	0.0041**
SN	141.44	<0.0001****	64.84	<0.0001****	88.83	<0.0001****	69.09	<0.0001****

Significant p-values:

^*p<0.05

^**p<0.01

^***p<0.001

^****p<0.0001

^$Significant 2-way interactions for regions that also have significant 3-way interactions are not mentioned in the main text

3.3. Attenuation of nigrostriatal loss by VNS was prevented with ANA-12

Analysis of TH in the nigrostriatal system was conducted via TH-ir of terminals in the striatum and TH-positive cell counts in the SN. These assessments allowed us to determine whether ANA-12 can block the VNS-induced attenuation of nigrostriatal degeneration in this PD model. A 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA showed a significant main effect of Lesion with interaction of Lesion x Drug in the striatum, as well as main effects of Lesion and Drug with a Lesion x Stimulation x Drug interaction in the SN (Table 1,2). This was followed up with Bonferroni’s correction for multiple comparisons to show that lesion nonVNS vehicle rats had lower striatal TH-ir and SN TH-positive cells compared to saline nonVNS vehicle rats (p<0.0001; Figure 4A,C,I,J,L,R). VNS resulted in increased striatal TH-ir (p<0.05) and SN TH-positive cells (p<0.01) for lesion rats (Figure 4C–D,I,L–M,R). ANA-12 administration prevented increases in TH-ir (p<0.05) and TH-positive cells (p<0.0001) for lesion +VNS rats only, such that levels were comparable to lesion nonVNS ANA-12 rats (Figure 4D,G–I,M,P–R). Our results from this section show that while VNS resulted in greater DA terminals in the striatum and SN-DA neurons of lesioned rats, ANA-12 prevented these effects, indicating a role for TrkB in the ability of VNS to protect SN-DA neurons.

3.4. ANA-12 increased intrasomal α-synuclein in remaining SN-DA neurons

Analysis of intrasomal α-synuclein density in TH-positive neurons was conducted in the SN to determine the role of TrkB in α-synuclein accumulation for SN-DA neurons, as this intrasomal accumulation has been previously observed in both post-mortem PD brains and in rodent models of PD [31,32]. A 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA revealed significant main effects of Lesion and Drug (Table 1), with an interaction between Lesion x Stimulation x Drug (Table 2). To investigate the nature of the interaction, subsequent analysis using Bonferroni’s correction was conducted. Lesion nonVNS vehicle rats had significantly higher intrasomal α-synuclein-ir (Figure 5, shown in green) than saline nonVNS vehicle rats (p<0.0001, Figure 5A,C,I). Intrasomal α-synuclein-ir was reduced after VNS in lesion vehicle rats (p<0.001, Figure 5B–D,I). ANA-12 prevented the decrease of intrasomal α-synuclein-ir observed in lesion +VNS rats (p<0.0001, Figure 5D,G–I). ANA-12 also significantly increased intrasomal α-synuclein-ir in saline nonVNS rats (p<0.01) and saline +VNS rats (p<0.01, Figure 5A–B,E–G,I). Our results demonstrate that this PD-like lesion increases intrasomal α-synuclein accumulation, which is prevented after VNS. ANA-12 blocks this effect of VNS in lesion +VNS rats, and increases intrasomal α-synuclein accumulation in saline rats regardless of VNS treatment. These data provide evidence for a role of BDNF-TrkB both in regulation of intrasomal α-synuclein expression, and in the ability of VNS to reduce intrasomal α-synuclein accumulation in SN-DA neurons.

Fig. 5. ANA-12 increased intrasomal α-synuclein in remaining SN-DA neurons.

Confocal z-stacks of TH-positive cells (red) are overlayed with α-synuclein (green) (A-H, scale = 25 μm, quantified in I). Insets show α-synuclein within a single representative TH-positive neuron from each group. Lesion nonVNS vehicle rats have higher intracellular α-synuclein density than saline nonVNS vehicle rats (A-B, C, G, ****p < 0.0001). VNS alone did not alter intracellular α-synuclein density (A-B), but reduced intracellular α-synuclein density in lesion vehicle rats (C-D, ***p < 0.001). ANA-12 increased intracellular α-synuclein density in both saline nonVNS and saline +VNS rats (A-B, E-F, **p < 0.01), though saline nonVNS ANA-12 rats still had lower levels than lesion nonVNS ANA-12 rats (*p < 0.05), and ANA-12 prevented the effects of VNS on intracellular α-synuclein density in lesioned rats (D, H, ****p < 0.0001).

3.5. Anti-inflammatory effects of VNS were prevented with ANA-12 in the nigrostriatal system

Microglia were assessed by Iba-1-ir in the striatum and SN. While microglial density can be influenced by number of cells or by activation state, visual inspection revealed that the majority of microglia seen in non-lesion saline rats are in a resting state, as indicated by small cell bodies and long processes (Figure 6J–K,N–O insets). In lesion nonVNS rats, microglia appear to be in an activated state, as shown by large, dark cell bodies and shortened processes, indicative of neuroinflammation (Figure 6L,P insets). In lesion +VNS vehicle rats, the majority of microglia appear to be in a resting state, although some activation is still present (Figure 7M inset). Microglia for lesion +VNS ANA-12 rats, however, appear be activated similarly to the lesion nonVNS rats (Figure 6Q inset). Analysis of Iba-1-ir using a 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA revealed significant effects of Lesion and Stimulation in the striatum, and of Lesion, Stimulation and Drug in the SN (Table 1). ANOVA results for the striatum showed significant interactions of Lesion x Stimulation and Stimulation x Drug, with a Lesion x Stimulation x Drug interaction in the SN (Table 2). Subsequent investigation of the interactions using Bonferroni’s method showed lesion nonVNS vehicle rats had a significant increase of Iba-1-ir than saline nonVNS vehicle rats in both regions (p<0.0001, Figure 6A,C,I–J,L,R). VNS resulted in a reduction of Iba-1-ir in the striatum (p<0.001) and SN (p<0.0001) for lesion +VNS vehicle rats compared to lesion nonVNS vehicle rats (Figure 6C–D,I,L–M,R). Lesion +VNS ANA-12 rats had an increase in lba-1-ir compared to lesion +VNS vehicle rats in striatum (p<0.05) and SN (p<0.0001, Figure 6D,H–I,M,Q–R).

Density of astrocytes was determined by GFAP-ir, and a 2(Lesion) x 2(Stimulation) x 2(Drug) ANOVA of these data revealed significant effects of all three ANOVA factors: Lesion, Stimulation, and Drug in the striatum and SN (Table 1), as well as a significant interaction of Lesion x Stimulation x Drug in both regions (Table 2). Using Bonferroni’s method to correct for multiple comparisons, we found that nonVNS vehicle rats had a significant increase of GFAP-ir after lesion in both regions (p<0.0001, Figure 7A,C,I–J,L,R). When treated with VNS, lesion vehicle rats had a reduction in GFAP-ir for both brain regions compared to their nonVNS counterparts (p<0.0001, Figure 7C–D,I,L-M,R). ANA-12 rats had an increase of GFAP-ir in the lesion +VNS group for the striatum and SN compared to vehicle-treated counterparts (p<0.0001, Figure 7D,H–I,M,Q–R). Taken together, the glial data demonstrate that a PD-like lesion increases gliosis in the nigrostriatal system, and that VNS reduces gliosis in lesioned rats. ANA-12 prevents this effect in the nigrostriatal system, indicating that TrkB plays a role in the anti-inflammatory effects of VNS.

4. DISCUSSION

This study is the first to examine a potential mechanism of action for VNS in a PD model. Here, we investigate the role of BDNF-TrkB as a key mediator of the beneficial behavioral and neuronal effects of VNS in a sequential NE and DA lesion model of PD. The current study replicates our previous findings that VNS in a PD model results in greater locomotor activity, higher numbers of TH-positive neurons in the LC and SN, as well as greater TH-ir in the striatum, decreased intrasomal α-synuclein accumulation in the SN, and decreased neuroinflammation in the nigrostriatal system [20]. ANA-12 inhibition of BDNF binding to TrkB does not prevent the beneficial effects of VNS on locomotor activity, and only partially prevents attenuation of LC-NE loss. However, ANA-12 does prevent the nigrostriatal effects of VNS on DA neurons, intrasomal α-synuclein accumulation, and nigrostriatal neuroinflammation. These findings suggest a role for BDNF-TrkB in the VNS effects on the degenerating nigrostriatal system, and in part on the LC in a double lesion model of PD (Table 3).

Table 3.

Summary of results

Region	Lesion nonVNS (compared to saline nonVNS)	Saline +VNS (compared to saline nonVNS)	Lesion +VNS (compared to lesion nonVNS)	Saline nonVNS ANA-12 (compared to saline nonVNS vehicle)	Lesion +VNS ANA-12 (compared to lesion +VNS vehicle)
Locomotor activity	⬇	⬌	⬆	⬇	⬌
TH
LC	⬇	⬆	⬆	⬌	⇣
Striatum	⬇	⬌	⬆	⬌	⇣
SN	⬇	⬌	⬆	⬌	⇣
Intrasomal α-synuclein
SN	⬆	⬌	⬇	⬆	⇡
Iba-1
Striatum	⬆	⬌	⬇	⬌	⇡
SN	⬆	⬌	⬇	⬌	⇡
GFAP
Striatum	⬆	⬌	⬇	⬌	⇡
SN	⬆	⬌	⬇	⬌	⇡

➞ Detrimental effects of lesion or ANA-12

➞ Beneficial effects of VNS

⇢ Blockade of VNS effects with ANA-12

⬌ No change

4.1. How does ANA-12 influence behavior in this PD model?

A previous study using ANA-12 to inhibit effects of DBS has shown that acute TrkB inhibition can prevent stimulation effects on limb akinesia at 6 weeks post-unilateral 6-OHDA lesion. However, chronic TrkB inhibition beginning 10 days post lesion does not prevent stimulation-induced behavioral improvements [35]. Similarly, in our current study using a PD model, we have found that chronic administration of ANA-12 has no effect on VNS-induced increases in locomotor activity. These findings could be due in part to a compensatory increase in either cortical or cerebellar neuronal activity for lesion rats, as increased cerebellar firing and strengthened connectivity of motor cortex to the striatum and subthalamic nucleus has been previously observed in rodent 6-OHDA models [36,37]. This compensatory recruitment of motor cortex and cerebellum also occurs in humans with early stage PD [38,39], and if similar compensation is occurring in our model, that would explain why ANA-12 does not affect locomotor activity for lesion nonVNS or lesion +VNS rats. It is also possible that since VNS is activating the LC, increased LC firing is increasing activity of remaining SN-DA neurons downstream for rats that received VNS. This effect of LC on SN-DA firing has been shown in previous electrophysiological studies [40,41]. Increases in SN-DA firing lead to increased striatal DA release, thereby helping to regulate basal ganglia firing to improve behavior [42]. Therefore, it is possible that similar alterations in neuronal activity are occurring in this lesion model both before and after VNS, thereby potentially preventing reductions in locomotor activity after ANA-12.

While there are not alterations in numbers of SN-DA neurons for saline rats, there is an increase of intrasomal α-synuclein accumulation within SN-DA neurons of saline rats after ANA-12 in this study. Reduced locomotor activity in the saline ANA-12 rats is consistent with previous studies that show a genetic overexpression of α-synuclein in mice leads to motor deficits with no SN-DA loss, indicating that dysfunction in either clearance or transport mechanisms of α-synuclein in our current study could lead to the decreased locomotor activity in saline ANA-12 rats [43]. Another possible explanation for reduced locomotor activity in this group could be reduced striatal DA release. Activity of the BDNF-TrkB pathway in the nigrostriatal system has been shown to increase striatal DA release and increase locomotion in adult rats [44,45]. Injections of amphetamine have also been shown to increase DA release and TrkB expression in the dorsal striatum, and this effect is thought to contribute in part to the hyperactivity observed after amphetamine administration [46]. Consequently, studies have shown that infusion of TrkB antibodies into the nucleus accumbens reduces DA release and locomotor activity after treatment with methamphetamine, demonstrating that reduced TrkB activity can influence DA release to ultimately reduce locomotion [47]. Therefore, in order to better understand the role of ANA-12 in reducing locomotor activity, studies need to be conducted assessing DA release and the potential role ANA-12 may play in the trafficking of α-synuclein.

4.2. What is the role of TrkB in the LC for this model?

It is well-established that VNS increases LC firing, and that VNS effects on higher brain regions are mediated through the LC [33,41]. Other studies have demonstrated that VNS can upregulate BDNF in target regions of the LC and that increased BDNF in LC targets corresponds to increased firing rates of LC-NE neurons [48,49]. In the current study, a single systemic injection of DSP-4 results in a 45% loss of LC TH-positive neurons. Therefore, it is possible that the remaining neurons act in a hyperactive manner following VNS to compensate for LC-NE cell loss, thus allowing for increased BDNF, reduced neuroinflammation and improved neuronal survival [50]. In support of this idea, we have previously observed BDNF increases after VNS in the striatum of this lesion model [24]. In this study, we only find a partial prevention of VNS effects on LC-NE neurons after ANA-12 administration for lesion rats. Because ANA-12 is a TrkB-specific antagonist, it is possible that activity of additional neurotrophic factors such as nerve growth factor or neurotrophin-3 acting on other Trk receptors (TrkA and TrkC, respectively) become hyperactive to compensate for reduced TrkB activity, thereby preserving a partial attenuation of LC-NE loss for this group. Both of these neurotrophins are important for health and maintenance of NE neurons, and TrkC has been shown to be highly expressed in LC-NE neurons [51,52,53]. While no previous studies have assessed the effects of VNS on neurotrophin-3 levels, VNS is known to alter expression of growth factors other than BDNF in models of ischemia and chronic heart failure including nerve growth factor, vascular endothelial growth factor, and basic fibroblast growth factor [54,55,56]. Therefore, it is possible that ANA-12 only partially prevents VNS effects on LC-NE neurons due to increased activity of these additional neurotrophins, although future studies are required to confirm this hypothesis.

4.3. How does TrkB influence the nigrostriatal system in this model of PD?

BDNF is known to be important for the health and survival of SN-DA neurons [7]. Infusion of BDNF can increase firing rates of SN-DA neurons [57], and BDNF administration can protect against 6-OHDA-induced degeneration, even when given after lesioning [58]. We have previously found that VNS increases BDNF levels in the striatum and frontal cortex of double lesioned rats [24]. Similarly to previous studies, the inhibition of TrkB with ANA-12 prevented neurostimulation-induced protection of TH-positive neurons in the SN [35]. These studies combined with our findings indicate the importance of BDNF and TrkB for protection of SN-DA neurons from toxic insult.

A previous link has been established between TrkB and α-synuclein expression. While we have demonstrated that VNS attenuates intrasomal accumulation of α-synuclein, TrkB inhibition prevented this VNS effect in our double lesion model. Similar to our findings, a study examining the gastrointestinal tract of mutant α-synuclein mice (A53T mutation) has shown increased α-synuclein aggregation in enteric neurons after TrkB inhibition with ANA-12, but reduced α-synuclein aggregation after fingolimod administration, a compound known to increase BDNF [18]. Administration of BDNF has been shown to prevent formation of α-synuclein aggregates in a mouse model overexpressing mutant α-synuclein, and this occurred through inactivation of a cleavage protein by Akt [59]. Conversely, a genetic reduction of TrkB expression in mice has led to increased α-synuclein deposits in TH-positive neurons, indicating that reduced TrkB can increase α-synuclein accumulation [15]. These findings combined with our observation that VNS can increase tissue concentrations of BDNF in the striatum of lesion rats [24] suggest that VNS may utilize BDNF to reduce intrasomal α-synuclein accumulation in our model.

In the current study we demonstrate that inhibition of TrkB prevents the antiinflammatory effects of VNS as measured by microgliosis and astrogliosis in the nigrostriatal pathway. Traditionally, it was thought that glial cells solely express a truncated form of the TrkB receptor that is incapable of inducing pro-survival signaling [60]. Flowever, we now know that the truncated TrkB receptor can induce MAPK/Akt signaling, and this is thought to occur via activation of Rho [61]. Since ANA-12 binds to the extracellular domain of TrkB, which is identical in full length and truncated isoforms, ANA-12 prevents BDNF binding to both isoforms similarly [25]. In addition, the full length form of TrkB can be expressed on microglia and astrocytes to reduce pro-inflammatory cytokine production via inhibtion of nuclear factor-kappa B (NFkB) [62,63]. In a rat model of meningitis, intraventricular administration of BDNF reduces neuroinflammation in both hippocampus and cortex via TrkB-Akt signaling, and these effects can be prevented via tyrosine kinase inhibition with acute administration of K252a [17]. Further, administration of antidepressants known to increase BDNF prevent production of pro-inflammatory cytokines [64]. Pretreatment with BDNF also reduces release of pro-inflammatory cytokines to protect against ischemic damage [65]. These findings suggest that increased BDNF after VNS may bind to TrkB on glial cells to decrease inflammatory cytokine production via inhibition of NFkB and induce activation of pro-survival pathways via Rho. Both of these processes would ultimately reduce inflammation to aid in attenuation of neuronal loss and downstream behavioral improvements for our PD model.

5. CONCLUSIONS

Overall, our data suggest that an increase in BDNF is an important mechanism of action for VNS in our Parkinson’s model, which results in reduced neuroinflammation that promotes neuronal survival in both the LC-NE and nigrostriatal systems. In our current study, by blocking BDNF-TrkB via ANA-12 administration, we show that the nigrostriatal effects of VNS on TH-positive neurons in the SN, accumulation of α-synuclein in SN-DA neurons, and nigrostriatal inflammation are BDNF-dependent. However, ANA-12 administration only partially reduces TH-positive neurons in the LC and does not affect locomotor activity after VNS, suggesting that BDNF is not the only mechanism of action for VNS. Future studies of mechanisms of action for VNS in PD models should assess the possible interplay between neurodegenerative mechanisms, including the role of additional growth factors, such as nerve growth factor or neurotrophin-3.

Highlights.

• TrkB inhibition prevents attenuation of dopaminergic neuron loss after VNS

• ANA-12 partially prevents beneficial effects of VNS on noradrenergic neurons

• Anti-inflammatory VNS effects in the nigrostriatal system are prevented with ANA-12

• Inhibition of TrkB does not prevent locomotor benefits after VNS in a PD model

• Beneficial effects of VNS in a PD model are partially mediated via TrkB