Intranasal delivery of mitochondria for treatment of Parkinson’s Disease model rats lesioned with 6-hydroxydopamine

The feasibility of delivering mitochondria intranasally so as to bypass the blood–brain barrier in treating Parkinson's disease (PD), was evaluated in unilaterally 6-OHDA-lesioned rats. Intranasal infusion of allogeneic mitochondria conjugated with Pep-1 (P-Mito) or unconjugated (Mito) was performed once a week on the ipsilateral sides of lesioned brains for three months. A significant improvement of rotational and locomotor behaviors in PD rats was observed in both mitochondrial groups, compared to sham or Pep-1-only groups. Dopaminergic (DA) neuron survival and recovery > 60% occurred in lesions of the substantia nigra (SN) and striatum in Mito and P-Mito rats. The treatment effect was stronger in the P-Mito group than the Mito group, but the difference was insignificant. This recovery was associated with restoration of mitochondrial function and attenuation of oxidative damage in lesioned SN. Notably, P-Mito suppressed plasma levels of inflammatory cytokines. Mitochondria penetrated the accessory olfactory bulb and doublecortin-positive neurons of the rostral migratory stream (RMS) on the ipsilateral sides of lesions and were expressed in striatal, but not SN DA neurons, of both cerebral hemispheres, evidently via commissural fibers. This study shows promise for intranasal delivery of mitochondria, confirming mitochondrial internalization and migration via RMS neurons in the olfactory bulb for PD therapy.


Results
Improvement of rotational and locomotor behavior. Behavior of PD rats was assessed in terms of apomorphine-induced rotations (Fig. 1A) and performance in the open field test (Fig. 1B). Rotation data for the WT group are not presented due to the extremely low frequency of apomorphine-induced turning in normal rats (Fig. 1A). The Sham group showed a dramatic increase of apomorphine-induced rotational activity relative to the inactivity of the WT group (Fig. 1). Unilateral intranasal infusion (ipsilateral to the lesioned side) of Mito or P-Mito caused a significant decrease in rotational activity 11 weeks post-treatment (Fig. 1A) and recovery of normal locomotor activity at 12 weeks, as measured by mean velocity, mobility duration, frequency of crossed zones, and traveling distance, compared to the Sham group (Fig. 1B). There was no significant difference in behavior between the Sham and Pep-1-only groups, or between the Mito and P-Mito groups (Fig. 1).

Support of nigral dopaminergic neurons against lesion-induced cell death in SN and ST.
Neuron survival in the substantia nigra (SN) was examined by Nissl staining 3 months post-treatment ( Fig. 2A, B) and survival of dopaminergic (DA) neurons was further confirmed by TH immunofluorescence staining (Fig. 2C,D). Magnified images of TH fluorescence with 4' ,6-diamidino-2-phenylindole (DAPI) nuclear counterstaining in right halves of WT and PD brains are shown in the third column of Fig. 2C. Obvious asymmetry was consistently apparent in both Nissl-stained ( Fig. 2A) and DA neurons (Fig. 2C) between the left (intact) and right (6-OHDA-lesioned) halves of the brain in the SN of the Sham group versus the WT group. In the sham group, there was a significant loss of both signals in the lesioned side of the SN, with an average survival of ~ 18.5% relative to the intact side (Fig. 2B,D). In contrast, a significant increase in survival of SN (Fig. 2B) and DA neurons (Fig. 2D) in the lesioned side of the SN was found in the Mito (µ = 68.4%) and P-Mito (µ = 73.2%) groups, with similar levels of performance indicating the effectiveness of mitochondrial treatment for neuron survival. No effect was seen in the Pep-1-only group (Fig. 2).
Furthermore, immunohistochemical TH staining also showed consistent results in the striata (ST) of the Mito and P-Mito groups, which were comparable to those of the sham and Pep-1 groups (Fig. 3A). The significant loss of DA neurons that project to the ST, disrupting the motor circuit of the basal ganglia, was restored by Mito and P-Mito treatments, but not in the sham or Pep-1-only groups (Fig. 3B). The difference of TH performance between lesioned and intact sides diminished significantly in the P-Mito and Mito groups (Fig. 3B), though there was no significant difference between them (p = 0.08).  The open-field test quantified the average velocity, zone-crossing frequency, duration of mobility, and distance traveled. In all graphs, values from treated animals were normalized to those of wild-type controls. +P < 0.05, vs. WT; *P < 0.05, vs. sham; WT wild-type controls, PD Parkinson's disease, Sham vehicle alone, Mito mitochondrial alone, P-Mito Pep-1-labelled mitochondria. In contrast, complexes II-IV (CII-CIV) showed significant increases in band intensity and obvious shifts in molecular weight (Fig. 4A). Both the Mito and P-Mito groups showed patterns of mitochondrial complex proteins like those of WT, documenting a significant recovery of CI and explaining the performance normalization of CII-CIV relative to the sham group (Fig. 4A). Quantification revealed that CI-CIV substantially returned to normal in the Mito and P-Mito groups and that the adjustment in CIV expression in the P-Mito group was more significant than in the Mito group (Fig. 4B). Meanwhile, IHC staining of nuclei in SN neurons exhibited strong 8-OHdG signals (Fig. 4C) in the Sham group compared to the WT group, whereas nuclei were significantly less  www.nature.com/scientificreports/ stained in both the Mito and P-Mito groups, though their staining levels were still higher than those observed in the WT group (Fig. 4C) . There was no significant difference in plasma levels of inflammatory cytokines between WT and Sham groups, except for Interleukin (IL)-1alpha (IL-1α), which showed a significant decrease in the Sham group (Fig. 5). Presence of allogeneic mitochondria in nigral dopaminergic neurons innervating the striatum through the rostral migratory stream. BrdU IHC images of brain sagittal sections (lesioned side) visualized the uptake of intranasally infused allogeneic mitochondria labelled with BrdU in different parts of the brain, including the corpus callosum (CC) and ST, accessory olfactory bulb (AOB), rostral migratory stream (RMS) track, and glomerular layer (GL) layer of the main olfactory bulb (Fig. 6A). Compared to the untreated WT group, despite having a mild antibody background around the AOB area (asterisks), mitochondrial treatment groups revealed the marked presence of allogeneic mitochondria in the region of an RMS-like track pen- The latter were visibly more numerous in the P-Mito group than in the Mito group, as revealed by denser BrdU signals, but exhibited similar fluorescence in the lesioned side (Fig. 7A,B). To explain allogeneic mitochondrial internalization in both cerebral hemispheres, double immunofluorescence staining (DCX/BrdU) of migrating neuroblasts was performed in coronal brain sections of the decussation of the anterior commissure (AC) (Fig. 7C). BrdU immunoreactivity in the Mito and P-Mito groups revealed expression of foreign mitochondria in neurons of the AC between the ipsilateral and contralateral hemispheres. Mitochondrial transmission in AC neurons from the treated side to the contralateral (intact) side was also more obvious in the P-Mito group than in the Mito group (Fig. 7C), which was consistent with BrdU expression on the contralateral side in the ST (Fig. 7A). The Z-stack for 3D reconstruction at high-magnification merged confocal fluorescence images of DCX-positive cells on the lesioned side AC (Fig. 7D). Co-expression of DCX and BrdU signals in merged images showed mitochondrial internalization in DCX cells with a typical fusiform morphology.

Discussion
Mitochondrial transplantation has been extensively studied in recent years, with efforts to apply it to treatments for a variety of mitochondrial and non-mitochondrial diseases 21,22 . Based on the pathogenesis of each disease, various invasive approaches have been used for transplantation, including intravascular injection, subcutaneous injection, and in situ injection 5,22,23 . Non-invasive mitochondrial transplantation, however, has rarely been studied due to certain limitations, including the inability to preserve mitochondrial activity after isolation, the lack of specificity in targeted delivery, and the problem of poor delivery efficiency. To the best of our knowledge, this is the first study to demonstrate the therapeutic feasibility of brain targeting via intranasal administration of allogeneic mitochondria in a neurotoxin-induced PD rat model. In addition, we utilized BrdU for mitochondrial   24,25 . In contrast to our previous study using a medial forebrain bundle (MFB) injection of mitochondria for PD treatment 7 , the present results show that when mitochondria were twice administered intranasally, restoration of mitochondrial complex I protein in substantia nigra neurons (~ 52%) was significantly lower than with local injection (~ 85%) using the same treatment frequencies and intervals. However, intriguingly, the improvement of animal behavior (except for the 40% lower index of cross-zoom frequency) and the increase in survival of nigra DA neurons were similar. This may be related to differences of mitochondrial uptake in specific regions of PD brains, due to different interventions. Indeed, neurons, astrocytes and glial cells of the cerebrum differed in mitochondrial internalization efficacy, which was also affected by different transplantation routes in the rescue model of stroke rats 25 . By tracking BrdU-labelled mitochondria, we found that mitochondria delivered intranasally could only be observed in the ST (the terminal of DA neurons) in contrast to mitochondria injected  10 . Although the delivery timing and frequency differed, delivered mitochondria and cells both resulted in higher performance in lesioned ST relative to lesioned SN and manifested a 60-80% improvement in motor function after > 70 days of treatment 10 . Interestingly, in a D-amphetamine-induced rotational behavior test to evaluate neuronal DA loss 27 , significant inhibition of rotational behavior occurred in the 11th week of continuous treatment with mitochondria versus in 136 days of end treatment with MSCs 10 . Mitochondrial treatment restored DA neuron viability sooner than MSCs treatment, though the restored level of TH in lesioned ST and SN was similar to stem cell treatment showed in Danielyan et al. 10 . In contrast to MSCs, the absence of delivered mitochondria in the SN www.nature.com/scientificreports/ makes it difficult to explain this phenomenon based on our limited data. We suggest that functional support of nigral DA neuron nerve terminals is more important in ST than in SN in PD therapy, because consistently higher expression level of delivered MSCs were observed in lesioned ST than in lesioned SN more than 6 months after treatment ended 10 . The latest article published in the period of our article being reviewed supports indirectly our finding that nasal administration of mitochondria reverses chemotherapy-induced cognitive deficits 28 . ST and its cortical connections is well-known critical regulator for cognitive symptoms 29 . Compared to injected MFB mitochondria, which need to be modified for uptake into nigra DA neurons 11 , nose-to-brain delivery affords easier access to the brain since mitochondria do not need to be modified to penetrate the AOB and to enter the ST via the RMS pathway. The RMS pathway provides a "conduit" for intranasal mitochondrial delivery, guiding internalized mitochondria into the ST for restoration of dysfunctional DA neurons in PD. Moreover, naked mitochondria can also enter neurons by perfusion ipsilateral to the lesioned side and can be delivered to DA neurons of the ST on the contralateral, non-lesioned side via axons of interhemispheric commissures, such as the AC and CC 30 . Although contralateral delivery efficiency of naked mitochondria was not as high as that of peptide-modified mitochondria, therapeutic effects were not affected in the unilateral lesioned rat PD model used here. Further studies are required to determine whether better contralateral mitochondrial delivery efficiency would enhance therapeutic effects in bilateral SN lesions of PD model rats. It would be beneficial to increase delivery efficiency of mitochondria from nose-to-brain and to simplify the process of transplantation. In addition, studies have shown that proteins and nanoparticles can be delivered intranasally to brain parenchyma, the third and fourth ventricles, midbrain, and hippocampus, via the same pathway (the RMS pathway) in transgenic mouse models of Alzheimer's disease 31 . However, in contrast to these techniques, which could also utilize intracellular transport of olfactory axons and extracellular transport processes along the olfactory nerve and the trigeminal nerve, mitochondria could only be delivered via the extracellular pathway to enter the AOB (since no significant presence of mitochondria was observed in olfactory neurons and the olfactory-trigeminal pathway, even after long-term treatments) and then reached the basal ganglia of the forebrain and lateral ventricles through the RMS pathway. This may be caused by variation in transport of mitochondria internalized in different types of neurons 32 . Otherwise, the diversity of intracellular signal transduction pathways in response to extracellular guide signals leads to variable cellular uptake of mitochondria 33 .
Moreover, we found that expression of plasma inflammatory cytokines was affected by different interventional approaches. Contrary to more serious contraindications resulting from local injections of vehicle relative to PD non-treatment controls 7 , high plasma inflammatory cytokine levels of PD rats were significantly reduced by intranasal vehicle perfusion, implying that this intervention will not exacerbate the disease during treatment 34 . Saline nasal irrigation is clinically proven for postoperative inflammation and to minimize antibiotic resistance 35 . One explanation for its utility is that the nasal inflammatory mediator, leukotriene C4, is substantially less elevated 2-6 h after treatment 35 . Clinical studies have shown that intranasal delivery of a placebo (saline) or an antioxidant enzyme (glutathione) alone over a three-month period provided symptomatic improvements in PD patients 36 and anti-inflammatory benefits for PD 22 . Moreover, our results are consistent with our previous finding that modifying mitochondria with Pep-1 reduces their induction of plasma pro-inflammatory cytokines 7 . Although this outcome did not significantly affect therapeutic efficacy after just three months of treatment, further investigation is necessary to study its long-term post-operative effects. Moreover, while mitochondria delivered via brain injection induced higher levels of proinflammatory cytokines 7 , intranasal mitochondrial delivery only induced expression of interleukin (IL)-1α, IL-1β, IL-10, and IL-17A, further demonstrating the immunological safety of intranasal drug delivery.

Conclusions
Various nose-to-brain delivery routes for nanomedicines, oligonucleotides, enzymes, and stem cells have been applied in PD treatments. This study demonstrates preliminarily that mitochondrial delivery from nose-to-brain is a feasible approach, and is safer than brain injection. While functional restoration of mitochondria to DA neurons using this approach is not as effective as by direct injection, neuronal survival and behavioral improvement are similar. Further, by this method, mitochondria can be delivered to the brain without modification. Unfortunately, mitochondria were only observed in the terminals of DA neurons in the ST and were not found in somata of DA neurons in the SN. Our previous studies have shown that injected mitochondria improve mitochondrial homeostasis and turnover by restoring dynamic mitochondrial protein loss in DA neurons, including increasing mitochondrial fusion-fission proteins and autophagy of damaged mitochondria in the SN 7 . Selectivity of mitochondrial fusion is particularly beneficial under conditions of increased mitochondrial damage, because it facilitates fusion frequency without compromising removal of damaged organelles by mitophagy, resulting in improved mitochondrial quality 37 . Whether this regulation is affected by different mitochondrial transplantation routes is under investigation. If retrograde transport of mitochondria from the minus-end in striatal terminal axons toward the cell body in the SN can be increased, therapeutic efficacy may be improved. Furthermore, therapeutic benefit of transfer of mtDNA-laden extracellular vesicles (EVs) have been reported not only in aggressive breast cancer 38 but also in PD 39 . Thus, whether the presence of exogenous mtDNA in EVs from neurons rescused with mitochoindrial transplanation is worthy to study further to clarify the diversified roles of mitochondrial regulation on the PD pathogenesis.
Mitochondria conjugated with Pep-1. Detailed procedures for Pep-1 conjugation have been described previously 24 . 200 μg allogeneic mitochondria suspended in respiration buffer were conjugated with 0.11 mg Pep-1 diluted with sterile water (Anaspec, San Jose, CA, USA). Incubation for 10 min at room temperature was performed to ensure complex assembly.
Animal study and nasal mitochondria administration. Adult female Sprague-Dawley rats (8 weeks old, weighing 250-300 × g) were purchased from BioLASCO (Taipei, Taiwan) and were maintained under standard laboratory conditions with free access to food and tap water in the Laboratory Animal Center, Changhua Christian Hospital, Changhua, Taiwan. All experimental procedures involving animals were performed in accordance with guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) and approval by the Animal Experiments and Ethics Committee of Changhua Christian Hospital (approval number CCH-AE-105-018). All animal and related experiments were carried out in compliance with the ARRIVE guidelines and followed its instructions of ARRIVE guidelines checklist. A detailed procedure for creating a neurotoxic rat model of Parkinson's disease using a unilateral injection of 6-OHDA into the medial forebrain bundle (MFB) has been described previously 7 . Successful induction of PD rats after three weeks was validated by a rotational behavior test, described below. PD rats were assigned randomly to five groups of six rats: Open field test. To assess general motor behavior, an open field test was employed 3 months after mitochondrial transplantation. Animals were placed in a 50 cm × 50 cm white plexiglass box and allowed an adaptation period of 30 min prior to being analyzed. Activity was recorded for two consecutive sessions, each lasting 15 min, using a ceiling-mounted video camera. Ethovision software (Noldus, Leesburg, VA, USA) (https:// www. noldus. com/ ethov ision-xt) was used to measure distance, velocity, total number of zone boundaries crossed, and duration of movement (seconds). Locomotion frequencies were assessed by dividing the floor of the box into four quadrats and by counting the number of quadrats entered. Entry was counted when a rat entered a new quadrat with all four paws. The apparatus was washed with 5% ethanol between tests to eliminate possible bias due to odors left by previous rats.

Histological and immunohistochemical staining. Rats that survived 3 months after transplantation
were sacrificed with an overdose of chloral hydrate (800 mg/kg i.p.) and fixed by intracardiac perfusion with 300 mL of saline followed by 300 mL of 4% paraformaldehyde (Sigma-Aldrich). Brains were removed, post-fixed in 4% paraformaldehyde for 4 h, and cryoprotected in 30% w/v sucrose in PBS for 20 h. Brains were then frozen and embedded in OCT medium (Tissue-Tek, Sakura Finetek, USA) and sectioned at 5-10-μm. For Nissl staining, sections mounted on glass slides were dried overnight. Slides were immersed in 0.025% cresyl violet (Sigma-Aldrich) in 90 mM acetic acid (Merck, Darmstadt, Germany) and 10 mM sodium acetate (Sigma-Aldrich) for 3 h, followed by dehydration in ascending ethanol and xylene series. Slides were then coverslipped with Histochoice mounting media (AMRESCO).
For immunohistochemical staining (IHC), fixed sections were subjected to heat-induced epitope retrieval in 10 mM citrate buffer pH 6.0 for 25 min at 100 °C. Sections stained with BrdU were additionally treated with 2 N HCl for 30 min at 37 °C. After washing and blocking non-specific sites with blocking buffer (5% BSA and 0.5% Tween-20 in PBS, pH7. After being washed three times in PBS to remove excess secondary antibody, chromogenic detection of immunoreactivity was performed using a DAKO 3,3'-diaminobenzidine (DAB) kit (DakoCytomation, Carpinteria, CA, USA) according to the manufacturer's protocol. A counterstain, Mayer's hematoxylin (Sigma-Aldrich), was then added for 1 min. Color development was terminated by washing, and slides were visualized using light microscopy and image analyzed with Image J Software (NIH, Bethesda, MD, USA). For fluorescent detection in IHC, fluorophore-conjugated secondary antibodies (1:500 dilution in 0.5% BSA/PBS, Jackson ImmunoResearch, West Grove, PA, USA,) were used and nuclei were visualized with 4′,6-diamidino-2-phenylindole (DAPI) (Abcam) counterstaining. Fluorescent signals were detected and Z-stacks were acquired and analyzed with a confocal microscope (Olympus Fluoview FV1200, Olympus, Tokyo, Japan).
Multiplex cytokine assay. Rat blood plasma was collected after 3 months of treatment and after addition Statistical analysis. All analyses were performed in triplicate or quadruplicate in each group of experiments. Biochemical data are presented as means ± standard deviations, except results of the animal behavior test, presented as means ± standard errors of the means. There were six animals per group. The two mitochondrial treatments were evaluated using paired Student's t-tests and differences with p < 0.05 were considered statistically significant. All statistical analyses and graphics were made with GraphPad Prism 5.0 software (GraphPad Software Inc., San Diego, CA, USA).