Tissue engineered nigrostriatal pathway for treatment of Parkinson’s disease

Abstract

The classic motor deficits of Parkinson’s disease are caused by degeneration of dopaminergic neurons in the substantia nigra pars compacta, resulting in the loss of their long-distance axonal projections that modulate the striatum. Current treatments only minimize the symptoms of this disconnection as there is no approach capable of replacing the nigrostriatal pathway. We are applying micro-tissue engineering techniques to create living, implantable constructs that mimic the architecture and function of the nigrostriatal pathway. These constructs consist of dopaminergic neurons with long axonal tracts encased within hydrogel micro-columns. Micro-columns were seeded with dopaminergic neuronal aggregates, while lumen extracellular matrix (ECM), growth factors, and end targets were varied to optimize cytoarchitecture. We found a 10-fold increase in axonal outgrowth from aggregates versus dissociated neurons, resulting in remarkable axonal lengths of over 6 mm by 14 days and 9 mm by 28 days in vitro. Axonal extension was also dependent upon lumen ECM, but did not depend on growth factor enrichment or neuronal end target presence. Evoked dopamine release was measured via fast scan cyclic voltammetry and synapse formation with striatal neurons was observed in vitro. Constructs were microinjected to span the nigrostriatal pathway in rats, revealing survival of implanted neurons while maintaining their axonal projections within the micro-column. Lastly, these constructs were generated with dopaminergic neurons differentiated from human embryonic stem cells. This strategy may improve Parkinson’s disease treatment by simultaneously replacing lost dopaminergic neurons in the substantia nigra and reconstructing their long-projecting axonal tracts to the striatum.

1. Introduction:

Parkinson’s disease is a progressive neurodegenerative disease that affects 1–2% of people over 65, causing significant morbidity across a prolonged and escalating disease course. Parkinson’s disease is characterized by resting tremor, bradykinesia, rigidity, and other symptoms that decrease quality of life, ultimately leading to significant disability via the inability to control motor function (Davie, 2008). The motor symptoms of Parkinson’s disease are due to the selective loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc). As SNpc neurons send long-projecting axons to the striatum, this stereotypical neurodegeneration deprives the striatum of crucial dopaminergic inputs and thereby renders an important motor feedback pathway ineffective.

While several Parkinson’s disease treatments have been developed, most options focus on mitigating symptoms resulting from the neurodegeneration of the dopaminergic neurons rather than treating the underlying pathology itself. Dopamine replacement therapies, including 3,4-dihydroxy-L-phenylalanine (L-DOPA) and dopamine agonists, attempt to correct the underproduction of dopamine due to neuron/axon loss (Davie, 2008). These dopamine-related therapies provide symptom improvements, but side effects, such as dyskinesias and motor fluctuations, may develop and efficacy declines over time (Katzenschlager and Lees, 2002). Alternatively, surgery can be performed to implant deep brain stimulation (DBS) systems that modulate basal ganglia network function to alleviate motor symptoms. While DBS improves motor function, electrode stimulation can have unwanted side effects impacting cognition and speech (Moro, et al., 2010). Similar to pharmaceutical therapies, DBS efficacy decreases over time as the disease progression continues. Newer treatments have focused on neuroprotection, but trials have only slightly slowed progression of symptoms (Obeso, et al.,, 2010). Even if an effective neuroprotective treatment could be developed, the treatment could not replace the ≥60% of SNpc neurons and their axonal projections to the striatum that have already degenerated by the time of symptom onset (Dauer and Przedborski, 2003).

To address stereotypical neurodegeneration of Parkinson’s disease, significant effort has been focused on replacing components of the nigrostriatal pathway via (1) tissue grafts, (2) cell implantation, and (3) scaffold-based methods. Trials have indicated beneficial functional results from tissue grafts when methods sustain a novel population of at least 80,000 dopaminergic neurons (Bjorklund and Lindvall, 2000). However, while grafts and cells implanted into the striatum create new “factories” for dopamine, these cells do not reconstruct the nigrostriatal pathway and thus lack normal feedback signals that target the SNpc. As a result, striatal neurons still do not receive appropriately timed dopamine signals (Freed, et al.,, 2001, Kim, 2011, Olanow, et al.,, 2001, Redmond, et al.,, 2001). Alternatively, implantations into the SNpc have largely failed because axonal regrowth to the striatum is limited and techniques to coax axonal outgrowth do not ensure specific outgrowth to the correct and/or distant target(s) (Grealish, et al.,, 2014, Nikkhah, et al.,, 1995, Thompson, et al.,, 2009). Stem cell implantations have been limited by cell loss, nonspecific cell type differentiation, and neoplastic transformation. Lastly, while biomaterial scaffolds have facilitated axonal regrowth, these treatments do not restore the loss of neural cells and underlying circuitry that is the cause of the motor symptoms (Borisoff, et al.,, 2003, Moore, et al.,, 2006, Silva, et al.,, 2010). Thus, there is currently no strategy that addresses both the loss of SNpc neurons as well as the dopaminergic axonal inputs into the striatum.

As a strategy to physically reconstruct lost long-distance axonal pathways in the brain, we are developing micro-tissue engineered neural networks (micro-TENNs), a novel class of transplantable, anatomically-inspired three-dimensional (3D) cylindrical constructs comprised of discrete neuronal population(s) spanned by internalized long-projecting axonal tracts (Cullen, et al.,, 2012, Harris, et al.,, 2016, Struzyna, et al.,, 2015a, Struzyna, et al.,, 2014, Struzyna, et al.,, 2015b). Despite being only several hundred microns in diameter, micro-TENNs may feature axonal extension of at least several centimeters to reconstruct long-distance brain pathways lost due to trauma or neurodegenerative disease (Cullen, et al.,, 2012, Harris, et al.,, 2016, Struzyna, et al.,, 2015a, Struzyna, et al.,, 2014, Struzyna, et al.,, 2015b). We employ novel micro-tissue engineering techniques to create the neuronal-axonal constructs within a miniature tubular hydrogel featuring an interior extracellular matrix (ECM) core that supports neurite extension in vitro. The miniature form factor allows for minimally invasive delivery into the brain (Harris, et al.,, 2016), while the biomaterial encasement protects the preformed cytoarchitecture and gradually degrades over several weeks (Struzyna, et al.,, 2015b). In rat models, we have previously demonstrated that allogeneic micro-TENNs comprised of cerebral cortical neurons maintained their axonal architecture, and structurally integrated into the native nervous system (Struzyna, et al.,, 2015b).

In the current study, we have significantly advanced this micro-tissue engineering approach by creating transplantable constructs that mimic the general cytoarchitecture of the nigrostriatal pathway. Specifically, we report the construction of the first micro-TENNs comprised of a discrete population of dopaminergic neurons with long-projecting, unidirectional axonal tracts within transplantable miniature tubular hydrogels. We advanced our micro-fabrication techniques by seeding tubular micro-columns with engineered neuronal aggregates generated from primary embryonic rat dopaminergic neurons. We systematically varied ECM constituents, growth factors, and the presence of a neuronal target population in order to optimize the length of unidirectional axonal extension. In vitro, we confirmed that these micro-TENNs exhibited evoked dopamine release and found immunocytochemical evidence that they formed synapses with striatal neurons. Moreover, we stereotaxically implanted preformed dopaminergic micro-TENNs en masse to mimic the nigrostriatal pathway in rats, revealing construct neuronal survival and maintenance of axonal architecture to at least 1 month post-implant. These micro-constructs mimic the cytoarchitecture and functional characteristics of the nigrostriatal pathway, while providing the first evidence of dopaminergic axonal extension >6 mm and evoked dopamine release within implantable three-dimensional constructs. We also showed proof-of-concept for clinical translation by generating micro-TENNs with dopaminergic neurons differentiated from human embryonic stem cells (hESCs), revealing at least several millimeters of unidirectional axonal extension within implantable hydrogel micro-columns. This novel strategy may uniquely address gaps in current Parkinson’s disease treatments by allowing simultaneous replacement of dopaminergic neurons in the substantia nigra as well as their long-distance axonal projections to the striatum (Fig. 1).

Figure 1:

Reconstruction of the Nigrostriatal Pathway Using Micro-Tissue Engineered Neural Networks. A diffusion tensor imaging representation of the long-distance axonal tracts that connect discrete populations of neurons in the human brain. This conceptual rendition shows how a unidirectional micro-TENN – consisting of a population of dopaminergic neurons extending long, aligned processes – can be used to recreate the nigrostriatal pathway (yellow) that degenerates in Parkinson’s disease. Axons in the substantia nigra are expected to functionally integrate with the transplanted dopaminergic neurons in the micro-TENN, while the transplanted dopaminergic axons are expected to functionally integrate with neurons in the striatum. After receiving appropriate inputs from the substantia nigra, the transplanted neurons will release dopamine in the striatum, thereby recreating the circuitry lost in Parkinson’s disease.

2. Materials and Methods:

All procedures were approved by the IACUCs at the University of Pennsylvania and The Michael J. Crescenz Veterans Affairs Medical Center and were carried out in accordance with Public Health Service Policy on Humane Care and Use of Laboratory Animals (2015).

2.1. Three-dimensional Micro-TENN Fabrication

All supplies were from Invitrogen, BD Biosciences, or Sigma-Aldrich unless otherwise noted. Micro-TENNs were comprised of an agarose (Sigma; A9539) ECM hydrogel molded into a cylinder through which axons could grow. The outer hydrogel structure consisted of 1% agarose in Dulbecco’s phosphate-buffered saline (DPBS). The agarose cylinder, with an outer diameter of 398 μm, was generated by drawing the agarose solution into a capillary tube (Drummond Scientific) via capillary action. An acupuncture needle (diameter: 160 μm) (Seirin) was inserted into the center of the agarose-filled capillary tube in order to produce an inner column. Cured micro-columns were pushed out of the capillary tubes using a 30 gauge needle (BD; 305128) and placed in DPBS where they were cut to 6 – 12 mm in length and sterilized under UV light (1 hour). 5 μL of the appropriate ECM cocktail was added to each micro-column. ECM cocktails included: rat tail type 1 collagen, 1.0 mg/mL; rat tail type I collagen, 1.0 mg/ml mixed with mouse laminin, 1.0 mg/ml; mouse laminin, 1.75 mg/ml; and rat tail type 1 collagen, 1.0 mg/mL in 11.70 mM N-(3-Dimethylaminopropyl)-N’-ethylcarboiimide hydrochloride, 4.3 mM N-Hydroxysuccinimide, and 35.6 mM sodium phosphate monobasic. These micro-columns were then incubated at 37°C for 15–30 minutes, after which DPBS was added to the petri dish.

2.2. Neuronal Cell Culture

Female Sprague-Dawley rats (Charles River) were the source for primary ventral mesencephalic neurons, a midbrain region previously shown to be enriched in dopaminergic neurons (Weinert, et al.,, 2015). Carbon dioxide was used to euthanize timed-pregnant rats (embryonic day 14), following which the uterus was extracted. The brains were removed in Hank’s balanced salt solution (HBSS) and the ventral midbrain was isolated (Pruszak, et al.,, 2009). The ventral midbrains were dissociated in accutase for 10 minutes at 37°C. The cells were centrifuged at a relative centrifugal force (RCF) of 200 for 5 minutes and resuspended at 1–2 million cells/mL in standard media consisting of Neurobasal medium + 2% B27 + 1% fetal bovine serum (Atlanta Biologicals) + 2.0 mM L-glutamine + 100 μM ascorbic acid + 4 ng/mL mouse basic fibroblast growth factor (bFGF) + 0.1% penicillin-streptomycin). High concentration growth media consisted of Neurobasal medium + 2% B27 + 1% fetal bovine serum (Atlanta Biologicals) + 2.0 mM L-glutamine + 100 μM ascorbic acid + 0.1% penicillin-streptomycin + 12 ng/mL mouse bFGF + 10 ng/mL brain-derived neurotrophic factor (BDNF) + 10 ng/mL glial cell-derived neurotropic factor (GDNF) + 10 ng/mL ciliary neurotropic factor (CNTF) + 10 ng/mL cardiotrophin. Dopaminergic neuron aggregates were created based on protocols adapted from Ungrin et al. (Ungrin, et al.,, 2008). Custom-built arrays of inverted pyramidal wells were fabricated using polydimethylsiloxane (PDMS) (Sylguard 184, Dow Corning) cast from a 3D printed mold and placed in a 12-well plate. 12 μL of the cell solution was transferred to each pyramidal well, and the 12-well plate was centrifuged at 1500 rpm for 5 minutes, after which 2 mL of standard media was placed on top of each array. The centrifugation resulted in forced aggregation of neurons (approximately 3,200 cells per aggregate). The wells were then incubated overnight. At the time of plating, the DPBS was removed from the dishes containing the micro-columns and replaced with media. Using forceps, the aggregates were inserted into one (unidirectional) or both (bidirectional) ends of the micro-columns, and the cultures were placed in an incubator (total micro-TENNs created with primary dopaminergic neurons: n=310).

H9 human embryonic stem cells (passages 90–100, NIH code WA09; Wicell, Madison, WI) were maintained on mouse embryonic fibroblast feeder cells (CF-1; MTI-Globalstem, Gaithersburg, MD), and cultured in human embryonic stem cell (hES) media containing the following components: DMEM/F12 (Invitrogen, Carlsbad, CA), 20% KnockOut serum replacement (Invitrogen), 1 mM non-essential amino acids (Invitrogen), 1X GlutaMax (Invitrogen), 0.1 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO), 100 U/mL penicillin/100 μg/mL streptomycin (Invitrogen), and 6 ng/mL bFGF (R&D Systems, Minneapolis, MN). The stem cells were differentiated into dopaminergic neurons using the protocol outlined in Kriks, et al. (Kriks, et al.,, 2011). Differentiated dopaminergic neurons (Day 45) were then dissociated using accutase, after which aggregates were created and inserted into micro-columns as described above (total micro-TENNs created with differentiated dopaminergic neurons: n=5). They were maintained in media consisting of Neurobasal medium + 2% B27 + 2.0 mM L-glutamine + 0.2 mM ascorbic acid + 20 ng/mL BDNF + 20 ng/mL GDNF + 1 ng/mL transforming growth factor type β3 + 0.5 mM dibutyryl cAMP + 10 nM DAPT.

For micro-TENNs containing dissociated cells with no ECM core, dopaminergic cells were suspended in standard media at 10 million cells/mL and 5 μL of this cell suspension was added to each micro-column. The micro-TENNs were incubated for 60 minutes, after which media was added. For micro-TENNs containing dissociated cells with an ECM core, dopaminergic cells were suspended in rat tail type 1 collagen, 1.0 mg/mL (10,000,000 cells/mL) at the time of plating and 5 μL of this mixture was added to each micro-column. The micro-TENNs were incubated for 15 minutes, after which media was added.

Pre-warmed media was used to replace the culture media every 3–4 days in vitro (DIV). In some instances, micro-TENNs were transduced with an adeno-associated virus (AAV) vector (AAV2/1.hSynapsin.EGFP.WPRE.bGH, UPenn Vector Core) to express green fluorescent protein (GFP) in the neurons. Here, at 3 DIV the micro-TENNs were incubated overnight in media containing the vector (3.2×10¹⁰ genome copies/mL) and the cultures were rinsed with media the following day.

Female Sprague-Dawley rats (Charles River, Wilmington, MA) were the source for primary striatal neurons. Carbon dioxide was used to euthanize timed-pregnant rats (embryonic day 18), after which the uterus was extracted. To isolate striatal neurons, the brains were removed in HBSS and striata were isolated. The striata were dissociated in trypsin (0.25%) + ethylenediaminetetraacetic acid (EDTA) (1 mM) for 12 minutes at 37°C. The trypsin-EDTA was then removed and the tissue was triturated in HBSS containing DNase I (0.15 mg/mL). The cells were centrifuged at 1000 rpm for 3 minutes and resuspended at 1–2 million cells/mL in Neurobasal medium + 2% B27 + 0.4 mM L-glutamine. Striatal aggregates were created and inserted into micro-TENNs as previously described. When testing if dopaminergic aggregates would form synapses with striatal aggregates, striatal aggregates were inserted into the vacant ends of dopaminergic micro-TENNs at 10 DIV. When testing if striatal aggregates would increase the growth rate of dopaminergic micro-TENNs, they were inserted at 3 DIV.

2.3. Immunocytochemistry

Micro-TENNs were fixed in 4% formaldehyde for 35 min and permeabilized using 0.3% Triton X100 plus 4% horse serum for 60 minutes. Primary antibodies were added (in phosphate-buffered saline (PBS) + 4% serum) at 4°C for 12 hours. The primary antibodies were the following markers: (1) β-tubulin III (1:500, Sigma-Aldrich, cat # T8578), a microtubule element expressed primarily in neurons; (2) tyrosine hydroxylase (TH; 1:500, Abcam, cat # AB113), an enzyme involved in the production of dopamine; (3) microtubule-associated protein 2 (MAP-2) (1:500, Millipore, cat # AB5622), a microtubule-associated protein found in neuronal somata and dendrites; (4) dopamine-and-cAMP-regulated neuronal phosphoprotein (DARPP-32) (1:250, Abcam, cat # AB40801) a protein found in striatal medium-sized spiny neurons; and (5) Synapsin 1 (1:1000, Synaptic Systems, cat # 106001), a protein expressed in synaptic vesicles of the central nervous system. Appropriate fluorescent secondary antibodies (Alexa-488, −594 and/or −649 at 1:500 in PBS + 30 nM Hoechst + 4% serum) were added at 18–24°C for 2 hours.

2.4. Fast Scan Cyclic Voltammetry

Micro-TENNs were fabricated for these studies as described above but using a larger diameter (500 μm ID, 973 μm OD) to allow for greater DA signal. At 24 DIV, micro-TENNs were transferred to a testing chamber and flushed with culture media containing 100 μM L-DOPA. A carbon fiber electrode (150–200 μm length x 7 μm diameter) was inserted either into the dopaminergic aggregate (n=3) or at the axon terminals (n=1), and a bipolar stimulating electrode (Plastics One, Roanoke, VA) was placed across the aggregate. Dopamine release was elicited using electrical pulse trains (30 Hz, 5 ms pulse width, 1 s, monophasic) every 5 minutes and recorded using Demon Voltammetry and Analysis Software (Yorgason, et al.,, 2011). The potential of the carbon fiber electrode was linearly scanned from −0.4 to 1.2 V and back to −0.4 V vs. Ag/AgCl. A voltammeter/amperometer (Chem-Clamp; Dagan Corporation) was used to scan at a rate of 400 V/s, and cyclic voltammograms were recorded every 100 ms. Electrode calibrations of known concentrations of DA (1–10 μM) were used to calculate the concentration of electrically evoked dopamine release at the peak oxidation potential for dopamine in consecutive voltammograms.

2.5. Transplantation of Micro-TENNs

Male Sprague-Dawley rats (325–350g) were anesthetized with isoflurane and mounted in a stereotactic frame. The scalp was cleaned with betadine, bupivacaine was injected along the incision line, and a midline incision was made to expose Bregma. A 5 mm craniectomy was centered at the following coordinates in relation to Bregma: +4.8 mm (AP), 2.3 mm (ML). The micro-TENN was loaded into a needle (OD: 534 μm, ID: 420 μm; Vita Needle, Needham, MA) attached to a Hamilton syringe mounted on a stereotactic arm. The stereotactic arm was positioned at 34° relative to the horizontal plane, the dura was opened, and the needle lowered into the brain to a depth of 11.2 mm. The needle was kept in place for 10 seconds, at which time a stationary arm was positioned to contact the plunger of the Hamilton syringe to ensure that the micro-TENN was expelled. The needle containing the micro-TENN was then withdrawn from the brain. The scalp was sutured closed and buprenorphine was provided for postoperative analgesia. Animals receiving micro-TENNs were survived for 15 minutes, (n=5), 1 week (n=5), or 1 month (n=5). At the time of sacrifice, animals were anesthetized and underwent transcardial perfusion with heparinized saline followed by 10% formalin.

2.6. Immunohistochemistry

After 24 hour post-fix in 4% paraformaldehyde, brains were prepared for optical clearing, parafin processing, or cryosectioning.

For optical clearing, Visikol Inc.’s histology protocol was followed (Merz, et al.,, 2017). Briefly, brains were cut into 1 mm sections using a vibratome, dehydrated in a series of methanol washes, treated with hydrogen peroxide, and then rehydrated. Sections were permeabilized in a buffer containing PBS, 0.2% Triton X-100, 0.3 M glycine, and 20% DMSO. Sections were blocked with 6% horse serum for three days, and then incubated in primary antibodies (Sheep anti-TH, 1:100, Abcam, cat #ab113; Rabbit anti-NeuN, 1:100, cat #abn78) for three days. Next they were incubated in secondary antibodies (1:250) for three days. They were then dehydrated, treated with Visikol® HISTO-1™ for 18 hours, and imaged in Visikol® HISTO-2™.

For paraffin processing, brains were blocked sagittally, processed through paraffin, cut at 8 μm, mounted on slides, and processed for immunohistochemistry. Paraffin sections were deparaffinized and then rehydrated. Endogenous peroxidase was quenched using 3% hydrogen peroxide in water (Fisher, cat #S25359) followed by heat-induced epitope retrieval in TRIS-EDTA. Sections were blocked with horse serum (ABC Universal Kit, Vector Labs, cat #PK-6200) for 30 min. Rabbit anti-TH (1:750; Abcam, cat #ab112) was applied in Optimax buffer overnight at 4°C. The antigen of interest was visualized using DAB (Vector Labs, cat # SK-4100).

For cryosectioning, brains were blocked sagittally, put into 30% sucrose until saturated, and then frozen. Sections were cryosectioned at 35 μm, mounted on slides, and processed for immunohistochemistry. Frozen sections were blocked with 5% normal horse serum in 0.1% Triton-x/PBS for 30–45 minutes. Primary antibodies (Rabbit anti-TH, 1:750, Abcam AB112; Mouse anti-Tuj1, 1:1000, Sigma T8578) were applied to the sections in 2% horse serum/Optimax buffer for two hours at room temperature. Secondary antibodies (1:1000) were applied in 2% horse serum/PBS for one hour at room temperature. Sections were counterstained with Hoechst.

2.7. Microscopy and Data Acquisition

For in vitro analyses, micro-TENNs were imaged using phase-contrast and fluorescence on a Nikon Eclipse Ti-S microscope with image acquisition using a QiClick camera interfaced with Nikon Elements. In order to determine the length of neurite penetration, the longest observable neurite in each micro-TENN was measured from the proximal end of the neuronal aggregate after fixation. For in vitro immunocytochemistry analyses, cultures and micro-TENNs were fluorescently imaged using a Nikon A1RSI Laser Scanning Confocal microscope. All micro-TENN confocal reconstructions were from full thickness z-stacks. For analysis of micro-TENNs post-transplant into the brain, micro-TENNs were fluorescently imaged using a Nikon A1RSI Laser Scanning Confocal microscope. Each section was analyzed to assess the presence, architecture, and outgrowth/integration of micro-TENN neurons/neurites.

2.8. Statistical Analyses

No method was used to pre-determine the sample sizes of groups. Due to obvious visual differences between experimental groups (e.g. dissociated versus aggregate neuron micro-TENNs), in many cases it was not possible for investigators to be blinded to treatment group during experiments or data assessment. For in vivo transplant studies, rats were randomly assigned for use in this experiment. The normality of all data was examined, and adjustments were made for non-normal data. An unpaired, parametric two-sided t-test was performed to determine if there were statistically significant differences in axonal outgrowth between uni-directional versus bi-directional micro-TENNs containing a dopaminergic end target. Unpaired, non-parametric, two sided Mann-Whitney tests were performed to determine if there were statistically significant differences in axonal outgrowth between the following treatment pairs: dissociated versus aggregated cells, high versus regular growth factor concentration, and uni-directional versus bi-directional micro-TENNs containing a striatal end target. An unpaired, non-parametric, two sided Mann-Whitney test was performed to determine if there were statistically significant differences between the lengths of TH+ axons as a percentage of total axonal length with collagen I versus collagen I-laminin cocktail inner cores. ANOVA was performed for the extracellular matrix studies. When differences existed between groups, post-hoc Tukey’s pair-wise comparisons were performed. For all statistical tests, p < 0.05 was required for significance. Data are presented as mean ± standard deviation.

3. Results:

3.1. Micro-TENNs Plated with Dissociated Neuronal Suspensions

Dopaminergic neurons were isolated from the ventral mesencephalon of embryonic rats. In planar culture, these neurons demonstrated a healthy neuronal morphology, the presence of dopaminergic neurons (based on TH expression), significant neurite outgrowth (based on β-tubulin III expression), and network formation out to 28 DIV (Supplemental Figure 1A-C). To create dopaminergic micro-TENNs, we initially seeded micro-columns using dissociated neuronal suspensions. These dissociated cells infiltrated the length of the inner lumen and generally did not produce the desired cytoarchitecture of a discrete cell body region projecting axons across the length of the inner core (Supplemental Figure 1D-H). However, the dissociated neurons within the micro-tissue constructs presented a healthy morphology, and occasionally self-organized into the desired cytoarchitecture by chance (Supplemental Figure 2). In these cases, unidirectional axonal projections achieved lengths of several millimeters, and, importantly, the health of these constructs was also maintained out to 28 DIV (Supplemental Figure 2). In order to see if the inclusion of additional ECM in the inner core increased the consistency with which the correct architecture was generated, we suspended dissociated cells into collagen, and injected the mixture to gel inside the micro-columns. Unfortunately, the presence of the collagen did not aid in producing the desired cytoarchitecture, and the dissociated cells continued to spread throughout the length of the inner core (Fig. 2A1-A3). While these results demonstrated our ability to culture dopaminergic neurons that formed extensive neurite networks within hydrogel micro-columns, these techniques were not sufficient to consistently generate our desired cytoarchitecture.

Figure 2:

Improved Micro-TENN Cytoarchitecture Using Forced Aggregation Method. Phase contrast and confocal reconstructions of micro-TENNs plated with primary dopaminergic neurons at 14 DIV. (A) Representative micro-TENN plated with dissociated neurons labeled via immunocytochemistry to denote neurons/axons (β-tubulin III; green) and cell nuclei (Hoechst; blue). Dissociated micro-TENNs did not demonstrate the desired cytoarchitecture as they showed cell infiltration throughout the entire length of the inner core. (B,C) Phase contrast images depicting micro-TENNs plated with engineered dopaminergic neuron aggregates. Based upon plating technique, aggregates either (B) attached directly outside the agarose micro-column, or (C) inside the inner core. Higher magnification images from demonstrative regions in (B,C) show that while the (D1,D3) cell body regions differed between the two aggregate plating techniques, their (D2,D4) axonal regions were similar. (E) Representative aggregate micro-TENN labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with cell nuclei counterstain (Hoechst; blue). Aggregate micro-TENNs demonstrated the ideal cytoarchitecture, with (E1) discrete cell body regions and (E2,E3) axonal regions. (F) A higher magnification reconstruction from a demonstrative region in (E) depicts the aggregated cell bodies. (G) Micro-TENNs generated using aggregates demonstrated a greater extent of axonal outgrowth than micro-TENNs plated with dissociated neurons (n=13 micro-TENNs each group; Mann-Whitney test, p<0.0001). Data are presented as mean ± standard deviation. Scale bar (A)=250 μm. Scale bar (B,C)=500 μm. Scale bar (D1)=200 μm. Scale bar (D2-D4)=100 μm. Scale bar (E)=250 μm. Scale bar (F)= 50 μm.

3.2. Forced Neuronal Aggregation Method

As our previous micro-TENN fabrication methods did not reliably generate the desired cytoarchitecture, we adapted a method to mechanically group cells into aggregates (Ungrin, et al.,, 2008). After dissociating embryonic tissue into a single cell suspension, this solution was centrifuged in inverted pyramidal wells in order to pellet the neurons at the bottom of the wells. The wells were left in the incubator overnight, during which the pelleted cells became aggregated spheres of neurons (Supplemental Figure 3). Once formed, the aggregates were inserted into the ends of the agarose micro-columns. This method consistently produced micro-TENNs with distinct cell body and axonal regions. Furthermore, we found that based upon the depth and placement of the aggregate within the micro-column, we could create micro-TENNs that exhibited either an externalized or internalized cell body region (Fig. 2B-D). Moreover, this technique produced long-projecting unidirectional axonal tracts, as demonstrated based on TH and β-tubulin III immunoreactivity (Fig. 2E-F). Indeed, as measured by the length of the longest neurite in each micro-TENN, it was determined that the axons projecting from the aggregates grew approximately 10x longer than analogous axons extending within micro-columns seeded with dissociated neurons (Fig. 2G).

3.3. Optimization of Micro-TENN Length

Once we were able to consistently produce micro-TENNs with the desired cytoarchitecture, we optimized micro-column features and environmental conditions to produce micro-TENNs with the longest possible axonal outgrowth. As the nigrostriatal pathway measures approximately 6 mm in the rat, we attempted to generate micro-TENNs at least 6 mm in length. Here we tested the effects of the ECM constituents in the inner lumen, presence of growth factors, and the micro-TENN directionality on outgrowth. It was found that collagen I and collagen I + laminin resulted in the longest axonal outgrowth, as measured by the length of the longest neurite in each micro-TENN (Fig. 3). The average axonal outgrowth for the collagen I and collagen I + laminin cores was 4892 ± 703 μm and 4686 ± 921 μm respectively. In contrast, it was found that crosslinked collagen (1227 ± 481 μm), laminin-coated (205 ± 615 μm), and empty cores (~0 μm) resulted in significantly reduced neurite outgrowth. For the two highest performing groups (lumen comprised of collagen or collagen + laminin), it was determined that TH⁺ dopaminergic axonal projections attained at least 60% of the maximal axonal length (Fig. 3F).

Figure 3:

Effect of Extracellular Matrix on Axonal Outgrowth within Micro-TENNs. Representative confocal reconstructions of dopaminergic micro-TENNs plated with different ECM cores. At 14 DIV, all micro-TENNs were labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with nuclear counterstain (Hoechst; blue). The type of ECM strongly influenced axonal outgrowth, with (A) collagen I (n=12 micro-TENNs) and a (C) collagen I + laminin cocktail (n=12) supporting the longest axonal outgrowth. Micro-TENNs with (B) empty cores (n=9) or (D) crosslinked collagen cores (n=11) demonstrated significantly less outgrowth. (a-d) Higher magnification reconstructions from demonstrative regions in (A-D) show similar expression of TH across groups. (E) A one way ANOVA (p<0.0001) followed by a post-hoc Tukey’s test determined that collagen I and collagen I-laminin cocktail cores were statistically equal (p=0.8590), and that they each supported axonal outgrowth that was statistically longer then outgrowth in empty (p<0.0001), laminin-coated (p<0.0001), or crosslinked collagen (p<0.0001) cores (* denotes significance). (F) As determined by a Mann-Whitney test, the lengths of TH+ axons as a percentage of total axonal length were statistically equivalent between the collagen I (n=12) and collagen I+ laminin (n=12) inner cores (p=0.9723). Data are presented as mean ± standard deviation. Scale bar (A,C,D)= 500 μm. Scale Bar (B)= 250 μm. Scale bar (E-H)= 50 μm.

We also tested the effect of the media growth factor concentration on axonal outgrowth within the micro-columns. A media containing a relatively low concentration of bFGF (4 ng/mL) was compared to media containing high concentrations of growth factors previously shown to increase dopaminergic neuron outgrowth and survival (Hyman, et al.,, 1991). At 14 DIV, we found that the high growth factor concentration media did not result in denser or longer axonal outgrowth compared to the low concentration media (n=14 micro-TENNs in each group; Supplemental Figure 4). Lastly, we investigated whether the use of a target population of dopaminergic cells would increase axonal outgrowth. Bidirectional dopaminergic micro-TENNs were plated by inserting dopaminergic aggregates into both ends of the micro-columns. Since the two dopaminergic neuron populations were separated by 1.2 cm, we sought to assess whether chemotactic signaling between the populations would increase outgrowth. However, at 14 DIV we determined that the axonal outgrowth in bidirectional micro-TENNs was not greater than axonal outgrowth in unidirectional micro-TENNs (n=14 micro-TENNs in each group; Supplemental Figure 4).

Overall, we found that the use of engineered neuronal aggregates and specific ECM constituents were critical factors in axonal extension, while high growth factor media and the presence of a target neuron population did not affect axonal outgrowth. Of note, the mean neuronal aggregate length at 14 DIV was 1165 ± 212 μm; therefore, the total micro-TENN length (neuronal aggregate + axon length) attained using dopaminergic aggregates in collagen was >6 mm by 14 DIV – suitable to span the nigrostriatal pathway in rats.

Following these optimization studies, we fabricated dopaminergic micro-TENNs with an inner core of collagen I and allowed them to grow over 28 DIV to ascertain if axonal extension progressed further within the micro-columns. We found continued axonal extension out to 28 DIV, with lengths of 6046 ± 670 μm for dopaminergic axons, 7697 ± 1085 μm for all axons, and 8914 ± 1187 μm for total aggregate + axon lengths. In some cases, maximum total micro-TENN lengths at this time point were over 10 mm, well beyond what would be required to span the nigrostriatal pathway in rats (Fig. 4).

Figure 4:

Long-Projecting Dopaminergic Micro-TENNs. (A-F) Confocal reconstructions of a representative micro-TENN plated with a dopaminergic aggregate and collagen I inner core at 28 DIV. Micro-TENN labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with nuclear counterstain (Hoechst; blue). (A-D) Long-term dopaminergic micro-TENNs showed robust survival and axonal extension over 28 DIV. (E-F) Higher magnification reconstructions from demonstrative regions in (C) show healthy TH+ neurons and axons, with apparent axonal varicosities suggesting sites of dopamine release. (G) Micro-TENN length measurements taken at 28 DIV (n=7 micro-TENNs) demonstrated TH+ axons measuring 6046 ± 670 μm, and a total TH+ length of 7264 ± 672 μm with the inclusion of the dopaminergic aggregate. Importantly, these lengths are more than sufficient to span the nigrostriatal pathway in rats. Data are presented as mean ± standard deviation. Scale bar (A-D)= 250 μm. Scale bar (E-F)= 50 μm.

3.4. Dopamine Release In Vitro

Fast scan cyclic voltammetry (FSCV) was used to examine the capacity of the micro-TENNs to release dopamine in vitro. At 24 DIV, a carbon fiber electrode was used to record evoked dopamine release in both the dopaminergic aggregate as well as at the distal end of the micro-column containing the terminals of the axonal tracts. Following incubation in media containing L-DOPA, it was found that dopamine release could be elicited in both the somatic and axonal regions (Fig. 5). Cyclic voltammograms recorded during dopamine release in the neuronal aggregate region all demonstrated characteristic oxidation between 0.55 and 0.65 V, and reduction between −0.20 and −0.30 V (Heien, et al.,, 2003). Under the parameters used in these studies, the pattern of oxidation and reduction may be used to correctly identify currents produced by action potential dependent release of catecholamines. Current traces recorded in the aggregate region across several micro-TENNs demonstrated an average peak extracellular dopamine concentration of ~30 nM immediately following electrical stimulation (Fig. 5). Of note, this protocol was also applied to multiple micro-TENNs at 9 DIV but measurements of evoked dopamine release were not attained (data not shown), suggesting the need for functional maturation of dopaminergic neurons/axons over 24 DIV.

Figure 5:

Evoked Dopamine Release from Micro-TENNs In Vitro. (A) Recording set up for FSCV in vitro. A bipolar stimulating electrode (red) was placed to span the dopaminergic aggregate in the micro-column, and was used to evoke dopamine release. A carbon fiber electrode (white) was used to record dopamine release either within the dopaminergic aggregate (R1) or at the axon terminals (R2). (B) Representative color plot displaying the current recorded as the potential of the carbon fiber electrode was linearly scanned from −0.4 to 1.2 V and back to −0.4 V vs. Ag/AgCl every 100 ms. The magnified cross section from the color plot displays an individual cyclic voltammogram exhibiting an oxidation peak at 0.65 V and a reduction trough at −0.3 V, which are characteristic of dopamine. (C) The averaged concentration of dopamine released in the dopaminergic aggregate (R1) following electrical stimulation. Scale bar A= 500 μm.

3.5. Formation of Synapses with Striatal Population

As the dopaminergic axons comprising the nigrostriatal pathway synapse with striatal neurons in the brain, we probed the ability of our tissue engineered nigrostriatal pathway to synapse with a population of striatal neurons in vitro. Dopaminergic micro-TENNs were generated and, after 10 DIV, embryonic rat striatal aggregates were inserted into the vacant ends of the micro-columns. After 4 more DIV, immunocytochemistry was performed in order to assess potential synaptic integration between the two populations. This analysis confirmed the presence of the appropriate neuronal sub-types in the two aggregate populations, specifically TH+ dopaminergic neurons and DARPP-32+ medium spiny striatal neurons (Fig. 6). Moreover, confocal microscopy revealed extensive axonal-dendritic integration and putative synapse formation involving the dopaminergic axons and striatal neurons (Fig. 6D,E,G,H). Also, immunocytochemistry confirmed that the majority of the striatal (DARPP-32+) neurites were also MAP-2+, suggesting that these were dendrites (data not shown). In order to determine if chemotactic cues generated by the striatal population influenced axonal outgrowth from the dopaminergic neuron aggregates, the length of axonal outgrowth was quantified with and without the striatal end target. It was found that the axonal outgrowth in dopaminergic micro-TENNs containing a target population of striatal neurons was not statistically greater than axonal outgrowth in unidirectional dopaminergic micro-TENNs (n=9 micro-TENNs in each group; Fig. 6F).

Figure 6:

Synapse Formation Between Micro-TENN Dopaminergic Axons and Striatal Neurons In Vitro. (A) Representative confocal reconstruction at 14 DIV of a dopaminergic micro-TENN plated with an aggregated striatal end target. The micro-TENN was labeled via immunocytochemistry to denote dopaminergic neurons/axons (TH; pink), striatal (medium spiny) neurons (DARPP-32; green), and synapses (synapsin; purple), with nuclear counterstain (Hoechst; blue). (B-E) Higher magnification reconstructions from demonstrative regions in (A) depict the (B) dopaminergic neuron aggregate, (C) robust, aligned TH+ axons, and (D) neurite outgrowth from the striatal neuron population. (E) A high degree of synapsin labeling (purple) along the trajectory of TH+ axons (pink) suggests that dopaminergic axons formed synapses with the striatal neurons. (F) Micro-TENNs containing striatal end targets did not result in statistically longer axonal outgrowth when compared to unidirectional dopaminergic micro-TENNs with no end target (n=9 micro-TENNs each group; Mann-Whitney test, p=0.9182). Data are presented as mean ± standard deviation. (G-H) Synapsin+ puncta (purple) can be seen decorating putative dendrites projecting from striatal neurons (green) shown with dopaminergic axonal varicosities (pink), further suggesting synaptic integration. Scale bar (A)= 250 μm. Scale bar (B-E)= 50 μm. Scale bar (G-H)= 20 μm.

3.6. Transplant and Survival of Preformed Dopaminergic Micro-TENNs In Vivo

We sought to demonstrate the ability to precisely deliver preformed dopaminergic micro-TENNs into the brain as well as their survival and architecture at various time points post-implant. Accordingly, dopaminergic aggregate micro-TENNs with an inner lumen containing collagen I (in some instances transduced to express GFP) were grown for 14 DIV, after which time they were drawn into a custom needle and stereotaxically micro-injected to approximate the nigrostriatal pathway in adult male Sprague-Dawley rats (Fig. 7A,B). Animals were sacrificed either after 15 minutes, or at 1 week and 1 month time points (n=5 each). The tissue of animals that were sacrificed after 15 minutes was cleared and labeled with the dopaminergic marker TH in order to confirm that the transplantation process itself did not harm the micro-TENN cytoarchitecture, revealing surviving construct neurons in the substantia nigra and maintenance of their axonal projections within the micro-column towards the striatum (Fig. 7C,D). At 1 week and 1 month time points, surviving GFP+ neurons and axons were found within the micro-TENN lumen, which were easily identified spanning the nigrostriatal pathway since the agarose micro-column had only partially degraded at these time points (Fig. 7E,F). Histological sections were co-labeled for the axonal marker β-tubulin III and the dopaminergic marker TH, revealing the preservation of a robust neuronal and dopaminergic axonal population. In particular, longitudinally projecting TH+ axons were present, which confirmed that the micro-TENNs were mostly able to maintain their cytoarchitecture following longer-term transplantation into the brain.

Figure 7:

Micro-TENN Neuronal Survival and Maintenance of Axonal Cytoarchitecture In Vivo. (A) Micro-TENN implant trajectory and dimensions drawn to scale (adapted from Gardoni et al. (Gardoni and Bellone, 2015)). (B) Micro-TENN orientation (not to scale). (C) A thick-tissue longitudinal section showing a micro-TENN 15 minutes following transplantation to span the nigrostriatal pathway and labeled via immunohistochemistry to denote dopaminergic neurons/axons (TH; red). (D) A higher magnification reconstruction from a demonstrative region in (C) depicting both the cell aggregate region and a portion of the axonal region in vivo. (E) At 1 week post-implant, a representative sagittal section showing a longitudinal view of a dopaminergic micro-TENN with all neurons labeled green (expressing GFP on the synapsin promoter) and labeled via immunohistochemistry to denote dopaminergic neurons/axons (TH; red). This demonstrates that micro-TENN neurons survived and the longitudinally aligned cytoarchitecture was at least partially maintained. (F) At 1 month post-implant, a representative oblique section providing a cross-sectional view of a GFP+ dopaminergic micro-TENN (green) labeled via immunohistochemistry to denote dopaminergic neurons/axons (TH; red) and all neurons/axons (β-tubulin III; purple). This demonstrates healthy transplanted neurons/axons with robust dopaminergic axonal projections at 1 month in vivo. Scale bar (C-D)=100 μm. Scale bar (E)=20 μm. Scale bar (F)=50 μm.

3.7. Generation of hESC-Derived Dopaminergic Micro-TENNs

As our eventual goal is to enable translation of this strategy, we sought to assess the feasibility of generating micro-TENNs with dopaminergic neurons derived from human sources. As such, we assessed the ability to fabricate micro-TENNs with dopaminergic neurons differentiated from human embryonic stem cells (hESCs). hESCs were differentiated following the protocol outlined in Kriks, et al. (Kriks, et al.,, 2011), after which they were lifted from two-dimensional culture (Fig. 8A), aggregated, and inserted into the micro-columns. At 14 days following plating in the micro-columns, it was found that these micro-TENNs displayed the correct cytoarchitecture of a discrete somatic zone with unidirectional axonal tracts within the lumen of the hydrogel micro-column. These axonal extensions measured over 4 mm in length at this time point (Fig. 8B). These human dopaminergic micro-TENNs co-labeled for the axonal marker β-tubulin III and the dopaminergic marker TH, and showed a purity of roughly 50% dopaminergic neurons (data not shown).

Figure 8:

Micro-TENNs Generated with Human Embryonic Stem Cell-Derived Dopaminergic Neurons. H9 human embryonic stem cells were differentiated into dopaminergic neurons and labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with nuclear counterstain (Hoechst; blue). (A) Differentiated dopaminergic neurons shown in two-dimensional culture. (B) Differentiated dopaminergic neurons were dissociated, aggregated, and inserted into micro-columns, were they extended robust processes measuring over 4 mm by 14 days post plating. Scale bar (A)=100 μm. Scale bar (B)=225 μm.

4. Discussion:

Current treatments for Parkinson’s disease, including the use of dopamine replacement strategies and DBS, are aimed at alleviating motor disabilities rather than correcting the underlying cause of the motor symptoms. Furthermore, while dopaminergic neuron and/or fetal graft implants into the striatum may provide a local source of dopamine, these approaches do not recreate the nigrostriatal circuit and thus do not provide appropriate feedback control of dopamine release. To address these gaps, we sought to develop a tissue-engineered solution that could be precisely delivered to physically restore lost dopaminergic neurons in the SNpc and their axonal projections to the striatum. To achieve this objective, we built upon our lab’s previously developed micro-tissue engineering platform by generating for the first micro-TENNs utilizing primary dopaminergic neurons. It was found that micro-TENNs plated with neuronal aggregates grew more than 6 mm in length when fabricated with the optimal inner core ECM. Furthermore, the dopaminergic micro-TENNs exhibited evoked dopamine release and were capable of synapsing with a population of striatal cells in vitro, and showed survival and maintenance of cytoarchitecture upon transplant in vivo.

Dopaminergic micro-TENNs were fabricated using a population of ventral mesencephalic neurons that, while enriched in dopaminergic neurons, are not a pure dopaminergic population. The impurity of the population may or may not matter functionally following transplantation, as glutamate is co-released with dopamine in some cases (Broussard, 2012). However, if we find that a higher purity of dopaminergic neurons is necessary for functional efficacy, then we could use a later developmental time point for midbrain isolation (Pothos, et al.,, 1998), cell sorting, and/or differentiation from stem cell sources, all of which have been shown to further enrich dopaminergic populations.

Our method to plate micro-TENNs with dopaminergic “aggregates” alleviated the lack of separation between the neuronal somata and neurites that we observed in some cases when micro-TENNs were plated with dissociated cells. It was particularly important to ensure that the micro-TENNs demonstrated the desired cytoarchitecture of a discrete cell body region projecting axons through the length of the inner core since separate somatic and axonal regions is a key feature of the nigrostriatal pathway. We believe that better approximation of the cytoarchitecture of the nigrostriatal pathway will lead to improved functional outcomes following micro-TENN implantation to reconstruct the degenerated dopaminergic neurons in the SNpc and their projections to the striatum. In particular, the dopaminergic neurons in the SNpc synapse directly onto striatal neurons without exerting their effects through intermediate synapses and/or neurons. Therefore, in order for the connectivity and timing of our micro-TENNs to be correct upon integration with the host, our micro-TENNs will likely need to achieve modulation of striatal neurons by propagating signals from the SNpc through a mono-synaptic pathway. Since we were able to create dopaminergic micro-TENNs that approximate the architecture of the nigrostriatal pathway, this increases the likelihood that functional integration in vivo will emulate the native mono-synaptic pathway.

By utilizing the neuronal aggregates as well as optimizing the inner core ECM, we confirmed that the micro-TENNs were capable of spanning the length of the nigrostriatal pathway in rats, which is approximately 6 mm. In fact, as far as we know, our micro-TENNs exhibit the longest in vitro dopaminergic axonal outgrowth recorded in the literature (Blakely, et al.,, 2011, Denis-Donini, et al.,, 1983, Li, et al.,, 2003, Lin and Isacson, 2006, Nakamura, et al.,, 2000, Park, et al.,, 2006, Shim, et al.,, 2004, Tonges, et al.,, 2012, Wakita, et al.,, 2010, Yue, et al.,, 1999, Zhang, et al.,, 2004). The dopaminergic cell aggregates produced neurite outgrowth that was approximately 10X longer than projections from individual cells. This may be influenced by the fact that the cellular density within the neuronal aggregates is more representative of the density within the brain. This higher cell density may give the aggregated neurons better control over their 3D microenvironment than dissociated cells, which in turn promoted better cell viability and health and therefore enhanced axonal extension. Moreover, the gene expression of cells in the aggregates may also be more representative of cells in the brain and allows the cells to tap into developmental programs to initiate axonal outgrowth (Ungrin, et al.,, 2008). Furthermore, many neuronal subtypes are programmed for either short or long distance communication (Caputi, et al.,, 2013). In dissociated micro-TENNs, “long distance” axons likely meet synaptic partners at intermediate distances along the length of the micro-column. In contrast, for the aggregate micro-TENNs the absence of any intermediate neighbors may have prompted the “long distance” neurons to up-regulate proteins for long distance outgrowth and thus project the length of the micro-columns. Lastly, the aggregates give rise to grouped and fasciculated axonal projections, which may produce a sustained drive for axonal extension due to concentrated and persistent pro-growth signaling, and/or physical/structural advantages. The rates and lengths of dopaminergic axonal extension from the neuronal aggregates were unprecedented, although clearly further studies are required to elucidate the mechanisms of ultra-long axonal projections from the aggregates.

In our attempts to maximize axonal outgrowth, we found that collagen I and a collagen I + laminin cocktail produced the longest outgrowth when used as the biomaterial for the inner core. We believe this occurred because, unlike the laminin coating and empty cores, the collagen I and collagen I + laminin cocktail both provided a continuous, 3D scaffold that supported axonal outgrowth. While the crosslinked collagen also provided a continuous scaffold, it was much stiffer and likely more resistive to enzymatic degradation upon growth cone extension. Interestingly, utilizing media enriched in growth factors or providing striatal neurons as a surrogate end target did not increase axonal outgrowth. As engineered growth factor gradients were not employed, our results suggest that the axons may be relatively non-responsive to global changes in growth signals. Likewise, neither the use of a dopaminergic neuron aggregate nor the use of a striatal neuron aggregate end target increased axonal outgrowth, which does not support our hypothesis that chemotactic cues generated from an end target would accelerate axonal outgrowth. Potentially, supra-threshold peak growth factor concentrations and adequate chemotaxic gradients could not be established due to the significant diffusional distance (1.2 cm) required to span the two populations.

While investigating the in vitro functionality of the dopaminergic micro-TENNs, we used FSCV to confirm that our micro-TENNS could release dopamine both within the neuronal aggregate and at the axon terminals. The FSCV experiments were performed within scaled up micro-TENNs to allow for ease of access for stimulating and recording electrodes. Likewise, while the micro-TENNs were incubated in media containing L-DOPA in order to amplify the dopamine signal for ease of detection, our ability to record evoked dopamine release demonstrates that the machinery necessary for dopamine synthesis and release was present. Allowing the dopaminergic neurons within the micro-TENNs to mature in culture was crucial, as it was difficult to record dopamine release at earlier time points (data not shown). While the cyclic voltammograms demonstrated oxidation between 0.55 and 0.65 V and reduction between −0.20 and −0.30 V, which are characteristic of dopamine (Heien, et al.,, 2003), the voltammograms also exhibited an oxidation peak around 0 V. We believe this additional peak is a result of the release of neurotransmitters from non-dopaminergic neurons present in our impure midbrain populations.

Our in vitro test bed to investigate the micro-TENNs’ ability to synapse with an aggregate of striatal cells revealed extensive synaptic labeling in areas where the dopaminergic micro-TENN axons grew into the striatal neurons. These findings suggest that the dopaminergic axons integrated with the striatal neuron population and provides proof-of-concept for our tissue engineered nigrostriatal pathway in vitro. However, higher resolution tract tracing, transsynaptic markers and/or electrophysiological measures will be required in the future to confirm functional integration in this test bed.

Animals sacrificed immediately following injection of micro-TENNS to span the nigrostriatal pathway revealed that the transplantation process itself did not alter the cytoarchitecture of the micro-TENNS. Furthermore, at one week and one month following injection of GFP+ micro-TENNs, evidence of micro-TENN survival and maintenance of cytoarchitecture was found. Specifically, surviving GFP+ and TH+ neurons were located within the lumen along the injection trajectory. Out to one month post-implant, histological sections orthogonal to the implant trajectory revealed dense GFP+ neurons/axons and TH+ axons in cross-section within the micro-column. While these findings are sufficient to demonstrate proof-of-concept for the implant paradigm and neuronal survival in vivo, future chronic studies will utilize immune suppressed animals in order to mitigate potential loss of transplanted neurons/axons due to host immune responses. In addition, the circuitry of the nigrostriatal pathway is complex, and while the linear trajectory of implantation was adequate to physically span the substantia nigra and striatum, our implantation trajectory may be refined in future studies using curved micro-TENNs to take into account relevant sub-structures within these anatomical substrates. Most importantly, it is imperative that the micro-TENNs undergo further testing in a rat model of Parkinson’s disease in order to determine their ability to ameliorate motor symptoms and restore dopamine levels in the striatum.

Our proposed strategy may be the first approach capable of simultaneously replacing dopaminergic neurons in the substantia nigra while physically reconstructing their long axonal tracts to the striatum. This will allow the implanted micro-TENNs to be subject to the normal cellular modulation that dopaminergic SNpc neurons are subject to in vivo to “close the loop” and restore a crucial circuit for motor control feedback. Our process to generate micro-TENNs enables a precisely engineered structure where the number of neurons and generation of dopamine can be known prior to implantation. Furthermore, in principle our method surpasses graft and cell-based methods by simultaneously restoring neurons and their long-distance axons to the striatum. In order to supply the 80,000 dopaminergic cells and >3.0 cm axonal lengths needed for functional improvement in humans, our micro-TENNs will eventually need to be scaled up considerably for clinical trials. As a first step towards this benchmark, our successful generation of micro-TENNs using dopaminergic neurons derived from hESCs demonstrates feasibility of using human cell sources. In the future, we foresee the generation of autologous micro-TENNs using cellular reprogramming techniques, which may eliminate the need for immunosuppression (Swistowski, et al.,, 2010). While it is possible that these autologous micro-TENNs would eventually degenerate similar to native tissue, previous research has shown that dopaminergic neurons from tissue grafts can survive decades in vivo despite ongoing degeneration of the native dopaminergic system (Cai, et al.,, 2005, Kim, et al.,, 2008). Furthermore, we envision the development of Parkinson’s disease-resistant micro-TENNs via genetic approaches to minimize α-synuclein (the primary pathological protein in Parkinson’s disease) induced pathology to lengthen the micro-TENN lifespan. Thus, as micro-TENNs are more targeted than systemically delivered drugs, have a smaller form factor than DBS electrodes, and provide the opportunity to reconstruct lost neuroanatomy, they have the potential to revolutionize Parkinson’s disease treatment and dramatically improve patient outcomes.

Supplementary Material

Supplemental Figure 1:

Dissociated Ventral Mesencephalon Neurons in Planar Culture and Within Micro-Columns. Neuronal cultures labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with nuclear counterstain (Hoechst; blue). (A-C) Dopaminergic neurons in planar culture showing significant neurite outgrowth and random network formation out to 28 DIV. (D-G) Representative micro-TENN at 7 DIV initially plated as a cell suspension. The single cell suspension infiltrated the length of the inner core of the micro-column and did not demonstrate the desired cytoarchitecture requiring separate cell body and axonal regions. (H1-H3) Higher magnification reconstructions from demonstrative regions in (D-G). Neurons and neurites present in a micro-TENN plated with a cell suspension show a lack of organization and directionality. Scale bar (C)= 250 μm. Scale bar (G)=200 μm. Scale bar (H3)=125 μm.

Supplemental Figure 2:

Chance Cell Aggregation in Micro-TENNs. Micro-TENNs plated with single cell suspensions labeled via immunocytochemistry to denote all neurons/axons (β-tubulin III; green) and dopaminergic neurons/axons (TH; red), with nuclear counterstain (Hoechst; blue). (A-D) At 14 DIV, a micro-TENN showed the desired cytoarchitecture due to chance re-aggregation of the dissociated neurons. (E-H) At 28 DIV, a micro-TENN exhibiting chance re-aggregation of dissociated neurons; however, axonal extension had only reached ~2800 μm in length. (I,J) Higher magnification reconstructions from demonstrative regions in (E-H) showing that chance re-aggregation created discrete regions of (I) cell bodies and (J) axons in some cases. Scale bar (D)= 100 μm. Scale bar (H)=250 μm. Scale bar (I,J)=125 μm.

Supplemental Figure 3:

Neuronal Aggregate Fabrication and Planar Outgrowth. (A) A custom-designed, 3D printed mold was used to generate the inverted PDMS wells. (B) Representation of PDMS wells used to aggregate cells. Each individual well measured 4 mm wide by 4 mm long by 3.46 mm deep. (C) An array of PDMS wells inserted into a 12-well culture plate. (D) Concentrated neuronal cell suspensions were pipetted into inverted pyramidal wells. The wells were centrifuged to aggregate the neurons by force, after which the wells were incubated overnight to allow the pelleted cells to adhere to each other. (E) A neuronal aggregate pelleted at the bottom of a custom-fabricated inverted PDMS well. (F) A dopaminergic aggregate plated in planar culture was labeled via immunocytochemistry to denote dopaminergic neurons/axons (TH; purple). Aggregates achieved neurite outgrowth of over 5000 μm in length by 8 DIV. Scale bar (E)=150 μm. Scale bar (F)=500 μm.

Supplemental Figure 4:

Effects of Growth Factor Concentration and Target Neuronal Population on Axonal Outgrowth within Micro-TENNs. (A) Micro-TENNs cultured in media containing high growth factor concentrations did not result in longer axonal outgrowth than micro-TENNs cultured using standard media (n=14 micro-TENNs each group; Mann-Whitney test, p=0.9851). (B) Micro-TENNs plated bi-directionally with dopaminergic aggregates on both ends (n=14) did not result in longer axonal outgrowth than micro-TENNs plated uni-directionally (n=14 each group; t-test, p=0.5219). Data are presented as mean ± standard deviation