Synaptic vesicle characterization of iPSC-derived dopaminergic neurons provides insight into distinct secretory vesicle pools

Introduction

The selective degeneration or loss of dopaminergic (DA) neurons in the brain plays a central role in the development of Parkinson’s disease (PD). Several studies have directly linked abnormalities in synaptic vesicle (SV) recycling as critical contributors to the preferential vulnerability of DA neurons, particularly in the development of early-onset parkinsonism (EOP)^1,2,3,4. DA neurons are known for their large and highly arborized axon terminals comprising of dense varicosities innervating the striatum^5,6,7. Moreover, DA neurons have an innate ability to fire in two patterns: tonic (4–5 Hz) and bursting (~20 Hz) phases⁸. Since the type of firing pattern can drastically affect neurotransmitter release^9,10, understanding the spatial architecture of DA synapse is crucial in elucidating the preferential vulnerability of these neurons to synaptic dysfunction.

Synapses can be broadly categorized into classical or neuromodulatory depending on the type of neurotransmitters released. While classical chemical synapses for glutamate and γ-aminobutyric acid (GABA) are fast and target receptors in nanoscale proximity (presynaptic terminals and post-synaptic densities), neuromodulators such as dopamine operate at different spatiotemporal scales¹¹. In fact, DA neurons are thought to have predominantly in-passing presynaptic boutons with no clear postsynaptic component in the brain, and to a lesser extent the classical chemical synapses^12,13,14.

A crucial aspect that is lacking in the synaptic biology of DA neurons is the organelle(s) responsible for dopamine release. The storage of monoamines (serotonin, dopamine, histamine, adrenaline, and noradrenaline) is performed by the vesicular monoamine transporters (VMATs) 1 and 2. Both transporters exhibit tissue-specific abundance, with VMAT1 in neuroendocrine while VMAT2 found in DA and serotonin-containing neurons of the central nervous system^15,16. Unlike glutamate and GABA transmitters which are primarily stored in small clear synaptic vesicles (SSVs), the vesicles storing dopamine are reportedly heterogenous in their identities. For instance, VMAT2, the primary transporter for neuronal dopamine, is described to localize on SSVs and secretory dense core vesicles (DCVs)^17,18 or exclusively in DCVs^19,20. The presence of VMAT2 in secretory vesicles is consistent with the localization of VMAT1 on DCVs of the adrenal chromaffin cells^19,21,22. However, little is known on the role of secretory vesicles on dopamine release in DA neurons. Moreover, evidence for SV transmitter pleiomorphism in DA neurons is primarily attributed to electrophysiological and behavioral analysis of experiments relating to transmitter co-release mechanisms^{23,24,25,26,27}. Recent evidences have also pointed to distinct variations in sorting pathways for VMAT2- and vesicular glutamate transporter (VGLUT)-containing vesicles, highlighting intrinsic differences in trafficking mechanisms for different SV transmitter pools^23,28,29. Hence, a detailed characterization of the vesicles secreting dopamine becomes even more critical in our understanding of synaptic malfunction in neurodegenerative diseases such as Parkinson’s disease.

Towards this aim, our work seeks to gain insight into the anatomical and molecular organization of SV transmitter pools in DA neurons. Using a combination of cellular-based systems including glutamatergic iPSC-derived neurons (i³Neurons), iPSC-derived DA neurons and co-cultures of iPSC-derived DA neurons with their striatal target medium spiny neurons (MSNs), we found that DA neurons form distinct vesicle pools of varying sizes and nature: classical SSVs, large vesicles and DCVs. Using an ectopic SV-like organelle reconstitution system in fibroblasts, we show that VMAT2-positive vesicles can be clustered by synapsin and are sorted in heterogenous vesicular populations distinct from the classical VGLUT-containing vesicle clusters. In DA neurons, vesicles positive for VMAT2 do not resemble bona fide SSVs in terms of morphology, size and clustering properties. Our findings provide evidence that DA neurons use different classes of vesicles for the secretion of dopamine and of glutamate during synaptic transmission.

Results

Characterization and maturation of iPSC-derived DA neurons

We used previously described DA differentiation protocol^30,31 to derive DA neurons from iPSCs (See Methods). Analysis of the cultures by immunofluorescence showed that 30 days from the induction of differentiation, 76.23% ± 7.8 (mean ± S.D.) of the cells were positive for the neuronal marker βIII-tubulin, and 89.21% ± 6.7 (mean ± S.D.) of the βIII-tubulin-positive neurons were positive for the DA neuron marker tyrosine hydroxylase (TH) (Fig. 1A, B). Western blot analysis confirmed expression of TH, as well as expression of the plasma membrane DA transporter (DAT) in these cells (hence referred to as DA neurons), but not in neurogenin-2 induced iPSC-derived glutamatergic cortical-like neurons (i³Neurons, day 19) used as a control (Fig. 1C).

Conversely, expression of typical SV markers, such as synaptophysin, synapsin, VAMP2 and Rab3, were not present at relevant level in DA neurons at 30 days, but robust expression was observed at 50 days by immunofluorescence, which revealed the expected punctate synaptic pattern (Fig. 1D, E, Supplementary Fig. 1A–C).

Characterization of synaptic function with the calcium indicator Fluo-4 showed basal spontaneous activity within the neuronal population (Fig. 1F, G) while acute depolarization with high K⁺ stimulation (30 mM KCL) induced a synchronized rise in calcium levels in these neurons (Fig. 1H, I). Collectively, these results indicate that iPSC-derived DA neurons exhibit synaptic properties typical of a neuron from day 50 onwards in culture.

iPSC-derived DA neurons show vesicle pools distinct from those in i³Neurons

We next used transmission electron microscopy (EM) to examine the presence and type of secretory vesicles in neurite varicosities of cortical-like i³Neurons and DA neurons. In i³Neurons, numerous tightly-packed typical presynaptic clusters of ~40 nm vesicles were already clearly visible at day 19 post-differentiation (Fig. 2A–C). In DA neurons, the appearance of clusters of similar vesicles along axonal processes was much delayed relative to i³Neurons (Fig. 2D–F), consistent with the delayed expression of typical SV marker proteins (Fig. 1 and Supplementary Fig. 1). In favorable sections, many such clusters were clearly anchored to a region of the plasma membrane directly opposed to a region of neighboring cells positive for a thick or thin plasma membrane undercoating, confirming that they represent synaptic sites. Moreover, three morphologically distinct populations of such clusters were observed.

**Fig. 2: i³Neurons show presence of classical SSVs while iPSC-derived DA neurons show size pleiomorphism in SV pools.**

In one population, the ~40 nm vesicles were found as in i³Neurons (Fig. 2D, see also Fig. 2B, C), and the post-synaptic membrane had the thick undercoating typical of post-synaptic densities of asymmetric synapses. These synapses are likely glutamatergic, consistent with the reported property of DA neurons in situ to form VGLUT-positive glutamatergic synapses^23,32.

In another population, clusters of irregularly shaped vesicles with a clear content and around 60–100 nm in diameter were frequently observed (Fig. 2E). These vesicles were tightly packed, completely segregated from other organelles and generally localized in outpocketing of neuronal processes, reminiscent of bouton-like structures lacking clear postsynaptic densities. In fact, staining for postsynaptic density protein 95 (PSD95), a postsynaptic marker, revealed a marked reduction in PSD95 labeling adjacent to synapsin-positive structures in DA neurons, compared to those found in i³Neurons (Supplementary Fig. 2A, B). Interestingly, appearance of these large vesicles has been observed in DA axonal boutons of the mouse nucleus accumbens¹³, whose identity has yet to be characterized. Analysis of the vesicle size distribution of both populations in DA neurons showed an average diameter of 54.9 ± 20.5 nm (mean ± S.D.), in comparison to those in i³Neurons (49.3 ± 5.8 nm and mean ± S.D., Fig. 2G, H).

In addition to the two populations of SV clusters described above, another vesicle population observed in DA neurons along cell processes was represented by 60–100 nm round, oval or irregularly shaped larger vesicles with an electron dense content. These vesicles were enriched in neurite varicosities but were much sparser than the other two populations described above. Their appearance suggests that they represent neuro-peptide containing large dense core vesicles (DCVs) (Fig. 2F). The occurrence of DCVs was previously observed in nigrostriatal axons²⁰. Their presence is of special interest, as in cells specialized for amine secretion, such as chromaffin cells of the adrenal gland, amines are stored in DCVs that also contain a variety of peptides.

To determine whether both small and large vesicles with clear contents are endocytic organelles, cholera toxin-horseradish peroxidase (CTX-HRP) was added to unstimulated, K⁺ stimulated, and during recovery from K⁺ stimulation. CTX-HRP labels endocytic organelles from the cell surface and is an efficient marker for tracing recently formed SVs. Incubation with CTX-HRP for 2 h revealed a few vesicles positive for HRP most likely due to spontaneous neuronal activity, and a further increase in the number of HRP-positive vesicles after 30 h (Supplementary Fig. 2C–E). Notably at 30 hours, most of the vesicles including the small and large ones were positive for HRP (Supplementary Fig. 2B). High K⁺ stimulation for 90 s resulted in massive formation of large HRP-positive vacuoles previously recognized as bulk endosomes (Supplementary Fig. 2D), often detected in the presynapse of mouse primary neuronal cultures after high K⁺ stimulation³³. Following recovery from high K⁺ stimulation, the majority of SVs were positive for HRP, indicating that the large pool of newly-formed SVs originated from the large vacuoles induced by high K⁺ stimulation (Supplementary Fig. 2F–H).

Confirming the presence of large vesicles and DCVs in axon terminals

The presence of organelles reminiscent of large vesicles and DCVs which could be storage sites for DA raises the possibility that at least a pool of DA secreted from DA neurons in the striatum may be released from these organelles. As in neurons, these vesicles can be present both in dendrites and axons, we wanted to confirm that the neurites of DA neuron containing them were axons. To address this question, we used a microfluidic compartmentalization device in which neurons are seeded in one chamber but can extend axons to another chamber through long microchannels (640μm in length, Fig. 3A, Supplementary Fig. 3A, B).

**Fig. 3: Co-culturing of iPSC-derived DA neurons with their striatal target neurons show size pleiomorphism in SV pools.**

DA neurons from a 30-days old culture were replated in one of the chambers and allowed to grow their axons through the microchannels. To facilitate visualization of the axons, DA neurons derived from tdTomato-tagged TH iPSCs were used to track axonal outgrowth in the microchannels (Fig. 3B–D). Moreover, to facilitate axonal outgrowth and formation of synapses, 10 days after plating the DA neurons, iPSC-derived striatal medium spiny neurons (MSNs), which we confirmed to be positive for DARPP32 immunoreactivity by western blotting and immunofluorescence (Supplementary Fig. 3C–E), were seeded in the target chamber. A major synaptic target of DA neurons in the striatal region (nigrostriatal pathway) is the medium spiny neurons (MSNs), which represent 90% of the striatal neuronal population³⁴. Following an additional 10 days, anti-synapsin immunofluorescence of the target chamber yielded abundant puncta revealing presence of axon varicosities (Fig. 3E, F). Importantly, EM revealed the same abundant presence of the organelles reminiscent of irregularly shaped large vesicles and DCVs, revealing that these organelles populate axons (Fig. 3G–J).

Distinct localization of VMAT2 and VGLUT2 in DA neurons

VMAT2 is the vesicular transporter responsible for the loading of dopamine into secretory vesicles of nigrostriatal DA neurons. A non-overlapping localization of VMAT2 and VGLUT2 is in fact supported by previous studies of axons of DA neurons in the ventral striatum and primary cultures of DA neurons^23,27,32. To further validate this difference, we assessed the localization of VMAT2 and VGLUT2 in our iPSC-derived DA neurons at 50–55 days of differentiation, i.e., stage when synapses have matured (see Fig. 1). Since individual axons can be easily identified from these neurons by tracing their length, immunofluorescence of these cultures for VGLUT2 and synapsin demonstrated close colocalization of the two proteins in typical synaptic puncta (93.9% ± 7.6; mean ± S.D.), (Fig. 4A, A’, D), confirming presence of glutamate-containing SVs. Since antibodies that yield reliable VMAT2 immunofluorescence were not available, expression of VMAT2 tagged at its C-terminus with either GFP or mCherry was used in this study. VMAT2-GFP colocalized with synapsin (like VGLUT2 vesicles), suggesting that they are indeed SVs and not just endosomes resulting from overexpression (91.8% ± 6.7; mean ± S.D.), Fig. 4B, B’, D). When DA neurons expressing tagged-VMAT2 were stained for VGLUT2, there was indeed only a partial overlap between VGLUT2 and VMAT2 signal (43.5% ± 22.4; mean ± S.D.), Fig. 4C, C’, D). In addition, we further validated the presence of endogenous VMAT2 and VGLUT2 in these neurons by western blot from day 30 and day 60 cultures (Supplementary Fig. 4A). Detection of VMAT2 in striatal mice lysates showed presence of more than one band (Supplementary Fig. 4B, C), which could arise from region-dependent modifications to VMAT2³⁵. Of note, VGAT, the GABA vesicular transporter, was also detected in these neurons (Supplementary Fig. 4D), and its presence in mouse DA neurons has also been previously documented¹⁴, suggesting strong resemblance of mouse SV pools in iPSC-derived DA neurons. Collectively, our results indicate that the types of transporters, specifically VMAT2 and VGLUT2, do not have the same localization, phenocopying what had been observed in primary neuronal cultures and striatal brain slices^23,25.

**Fig. 4: Fluorescent-tagged VMAT2 show minimal overlap with VGLUT2 in iPSC-derived DA neurons.**

To further understand the type of vesicles VMAT2 is localized on in these neurons, we first performed correlative light and electron microscopy (CLEM) for VMAT2 in DA neurons. Towards this aim, VMAT2-GFP was transfected into DA neurons at day 50 on gridded glass dishes and visualized for VMAT2 fluorescence at day 53. VMAT2-GFP-positive structures were then identified on the gridded dish, and selected regions were processed for EM analysis (Fig. 4E). EM observation revealed that clusters of VMAT2-GFP fluorescence in control DA neurons corresponded to a mixture of predominantly larger vesicles (>60 nm, Fig. 4E’, F), which are mostly empty and occasionally electron-dense, indicating that VMAT2 proteins are indeed present on larger-sized secretory vesicles.

Differential sorting of VMAT2 and synaptophysin when expressed in an ectopic system

To further assess differences in the intracellular sorting of VMAT2 relative to other SV proteins, we capitalized on a recent demonstration that clusters of SV-like organelles – condensates that appear as large droplets in fluorescence microscopy – can be generated in fibroblastic cells (COS7 cells) by exogenous expression of synapsin and synaptophysin^36,37. We had found that when additional bona fide SV proteins, such as VAMP2, SCAMP5, synaptotagmin 1, VGLUT1, VGAT1 and Rab3A are expressed together with synaptophysin and synapsin in these cells, they co-assemble with synaptophysin in the same vesicles. Thus, we examined whether also VMAT2, like these other proteins, can assemble into vesicles generated by synaptophysin expression.

First, we expressed VMAT2-GFP alone in COS7 cells and found that it localizes to the cis- and trans– Golgi complex, in addition to scattered vesicular puncta throughout the cytoplasm (Supplementary Fig. 5A). When co-expressed together with mCherry-synapsin, VMAT2 co-assembled with synapsin into droplet-like condensates (Fig. 5A, B, Supplementary Fig. 5B, C) reminiscent of those generated by synaptophysin and synapsin (Fig. 5C, D). However, these condensates were composed of larger and irregularly shaped vesicles (81.1 ± 34.9 nm; mean ± S.D.), clearly different from those found in the synaptophysin-synapsin condensates (43.3 ± 8.2 nm ; mean ± S.D.) (Fig. 5B, D, E, F). Strikingly, when VMAT2-GFP, synaptophysin and mCherry-synapsin were co-expressed together, VMAT2 vesicles and synaptophysin vesicles segregated from each other and assembled into distinct phases within the mCherry-synapsin phase, with synapsin being more concentrated (based on a higher fluorescence intensity) in the synaptophysin subphase (Fig. 5G–J). Most interestingly, CLEM of cells co-expressing synaptophysin, synapsin and VMAT2 revealed that the two phases detectable by fluorescence correlated with two classes of vesicles: small SV-like vesicles in the synaptophysin phase and larger vesicles in the VMAT2 phase (Fig. 5K). Synaptophysin and synapsin condensates have previously been shown to have liquid-like property³⁶. To test whether VMAT2-synapsin droplets are phase separated vesicles, both synaptophysin (untagged) and VMAT2 (VMAT2-GFP) were co-expressed with synapsin (mCherry-synapsin) separately and photobleached for measurement of fluorescence recovery kinetics (Fig. 5L, M). Both types of condensates showed a similar fast fluorescence recovery after photobleaching for mCherry-synapsin, supporting the role of phase separation mechanisms mediated by synapsin in the formation of both VMAT2 and synaptophysin condensates. Furthermore, other SV proteins, such as VAMP2, Rab3, synaptotagmin-1, SCAMP5 or SV2C (an SV protein highly expressed by nigrostriatal DA neurons), were all positively colocalized with the VMAT2-synapsin clusters (Fig. 6A–E). Moreover, either VGLUT2 (VGLUT2-GFP, Fig. 7A, B) or VGLUT1 (VGLUT1-GFP, Fig. 7C, D), when co-expressed with VMAT2 (VMAT2-FLAG), synaptophysin (untagged) and synapsin (mCherry-synapsin) showed selective preference of the glutamate transporters for the synaptophysin-synapsin condensates over the VMAT2-synapsin clusters. In addition, since VGAT, the GABA transporter was also detected by western blot in DA neurons (Supplementary Fig. 4D), we co-express VGAT (VGAT-FLAG) with VMAT2 (VMAT2-GFP), synaptophysin (untagged) and synapsin (mCherry-synapsin). Like the VGLUTs, VGAT preferred the synaptophysin-synapsin (Fig. 7E, F) over the VMAT2-synapsin condensates. The localization of VGAT to synaptophysin vesicle condensates is consistent with previous findings³⁷, suggesting that VGAT-positive vesicles are most likely comprised of SSVs.

**Fig. 5: Co-expression of VMAT2-GFP with mCherry-synapsin results in clusters of large vesicles distinct from the synaptophysin-synapsin vesicle clusters.**

**Fig. 6: VMAT2 and synapsin vesicle clusters recruit other SV proteins.**

**Fig. 7: VGLUT1, VGLUT2 and VGAT show preferential recruitment to synaptophysin-synapsin over VMAT2-synapsin vesicle clusters.**

Knockdown (KD) of AP3, an adaptor protein known to be important for the intracellular sorting of VMAT2²³, prevented the formation of VMAT2-synapsin condensates (Supplementary Fig. 6A–E) but did not impact the formation of synaptophysin-synapsin condensates (Supplementary Fig. 6F, G), pointing to different sorting pathways for the two proteins. Interestingly, knockdown of AP2, an adaptor protein essential for clathrin-mediated endocytosis, did not impede formation of both VMAT2 and synaptophysin vesicle clusters (Supplementary Fig. 7), suggesting Golgi-dependent mechanisms for these vesicles, in addition to endocytic origins³⁷. The separation of VMAT2-positive vesicles from vesicles that share greater similarity to bona fide SVs is consistent with the differential localization of VGLUT2 and VMAT2-postive vesicles in mouse DA neurons of the ventral striatum^23,25,32 and in iPSC-derived DA neurons as shown above. Note that while DA neurons have both large vesicles and DCVs, lack of a dense core in these vesicles is not unexpected, as fibroblastic cells do not express the molecular machinery to generate the peptide containing secretory granules of the classical regulated secretory pathway³⁸. Hence it remains possible that VMAT2 is localized to both large clear and DCVs in DA neurons.

Presence of small and large vesicles in axonal varicosities of mouse striata

To gain direct insight on the structural differences in SV populations of DA axonal terminals in situ, EM of striatal homogenate from brains of adult mice (3–6 months old) (Fig. 8A) was performed, where the somatodendritic regions of DA neurons are absent in this region. Examination of the crude striatal homogenate embedded in agarose showed preservation of individual synaptosomes, with dense appearance in the cytoplasm. In some nerve terminals, tightly-packed SSVs with clear post-synaptic densities (PSDs) were observed (Fig. 8B), however, the presence of varicosities lacking clear PSDs (bouton-like) comprising of pleiomorphic vesicle pools were frequently observed (Fig. 8C). Since available antibodies to VMAT2 are not specific, we immunolabel striatal homogenate isolated from HA-tagged VMAT2 transgenic mice with HA (to label VMAT2) and synaptophysin antibodies in the agarose-embedded synaptosomes (Fig. 8D, E). Indeed, the sizes of vesicles labeled by HA were much larger than those labeled with synaptophysin (Fig. 8G, H). Immunogold antibody staining for synapsin showed labeling of both SSVs and large vesicles in the agarose-embedded striatal synaptosomes (Fig. 8F, G, H), consistent with those visualized in axonal varicosities of iPSC-derived DA neurons (Fig. 5). These data suggest that DA neurons are comprised of different SV populations with distinct vesicle size identities in vivo.

**Fig. 8: DA axonal varicosities show presence of small and large vesicles in situ.**

Discussion

Our results show a striking morphological difference in the SV pool populations of iPSC-derived DA neurons. In mice, DA nerve terminals were previously described to contain SVs positive for both VMAT2 and VGLUT2, necessary for the storage of dopamine and glutamate transmitters, respectively^{23,25,27,32,39}. While both VMAT1 and VMAT2 are known to localize to secretory vesicles like DCVs in neuroendocrine cells^19,40,41, it is generally believed that dopamine-containing vesicles in neurons are morphologically identical to the VGLUT- and VGAT-containing SV counterparts: clear vesicle clusters with a vesicular diameter of about 40 nm^1,42. We now demonstrate that DA nerve terminals show three pools of secretory vesicles: SSVs, large empty vesicles and DCVs in iPSC-derived DA neurons and synaptosomes from DA nerve terminals of mouse striata.

The large vesicles were positive for VMAT2 in DA neurons and in an SV-like cluster reconstitution system in fibroblasts. This finding opens the possibility of using iPSC-derived DA neurons as a synaptic model to gain new insight into the still open question of how DA neurons regulate neurotransmitter release in nerve terminals, which are predominantly comprised of non-classical synapses^11,12,13,43.

An important result of our study, which was also previously documented in mouse and rat striata^23,32, is the partial segregation of VGLUT2 and VMAT2-positive SVs in DA neurons. We show that VMAT2 localize on vesicles larger than bona fide SVs in iPSC-derived DA neurons and are also positive for synapsin. In addition, while vesicular transporters such as VGLUT1 and VGAT self-organize into small SV-like vesicle clusters by co-expressing synaptophysin and synapsin in fibroblasts³⁶, the co-expression of VMAT2 with synaptophysin and synapsin resulted in the segregation of two distinct clusters: synaptophysin-synapsin and VMAT2-synapsin vesicle pools. The larger-sized VMAT2-positive vesicles in fibroblasts were strikingly reminiscent of those large vesicles (including DCVs) in DA neurons, which we further confirmed to be VMAT2-positive by CLEM. More importantly, co-expression of several SV proteins including Rab3, VAMP2, synaptotagmin 1, SCAMP5 and SV2C but not VGLUT1 or VGLUT2 were found to localize to the VMAT2-positive vesicles in fibroblasts, suggesting that both transporters are differentially sorted in cells. Furthermore, it has been previously reported that the lack of AP3 abolished the formation of VMAT2-postive vesicles but not the VGLUT2 vesicles in DA neurons²³, suggesting differential recycling pathways for both dopamine and glutamate transporters. Indeed, we showed that knockdown of AP3 in fibroblasts abolished formation of VMAT2-synapsin vesicle clusters, but not the synaptophysin-synapsin clusters.

More recently, large scale tracing of DA neurons in mouse nucleus accumbens using volume-based serial EM revealed presence of small and large vesicle clusters in DA axonal varicosities, predominantly lacking classical synapses¹³. These large vesicles were either clear or electron-dense (reminiscing those of DCVs) in nature, which were identified to be predominantly in non-classical synapses lacking PSDs. It is very likely that these vesicles are VMAT2-positive, based on our CLEM findings in iPSC-derived DA neurons, SV-like overexpression system in fibroblasts, and EM morphometric and immunolabeling analyses of the mouse striata.

Collectively, our study demonstrates that, both in dopaminergic nerve terminals and in an exogenous SV-like organelle reconstitution system, major SV proteins and VMAT2 are differentially sorted and represent an important step towards the elucidation of how dopamine transmitters are stored and released during synaptic transmission. iPSC-derived DA neurons would serve as a fantastic model to study how classical and non-classical synapses are interconnected to each other, a facet of dopaminergic neurobiology that is still insufficiently characterized.

Methods

Antibodies and plasmids

All VMAT2 and VGLUT1/2 constructs used in this study utilized a human codon-optimized sequence and were cloned using the Gateway recombination cloning approach. The following donor constructs were obtained from RESOLUTE consortium SLC collection on Addgene: pDONR221_SLC18A2 (#131997), pDONR221_SLC17A7(#131960), pDONR221_SLC17A6 (#131948), pDONR221-SLC6A3_STOP (#161326), pDONR221_SV2C (#132297). Briefly, VMAT2 (SLC18A2), VGLUT1 (SLC17A7), VGLUT2 (SLC17A6) and SV2C were tagged at the c-terminus with EGFP, mCherry or 2xFLAG in the following destination vectors pDEST-eGFP-N1(#31796), pDEST-mCherry-N1 (#31907) and 2xFlag-pDEST-C (#118372), respectively. The SLC6A3 (DAT) was n-terminally tagged with mCherry in the pDEST-CMV-N-mCherry (#123215) destination vector. C-terminal FLAG tagged VGAT was generated by PCR amplification from VGAT-pHluorin and ligated into linearized vector from synaptophysin-FLAG by In-Fusion HD Cloning (Takara Bio). The following plasmids: untagged synaptophysin, synaptophysin-FLAG, untagged synapsin 1a, mCherry-synapsin 1a, VAMP2-mCherry, mCherry-Rab3a, mCherry-SCAMP5, Synaptotagmin1-mCherry were previously generated in the laboratory of Pietro De Camilli. Each construct was validated by DNA sequencing. All antibodies used in this study are listed in Supplementary Table 1.

Human iPSC culture, i³Neuron and DA differentiation

The following iPSC lines were obtained from the iPSC Neurodegeneration Initiative (iNDI) consortium and genome-edited by Jackson Laboratories (JAX): KOLF2.1, KOLF2.1 with the NGN2 cassette at the AAVS locus (used for the i³Neurons experiments) and tdTomato-tagged TH KOLF2.1 iPSCs. For the maintenance of iPSCs in culture, iPSCs were cultured on Geltrex (Life Technologies) coated dishes and maintained in Essential 8 Flex media (Thermo Fisher Scientific). The Rho-kinase (ROCK) inhibitor Y-27632 (EMD Millipore, 10 μM) was added to Essential 8 Flex media on the first day of plating and replaced with fresh media without ROCK inhibitor on the following day.

For i³neuronal differentiation, iPSCs were differentiated into cortical-like i³Neurons according to a previously described protocol based on the doxycycline inducible expression of Ngn2⁴⁴. Briefly, iPSCs were dissociated with Accutase (Thermo Fisher Scientific) and re-plated at a density between 1.5–3 × 10⁵ cells on geltrex-coated dishes in induction medium [(KnockOut DMEM/F-12 (Thermo Fisher Scientific) containing 1% N2-supplement (Thermo Fisher Scientific), 1% MEM non-essential amino acids (Thermo Fisher Scientific), 1% GlutaMAX (Thermo Fisher Scientific) and 4 μg/mL doxycycline (Sigma-Aldrich)]. After 3 days, pre-differentiated i³Neurons were dispersed using Accutase and plated on 0.1 mg/ml poly-L-ornithine (Sigma-Aldrich) in borate buffer and 10 μg/ml laminin (Thermo Fisher Scientific) coated 35 mm glass-bottom dishes (MatTek) or 6-well plates (Corning) for imaging and immunoblotting, respectively. These i³Neurons were cultured and maintained in cortical medium (induction medium supplemented with 2% B27 (Thermo Fisher Scientific), 10 ng/mL BDNF (PeproTech), 10 ng/mL NT-3 (PeproTech) and 10 μg/mL laminin). Fresh cortical media was added to the existing media every 5 days. The iPSCs and i³Neurons were kept at 37 °C with 5% CO₂ in an enclosed incubator. A detailed protocol can be found at https://www.protocols.io/view/culturing-i3neurons-basic-protocol-6-n92ld3kbng5b/v1.

For the differentiation of iPSCs to DA neurons, we used the following protocols described in Kriks, et al.³¹ and Bressan, et al.³⁰. Briefly, iPSCs were dissociated with Accutase (Thermo Fisher Scientific) and re-plated at a density of 8 × 10⁵ cells per well (of a 6-well plate) on geltrex-coated dishes in Essential 8 Flex media with Rock inhibitor. On the next day (Day 0 of differentiation), the media was replaced with knockout serum replacement (KSR) media containing 500 nM LDN193189 (STEMCELL Technologies) and 10 μM SB431542 (STEMCELL Technologies). KSR medium is comprised of Knockout DMEM/F12 medium, 15% Knockout serum replacement (Thermo Fisher Scientific), 1% MEM NEAA, 1% glutaMAX, 0.1% 2-mercaptoethanol (Thermo Fisher Scientific) and 0.2% penicillin-streptomycin (Thermo Fisher Scientific). Starting the following day (Day 1) 75% of the differentiation medium was replaced with new medium each day from day 1 to day 15, then every 2 days until day 20. For days 1–4, KSR medium containing 500 nM LDN193189, 10 μM SB431542, 200 ng/ml SHH C25II (R&D Systems), 2 μM Purmorphamine (Cayman Chemical Company) and 100 ng/ml FGF-8b (PeproTech) was added daily, supplemented by the addition of 4 μM CHIR99021 on day 3 and 4. For days 5 and 6, a mixture of 75% KSR + 25% N2 medium also containing 500 nM LDN193189, 10 μM SB431542, 200 ng/ml SHH C25II (R&D Systems), 2 μM Purmorphamine (Cayman Chemical Company), 100 ng/ml FGF-8b (PeproTech) and 4 μM CHIR99021 (Tocris) was added to the cells followed by equal amounts of KSR and N2 media on days 7–8, and 25% KSR + 75% N2 media on days 9–10 also containing 500 nM LDN193189, 10 μM SB431542, 200 ng/ml SHH C25II and 4 μM CHIR99021. The N2 medium is comprised of Neurobasal Plus media (Thermo Fisher Scientific), 2% B27 supplement without vitamin A (Thermo Fisher Scientific), 1% N2 supplement, 1% glutaMAX and 0.2% penicillin-streptomycin. For days 11–20, complete NB/B27 medium was added to cells, with the addition of 4 μM CHIR99021 on days 11 and 12 only. Complete NB/B27 medium is comprised of N2 medium (without the N2 supplement) and the following components: 20 ng/ml BDNF (PeproTech), 0.2 mM ascorbic acid (Sigma-Aldrich), 20 ng/ml GDNF (PeproTech), 0.5 mM db-cAMP (Sigma-Aldrich), 1 ng/ml TGFβ3 (R&D Systems) and 10 μM DAPT (Cayman Chemical Company). After 20 days of culture, DA progenitor cells were frozen in Synth-a-freeze cryopreservation media (Thermo Fisher Scientific) and stored at -80 ^°C or liquid nitrogen.

For long-term culture of DA neurons, cells were re-plated on 0.1 mg/ml poly-L-ornithine in PBS (Sigma-Aldrich) and 10 μg/ml laminin (Thermo Fisher Scientific) coated 35 mm glass-bottom dishes (MatTek) or 6-well plates (Corning) for imaging and immunoblotting, respectively. These neurons were cultured and maintained in complete NB/B27 medium followed by the addition of 0.1% anti-mitotic inhibitor (Supplement K, Brainxell) at day 25 to terminate division of non-neuronal cells. Fresh NB/B27 medium was added to the existing plates or dishes every 7 days and kept at 37 °C with 5% CO₂ in an enclosed incubator. A detailed protocol can be found at dx.doi.org/10.17504/protocols.io.dm6gp39m8vzp/v1.

Cell culture and transfections

COS7 cells were grown in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS (Thermo Fisher Scientific) and 1% penicillin-streptomycin. Cells were kept at 37 °C with 5% CO₂ in an enclosed incubator. For transfection of COS7 cells, 1 μl Lipofectamine™ 2000 Transfection Reagent (Invitrogen) was used with the respective plasmids and visualized within 24–48 h. For both i³Neuron and DA neuron transfections, plasmids were transfected with 4 μl of Lipofectamine™ Stem Transfection Reagent (Invitrogen) and visualized at least 48 h later. For siRNA-mediated knockdown in COS7 cells, 3 μl Lipofectamine™ RNAiMAX Transfection Reagent was used with the following siRNAs: adaptor related protein complex 3 delta 1 subunit (AP3D1) (Dharmacon, ON-TARGETplus SMART siRNA, LQ-016014-00-0002) and adaptor related protein complex 2 mu 1 subunit (AP2M1) (Dharmacon, ON-TARGETplus SMART siRNA, L-008170-00-0005) at a final concentration of 10 nM. Knockdown efficiency was validated by immunoblotting for AP3D1 and AP2M1, 48 h after transfection.

Neuronal co-culture device

tdTomato-tagged TH DA neurons (day 30) were replated on one side of the two-chamber microfluidic compartmentalization device (OMEGA⁴, eNuvio), where only axonal processes can migrate through the microfluidic channels connected to the adjacent chamber. After an additional 25 days in the co-culture device, frozen iPSC-derived medium spiny neurons (MSN) from Brainxell were plated on the other half of the device (where only the axons of DA neurons are present). The DA-MSN co-cultures were then fixed 7–10 days later for immunofluorescence or EM analysis. A detailed protocol can be found at dx.doi.org/10.17504/protocols.io.dm6gpze38lzp/v1.

Immunofluorescence, live imaging and fluorescent microscopy

For all imaging, cells were seeded on glass-bottom matTek dishes (MatTek corporation). For immunofluorescence visualization, cells were fixed with 4% (v/v) paraformaldehyde (Electron Microscopy Sciences) in 1x phosphate-buffered saline (PBS) for 20 min followed by three washes in PBS. Cells were permeabilized with 0.25–0.5% (v/v) Triton X-100 in PBS for 5 min followed by three washes in PBS. Cells were then incubated with fresh 1 mg/ml sodium borohydride (Sigma-Aldrich) in PBS for 7 min to reduce autofluorescence, and then washed thrice in PBS. They were further blocked for 30 min in 5% bovine serum albumin (BSA, Sigma-Aldrich) in PBS and then incubated overnight at 4 °C with the primary antibodies listed in Supplementary Table 1. Subsequently, cells were washed with PBS thrice the following day and incubated with Alexa Fluor-conjugated secondary antibodies (Thermo Fisher Scientific) for 1 h at room temperature, followed by three washes in PBS. DAPI (Thermo Fisher Scientific) was used for nuclear staining.

For calcium imaging, cells were incubated with FLUO-4 (Thermo Fisher Scientific) at a final concentration of 1 μM for 15 min followed by 2 washes in neuronal media. Transfections were carried out as described above. For live imaging, cells were maintained in Live Cell Imaging buffer (Life Technologies) for COS7 cells, while both i³Neurons and DA neurons were maintained in CM and NB/B27 media, respectively, in a caged incubator with humidified atmosphere (5% CO₂) at 37 °C. The Yokogawa spinning disk field scanning confocal system with microlensing (CSU-W1 SoRa, Nikon) controlled by NIS elements (Nikon) software was used for neuronal imaging. Excitation wavelengths between 405 and 640 nm, CFI SR Plan ApoIR 60XC WI objective lens and SoRa lens-switched light path at 1x, 2.8x or 4x were used. SoRa images were deconvolved using the Batch Deconvolution (Nikon) software. The Andor Dragon Fly 200 (Oxford Instruments) inverted microscope equipped with a Zyla cMOS 5.5 camera and controlled by Fusion (Oxford Instruments) software was used for imaging of COS7 cells. Excitation wavelengths between 405-640 nm and Plan Apo 60x (1.45-NA) objective lens were used to visualize COS7 cells.

For fluorescence recovery after photobleaching (FRAP) assay, transfected COS7 cells in live imaging buffer were imaged for 10 s before bleaching, then regions of interest were bleached with 561 nm wavelength at 30% power for 350 ms controlled by NIS elements (Nikon operating software). After bleaching, fluorescence recovery was observed at the same power as before photobleaching. Further analysis on fluorescence recovery was performed with ImageJ plugin, FRAP Analysis.

Immunoblotting

i³Neurons, DA neurons and MSNs were grown on six-well plates (3–5 × 10⁵ cells/well). After differentiation in their respective maturation media, neurons were washed with ice-cold PBS and then lysed in 1xRIPA lysis buffer (10X RIPA lysis buffer, Sigma-Aldrich) supplemented with cOmplete™ EDTA-free protease inhibitor cocktail (Roche) and PhosSTOP phosphatase inhibitor cocktail (Roche), followed by centrifugation at 13,000 × g for 6 min. For striatal brain lysate, dissected striatal regions of the mice were lysed and sonicated in 2% SDS, followed by centrifugation at 13,000 × g for 5 min. The supernatant was collected and incubated at 95 °C for 5 min in SDS sample buffer containing 1% 2-mercaptoethanol (Sigma). The extracted proteins were separated by SDS-PAGE in Mini-PROTEAN TGX precast polyacrylamide gels (Bio-Rad) and transferred to nitrocellulose membranes (Bio-Rad) at 100 V for 1 h or 75 V for 2 h (for high molecular weight proteins: >150 kDa). Subsequently, the nitrocellulose membranes were blocked for 1 h with 5% non-fat milk (AmericanBIO) in TBST (tris-buffered saline [TBS] + 0.1% tween 20), then incubated overnight at 4 °C with primary antibodies and then incubated with IRDye 680RD or 800CW (LI-COR) secondary antibodies (1:8000) for 1 h at room temperature in TBST. For VGLUT2 immunoblotting, blot was re-probed after stripping with Restore™ PLUS Western Blot Stripping Buffer (Thermo Fischer Scientific) for 10 min, followed by three washes with TBST three times, and re-blocked as described above. Finally, blots were imaged using the Gel Doc imaging system (Bio-Rad) using manufacturer’s protocols. A detailed protocol can be found at dx.doi.org/10.17504/protocols.io.3byl49eqjgo5/v1.

EM sample preparation and CLEM

COS7 cells were plated on 35 mm gridded, glass-bottom MatTek dish (P35G-1.5-14-CGRD) and transfected as described above. Cells were prefixed with 4% (v/v) PFA in Live Cell Imaging Buffer (Life Technologies) for 15 min followed by three times washing with the same buffer. Regions of interest were selected by fluorescence light microscopy imaging and their coordinates were identified using phase contrast. Cells were further fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer for 1 h at room temperature followed by 4 times washing with 0.1 M sodium cacodylate buffer for 5 min each. Cells were postfixed in 2% OsO₄ and 1.5% K₄Fe(CN)₆ (Sigma-Aldrich) in 0.1 M sodium cacodylate buffer on ice, then washed 4 times with Milli-Q water for 5 min each. DA neurons were prefixed with 4% PFA as described, further fixed with 2.5% glutaraldehyde and 2 mM CaCl₂ in 0.15 M sodium cacodylate buffer for 1 h at room temperature, and postfixed in 2% OsO₄, 1.5% K₄Fe(CN)₆ and 2 mM CaCl₂ in 0.15 M sodium cacodylate buffer for 1 h on ice. The specimens were stained with 2% uranyl acetate dissolved in Milli-Q water for 1 h. After washing 4 times with Milli-Q water, specimens were dehydrated in a graded series of ethanol (50%, 75%, and 4 times in 100% for 5 min each), followed by embedding in Embed 812. Regions of interest were relocated based on the pre-recorded coordinates, sectioned and imaged.

For HRP uptake assay, neurons were incubated with 15 μg/ml HRP conjugated CTX (Thermo Fischer Scientific) at 37 °C. HRP reaction was carried out with diaminobenzidene (Sigma-Aldrich, D-5637) (0.5 mg/ml) and H₂O₂ (JTBaker, 2186-01) (0.01%) in 0.1 M ammonium phosphate buffer (pH7.4) after the glutaraldehyde fixation step. Ultrathin sections (50–60 nm) were observed in Talos L120C TEM microscope at 80 kV. Images were taken with Velox software and a 4k × 4 K Ceta CMOS Camera (Thermo Fischer Scientific). Except when noted, all EM reagents are from Electron Microscope Sciences.

Agarose embedding of striatal brain homogenate

All mice were maintained on a 12 h light/dark cycle with standard mouse chow and water ad libitum. HA-VMAT2 bacterial artificial chromosome (BAC) transgenic mice were generous gifts from Robert Edwards (UCSF)²³. All research and animal care procedures were approved by the Yale University Institutional Animal Care and Use Committee. The preparation of striatal homogenate containing synaptosome fractions was adapted from previously published protocols with some modifications^45,46. Briefly, animals were euthanized, decapitated and the striatal regions of the mice were quickly dissected under a stereomicroscope within 5–10 min from death on ice. Tissues were mildly homogenized by several manual strokes in about 1–2 volumes of ice-cold homogenate buffer (0.25 M sucrose, 5 mM sodium phosphate (PB), pH 7.4) in a glass-Teflon homogenizer. The homogenate was then fixed by the addition of 1 volume of ice-cold fixation buffer (6% PFA, 0.25% glutaraldehyde, 0.5 M sucrose, 10 mM PB, pH 7.4) for 30 min on ice. The fixed subcellular particles were pelleted by centrifugation (13,000 g for 45 min at 4 °C). The pellet was scraped with a spatula from the bottom of the tube, resuspended in 100–200 μl of 120 mM PB buffer, and gently re-homogenized. The homogenate was mixed with 1 volume of warm liquid 3% agarose (Sigma-Aldrich, A9045-25G) at 58 °C to prepare blocks of agarose-embedded particles as described⁴⁵.

Immunogold labeling of agarose embedded striatal brain homogenate

Fixed agarose blocks were incubated with 10 mM Tris-HCl buffer (pH 7.4) for 30 min to quench unreacted aldehyde groups. These blocks were then incubated in blocking buffer (3% BSA or 3% goat serum (Gibco) in 0.02 M PB) for 30 min, followed by incubation with primary antibodies diluted in antibody buffer (0.5% BSA or goat serum in 0.02 M PB) at 4 °C overnight. After 6 washes with 0.1 M PB for 5 min each, the antibody-bound agarose blocks were further fixed with 1% GLA in 0.12 M PB for 30 min, followed by three washes in 120 mM PB. The agarose blocks were labeled by immunogold in antibody buffer for 3 h, then washed 6 times in 0.1 M PB for 5 min each. These blocks were further fixed with 1% glutaraldehyde in 120 mM PB, then washed thrice in 120 mM PB. Subsequently, the blocks were post-fixed in 1% OsO₄ in 100 mM PB on ice, followed by 4 washes of 100 mM PB and three washes of MilliQ water on ice. The blocks were then dehydrated and embedded in Epon as described above. Ultrathin sections (50–60 nm) were obtained from these blocks and imaged using EM as described above.

Fluorescence image analysis, EM and statistics

Images were pseudocolor-coded, adjusted for brightness and contrast, cropped and/or rotated using the open-source image processing software FIJI (ImageJ)⁴⁷. For Fig. 1B, semi-automatic segmentation using Cellpose in Napari was used to segment the nucleus (DAPI staining), followed by manual identification of βIII-tubulin and/or TH signals. Immunoblot data were processed using Image Lab software (Bio-Rad) and quantified using the “Gels” plugin in FIJI. The methods for statistical analysis and sizes of the samples (n) are specified in the results section or figure legends for all of the quantitative data. Student’s t-test or Mann–Whitney test was used when comparing two datasets. Quantification of vesicle diameter was performed using randomly sampled images from clustered vesicles in classical synapses and bouton-like structures for DA neurons, and large droplets in COS7 cells. The diameter of synaptic vesicles was measured from the average of long and short axes. Measurements were performed using FIJI (ImageJ). Differences were accepted as significant for P < 0.05. Prism version 9 (GraphPad Software) was used to plot, analyze and represent the data.

Statistical analysis

The methods for statistical analysis and sizes of the samples (n) are specified in the results section or figure legends for all quantitative data. Student’s t-test or Mann–Whitney test was used when comparing two datasets. Differences were accepted as significant for P < 0.05. Prism version 9 (GraphPad Software) was used to plot, analyze and represent the data.