Quantitative mapping of transcriptome and proteome dynamics during polarization of human iPSC-derived neurons

The differentiation of neuronal stem cells into polarized neurons is a well-coordinated process which has mostly been studied in classical non-human model systems, but to what extent these findings are recapitulated in human neurons remains unclear. To study neuronal polarization in human neurons, we cultured hiPSC-derived neurons, characterized early developmental stages, measured electrophysiological responses, and systematically profiled transcriptomic and proteomic dynamics during these steps. The neuron transcriptome and proteome shows extensive remodeling, with differential expression profiles of ~1100 transcripts and ~2200 proteins during neuronal differentiation and polarization. We also identified a distinct axon developmental stage marked by the relocation of axon initial segment proteins and increased microtubule remodeling from the distal (stage 3a) to the proximal (stage 3b) axon. This developmental transition coincides with action potential maturation. Our comprehensive characterization and quantitative map of transcriptome and proteome dynamics provides a solid framework for studying polarization in human neurons.


INTRODUCTION
Neuronal development is a complex multistep process in which neurons undergo dramatic morphological changes, including migration, axon outgrowth, dendritogenesis, and synapse formation. Much of the fundamental knowledge about neuronal development is based on experimental studies in non-human model systems, such as Drosophila, C. elegans, mice and rats (Zhao and Bhattacharyya 2018). However, to what extent the knowledge obtained in animal models can be extrapolated to human neuronal development remains largely unclear. Moreover, analysis of humanspecific characteristics is hindered by the difficulty in obtaining human brain tissue. The generation of human induced pluripotent stem cells (iPSCs) has provided a critical step forward for studying the development and function of human neuronal cells.
In recent years, many labs have used human iPSC-derived neuronal cultures to study fundamental neurobiological questions. This has contributed to our understanding of processes such as neuronal polarity, spine development and synaptic plasticity in human cells. For example, human iPSC-derived model systems have been used to study dynamic changes in gene expression during early neurogenesis, and to study polarization of neuronal progenitors (Compagnucci et al. 2015;Grassi et al. 2020). In addition, human synaptic transmission and plasticity have been studied at single cell level in hiPSC-derived neurons, and human-specific protein functions have been shown to regulate excitatory synaptic transmission specifically in human neurons (Meijer et al. 2019;Marro et al. 2019).
These examples illustrate how the use of human iPSC-derived neurons as a model system can lead to novel findings for human neurodevelopment.
One of the classic model system to study neuronal development consists of are dissociated rat hippocampal neurons as developed by Banker and collaborators (Dotti, Sullivan, and Banker 1988).
These neurons undergo five well-defined developmental stages, transforming from round, spherical cells to fully mature, polarized neurons (Craig and Banker 1994). First, the symmetric young neuron forms small processes (stage 1) and multiple neurites (stage 2). Next the cells undergo polarization, where one neurite is specified as the axon (stage 3), while the remaining neurites will further develop into dendrites. The axon rapidly extends and further matures by the formation of the axon initial segment (AIS) (Leterrier 2018). The AIS is required for generating action potentials (APs) and maintaining neuronal polarity. In addition to the classic AIS component Ankyrin-G (AnkG), the microtubule binding protein Trim46 also localizes to the AIS and is critical for axon formation by forming parallel microtubule bundles in the proximal axon (van Beuningen et al. 2015;Gumy et al. 2017;Harterink et al. 2018). As the neuron matures, the developing axons and dendrites undergo significant morphological and molecular changes and form dendritic spines (stage 4-5), which allow for the formation of synaptic contacts and the establishment of functional neuron-to-neuron interactions (Harris and Kater 1994;Fletcher, De Camilli, and Banker 1994;Grabrucker et al. 2009). In depth proteomic analysis of primary rat neurons in culture have identified a number of specific pathways and unique protein profiles that contribute to various aspects of neurodevelopment processes (Frese et al. 2017). Proper characterization and quantitative profiling of transcriptome and proteome dynamics is essential to study the specific neurodevelopment events in human iPSC-derived neuronal cultures, including early developmental changes such as neuronal polarization and axon specification.
In this study, we performed extensive characterization of the early developmental stages of hiPSCderived neurons by immunocytochemistry, electrophysiology, RNA sequencing, and stable isotope labeling combined with high-resolution liquid chromatography-tandem mass spectrometry (LC-MS/MS). We established transcriptomic and proteomic profiles of the early developmental stages (stage 1-3), comprising 14,551 transcripts and 7,512 protein identifications, of which we assessed 1,163 and 2,218 factors that showed differential expression, respectively. These transcriptomic and proteomic profiles point to the importance of microtubule cytoskeleton remodeling in the early stage of neuronal development. Combining this framework with additional methods such as genetic manipulation and live-cell imaging allowed us to investigate the cellular and molecular processes during neuronal polarization and axon outgrowth. Specifically, we identified a distinct, previously unrecognized developmental stage during early axon development, characterized by the reorganization of the axonal microtubule network and relocation of AIS proteins from the distal to proximal axon. The transition through these early axon developmental stages coincided with the time window in which maturation of action potentials occurred. Together, our study provides a quantitative description of transcriptomic and proteomic profiles of hiPSC-derived neuron cultures, which is a rich resource for further analyses of critical signaling pathways during early human neurodevelopment.

Characterization of developmental stages in human iPSC-derived neurons
We first systematically assessed if the human iPSC-derived neurons proceed through the initial neurodevelopmental stages, which have previously been described in dissociated rat neurons (Dotti, Sullivan, and Banker 1988). The hiPSC-derived neuron cultures were obtained by neuronal induction of neuronal stem cells (NSCs) and maintained up to ~15 days (Axol Bioscience, Protocol Version 5.0, Human iPSC-derived Neural Stem Cells). hiPSC-derived neurons were transduced with FUGW-GFP lentivirus to visualize cell morphologies, and immunostained at different timepoints (day 1, 5 and 14) for proliferation marker Ki67 and NSC marker Nestin to identify NSCs (stage 1), neuron specific markers ß3-Tubulin and MAP2 to identify differentiated neurons (stage 2), and AIS markers AnkG and Trim46 to identify polarized neurons (stage 3) (FIG 1A,B). The presence of these markers indicated a clear developmental transition over time: stage 1 NSCs (~day 1) differentiated into stage 2 neurons with a characteristic neuronal morphology (~day 5), and subsequently developed axons containing AIS structures (~day 14) (FIG 1C-E). Accordingly, axon width showed a developmental decrease over time, whereas dendrite width remained relatively stable (FIG S1A,B). These observations indicate that hiPSC-derived neurons have a relatively prolonged development compared to dissociated rat neurons, consistent with the protracted development of the human brain (Dotti, Sullivan, and Banker 1988;Petanjek et al. 2011). This further supports the emerging evidence showing species-dependent differences in developmental timing of human and non-human neurons in vivo and in vitro (Shi, Kirwan, and Livesey 2012;Espuny-Camacho et al. 2013;Nicholas et al. 2013;Otani et al. 2016;Linaro et al. 2019). We further analyzed the structural AIS organization in axons of hiPSC-derived neurons by quantifying the average fluorescence intensity profiles of Trim46 and AnkG (FIG 1F,G). Consistent with previous reports in dissociated rat neurons, we found that Trim46 and AnkG localization largely overlapped, with the peak of AnkG intensity located ~6 µm more distally than the peak Trim46 intensity (FIG 1F,G) (van Beuningen et al. 2015). The AIS structure was also enriched for voltage-gated sodium channels (NaV), which strongly overlapped with AnkG structures (FIG S1C). In summary, these data demonstrate that human iPSC-derived neurons follow the characteristic sequence of developmental stages during neuronal polarization, which occurs at a relatively slower rate than non-human neurons.

Action potential firing of polarized human iPSC-derived neurons
To determine whether the polarized human iPSC-derived neurons are able to fire APs, we performed whole-cell patch clamp recordings of stage 3 neurons. We observed AP firing upon positive somatic current stimulation in nearly all cells (59/61). Neuronal excitability was quantified as the number of APs in response to increasing current stimuli (steps of 5 pA; 400 ms) in 54 neurons (FIG 1H,I). Of these, 22 neurons (41%) fired multiple times upon higher current stimulation, while 32 neurons (59%) fired only once independent of stimulus strength (FIG S1D). Neurons that fired one AP showed more immature intrinsic cell properties, including depolarized resting membrane potential, lower input resistance, smaller maximum sodium currents and smaller after-hyperpolarization (FIG S1E-H).
Spontaneous AP firing or incoming spontaneous synaptic responses were only observed in a few neurons (4/22). This is in line with previous studies reporting that synapse formation typically starts around two weeks after neuronal induction in human iPSC-derived neurons (Zhang et al. 2013).
Together, the immunofluorescence and electrophysiology data indicate that human iPSC-derived neurons develop functional axons after neuronal polarization.

Transcriptomic profiling of developing human iPSC-derived neurons
To assess global changes in gene expression during differentiation and neuronal polarization, we next performed an unbiased, in-depth analysis of the transcript expression profiles during early neuronal development. To this end, we collected human iPSC-derived neurons at the previously described developmental stages and monitored mRNA expression changes using quantitative population-based transcriptome analysis. The synchronized differentiation and relatively slow development of these cultures enabled us to select time points at which particular stages manifested in the majority of the cells. The cells were sampled for RNA analysis at days 1, 3, and 7, corresponding to stage 1, the onset of stage 2, and the onset of stage 3, respectively. At the same days, we collected samples for in-depth proteome analysis, which is discussed below. In two biological replicates with each two technical replicates, we identified transcripts corresponding to 14,551 genes by RNA sequencing (TABLE 1,   FIG S2A). Of these, 9,655 transcripts were successfully quantified at all time points and normalized to reads per million for further analysis (TABLE 2). As expected, most genes with significantly altered expression were found between day 1 and day 7: 614 genes were downregulated and 549 genes were upregulated at day 7 compared to day 1 (FDR <0.05 (Yoav Benjamini and Yosef Hochberg, 1995)) (TABLE 3,  Upregulated genes correspond to several cellular components and biological processes including nervous system development, neuron projection, and axonogenesis (TABLE 4, FIG 2C). Consistently, upregulation of many neurodevelopmental-related and axonogenesis-related genes was previously observed in differentiating mouse embryonic stem cells (ESCs) and iPSCs as well as human ESCs (Wu et al. 2010;Chen et al. 2013). Interestingly, upregulated genes also showed enrichment of several GO terms related to the microtubule cytoskeleton (FIG 2C). Among the highly downregulated genes were several proliferation factors, such as SOX2, NOTCH1, and OTX2 (TABLE 3). Furthermore, the chromokinesin motor proteins KIF4a and KIF22, involved in cell proliferation by regulating spindle microtubule dynamics during mitosis, were downregulated (Almeida and Maiato 2018;Bisht, Tomschik, and Gatlin 2019). Conversely, genes of several neuronal and axonal kinesins are upregulated, including KIF1a, KIF5c, and KIF21b (TABLE 3) (Hirokawa and Tanaka 2015).
Additionally, the neuron-specific tubulin TUBB3 and the axonal microtubule associated protein MAPT are upregulated (TABLE 3). The dynamic shift in the transcriptome reflects a population-wide transformation from proliferating cells to terminal differentiating cells with intrinsic neuronal properties, and highlights changes in the microtubule cytoskeleton expression profiles at the onset of stage 3.

Proteomic profiling of developing human iPSC-derived neurons
In addition to transcriptome analysis, we assessed gene expression on the translational level by performing mass spectrometry based quantitative proteomics. We collected samples for proteome analysis on the same days as described above for RNA sequencing. We identified 7,512 proteins in two replicates and quantified 5,620 proteins across all three time points (TABLE 5,  GO enrichment analysis revealed enrichment of several terms related to neuronal differentiation and intracellular transport mechanisms, which reflects cell-autonomous remodeling of molecular processes (FIG2D, TABLE5). One of the upregulated proteins in cluster 1 is KLC1, a subunit of the microtubule motor protein, which was found to be required for neuronal differentiation from human embryonic stem cells (Killian et al. 2012). The AIS protein Trim46, which is known to regulate neuron polarity and axon specification by controlling microtubule organization during development, was also found in this cluster (van Beuningen et al. 2015). Furthermore, this cluster contains Camsap1, Camsap2 and Camsap3, proteins which localize to the minus ends of microtubules to stabilize them, thereby regulating neuronal polarity . Proteins in cluster 2 present a considerable increase in relative expression from day 3 to day 7, which coincides with the onset of axon formation and development (stage 3). Accordingly, enriched GO terms include proteins associated with neuronal development, axonogenesis, and other axon-related mechanisms (FIG2D).
Similarly, the GO terms neuronal development, axon, and synapse were also upregulated during differentiation of immortalized human neural progenitor cells (Song et al. 2019). Among the highly upregulated proteins in this cluster are several members of the Septin family: neuronal-specific Sept3, Sept5, and Sept6 (TABLE 5). Although mechanistic insights remain unclear, emerging evidence implicates Septins as potential factors for establishing neuronal polarity (Falk, Boubakar, and Castellani 2019). Septins interact with actin and microtubule networks and could affect neuronal polarity by regulating cytoskeleton dynamics (Spiliotis 2018;Falk, Boubakar, and Castellani 2019).
Sept6 specifically is suggested to play a role in axonal filopodia formation as well as in dendritic branching, and its increased expression coincides with axonal outgrowth (Cho et al. 2011;Hu et al. 2012). Moreover, examples of proteins with the highest relative expression in this cluster are DCX, Tau, Ncam1, Basp1, Snap91, and Syt1, which are generally considered to be neuronal differentiation and polarization markers (TABLE 5). These data confirm the neuronal identity of the human iPSCderived cells, and the presence of cellular machinery involved in axon development. Cluster 3 represents proteins with increased expression from day 1 to day 3, and minimal changes in expression from day 3 to day 7. This cluster comprises proteins enriched in GO terms that are associated with cell metabolism and (re)localization of intracellular and extracellular components, which correspond to substantial changes in the cellular proteome (FIG2D). Proteins in this cluster that show differential expression from day 1 to day 3 include Sox4 and Sox11, both members of the SoxC transcription factor family (TABLE 5). These factors are involved in neurogenesis and their expression induces subsequent expression of neuron-specific genes (Kavyanifar, Turan, and Lie 2018). Also represented in this cluster are Arpc2 and Arpc4, subunits of the Arp2/3 complex. The Arp2/3 complex mediates actin polymerization and is required for formation of lamellipodia and filopodia as well as axon guidance (Chou and Wang 2016). Clusters 4, 5, and 6 encompass proteins that are downregulated during the differentiation of NSCs into polarized neurons. GO analysis of these clusters reveals that they contain proteins involved in intracellular metabolism and homeostasis, genomic translation, the cell cycle, and biosynthesis of amino acids and peptides (FIG 2E). Downregulation of DNA replication and cell cycle-related proteins is also reported to coincide with terminal differentiation in neuroblastoma cells and with development of cultured rat neurons (Murillo et al. 2017;Frese et al. 2017). These results suggest that the overall proteome dynamics are indicative of cellular processes such as cell cycle exit and neuronal differentiation.

Comparison of transcriptomic and proteomic profiles of developing human iPSC-derived neurons
To compare the transcriptome and proteome dynamics, we performed correlative analysis of the relative RNA and protein expressions on day 7 compared to day 1 (TABLE 6). Based on their annotated gene names we were able to compare the expression dynamics of 7,021 factors. Of these, 4,536 followed the same trend for transcriptomic and proteomic expression changes, and overall, we found a significant correlation between the relative transcriptomic and proteomic expression profiles (FIG 2F). In agreement with the observed immunofluorescence, typical stem cell markers (NOTCH1, SOX2, MKI67, LIN28A, OTX2, and NES) showed a downregulation of both RNA and protein levels during neuronal differentiation. RNA levels as well as protein levels of typical neuron markers (DCX, ENO2, SYP, MAP2, STMN1, and TUBB3) and of axonal markers (TRIM46, MAPT, BASP1, ANK3, NCAM1, GAP43, and NFASC) displayed a marked increase during neuronal development (FIG 2G,H). Through quantitative analysis of transcriptomic and proteomic dynamics we were able to characterize human iPSC-derived neuronal differentiation and identify early neurodevelopmental processes in an unbiased manner. This quantitative map of neuronal transcriptome and proteome dynamics provides a rich resource for further analyses and may identify molecular mechanisms involved in neuronal polarity and axon specification.

Identification and characterization of intermediate stages during axon specification
Transcriptomic and proteomic profiling of developing neurons revealed that axonal components are upregulated after ~7 days, and assembled AIS structures were detected at proximal axons after ~14 days. Next, we studied the process of axon specification in human iPSC-derived neurons in more detail, and investigated the appearance of AIS proteins at different timepoints between day 5 and 11.
In stage 2 neurons, in which neurites are unpolarized and have similar lengths, Trim46 and AnkG proteins appeared as punctate structures in one or more neurites in a subset of neurons (FIG 3A,B).
Quantification of the relative abundance of stage 2 neurons lacking AIS proteins (referred to as stage 2a), or containing AIS proteins at one or more neurites (referred to as stage 2b), showed a developmental transition over time from stage 2a to stage 2b neurons (FIG 3A,B). Stage 3 neurons were morphologically defined by the appearance of a single elongated neurite, the future axon, which has grown at least twice as long as the other neurites. Interestingly, we found that AIS proteins in stage 3 neurons first appeared as noncontinuous structures consisting of multiple smaller puncta and stretches that cover distal parts of the axon (referred to as stage 3a), prior to their more conventional localization in the AIS at the proximal axon (referred to as stage 3b) (FIG 3A,B). Quantification of the abundance of these neurodevelopmental stages over time revealed a developmental decline of stage 2b neurons that was accompanied with an increase of stage 3a neurons, as well as an increase of stage 3b neurons with a relatively later onset (FIG 3C). We further characterized the distribution of AIS proteins by measuring their lengths and distances to the soma at each stage. Developmental changes in the length of AnkG and Trim46 structures were observed, as the total neurite length covered by Trim46 or AnkG signals was strongly increased by ~40% in stage 3a neurons, and significantly reduced by ~55% in stage 3b neurons (FIG 3D; FIG S3A). Movereover, the axonal Trim46 and AnkG structures were localized more distally in stage 3a neurons, as the distance from the soma to both the start as well as the end of the Trim46 and AnkG appearance was significantly larger compared to stage 3b neurons (FIG 3E; FIG S3B-D). The axonal localization of NaV during development shows a similar dynamic profile as Trim46 and AnkG (FIG S3D). These data imply that axon specification (transition stage 2-3) in human iPSC-derived neurons can be subdivided in 4 steps (stage 2a, 2b, 3a and 3b) based on the subcellular localization of AIS proteins. Here, AIS proteins first form relatively long, noncontinuous structures in the distal axon before accumulating at the proximal axon to form the AIS structure.

Action potential maturation coinciding with onset of axon development
We next investigated whether the different organization of AIS components in stage 3a and 3b neurons is accompanied by differences in electrical properties. Local clustering of voltage-gated ion channels at AIS structures, as observed in stage 3b neurons, are important to facilitate mature APs (Kole et al. 2008). Hence, we hypothesized that the noncontinuous appearance of NaV channels at No differences in AP amplitude and half-width were found between neurons of 7 and 10-11 days.
However, on day 7, neurons fired APs with smaller after-hyperpolarization (data not shown), possibly reflecting a developmental increase in potassium channels (Song et al. 2013). Other intrinsic properties, like resting membrane potential, input resistance, AP firing threshold and maximum sodium current, remained stable during this developmental period. Together, these results indicate a developmental maturation of specific AP properties, which coincides with the timing of the developmental transition from stage 3a to stage 3b neurons.

Mapping microtubule reorganization in the newly identified developmental stages
The transcriptomic and proteomic profiles point to the importance of microtubule cytoskeleton remodeling during axon specification. Indeed, AIS dynamics are reported to be regulated by the underlying axonal microtubule cytoskeleton (Stepanova et al. 2003;Kleele et al. 2014;van Beuningen et al. 2015;Yau et al. 2016). Therefore, we assessed changes in the microtubule network in axons and In developing axons, the shift towards unidirectional, plus-end out microtubules was more notably

DISCUSSION
To better understand neuronal differentiation and polarization in human cells, we performed an indepth characterization of human iPSC-derived neurons during these developmental processes. We systematically assessed the early stages of human neurodevelopment in culture, including axon specification (transition stage 2-3), and performed transcriptomic and proteomic profiling during these steps. We describe previously undescribed intermediate stages of axonal outgrowth (stage 2a, 2b, 3a and 3b), which is characterized by a distal to proximal reorganization of the axonal microtubule network and relocation of AIS proteins.

Development of polarized and functional human iPSC-derived neurons
In this study we showed that NSCs consistently gave rise to polarized and functional human neurons, which was demonstrated by the loss of cell proliferation and NSC markers, and the appearance of neuron-specific and AIS markers. These neurons form a single axon with a functional AIS, and exhibit AP firing. As expected, AP properties and cell intrinsic variables appeared immature compared to other studies performed at later developmental stages (Bardy et al. 2016;Gunhanlar et al. 2018). Passive physiological properties were comparable to neurons recorded from ex vivo fetal cortical brain tissue (Moore et al. 2009). We consistently observed neurons that fired a single AP and neurons that fired multiple APs at different developmental stages. The reason for this difference remains unknown, but may indicate variation in the maturation, cell morphology or other factors that could contribute to cellular heterogeneity of the culture. Non-human neurons develop at faster rates: for example, rat dissociated neurons reach stage 3 after approximately 1.5 days in culture, and cortical development in maturation in mammals ranging from mouse to primate is both faster and less complex than in humans (Dotti, Sullivan, and Banker 1988;Clowry, Molnar, and Rakic 2010;Molnar and Clowry 2012;Silbereis et al. 2016;Marchetto et al. 2019). We found that human iPSC-derived neurons transition to stage 3 in approximately 7 days, which is consistent with the described prolonged development of human neurons in vivo and in vitro (Grabrucker et al. 2009). The slow rate of development is also reflected in the lack of spontaneous AP firing or incoming synaptic responses as mature synapses have likely not formed yet. Co-culturing with astrocytes could benefit synaptogenesis, as this has been shown to promote synaptic connectivity . The prolonged development of human iPSC-derived neurons allows studying neurobiological processes, such as AP maturation and AIS assembly, with higher temporal resolution.

Quantitative profiles of transcriptomic and proteomic of early human neurodevelopment
Quantitative transcriptome analysis identified 549 upregulated genes and 614 downregulated genes during early stages of neurodevelopment. As expected, and in agreement with changes observed in other differentiating cell types, GO term enrichment analysis showed strong downregulation of genes related to cell proliferation during terminal differentiation Gao, Yourick, and Sprando 2014;Tripathi et al. 2014). Simultaneously, genes related to neurodevelopmental processes such as neurite formation and axonogenesis are upregulated, which is in line with changes previously observed in mouse iPSCs and human ESCs (Wu et al. 2010;Chen et al. 2013). To determine if regulatory changes of transcripts are reflected in protein expression, we performed quantitative proteome analysis. We identified 2,218 proteins with more than two-fold expression changes during the first stages of neurodevelopment, indicating that significant remodeling of the proteome takes place during early stages. We identified six clusters of expression profiles and conducted GO enrichment analysis, which revealed a coordinated proteomic rearrangement during neurodevelopment. Proteins related to neuronal differentiation are upregulated and proteins related to cell proliferation are downregulated. Similar changes are observed in dissociated rodent neurons, differentiating neural crest and immortalized human neural progenitor cells, confirming the neuronal identity adopted by our human iPSC-derived cells (Kobayashi et al. 2009;Frese et al. 2017;Song et al. 2019). Similar to changes in protein expression levels, RNA expression also showed a strong increase of factors involved in neuronal differentiation and polarization, such as TUBB3 and TRIM46. Indeed, RNA and protein dynamics were generally correlated, indicating a coordinated cellular reprogramming into neurons. Furthermore, reorganization of the microtubule cytoskeleton is reflected in both transcriptome and proteome profiles during early neurodevelopment. However, differences between transcriptome and proteome dynamics during early neurodevelopment are also observed. This may reflect a temporal shift in regulation of transcription and translation. In addition, possible regulatory mechanisms on the protein level might be at play. Together, these data provide a rich resource for both the transcriptome and proteome dynamics in developing human neurons, which can be used in future studies to advance our understanding of the molecular mechanisms involved in neuronal differentiation and polarity.

Distal to proximal relocation of AIS proteins during axon development
A critical event in early neurodevelopment is the polarization of a symmetric cell into a neuron, which is initiated by the formation of a single axon (Dotti, Sullivan, and Banker 1988;Craig and Banker 1994). Here, we studied the onset of polarity in human iPSC-derived neurons in detail, and found that axon specification proceeds through a multistage process with multiple intermediate steps. We observed that AIS proteins Trim46, AnkG and NaV first appear as long noncontinuous structures in distal regions of the axons at the onset of stage 3 (stage 3a), and later relocate and cluster at proximal regions to form the stable AIS structures (stage 3b). In conjunction, we also found a distal-to-proximal reorganization of the microtubule cytoskeleton network in axons. This is marked by a developmental shift towards the characteristic uniform, plus-end out orientation in growing axons. Although the precise relation between AIS formation and microtubule remodeling in early axon development is unclear, various studies have shown the cooperative interaction between AIS components and microtubule structures during axon development (Leterrier et al. 2011;Freal et al. 2016). Recently Fréal et al., described a feedback-based mechanism that drives AIS assembly, in which membrane, scaffolding, and microtubule(-associated) proteins, including AnkG and Trim46, cooperate to form a stable AIS-microtubule structure in the proximal axon (Freal et al. 2019). It has been shown that AnkG can act as a scaffold to recruit Trim46-positive microtubules and subsequently direct AIS protein trafficking to the proximal axon (Freal et al. 2019). The possible function of AIS proteins in the distal axon remains elusive. Similar to their function in stabilizing microtubules in the proximal axon following proper AIS formation, redistribution of AIS proteins in the distal axon may help to locally stabilize microtubules and drive axon outgrowth. AnkG in distal axons may provide additional support to the formation of Trim46-positive parallel microtubules during outgrowth. Then, as Trim46 moves proximally, the axonal microtubule network close to the cell body is remodeled to the characteristic uniform plus-end out orientation. This idea is consistent with the observed shift of the unidirectional parallel microtubule organization in the distal parts of the axon first, followed by proximal reorganization. Finally, AnkG and Trim46 together may drive AIS assembly at the proximal axon, as previously described in dissociated rodent neurons (Freal et al. 2019). It remains unknown if the observed intermediate step of distal AIS protein accumulation is unique to human neurons. Axons in humans grow significantly longer compared to rodents, thus additional regulatory mechanisms enhancing axon outgrowth might be at play in humans. Alternatively, it is possible that these changes have not been observed in rodent neurons because of their relatively faster development. We also observed differences in microtubule dynamics, as MT+TIP growth speeds are higher than those found in rodent neurons in vitro and in vivo, hinting at species-specific regulation of microtubule dynamics in human neurons (Stepanova et al. 2003;Kleele et al. 2014;Yau et al. 2016). Future studies are required to examine the potential human-specific attributes of neuronal polarity and axon outgrowth.
In summary, our quantitative map of neuronal transcriptome and proteome dynamics provides a rich resource for future analysis of early neurodevelopmental processes in human iPSC-derived neurons.
We investigated early development in human neurons and uncovered an intermediate axon developmental step, thereby illustrating the potential of this model system to study neurobiological processes in human cells in stage 3 neurons. This study also provides a framework and excellent starting point for further studies that aim to complement our understanding of neuronal polarization in human cells.

DECLARATION OF INTERESTS
Casper Hoogenraad is an employee of Genentech, Inc., a member of the Roche group. The authors declare that they have no additional conflict of interest. and plated on 12 mm (~100k per well) or 18 mm (~200k per well) pre-coated glass coverslips in respectively a 24-wells or 12-wells plate at 37°C with 5% CO2. Coating of coverslips was performed directly before plating: coverslips were first incubated with ReadySet (45 min, 37°C; ax0041+, Axol Bioscience), washed four times with sterilized water, and incubated with freshly-thawed 1x SureBond (1h, 37°C; ax0041+, Axol Bioscience) in PBS. After 24 hours (day 1), the medium was fully replaced by Neuronal Maintenance-XF Medium (ax0032, Axol Bioscience), and after another 24 hours (day 2) by Neuronal Differentiation-XF Medium (ax0034, Axol Bioscience). Three days later (day 5), half of the medium was replaced by Differentiation-XF Medium. Next day (day 6), half of the medium was replaced by Neuronal Maintenance-XF Medium, again one day later (day 7), and every three days during further maintenance. Cells were kept in culture for maximum ~15 days to ensure high quality of the cultures.

Lentivirus and lentiviral infection
The constructs used for lentiviral transduction are FUGW-GFP (Addgene #14883, (Lois et al. 2002)) and Marcks-tagRFP-T-pIres-GCN4-MacF18. Marcks-tagRFP-T-pIRES-GCN4-MacF18 cloning is previously described (Yau et al. 2016). The construct was subcloned into the pSIN-TRE-mSEAP-hPGK-rtTA2sM2 lentiviral vector (kindly provided by Dr. Didier Trono, Ecole Polytechnique Fédérale de Lausanne, Switzerland) wherein the neuron-specific synapsin promotor has substituted the PGK promotor. Generation of lentiviral particles was performed as previously described . Lentiviral transduction of cells was performed two hours after plating. The tetracyclinedependent expression was induced by adding500 ng/ml doxycycline to the medium two days before imaging. Immunofluorescence Cells were fixed for 5-10 min in PBS with 4% paraformaldehyde/4% sucrose at room temperature, and washed three times with PBS. For immunofluorescence stainings, fixed cells were sequentially incubated with primary and secondary antibodies dissolved in gelate dilution buffer (GDB; 0.2% BSA, 0.8 M NaCl, 0.5% Triton X-100, 30 mM phosphate buffer, pH 7.4), and washed three times with PBS after every antibody incubation step. Coverslips were mounted on glass slides using Vectashield mounting medium (Vector laboratories) with or without DAPI.

Microscopy
Fixed cells were imaged using a LSM700 confocal laser-scanning microscope (Zeiss) with a Plan-Apochromat 63x NA 1.4 oil DIC; EC Plan-Neofluar 40x, NA 1.3 Oil DIC; and a Plan-Apochromat 20x, NA 0.8 objective. Live-cell acquisition was performed using spinning-disk confocal microscopy on an inverted research microscope Nikon Eclipse Ti-E, equipped with a perfect focus system (Nikon) and a spinning disk-based confocal scanner unit (CSU-X1-A1, Yokogawa). The system was also equipped with an ASI motorized stage with the piezo plate MS-2000-XYZ (ASI), Photometric Evolve Delta 512 EMCCD camera (Photometric) controlled by the MetaMorph 7.8 software (Molecular Devices), or Photometric PRIME BSI sCMOS camera (version USB 3) as upgrade of EMCCD and controlled by the MetaMorph 7.10 software (Molecular Devices). The system was equipped with Plan Apo VC 60x NA 1.4 oil-immersion objective (Nikon) and S Fluor 100x NA 0.5-1.3 oil-immersion objective (Nikon) for photoablation experiments. A 491 nm 100 mW Calypso (Cobolt) and a 561 nm 100 mW Jive (Cobolt) laser were used as the light sources. We used an ET-GFP filter set (49002, Chroma) for imaging of proteins tagged with GFP and an ET-mCherry filter set (49008, Chroma) for imaging of proteins tagged with tag-RFP. For the photoablation experiments we used an ILas system (Roper Scientific France/ PICT-IBiSA, Institut Curie, currently Gataca Systems) mounted on a Nikon Eclipse microscope described above. A 355 nm passively Q-switched pulsed laser (Teem Photonics) was used for the photoablation together with a S Fluor 100x 0.5-1.3 NA oil objective (Nikon).

Image quantification and analysis
Quantifying neuronal differentiation and polarization: To measure neuronal differentiation and polarization over time, cells were identified using DAPI staining and scored to be positive or negative for the indicated NSC, neuron differentiation and axon markers. Sample preparation RNA sequencing ~100,000 hiPSC-derived NPCs were plated per well for bulk RNA sequencing samples. Prior to sample preparation, all equipment and surfaces were cleaned with RNaseZap (Sigma-Aldrich).
Replicates of hiPSC-derived neurons were harvested at three different timepoints of differentiation (days 1, 3, and 7) with 200 µl Trizol (Invitrogen) per sample and stored at -80 °C until sequencing.
RNA extraction, cDNA library preparation (CEL-Seq2 protocol), quality control for aRNA and cDNA, and sequencing on a NextSeq500 High output 1x75 bp paired end run with 2% sequencing depth were performed by Single Cell Discoveries (Utrecht, The Netherlands).

Bioinformatic analysis RNA sequencing
Mapping to reference transcriptome Hg19 was performed by Single Cell Discoveries (Utrecht, The Netherlands). The following investigations were done in R statistical software (R Core Team, 2019) with the use of packages ggplot2 (Wickham, 2009) and pheatmap (Kolde, 2019). The raw read counts were normalized to reads per million for each gene. Genes observed in less than 5 samples were excluded from further analysis. To determine differential expression a linear regression ANOVA model was used where gene expression is explained by timepoint + biological replicate and technical replicate (nested in the biological replicate). To obtain the differences (in both p-value and effect between the timepoints) a Tukey-test was performed for each gene using the same model. We corrected for multiple-testing by pairwise comparison using the p.adjust() function in R with the 'BH' setting (R Core Team, 2020). Genes with an adjusted p-value < 0.05 were used for further investigation. GO enrichment was done using the hypergeometric test phyper() in R (R Core Team, 2020), the set of GO terms was obtained from Ensembl Biomart for Human genes version GRCh38.p13. Comparison between transcriptomics and proteomics was done by selecting genes present in both datasets based on their public name.

Sample preparation for mass spectrometry (TMT labeling)
Replicates of hiPSC-derived neurons were harvested with lysis buffer (8 M Urea, 50 mM ammonium bicarbonate (Sigma), EDTA-free protease inhibitor Cocktail (Roche)) at three distinct differential time points (days 1, 3, and 7). Lysates were sonicated on ice with a Bioruptor (Diagenode) and cleared by centrifugation at 2500 g for 10 min. The protein concentration of the samples was determined by Bradford assay. Per sample 100 µg of proteins were reduced (5 mM DTT, 55˚C, 1 hour), alkylated (10 mM Iodoacetamide, 30 min in the dark) and sequentially digested by LysC (Protein-enzyme ratio 1:50, 37˚C, 4 h) and trypsin (Protein-enzyme ratio 1:50, 37˚C, overnight). After digestion (overnight), formic acid (final concentration 3%) was used to acidify the samples and resulting peptides were afterwards desalted with Sep-Pak C18 columns (Waters). Samples were labeled with stable isotope TMT-6plex labeling, according to manufacturer's instruction (Thermo Fisher Scientific). In short, peptides were resuspended in 80 µl of 50 mM HEPES buffer, 12.5% ACN (pH 8.5), while TMT reagents were dissolved in 50 µl anhydrous ACN. We added 25 µl of each dissolved TMT reagent to a correspondent sample according to the following scheme: Following incubation (room temperature) for 1 hour, the reaction was quenched with 5% hydroxylamine. Differentially TMT-labeled peptides were mixed in equal ratios and dried in a vacuum concentrator.

Peptide fractionation
Mixed TMT-labeled peptides were solubilized in 10mM ammonium hydroxide, pH 10.0 and subsequently fractionated using basic pH reverse phase HPLC. Peptides were loaded on a Gemini 3µm C18 110A 100 x 1.0 mm column (Phenomenex) using an Agilent 1100 pump equipped with a degasser and a photodiode array (PDA) detector. Peptides were concentrated on the column at 100µl/min using 100% buffer A (10mM ammonium hydroxide, pH 10) after which the fractionation gradient was applied as follow: 5% solvent B (10mM ammonium hydroxide in 90% ACN, pH 10) to 30% B in 53 mins, 70% B in 7 min and increased to 100% B in 3 min at a flow rate of 100µl/min. In total 60 fractions of 1 min were collected using an Agilent 1260 infinity fraction collector and then concatenated into 12 final fractions. Collected fractions were vacuum-dried, reconstituted in 5% formic acid/5% DMSO and stored at -80°C prior to mass spectrometry analysis.

Mass spectrometry analysis
We analyzed the samples on an Orbitrap Q-Exactive HF mass spectrometer (Thermo Fisher Scientific) coupled online to an Agilent UPLC 1290 system (Agilent Technologies). Peptides were loaded onto a trap column (Reprosil C18, 3 µm, 2 cm × 100 µm; Dr. Maisch) and separated on an analytical column (Poroshell 120 EC-C18, 2.7µm, 50cm x 75µm; Agilent Technologies). Peptides were trapped for 10 min at 5 µl/min in solvent A (0.1M acetic acid in H20) and then separated at a flow rate of approximately 300 nl/min (split flow from 0.2ml/min) by applying a 120 min linear gradient as follows: 13% up to 40% solvent B (0.1M acetic acid in 80% ACN) in 95 min, 40-100% in 3 min and finally 100% for 1 min. The mass spectrometer was operated in data-dependent acquisition mode. Full MS spectra from m/z 375-1600 were acquired at a resolution of 60.000 with an automatic gain control (AGC) target value of 3e6 and maximum injection time (IT) of 20 ms. The 15 most intense precursor ions were selected for HCD fragmentation. HCD fragmentation was performed at a normalized collision energy (NCE) of 32%. MS/MS spectra were obtained at a 30.000 resolution with an AGC target of 1e5 and maximum injection time (IT) of 50 ms. Isolation window was set at 1.0 m/z and dynamic exclusion to 16.0s.

Data processing proteomics
Raw MS files were processed for data analysis with Proteome Discoverer 1.4 (Thermo Fisher Scientific). A database search was performed using the Swissprot homo sapiens database and Mascot (version 2.5.1, Matrix Science, UK) as search engine. Carbamidomethylation of cysteines was set as a fixed modification, and oxidation of methionine, acetylation at the N-termini, TMT-6plex of lysine residues and TMT-6plex at the peptide N-termini were set as variable modifications. Trypsin was set as cleavage specificity, allowing a maximum of two missed cleavages. Data filtering was performed using percolator, resulting in 1% false discovery rate (FDR). Additional filters were search engine rank 1 and mascot ion score > 20. Only unique peptides were included for quantification and the obtained TMT ratios were normalized to the median. Common contaminant proteins (such as keratins and albumin) were removed from the list.

Bioinformatic analysis proteomics
All mass spectrometry data were analyzed using R statistical software (R Core Team, 2017). To infer protein dynamics upon differentiation, TMT reporter intensity values of hiPSC neurons at time point day 3 and day 7 were normalized to their correspondent day 1 or alternatively day 7 was normalized to day 3. TMT generated ratios (previously normalized to the median) were then log2-transformed. A log2-transformed mean of the TMT-ratios of the individual replicates was calculated. Proteins with less than 3 peptides used for TMT quantification or with a reporter ion variability > 100% in at least one TMT-ratio (high reporter ions variability) or with a median log2 fold-change > 0.4 between the replicates in at least one TMT-ratio (high replicate variability) were excluded from the analyses. Good correlation of replicates was assessed by comparing TMT ratios of all quantified proteins at different time points using Pearson correlation. Proteins with an absolute log2 fold-change > 0.3 between day 3 and day 1 or between day 7 and day 3 or between day 7 and day 1 were considered significantly regulated. Only significant regulated proteins were subjected to cluster analysis by using K-means Heatmaps in the figures were prepared applying hierarchical clustering using Euclidean distance.

Statistical Analysis
Data processing and statistical analyses were performed using Prism GraphPad (version 8.0) software. All statistical tests are described in the corresponding figure legends. Differences were considered significant when P < 0.05, and P-values are represented as: * P < 0.05, ** P < 0.01, and *** P < 0.001.

Supplementary movie 1
Representative movie of MT+TIP comet dynamics in a dendrite (day 7). Scale bar: 5 µm. Time in minutes:seconds.

Supplementary movie 2
Representative movie of MT+TIP comet dynamics following laser severing in a dendrite (day 7).