Human embryonic kidney (HEK) 293T cells [American Type Culture Collection (ATCC)] were grown in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) containing 10% heat-inactivated fetal bovine serum (FBS, Gibco) and 1% penicillin/streptomycin (Gibco). Primary skin fibroblasts from apparently healthy human individuals were obtained from the cell repository at the Coriell Institute for Medical Research (Table 1). They were cultured in DMEM supplemented with 15% FBS and 1% penicillin/streptomycin. Two different cell culture media were sequentially employed in reprogramming human neurons. (1) Neuronal induction medium consists of DMEM: F12: neurobasal (1:1:1), 0.8% N2 (Invitrogen), 0.8% B27 (Invitrogen), 1% penicillin/streptomycin, and supplemented with 10 μM forskolin (FSK; Sigma-Adrich, St. Louis, MO, USA), 1 μM dorsomorphin (DM; Millipore, Kankakee, IL, USA), and 10 ng/ml basic fibroblast growth factor (bFGF; PeproTech, Rocky Hill, NJ, USA). (2) Neuronal maturation medium consists of DMEM: F12: neurobasal (1:1:1), 0.8% N2, 0.8% B27, 1% penicillin/streptomycin, and supplemented with 5 μM FSK and 10 ng/ml each of brain-derived neurotrophic factor (BDNF; PeproTech, Rocky Hill, NJ, USA), glial cell-derived neurotrophic factor (GDNF; PeproTech, Rocky Hill, NJ, USA), and neurotrophin-3 (NT3; PeproTech, Rocky Hill, NJ, USA).

As previously descried (Liu et al., 2016), a third-generation lentiviral vector (pCSC-SP-PW-IRES-GFP) was used to express NEUROG2-IRES-GFP-T2A-Sox11 and ISL1-T2A-LHX3. The dual reporter 2Gi2R was generously provided by Dr. Fred H. Gage (Salk Institute for Biological Studies; Mertens et al., 2015). Replication-incompetent lentiviruses were produced in HEK293T cells, and viral supernatants were collected at 48 and 72 h post-transfection. Viral supernatants were filtered through 0.45-μm syringe filters and titered as previously described (Ding and Kilpatrick, 2013). Lentiviruses were stored at 4°C prior to cell transduction.

Lentiviral delivery of four factors (NGN2, Sox11, ISL1, and LHX3) was used as previously described (Liu et al., 2016; Tang et al., 2017). In brief, fibroblasts were plated at a density of 1 × 10⁴ cells/cm² onto Matrigel-coated dishes. Cells were transduced the next day at an approximate multiplicity of infection (MOI) of 10 with lentiviral supernatants supplemented with 6 μg/ml polybrene (Millipore, Kankakee, IL, USA). Fibroblast medium was refreshed after overnight incubation. One day later, the culture was replaced with neuronal induction medium and half-changed every other day until 14 days post infection (dpi). Induced neurons were then partially enriched with a replating procedure (Liu et al., 2013) and cocultured with primary astrocytes on Matrigel-coated coverslips. The medium was half-changed twice a week until analysis. The neuronal conversion efficiency was calculated with the percentage of GFP (+) cells expressing TUBB3 (TUBB3/GFP), and the MN fraction was calculated with the percentage of TUBB3 (+) cells expressing HB9 (HB9/TUBB3).

Primary cortical neurons (PCNs) and cerebellar granule neurons (CGNs) were isolated from CD1 embryonic day-18 mice (E18) and 6-day-old pups (P6) of either sex, respectively, as previously described (Ding et al., 2013, 2016, 2018). All animal work was carried out with permission from the Institutional Animal Care and Use Committee (IACUC) of the University of Texas Southwestern Medical Center. Individual culture experiments were performed using cells prepared from the same litter. Cells were plated at a density of 5 × 10⁴ cells/cm² onto coverslips (Thermo Fisher Scientific, Waltham, MA, USA) in cell culture plates coated with poly-D-lysine/laminin (Invitrogen) in Neurobasal medium (Invitrogen) containing B-27 serum-free supplement (50×; Invitrogen). The medium was half-changed twice a week until analysis.

Cells were cultured to the desired time points. They were fixed with 4% paraformaldehyde (PFA; Sigma-Adrich, St. Louis, MO, USA) in phosphate buffered saline (PBS) for 15 min at room temperature (RT) and permeabilized with blocking buffer [PBS containing 3% bovine serum albumin (BSA) and 0.2% Triton X-100] for 1 h. Cells were then incubated with primary antibodies in blocking buffer overnight at 4°C, followed by washing and incubation with fluorophore-conjugated corresponding secondary antibodies. Primary antibodies used in this study are goat anti-CHAT (Millipore, Kankakee, IL, USA; 1:200), mouse anti-HB9 (DSHB; 1:500), chicken anti-MAP2 (Abcam; 1:10,000), mouse or rabbit anti-TUBB3 (Covance; 1:2,000). Nuclei were stained with Hoechst 33342 (HST; Thermo Fisher Scientific, Waltham, MA, USA).

FISH was performed as described previously (Li et al., 2016) with modifications. Briefly, cultured neurons were washed once with DPBS (Gibco) and fixed with 4% PFA in PBS for 30 min at RT. Cells were further permeabilized with 0.2% Triton X-100 (Sigma-Adrich, St. Louis, MO, USA) for 10 min at RT and twice washed with ice-cold PBS. For RNase A-treated controls, samples were incubated with 0.5 mg/ml of RNase A (New England BioLabs) in 10 mM Tris-HCl buffer (pH 7.4) for 1 h at 37°C. The samples were equilibrated with hybridization buffer composed of 2× saline sodium citrate (SSC), 10% dextran sulfate (average molecular weight 500,000; Sigma-Adrich, St. Louis, MO, USA), 10 mM ribonucleoside–vanadyl complex (RVC; New England Biolabs), and 20% formamide (Thermo Fisher Scientific, Waltham, MA, USA). A mixture of DIG-labeled oligo-dT or dA probes (0.2 ng/μl) containing yeast tRNA (0.2 μg/μl) were heated to 95°C for 5 min and chilled on ice immediately. Probes were then combined with equal volumes of 2× hybridization buffer and added to samples for overnight incubation at 37°C. Anti-DIG antibody (sheep anti-DIG; Sigma-Adrich, St. Louis, MO, USA) and fluorophore-conjugated corresponding secondary antibodies (Thermo Fisher Scientific, Waltham, MA, USA) were used to detect oligo-dT signals. Anti-MAP2 antibody (Abcam) and HST were used to determine neuronal soma and nucleus, respectively. All reagents and solutions were prepared in nuclease-free water.

Protein nuclear transport was analyzed by using a dual reporter 2Gi2R (Mertens et al., 2015). Cultured cells were transduced with a lentivirus, which expressed a fused double-GFP with an NES (2GFP-NES) and a double RFP containing an NLS (2RFP-NLS). To validate this reporter system, transduced HEK cells were treated with 50 nM leptomycin B (Alfa Aesar) and then analyzed by confocal microscopy during a time course. For directly reprogrammed neurons, the reporter lentivirus was cotransduced with reprogramming factors, and the signal densities and distributions of GFP and RFP were analyzed at the indicated time points.

Confocal images were obtained with a Zeiss-LSM700 or Leica SP5 confocal microscopes. The ImageJ software (NIH) was used to quantify fluorescence intensity from images of confocal slices. For each neuron, as large an area as possible was measured within the nucleus or cytoplasm. An unbiased approach for data collection was employed. The person who analyzed the images was completely blinded to the sample information. The mean values were used to quantify an average signal density. For FISH assays, the average of oligo-dT signals in soma (both nucleus and cytoplasm) was used to evaluate the overall mRNA expression levels. mRNA subcellular distribution was measured by the ratio of nuclear to cytoplasmic oligo-dT signals (dT_nuc/dT_cyt). For the dual reporter assay, GFP and RFP signals were separately measured in the nucleus and cytoplasm. The ratio of nuclear to cytoplasmic GFP (GFPnuc/GFPcyt) was used to evaluate protein export, while the ratio of cytoplasmic to nuclear RFP (RFPcyt/RFPnuc) was used to measure protein nuclear import.

An unpaired two-tailed Student’s t-test was used for experiments consisting of two groups. For experiments consisting of more than two groups, one-way analysis of variance (ANOVA) was used with either Tukey (when comparing samples with each other) or Dunnett (when comparing samples to a control) post hoc tests. The results are expressed as mean ± SD of three biological replicates, and P < 0.05 was treated as significant.

Disruption of NCT is particularly involved in movement disorders (Da Cruz and Cleveland, 2016; Li and Lagier-Tourenne, 2018), such as ALS which is also characterized with progressive loss of MNs (Freibaum et al., 2015; Jovicic et al., 2015; Zhang et al., 2015; Boeynaems et al., 2016; Chou et al., 2018; Burk and Pasterkamp, 2019). To understand how defective NCT contributes to diseases, measuring NCT activity directly in human MNs became extremely important. In this study, we generated human MNs from adult fibroblasts through direct reprogramming (Liu et al., 2016). Transduced fibroblasts can be identified with co-expressed GFP, and they gradually became neuron-like and grow complex neurites over time (Figure 1A). Immunocytochemistry confirmed their expression of the neuronal marker tubulin beta 3 class III (TUBB3) and early MN marker HB9 (Figure 1B; Arber et al., 1999). At later stages, such as 6–8 weeks post viral infection (wpi), reprogrammed neurons expressed high levels of choline acetyltransferase (CHAT), the mature MN marker (Figure 1C). Thus, adult fibroblasts could be efficiently and directly converted into MNs, which were hereafter referred as directly induced MNs (diMNs). We examined skin fibroblasts from six apparently healthy donors, three males and three females (Table 1). No significant difference was observed between male and female donors at the neuronal conversion efficiency and the MN fraction (Figures 1D,E). About 80% transduced GFP+ cells expressed TUBB3 (Figure 1D). Among these TUBB3+ neurons, approximately 75% of them highly expressed nuclear HB9 (Figures 1B,E).

Based on this study and previous reports (Mertens et al., 2015; Liu et al., 2016; Smith et al., 2016), we summarized the major features associated with the reprogramming process in Table 2. During the first week, the transduced cells highly express the reprogramming factors, which remodel the chromatin structure and induce global changes on gene expression (Smith et al., 2016). The cells rapidly change morphology with outgrowing neurites and high expression of markers for general neurons at 2–4 wpi (Mertens et al., 2015; Liu et al., 2016). By 6–8 wpi, the reprogrammed neurons become functionally mature, characterized by synaptogenesis, firing of repetitive action potentials, and formation of neuromuscular junctions (NMJs; Liu et al., 2016).

The distribution of total poly(A) RNAs in single diMNs was measured by FISH with oligo-dT probes as previously described (Packard et al., 2015; Li et al., 2016). Following hybridization, immunostaining of microtubule-associated protein 2 (MAP2) was used to identify the soma, and the DNA dye HST was used to define the nucleus. At the late development stage of 6 wpi diMNs, strong oligo-dT signals could be detected in both the nucleus and cytoplasm, but the majority of signals localize in the cytoplasm (Figure 2A). No obvious signals were detected with the oligo-dA probe (Figure 2B) or in diMNs that were treated with RNase A (Figure 2C), suggesting that the oligo-dT signals in FISH assays are specific.

To examine whether neuronal maturation affects mRNA subcellular distribution, we conducted a time-course analysis of diMNs by FISH. At 3 wpi, oligo-dT signals were highly enriched in the nucleus (Figure 3A). Although diMNs at this stage outgrow neurites and express markers for general neurons (MAP2 and TUBB3) and MNs (HB9; Figures 1, Figure 3), their somas are still under transformation from an oval morphology of fibroblasts to the multipolar morphology of typical MNs. With diMN maturation, subcellular distribution of mRNAs gradually changed, with decreased signals in the nucleus and conversely increased signals in the cytoplasm. By 7 wpi, the signal density was obviously higher in the cytoplasm than that in the nucleus (Figure 3A). To quantify the subcellular distribution of poly (A) RNAs, the average signal density was separately measured in the nucleus and cytoplasm. The ratio of nuclear to cytoplasmic oligo-dT signal density (Nuc/Cyt) was significantly decreased within the first 6 wpi but reached a steady state thereafter (Figure 3B). These findings indicate that mRNA subcellular distribution is significantly affected by neuronal maturation stages and that it can be steadily measured at 6 wpi or later. A thorough analysis, however, failed to show any effect of sample sexes on mRNA subcellular localization in diMNs (Figure 3B).

To determine whether the influence of neuronal maturation on mRNA subcellular distribution is a common feature or unique to diMNs, we examined primary mouse neurons at different maturation stages. Primary mouse cortical neurons (PCNs) were isolated from E18 embryos. After culturing for 2 days in vitro (DIV), they showed strong oligo-dT signals in both the nucleus and the cytoplasm (Figure 4A). The specificity of FISH assays was verified by using the oligo-dA probe that showed no any signals (Figure 4B). The overall oligo-dT signals and the ratio of nuclear to cytoplasmic signal density (Nuc/Cyt) significantly decreased when examined at 14 DIV (Figures 4C–E). We also analyzed mRNA distribution in primary mouse CGNs, which were isolated from P6 pups (Figures 5A–C). These postnatal neurons mature faster than those prenatal PCNs in culture, and they are considered as mature with robust marker expression at 6 DIV (Ding et al., 2013; Leto et al., 2016). Consistent with the results in PCNs, total mRNA levels and the ratio of nuclear to cytoplasmic signal density (Nuc/Cyt) gradually decreased with the CGN maturation process (Figures 5D,E). These results clearly indicate that the influence of neuronal maturation on mRNA subcellular distribution is not unique to diMNs but also occurs in primary mouse neurons.

We employed a dual reporter system (2Gi2R) to measure protein NCT activity in single diMNs. This reporter system consists of a fused double-GFP with an NES (2GFP-NES) and a double RFP containing an NLS (2RFP-NLS; Mertens et al., 2015). NES and NLS are conserved signal sequences that are recognized by nuclear transport receptors of exportins and importins, respectively (Christie et al., 2016; Cagatay and Chook, 2018). Thus, both protein export and import can be analyzed with this dual reporter in single cells. In cells with normal NCT activity, GFP and RFP are respectively localized in the cytoplasm and nucleus, whereas such distribution will be disrupted when NCT activity is impaired (Figure 6A). An increased ratio of nuclear to cytoplasmic GFP (GFP_nuc/GFP_cyt) represents disrupted protein export, while a higher ratio of cytoplasmic to nuclear RFP (RFP_cyt/RFP_nuc) indicates a compromised protein nuclear import.

We first validated the 2Gi2R reporter in HEK cells with the treatment of leptomycin B, a nuclear export inhibitor (Jang et al., 2003). Both GFP and RFP were exclusively localized in the cytoplasm and nucleus of transduced cells, respectively (Figure 6B). In the presence of leptomycin B, GFP gradually accumulated inside the nucleus. A significant increase of the ratio of nuclear to cytoplasmic GFP (GFP_nuc/GFP_cyt) was observed at 10 min and later following exposure to leptomycin B (Figures 6B,C). On the other hand, the distribution of RFP was not significantly affected (Figures 6B,C). These results agree well with previous reports that leptomycin B is a highly specific inhibitor of nuclear export receptor CRM1 (Jang et al., 2003; Mutka et al., 2009), suggesting that this reporter system is reliable in protein NCT measurement.

Human fibroblasts from healthy donors were then transduced with lentiviruses expressing the 2Gi2R reporter and the reprogramming factors. In the resulting diMNs, both GFP and RFP were mainly localized in the cytoplasm and nucleus, respectively (Figure 6D). We also examined but failed to observe an effect of neuronal maturation on protein NCT activity, indicated by unaltered subcellular distribution of the GFP and RFP reporters at 4, 6, or 8 wpi (Figures 6E,F). Similarly, protein NCT activity is not affected by sample sexes during this time-course analysis (Figures 6E,F).