Transcriptional Programming of Human Mechanosensory Neuron Subtypes from Pluripotent Stem Cells

Abstract

Efficient and homogeneous in vitro generation of peripheral sensory neurons may provide a framework for novel drug screening platforms and disease models of touch and pain. We discover that, by overexpressing NGN2 and BRN3A, human pluripotent stem cells can be transcriptionally programmed to differentiate into a surprisingly uniform culture of cold- and mechano-sensing neurons. Although such a neuronal subtype is not found in mice, we identify molecular evidence for its existence in human sensory ganglia. Combining NGN2 and BRN3A programming with neural crest patterning, we produce two additional populations of sensory neurons, including a specialized touch receptor neuron subtype. Finally, we apply this system to model a rare inherited sensory disorder of touch and proprioception caused by inactivating mutations in PIEZO2. Together, these findings establish an approach to specify distinct sensory neuron subtypes in vitro, underscoring the utility of stem cell technology to capture human-specific features of physiology and disease.

Keywords: DRG; PIEZO2; TRPM8; differentiation; human; iPSC; mechanosensation; neural crest; neuron; sensory.

Conflict of interest statement

Declaration of Interests The authors declare no competing interests.

Published by Elsevier Inc.

Figures

Figure 1.. NGN2-BRN3A Programming of Human iPSCs Efficiently Yields Induced Sensory Neurons

(A) Protocol for sensory neuron induction using NGN2-BRN3A-engineered iPSCs. Phase-contrast images were captured on days 2, 4, 8, and 14 after doxycycline addition. Neurons induced with NGN2 alone are shown on the bottom for comparison. (B) Phase-contrast and immunocytochemistry images of day 21 neurons induced by NGN2-BRN3A and NGN2 only. (C) Immunocytochemistry of day 21 NGN2-BRN3A neurons to detect proteins found in sensory neurons. (D) Quantification of percent staining NeuN (78.9% ± 2.6%), BRN3A/NeuN (82.0% ± 1.7%), and ISL1/NeuN (90.1% ± 1.0%). For NeuN stains, n = 6 independent coverslips were used and were split for co-staining into n = 3 coverslips for BRN3A and n = 3 for ISL1. At least 200 cells were counted per stain. Values are expressed as mean ± SEM. Scale bars, 100 μm. Δ, medium change; 1/2Δ, half medium change; NTF, neurotrophic factor; Y-27632, ROCK inhibitor. See also Figure S1 and Videos S1 and S2.

Figure 2.. iSNs and Human DRG Neurons Co-express TRPM8 and PIEZO2

(A) Heatmap of RNA sequencing results on iPSCs, iSNs, and adult human donor DRG (hDRG). Each column represents an individual sample’s normalized log2 transcripts per million (TPM). The hDRG is a pooled RNA sample from 21 individuals. (B) RNA in situ hybridization on iSNs for transcripts of neurotrophin and sensory receptors. Each small fluorescent punctum roughly indicates one transcript. TUBB3 encodes a general neuron class of microtubule, βIII-tubulin. Scale bars, 10 μm. (C) Quantification of in situ hybridization in iSNs using a cutoff criteria of ≥ 5 puncta per transcript for positive cells. TRPM8+ (17.9% ± 1.5%), PIEZO2+ (14.0% ± 4.6%), TRPM8+/PIEZO2+ (56.0% ± 6.3%), and TUBB3+ only (11.5% ± 3.4%). 199 cells were counted across n = 5 coverslips. (D) RNA in situ hybridization for TRPM8 and PIEZO2 in human DRG (hDRG). Labeled neurons are shown in wide-field view (left) with dotted, dashed, and solid circles indicating TRPM8+, PIEZO2+, and TRPM8+/PIEZO2+ neurons, respectively. Zoomed images of single neurons (right), individually marked by numbers in the wide-field view. Scale bars, 25 μm. (E) Quantification of human DRG in situ hybridization. TRPM8+ (12.8% ± 4.4%), PIEZO2+ (59.7% ± 6.8%), and TRPM8+/PIEZO2+ (27.4% ± 6.9%). 622 positive cells were counted across n = 4 DRG donors. All values are expressed as mean ± SEM. See also Figure S2 and Table S1.

Figure 3.. iSNs Detect Cold Temperature and Mechanical Force

(A) Fluo-4 fluorescence images recording iSNs at 25°C and 4°C. The TRPM8 inhibitor RQ-00203078 was incubated at 10 μM for 5 min before resuming recording. A 5-min room temperature incubation was used in between each 4°C treatment. Scale bar, 500 μm. (B) Representative fluorescence traces of calcium imaging trials, with 5 cells shown per trial. Values are graphed as the change in fluorescence intensity from baseline and divided by the baseline (ΔF/F0). 100 μM AITC, 10 μM capsaicin, 500 μM menthol, and 50 μM α,β-methylene-ATP (α,β-meATP) were used. (C) Scatter dot plots of peak ΔF/F0 of individual cells after application of temperature or chemicals. Each cell is normalized to a percentage of its own KCl response at the end of the recording period. For each condition, ≥ 80 cells were analyzed. Post hoc comparisons of each treatment to vehicle with one-way ANOVA and Dunnett’s correction for multiple comparisons. p values: 50°C, p = 0.99; 4°C pre, p = 0.0001; 4°C RQ, p = 0.5135; 4°C post, p = 0.001; AITC, p = 0.1379; capsaicin, p = 0.5242; menthol, p = 0.0001; α,β-meATP, p = 0.0001. ****p < 0.0001. (D) Example whole-cell current-clamp recording of iSNs in response to depolarizing currents. A total of n = 4 cells were recorded without tetrodotoxin (TTX) andn = 5 cells with 1 μM TTX. (E) Mechanical stimulation of iSNs in whole-cell voltage-clamp mode. In the phase-contrast image, the recording pipette is on the left, and the stimulator probe is on the right. Scale bar, 20 μm. In the recording, the top trace indicates micrometer steps of the stimulator probe, whereas the bottom trace shows whole-cell currents. (F) Quantification of the current amplitude at each probe indentation depth. A total of n = 10 cells were recorded, with 10/10 cells displaying a peak mechanically evoked current above 50 pA. Values are expressed as mean ± SEM. See also Videos S3 and S4.

Figure 4.. Varied Induction of NGN2-BRN3A in Neural Crest Yields Divergent Sensory Neuron Populations

(A) Protocol and phase-contrast images of neural crest differentiation with two strategies for sensory neuron induction (NC-iSN1 and NC-iSN2). Day 0 depicts neurectodermal spheroid formation in a microwell plate. Day 12 shows neural crest migration from an individual spheroid. Scale bars, 100 μm. (B) Quantification of cell soma area based on phase-contrast images across three differentiation experiments of iSNs (n = 301 cells), NC-iSN1s (n = 228 cells), and NC-iSN2s (n = 332 cells). Values are expressed as mean ± SEM. Post hoc comparisons with Kruskal-Wallis test and Dunn’s correction for multiple comparisons, ****p in situ hybridization experiments on NC-iSN1s and NC-iSN2s. Scale bars, 25 μm. (D) Representative mechanical stimulation of NC-iSN2s in whole-cell voltage-clamp. A total of n = 6 cells were recorded. (E) RNA sequencing heatmap expression results of selected genes grouped by known sensory neuron subtype. Each heatmap column depicts the average log2 TPM of three independent samples. Δ, medium change; 1/2Δ, half medium change; C-LTMR, C fiber LTMR; LTMR, low-threshold mechanoreceptor; NC, neural crest; NP, nonpeptidergic nociceptor; PEP, peptidergic nociceptor; Proprio., proprioceptor. See also Figure S3.

Figure 5.. Single-Cell RNA Sequencing of Neural Crest-Derived iSNs

(A) Merged UMAP plot representing individual cells from a single differentiation of both NC-iSN1 (22,804 cells) and NC-iSN2 (15,004 cells). (B) Expression UMAP plots indicating cells in the upper 90% of the population distribution with mapped reads to specific genes, based on unique molecular identifier counts.

Figure 6.. PIEZO2LOF iSNs Are Insensitive to Mechanical Stimuli

(A) Representative immunocytochemistry from at least three differentiations of control and patient iSNs. Scale bar, 50 μm. (B) qRT-PCR of sensory-related genes in day 21 iSNs, normalized to the housekeeping gene RPLP0 and expressed as −ΔCt for color coding. Each column represents an independent sample, three samples per cell line. Human DRG (hDRG) total RNA was used as a positive control, and WTC11 iPSCs were used as a negative control. Text color denotes gene category: orange, pluripotency; magenta, pan-sensory; purple, LTMR; blue, proprioceptor; green, nociceptor and/or thermoreceptor. (C) Whole-cell voltage-clamp recordings of iSNs during mechanical stimulation. Traces are examples from n = 10 neurons for each cell line. (D) Quantification of mechanically activated current peak amplitudes in n = 10 cells per cell line. Values are expressed as mean ± SEM. Post hoc comparisons with Kruskal-Wallis test and Dunn’s correction for multiple comparisons. *p F/F0 of individual cells after exposure to 500 μM menthol, normalized to KCl response. For each cell line, ≥80 cells were analyzed. In post hoc comparisons between each sample with one-way ANOVA and Dunnett’s correction for multiple comparisons, no comparison was significantly different (p > 0.05). See also Figures S4–S7 and Table S2.