An in vitro organ-on-chip model for studying neuron–keratinocyte interactions in sensory response through electrophysiology

Abstract

This study introduces a human-relevant in vitro model using iPSC-derived sensory neurons and keratinocytes in MEA-integrated microfluidic chips. Neurons expressed nociceptor markers, showed TRPV activity, and formed contacts with keratinocytes. Stimuli evoked electrophysiological responses, highlighting neuron–keratinocyte interactions relevant to pruritus, pain, and skin disorders, supporting therapeutic development.

---

Introduction

The skin serves as a primary sensory organ, allowing the perception of various stimuli such as touch, pain, temperature, and itch. Sensory neurons play a crucial role in mediating these sensations by transmitting signals from the periphery to the central nervous system. Among these, specialized neurons are responsible for detecting painful (nociceptive) and pruritic stimuli, contributing to sensations such as discomfort and itch. Dysregulation or hypersensitivity of these neurons is implicated in various skin disorders, including chronic itch and hypersensitive skin syndromes.^1,2 Understanding the mechanisms governing sensory neuron function and their interactions with skin cells is therefore critical for developing targeted therapeutic strategies for sensory-related skin disorders.

Despite the prevalence of skin-related sensory disorders, the development of effective therapeutic compounds remains challenging due to the lack of physiologically relevant human in vitro models. Most existing in vitro skin models lack neuronal components, hindering their ability to accurately recapitulate sensory transduction mechanisms.

The recent advent of human iPSC-derived sensory neurons has proven to be a promising tool for studying neuronal function and screening potential therapeutics. These neurons exhibit key functional properties of nociceptors, and their responses to various stimuli are assessed using electrophysiological and biochemical readouts, including patch-clamp electrophysiology,^3–9 microelectrode array (MEA),^4,7,10–13 calcium imaging,^{3–5,8,14,15} and substance P release.^7,9,14

A physiologically relevant model of skin sensation requires more than neurons alone. Keratinocytes play a significant role in sensory transduction, contributing to mechanosensation and thermosensation. They express receptors such as transient receptor potential vanilloid 1 (TRPV1), which is involved in pain and heat sensation, and are believed to be involved in many other sensory mechanisms.¹⁶ Additionally, keratinocytes actively participate in neuroimmune interactions, releasing cytokines and inflammatory mediators that modulate neuronal activity and contribute to conditions such as neurogenic inflammation. Therefore, incorporating both sensory neurons and keratinocytes in an in vitro model expands its applicability by better capturing the diverse physiological interactions occurring in vivo.

Efforts to develop innervated skin models have focused on incorporating sensory neurons with keratinocytes or in reconstructed skin.¹⁷ These models range from simple co-culture systems, where sensory neurons are in direct contact with keratinocytes,^{14,15,18–20} to more complex approaches utilizing microfluidic devices for a physiological organization,^21–24 or reconstructed epidermis/skin explant with integrated neurons.^3,25–27 While these models offer improvements over neuron-free systems, they often lack the capacity to integrate large scale electrophysiological readouts, making it difficult to accurately assess neuronal responses to stimuli. Additionally, most models investigating sensory responses rely on animal dorsal root ganglia (DRG), with the limitations brought by molecular differences in the skin sensory signaling between rodents and humans.²⁸ Only a limited number employing fully human systems, which may offer greater physiological relevance. Moreover, current models often treat neurons as a single unit, without distinguishing between the soma and distal axonal terminals. This does not accurately reflect in vivo conditions, where sensory neurons primarily interact with target tissues via their peripheral endings. Notably, sensory neuron responsiveness has been shown to vary depending on the specific neuronal segment being stimulated.²⁹

Microfluidic technology offers a promising platform to address the limitations of existing innervated skin models. The primary advantage of microfluidic devices lies in their ability to compartmentalize different cell types, enabling controlled interactions between neurons and keratinocytes while maintaining physiological organization. Microfluidic channels have been extensively used in neuronal cultures to separate neuronal somas from axonal terminals, facilitating targeted treatments on nerve endings. This separation is particularly valuable for studying the specific effects of compounds or stimuli on distal neuronal terminals, which more closely mimics in vivo conditions.

Questions addressed

In this study, we developed a physiologically relevant in vitro model of MEA cutaneous sensitivity using a commercially available MEA-integrated microfluidic chip (DuaLink-MEA, NETRI). This system compartmentalizes sensory neurons and keratinocytes, enabling controlled interactions between these cell types while maintaining their spatial organization. Microchannels enable the selective exposure of neuronal endings to stimuli while keeping the somas isolated. The integration of MEA electrodes within the microchannels allows real-time and non-invasive electrophysiological recordings of neuronal activity in response to stimuli applied to the keratinocyte compartment. This model presents a novel approach for studying sensory neuron–keratinocyte interactions under physiologically relevant conditions. By allowing targeted treatment of neuronal axonal endings and real-time functional assessment, this platform offers significant advantages for investigating skin sensation mechanisms and evaluating potential therapeutic compounds.

Results

Compartmentalized culture of human iPSC-derived sensory neurons in MEA-compatible microfluidic chips

To establish a human-relevant in vitro platform for studying skin sensory responses, we cultured human iPSC-derived sensory neurons in a commercially available microfluidic chip (DuaLink, NETRI) integrated with microelectrode arrays (MEAs). This device comprises three longitudinally aligned compartments connected by microchannels, which permit axonal growth while preventing neuronal cell bodies migration. The leftmost compartment was seeded with sensory neuron somas, and axons progressively extended through the microchannels into the terminal endings compartment on the right-hand side (Fig. 1A; S1A). The central compartment provided enhanced fluidic isolation between the distal compartments, enabling selective treatment of axonal endings without affecting neuronal somas. The chip's ANSI/SLAS-format-compatible layout and 96-well plate-like footprint facilitate parallel experiments and high-content analysis (Fig. S1B).

Fig. 1 Compartmentalized culture and characterization of human iPSC-derived sensory neurons in MEA-compatible microfluidic chips. A) Schematic representation of the NETRI device architecture (left) and immunofluorescence image of iPSC-derived sensory neurons stained for β-III-tubulin after 21 days in culture (right). B) Immunofluorescence staining for Nav1.7, TRPV1, P2X3, substance P, and TRPA1 (cyan), co-stained with either βIII-tubulin or neurofilament (magenta). C) Neuronal response to a temperature challenge, (Top) representative traces showing activity modulation. (Bottom) Quantification of the relative change between defined segments (separated by dotted lines). Scale bars: 200 μm.

Human iPSC-derived sensory neurons (Axol®) were cultured in a microfluidic device. Surprisingly, despite homogeneous seeding, the neurons formed ganglion-like clusters near the microchannel entrances after three days of culture—immediately after mitomycin treatment—unlike in conventional culture plates (Fig. S2). Axonal projections reached the middle compartment by day 3, extending into the terminal endings compartment by day 9. By day 17, axons spanned the entire width of the terminal endings compartment and continued to elongate vertically, reaching lengths of at least 1.5 mm (Fig. 1A; S2).

To confirm the proper neuronal differentiation within the microfluidic device, immunofluorescence staining was performed at day 21 using key nociceptor differentiation markers, and counterstained with a nonspecific neuronal marker such as neurofilament or tubulin beta 3. The assay confirmed the expression of general sensory neuron differentiation markers, including BRN3A, Islet 1, NaV1.7, TrkA and substance P, indicating successful sensory neuron specification (Fig. 1B; S3). Also, neurons expressed ion channel receptors essential for nociceptive and pruritic signaling, including TRPV1, TRPA1, and P2X3 (Fig. 1B). The presence of both P2X3 and substance P staining indicates a mixed population of peptidergic and non-peptidergic neurons resulting from the differentiation.

Spontaneous electrical activity was detected in all compartments using the integrated MEA system (Fig. S4A). Notably, spike amplitudes were significantly higher in the microchannel electrodes, with values reaching up to 1.2 mV (Fig. S4A and B). This amplification correlated with the increased resistance observed in the microchannels (Fig. S4C) consistent with the previous observations in MEA systems combined microchannel architectures. The small dimension of microchannels (18 μm² section over 125 μm length) result in a smaller available volume of conductive medium compared to an open well. Additionally, axonal growth within the microchannels decreases even more the amount of available conductive medium. According to Ohm's law, electrodes record a higher voltage amplitude of the extracellular action potential induced by a fixed current flow.

In addition to enhanced signal amplification, microchannels also topographically constrain axons to grow over electrodes. Microchannel electrodes therefore consistently recorded higher mean firing rates (MFRs). Specifically, while the soma and terminal ending compartments exhibited average MFRs of ∼0.22 Hz, the microchannel electrodes recorded MFRs averaging 45 Hz—an approximately 20-fold increase (Fig. S4D). Therefore, subsequent analyses were conducted using data from microchannel electrode, reflecting afferent signals from distal axonal terminals to somas.

To assess the functional responsiveness of the sensory neurons to external stimuli, we applied a global temperature increase from 37 °C to 41 °C and quantified neuronal responses via MEA recordings from the microchannels. A significant increase in spike frequency was observed, with 107 out of 129 active electrodes (>83%) exhibiting an increase in activity exceeding 10% (Fig. 1C). These results demonstrate that the neurons can transduce temperature increases into electrical activity, indicating functional TRPV channel activity and validating the platform's functional relevance.

Sensory neurons interact with keratinocytes in microfluidic chips

To mimic the complex cellular interactions involved in skin sensory transduction, we incorporated primary Normal Human Epidermal Keratinocytes (NHEKs) into the microfluidic platform. Keratinocytes play a key role in skin sensation, contributing to mechanosensation, thermosensation, and neuroimmune signaling.¹⁶ Their inclusion is essential for creating a model suitable for sensory transduction studies.

Immunofluorescence confirmed that the NHEKs expressed key functional markers consistent with their known sensory role,^20,30,31 including cytokeratin 14 (K14), TRPV1, TRPA1 and P2X3 (Fig. 2A; S5).

Fig. 2 Sensory neuron–keratinocyte interaction in microfluidic chips. A) Immunofluorescence staining of primary human keratinocytes for cytokeratin 14 (CK14). B–C) Immunofluorescence images of the terminal endings compartment in sensory neuron–keratinocyte co-cultures: B) βIII-tubulin (β3T) (magenta), CK14 (yellow), and DAPI (cyan); C) Synaptophysin (magenta), connexin 43 (yellow), and DAPI (cyan). D) Representative images of neurites regrowth 4 days post-axotomy. Neurites were cultured in neuron medium (top), keratinocyte medium (middle), and keratinocyte medium with keratinocytes (bottom). E) Quantification of neurite regrowth 4 days post-axotomy. Scale bars: 100 μm.

When keratinocytes were seeded into the terminal endings compartment of the microfluidic device, imaging revealed physical contact between the two cell types; however, no ensheathment or synaptic-like structures could be identified at this imaging resolution (Fig. 2B).

To explore potential interactions between sensory neurons and keratinocytes, we performed immunostaining for two previously reported markers of cellular connectivity:^15,18,32 Connexin 43 (Cx43), a gap junction protein, and synaptophysin, a synaptic vesicle marker. Synaptophysin was detected exclusively in neuronal projections and was not observed in keratinocytes (Fig. 2C; S3), although higher-resolution imaging might reveal low-level expression or synaptic structures in these cells. Connexin 43 appeared as punctate spots within the keratinocyte layer and localized near neurites, suggesting potential interaction sites (Fig. 2C). These observations are consistent with earlier findings in co-culture models of iPSC-derived sensory neurons and keratinocytes,¹⁵ supporting the existence of neuron–keratinocyte communication.

Assessing the impact of keratinocytes on neurites growth is challenging when neurons have fully colonized the terminal endings compartment. To address this, we employed an axotomy/regrowth assay, to quantify neurite regrowth in the presence or absence of keratinocytes. Following axotomy, sensory neurons exhibited robust regrowth in standard neuronal culture medium supplemented with neurotrophic factors (BDNF, β-NGF, NT3, GDNF). In contrast, neurite regrowth was negligible when neurons were cultured solely in keratinocyte medium (Fig. 2D), indicating that standard keratinocyte culture conditions do not support neuronal regeneration.

Interestingly, the presence of keratinocytes in the distal compartment appeared to enhance neurite regrowth. In co-culture conditions without added growth factors, axons extended into the keratinocyte layer. Neurotrophic factors supplementation further promoted this growth (Fig. 2D). Quantitative analysis revealed comparable neurite coverage and length between neurons cultured in supplemented neuronal medium alone (coverage: 6.2%, mean length: 332 μm) and those co-cultured with keratinocytes, either in unsupplemented (coverage: 5.8%, mean length: 234 μm) or supplemented conditions (coverage: 7.6%, mean length: 314 μm). In contrast, there was next to no regrowth in keratinocyte medium alone (coverage: 0.3%, mean length: 12 μm) (Fig. 2E).

These results provide evidence of functional interaction between the two cell types, potentially mediated by keratinocyte-secreted neurotrophic factors.³³ Altogether, this highlights the importance of incorporating keratinocytes in the model to better replicate the cellular crosstalk present in native skin tissue and facilitate future investigations into neurocutaneous signal integration, transmission, and sensory modulation.

Stimulation of sensory neurons and co-cultures reveals functional responses to pruritogenic and acidic stimuli

Next, the functional sensitivity of the microfluidic neuron–keratinocyte model was evaluated. To maintain physiological relevance, treatments were applied exclusively to the nerve endings compartment, mimicking the in vivo exposure of peripheral terminals to environmental stimuli, endogenous factors and topical treatments (Fig. 3A). This spatially restricted stimulation was facilitated by the device's microfluidic design, which ensures fluidic isolation between compartments.

Fig. 3 Functional responses of sensory neurons and co-cultures to pruritogenic and acidic stimuli. A) Schematic of sensory neuron culture (left) and sensory neuron–keratinocyte co-culture (right) in the MEA microfluidic device, illustrating selective stimulation of the distal axon terminal endings. B) Immunofluorescence image of the co-culture stained with βIII-tubulin (yellow) and phalloidin (green). C) Representative spike count over time (0.5 s bins), showing sequential treatment with vehicle, ATP (20 μM), and KCl (8 mM). D) Change in Fura-2 fluorescence (ΔF/F₀) (calcium probe intensity changes) in keratinocytes following sequential treatment with vehicle, ATP (20 μM), or lactic acid (0.3%). E) Quantification of the maximum neuronal activity after stimulation with ATP or lactic acid, normalized to baseline. F) Quantification of maximum neuronal activity following co-culture sequential treatment with ATP and lactic acid, normalized to baseline. Scale bar: 200 μm.

Two compounds were selected for their known capacity to activate sensory pathways: ATP and lactic acid (LA). ATP, a well-established pruritogen, activates purinergic receptors such as P2X3 in nociceptors and participates in neuron–keratinocyte sensory transduction.²⁰ Lactic acid is associated with non-specific proton sensitivity and discomfort, particularly in the context of inflamed or sensitive skin, it is used in clinical trials.³⁴

Prior to co-culture evaluation, Human iPSC-derived sensory neurons and keratinocytes were tested in monoculture to confirm their responsiveness. Sensory neurons exhibited robust functional responses to both ATP and LA treatment at their terminal endings. In both cases, stimulation induced a transient increase in neuronal activity recorded via microchannel electrodes, followed by a temporary suppression of firing and a subsequent return to baseline activity levels (Fig. 3C). This dynamic response pattern is typical of sensory neuron firing behavior following acute stimulation. Quantitative analysis confirmed a statistically significant increase (p value < 0.0001) in spike frequency after both ATP and LA treatment (Fig. 3E), validating the platform's ability to detect rapid changes in neuronal excitability.

Given the established role of keratinocytes in sensory modulation, we next assessed the functional responsiveness of keratinocyte alone. To ensure reproducibility and account for inter-donor variability, pooled-donor primary keratinocytes were used. Calcium imaging with a Fura-2 probe confirmed that keratinocytes in monoculture respond directly to both ATP and LA stimulations (Fig. 3D; S6), consistent with their expression of purinergic and acid-sensitive receptors.

Subsequently, the response of the axon keratinocyte complex was trialed. To this end, keratinocytes were seeded into the distal compartment. Keratinocytes were seeded at high confluence to create a layer overlapping most of the neuronal terminals and maximizing potential neuron–keratinocyte interaction points while minimizing the neurites direct exposure to compounds (Fig. 3A and B).

When ATP and LA were applied sequentially to the axon–keratinocyte complex, a clear electrophysiological response was again observed (p value < 0.0001). The response profile remained consistent with that of neurons alone, indicating the responsiveness of the axon–keratinocyte complex (Fig. 3F). However, subtle signal modulations by keratinocytes are unlikely to be detected by MEA recordings, which primarily capture discrete action potential. Patch-clamp analysis would be more appropriate for revealing such subtle changes in ionic currents. Future studies employing more specific molecular stimuli may better reveal keratinocyte-mediated modulation of neuronal signal integration.

These data validate the platform's capacity to model physiologically relevant sensory responses in both monoculture and co-culture, particularly the response to ATP, a key mediator of sensory transduction.

Experimental design

MEA microfluidic device, preparation and coating

DuaLink-MEA devices (NETRI, NF-DL-MEA) were produced following previously published protocols.^35,36 Devices were left for one hour under UV and one hour degassing in a vacuum chamber before use. Microfluidic channels were opened with 70% ethanol and immediately rinsed with PBS. Channels were then coated with 0.1 mg mL⁻¹ poly-D-lysin (PDL) (Thermofischer, A38904) overnight in a cell culture incubator. After 3 washes with PBS, channels were incubated for 4 hours with Surebond-XF (Axol, ax0053) diluted 1 : 200 in PBS. The Surebond-XF solution was removed prior to cell seeding.

Cell culture and treatments

Sensory neurons were obtained from neural crest progenitors (Axol, ax0055) differentiation. 40 k cells were seeded in the neuronal cell body compartment in 3 μL of Sensory Neuron Maintenance Medium (SNMM, Axol, ax0060) supplemented with Surebond-XF (1 : 400) and the ROCK inhibitor Y-27632 (Focus Biomolecules, 10-2301) at 10 μM. Surebond-XF was maintained in the culture medium up until day 9. On day 1, the medium was fully replaced with SNMM supplemented with sensory maturation MAXIMIZER supplement (Axol, ax0058), 10 μg mL⁻¹ GDNF (Axol, ax139855), 20 μg mL⁻¹ NGF (Axol, ax139789), 10 μg mL⁻¹ BDNF (Axol, ax139789), 10 μg mL⁻¹ NT-3 (Axol, ax139811). On day 3, cells were treated with 1.9 μg mL⁻¹ mitomycin C (Sigma, M4287) for 2 hours and then washed with supplemented SNMM. Subsequently, 75% of the medium change was replaced on day 6 and 50% was replaced two times per week. Cells were cultured for 21 days before any assay.

Human, pooled donor, juvenile primary keratinocytes (Promocell, C-12005) were cultured in fully supplemented Keratinocyte Growth Medium 2 (KGM2, Promocell C-20011). Human, single donor, juvenile primary keratinocytes (EPISKIN) were cultured in Green 7 factor medium (GR7F). Cells were used at maximum passage number of 4.

Cells were treated with ATP 20 μM (Sigma, A3377) and lactic acid 0.3% (Sigma, L1750).

Axotomy and co-culture

Axotomy was performed on day 17. The medium was removed from neuronal ending channel and 30 μL of 0.5% Triton X100 solution was added in the inlet. After 30 seconds of incubation, the solution was aspirated, and the channel was washed twice with medium. Chips were incubated 5 min at 37 °C before full medium renewal in the channel. Post axotomy, the medium in the terminal ending channel was replaced with GR7F, followed by the seeding of 1000 EPISKIN keratinocytes. Cells were fixed on day 21 and stained with an ubiquitous neuronal marker. Neurite morphology was analyzed with ImageJ.

For co-cultures, on day 20 of neuronal culture, the medium in terminal ending channel was replaced with KGM2, followed by seeding 40 k Promocell keratinocytes. Co-cultures were stimulated 24 hours after keratinocyte seeding.

Immunostaining and image acquisition

Cells were fixed with 4% paraformaldehyde (Thermofisher, 28908) and permeabilized and blocked with 3% BSA (Sigma, A9418).

Cells were stained with the following primary antibodies or dyes: β-III tubulin (Merck, MAB1637), Nav1.7 (Abcam, ab65167), TRPV1 (Abcam, ab3487), P2X3 (Invitrogen, PA5-115707), substance P (Abcam, ab14184), TRPA1 (Invitrogen, PA1-46159), Neurofilament (Abcam, ab254348), Synaptophysin (Abcam, ab309493), Islet 1 (Abcam, ab20670), CK14 (Abcam, ab7800), Connexin 43 (Abcam, ab314908) or phalloidin (Abcam, ab176753).

Secondary antibodies were goat anti-rabbit IgG Alexa Fluor 488/633 (Invitrogen, A11070/A21072) and goat anti-mouse IgG Alexa Fluor 488/633 (Invitrogen, A11017/A21053).

Nuclei were counterstained with 3 μM DAPI (Sigma, D8417).

Fluorescent images were acquired with an Axio Observer 7 confocal microscope (Carl Zeiss).

Calcium assay

Calcium assays were performed using a Flexstation 3 (Molecular Devices). Promocell keratinocytes were seeded in 96-well black plates (Greiner, 655090) and incubated for 24 h. Cells were loaded with 5 μM Fura-2 (Molecular Probes, F1221), 0.05% pluronic acid (Molecular Probes, P6866) in HBSS (Invitrogen, 14065) containing 20 mM HEPES and 4 mM NaOH and then incubated at 37 °C for 30 min. Responses were measured every 6 s for 300 s and, analyzed using GraphPad Prism.

Microelectrode array recording

All microfluidic chips underwent a full medium change at least 3 hours prior to recording. Medium volume within reservoirs was then minimized. Plates were equilibrated for 15 minutes in the MaestroPro (Axion Biosystems) before initiating spontaneous activity recordings. For the temperature challenge, three consecutive 5-minute recordings were performed: at 37 °C, during a shift to 41 °C (first 2 minutes), and again at 37 °C. For chemical stimulation, a continuous recording was made throughout baseline, vehicle addition, treatment, and 8 mM KCl addition. Between each step, medium was partially removed to standardize starting volumes. All treatments (20 μL) were applied exclusively to the neuronal processes compartment, with 3-minute undisturbed intervals between additions.

MEA quantification and statistical analysis

Using AxIS Navigator software (Axion Biosystem), spikes were detected using a spike detector set with an adaptative threshold at 6 standard deviations above the noise. The resulting spike list with the time tag of each spike was extracted and compiled with a Python script to compute the activity rate of microchannel electrodes over time using bin size of 0.5 second. For each analyzed event, the maximum activity following the event was detected and normalized by the mean activity over the 60 seconds preceding the event.

Statistical analysis was performed using a non-parametric one-way ANOVA (Friedman test), with paired data representing different events within each chip. Significant levels are indicated as follows: * if p < 0.05, ** if p < 0.01, *** if p < 0.001 and **** if p < 0.0001.

Conclusions

This study presents an in vitro model integrating human iPSC-derived sensory neurons with primary keratinocytes in a microfluidic, MEA-compatible system. By enabling compartmentalized culture and selective stimulation of the axon terminals–keratinocyte complex, the platform more accurately reflects peripheral sensory architecture than conventional models. Its compatibility with moderate-throughput formats further positions it as a promising tool for screening sensory-modulating compounds.

While the co-culture system successfully demonstrates neuronal responsiveness to ATP and LA stimulation, the functional contribution of keratinocytes to this response remains unaddressed and warrants further investigation. The presence of gap junctions, the potential release of soluble factors, and neuronal sensitivity to ATP suggest a possible role for keratinocytes in sensory transduction. This model provides a valuable platform to study neuron–keratinocyte interactions in signal transmission and sensory modulation.

Future work will focus on mechanistic studies to elucidate the precise role of keratinocytes in signal integration, including ion channel knockdown experiments or the use of gap junction blockers. These approaches will help determine whether keratinocytes modulate neuronal sensitivity and identify specific pathways involved in neurocutaneous communication.

Author contributions

T. Bessy, A. Guichard, A. Martinez, D. Lelièvre were involved in experiment design and conduct, as well as in data collection, analysis and interpretation. They were involved in manuscript preparation and editing. They have full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. A. Martinez, C. Baquerre, L. Dubuisson, A. Batut, A. Berthelin, C. Grégoire, S. Teluob and A. Azéma were involved in experiment conduct, data collection and analysis. All authors have read and approved the final manuscript.

Conflicts of interest

All authors are employees of either EPISKIN or NETRI. This research received no specific grant from any external funding agency.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request. Supplementary information (SI): the SI details microfluidic device schematics, iPSC-derived sensory neuron morphological and electrophysiological characterization, and primary human keratinocyte sensory function assessed via immunofluorescence and calcium imaging. See DOI: https://doi.org/10.1039/d5lc00867k.

Acknowledgements

All laboratory members for their technical support and insightful discussions. Serge Roux for his feedback on figure organization. Juliane Robert for her initial guidance and training in the use of microfluidic chips and sensory neurons culture.

Notes and references

1. L. K. Oetjen, M. R. Mack, J. Feng, T. M. Whelan, H. Niu, C. J. Guo, S. Chen, A. M. Trier, A. Z. Xu, S. V. Tripathi, J. Luo, X. Gao, L. Yang, S. L. Hamilton, P. L. Wang, J. R. Brestoff, M. L. Council, R. Brasington, A. Schaffer, F. Brombacher, C. S. Hsieh, R. W. Gereau, M. J. Miller, Z. F. Chen, H. Hu, S. Davidson, Q. Liu and B. S. Kim, Cell, 2017, 171, 217–228.e13.
2. J.-K. Cheng and R.-R. Ji, Neurochem. Res., 2008, 33, 1970–1978.
3. Z. Guo, C. K. Tong, J. Jacków, Y. S. Doucet, H. E. Abaci, W. Zeng, C. Hansen, R. Hayashi, D. DeLorenzo, A. Rami, A. Pappalardo, E. A. Lumpkin and A. M. Christiano, Bioeng. Transl. Med., 2021, 7(1), e10247.
4. A. K. Kalia, C. Rösseler, R. Granja-Vazquez, A. Ahmad, J. J. Pancrazio, A. Neureiter, M. Zhang, D. Sauter, I. Vetter, A. Andersson, G. Dussor, T. J. Price, B. J. Kolber, V. Truong, P. Walsh and A. Lampert, Stem Cell Res. Ther., 2024, 15(1), 99.
5. A. K. Holzer, C. Karreman, I. Suciu, L. S. Furmanowsky, H. Wohlfarth, D. Loser, W. G. Dirks, E. Pardo González and M. Leist, Stem Cells Transl. Med., 2022, 11, 727–741.
6. P. Röderer, A. Belu, L. Heidrich, M. Siobal, J. Isensee, J. Prolingheuer, E. Janocha, M. Valdor, S. Hagendorf, G. Bahrenberg, T. Opitz, M. Segschneider, S. Haupt, A. Nitzsche, O. Brüstle and T. Hucho, Pain, 2023, 164, 1718–1733.
7. T. Deng, V. M. Jovanovic, C. A. Tristan, C. Weber, P. H. Chu, J. Inman, S. Ryu, Y. Jethmalani, J. Ferreira de Sousa, P. Ormanoglu, P. Twumasi, C. Sen, J. Shim, S. Jayakar, H. X. Bear Zhang, S. Jo, W. Yu, T. C. Voss, A. Simeonov, B. P. Bean, C. J. Woolf and I. Singeç, Stem Cell Rep., 2023, 18, 1030–1047.
8. J. W. Blanchard, K. T. Eade, A. Szůcs, V. Lo Sardo, R. K. Tsunemoto, D. Williams, P. P. Sanna and K. K. Baldwin, Nat. Neurosci., 2015, 18, 25–35.
9. M. Z. P. Guimarães, R. De Vecchi, G. Vitória, J. K. Sochacki, B. S. Paulsen, I. Lima, F. Rodrigues da Silva, R. F. M. da Costa, N. G. Castro, L. Breton and S. K. Rehen, Front. Mol. Neurosci., 2018, 11, 277.
10. B. J. Black, R. Atmaramani, S. Plagens, Z. T. Campbell, G. Dussor, T. J. Price and J. J. Pancrazio, Biosens. Bioelectron., 2019, 126, 679–689.
11. M. Hiranuma, Y. Okuda, Y. Fujii, J. P. Richard and T. Watanabe, Sci. Rep., 2024, 14, 6011.
12. S. Nimbalkar, X. Guo, A. Colón, M. Jackson, N. Akanda, A. Patel, M. Grillo and J. J. Hickman, Front. Cell Dev. Biol., 2023, 11, 1011145.
13. W. Plumbly, N. Patikas, S. F. Field, S. Foskolou and E. Metzakopian, Cells Rep. Methods, 2022, 2(11), 100341.
14. C. Bocciarelli, N. Cordel, R. Leschiera, M. Talagas, C. Le Gall-Ianotto, W. Hu, P. Marcorelles, G. Bellemere, S. Bredif, J. Fluhr, L. Misery and N. Lebonvallet, Exp. Dermatol., 2023, 32, 1563–1568.
15. C. Erbacher, S. Britz, P. Dinkel, T. Klein, M. Sauer, C. Stigloher and N. Üçeyler, Elife, 2024, 13, e77761.
16. M. Denda and S. Nakanishi, Exp. Dermatol., 2022, 459–474.
17. A. Guichard, N. Remoué and T. Honegger, Cosmetics, 2022, 9, 67.
18. M. Talagas, N. Lebonvallet, R. Leschiera, G. Sinquin, P. Elies, M. Haftek, J. P. Pennec, D. Ressnikoff, V. La Padula, R. Le Garrec, K. L'herondelle, O. Mignen, L. Le Pottier, N. Kerfant, A. Reux, P. Marcorelles and L. Misery, Ann. Neurol., 2020, 88, 1205–1219.
19. L. Ulmann, J. L. Rodeau, L. Danoux, J. L. Contet-Audonneau, G. Pauly and R. Schlichter, Eur. J. Neurosci., 2007, 26, 113–125.
20. S. Koizumi, K. Fujishita, K. Inoue, Y. Shigemoto-Mogami, M. Tsuda and K. Inoue, Biochem. J., 2004, 380(Pt 2), 329–338.
21. P. G. Mazzara, E. Criscuolo, M. Rasponi, L. Massimino, S. Muggeo, C. Palma, M. Castelli, M. Clementi, R. Burioni, N. Mancini, V. Broccoli and N. Clementi, Biomedicines, 2022, 10(9), 2068.
22. C. Tsantoulas, C. Farmer, P. Machado, K. Baba, S. B. McMahon and R. Raouf, PLoS One, 2013, 8(11), e80722.
23. J. Ahn, K. Ohk, J. Won, D. H. Choi, Y. H. Jung, J. H. Yang, Y. Jun, J. A. Kim, S. Chung and S. H. Lee, Nat. Commun., 2023, 14(1), 1488.
24. J. Kumamoto, M. Nakatani, M. Tsutsumi, M. Goto, S. Denda, K. Takei and M. Denda, Exp. Dermatol., 2014, 23(1), 58–60.
25. Q. Muller, M. J. Beaudet, T. De Serres-Bérard, S. Bellenfant, V. Flacher and F. Berthod, Acta Biomater., 2018, 82, 93–101.
26. R. Pepin, J. Ringuet, M. J. Beaudet, S. Bellenfant, T. Galbraith, H. Veillette, R. Pouliot and F. Berthod, Acta Biomater., 2024, 182, 1–13.
27. N. Lebonvallet, J. W. Fluhr, C. Le Gall-Ianotto, R. Leschiera, M. Talagas, A. Reux, A. Bataille, C. Brun, T. Oddos, J. P. Pennec, J. L. Carré and L. Misery, Skin Health Dis., 2021, 1(4), e66.
28. M. Sun, Z. Chen, H. Ding and J. Feng, Acta Pharmacol. Sin., 2025, 46, 539–553.
29. A. J. Clark, G. Menendez, M. AlQatari, N. Patel, E. Arstad, G. Schiavo and M. Koltzenburg, Pain, 2018, 159, 1413–1425.
30. K. Inoue, M. Denda, H. Tozaki, K. Fujishita, S. Koizumi and K. Inoue, J. Invest. Dermatol., 2005, 124, 756–763.
31. R. Atoyan, D. Shander and N. V. Botchkareva, J. Invest. Dermatol., 2009, 129(9), 2312–2315.
32. C. Seto, K. Toyoda, K. Inada, K. Oka and E. Ito, Biophys. Physicobiol., 2022, 19, e190041.
33. E. Di Marco, P. Carlo MarchisioS, S. Bondanza, A. Tito FranziP, R. Cancedda and M. De Lucall, Growth-regulated Synthesis and Secretion of Biologically Active Nerve Growth Factor by Human Keratinocytes, 1991, vol. 266.
34. L. Misery, K. Loser and S. Ständer, J. Eur. Acad. Dermatol. Venereol., 2016, 30(Suppl 1), 2–8.
35. T. Gabriel-Segard, J. Rontard, L. Miny, L. Dubuisson, A. Batut, D. Debis, M. Gleyzes, F. François, F. Larramendy, A. Soriano, T. Honegger and S. Paul, Int. J. Mol. Sci., 2023, 24(13), 10568.
36. L. Miny, J. Rontard, A. Allouche, N. Violle, L. Dubuisson, A. Batut, A. Ponomarenko, R. Talbi, H. Gautier, B. G. C. Maisonneuve, S. Roux, F. Larramendy, T. Honegger and I. Quadrio, Sci. Rep., 2025, 15(1), 29738.

---

Footnote

† Co-first authors.

---