Hydrogen peroxide production by epidermal dual oxidase 1 regulates nociceptive sensory signals

Abstract

Keratinocytes of the mammalian skin provide not only mechanical protection for the tissues, but also transmit mechanical, chemical, and thermal stimuli from the external environment to the sensory nerve terminals. Sensory nerve fibers penetrate the epidermal basement membrane and function in the tight intercellular space among keratinocytes. Here we show that epidermal keratinocytes produce hydrogen peroxide upon the activation of the NADPH oxidase dual oxidase 1 (DUOX1). This enzyme can be activated by increasing cytosolic calcium levels. Using DUOX1 knockout animals as a model system we found an increased sensitivity towards certain noxious stimuli in DUOX1-deficient animals, which is not due to structural changes in the skin as evidenced by detailed immunohistochemical and electron-microscopic analysis of epidermal tissue. We show that DUOX1 is expressed in keratinocytes but not in the neural sensory pathway. The release of hydrogen peroxide by activated DUOX1 alters both the activity of neuronal TRPA1 and redox-sensitive potassium channels expressed in dorsal root ganglia primary sensory neurons. We describe hydrogen peroxide, produced by DUOX1 as a paracrine mediator of nociceptive signal transmission. Our results indicate that a novel, hitherto unknown redox mechanism modulates noxious sensory signals.

Keywords: DUOX1; Dual oxidase 1; Hydrogen peroxide; NADPH oxidase; Nociception; Skin.

Graphical abstract

Fig. 1

Expression and activity of Duox1 in epithelial cells. Quantitative PCR analysis of the expression of NADPH oxidase family components in mouse tail skin (A), hindpaw skin (B), primary mouse keratinocytes (C) and HaCaT cells (D). Dot plots represent mean ± SEM from 3 to 4 independent experiments. Western blot analysis of Duox1 protein expression in tail skin and hindpaw skin tissue or primary kertinocytes from wild-type and Duox1 knockout mice (E and F) or HaCaT wild type cells or CRISPR-modified Duox1 knockout cells (G). Biotinyl tyramide assay in HaCaT wild-type cells or CRISPR-modified Duox1 knockout cells (H). Cells were treated with 1 μM thapsigargin in the presence or absence of horse radish peroxidase. Reaction solution also contained 27.5 μM biotinyl tyramide. After treatment, biotinylated molecules were labeled with fluorescent streptavidin and fixed. Cell nuclei were stained with To-Pro-3. Scale bars: 10 μm.

Fig. 2

Measurement of changes in Amplex Red fluorescence. (A) Primary mouse back skin keratinocytes were stimulated with 2 nM GSK 1016790A or 100 nM thapsigargin for 30 min. (B) HaCaT wild-type or CRISPR-modified Duox1 knockout cells were stimulated with 2 nM GSK 1016790A or 10 μM ATPγS or 1 μM thapsigargin for 30 min. Cumulative change in Amplex Red fluorescence after 30 min was normalized to the initial fluorescence signal. Representative plot (2 independent experiments for primary cells and 3 independent experiments for HaCaT cells) shows mean ± SD of triplicate. *p < 0.05, ***p < 0.001 compared with corresponding controls, by 2-way ANOVA.

Fig. 3

Histological analysis of wild-type and DUOX1-deficient mouse skin. (A) H&E staining of paraffin embedded tissue sections from wild-type and Duox1 KO mouse tail and paw skin. Scale bars: 50 μm. (B) Wild-type or Duox1 KO mouse paw skin sections were labeled with keratin-14, keratin-10 and loricrin antibodies. Scale bars: 25 μm. (C) Analysis of WT and Duox1 KO mouse tongue by transmission electron microscopy.

Fig. 4

Allyl isothiocyanate-induced thermal hyperalgesia on the tail of wild-type and Duox1 knockout mouse. (A and B) Thermal hyperalgesia was induced by allyl isothiocyanate (AITC)-contained mustard oil for 30 s. Comparison of the tail thermonociceptive threshold and its decrease after AITC treatment in wild-type and Duox1 knockout mice. Noxious heat threshold measurements were repeated at 10 min intervals for 60 min after treatment. Plots show mean ± SEM of n = 16–17 animals/group. ***p < 0.001 compared with corresponding controls, by 2-way ANOVA.

Fig. 5

Formalin-induced nociception in wild-type and Duox1 knockout mouse. After intraplantar 20 μl 2.5% formalin injection, spontaneous nocifensive behaviour was observed between 0-5 min and 20–45 min in wild type and Duox1 KO animals. (A) The duration of paw licking was measured by a stopwatch. (B) Paw flinching responses were counted. (C) Composite Pain Score (CPS) was also calculated by the following formula: CPS=(2x paw licking time + 1x paw flinchings)/observation time. Data are mean ± SEM of n = 5–5 animals/group. *p < 0.05.

Fig. 6

TRPV4 agonist, GSK 1016790A induced ATP secretion from wild-type and Duox1 knock-out HaCaT cells. (A) Extracellular ATP-sensitive, fluorescent GRABATP sensor expressing HEK293A cells were cocultured with WT or Duox1 knock-out HaCaT. After a wash to remove any residual ATP in the media, the cells were treated with 2.5 nM GSK 1016790A and then with 1 μM ATP, as a positive control. Gray lines represent the normalized mean GRABATP intensity of every cell over time. Blue line shows the overall average of the normalized mean intensity of all the cells ± SEM (WT: n = 38, KO: n = 53 cells from 3 independent experiments). (B) The area under the curve (AUC) was measured for each cell during the time of the GSK 1016790A stimulus, between 5 and 27 min and was plotted per condition (mean ± SEM, gray dots represent values of individual cells). (C) Fluorescent images from different time points showing local oscillations (marked by asterisks) and sustained elevations of the GRABATP signal. Scale bars: 25 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 7

Expression of ion channels and Duox1 in dorsal root ganglion and tail skin. (A) Quantitative PCR analysis of the expression of potassium channels, Duox1, DuoxA1 and TRP channels in wild-type and Duox1 knockout mouse dorsal root ganglion. Bars represent mean ± SEM from 2 independent experiments. (B) Quantitative PCR analysis of the expression of TRP channels in mouse tail skin. Bars represent mean ± SEM from 2 to 4 independent experiments.

Fig. 8

Ca²⁺ measurements with Fura-2-AM in HEK293 cells expressing TRPA1. (A) TRPA1-dependent effect of H₂O₂ on [Ca²⁺]_ic. [Ca²⁺]_Ic responses evoked by 500 μM H₂O₂ in the presence of TRPA1. (B) Change in [Ca²⁺]_ic of TRPA1 transfected HEK cells in response to sequential applications of 500 μM H₂O₂ and 10 μM AITC. Representative plot from at least 3 independent experiments shows the mean ± SD of triplicate.

Fig. 9

Hydrogen peroxide mediated redox changes of voltage-gated potassium channel, Kcnq4. (A) Quantitative PCR analysis of the Kcnq4 and Duox1 in wild-type mouse dorsal root ganglion, tail skin and paw skin. Bars represent mean ± SEM from 3 independent experiments. (B) Expression of Duox1 in spinal cord, tail and paw skin. (C) HEK293 cells expressing FLAG-tagged Kcnq4 were treated with H₂O₂ (0, 20, 100, 500 μM) and lysed in presence of BIAM. Following anti-FLAG immunoprecipitation, BIAM signal was detected by streptavidin-HRP on Western blot. (D) After H₂O₂ treatment (0, 20, 100 μM) non-oxidized thiols were alkylated with N-ethylmaleimide, then lysates were reduced with dithiothreitol and labeled with BIAM. We continued with anti-FLAG immunoprecipitation and streptavidin-HRP detection.