The Cytokine TGF-β Induces Interleukin-31 Expression from Dermal Dendritic Cells to Activate Sensory Neurons and Stimulate Wound Itching

Abstract

Cutaneous wound healing is associated with the unpleasant sensation of itching. Here we investigated the mechanisms underlying this type of itch, focusing on the contribution of soluble factors released during healing. We found high amounts of interleukin 31 (IL-31) in skin wound tissue during the peak of itch responses. Il31^-/- mice lacked wound-induced itch responses. IL-31 was released by dermal conventional type 2 dendritic cells (cDC2s) recruited to wounds and increased itch sensory neuron sensitivity. Transfer of cDC2s isolated from late-stage wounds into healthy skin was sufficient to induce itching in a manner dependent on IL-31 expression. Addition of the cytokine TGF-β1, which promotes wound healing, to dermal DCs in vitro was sufficient to induce Il31 expression, and Tgfbr1^f/f CD11c-Cre mice exhibited reduced scratching and decreased Il31 expression in wounds in vivo. Thus, cDC2s promote itching during skin would healing via a TGF-β-IL-31 axis with implications for treatment of wound itching.

Keywords: dermal dendritic cells; interleukin-31; itch; neuro-immune interaction; transforming growth factor beta; wound healing.

Graphical abstract

Figure 1

Pruritus Elicited by Acute Wound Healing and IL-31 Is Important for Itch during Wound Healing (A) 12 patients with 1- to 3-cm sterile surgical wounds who experienced pruritis were surveyed over 7 days and asked to score the level of itching they felt on each day. One-way ANOVA was used for comparisons. Bars represent means ± SEM. ^∗p < 0.05. (B) A mouse wound healing model was established by making 2 equidistant, 1 cm, full-thickness incisional wounds through the dorsal skin and left to heal. The number of scratching bouts per 30 min of observation were measured for 7 days following wounding. Data are from 3 independent experiments (n = 9) and were analyzed with a one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗p < 0.05. (C and D) Mouse wound skin at different time points was harvested, followed by RNA-seq of the mRNAs. The Venn diagram (C) shows overlapping of the genes that were significantly changed (p < 0.01) on the fifth day versus other time points of wound healing and genes related to itching based on published literature. The heatmap (D) shows all the itch-associated genes which changed on the fifth day were compared with any other time points. For RNA-seq, 2 wound tissues from one animal were pooled, and 3 animals were used per time point. (E) The expression of IL-31 in wounds was confirmed by qPCR. Data are from 3 independent duplicated experiments (n = 12) and were analyzed with a one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (F) Itching behaviors of Il31^−/− mice during wound healing were compared with wild-type mice. Data are pooled from 2 independent experiments (n = 6) and represented as mean ± SEM; two-way ANOVA was used. ^∗∗∗∗p < 0.0001. See also Figure S2 and Tables S1 and S2 for more details.

Figure 2

IL-31 Increases Itch Sensory Neuron Sensitivity (A–C) TRPV1+ cells from DRGs that innervate the fifth day’s wounds in Trpv1-lineage reporter mice (Trpv1-tdTomato mice) were sorted by flow cytometry, and the expression of Il31ra (A), Trpv1 (B), and Nppb (C) in these TRPV1+ cells that innervate the fifth day’s wounds were compared with TRPV1+ cells that innervate naive skin in normal controls by qPCR. Data were from 2 independent experiments (n = 5), and each sample was pooled from 2 mice. Student’s t test was used for comparisons. Bars represent means ± SEM. ^∗p < 0.05, ^∗∗p < 0.01. (D) The calcium transient in capsaicin (50 nM) was observed in DRG neurons treated with IL-31 (10 ng/mL) for 24 h and in untreated neurons. The solid blue and red lines were representative images (mean values) and dash lines were individual traces. (E) Phosphorylation of Stat3 was detected by western blot in DRG neurons treated with IL-31 (10 ng/mL) for 24 h. Data are representative of 3 independent experiments. (F) Scratching bouts of mut-Stat3 mice were counted after the first IL-31 injection (1 μg/site, i.d.) and compared with wild-type mice. 8 h after the first IL-31 injection, the second IL-31 injection was administered, and the itching behaviors were again observed. Data were from 2 independent experiments (n = 6) and analyzed with a two-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. See also Figure S3 for more details.

Figure 3

Dermal DCs Are a Key Source of IL-31 in Wounds (A) Il31 gene expression was determined in CD45⁻ cells, CD45⁺CD3⁺ cells, and CD45⁺CD3⁻ cells sorted from fifth-day wounds tissue by qPCR. Data were from 2 independent experiments with 2 samples each time, and each sample was pooled from 4 wounds in 2 mice. Data were analyzed with one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (B) Expression of IL-31 in wounds on the fifth day were tested in Rag1^−/− mice and wild-type mice by real-time PCR. Data are from 3 independent experiments (n = 6), and two-way ANOVA was used for comparison. Bars represent means± SEM. (C) Scratching behaviors were observed in Rag1^−/− mice before and on the fifth day of wound healing. Data are from 2 independent experiments (n = 5); Student’s t test was used. Bars represent means ± SEM. (D) DCs (CD45⁺CD3⁻CD11c⁺MHC II⁺), macrophages (CD45⁺CD3⁻CD11c⁻CD11b⁺F4/80⁺), granulocytes (CD45⁺CD3⁻CD11c⁻CD11b⁺F4/80⁻), and mast cells (CD45⁺CD3⁻CD11c⁻CD11b⁻FceR1a⁺) were sorted from fifth-day wounds, and the expression of Il31 was determined by real-time PCR. Data were from 2 independent experiments with 2 samples each time, and each sample was pooled from 4 wounds in 2 mice. Data were analyzed with one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (E and F) The frequency (E) and numbers (F) of CD11c⁺ cells from day 5 wounds was calculated and compared with those from normal skin. Data were from 3 independent experiments (n = 5-6), analyzed with Student’s t test for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (G) Itching behaviors during wound healing were observed in diphtheria toxin (DT)-treated (i.p., 500 ng/mouse for the first time and 100 ng/mouse on all subsequent days) zDC^DTR-to-C57BL/6 bone marrow chimera mice and were compared with DT-treated C57BL/6-to-C57BL/6 bone marrow chimera mice (WT [wild-type]). Data were pooled from 6 mice for each condition in two independent experiments and are represented as mean ± SEM. Two-way ANOVA was used for multiple comparisons. ^∗∗∗p < 0.001. See also Figure S4 for more details.

Figure 4

IL-31 in Wounds Was Mostly from Dermal cDC2s (A) Macrophages (CD45⁺CD3⁻CD11c⁻CD64⁺CD11b⁺F4/80⁺), LCs (CD45⁺CD3⁻CD11c⁺CD326⁺), dermal cDC2s (CD45⁺CD3⁻CD11c⁺CD326⁻CD64⁻CD11b⁺MHC II⁺), and dermal cDC1s (CD3⁻CD11c⁺CD326⁻CD64⁻CD11b⁻MHC II⁺) were sorted from fifth-day wounds, and Il31 expression was determined by real-time PCR. Data were from 2–3 independent experiments with 2 samples each time, and each sample was pooled from 4 wounds in 2 mice. Data were analyzed with one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (B) Il31 gene expression in dermal cDC2 from fifth-day wounds was compared with cDC2s from normal skin by real-time PCR. Data are from 3 independent experiments with 2 samples each time, and each sample was pooled from 4 wounds in 2 mice. Data were analyzed with Student’s t test for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (C–E) Flow cytometer analysis showed that the frequency (C and D) and total numbers of dermal cDC2s (CD45⁺CD11c⁺CD64⁻CD11b⁺MHC II⁺ CD326⁻) (E) were increased in fifth-day wound skin tissues compared with normal skin. In (D) and (E), each circle represents one mouse. Data were from 3 independent experiments (n = 5–6), and Student’s t test was used. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001.

Figure 5

Administration of Late Wound Stage Dermal cDC2s Evokes Itch Responses (A–D) Dermal cDC2s sorted from fifth-day wounds or normal skin was injected intradermally to the dorsal area of normal B6 mice at 15,000 per site, and the scratching behaviors were counted 3 h after injection (A). The expression of Il31ra (B), Trpv1 (C), and Nppb (D) in DRGs that innervate the back skin around the dermal cDC2-injected area was also determined by real-time PCR. Data were from 2 independent experiments with 3–4 mice each time. Student’s t test was used for comparison. Bars represent means ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001. (E and F) The numbers of scratching behaviors (E) and wiping behaviors (F) were counted after injection of cDC2s from fifth-day wounds or normal skin to cheeks of mice. Each circle represents one mouse. Data are from 2 independent experiments (n = 8), and Student’s t test was used. Bars represent means ± SEM. ^∗∗∗p < 0.001. (G) Scratching bouts were counted in recipient mice that were injected with cDC2s from fifth-day wounds of Il31^−/− mice or WT mice. Data were pooled from 3 independent experiments (n = 9). Each circle represents one mouse. Student’s t test was used. Bars represent means ± SEM. ^∗∗p < 0.01. See also Figure S5 for more information.

Figure 6

TGF-β Increases IL-31 in Dermal cDC2 (A) IL-1β (10 ng/mL), IL-6 (50 ng/mL), IL-17a (10 ng/mL), TGF-β1 (2 ng/mL), and TNF-α (10 ng/mL) were used in vitro to treat cDC2s sorted from healthy skin individually, and Il31 expression in dermal cDC2s was determined 6 h after treatment. Data were from 3 independent experiments (n = 6), and each circle represents one culture well. Data were analyzed with one-way ANOVA for comparisons. Bars represent means ± SEM. ^∗∗∗p < 0.001. (B and C) The expression of Il31 in TGF-β1-treated dermal cDC2s from Tgfbr1-deficient mice (tamoxifen-treated Tgfbr1^f/fErt2-Cre mice) (B) and Smad3^−/− mice (C) were determined and compared with WT dermal cDC2s. Data were from 2 independent experiments (n = 4–8). Each circle represents one culture well, and two-way ANOVA was used for comparisons. Bars represent means ± SEM. ^∗∗∗∗p < 0.0001. (D and E) The wound healing model was set up in Tgfbr1^f/fCd11c-Cre (D) and Smad3^−/− mice (E). Itching behaviors were observed in Tgfbr1^f/fCd11c-Cre mice and Smad3^−/− mice and compared with WT controls. Data were from 3 independent experiments (n = 6–8) and represented as mean ± SEM. Two-way ANOVA was used for comparisons. ^∗∗∗p < 0.001. (F) Dermal cDC2 from normal skin treated with or without TGF-β1 for 24 h were injected (i.d., 15,000 cells per site) to the dorsal area of normal B6 mice, and scratching bouts on the dorsal area of dermal cDC2 recipients were counted 3 h after injection. Data were from 3 independent experiments, each circle represents one mouse (n = 6). Student’s t test was used for comparison. Bars represent means ± SEM. ^∗p < 0.05. (G) Dermal cDC2 from normal or Tgfbr1-deficient mice (tamoxifen-treated Tgfbr1^f/fErt2-Cre mice) were also treated with TGF-β1 for 24 h and injected into the dorsal area of normal B6 mice, and scratch counts were determined in recipients of dermal cDC2s. Data were from 3 independent experiments. Each circle represents one mouse (n = 6). Student’s t test was used for comparison. Bars represent means ± SEM. ^∗p < 0.05. See also Figure S6 for more details.