TLR7 promotes skin inflammation via activating NFκB-mTORC1 axis in rosacea

Human samples

All the skin biopsies were obtained from the central face of 24 female patients (20–50 years old) diagnosed with rosacea by clinical and pathologic examination and 17 age-matched female healthy volunteers from the Department of Dermatology in Xiangya Hospital, Central South University. The patients details were shown in the supplementary materials of our previous study (Deng et al., 2021a). The severity of rosacea was evaluated by Investigator Global Assessment (IGA) scores and Clinician’s Erythema Assessment (CEA) scores according to previous study (Wang et al., 2021). The utilization of all human samples was approved by the ethical committee of the Xiangya Hospital of Central South University (IRB number 201404361), and all the participants signed the written informed consent. This program was conducted in accordance with the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

Mice and treatments

Six weeks of BALB/c mice were purchased from Slack Company and bred under SPF conditions. The LL37-induced rosacea-like mouse model was generated as previously described (Yamasaki et al., 2007). Simply put, mice, anaesthetized by intraperitoneal injection of avertin, were shaved a day before injection and then were injected with LL37 peptide (40 μl, 320 μM) or control vehicle (PBS) on back skin twice daily for 2 days. The skin inflammation of mouse models was evaluated by the severity of erythema and edema according to previous methods (Chen et al., 2019). The redness was scored ranging from 1–5, and 5 being the reddest. The area of redness was measured by stereomicroscope measurements (Leica S8AP0; Leica, Wetzlar, Germany). Mice were anaesthetized and decapitated to obtain the back skin for histological Analysis, immunohistochemistry, immunofluorescence and RT-qPCR. For TLR7 knockdown treatment, 6 weeks of BALB/c mice were intradermally injected with 1.5 nmol/L scrambled/Tlr7 siRNA (obtained from Shanghai GenePharma Co., Ltd.) three days and one day before LL37 injection. Procedures performed were approved by the ethical committee of the Xiangya Hospital, Central South University (IRB number 201611610).

RNA-sequencing

Total RNA from human skin and HaCaT keratinocytes were extracted using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA). Each group of samples were analyzed as three biological replicants. Library preparation and transcriptome sequencing were prepared using Illumina HiSeq X Ten (Novogene, Beijing, China). The mapping of 100-bp paired-end reads to genes was conducted with HTSeq v0.6.0. The fragments of every kilobase of transcript per million fragments mapped (FPKM) were analyzed, and differential expression analysis was produced by the DESeq R package (1.10.1). The hierarchical clustering heat map was generated with the ggplot library. Differential expression data were ranked through log₂ fold change and performed Preranked GSEA to figure out enrichment for KEGG pathways (Subramanian et al., 2005). The statistical power of this experimental design, calculated in RNASeqPower is 1.0. The sequencing depth is 37.00 (https://doi.org/doi:10.18129/B9.bioc.RNASeqPower). The nominal P value was adjusted with the Benjamini and Hochberg methods to correct the output results for multiple comparisons.The raw sequence data of HaCaT keratinocytes have been deposited in Figshare that are publicly accessible at https://doi.org/10.6084/m9.figshare.22725614.v2. Sequencing data from rosacea patients have been deposited in the genome sequence archive under accession number HRA000378.

Histological analysis

Histological analysis was performed as previously described (Wu et al., 2018; Zhao et al., 2018). Skin samples of mice were fixed in formalin and then embedded in paraffin. Skins were cut into 5 μm sections and then were stained with hematoxylin and eosin (H&E). The number of infiltrating cells in dermis was identified as histological feature. The average number of infiltrating cells in six randomly selected microscopic areas (original magnification, 200×) in each sample was determined as the amount of infiltrating cells in dermis.

Immunohistochemistry

Skin samples of human were fixed in formalin and then embedded in paraffin. Skins were cut into 5 μm sections. Immunohistochemistry was carried out according to previous methods (Deng et al., 2019). Skin sections were incubated with primary TLR7 antibody (1:100, Abcam). Pictures were captured from six typical areas in each sample.

Immunofluorescence

Immunofluorescence of mice skin sections was carried out as previously described (Tang et al., 2016). In short, mouse skin samples were fixed in 4% paraformaldehyde (PFA) and frozen in OCT. After being cut into 8 μm, skin sections were washed with phosphate-buffered saline (PBS) three times and then blocked by blocking buffer (5% NDS, 1% BSA, 0.3% Triton X-100) for 1 hour. Primary antibodies (TLR7, 1:100, Abcam; pS6, 1:100, Cell Signaling; p-p65, 1:100, Cell Signaling; CD4, 1:100, Invitrogen) were incubated overnight at 4 °C. After that, sections were washed with PBS and incubated with Alexa Fluor 488- or 594-conjugated secondary antibody (Thermo Fisher Scientific) at room temperature for 1 hour. Later, sections were counterstained with 4′, 6-diamidino-2-phenylindole (DAPI) after wash. All pictures were captured using an ECLIPSE Ni-U Microscope. The fluorescence intensity among six randomly selected microscopic areas (original magnification, 200×) in each sample was measured with ImageJ.

Cell isolation and culture

Primary human keratinocytes were isolated from human foreskin (aged 2–5) and cultured in CnT-07 (CELLnTEC, Bern, Switzerland, USA). HaCaT keratinocyte cell line obtained from NTCC (Biovector Science Lab, Beijing, China) was cultured in DMEM supplemented with 10% fetal bovine serum, penicillin–streptomycin, and 2 mM glutamine (Invitrogen). According to previous study, PBMCs were isolated from the peripheral blood of human through density gradient centrifugation (GE Healthcare) and CD4⁺ T cells were separated by positive selection using Miltenyi beads according to the manufacturer’s instructions (Miltenyi Biotec), then cultured in RPMI 1640 medium (Gibco, Billings, MT, USA) with 10% fetal bovine serum (Biological Industries) (Wu et al., 2018). For R848 treatment, keratinocytes were incubated with rapamycin (50 nM, 2 h), QNZ (1 nM, 12 h) or SC75741 (20 uM, 24 h), and then were treated with R848 (10 ng/ml) for the indicated time. All experiments were performed at least three times.

RT–qPCR

Total RNA was extracted from human skin, mice skin tissues, HaCaT keratinocytes, and primary human keratinocyte cells with TRIzol Reagent (Thermo Fisher Scientific). The quality control of RNA was conducted by a NanoDrop spectrophotometer (ND-2000, Thermo Fisher Scientific). The Maxima H Minus First Strand cDNA Synthesis Kit with dsDNase (Thermo Fisher Scientific) was applied for mRNA reverse-transcription in line with the manufacturer’s instructions. RT-qPCR was carried out on a LightCycler 96 (Roche) thermocycler with iTaqTM Universal SYBR Green Supermix (Bio-Rad). Each gene’s relative expression was calculated by delta-CT relative to GAPDH and the fold change was standardized into the control group.

Enzyme-linked immunosorbent assay

The conditioned media of keratinocytes was collected after 24 h of treatment. The mice skin tissues were homogenized in the complex of RIPA buffer and protease inhibitors (Thermo Fisher Scientific). Then samples were centrifuged to obtain the supernatant at 4 °C. The production of cytokines and chemokines was measured by an enzyme-linked immunosorbent assay (ELISA), using ELISA kits for human IL-1β, human CXCL10, mouse Cxcl1 (Solarbio) and mouse Cxcl2 (Minneapolis), respectively, according to the manufacturer’s instructions.

Immunoblotting

The collected cells were dissolved in the complex of RIPA buffer and protease inhibitors (Thermo Fisher Scientific) after being washed with cold PBS. Centrifugation to obtain the supernatant and then quantifying protein with bicinchoninic acid assay (Thermo Fisher Scientific). The prepared protein samples were separated on SDS–polyacrylamide gel electrophoresis and electroblotting to PVDF membranes. Membranes were blocked in 5% non-fat milk for 1 hour at room temperature, then incubated with primary antibody at 4 °C overnight. After being washed, membranes were probed with HRP-conjugated secondary antibodies (Santa Cruz). The ChemiDocTM XRS+ system (Bio-Rad) was employed to visualize the immunoreactive bands via the HRP substrate (Luminata, Millipore). GE-ImageQuant LAS 4000 mini (GE Healthcare) was employed for data analysis. Quantification of pS6, p-p65 and pAKT was normalized to total S6, p65 and AKT, respectively, while quantification of other proteins was normalized to GAPDH, β-Actin or α-Tublin by densitometry. The primary antibodies are listed as follows: TLR7, 1:1000, Abcam; pS6, 1:2000, Cell Signaling; S6, 1:2000, Cell Signaling; p-p65, 1:2000, Cell Signaling; p65, 1:2000, Cell Signaling; pAKT, 1:1000, Cell Signaling; AKT, 1:1000, Cell Signaling; MYD88, 1:500, Proteintech; p100/p52, 1:2000, Cell Signaling; GAPDH, 1:20000, Proteintech; β-Actin, 1:5000, Santa Cruz; α-Tublin, 1:5000, Abcam

Plasmid and transfection

TLR7 ectopic expression vector was generated using lentivirus vector pLVX-IRES-Neo (Addgene). Primary human keratinocytes were transfected with plasmid using Amaxa^® Human Keratinocyte Nucleofector^® Kit and Nucleofector^® Device (LONZA) following the manufacturer’s instructions, then were planted into six-well plates for additional treatments. HaCaT keratinocytes in six-well plates incubated in serum-free media were transfected with plasmid using FuGENE^®HD Transfection Reagent (Promega) following the manufacturer’s instructions.

Migration assay of T cells

Primary human keratinocytes were plated at 10⁵ cells per well in 24-well plates, and stimulated with R848, QNZ or RAPA according to the indicated dosage and time. 2 × 10⁶ purified CD4⁺ T cells were plated in the upper transwell with 8 μm pores (Falcon) and incubated 6 hours for migration.

Statistical analysis

Statistical analysis was conducted via GraphPad 8.3. Data were represented as the mean ± SEM. Normal distribution and similar variance between groups were examined. Two-tailed unpaired Student’s t-test was for 2 groups’ comparisons while 1-way analysis of variance (ANOVA) with relevant post hoc tests was for the comparison between multiple groups. P value stands for statistical significance (*P < 0.05, **P < 0.01). Data with nonnormal distribution and different variance were analyzed with two-tailed Mann–Whitney U-test. Pearson’s r-test or Spearman’s r-test (for abnormally distributed data) was for correlation analysis.

TLR7 is overexpressed in rosacea

We collected the lesional skin biopsies in the central face of 10 rosacea patients and normal skin biopsies in the similar area of 10 healthy samples (HS), and performed RNA-sequencing to explore the pathogenesis of rosacea (Deng et al., 2021a). After re-analysis these data of the same participants, gene set enrichment analysis (GSEA) showed Toll-like receptor signaling pathway ranked first among the differential KEGG pathways between rosacea and HS (Figs. 1A and 1B). In addition, the fold change of TLR8 and TLR7 ranked as the top two among these TLRs, suggesting that they may take part in the procession of rosacea (Fig. 1C).

In order to figure out whether TLR8 and TLR7 act in the etiology of rosacea, we detected the expression of TLR8 or TLR7 in skin samples of participants and conducted a correlation analysis between its mRNA levels and disease severity (including CEA scores and IGA scores) among 24 patients. Although the fold change of TLR8 is more obvious than TLR7 in RNA-sequencing, there was no difference in the mRNA expression of TLR8 between rosacea patients and HS (Fig. S1A). Meanwhile, TLR8 had no correlation with CEA and IGA scores (Figs. S1B and S1C). Instead, TLR7 significantly increased in rosacea patients and was positively correlated with the IGA scores (indicative of inflammation severity in rosacea) (Figs. 1D– 1F). We further detected the expression of TLR7 in the skin biopsies from rosacea patients and HS through immunohistochemistry (IHC), and found TLR7 was abundantly located in cytoplasm mainly in epidermal keratinocytes of lesional skin from rosacea patients (Fig. 1G).

Furthermore, we established a cathelicidin LL37-induced rosacea-like skin inflammation mouse model based on previous studies, which was similar to the human phenotype (Yamasaki et al., 2007; Schwab et al., 2011; Yamasaki et al., 2011). Mice treated with LL37 presented obvious rosacea-like clinical phenotype and pathological changes as expected (Figs. S1D and S1E). Consistent with our observation in human, the mRNA and protein expression of TLR7 in epidermis was increased remarkably in LL37-injected mice compared with control mice (Figs. 1H and 1I).

These data above suggested that TLR7, rather than TLR8, is overexpressed in the lesional skin of rosacea both in human and mice.

TLR7 deficiency alleviates rosacea development

To determine the function of TLR7 in rosacea development, mice were intradermally administrated with scrambled (Scr) siRNA or Tlr7 siRNA twice before injection with cathelicidin LL37 (Fig. 2A). Immunofluorescence was conducted to confirm the silencing efficacy of TLR7 in the skin in siRNA-administered mice (Fig. S2). Mice results showed that 12 hours after the last time of LL37 injection, LL37-induced mice presented typical rosacea-like skin inflammation in Scr siRNA groups while Tlr7 siRNA groups with LL37 exhibited alleviated rosacea characteristics (Fig. 2B). Similarly, silencing TLR7 reduced the severe redness area and score caused by LL37 (Figs. 2C and 2D). Histological analysis indicated that the infiltration of inflammatory cells in the dermis induced by LL37 was improved in Tlr7 siRNA mice compared with Scr siRNA mice (Figs. 2E and 2F). Meanwhile, RT-qPCR reflected that rosacea-related pro-inflammatory cytokines increased by LL37 were improved when epidermal TLR7 was silenced, for instance, Tnf-α, Il6 and Il-1β (Fig. 2G). Collectively, these results proved that TLR7 deficiency could prevent the development of rosacea, which may put forward a novel treatment strategy.

Stimulation of TLR7 activates NFκB signaling in keratinocytes

Considering that TLR7 is mainly upregulated in epidermal keratinocytes in rosacea, we transfected keratinocytes with ectopic expression vector to overexpress TLR7 to explore the underlying molecular mechanism regulating rosacea formation (Figs. S3A and S3B). The overexpression and hyperactivation of TLR7 increased the expression of its downstream adaptor protein MYD88 (Fig. S3C). We conducted RNA-sequencing from TLR7-overespressing HaCaT keratinocytes with or without R848 (a specific ligand of TLR7) treatment and identified 1,501 differentially expressed genes (DEGs) between two groups (P < 0.05; Fig. S3D). Gene set enrichment analysis (GSEA) exhibited that NFκB signaling pathway ranked at top two (Figs. 3A and 3B).

For the purpose of determining whether TLR7 regulates NFκB, we examined the phosphorylation level of p65/NFκB (p-p65) in TLR7-overexpressing human keratinocytes by immunoblotting, and figured out activation of TLR7 with R848 dramatically increased the activation of NFκB signaling (Fig. 3C). However, the non-canonical NFκB pathway (P100/P52) can’t be stimulated after hyperactivating TLR7 (Fig. S3E). Moreover, NFκB family of transcription factors (including NFKB1, NFKB2, RELA, and RELB) were also upregulated after R848 stimulation (Fig. 3D) (Hiscott et al., 1993; Mori & Prager, 1996). To further verify that TLR7 regulates NFκB signaling in rosacea, we performed immunofluorescence staining on mouse skin lesions and found cathelicidin LL37 increased the expression of p-p65 in epidermis, which was not apparent in Tlr7 siRNA mice (Figs. 3E and 3F).

Taken together, this evidence explained that stimulated TLR7 in keratinocytes activates the NFκB signal pathway in rosacea.

Hyperactivated TLR7 promotes rosacea-characteristic cytokine and chemokine production via NFκB signaling in keratinocytes

Since the NFκB signal pathway is closely linked to the production of inflammatory chemokines and cytokines and is reported to participate in the process of rosacea (Deng et al., 2021a; Vallabhapurapu & Karin, 2009; Taniguchi & Karin, 2018; Li et al., 2022), we detect rosacea-related cytokines and chemokines. According to the results, the expression of pro-inflammatory cytokines (IL-1β, IL-6, IL-10 and IL-12) and chemokines (CCL2, CCL7, CCL20, CXCL1, CXCL2 and CXCL10) raised largely in TLR7-hyperactivated keratinocytes (Fig. 4A). Meanwhile, the protein epxression of IL-1β and CXCL10 in culture supernatants was increased in TLR7-hyperactivated keratinocytes (Fig. 4B). Correspondingly, the abnormal activation of chemokines (Ccl3, Ccl5, Ccl7, Cxcl1, Cxcl2, Cxcl10 and Cxcl11 at mRNA level, as well as Cxcl1 and Cxcl2 at protein level) in mice skin resulted from LL37 was attenuated by Tlr7 siRNA (Figs. 4C and 4D). In consideration of the chemotactic effect of chemokines on immune cells, as well as the enrichment of immune cells in rosacea lesions, we performed CD4 immunostaining and found infiltration of CD4⁺ T cells induced by LL37 was eliminated by Tlr7 siRNA in the dermis of rosacea-like mice (Figs. S4A and S4B). Likewise, the migration of human T cells was also promoted by TLR7-overexpressing human keratinocytes (Figs. S4C and S4E).

Next, we administrated an NFκB inhibitor QNZ (Bottomly et al., 2022) in TLR7-overexpressing human keratinocytes to identify the function of NFκB signaling in these processes (Fig. S4F). Not surprisingly, a number of chemokines (IL-1β, CCL7, CCL20 and CXCL10 at mRNA level, as well as IL-1β and CXCL10 at protein level) induced by R848 was suppressed due to attenuated NFκB signaling (Figs. 4E and 4F). Cell migration assay also demonstrated that the migrating number of human T cells induced by TLR7-overexpressing keratinocytes treated with R848 was repressed by QNZ (Figs. S4G and S4I).

To summarize, excess stimulation of TLR7 regulating rosacea-associated cytokines and chemokines to recruit T cells is dependent on NFκB signaling.

Attenuated TLR7/NFκB signaling regulates rosacea-associated cytokine and chemokine production through inhibition of mTORC1 signaling

The mammalian target of rapamycin (mTOR), as a serine/threonine protein kinase, has been reported to promote skin inflammation in rosacea (Deng et al., 2021a). We detected the expression of the phosphorylated S6, the downstream molecule of mTORC1, in the epidermis of mice, and the result showed LL37 increased the expression of pS6 in Scr siRNA group but not in Tlr7 siRNA groups (Figs. 5A and 5B). Similarly, we discovered the protein level of pS6 was also upregulated after being stimulated with R848 in TLR7-overexpressing keratinocytes (Fig. 5C), but the phosphorylation of AKT, the downstream molecule of mTORC2, remained unchanged after TLR7 activation (Fig. S5A), which confirmed the moderation effect TLR7 had on mTORC1.

Numerous studies have indicated that NFκB and mTOR signals interact with each other (Wang et al., 2017). Consequently, we evaluated the phosphorylation of S6 after inhibiting p-p65 by QNZ in keratinocytes, and found mTORC1 was suppressed by NFκB inhibitor (Fig. 5D). This result was also confirmed by another NFκB inhibitor SC75741 (Fig. S5B) (Reda et al., 2022; Zhao et al., 2022). Curiously, when treating cells with rapamycin (RAPA), a mTORC1-specific inhibitor, the activation of NFκB cannot be repressed, which may explain the unidirectional regulatory effect of NFκB on mTORC1 (Figs. S5C and 5E).

Similar to the result of NFκB inhibition, inhibition of mTORC1 by RAPA downregulated the production of cytokines and chemokines, as well as the migration of human T cells raised by R848 (Fig. 5F, Figs. S5D–S5F).

In summary, hyperactivated TLR7/NFκB signaling may stimulate mTORC1 signaling, and encourage the production of cytokines and chemokines to recruit T cells afterward.