Sterol intermediates of cholesterol biosynthesis inhibit hair growth and trigger an innate immune response in cicatricial alopecia

Abstract

Primary cicatricial alopecia (PCA) is a group of inflammatory hair disorders that cause scarring and permanent hair loss. Previous studies have implicated PPARγ, a transcription factor that integrates lipogenic and inflammatory signals, in the pathogenesis of PCA. However, it is unknown what triggers the inflammatory response in these disorders, whether the inflammation is a primary or secondary event in disease pathogenesis, and whether the inflammatory reaction reflects an autoimmune process. In this paper, we show that the cholesterol biosynthetic pathway is impaired in the skin and hair follicles of PCA patients. Treatment of hair follicle cells with BM15766, a cholesterol biosynthesis inhibitor, or 7-dehydrocholesterol (7-DHC), a sterol precursor, stimulates the expression of pro-inflammatory chemokine genes. Painting of mouse skin with 7-DHC or BM15766 inhibits hair growth, causes follicular plugging and induces the infiltration of inflammatory cells into the interfollicular dermis. Our results demonstrate that cholesterologenic changes within hair follicle cells trigger an innate immune response that is associated with the induction of toll-like receptor (TLR) and interferon (IFN) gene expression, and the recruitment of macrophages that surround the hair follicles and initiate their destruction. These findings reveal a previously unsuspected role for cholesterol precursors in PCA pathogenesis and identify a novel link between sterols and inflammation that may prove transformative in the diagnosis and treatment of these disorders.

Figure 1. Decreased expression of genes related to cholesterol biosynthesis and lipogenesis in PCA.

Principal component analysis of microarray data (downregulated genes) from lymphocytic (1A–1C) and neutrophilic (1D) cicatricial alopecia was performed with the Partek Genomics Suite. The results of the analysis for LPP are shown in 1A, for CCCA in 1B, for FFA in 1C and for TF in 1D. The horizontal axis corresponds to principal component 1 (PC1), the vertical axis corresponds to PC2 and the depth axis corresponds to PC3. The points are colored by group status: blue represents normal samples (pooled), green represents unaffected samples and red represents affected cicatricial alopecia samples. The clustering of data by samples suggests similarities in gene expression profiles. Unaffected and affected samples are clustered together in each subtype, which suggests that the expression profiles of genes involved in cholesterol biosynthesis and lipogenesis are not significantly different among these samples. The normal scalp tissue was significantly different from the unaffected and affected scalp samples from patients with LPP, CCCA, FFA and TF. (E) Heat map of the 39 most significantly downregulated genes in patients with LPP (6 affected and 5 unaffected scalp samples) CCCA, FFA and TF (3 affected and 3 unaffected scalp samples each). The majority of the genes participated in cholesterol biosynthesis. The color bar below indicates the level of expression. (F) Real-time PCR validation of DHCR7 gene expression in normal skin and in the PCA subtypes LPP, CCCA, FFA, TF and DF (*p<0.05, **p<0.01). Compared with normal tissue, DHCR7 expression was significantly decreased in all PCA samples. The unpaired t-test was used for statistical analysis. (G) Real-time PCR validation of EBP gene expression in skin from normal controls and patients with the PCA subtypes LPP, CCCA, FFA, TF and DF (*p<0.05, **p<0.01). EBP expression was significantly decreased in the PCA subtypes TF, FFA and CCCA but not in DC, FD or LPP. The unpaired t-test was used for statistical analysis. See also Figure S1 and Table S1.

Figure 2. Ingenuity Pathways Analysis of the top toxic pathways in cicatricial alopecia.

IPA-Tox®, a data analysis capability tool within the Ingenuity Pathways Analysis, was used to analyze the microarray data and to determine the toxicity associated with the observed gene expression changes in PCA. The figure shows the top toxicity lists (Toxlists) associated with gene expression changes in samples from unaffected and affected scalp areas in patients with LPP, CCCA, FFA and TF. Cholesterol biosynthesis appears to be the most significant toxicity-related pathway associated with the lymphocytic PCA subtypes.

Figure 3. Increased expression of immune and inflammatory genes in PCA.

The principal component analysis results for upregulated genes from lymphocytic (A) LPP, (B) CCCA, (C) FFA and (D) neutrophilic (TF) cicatricial alopecia are shown. The normal and unaffected samples are clustered together in each subtype, which suggests that the expression of immune and inflammatory genes is not significantly different among these samples. Most samples from affected scalp areas in patients with PCA are clustered separately from normal controls and from samples of unaffected scalp skin from PCA patients, which suggest that the expression of these genes differs between affected and unaffected samples. The top canonical pathways from gene expression profiles in patients with (E) LPP, (F) CCCA, (G) FFA and (H) TF are shown. Red represents upregulated and green represents downregulated genes in these pathways. The yellow graph line in E, F, G and H represents –log (p values). (I) Heat map of the most significantly altered immune and inflammatory genes in LPP (6 affected and 5 unaffected samples), CCCA, FFA and TF (3 affected and 3 unaffected samples each) is shown. The color bar below indicates the level of expression.

Figure 4. Innate immune genes are upregulated in PCA.

Real-time PCR validation of (A) TLR4, (B) TLR6, (C) IFNα, (D) IFNα7, (E) NFkB, (F) IFNγ, (G) MMD and (H) MCP1 in mixed (DC, DF), neutrophilic (TF, FD) and lymphocytic (FFA, LPP, CCCA) PCA. These genes are significantly upregulated in affected tissue compared to unaffected tissue from the same patients (*p<0.05, **p<0.01). The unpaired t-test was used for statistical analysis. Differences in the pattern of expression of these genes were observed in the different PCA subtypes. (I) IPA identified the interferon signaling pathway or the “interferon-responsive signature” in the gene expression profiles of LPP. The intensity of the node color red indicates the degree of upregulation, and the intensity of the color green indicates the degree of downregulation. Genes shown as uncolored nodes were not identified as differentially expressed in our experiment and were integrated into the computationally generated networks based on the evidence stored in the IPA knowledge base, which indicated a relevance to this network. The node shapes denote enzymes, phosphatases, kinases, peptidases, transmembrane receptors, cytokines, transporters, translation factors, nuclear receptors and transcription factors. The interferon target genes IRF1, IRF8, IFNA5, IFNAR2, IFIT3, IFITM1, MX1, OAS1 and IFI35 are significantly upregulated in LPP. See also Table S2.

Figure 5. Sterol intermediates of cholesterol biosynthesis trigger an inflammatory response in hair follicle cells.

Inflammatory and immune responses were the top biological functions affected by treatment of HHFORS cells with (A) 7-DHC or (B) BM15766. The most significant biological functions affected, their p values and the number of differentially expressed genes (molecules) after each treatment were identified using IPA. The difference between vehicle and 7-DHC or BM15766 treatments was defined as significant if a 1.5-fold or greater difference in the average hybridization signal intensity with a p<0.05 using a two-tailed unpaired t-test was observed. Predicted interaction networks in hair follicle cells after treatment with (C) 7-DHC and (D) BM15766 are shown. The TLR4 gene network was activated after 7-DHC treatment, and the TLR6 network was activated after BM15766 treatment. Solid lines denote a direct relationship, and dotted lines denote an indirect relationship between two genes in the network. A red node denotes an upregulated gene, and a green node denotes a downregulated gene, with the difference in intensity reflecting the degree of change in the expression of differentially expressed genes in our dataset. The inflammatory functions and disease (fx) associated with each TLR network were determined using IPA. See also Table S3. The real-time PCR validation of (E) TLR4, (F) TLR6, (G) IFNα, (H) IFNα7 and (I) NFkB gene expression in 7DHC- and BM15766-treated hair follicle cells (*p<0.05, **p<0.01) is shown. The unpaired t-test was used for statistical analysis.

Figure 6. Sterol intermediates of cholesterol biosynthesis inhibit hair growth and activate the inflammatory response in C57BL/6J mice.

(A) Topical 7-DHC and (B) BM15766 treatment inhibited hair growth compared with vehicle treatment (ethanol or DMSO). Hair growth was restored in vehicle-treated but not in 7-DHC and BM15766 treated mice. (C, D) Histology of mouse skin treated with ethanol, DMSO, 7-DHC and BM15766 (20X). H&E staining of vehicle-treated mice showed normal hair follicles and sebaceous glands. In contrast, H&E staining of mice treated with 7-DHC and BM15766 showed hyperkeratosis as well as dystrophic hair follicles and sebaceous glands. Scale bar = 50 µm. (E, F) H&E staining showed peri-follicular and inter-follicular inflammation in mouse skin that had been treated with 7-DHC (40×). At higher magnification (100×), infiltration of tissue histiocytes is observed. (G, F) A higher percentage of F4/80 cells was present in mouse skin treated with 7-DHC (Q5-LR) and BM15766 (Q4-LR) than in the vehicle-treated mouse skin. See also Figures S3 and S4.

Figure 7. Inflammatory pathways and networks activated in C57BL/6J mouse skin after topical treatment with 7-DHC and BM15766.

(A) The most significant signaling pathways altered by 7-DHC treatment participated in the inflammatory and immune responses and were identified using IPA. Fisher’s exact test was used to calculate p values to determine the probability that the association between the genes in the dataset and the pathway could be explained by chance alone. The yellow line indicates the threshold of significance (p<0.05) and represents the ratio of the number of molecules from the data set that map to the pathway to the total number of molecules that map to the pathway. (B) The top differentially regulated pathways in BM15766-treated mouse skin. The majority of the upregulated pathways participated in the inflammatory and immune responses. (C, D) The top two predicted networks in 7DHC-treated mouse skin, determined using IPA. The TLR4 and IFN gene networks are significantly upregulated by 7-DHC. Solid lines denote direct relationships between genes. Dotted lines denote an indirect relationship between two genes. A red node denotes an upregulated gene, and a green node denotes a downregulated gene. See also Table S4. Real-time PCR validation of (E) TLR4, (F) TLR6, (G) IFNα, (H) IFNα7, (I) NFkB, (J) IFNγ, (K) MMD and (L) MCP1 gene expression in mouse skin treated with 7-DHC or BM15766 compared with vehicle-treated (ethanol or DMSO) controls (n = 3; *p<0.05, **p<0.01). The unpaired t-test was used for the statistical analysis. Treatment with 7-DHC and BM15766 can induce the expression of some or all of these genes.