1 INTRODUCTION

Frontal fibrosis alopecia (FFA) is a type of primary lymphocytic cicatricial alopecia, characterized by the progressive hair loss of eyebrow and frontotemporal.¹ It mainly occurs in post-menopausal women and has a negative impact on the quality of patients’ life, and even might inducing psychological distress.^{2, 3} Some risk factors were found in recent studies, including sunscreen usage, tobacco consumption, thyroid disorders, and rosacea.^4-6

More and more studies have showed increasing interests in cicatricial alopecia.⁷ In terms of the pathogenesis of FFA, it is still unclear and seldomly reported. E. Del Duca et al. preliminary investigated the biomarkers in scalp biopsies of patients with FFA.⁸ And they emphasized the robust immune activation with CD8+ cytotoxic T cells, increased regulatory T cells, increased interferon-γ aaZ production and JAK/STAT signaling. Similarly, Celina Dubin, BA et al .also studied the scalp and serum profiling of FFA patients and found significant gene marker related to Th1, T cell activation, fibrosis, T regulatory, and Janus kinase.⁹ In addition, major primary cicatricial alopecia variants, including FFA, were found to share dysregulated pathways involving mast cells.¹⁰ However, the mechanism of FFA has not been fully elucidated and needs to be further determined.

Ferroptosis is an iron-dependent and lipid peroxidation-involved programmed cell death type proposed in 2012.¹¹ Distinguished from cell necrosis, apoptosis, and cell autophagy, it is characterized by mitochondrial atrophy, increased mitochondrial membrane density, participation of specific genes, and accumulation of iron and lipid reactive oxygen species.¹² It has been found to have an impact on many diseases, including tumors, neurodegenerative diseases, inflammation, and infection.¹³ In recent years, the role of ferroptosis has been revealed in many skin diseases, such as melanoma, psoriasis, and vitiligo.^14-16 Nevertheless, the relationship between ferroptosis and FFA, even alopecia, is unknown.

Herein, we aimed to analyze the gene expression profile of FFA from clinical scalp lesions through bioinformatics methods and identify key differentially expressed genes (DEGs) and functional modules. Furthermore, we explored the diagnostic performance of ferroptosis-related genes (FRGs) as the accurate biomarkers in FFA and analyzed the immune microenvironment of FFA. Considering that there are currently few studies on the mechanism of FFA, the present study is expected to provide new insight of it and contribute to future research on treatment and prevention strategies.

2 METHODS

3 RESULTS

3.1 DEGs and key modules in WGCNA

We processed the differential expression analysis in FFA, which was showed in the heatmap and volcano plots (Figure 1A and 1B). Finally, a total of 394 upregulated and 242 downregulated DEGs were found. The WGCNA analysis constructed the FFA network and clustered 19 modules (Figure 1C and 1D). And the turquoise and red modules were found significantly related to FFA and selected as the key modules. After the intersection of 636 DEGs with genes in the key modules, 458 key DEGs were identified (Figure 1E).

3.2 Functional enrichment of key DEGs

The GSEA-GO analysis on 458 key DEGs mainly suggested the role of immune response, stimulus response, and lipid metabolic process in FFA (Figure 2A). And in GSEA-KEGG results, these genes were mostly enriched in some autoimmune diseases, antigen processing and presentation, and immune response (Figure 2B). Later, five functional modules were identified by MCODE of Cytoscape (Figure 2C-G). The results of enrichment analysis of functional modules showed that module 1, 2, and 3 were all related to immune response and immune cells; module 4 played a role in fatty acid metabolism; and module 5 was associated with membrane potential and synapse (Figure S1).

3.3 DE-FRGs and validation

We further investigated the ferroptosis-related genes in FFA. We acquired 484 FRGs from FerrDb database and took the intersection of FRGs and key DEGs, indicating 19 DE-FRGs, including SLC7A11, ACSL1, CYP4F8, GCLC, PARP3, PARP9, PARP10, PARP12, PARP14, ALOX5, PLIN2, FADS1, ELOVL5, ALOX15B, CYBB, PTPN6, FADS2, BCAT2, and HILPDA (Figure 3A). Furthermore, we explored the functional enrichment of 19 DE-FRGs. The results of GO analysis on molecular function, cellular component, and biological process and KEGG analysis were showed in Figure 4. The KEGG results pointed out the involving pathways including ferroptosis, fatty acid metabolism, biosynthesis of unsaturated fatty acids, arachidonic acid metabolism, metabolic pathways and PPAR signaling pathway. Subsequently, we verified the gene expression of 19 DE-FRGs in validation dataset GSE186075. And all DE-FRGs were found to be significantly differentially expressed (p < 0.05), except HILPDA was not detected for the different platforms (Figure 3B).

3.4 Diagnostic model and biomarkers

The LASSO algorithm obtained four DE-FRDs as biomarkers, including CYP4F8, GCLC, PARP14, and PARP3 (Figure 5A and 5B). We established a diagnostic model for FFA based on these four DE-FRDs and the CC shows a small difference between the observed and projected probability of FFA (Figure 5C and 5D). Then, the performance of the model in validation dataset GSE186075 were explored in ROC analysis with a good AUC of 0.924 (Figure 5E). Additionally, the ROC analyses of every biomarker in validation dataset were processed and the results were AUC of 0.72 for PARP3, 0.87 for PARP14, 0.88 for GCLC, and 0.74 for CYP4F8 (Figure 5F).

3.5 Immune infiltration of FFA

The immune microenvironment in FFA was explored utilizing the CIBERSORT algorithm. Memory B cells, activated dendritic cells, resting dendritic cells, M1 macrophages, M2 macrophages, monocytes, activated NK cells, CD8+ T cells, follicular helper T cells, and regulatory T cells were found to be significantly different between FFA lesions and controls (Figure 6A). And the correlation analysis of immune cells yielded out a high correlation of CD8+ T cells and M1 macrophages (Figure 6B). Besides, the relationshio analysis of immune cells and DE-FRDs showed a strong correlation between DE-FRDs and helper T cells, CD8+ T cells, M1 macrophages, and activated dendritic cells (p < 0.05) (Figure 6C).

4 DISCUSSION

This study analyzed transcriptome data of 14 clinical skin samples and collected 48 skin samples as validation dataset. After differential expression analysis and WGCNA, a total of 458 key DEGs were identified. They yielded five functional modules, including three immune-related modules, a fatty acid metabolism-related module, and a membrane potential and synapse-related module. Considering the robust immune infiltration of FFA suggested by the results of functional enrichment analysis and the functional modules, we further processed the immune infiltration analysis through CIBERSORT. And the results stressed the role of M1 macrophages and CD8+ T cells in FFA and ferroptosis. This was partly consistent with prior studies. Previous immunohistochemical studies on inflammatory microenvironment of hair follicles in FFA showed a CD8-biased T-cell infiltration.²⁹ And the study of E. Del Duca et al. also suggested a strong CD8+ cytotoxic T cells activation, juxtaposed to follicular stem cells, both in FFA lesions and non-lesional scalp.⁸ Besides, the role of M1 macrophages also was emphasized in the pathogenesis of FFA. FFA is commonly considered a variant of lichen planopilaris (LPP) because of their similar histological features. However, the macrophages polarization seemed to play a vital role in the determination of these two diseases. The M1 macrophage marker CD86 in LPP were found to be downregulated compared with FFA, while M2 marker CD163 were upregulated in LPP.³⁰

In present studies, the key DEGs also showed the involvement in fatty acid metabolism in mechanism of FFA. Genes on cholesterol biosynthesis, fatty acid biosynthesis and metabolism were found to be downregulated in FFA lesions in previous study.¹⁰ The lipid metabolism is highly correlated with ferroptosis and lipid peroxidation.³¹ Besides, oxidative stress has been found in multiple hair loss disorders, including alopecia areata and androgenetic alopecia.^32-34 The free radical-mediated lipid peroxidation could induce the apoptosis of hair follicle cells.³⁵ Notably, reactive oxygen species generation has been considered tightly linked to ferroptosis, as the result of the activation of ferroptosis by excessive reactive oxygen species production.³⁶ Additionally, ferroptosis were found to be associated with the development of many fibrotic diseases, such as renal fibrosis, liver cirrhosis, idiopathic pulmonary fibrosis, and cardiomyopathy.³⁷ Whereas FFA was described as a type of scarring alopecia clinically and histologically and many fibrosis markers were founded to be significantly elevated in FFA.^{9, 38} Based on these clues of the association between FFA and ferroptosis, we further identified potential FRGs in FFA. And finally, 19 DE-FRGs, named SLC7A11, ACSL1, CYP4F8, GCLC, PARP3, PARP9, PARP10, PARP12, PARP14, ALOX5, PLIN2, FADS1, ELOVL5, ALOX15B, CYBB, PTPN6, FADS2, BCAT2, and HILPDA, were selected and validated for differential expression in an external dataset.

Among the 19 DE-FRGs, PARP3, PARP9, PARP10, PARP12, PARP14 are the members of poly ADP-ribose polymerase (PARP) family, which are involved in DNA repair, DNA methylation, transcriptional regulation, and metabolic regulation.³⁹ Previous study has showed the role of PARP in promoting ferroptosis via regulating the expression of SLC7A11 in some diseases.^{40, 41} And many members of PARP family were found to be upregulated in FFA cases compared with healthy controls in this study. This might be a direction for FFA treatment that deserves further investigation. Moreover, we constructed a four-gene-model for FFA diagnosis based on four DE-FRGs through LASSO algorithm. The model achieved the AUC of 0.924 in validation dataset, suggesting its accuration and specificity in distinguishing FFA samples from healthy controls. This also explains to some extent that ferroptosis signaling is a potential pathway in the development of FFA.

The expression of Solute Carrier Family 7 Member 11, SLC7A11, in FFA was downregulated in our study. SLC7A11 is a key gene of ferroptosis, as its downregulation could indirectly inhibit the activity of GPX4 which can lead to the accumulation of lipid peroxides and finally induce ferroptosis in cells.⁴² The role of SLC7A11-GPX4 axis in ferroptosis may function in some inflammatory and immune dermatosis. For instance, in vitiligo, an autoimmune skin disease charactered by the disappearance of melanocytes due to the overactivity of CD8+ T cells, ferroptosis might be triggered by oxidative-related production, including the reduced activity and expression of GPX4.^{43, 44} It was tightly related to oxidative stress and might initiate autoimmunity of vitiligo.⁴⁵ Interestingly, vitiligo seems to be associated with FFA as some studies have reported cases of coexistence of FFA and vitiligo.^46-48 These two diseases are both immune-mediated disorders and the infiltration of most lymphocyte are of CD8+ cytotoxic origin. Besides, they both emphasized the interferon-γ-induced immune collapse and the potential role of oxidative stress. Mechanistically, CD8+ T cells can secrete interferon-γ to downregulate the expression of SLC7A11, thereby promoting ferroptosis induction secondary to GSH depletion.⁴⁹ Therefore, similar pathogenic pathways might explain the coexistence and ferroptosis might also take place in the development of FFA. Besides, although the role of macrophage in ferroptosis is unclear, a previous study indicated that M1 macrophage could enhance the functions of CD8+ T cell via stimulating the secretion of interferon-γ of CD8+ T cell.⁵⁰ This implies that M1 macrophage may play a role in ferroptosis by increasing interferon-γ from CD8+ T cell.

Despite the fact that we have gone as far as possible into the pathogenesis of FFA and preliminarily explored the ferroptosis in the development of FFA, there are still some limitations in this study. First, although the DE-FRGs have been validated in another dataset, further experimental verification of gene expression and functions in FFA is absent due to a lack of clinical cases. Secondly, the ferroptosis sensitivity might differ in multiple cells of the epidermis.⁵⁰ Thus, the specific situation of ferroptosis in different pathological locations of FFA is worthy of further study.

In conclusion, we identified 458 key DEGs and five functional modules through transcriptome data of FFA scalp samples, stressing the involvement of immune and fatty acid metabolism in FFA. And further study suggested that ferroptosis and immune cells, mainly M1 macrophage and CD8+ T cell, might contribute to the development of FFA. This study is expected to improve the understanding of the pathogenesis of FFA and provide potential therapeutic targets for the future.

CONFLICT OF INTEREST STATEMENT

None of the authors have any potential financial conflict of interest related to this manuscript.

ETHICS STATEMENT

No ethical approval is required.