Comparative transcriptome profiling provides new insights into mechanisms of AGA

[TheKingofFighters]

http://onlinelibrary.wiley.com/doi/1...bjd.14767/full

Dear Editor, Androgenetic alopecia (AGA) is the most common progressive form of hair loss, occurring in more than half of men aged > 50 years.[1] AGA onset is dependent on genetic predisposition and presence of androgen.[2, 3] In patients with AGA, hair undergoes a shortened anagen phase, progressive miniaturization and subsequent vellus transformation of the terminal hair follicle.

In this study, we present follicular bulb-centric transcriptome profiling of AGA and non-AGA control scalps. We sought to identify key transcriptomic differences among the hair bulbs of the occipital and vertex regions of 20 patients with AGA (Hamilton grade III–VI) and 10 controls aged 21–57 years (Table S1, Fig. S1; see Supporting Information). These were correlated with dermatoscopic scalp imaging and microscopic evaluation of follicular unit extractions (FUEs) from the leading edge of hair recession at the vertex and occipital regions. Patients had given informed consent and the study protocol was approved by the domain-specific review board of NHG Research.

We found no significant difference in hair density among the vertex and occipital regions in patients (PV and PO, respectively) and controls (CV and CO, respectively) (Fig. 1a, c). The absence of reduction in hair density in patients with AGA is concordant with previous findings where hair thinning was observed.[4] Furthermore, the number of hairs per follicular unit (FU) was reduced in PV (Fig. 1h), which provides support to the hypothesis that a subset of hair follicles within an FU in AGA is more prone to undergoing a reduction in hair shaft size and miniaturization.[5]



The hair shafts in PV (Fig. 1a, d), measured at the protrusion through the epidermis, were significantly thinner in all comparisons, while PO was thinner compared with CO samples. FUE samples from PV showed a well-correlated reduction in hair depth and hair bulb width compared with PO, which was comparable with CV and CO where no significant differences were found (Fig. 1e–g).

We then interrogated the transcriptome profile of the hair bulb-containing lower portion (Fig. 1b) of one vertex and one occipital FUE from each subject by RNA sequencing. We found low numbers of differentially expressed transcripts between anatomical sites (PV vs. CV, PV vs. PO, PO vs. CO; Fig. 2b) due to transcriptome heterogeneity within samples of each anatomical site (Fig. 2a). Genes upregulated in the three comparison pairs were highly enriched in lipid synthesis and electron carrier activity/transport (Fig. 2c–e; Fig. S2 and Table S2; see Supporting Information). Interestingly, the transcriptome profiles of occipital and vertex FUEs of controls revealed a high degree of similarity (Fig. 2b), despite the difference in embryonic origins.[6] We validated five upregulated genes in PV compared with PO (CYB5R3, FADS2, PTGDS, INSR and SRD5A1) with reverse-transcriptase quantitative polymerase chain reaction, and found that the expression trend concurred with RNA-seq observations (Fig. 2h; Table S3).



The transcriptomic heterogeneity observed within samples of each anatomical site (Fig. 2a) has shown that it is insufficient to classify and compare between them. Rather, identification of differentially expressed genes in clustered samples following unbiased transcriptome profile-based sample stratification was a more suitable approach. As such, we conducted principal-components analysis of all samples (all expressed transcripts; fragments per kilobase of transcript per million mapped reads > 0), and found that the majority of samples fell within three clusters (I, II, III; Fig. 2a). A mixture of samples from all anatomical sites was found in cluster I, suggesting that PV samples in cluster I are likely to be less advanced in AGA, as its transcriptome profile is more similar to non-AGA (CV and CO) samples. The distinct cluster II, consisting of five CV and six CO samples, was isolated from all patient samples and thus represented the healthiest hairs. Cluster III contained only morphologically miniaturized PV samples and thus represented hairs with advanced AGA profiles.

We identified 1339 differential transcripts between samples from clusters III and II (Fig. 2b, g), and found upregulation of metabolism (electron carrier activity, respiratory chain and monosaccharide metabolic process), lipid biosynthesis, response to hormone stimulus and steroid hormone biosynthesis-related genes (Fig. 2f; Table S2; see Supporting Information). The upregulation of genes in the respiratory chain (CYB5R3, INSR) may impact on the redox state in AGA-affected hairs.[7, 8]

Furthermore, the upregulation of antioxidation genes (GPX4 and PRDX3) suggests that PV scalps may be exposed to greater oxidative stress than control scalps, possibly resulting from increased respiratory chain activity.[9, 10] Increased levels of GPX4 would also protect the increased amount of lipid synthesized in the PV scalp from phospholipid hydroperoxide-mediated oxidation.[11] The occurrence of oxidative stress in dermal papilla cells may account for impaired hair growth in AGA, as elevated reactive oxidative species in balding dermal papilla cells are known to cause elevated secretion of the hair growth inhibitors, transforming growth factors β1 and β2, and cell senescence.[12, 13] Furthermore, it is proposed that low-level laser therapy alleviates AGA through increasing mitochondrial ATP production, providing support for the involvement of mitochondrial activity in AGA pathogenesis.[14]

We found downregulation of genes related to keratin, epidermis development, cell cycle and hair follicle morphogenesis in cluster III compared with cluster II (Fig. 2b, f; Table S4; see Supporting Information), likely attributed to reduced proliferation and differentiation of matrix cells into keratin-rich hair inner root sheath cells in PV scalps. This is concordant with the significantly decreased hair depth (P = 0·012, Fig. 2i) and hair bulb width (P < 0·001, Fig. 2j) observed in cluster III compared with cluster II, suggesting that downregulation of these genes is a hallmark of advanced AGA.

In conclusion, we present a potential link between altered redox state in the hair follicle and its potential contribution to AGA pathogenesis, which could be targeted in AGA treatments. Further studies to validate the spatial expression of candidate genes in hair follicle compartments is essential to address the contribution of different hair cell types to AGA pathogenesis. Our results emphasize that follicular-based transcriptomes should be compared and analysed with clinical and histological measurements. Evaluation by Hamilton score, while established as a robust clinical measurement of scalp hair pattern, is not designed to classify the state of individual hair follicles in the course of miniaturization and molecular changes in AGA progression. Our data illustrate the importance of evaluating each hair follicle individually, and suggest that future study designs should take these factors into consideration.




CYB5R3 > High ROS in balding scalp
FADS2 > Too much of Fatty acid production. Omega 3 and 6 r products of this gene
PTGDS > Protaglandin d2 synthase. Thus, reafffirming other researcher's findings.
INSR > Insulin receptor. Thus, too much of Insulin in balding scalp.
SRD5A1 > 5-alpha redutase. Thus, DHT is indeed responsible.