Scalp bacterial shift in Alopecia areata

The role of microbial dysbiosis in scalp disease has been recently hypothesized. However, little information is available with regards to the association between microbial population on the scalp and hair diseases related to hair growth. Here we investigated bacterial communities in healthy and Alopecia areata (AA) subjects. The analysis of bacterial distribution at the genus level highlighted an increase of Propionibacterium in AA subjects alongside a general decrease of Staphylococcus. Analysis of log Relative abundance of main bacterial species inhabiting the scalp showed a significant increase of Propionibacterium acnes in AA subjects compared to control ones. AA scalp condition is also associated with a significant decrease of Staphylococcus epidermidis relative abundance. No significant changes were found for Staphylococcus aureus. Therefore, data from sequencing profiling of the bacterial population strongly support a different microbial composition of the different area surrounded hair follicle from the epidermis to hypodermis, highlighting differences between normal and AA affected the scalp. Our results highlight, for the first time, the presence of a microbial shift on the scalp of patients suffering from AA and gives the basis for a larger and more complete study of microbial population involvement in hair disorders.


Introduction
Alopecia areata (AA) is the second most common type of hair loss disorder for human beings. It occurs in the form of a non-scarring alopecia which affects the scalp and, eventually, the entire body [1]. An incidence higher than 2% has been reported for AA, with a lifetime risk of 1.7% both in men and women [2].
For subjects affected by AA, the catagen phase is either extremely short or doesn't occur at all, and in turn proceeds rapidly to telogen phase. From a clinical point of view, this led to single or several annular or patchy bald lesions usually on the scalp [3,4]. These lesions can extend to the entire scalp (Alopecia totalis) or to the entire pilar area of the body (Alopecia universalis).
The management of AA still remains a challenge and is mainly aimed at containing it. Among treatments currently available [5], in 2012, the British Association of Dermatologists recommended two main treatments with a C grade of recommendation: i) topical and intralesional corticosteroid (limited patchy hair loss); ii) immunotherapy (extensive patchy hair loss and Alopecia totalis/universalis) [6]. PLOS  Causes behind AA are not yet fully understood, and there have been debates dating back to the beginning of the 1800s. Many associations have been proposed by researchers over the years [7]. However, clinical evidence and association with other immune disorders [8] underline the role of immunity and inflammation in the early development of AA [9][10][11]. Interestingly, authors [11] reported the efficacy of PRP (Platelet-rich plasma) on AA as a potent antiinflammatory agent by suppressing cytokine release and limiting local tissue inflammation [11].
Other comimon (common?) recognized offenders are hormonal imbalance, psychological stress, genetic tendencies, other local skin disorders and also nutritional deficiencies [5]. More recently, some authors reported evidence of the link between the gut microbiome and AA [12,13] but little information is currently available as regards microbial communities on the scalp [14,15]. Due to its unique features, the scalp is expected to harbor a specific microbiome, which is expected to play a peculiar role in scalp conditions related to hair growth [16].
In this work, we present data on bacterial communities in healthy and AA subjects, on a sample of Italian population. Our results highlight, for the first time, the presence of a significative bacterial disequilibrium on the scalp of AA subjects compared to healthy population; this disequilibrium also extends in the subepidermal compartments of the scalp.

Subjects recruitment
Fifteen healthy and AA subjects, respectively (20-60 years old; 40% male) were recruited from a private Italian dermatological clinic (Milan, Italy).
All subjects have been enrolled under dermatological control. AA subjects have been previously evaluated about their disease history and by means of clinical examinations. Subjects have been enrolled in control population after clinical examinations and in absence of any history of dermatological or scalp disorders.
All enrolled subjects had to meet the following criteria: i) no antibiotics in the 30 days leading up to the sampling; ii) no probiotics in the last 15 days; iii) the last shampoo was performed 48h before sampling; iv) no pregnancy or lactation; v) suffering from other dermatological diseases; vi) no anti-tumor, immunosuppressant or radiation therapy in the last 3 months; vii) no topical or hormonal therapy on the scalp in the last 3 months.
The study was approved by the Ethical Independent Committee for Clinical, not pharmacological investigation in Genoa (Italy) and in accordance with the ethical standards of the 1964 Declaration of Helsinki. All of the volunteers signed the informed consent.

Swab sample collection
The scalp surface has been sampled by means of swab procedure according to previously reported methods [17,18] with minor modifications. Sterile cotton swabs were soaked for at least 30s in ST solution (NaCl 0.15 M and 0.1% Tween 20) before sampling. A comb was used to separate hair fibers and collect samples from a total area of 16cm 2 from a different area of the scalp. After collection, the head of each swab was cut and stored in ST solution. Samples from the same subjects were collected together and stored at 4˚C until DNA extraction. Sterile cotton swabs placed in ST solution have been used as negative controls.

Biopsy samples collection
A total of 4 female subjects (two control and two AA, respectively) were also sampled for the microbial community in the subepidermal compartments of the scalp. A 4-mm punch biopsy specimen was collected from each subject. In AA subjects, the specimen was obtained from a well-developed lesion. The sampled area was disinfected prior to the surgery to avoid contamination from surface bacteria. Epidermis, dermis and hypodermis were aseptically separated and stored in Allprotect medium (Qiagen) according to manufacturer conditions until DNA extraction.

Bacterial DNA extraction
Bacterial DNA from scalp swabs was extracted by mean of QIAamp UCP Pathogen Mini Kit (Qiagen, Milan, Italy) according to manufacturer protocol, with minor modifications [19]. The DNeasy Tissue kit (Qiagen, Milan, Italy) was used for DNA extraction from biopsy specimens. Extracted DNA was finally suspended in DNAse free water and quantified by the QIAexpert system (Qiagen, Milan, Italy) before qRT-PCR and sequencing.
High throughput 16S amplicon generation, sequencing and analysis DNA samples extracted from scalp swabs were amplified for the variable region V3-V4 using the universal prokaryotic primers: 341 F CTGNCAGCMGCCGCGGTAA [20,21] and 806bR GGACTA CNVGGGTWTCTAAT [22][23][24] utilizing a modified dual-indexed adapter-linked single step protocol. Library preparation and Illumina MiSeq V3-V4 sequencing were carried out at StarSEQ GmbH, Mainz, Germany, according to the method of Caporaso et al. [25] and Kozich et al., [26] with minor modifications. Amplicons were generated using a high fidelity polymerase (AccuStart II PCR ToughMix, Quantabio, Beverly, MA). The amplicons were then normalized to equimolar concentrations using SequalPrep Plate Normalization Kit (ThermoFisher Scientific, Monza, Italy) and the final concentration of the library was determined using a fluorometric kit (Qubit, Life technologies, Carlsbard, CA, USA). Libraries were mixed with Illumina-generated PhiX control libraries and denatured using fresh NaOH. Runs were performed using Real-Time Analysis software (RTA) v. 1.16.18 and 1.17.22, MiSeq Control Software (MCS) v. 2.0.5 and 2.1.13, varying amounts of a PhiX genomic library control, and varying cluster densities. Four sequencing runs were performed with RTA v. 1.18.54, MCS v. 2.6, a target of 25% PhiX, and 600-700 k/mm2 cluster densities according to Illumina specifications for sequencing of low diversity libraries. We used 25% PhiX to balance the runs and use 600 bp V3 chemistry for sequencing. Basecalls from Illumina High Throughput Sequencing (HTS) machines were converted to fastQ files using bcl2fastq (Illumina) software, v2.20.0.42 and quality control carried out by mean of, v0.11.5. bcl2fastq (Illumina) software, v2.20.0.422. Quality control of fastq reads was carried out using FastQC v0.11.5. The quality trimming of primers and adaptors was carried out using Cutadapt, v. 1.14 [27] and Sickle v. 1.33 [28] toolkits, respectively.
Paired-end reads were assembled using Pandaseq v. 2.11 [29] using a threshold of 0.9 and a minimum overlap region length of 50. Clustering was carried out using closed-reference OTU picking and de novo OUT picking protocol of QIIME v1.9 [25] at �97% identity.
Greengenes database v13_8 was used as a reference for bacterial taxonomic assignment [30]. Amplicon reads were also analyzed as regards alpha diversity by mean of Shannon index, using QIIME v1.9.

Bacteria quantification by qRT-PCR
Relative abundance of bacterial DNA of main bacterial species on the scalp was assessed by means of real-time quantitative PCR (RT qPCR). Microbial PCR assay kit (Qiagen, Milan, Italy) with gene-specific primers and TaqMan MGB probe targeting Propionibacterium acnes, Staphylococcus epidermidis and Staphylococcus aureus 16S rRNA gene, respectively, were used. Genbank accession numbers of 16S rRNA gene sequences for P. acnes, S. aureus and S. epidermidis were ADJL01000005.1, ACOT01000039.1 and ACJC01000191.1, respectively. Samples were mixed with 12.5μL of Microbial qPCR Mastermix, 1 μL of Microbial DNA qPCR Assay, 5ng of genomic DNA sample and Microbial-DNA-free water up to a final volume of 25 μL.
Nine separate PCR reactions are prepared for each sample, including Positive PCR Control, No Template Control, and Microbial DNA Positive Control, as well as the Microbial DNA qPCR Assay. Pan-bacteria (Genebank accession number HQ640630.1) assays that detect a broad range of bacterial species are included to serve as positive controls for the presence of bacterial DNA. Assays for human GAPDH and HBB1 (Genebank accession numbers NT_009759.16 and NT_009237.18, respectively) have been included to determine proper sample collection and used to assess the presence of human genomic DNA in the sample and, eventually, subtracted from calculation. Thermal cycling conditions used were as follows; 95˚C for 10 min, 40 cycles of 95˚C for 15 sec, 60˚C for 2 min. PCR reactions were performed in duplicate using an MX3000p PCR machine (Stratagene, La Jolla, CA). Amplification-curve plotting and calculation of cycle threshold (Ct) values were performed using MX3000p software (v.3; Stratagene) and data were further processed by Excel. ΔΔCt method [31] was used to calculate bacterial load of each swab sample. Obtained values have been used for calculation of Bacterial Load-Fold Change (AA/Healthy subjects). Data is finally expressed as Log of the relative abundance of each sample versus the control group.

Statistical analysis
Data is expressed as log Relative abundance (RA) ± SEM for qRT-PCR analysis. Results were checked for normal distribution using D'Agostino & Pearson normality test before further analyses. Statistically significant differences on bacterial community between healthy and AA group were determined using Wilcoxon test (p � 0.05). All the comparisons were performed pairwise for each group. Analyses were performed with GraphPad Prism 7.0 (GraphPad Software, Inc., San Diego, CA). P-values equal to or less than 0.05 were considered significant.

Microbiota profiling of the scalp in AA subjects
The human scalp's bacterial composition of Control (n = 15) and AA (N = 15) subjects have been analyzed by IlluminaSeq (Fig 1). We obtaining about 585,219 and 544,578 high quality reads for the total V3-V4 sequences from control and AA subjects, respectively. About 56.3% of sequences from the control group were assigned to Actinobacteria phylum and 35.2% to Firmicutes. As regards, AA group Actinobacteria were around 57.4% and Firmicutes decreased to 29.2%. The analysis of bacterial distribution at the genus level, interestingly, highlighted an increase of Propionibacterium from 45.6% to 55.1% in AA subjects. Alongside data showed a general decrease of Staphylococcus from 32.6% to 27.4% (Fig 1A). Therefore, the percentage of other less abundant bacteria genus was similar (around 5%) both in control and AA subjects. Alpha-diversity (Shannon diversity index) was significantly higher (p � 0.001) in AA subjects than in the control group (Fig 1B).

Microbial shift of the scalp surface in AA subjects
As previously reported by other authors [14,15], P. acnes, S. epidermidis and S. aureus are the three major microbial species found on the scalp.
Relative abundance of predominant bacteria on scalps both of control and AA subjects has been analyzed by mean of RT q-PCR. Primers and TaqMan MGB probe specific for 16S region of P. acnes, S. epidermidis and S. aureus were used.
Pan bacteria specific targets designed to detect the broadest possible collection of bacteria involved in human biology were used as control. Student's test analysis of log Relative abundance comparing control and AA subjects showed a significant (p<0.01) increase of P. acnes (from 1.6 to 1.8 log RA) in AA subjects compared to control ones (Fig 2A). AA scalp condition is also associated with a significant (p<0.05) decrease of S. epidermidis relative abundance (from 1.4 to 1.01 log RA) (Fig 2B) while no significant changes were found for S. aureus (Fig 2C).

AA alteration of bacterial distribution in the subepidermal compartments of the scalp
Two bioptic samples were collected respectively from control and AA subjects and divided in the main subepidermal compartments. Extracted genomic DNAs were analyzed by Illumina-Seq and analyzed for bacterial distribution.
Similar proportions of Firmicutes (24.6% vs 27.6%) and Proteobacteria (16.2% vs 16.9%) were reported in epidermis of both control and AA subjects (Fig 3) while a higher proportion of Actinobacteria (33.3% vs 22.4%) and Bacteroidetes (20.1% vs 9.9%) were found in AA subjects compared to control (Fig 3). Bacterial community in dermis shifted to a lower proportion of Actinobacteria (6.1% vs 11.3%) in AA subjects while Proteobacteria (14.9% vs 8.1%) and Bacteroidetes (14.2% vs 4.0%) increased compared to control (Fig 3). Also hypodermis showed a peculiar bacterial distribution which results, also in this case affected by scalp condition. AA subjects showed a significative higher proportion of Proteobacteria, Bacteroidetes and especially Firmicutes than control subjects (Fig 3). In general less variability was observed for bacterial communities in AA subjects and this may reflect in a compromised healthiness of the scalp.
Most interesting, the analysis at species level of bioptic samples highlighted the presence of Prevotella copri in both AA samples, in all analyzed compartments.
Akkermansia muciniphila was also found (less than 1.5% of total population) in AA subcompartments of the scalp, in particular in the hypodermis.

Discussion
In this study, we reported, for the first time, the relationship between microbial shift on the scalp and hair growth disorder, in particular, Alopecia areata. We conducted analysis by mean of qRT-PCR and 16S sequencing.
A diversified and abundant microbial community host the skin [32] and this symbiotic relationship results, most of the time, as beneficial for both the host and microbial community [33][34][35]. Bacteria mainly belong to Corynebacteriaceae, Propionibacteriae, and Staphylococcaceae [36][37][38][39] and are differently distributed according to the physiochemical properties of each skin site they host [39,40]. Many scientific published evidence reported the strict correlation between microbial disequilibrium and skin conditions [41][42][43][44][45]. Little is still reported with regards to the microbiome inhabiting the scalp and hair growth disorders [14,15,46]. Clavaud and collaborators [15] and, more recently, Xu et al. [14] reported, the implication of microorganisms in the development of dandruff. Characterization of scalp bacterial species involved in hair disorders such as Alopecia androgenetica, Alopecia areata, and Lichen Planopilaris has been poorly investigated and, only recently, the piece bit of evidence has been reported [16].
We focused our attention on bacterial population of the scalp of healthy and AA subjects looking at main bacterial species on the scalp [15] (P. acnes, S. aureus, and S. epidermidis) and at their reciprocal balancing. We quantified their relative abundance by means of accurate gene-specific primers and probe targeting 16S region, by RT qPCR. Our results are concurrent with Wang's work [46] highlighting the reciprocal inhibition exerted by bacteria, each other, on the scalp (Propionibacterium vs Staphylococcus and vice-versa). AA subjects showed an increase in P. acnes and a decrease of Staphylococcus, especially S. epidermidis, suggesting the role of Propionibacterium/Staphylococcus balancing in AA. A role of P. acnes with hair casts and Alopecia has previously been hypothesized by Wang and collaborators [46] even though not deeply investigated. P. acnes is able to synthesize many enzymes involved in the metabolism of porphyrins that, once activated, may contribute to oxidation and follicular inflammation. Therefore, a speculation about the role of the hypoxic condition of the follicular region may be speculated in AA and this may encourage P. acnes overgrowth. A role of hypoxia has been reported in the progression of other skin condition such as psoriasis [47] and atopic dermatitis [48]. The presence of A. muciniphila, a strictly anaerobic bacteria, around the hair follicle in analyzed AA subjects may be suggestive of a hypoxic ecosystem in which this bacteria can find favorable growth conditions. Data from IlluniaSeq profiling also suggested a higher diversity of bacterial species inhabiting the scalp of AA subjects. These results are in line with previous work [15] on other scalp conditions. On the basis of the present and previous results, a link with a higher susceptibility of an unhealthy scalp to be colonized by microorganisms could be postulated but further analysis are needed to understanding the reason behind this high variety.
Beyond the superficial relationship between the microorganism with skin, microbes can also communicate with cells of the subepidermal compartments [49] and are involved also in deep immunological response [50][51][52][53][54]. As reported by Nakatsuji et al., [49] high interpersonal variability was observed as regards epidermal and subepidermal microbial population. In this study, data from sequencing profiling of the bacterial population strongly support a different microbial composition of different area surrounded hair follicle from the epidermis to hypodermis, highlighting differences between normal and AA affected scalp. We can hypothesize the role of this different microbial composition in AA symptoms and manifestations.
Microbial changing at different subepidermal compartment may be linked to an autoimmune component of the pathology as to skin barrier skin disruption, as previously shown for other skin disorders [55].
Most interesting, the analysis at species level of bioptic samples highlighted the peculiar presence of P. copri and A. muciniphila in both AA samples, in all analyzed compartments. These findings are very intriguing. The finding of Prevotella copri as one of the most abundant bacteria in subepidermal compartments of AA scalp may be linked to the autoimmune component of this hair condition. For example, P. copri has been found as relevant in the pathogenesis of rheumatoid arthritis [56], another chronic inflammatory autoimmune disorder that can affect other parts of the body including the skin. Therefore the identification of A. muciniphila in the subepidermal compartments of the scalp of AA subjects could open to new therapeutic approaches in the management of AA. The link between A. muciniphila and skin disease has been yet discussed as it has been considered a gut signature of psoriasis [57].
The present work reported data from an initial pilot study. Future studies should be aimed at better investigate both the role of microbial community shifts and hypoxia in hair scalp diseases. Also the study of additional factors such as inclusion of samples from non-lesional sites in AA and non-AA subjects and from other baldness disease besides AA and the role should be considered.

Conclusions
Our study highlighted, for the first time, the presence of a microbial shift on the scalp of patients suffering from AA and gives the basis for a larger and more complete study of microbial population involvement in hair disorders. Therefore, the reported findings as the availability of sophisticated and quick methods to evaluate the microbial composition of the scalp open to new therapeutic approaches in the management of hair disorders.
Larger studies are still needed for a more precise identification of bacterial community on the scalp as for the analysis of fungal component in AA subjects but the results of the present work permit to asses, for the first time, the involvement of microbial changing in hair disorder, in particular AA, also in the subepidermal compartments of the scalp.