Introduction

Human scalp is covered with high density of hair growing from hair follicles with connection to sebaceous glands, creating a lipid-rich hydrophobic niche that harbors a variety of microorganisms [1]. Cutaneous microorganism of fungi and bacteria are strongly associated with scalp disorders and diseases, including dandruff, seborrheic dermatitis (SD), scalp psoriasis, folliculitis decalvans, and alopecia [2]. Fungal Malassezia is one of the etiologic factors in the progression of dandruff [3] and the source of oxidative damage that leads to premature hair loss [4] and plays a role in the pathogenesis of alopecia [5]. Bacteria Cutibacterium and Staphylococcus also contribute to scalp inflammation, hair-follicle associated disorders, and hair diseases [6]. Metagenomic analysis showed that Malassezia, Cutibacterium, and Staphylococcus are the main fungal and core bacterial genera found on healthy and dandruff scalps, and the species of M. restricta, M. globosa, C. acnes, and S. epidermidis are further identified [7,8,9]. Furthermore M. restricta and M. globosa display contrary roles in scalp health, and C. acnes and S. epidermidis are reciprocally inhibited bacteria on the scalp [7, 8]. The role of scalp microbiota diversity in causing scalp disorders and diseases has been highlighted and investigated. For example, the diversity of microbial composition in hair follicle contributes to the pro-inflammatory environment in chronic inflammatory scalp disease [10]. High bacterial diversity was detected in the lesions of patients with psoriasis and SD compared with that in the healthy controls [11]. The intraspecific diversity of Malassezia from the scalp microbiota was found in patients with SD [12]. Scalp microorganisms can also affect the scalp situation through microbial cell–host interaction. The fungal microbiome is majorly implicated in cell–host adhesion in dandruff scalp, and the bacterial microbiome is related to the synthesis of biotin, vitamins, and other nutrients in healthy scalp [8].

Several bio-oils and plant extracts, including anti-dandruff and hair growth stimulating agents, are used as scalp-related products to maintain scalp health and hygiene [13,14,15]. Recent evidence has shown the beneficial effects of probiotics for skin health, such as maintaining barrier integrity, tissue homeostasis, and microbiome balance and inhibiting pathogen growth, biofilm formation, and local inflammation [1, 16, 17]. The beneficial effects of probiotic intake on modulating cell cycles in hair follicles and promoting hair growth have also been reported [18, 19]. Live probiotics and the compounds from bacteria or their metabolites play important role in skin care. For example, the fermentation metabolites of commensal S. lugdunensis could suppress fungi overgrowth, providing novel therapeutics for C. parapsilosis-associated infection in human dandruff [20]. The glycolipopeptide of Lactiplantibacillus pentosus and Lacticaseibacillus paracasei has anti-adhesion and anti-microbial activity against pathogen in the skin microflora [21]. Our previous studies demonstrated that heat-killed Lactobacilli preparations from L. plantarum GMNL-6 and L. paracasei GMNL-653 promote skin wound healing and preventing excessive fibrosis by suppressing TGF-β/pSmad signaling [22]. Lipoteichoic acid, the major cell wall component of GMNL-6, has a beneficial effect on skin care [23]. Heat-killed GMNL-653 also exhibits anti-osteoporotic effects by regulating the gut microbiota dysbiosis [24]. Given the safety concern of live probiotics in immunosuppressive populations, the preparations of inanimate microorganisms or their components have potential benefits and l applications for skin care [25]. In the present study, we investigated the potential role of heat-killed L. paracasei GMNL-653 in human scalp health using in vitro cell lines and recruiting human participants.

Results

Heat-killed GMNL-653 has beneficial effects on co-aggregation with M. furfur and inhibitory effects on the biofilm formation of M. furfur

We aimed to clarify whether heat-killed probiotics GMNL-653 has effects on scalp pathological microbes in vitro. First, we evaluated the ability of heat-killed GMNL-653 to co-aggregate with fungus M. furfur, a common pathological inhabitant on human scalp. We found that when mixed in a test tube, heat-killed GMNL-653 aggregated with M. furfur (Fig. 1A) and increased following times (Fig. 1B). The co-aggregation of heat-killed GMNL-653 and M. furfur was also confirmed by SEM (Fig. 1C). Furthermore, we evaluated whether GMNL-653-derived LTA, the major cell wall component, influence the biofilm formation of M. furfur on human scalp. The results showed that GMNL-653-derived LTA inhibited the biofilm formation of M. furfur on Hs86 cells in a dose-dependent manner (Fig. 1D). Owing to its capabilities of co-aggregating with M. furfur and reducing the biofilm formation of M. furfur, heat-killed GMNL-653 potentially has beneficial effects on scalp health.

Fig. 1
figure 1

Heat-killed GMNL-653 causes co-aggregation with M. furfur and inhibits its adhesion to Hs86 cells. A, B Heat-killed GMNL-653 (2 × 109 cells/ml), live M. furfur (2 × 109 cfu/ml), or mixtures of heat-killed GMNL-653 and live M. furfur with the ratio of 1:1 were added into tubes and stand at room temperature for 40 min to observe the formation of precipitates (A). Liquids of the suspended area from the tubes of M. furfur alone, GMNL-653 alone, and M. furfur-GMNL mixture were collected after mixing for 0, 20, and 40 min. The liquid was detected by the absorbance at a wavelength of 590 nm (B). The aggregation ability was quantified by the formula described in the Materials and Methods section. C The aggregation of GMNL-653 and M. furfur from (A) was visualized by SEM. D M. furfur were seeded into wells of a 96-well-plate with or without adding 25, 50 µg/ml GMNL-653 derived LTA for 24 h. The biofilm was visualized after staining with 0.1% crystal violet, dissolved with DMSO, and quantified by the absorbance at a wavelength of 590 nm. *, p < 0.05

Full size image

Heat-killed GMNL-653 up-regulates the expression of growth factors in skin cell lines

Next, we examined the effect of heat-killed GMNL-653 on the expression of growth factors in human skin cells. The mRNA expression of growth factor genes in Hs68 cells and HaCaT cells, human immortal keratinocytes, was determined. The results showed that the mRNA expression of IGF-1R (Fig. 2A), VEGF (Fig. 2B), and KGF (Fig. 2D) was up-regulated in Hs68 cells after treatment with heat-killed GMNL-653 in a dose-dependent manner. IGF expression was also up-regulated, though the difference was not statistically significant (Fig. 2C). We also observed that heat-killed GMNL-653 up-regulated the mRNA expression of IGF-1R (Fig. 3A), VEGF (Fig. 3B), and IGF (Fig. 3C) in HaCaT keratinocytes. Basing on the up-regulation of these growth factors, we suggested that heat-killed GMNL-653 could improve scalp hair growth.

Fig. 2
figure 2

Treatment of heat-killed GMNL-653 in Hs68 fibroblasts increases the mRNA expressions of growth factors. 1.5 × 105 cells/well of human fibroblast Hs68 cells were seeded in wells of 6-well-plate for attachment and then cultured with serum free medium for 24 h. Cells were treated with indicated concentration of heat-killed GMNL-653 for 24 h. The mRNA expression of IGF-1R (A), VEGF (B), IGF-1 (C), and KGF (D) were determined by quantified RT-PCR (n = 5). Data were presented as the relative fold changes (mean ± SEM) in compared to non-GMNL-653 treatment control (Ctrl) after normalization with the house-kee** of β-actin. *p < 0.05

Full size image
Fig. 3
figure 3

Heat-killed GMNL-653 increases the mRNA expressions of growth factors in HaCaT keratinocytes. Human epidermal keratinocyte HaCaT cells were seeded in 6-well-plate as a cell density of 3 × 105 cells/well and cultured with serum free medium after attachment for 24 h. Cells were treated with the indicated concentrations of heat-killed GMNL-653 for further 24 h. The mRNA expressions of IGF-1R (A), VEGF (B), and IGF-1 (C) were determined by qRT-PCR. Data were presented as the relative fold changes (mean ± SEM) in compared to non-GMNL-653 treatment control (Ctrl) after normalization with the house-kee** of β-actin. *p < 0.05

Full size image

Beneficial effects of heat-killed GMNL-653 on human scalp in a clinical trial

Owing to the ability of heat-killed GMNL-653 to suppress M. furfur biofilm formation and induce growth factors in skin cells, we examined the beneficial effects of heat-killed GMNL-653-containing shampoo on scalp conditions by conducting a.clinical trial as illustrated in Fig. 4. Twenty-two participants with age ranging 30–45 years and including 8 males and 14 females were recruited. At the first visit, all participants underwent baseline examinations, including oil counts, hair volume, dandruff, and microbiota determination. The baseline of scalp conditions among enrolled participants is summarized in Table 1. At the beginning, all participants were asked to wash their hair by using control shampoo without heat-killed GMNL-653 for 1 month and then change to the tested shampoo with heat-killed GMNL-653 for hair washing for 4 months. Data on oil count, hair volume, dandruff and scalp microbiota were collected from participant scalp according to the timing from the beginning (0M) to the 5th month (5M) as shown in Fig. 4A. The value of oil count was showed by the sum of the three regions (front, middle and back scalp) in each participant, and the value of hair volume was the average from the three regions (Fig. 4B). The samples of dandruff and microbiota were collected from the whole scalp by tape and wet swab, respectively (Fig. 4C), using the methods described in materials and methods.

Fig. 4
figure 4

The design of a clinical trial with the application of heat-killed GMNL-653-containing shampoo and examinations of scalp conditions among recruited participants. A All participants were requested to use the shampoo without heat-killed GMNL-653 (control shampoo) for the first month. Then, heat-killed GMNL-653-containing shampoo was used for the following 4 month. The collection time points of scalp conditions and scalp microbiota data were indicated by tick. B Data of oil count and hair volume were collected from front, middle, and back of scalp using Sebumeter 815 and Aram TSII. C Data of dandruff were collected from whole scalp by dandruff tapes and analyzed by Image J. Scalp samples for microbiota analysis were collected from whole scalp by wet cotton swabs containing 1 ml PBS buffer with 0.1% Triton X

Full size image
Table 1 Baseline scalp conditions of enrolled participants
Full size table

We next performed subgroup analysis according to several scalp conditions, namely, oil situation, dandruff formation, or hair volume. Among the participants, two cases with missing data during the collection time-points were removed. Relative to that at the beginning (0M), the oil count of scalps of 20 subjects after using control shampoo for 1 month (Ctrl; 1M) was not statistical different but was significantly decreased after using heat-killed GMNL-653-containing shampoo for the following 1–4 months (GMNL-653; 2M, 3M, and 5M) (Fig. 5A). Considering that the recruited volunteers initially had various scalp situation, we further classified them into two subgroups of high oil and normal oil depending on their oil situation at the beginning (Start, 0M). We found that the oil count was significantly decreased only in the subgroup of high oil (Fig. 5B), but no statistical difference was observed in the normal oil subgroup (Fig. 5C) after the use of heat-killed GMNL-653-containing shampoo for 1–4 months (GMNL-653; 2M, 3M, and 5M). Among the participants, four cases with missing data during the collection time-points were removed. Relative to that at the beginning (0M), the dandruff formation of 18 subjects statistically increased when they used control shampoo for 1 month (Ctrl; 1M) but significantly decreased when they changed to heat-killed GMNL-653-containing shampoo for 0.5 month (GMNL-653; 1.5M) when compared to the situation at 1 month using control shampoo (Ctrl; 1M) (Fig. 6A). In the subgroup comparisons, dandruff formation was significantly reduced only in the subgroup with high dandruff (Fig. 6B) but not in the subgroup with normal dandruff (Fig. 6C) after the use of heat-killed GMNL-653 shampoo for 0.5 month (GMNL-653; 1.5M) compared with that at the beginning (0M) and using control shampoo for 1 month (Ctrl; 1M). For hair volume analysis, only 19 subjects were analyzed after removing 3 cases with missing data during the collection time-points. The hair volume statistically increased after using heat-killed GMNL-653 shampoo for 2 and 4 months (GMNL-653; 3M, 5M) compared with that at the beginning (0M) among all subjects (Fig. 7A). In the subgroup analysis, the increase in hair level with time caused by using heat-killed GMNL-653 shampoo was observed only in the subgroup with low hair abundance (Fig. 7B) but not in the subgroup of normal hair abundance (Fig. 7C).

Fig. 5
figure 5

The sebum secretion is decreased in human scalp after using heat-killed probiotics GMNL-653-containing shampoo. Oil counts were collected from all participants scalp at the beginning (0 M), after using control shampoo (Ctrl) for 1 month (1 M), and followed using GMNL-653 containing shampoo (GMNL-653) for 1, 2, and 4 months (2 M, 3 M, and 5 M). Oil counts of the front, middle, and back regions of scalp were detected from each participant and calculated the sum values. A Oil counts of all participants (n = 20) were shown. B Oil counts of 7 participants scalp with high oil condition (> 400 µg/cm2) in the beginning were shown. C Oil counts from 13 participants with normal oil condition in the beginning were shown.* p < 0.05 compared with 0 M group. # p < 0.05 compared with the Ctrl group

Full size image
Fig. 6
figure 6

Dandruff formation is decreased in human scalp after using heat-killed GMNL-653-containing shampoo. Dandruff from participants were collected using dandruff tapes in the beginning (0M), after using control shampoo for 0.5 and 1 month (0.5M and 1M), and followed using heat-killed GMNL-653 shampoo for 0.5, 1, and 2 months (1.5M, 2M, 3M). A Dandruff level from all participant (n = 18) were shown. B Dandruff level of 8 participants with high dandruff condition (> 0.1%) at the beginning were shown. C Dandruff level of 10 participants with normal condition at the beginning were shown. * p < 0.05 compared with 0 M group. # p < 0.05 compared with the Ctrl group

Full size image
Fig. 7
figure 7

The hair volume is increased in human scalp after using heat-killed GMNL-653-containing shampoo. Hair level from participants were collected at the beginning (0M), after using control shampoo for 1 month (1M), and followed using GMNL-653 containing shampoo for 2 and 4 months (3M and 5M). The front, middle, and back regions of scalp in each participant were detected by Aram TSII. Hair volumes were presented as the sum of value of front, middle, and back scalps from each participant. A Hair volumes from all participant (n = 19) were shown. B Hair volumes of the 10 participants with less hair condition (< 125 hairs/cm2) at the beginning were shown. C Hair volumes of the 9 participants with normal hair condition (> 125 hairs/cm2) at the beginning were shown. * p < 0.05 compared with 0 M group. # p < 0.05 compared with the Ctrl group

Full size image

Changes of microbiota in human scalp after using heat-killed GMNL-653-containing shampoo

We focused on the two bacteria, C. acnes and S. epidermidis, and two fungi, M. restricta and M. globosa, to analyze the scalp microbial level of participants by quantitative PCR before or after using control and heat-killed GMNL-653 shampoo. First, we confirmed that the level of L. paracasei from the scalp samples of all participants significantly increased after using heat-killed GMNL-653 shampoo relative to that at the beginning (Start) and after using the control shampoo (Ctrl) using (Fig. 8A). We further observed that the abundance of C. acnes statistically decreased (Fig. 8B) and that of M. globosa statistically increased (Fig. 8D) after using heat-killed GMNL-653 shampoo relative to that at the beginning (Start). However, the level of S. epidermidis (Fig. 8C) and M. restricta (Fig. 8E) among all scalp samples did not show statistical differences. The decreasing level of C. acnes (Fig. 8B) and the increasing level of M. globosa (Fig. 8D) were observed in the group using control shampoo (Ctrl) relative to that at the beginning (Start), indicating that the control shampoo had the same effect as heat-killed GMNL-653 shampoo on the abundance of C. acnes or M. globose. We further analyzed the microbial abundance relative to that at the beginning scalp condition after classifying the participants into weak and normal subgroups as described above. In the three kinds of scalp situations (dandruff, oil secretion, and hair volume), the increased accumulation of L. paracasei was observed in the three normal subgroups (Supplementary Figs. 1A, C, and E) and in the subgroup with less hair (Supplementary Fig. 1F) after the use of heat-killed GMNL-653 shampoo. The abundance of M. restricta was not different among all subjects (Fig. 8E) but statistically decreased in the subgroup of high dandruff after the use of heat-killed GMNL-653 shampoo (Supplementary Fig. 1B). In addition, a decrease in C. acnes in the high dandruff subgroup and normal subgroups of oil and hair (Supplementary Figs. 1B, C, and E) and an increase in M. globosa in the normal subgroups of dandruff, oil, and hair (Supplementary Figs. 1A, C, and E) were observed regardless of using control or heat-killed GMNL-653 shampoo. However, the abundance of C. acnes in the subgroups of normal dandruff, high oil, and less hair (Supplementary Figs. 1A, 1D, and 1F) and the abundance of M. globosa in the three weak subgroups of high dandruff, high oil, and less hair (Supplementary Figs. 1B, D, and F) showed no statistical difference after the use of heat-killed GMNL-653 shampoo. These results suggest that the beneficial effects of heat-killed GMNL-653 shampoo on human scalp health may be related to the changes of scalp microbiota.

Fig. 8
figure 8

The abundance of fungal and bacterial microbiota on human scalp after using heat-killed GMNL-653-containing shampoo. Microbiota samples of each participant were collected from the whole scalp by wet cotton swabs with 0.1% Triton X /PBS buffer at the beginning (Start), after using control shampoo for 1 month (Ctrl), and followed using GMNL-653 shampoo for 1 months (GMNL-653). The liquid samples in cotton swabs were centrifuged to collect the pellets followed by extracting DNA for microbiota analysis. The accumulation of L. paracasei (A) and the abundance of C. acnes (B), S.epidermidis (C), M. globosa (D), and M. restricta (E) in Start, Ctrl, and GMNL-653 groups were quantified by qPCR method. The relative quantified data of L. paracasei, C. acnes, and S. epidermidis were normalized with total bacteria value (A-C) and M. globosa and M. restricta were normalized with total Malassezia value (D-E). * p < 0.05, and ** p < 0.01 compared with Start group. # p < 0.05 compared with the Ctrl group

Full size image

Correlation between microbial species in human scalp microbiota after using heat-killed GMNL-653 shampoo

We then investigated the association between the increase in L. paracasei on the scalp and four commensal microbes, namely, M. restricta, M. globosa, C. acnes, and S. epidermidis. Sixty-six data entries collected from three different time points of 0, 1, and 2 months during the clinical trial were used for analysis. The results showed that L. paracasei level was negatively correlated with that of C. acnes and positively correlated with that of M. globosa but was not statistically correlated with that of S. epidermidis and M. restricta (Table 2). We further observed the correlations in the normal and weak subgroups of the three scalp conditions, namely, hair growth, oil secretion, and dandruff. Similarly, L. paracasei abundance was negatively correlated with that of C. acnes and positively correlated with that of M. globosa in the normal subgroups of hair, oil, and dandruff. L. paracasei abundance was statistically positively correlated with that of S. epidermidis in the high dandruff subgroup. However, no statistically correlation was observed between the abundances of L. paracasei and M. restricta even in all subgroups (Table 2). Next, we determined the correlations among the two fungal and two bacterial species in the clinical trial of using heat-killed GMNL-653 shampoo. Table 3 lists 66 collected data used to observe the correlation between bacteria and fungus. S. epidermidis abundance was negatively correlated with that of M. globosa and positively correlated with that of M. restricta. C. acnes abundance was negatively correlated with that of M. globosa but not statistically correlated with that of M. restricta according to the 66 data observation. In the subgroup observation, S. epidermidis abundance was negatively correlated with that of M. globosa in the weak subgroup of low hair. A positive correlation between S. epidermidis and M. restricta abundance and a negative correlation between C. acnes and M. globosa abundance were observed in the weak subgroup of low hair and in the normal subgroups of oil and dandruff. For the correlation between two fungi, M. globosa abundance was positively correlated with that of M. restricta in the high dandruff subgroups. However, no statistical correlation was found between bacterial S. epidermidis and C. acnes abundance in total data and in the subgroups of three scalp conditions. These data suggest that the use of heat-killed GMNL-653 shampoo could change the scalp microbiome to achieve its beneficial effects on scalp health.

Table 2 Correlations of L. paracasei level with C. acnes, M. globosa, S. epidermidis, or M. restricta in the shampoo clinical trial
Full size table
Table 3 Correlations among the fungal and bacterial species in the shampoo clinical trial
Full size table

Potential role of bacterial shifts in human scalp health

We investigated whether the microbiota shift modulated by heat-killed GMNL-653 influences the scalp condition and health. By analyzing the correlations of microbial abundance with the three scalp conditions, namely, hair growth, sebum secretion and dandruff, we found that the abundance of bacterial C. acnes and S. epidermidis was positively correlated with sebum secretion and dandruff, respectively, after the use of GMNL-653 shampoo for 2 months (Fig. 9). However, the two bacteria were not correlated with hair growth of the 3rd and 5th months (Supplementary Table S2). The abundance of fungi M. restrica and M. globosa did not show statistical correlations with sebum secretion, hair growth, and dandruff (Supplementary Table S2). These results indicate that the beneficial effects of heat-killed GMNL-653-containing shampoo on sebum secretion and dandruff are correlated with the changes of C. acnes and S. epidermidis.

Fig. 9
figure 9

The correlations of C. acnes or S. epidermidis to human scalp conditions after using heat-killed GMNL-653-containing shampoo. The levels of C. acnes and S. epidermidis of all participants in three time points after using heat-killed GMNL-653 shampoo (0, 1, 2 months, n = 66) were used for analyzing the Pearson correlation coefficient (R) to oil counts. R and the p value were calculated by SPSS statistics

Full size image

Discussion

Reygagne et al. [26] reported that orally consumed active probiotics L. paracasei NCC2461 ST11 could restore human severe dandruff by modulating the scalp microbiota. Topical cream containing heat-killed L. platarum GMNL-6 also improves human skin health [23]. We previously demonstrated that the treatment of smeared gel containing heat-killed L. platarum GMNL-6 and L. paracasei GMNL-653 on wounded mouse tails exhibited excellent healing ability and the capacity to prevent excessive fibrosis through the suppression of TGF-β/pSmad signaling in the skin wound repair [22]. In the current study, we found that heat-killed L. paracasei GMNL-653 could ameliorate damaged human scalp health for the first time. Improvements in human scalp conditions including decreased oil secretion and dandruff formation and increased hair volume were observed after the use of heat-killed GMNL-653-containing shampoo (Figs. 5, 6 and 7). In particular, the effect of heat-killed GMNL-653 shampoo on improving scalp health was observed for the participants with weak scalp condition at the beginning, such as the weak subgroups of high oil (Fig. 5B), high dandruff (Fig. 6B), and less hair (Fig. 7B), but not apparent for the participants with the healthy scalp condition (Figs. 5C, 6C, and 7C). The results suggest that the beneficial effect of the heat-killed GMNL-653 shampoo is more significant for the people with weak scalp conditions than for those with healthy conditions.

We also showed that heat-killed L. paracasei GMNL-653 could co-aggregate with M. furfur, and GMNL-653-derived LTA could inhibit the biofilm formation of M. furfur in our in vitro study (Fig. 1). Adherence, hydrophobicity, and biofilm formation are important virulence factors to change skin M. furfur from commensal status to pathogen status [27], and the pathological role of M. furfur is strongly correlated with dandruff formation in human scalp [28]. However, our in vivo studies for M. furfur could not demonstrate the correlation (data not shown). This phenomenon can be attributed to the fact that our clinical project recruited participants with healthy scalp condition rather than those with scalp diseases. Other studies showed that the distribution of Malassezia species in scalps of different populations could vary depending on factors such as differences of scalp disease, age, gender, body site, and resident region [29,30,31]. For example, one study found that the isolation frequency of M. furfur was 13% in healthy subjects, 25% in patients with SD, 10% in patients with atopic dermatitis, and 52.5% in patients with pityriasis versicolor [29]. The specific population of recruited participants must be considered to study the correlation of M. furfur with scalp condition in the future. Our in vivo observation also showed that the scalp microbial abundance of fungus M. restricta and M. globosa or bacteria C. acnes and S. epidermidis was observed after the participants with normal scalp condition at the beginning used heat-killed GMNL-653 shampoo (Fig. 8). M. restricta and M. globosa are the predominated species in human scalp Malassezia colonization. In particular, M. restricta has a critical role in the pathogenesis of dandruff and SD among the different species of Malassezia [32,1.

Study design of clinical trial

A 5-month clinical study was approved by the Institutional Review Board in Antai-Tian-Sheng Memorial Hospital (TSMH IRB No./ Protocol No. 20–040-A, approval date: 06/07/2020), Antai Medical Care Cooperation, Donggang Township, **tung County, Taiwan) and performed in accordance with the relevant guidelines and regulations. Informed consent was obtained from all subjects and/or their legal guardians. The study was registered at ClinicalTrials.gov with approval number NCT04566549 (first posted date: 28/09/2020), and the data were collected at Chia Nan University of Pharmacy and Science (Tainan, Taiwan). Twenty-two healthy adults (age: 37 ± 6.2 years old; 8 males and 14 females) were recruited (Table 1). The participants underwent examinations including oil count, hair volume, dandruff, and scalp microbiota, the primary outcomes of this trial, according to the time points of using shampoo shown in Fig. 4. In the first month, all participants used the baseline control shampoo without heat-killed probiotics GMNL-653 to wash their hair. The ingredients of control shampoo contained aqua, sodium lauryl ether sulphate, cocamidopropyl betaine, cocoamide DEA, sodium, polyquaternium-10, disodium EDTA, acrylates copolymer, citric acid, phenoxyethanol, and fragrance. Afterward, the participants washed hair with heat-killed probiotics GMNL-653-containing shampoo for 4 months. This shampoo contained 0.5% heat-killed L. paracasei and the same ingredients found in the control shampoo. The frequency of hair washing would be once in each day or twice each day depending on their personal habits. For the checking of subjects’ compliance, the average use dosage of control shampoo was 152 ± 69.8 g/month and GMNL-653-containing shampoo was 161.5 ± 56.2 g/months. The application status of two kinds of shampoos in 22 subjects were similar (P-value = 0.812). The value of oil count was the sum of three regions of scalp (front, middle, and back) analyzed by Sebumeter 815 (Courage + Khazaka electronic GmbH, Köln, Germany) in each participant. The hair volume was the average of three scalp regions (front, middle, and back) analyzed and measured by Aram TSII (Visia Complexion Analysis, Canfield Scientific, Inc., Parsippany, NJ, USA). For the test of dandruff, the flakes were collected from the whole scalp by dandruff D-Squame tape (Clinical and Derm LLC., Dallas, TX, USA). The dandruff was evaluated by measuring the percentage of flakes in the total area of dandruff tape using Image J software (version 1.53n 7, National Institutes of Health, Bethesda, Maryland, USA). The scalp conditions of hair volume, oil secretion, and dandruff were classified into normal and weak subgroups according to the scalp situation at the beginning without shampoo treatment. The subgroups of hair volume (normal: > 125 hairs/cm2; weak: > 125 hairs/cm2), oil secretion (normal: < 400 μg/cm2; weak: > 400 μg/cm2), and dandruff (normal: < 0.1%; weak: > 0.1%) were defined. Each value of oil level was the sum of three measured sites from front, middle, and back scalp.

DNA extraction and analysis of scalp microbiota by quantitative PCR (qPCR)

For the analysis of scalp microbiota, samples of whole scalp were collected by wet cotton swab containing 1 ml of PBS buffer with 0.1% Triton X-100. The separate areas of scalp including front left, front right, back left, back right, edge of the hair line, and behind the ears were swabbed following the Z line back and forth at least three times. The swab head was placed in a 1.5 ml Eppendorf tube to vortex for 30 s and incubated in room temperature for 15 min. After incubation, the swab head was centrifuged at 13,000 rpm for 10 min and then transferred to sterile H2O. The samples were dissolved and released out by swirling and snap** the swab. After the swab was removed, the liquid sample was centrifuged at 13,000 rpm for 5 min and the supernatant was then removed. DNA from the pellets was extracted by Quick-DNA fungal/bacterial kit (Zymo Research Corporation, Irvine, CA, USA). The reaction mixtures were prepared by 5 μl of 2 × Rotor-Gene SYBR Green PCR Master Mix (QIAGEN), 2 μl DNA from scalp extract and 3 μl of forward and reverse primers. The primer sequences of target microbes and internal control (total bacteria for L. paracasei, C. acnes, or S. epidermidis or total Malassezia for M. globose or M. restricta) are shown in part B of Table S2. The reaction mixtures were analyzed on the Roter-Gene Q 2plex machine (QIAGEN). The relative quantification of changes in specific bacterial species were calculated by the equation 2−△Ct, where △Ct is Cttarget bacteria – Cttotal bacteria. The relative quantification of changes in specific Malassezia species were calculated by the equation 2−△Ct, where △Ct is Cttarget malassezia – Ct malassezia.

Statistical analysis

All data were presented as mean ± SEM. Comparisons between two groups were determined using unpaired t-test, and comparisons among three groups or above were performed by one-Way ANOVA and Tukey’s post hoc tests. Pearson correlations between two microbiota variables were computed by PASW Statistical 18 Software (SPSS Inc., Chicago, IL, USA). A p value less than 0.05 was considered as statistically significant.