From September 27, 2021 to November 19, 2022, the study was constructed in the form of a single-center, open-label, parallel groups study. Patient recruitment was conducted at the Shanxi Provincial People’s Hospital, where all patients had comprehensive clinical records available. Prior to their involvement in the experiment, all patients signed written informed consent. The following eligibility criteria were applied: patients of acne vulgaris aged between 18 and 30 had facial lesions between 20 and 50 and achieve grade III to IV according to the Chinese Acne Treatment Guidelines with more than 10 inflammatory acne lesions on their face. Patients who met the criteria had no history of systemic disease. The exclusion criteria were as follows: people who smoke, are pregnant, breastfeeding; patients with a history of allergy to drugs or probiotics; individuals had taken oral isotretinoin within the last 6 months or taken oral antibiotics or probiotics in the last 1 month for acne; individuals with a history any other form of topical or systemic anti-acne treatment in the last 1 month; patients had concurrent dermatological conditions that might influence the assessment of acne such as folliculitis or rosacea.

This trial was registered with the China Clinical Trials Center and assigned the registration number ChiCTR2200063499. This research received ethical approval from the Ethics Committee of Shanxi Provincial People’s Hospital.

In the present study (23), it was assumed that the expected treatment effect of control group would result in a 10.00% reduction in inflammatory skin lesions, while the intervention group would reach a 56.67% reduction (power = 0.8, α = 0.05) Accounting for a 20% attrition rate, each group was intended to enroll 19 participants at least. Thus, a total of 105 individuals were recruited in this study.

Patients (n = 105) meeting the eligibility criteria received a baseline treatment of oral doxycycline (20–30 mg/kg/day) for the initial 4 weeks. Following this phase, they were randomized into three groups: Group I (n = 35) receiving oral isotretinoin 50 mg/kg/day for 12 weeks; Group P (n = 35) receiving oral L. plantarum MH-301 (CGMCC No. 18618, Harbin Meihua Biotechnology Co., Ltd., Harbin, Heilongjiang, China) treatment at a dosage of 2 g/day (10⁹ CFU/g) for 12 weeks; Group IP (n = 35) receiving combined treatment with oral probiotics (2 g/day, 10⁹ CFU/g) and oral isotretinoin (50 mg/kg/day) for 12 weeks. The probiotic used in this experiment was a mixture of L. plantarum MH-301 and maltodextrin with 2 × 10⁹ colony-forming units per packet of live bacteria. The probiotic preparation stored in a refrigerator at 4°C. Skin samples collections and lesion counts were performed at four different time points: before treatment, at 4 weeks, 8 weeks, and 12 weeks during the treatment. Feces samples were collected at 12 weeks of the treatment. Feces samples were collected using 15 mL sterile tubes and stored at −80°C refrigerator for high-throughput sequencing analysis.

In this study, the lesion counting method was chosen as the clinical assessment approach for acne vulgaris. Facial photographs were captured under standardized conditions at 0 week, 4 weeks, 8 weeks, and 12 weeks of treatment, respectively, and skin lesion counting was conducted by the same dermatologist. This counting included both non-inflammatory lesions, such as open and closed comedones, and inflammatory lesions, including papules, and pustules according to the Chinese Acne Treatment Guidelines. Importantly, the dermatologist conducting the lesion counting was kept unaware of the patient groupings, ensuring an unbiased assessment.

To detect the difference of skin microbiota in each group, patient skin samples were collected at 4 weeks, 8 weeks and 12 weeks of treatment. After moistening a sterile cotton swab with physiological saline, the swab was gently used to scrape the surface of skin lesion. Following the sampling, the tip of the cotton swab was cut off and placed into a sterile EP tube. The tube was sealed tightly and stored in a −80°C freezer pending subsequent sequencing. Moreover, for the fecal microbiota, feces samples were collected at 12 weeks of treatment in centrifuge tubes and stored in a refrigerator at −80°C awaiting sequencing analysis.

Total genomic DNA was extracted from skin and feces samples of patients using the TIANamp Bacteria DNA Kit (QIAGEN) in accordance with the manufacturer’s instructions and from glass fiber filters. The genomic DNA concentration and quality extracted were evaluated using a spectrophotometer (NanoDrop, Thermo Fisher Scientific, Inc., United States). The DNA extracted underwent 16S sequencing at Shanghai Personal Biotechnology Company Ltd. The primer sets 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and 806R (5′-CGGACTACHVGGGTWTCTAAT-3′) were used to amplify the hypervariable V3V4 region of the 16S rRNA gene in the skin samples, while primer sets 520F (5′-AYTGGGYDTAAAGNG-3′) and 802R (5′-TACNVGGGTATCTAATCC-3′) were used to amplify the V4 hypervariable region in feces samples. The PCR-amplified products were double-end sequenced utilizing the Illumina MiSeq platform. FLASH was used to merge overlapped reads, and sequence analysis was carried out using UPARSE software package. Reads with quality scores lower than 20, ambiguous bases, and improper primers were discarded before clustering, and chimeras were detected and removed during the process. ASV/OTU signature sequences were acquired through the application of the DADA2 approach, followed by processing via Quantitative Insights into Microbial Ecology (QIIME). The taxonomic categorisation was carried out utilizing the Greengenes version 13.8 database (19), where the percentage of total bacteria in every sample was placed into dissimilar classification tiers (Phylum, Class, Order, Family, and Genera). The samples collected from Groups M, I, P, and IP underwent analysis for α-diversity, β-diversity, and species differences.

The α-diversity, including Chao1 and Observed species indices (represented richness), Shannon and Simpson indices (represented diversity), and Goods-coverage index (represented coverage), were calculated and visualized by QIIME2 and R software (v 3.6.3, R Foundation for Statistical Computing). The β-diversity, which is based on Jaccard distances and unweighted UniFrac distances, was demonstrated by PCoA to show the dissimilarity of species composition, and was calculated and visualized using QIIME2 and R software. The distributions of skin and feces microbiota at both phylum and genus levels were analyzed and visualized using QIIME2, and the results were shown as histograms.

Prism version 9.5.0 (GraphPad Prism, San Diego, CA, United States) or SPSS 27.0.1 were used for data analysis. Continuous variables of normality distributions were shown as mean ± standard deviation (SD), qualitative data are expressed as rates, and one-way ANOVA with Tukey’s multiple comparison test was employed to assess the variations between groups. Covariates were selected based on gender, age and total number of skin lesions prior to treatment. Descriptive statistics were used to describe the characteristics of the recruited participants and were assessed using chi-square tests and one-way ANOVA to assess baseline comparability of age and sex between the different groups, respectively. Two-way ANOVA was used to assess differences in skin lesion counts between groups of patients at 4 weeks, 8 weeks and 12 weeks of treatment. Statistical significance was set at ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

In order to assess the adjuvant efficacy of probiotics in conjunction with isotretinoin for the treatment of acne vulgaris, a total of 153 individuals were recruited to assess for eligibility, of whom 105 individuals were initially screened for eligibility and subsequently enrolled into the study (Figure 1). They were randomly assigned, with 35 participants allocated to each of the oral isotretinoin group (Group I), oral L. plantarum MH-301 group (Group P), and oral combination of isotretinoin and L. plantarum MH-301 group (Group IP). There was no significant difference in the patients’ gender (p > 0.05) and age (I vs. P vs. IP = 22.68 vs. 21.72 vs. 22.44, p > 0.05). Regarding the total skin lesion count of baseline, no significant difference was found among the three groups (I vs. P vs. IP = 61.70 vs. 59.50 vs. 61.03, p > 0.05). Thus, based on clinical records, the recruited patients across the three groups were homogeneous (Table 1). During the course of the study, one participant in Group I withdrew consent and 2 patients received discontinued intervention, while 2 participants were lost to follow-up. A total of 2 and 3 participants were lost to follow-up in Groups P and IP, respectively. Totally, 3 and 2 patients were received discontinued intervention in Groups P and IP. Ninety participants successfully completed the study and were subsequently incorporated into the final analysis (Figure 1).

To evaluate whether L. plantarum MH-301 can improve acne vulgaris and enhance the therapeutic effect of isotretinoin on acne vulgaris, we administered oral isotretinoin (Group I), oral L. plantarum MH-301 (Group P), and an oral combination of isotretinoin and L. plantarum MH-301 (Group IP) to three groups, respectively, and monitored the number of skin lesions in patients at weeks 0, 4, 8 and 12 of treatment. The result of each intervention approach was measured by calculating the average number of changes in skin lesions before and after treatment within the same group, and then calculated the percentage reduction in skin lesions. Clinical results showed that comparing to the amount of skin lesions of baseline, the severity of acne vulgaris significantly reduced in patients following oral administration of isotretinoin and L. plantarum MH-301 (Supplementary Figure S1).

The reduction in skin lesions of each group was statistically significant as early as the fourth week of treatment, oral isotretinoin treatment (Group I) exhibited a significant average reduction in skin lesions19.02% (baseline vs. the fourth week = 61.70 vs. 49.97, p < 0.001). Compared to baseline, oral L. plantarum MH-301 (Group P) showed a reduction of 26.89% (baseline vs. the fourth week = 59.50 vs. 43.50, p < 0.001), and oral combination of isotretinoin and L. plantarum MH-301 group (Group IP) exhibited a reduction of 19.99% (baseline vs. the fourth week = 61.03 vs. 48.83, p < 0.001), and this improvement was sustained until the completion of the treatment course (Figures 2A,B and Supplementary Table S1). Overall, the amount of skin lesions in Group I was less than Group P, while group IP treatment had the largest reduction in skin lesion, especially in the 8- and 12 weeks during course (reductions of the twelfth week in Group I vs. P vs. IP = 20.20 vs. 28.87 vs. 17.33, p < 0.01) (Figure 1). By the twelfth week, Group I achieved a reduction of 67.26% (baseline vs. the twelfth week = 61.70 vs. 20.20, p < 0.001), Group P achieved a reduction of 51.48% (baseline vs. the twelfth week = 59.50 vs. 28.87, p < 0.001), while Group IP showed a reduction of 71.60% (baseline vs. the twelfth week = 61.03 vs. 17.33, p < 0.001) (Supplementary Table S1).

Then we further identify and counted non-inflammatory lesions, namely comedones and inflammatory lesions including pustules of papules. For comedones, the patients’ symptoms improved as the treatment progressed, and the effect was most significant at week 12, with Group IP having the best results comparing to the baselines (baselines vs. week 12 of Group IP = 30.80 vs. 9.03, p < 0.001) (Figure 2C), likewise with pimples and pustules (Figures 2D–F). Collectively, these results suggested that L. plantarum MH-301 have the ability to improve skin lesions of acne vulgaris significantly in short time, and L. plantarum MH-301 can enhance the effects of isotretinoin on acne to achieve long-term and best intervention effect.

Studies indicated a close association between acne vulgaris and disturbances in the skin microbiota. Therefore, we collected and sequenced microbiota from skin lesion sites of patients. A total of 64 skin samples from lesion sites were collected at week 12 of treatment, comprising 16 patients in each of the Group M, I, P, and IP. Subsequently, we subjected these samples to 16S rRNA sequencing analysis to investigate alterations in the skin microbial community. The α-diversity analysis showed that the Shannon index (p < 0.01) (Figure 3A), Simpson index (p < 0.01) (Figure 3B), and Goods-coverage index (p < 0.001) (Figure 3C) of Group I, P and IP increased dramatically compared with the baseline (Group M, for medicine not used), suggesting that both isotretinoin and L. plantarum MH-301 may be effective in restoring the diversity of skin microbiota that was disrupted by acne vulgaris. Furthermore, results from Venn diagrams (Figure 3D) revealed the presence of 508 shared Operational Taxonomic Units (OTUs) across all four groups, and the unique OTU numbers discovered in groups M, I, P, and IP were 5,331, 6,347, 3,644 and 3,362 respectively, indicating that acne vulgaris may impact the types of species that inhabited the skin and that treatment with L. plantarum MH-301 alone can hardly reshape this change. Additionally, principal coordinate analysis (PCoA analysis) revealed slight differences in patients between Group I, P and IP compared to Group M (Figure 3E), which further suggested that acne vulgaris affected the structure of human skin microbiota and L. plantarum MH-301 could remodel the skin microbial profile.

At the phylum level, Actinobacteria, Firmicutes, Proteobacteria and Bacteroidetes were the most prevalent populations in these treated groups, accounting for 95.62, 93.11, 96.60 and 97.06% of the total sequencing results, respectively (Figure 4A). Specifically, the relative abundance of Proteobacteria (Figure 4B) was 27.37% in oral isotretinoin group, which was significantly higher than that in baseline (Group M) (8.65%, p = 0.038), oral combination of isotretinoin and L. plantarum MH-301 group was also significantly higher than the Group M (32.63% vs. 8.65%, p = 0.007). In contrast, the relative abundance of Actinobacteria (Figure 4C) in Group I and IP were significantly higher than that in Group M (I vs. IP vs. M = 32.12% vs. 33.89% vs. 54.58%, p = 0.02, p = 0.04). Additionally, the abundance of Firmicutes (M vs. I vs. P vs. IP = 31.51% vs. 30.05% vs. 23.49% vs. 28.12%) and Bacteroidetes (M vs. I vs. P vs. IP = 0.88% vs. 2.57% vs. 1.91% vs. 0.34%) had no differences between groups, suggesting that they were may not affected by acne vulgaris, yet the abundance of Firmicutes could be reduced by L. plantarum MH-301 preparation and Bacteroidetes could be increased (Figures 4D,E).

At the genus level, we selected representative skin microbiota closely associated with acne vulgaris for further analysis (Figure 4F). The abundance of Propionibacterium and Corynebacterium were lower in the oral isotretinoin and L. plantarum MH-301 treatment groups alone and in the combination treatment groups (Group I, P and IP) than in the baseline (Group M). In concrete terms, the relative abundance of Propionibacterium (Figure 4G) in Group I, P and IP was 9.02, 28.67, 8.83% respectively, while in Group M was 38.26% (M vs. I = 38.26% vs. 9.02%, p = 0.0004, M vs. IP = 38.26% vs. 8.83%, p = 0.004). The relative abundance of Corynebacterium (Figure 4H) in Group I, P and IP was 10.89, 4.14 and 7.87%, while in Group M was 12.95% (M vs. p = 12.95% vs. 4.14%, p = 0.04). These results suggest that L. plantarum MH-301 could partially reverse the change trend of the skin microbiome. Additionally, the content of Staphylococcus and Streptococcus were not influenced by acne vulgaris as their relative abundance were similar among Group M, I, P, and IP (Staphylococcus M vs. I vs. P vs. IP = 27.19% vs. 18.04% vs. 12.11% vs. 16.85%, Streptococcus M vs. I vs. P vs. IP = 0.85% vs. 4.90% vs. 4.41% vs. 3.85%), but after treatment, the abundance of Staphylococcus decreased and the abundance of Streptococcus increased (Figures 4I,J). Collectively, these results suggested that L. plantarum MH-301 could improve the dysbiosis of skin microbiota, and remodel the skin microbial taxa in acne vulgaris patients, particularly in decreasing Propionibacterium, Corynebacterium and Staphylococcus and increasing Streptococcus.

Previous studies have indicated a possible association between acne vulgaris and the gut microbiota. We prescribed patients oral isotretinoin and L. plantarum MH-301 and collected their feces for 16S rRNA sequencing analysis to assess alternations in the gut microbiota of the patients. A total of 64 feces samples were collected (16 from the Group M, 16 from the Group I, 16 from the Group P, and 16 from the Group IP). It turns out that, for α-diversity, there were differences among the four groups in terms of Chao1, Observed-species, and Goods-coverage, but these differences did not reach statistical significance (Figures 5A–C). In particular, the Chao 1 and Observed-species in oral L. plantarum MH-301 group (Group P and IP) were marginally greater than that of in baseline (Group M), while L. plantarum MH-301 could restore the diversity of gut microbiota that was disrupted by acne vulgaris. Venn diagram results (Figure 5D) indicated that there were 284 common operational taxonomic units (OTUs) across all four groups, with unique OTUs numbering 517, 585, 617, and 493 for the Group M, I, P and IP, which further shown that acne vulgaris might affect the gut microbiological composition of patients, but only L. plantarum MH-301 could hardly remodel this change. Moreover, principal coordinate analysis (PCoA analysis) did not differ significantly between Group M, I, P and IP (Figure 5E).

At the phylum level, Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria and Verrucomicrobia were the most common microbiota in the patients’ gut, and they accounted for 98.10, 99.67, 99.88 and 99.26% of the sequencing results for each group, respectively (Figure 6A). Remarkably, after oral administration of L. plantarum MH-301, the abundance of Actinobacteria in Group I and IP were slightly higher than that in Group M (I vs. IP vs. M = 3.89 and 4.48% vs. 3.87%) (Figure 6C). In contrast, the relative abundance of Proteobacteria and Verrucomicrobia in oral L. plantarum MH-301 group (Group P and IP) (0.84 and 0.87%, 0.01 and 0.72%) were lower than that in baseline (Group M) (1.54, 1.37%) (Figures 6B,F). Additionally, L. plantarum MH-301 treatment showed opposite changes in the abundance of Firmicutes and Bacteroidetes, but the same changes in Group I and IP (Figures 6D,E).

At the genus level (Figure 6G), the fecal microbiota of patients in the probiotic treatment groups including Group P and IP were richer in Lactobacillus (0.81 and 0.85%), Bifidobacterium (2.68 and 3.20%), Coprococcus (7.68 and 6.12%), and Bacteroides (18.36 and 21.68%) than that in baseline (Group M) (0.09, 2.09, 3.09 and 17.31%) (Figures 6H,I,K,L). On the contrary, the abundance of Faecalibacterium reduced in Group P and IP. These results suggested that L. plantarum MH-301 had the potential to counteract certain alterations within the gut microbiota and restore gut homeostasis (Figure 6J). Collectively, these results suggested that L. plantarum MH-301 could help regulate and improve gut dysbiosis caused by acne vulgaris, thereby remodel intestinal microbiota diversity, especially in decreasing Faecalibacterium and increasing Lactobacillus, Bifidobacterium, Coprococcus and Bacteroides.