Two methicillin-sensitive S. aureus strains (MSSA; ATCC 6538 and ATCC 25923) and two methicillin-resistant S. aureus strains (MRSA 33591 and MW2), an S. epidermidis strain (ATCC 14990), a fungal Candida albicans DAY185 strain, a uropathogenic E. coli O6:H1 strain CFT073 (ATCC 700928), and a Vibrio parahaemolyticus strain ATCC 17802 were used. Cultures of MSSA ATCC 6538, ATCC 25923, and S. epidermidis strains were performed in Luria-Bertani (LB) broth, and MRSA ATCC 33591 and MW2 strains were cultivated in LB medium containing 0.2% glucose at 37°C and 30°C. Candida albicans DAY185 was cultured in potato dextrose broth (PDB). UPEC and V. parahaemolyticus were cultured in nutrient broth (NB) and LB supplemented with 3% (w/v) NaCl (mLB) at 37°C, respectively. All-trans-retinoic acid, 9-cis-retinoic acid, 13-cis-retinoic acid, and crystal violet were obtained from Sigma-Aldrich (St. Louis, MO, United States). Dimethyl sulfoxide (DMSO) was used to dissolve retinoic acids, and DMSO (0.1% v/v) was used as a control and it did not affect cell growth or biofilm formation. For planktonic cell growth assay, colony-forming units (CFU) were determined after incubating S. aureus cells in 96-well plates in LB medium with or without 13-cis-retinoic acid for 24 h.

A crystal violet staining assay was conducted using 96-well plates, as previously reported (Lee et al., 2021). S. aureus cells (~10⁷ CFU/mL) were inoculated into LB medium and retinoic acids were added at 0, 2, 5, 10, 20, 50, or 100 μg/mL to the wells of 96-well plates and cultivated for 24 h at 37°C without agitation. Biofilm formation was measured by discarding planktonic cells and washing the plates three times with distilled water. Biofilm cells were then stained with 0.1% crystal violet (300 μL) for 20 min and washed three times with water to remove crystal violet. Crystal violet stained cells were then extracted with 95% ethanol (300 μL) by shaking vigorously. Absorbances were measured at 570 nm (OD₅₇₀) using a Multiskan plate reader (Thermo Fisher Scientific, Waltham, MA, United States). Biofilm formation results are obtained from three independent cultures of six replicate wells.

After forming S. aureus biofilms in 96-well plates in the presence or absence of 13-cis-retinoic acid (0, 2, 5, or 10 μg/mL) for 24 h at 37°C, planktonic cells were removed by washing three times with distilled water, and live biofilm cells were observed by the iRiS™ Digital Cell Imaging System (Logos Biosystems, Anyang, Korea). Color-coded 3D biofilm images were generated using ImageJ.¹

Also, SEM was used to observe biofilm reduction by 13-cis-retinoic acid, as previously reported (Park et al., 2022). S. aureus ATCC 6538 cells (~10⁷ CFU/mL) were inoculated into 1 mL of fresh LB medium with or without 13-cis-retinoic acid (0, 2, 5, or 10 μg/mL) in a 96-well plate. A piece of nylon membrane (~ 0.16 cm²) was placed in each well, and S. aureus cells were cultured for 24 h at 37°C without agitation. Biofilms developed on the membrane were then fixed with a glutaraldehyde (2.5%) and formaldehyde (2%) for 24 h, post-fixed with OsO₄ (1%), and dehydrated with ethanol and isoamyl acetate (99%). After drying biofilms using critical-point dryer (HCP-2, Hitachi, Tokyo, Japan), biofilm cells were coated with Precision Etching Coating System (Gatan, Inc., Pleasanton, United States) and observed under a field emission scanning electron microscope S-4800 (Hitachi, Tokyo, Japan).

The hemolysis of sheep blood cells (MBcell, Seoul, Korea) was investigated as described previously (Kim et al., 2022a). S. aureus ATCC 6538 cells (~2 × 10⁷ CFU/mL) were diluted in 2 mL of fresh LB medium, cultivated with retinoic acids (0, 0.5, 1, 2, 5, or 10 μg/mL) for 24 h with 250 rpm shaking. Fresh sheep blood cells were collected by centrifugation at 3,000 × g for 2 min, the red blood cells were then cleaned three times with PBS and resuspended gently in PBS buffer (3.3%). S. aureus cell culture (100 μL) was then added to 1 mL of red blood cells and incubated for 4 h at 37°C with shaking at 250 rpm. The mixtures were centrifugated at 16,600 × g for 10 min, and the absorbances of supernatants were measured at 543 nm.

Staphylococcus aureus ATCC 6538 cells (~2 × 10⁷ CFU/mL) were inoculated into LB medium (2 mL) in 14-mL tubes and incubated for 24 h at 37°C with 13-cis-retinoic acid (0, 10, 20, 50, or 100 μg/mL) with shaking at 250 rpm. Staphyloxanthin levels were assessed optically, as previously described (De Souza Feitosa Lima et al., 2019; Kim et al., 2022a).

To quantify the effect of 13-cis-retinoic acid on extracellular lipase production, S. aureus ATCC 6538 cells (~2×10⁷ CFU/mL) were inoculated into LB medium (2 mL) in 14-mL tubes and incubated for 20 h at 37°C with 250 rpm shaking with or without 13-cis-retinoic acid (0, 2, 5, 10, 20, or 50 μg/mL), as previously reported (Lee et al., 2022). Briefly, culture supernatants (0.1 mL) were mixed with 0.9 mL of substrate solution (10% of buffer A with 3 mg/mL of p-nitrophenyl palmitate in isopropyl alcohol and 90% of buffer B with 1 mg/mL of gum arabic and 2 mg/mL sodium deoxycholate in 50 mM Na₂PO₄ buffer and then heated at 40°C for 30 min). The reactions were stopped by adding 1 M Na₂CO₃. Absorbances of the reaction supernatant were measured at 405 nm.

S. aureus ATCC 6538 cells at OD₆₀₀ of 0.05 were inoculated to 15 mL of LB medium in a 250 mL flat-bottomed flask and incubated for 6 h at 37°C with shaking at 250 rpm with or without 13-cis-retinoic acid (100 μg/mL). Cells were then treated with RNase inhibitor (RNAlater, Ambion, TX, USA) for preventing RNA degradation and harvested by centrifugation at 16,600 × g for 1 min. Total RNA was purified using an RNA isolation/purification kit (Qiagen RNeasy Mini Kit, Valencia, CA, USA), and additional step for cell lysis was performed using glass beads to enhance cell disruption. Briefly, acid-washed glass beads (Sigma-Aldrich, 150–212 μm, ~10 x vol. of cell pellet) were added in lysis buffer. The mixture was vortexed vigorously for 50 s and chilled down on ice between each vortex for 50 s, which was repeated twelve times. After breaking cells, supernatant was collected by centrifugation at 16,000 x g for 10 min and the rest of the procedure was followed by the manufacturer’s guidelines. qRT-PCR was applied to analyze the transcript levels of 32 biofilm- and toxin-related genes (agrA, agrB, agrC, agrD, alsS, arlR, arlS, aur, clp9, coa, fibA, fibB, hla, icaA, icaR, isaA, lrgB, nuc1, nuc2, psmα, rbf, RNAIII, saeR, saeS, sarA, sarZ, seb, sigB, srrA, srrB, spa, and yycF) in S. aureus ATCC 6538 cells. Primers used are listed in Supplementary Table S1, and 16s rRNA was used as the housekeeping control. qRT-PCR was performed as previously described (Lee et al., 2022) using an SYBR™ Green qPCR Master Mix (Applied Biosystems, Foster City, United States) and an ABI StepOne Real-Time PCR System (Applied Biosystems). The changes of each gene expression were determined using two independent cultures and four reactions per gene.

Brassica rapa (Chinese cabbage) seeds were soaked in sterile H₂O for 16 h, rinsed with water three times, sterilized by soaking in 95% ethanol first and then 3% sodium hypochlorite (both for 15 min) at 25°C, and rinsed with sterile H₂O three times. Ten seeds per plate were carefully placed on Murashige and Skoog soft agar plates containing 0.7% agar and 0.86 g/L Murashige and Skoog (MS) and 13-cis-retinoic acid at 0, 20, 50, and 100 μg/mL, and then incubated at 25°C for 5 days. Seed germination rates and seedling lengths were then measured. Four independent cultures were used.

Caenorhabditis elegans fer-15(b26); fem-1(hc17) strain was used to investigate the chemical toxicity of 13-cis-retinoic acid, as previously described (Kim et al., 2022b). Synchronized nematodes were carefully washed two times with M9 buffer (3 g/L KH₂PO₄, 6 g/L Na₂HPO₄, 5 g/L NaCl, 1 mM MgSO₄). Approximately 40 worms were placed into the each well of 96-well plates containing M9 buffer (200 mL) and 13-cis-retinoic acid (50, 100, 200, or 500 μg/mL). Then, the plates were incubated for 10 days at 25°C. Four independent cultures were used. Survived nematode percentages was determined by responses to LED lights for 30 s and an iRiS™ Digital Cell Imaging System (Logos BioSystems).

The drug-like properties of 13-cis-retinoic acid were evaluated using ADME software. The online web servers, viz, PreADMET² Molinspiration³ and Gusar⁴ were accessed on March 15, 2023.

All results were analyzed by one-way ANOVA followed by Dunnett’s test in SPSS version 23 (SPSS Inc., Chicago, IL, United States). Results are presented as averages and standard deviations, and p values <0.05 are considered as a significant change.

The antibiofilm efficacies of three vitamin A₁ metabolites, namely, all-trans-retinoic acid, 9-cis-retinoic acid, and 13-cis-retinoic acid, were initially tested at concentrations up to 100 μg/mL to investigate their effects on methicillin-sensitive S. aureus (MSSA 6538). Of these compounds, 13-cis-retinoic acid significantly inhibited biofilm formation, all-trans-retinoic acid had a weak inhibitory effect, but 9-cis-retinoic acid had no effect (Figures 1A–C). More specifically, 13-cis-retinoic acid at 10 μg/mL reduced S. aureus biofilm formation by 91%, while all-trans-retinoic acid at 100 μg/mL inhibited the biofilm formation by 34%. Also, the antimicrobial activity of 13-cis-retinoic acid was investigated by measuring colony-forming units, and at 50 μg/mL, it slightly delayed planktonic cell growth with a MIC of >400 μg/mL (Figure 1D). These results showed that the observed antibiofilm activity of 13-cis-retinoic acid was mainly due to its ability to inhibit biofilm formation rather than cell growth inhibition.

Further biofilm assays were performed on another MSSA 25923 strain and two methicillin-resistant S. aureus strains (MRSA 33591 and MW2). 13-cis-Retinoic acid potently reduced biofilm formation by MSSA 25923, MRSA 33591, and MRSA MW2 strains with MICs of >400 μg/mL (Figures 2A–C). Specifically, 13-cis-retinoic acid at 10 and 20 μg/mL decreased biofilm formation by MSSA 25923, MRSA 33591, and MRSA MW2 strains by ≥84%.

Also, the antibiofilm activities of three retinoic acids were investigated with a S. epidermidis strain. As was observed for S. aureus, S. epidermidis biofilms were strongly inhibited by 13-cis-retinoic acid, weakly inhibited by all-trans-retinoic acid, but unaffected by 9-cis-retinoic acid (Figures 2D–F).

The impact of 13-cis-retinoic acid on other biofilms was also investigated with a uropathogenic Escherichia coli strain, an aquatic pathogenic Vibrio parahaemolyticus strain, and a fungal Candida albicans strain. Unlike that observed for the five Staphylococcal biofilms, 13-cis-retinoic acid up to 50 μg/mL did not reduce biofilm formation by two Gram-negative pathogens (E. coli and V. parahaemolyticus) (Figures 2G,H) and only exhibited weak antibiofilm activity against C. albicans (Figure 2I). These results indicate that 13-cis-retinoic acid is active against Gram-positive Staphylococcal biofilms but not against Gram-negative bacteria.

Live imaging microscopy and SEM were utilized to observe biofilm reduction. For untreated biofilms, 3D color images obtained by bright-field microscopy were green, indicating dense biofilms, whereas 13-cis-retinoic acid at 2–10 μg/mL produced yellow to red colors, indicating weak to no biofilm formation (Figure 3A). SEM analysis also showed that 13-cis-retinoic acid markedly diminished the numbers of S. aureus cells in biofilms but did not affect S. aureus cell morphology (Figure 3B). These observations confirmed that 13-cis-retinoic acid at 2–10 μg/mL significantly inhibited S. aureus biofilm formation without affecting cell morphology.

S. aureus is known to produce various virulence factors, including hemolysins, staphyloxanthin, and extracellular lipase, and thus, we investigated the effects of the three retinoic acids on their levels. Notably, all-trans-retinoic acid and 13-cis-retinoic acid dose-dependently reduced the red blood cell hemolytic activity of S. aureus, while 9-cis-retinoic acid showed only weak anti-hemolytic activity (Figure 4A). For example, all-trans-retinoic acid and 13-cis-retinoic acid at 50 μg/mL inhibited hemolytic activity by 84 and 83%, respectively, which partially reflected their antibiofilm activities (Figure 1).

Also, the effects of the three retinoic acids on yellow staphyloxanthin production in S. aureus were investigated. Interestingly, all-trans-retinoic acid and 13-cis-retinoic acid dose-dependently inhibited staphyloxanthin production, whereas 9-cis-retinoic acid did not, which paralleled our biofilm inhibition and hemolytic activity results (Figure 4B). However, 13-cis-retinoic acid did not affect extracellular lipase activity at concentrations of <50 μg/mL (Supplementary Figure S1).

To investigate the molecular mechanisms responsible for the antibiofilm and antitoxin activities of 13-cis-retinoic acid on S. aureus, we used qRT-PCR to assess the expressions of 32 selected biofilm-, toxin-related genes and regulatory genes in S. aureus MSSA 6538 cells. 13-cis-Retinoic acid for 6 h incubation significantly downregulated the gene expression of the arlRS two-component system, α-hemolysin (hla), nuclease (nuc1 and nuc2), coagulase coaA, staphylococcal antigen A isaA, antiholin-like protein lrgB, and psmα (phenol soluble modulins α) but slightly upregulated the expression of the transcriptional regulator sarA and sarZ. However, the expression of other genes tested was unchanged (Figure 5). Notably, 13-cis-retinoic acid suppressed hla expression 13-fold, which matches with its inhibitory effect on S. aureus hemolytic activity.

Further qRT-PCR has been conducted with a different growth condition of 10 h contact with 13-cis-retinoic acid (stationary growth phase) instead of 6 h contact (exponential growth phase). The changes of gene expression were attenuated since only the two-component kinase arlS was downregulated and other genes including QS-related genes were less affected after 10 h incubation (Supplementary Figure S2). The result indicates that gene expression is a dynamic process depending on incubation and growth stages.

Chemical toxicity assessments of 13-cis-retinoic acid were performed using a B. rapa germination assay and a C. elegans model. Interestingly, 13-cis-retinoic acid dose-dependently increased (not decreased) plant root growth for 4 days (Figures 6A,C) and also slightly increased the seed germination rate (Figure 6B). In the nematode model, incubation with 13-cis-retinoic acid up to 50 μg/mL for 10 days was non-toxic (Figure 6D). While incubation for 8 days at concentrations of ≤500 μg/mL had no acute toxic effect, mild toxicity was observed for old nematodes after 8 days. These results suggest that 13-cis-retinoic acid may not be toxic to plants or nematodes in its antibiofilm concentration range (2–10 μg/mL).

In silico ADME profiling showed 13-cis-retinoic acid violated one (miLogP <5) of Lipinski’s Rule of Five but had acceptable human intestinal adsorption and skin barrier permeability and no fish toxicity. The ADME parameters investigated are detailed in Supplementary Table S2.