In vitro Antimicrobial Activity of Acne Drugs Against Skin-Associated Bacteria

Acne is a common skin affliction that involves excess sebum production and modified lipid composition, duct blockage, colonization by bacteria, and inflammation. Acne drugs target one or more of these steps, with antibiotics commonly used to treat the microbial infection for moderate to severe cases. Whilst a number of other acne therapies are purported to possess antimicrobial activity, this has been poorly documented in many cases. We conducted a comparative analysis of the activity of common topical acne drugs against the principal etiological agent associated with acne: the aerotolerant anaerobic Gram-positive organism Propionibacterium acnes (recently renamed as Cutibacterium acnes). We also assessed their impact on other bacteria that could also be affected by topical treatments, including both antibiotic-sensitive and antibiotic-resistant strains, using broth microdilution assay conditions. Drugs designated specifically as antibiotics had the greatest potency, but lost activity against resistant strains. The non-antibiotic acne agents did possess widespread antimicrobial activity, including against resistant strains, but at substantially higher concentrations. Hence, the antimicrobial activity of non-antibiotic acne agents may provide protection against a background of increased drug-resistant bacteria.

be limited to the shortest possible duration with re-evaluation at 3-4 months to minimize the development of bacterial resistance, and co-application of benzoyl peroxide (BP) to help reduce the development of resistance 7 . Topical therapy is strongly suggested to follow the discontinuation of systemic antibiotics as a maintenance regimen 7 . The European Evidence-Based Guideline for the Treatment of Acne has similar recommendations 9 .
The rise in antimicrobial resistance is accompanied by an increasing awareness of the role of the human microbiome in the ability of pathogenic species to establish an infection. Several recent genomic studies have specifically examined the human skin microbiome [10][11][12][13][14][15][16][17] , and even the subpopulation of P. acnes in the human skin microbiome 18,19 . These studies identify abundant populations of P. acnes and Staphylococcus spp. (especially S. epidermidis, but also S. aureus and S. hominis, and lower levels of S. warneri, S. saprophyticus, S. lugdunensis, S. haemolyticus and S. capitis). High levels of Corynebacterium, Streptococcus mitis and the fungus Malassezia globosa have also been identified, with community composition varying depending on the skin region and skin type (sebaceous, dry skin or wet skin) 8,13 . The top ten bacteria present in sebaceous sites are P. acnes, S. epidermidis, Corynebacterium tuberculostearicum, S. capitis, Corynebacterium simulans, S. mitis, S. hominis, Corynebacterium aurimucosum, Corynebacterium kroppenstedtii, and Corynebacterium amycolatum 13 . Lactobacillus, β-Proteobacteria and Flavobacteriales are also found in sebaceous sites, though usually are more common in 'dry' or 'moist' skin environments 10 . Altering the skin microbiome with topical antibiotic treatment can have significant effects on the cutaneous host defense 20 , and some skin bacteria (such as Micrococcus luteus) have been found to enhance S. aureus pathogenesis 21 . A new anti-acne tetracycline, sarecycline, has been designed as the first narrow-spectrum tetracycline-class antibiotic being developed for acne treatment, reducing collateral damage on the microbiome (though in this case used systemically, not topically) 22 .
It is important to know the relative effects of antimicrobial agents on human microbiota in order to understand their potential to foster resistance and alter the microbiome composition. To date, there has not been a comparative assessment of the antimicrobial activity of commonly used antibiotics and topical acne agents against a set of representative commensal skin bacteria, including those not directly associated with acne. We now report such a study against standardized accessible organisms from reference collections, testing both specific antibiotics used to treat acne (tetracycline, erythromycin, clindamycin, oxacillin, dapsone, along with control antibiotics vancomycin/colistin) and other acne agents reported to have antimicrobial activity (salicyclic acid, azelaic acid, benzoyl peroxide) (see Fig. 1). These are assessed against both sensitive and resistant bacterial strains, under anaerobic and aerobic conditions. In addition to some of the most common strains identified by microbiome studies, we also include several additional pathogenic bacteria that can be found on the skin and/or involved in skin infections, such as Streptococci (S. pyogenes and, less commonly S. pneumoniae 23

Results and Discussion
The antimicrobial activity of the antibiotics and anti-acne agents, tested under standard broth microdilution (BMD) conditions, are summarized in Tables 1-3. The topical acne therapeutics originally developed as specific antimicrobial agents (tetracycline, erythromicin, oxacillin, and clindamycin) generally showed potent activity under both anaerobic and aerobic conditions against a range of bacteria, though erythromycin, oxacillin and www.nature.com/scientificreports www.nature.com/scientificreports/ clindamycin lost substantial activity against resistant bacteria, such as MRSA (methicillin-resistant S. aureus) and MDR (multidrug-resistant) S. pneumoniae. Dapsone, an aniline sulfone first made in 1908 but discovered as an antimicrobial agent in 1937 25 , was generally less effective than the other antibiotics but had widely varying activity that was dependant on the species (ranging from <2 µg/mL for S. pyogenes to >4100 µg/mL against a S. epidermidis strain, with the variable activity potentially partly due to poor solubility when diluting from stock solutions into media at high concentrations). Previous literature reports for broth Minimum Inhibitory Concentration (MIC) potency of tetracycline, erythromycin and clindamycin against P. acnes also showed a wide variation against 16 strains, with activity ranges of ≤0.06 to 31, ≤0.25 to >1000, and ≤0.125 to >500 µg/mL respectively for the three antibiotics 26 , with results from the current study generally fitting into these ranges. In sharp contrast, the 'non-antibiotic' acne agents (salicylic acid, azelaic acid and benzoyl peroxide) that are believed to help treat acne by multiple mechanisms, including bacterial inhibition, had substantially lower, but measurable, activity, compared to true antibiotics. Their potency, generally ranging from 2000-64,000 µg/mL, was approximately 1000-fold less active than the designated antibiotics. However, their activity was maintained against all of the resistant bacteria tested, including highly resistant strains of S. aureus, S. epidermidis, and S. pneumoniae where almost all the antibiotics failed.
Previous reports of the direct antimicrobial activity of salicyclic acid are limited, with disc diffusion measurements of activity in 1962 versus E. coli, Aerobacter aerogenes, Leuconostoc mesenteroides P-60, S. aureus, 'Streptococcus faecalis' [sic] and five fungi 27 . In 2014 the MIC and Minimum Bactericidal Concentration (MBC) of salicylic acid and other phytochemicals were assessed against E. coli (MIC = 3200 µg/mL) and S. aureus (MIC = 1600 µg/mL) 28 , compared to MIC = 16000 µg/mL and 32000-64000 µg/mL, respectively in this study. A 2007 study showed 5 mM salicylate (approx. 700 µg/mL) halted growth of SH1000 S. aureus after 5h 29 , though the same concentration only slightly slowed the growth of E. coli GC4468 30 . A 2011 article on new antimicrobial formulations compared their activity against P. acne to salicyclic acid, with MIC 90 for salicyclic acid of 1000 µg/mL 26 , compared to 8000 µg/mL in this study. A review of the effects of salicylate on bacteria was published in 2000 31 , which summarized research showing that, at concentrations that do not substantially affect bacterial growth, salicylate can: (a) induce antibiotic resistance, (b) reduce resistance to some antibiotics; and (c) affect production of bacterial virulence factors. More recent studies have supported the reduction in susceptibility of organisms such as S. aureus 29 or Salmonella enterica serovar Typhimurium 32 to common antibiotics or antiseptics in the presence of salicylate. Further studies are warranted to see if topical use of salicylate for acne reduces the effectiveness of topical acne antibiotics.
The antimicrobial potential of azelaic acid has been more thoroughly studied than that of salicyclic acid, with a review in 1993 33 . The first observation that it exerted a bacteriostatic effect on aerobic and anaerobic bacteria (including Propionibacterium) appeared as a comment in a 1983 clinical report 34 . A clinical trial noted a 224-fold reduction in the population of Micrococcaceae and 30-fold decrease in the density of Propionibacterium sp. on the skin after application of 20% azelaic acid cream (compared to no effect from tetracycline) 35 . Another report measured MIC in broth at pH 6.0 against S. epidermidis, S. capitis, and S. hominis (125 mM, approximately 23,500 µg/ mL, similar to our values of 8000-16,000 µg/mL), P. acne and P. granulosum (>250 mM ≈ >47,000 µg/mL, versus 8000-16,000 µg/mL in this study), Propionibacterium avidum (31 mM ≈ 5900 µg/mL) and Pityrosporum ovale (now known as Malassezia ovale) (250 mM ≈ 47,000 µg/mL) 36 . In 1991 concentrations of 500 mM (≈ 94,000 µg/ mL) were reported to exert bactericidal activity against P. acne in vitro at pH 6.0, with activity enhanced by lowering the pH to 5.6 but little activity at pH 7.0 37 . A 1992 report compared the in vitro activities of the topical antimicrobials azelaic acid, nitrofurazone, silver sulphadiazine and mupirocin against MRSA 38 . Against 80 MRSA strains, the MIC 50 and MIC 90 of azelaic acid, measured by agar dilution, were 850 µg/mL and 1150 µg/ mL respectively (no pH mentioned), with a range of 600-1200 µg/mL 38 , around 10-fold less than our BMD MIC values (8000-16,000 µg/mL). The corresponding MBC values were 1800 µg/mL and 3500 µg/mL respectively. Azelaic acid was slowly bactericidal at 2500 µg/mL, with around 3-log reduction from a starting inoculum of 10 6 cfu after 24 h; a resistance mutation rate of <1 × 10 −9 was observed 38 . The authors of the 1993 review also noted in the review that they had conducted an in vitro experiment to assess the development of resistance in P. acnes or S. epidermidis over 53 days exposed to 2-4 mM (400-800 µg/mL) azelaic acid, with no changes in MIC detected 33 .
Finally, benzoyl peroxide has long been known to have antimicrobial properties, with speculation of antiseptic action in the 1920's and treatment of acne/skin lesions in the 1930's 39 . The history of its application for the treatment of acne was reviewed in 1987 40 and 2009 39 . Survival curves of S. epidermidis, S. capitis, S. hominis, P. acne, P. granulosum, P. avidum and P. ovale have been measured in the presence of 10 −2 -10 −4 w/v% benzoyl peroxide, with bacteria showing varying sensitivity but all killed at the higher concentrations 41 . Another study looked at 10 sensitive and 10 erythromycin resistant strains of P. acne, P. granulosum, P avidum, and 10 sensitive and 10 erythromycin resistant strains of S. epidermidis, with benzoyl peroxide agar dilution MIC of 64-128 µg/ mL and 512 µg/mL respectively 42 (compared to BMD MIC of 2048 µg/mL in this study; their benzoyl peroxide parent solution had 5% w/w benzoyl peroxide but also contained carbomer 940, 14% alcohol, sodium hydroxide, dioctylsodium sulphosuccinate and fragrance). In 1989 MICs against nine P. acne strains were reported to be between 100-800 µg/mL 43 using a modified broth with added 2%Tween and glycerol to improve benzoyl peroxide  www.nature.com/scientificreports www.nature.com/scientificreports/ solubility, a 2 × 10 4 innoculum, and four day incubation (compared to 2048 µg/mL in this study with BHI broth, 5 × 10 5 innoculum, and 48 h incubation). It was also not clear what form of benzoyl peroxide was used in the 1989 report, as it was "obtained from commercial products" so likely contained other components. More potent BMD MICs of 62.5, 15.6 and >100 µg/mL were reported against P. acne, S. aureus and S. epidermidis in 2009, again employing different assay conditions from our study that included varied incubation times (72, 24 and 48 h respectively) 44 . A comparison of the activity of new antimicrobial formulations against P. acne used benzoyl peroxide as a standard, with MIC 90 for benzoyl peroxide of 50 µg/mL 26 (compared to 2048 µg/mL determined in this study). A 2016 study assessed the activity of benzoyl peroxide against 44 clinical isolates of P. acne using the Decker modified broth, with MIC 50 = 128 µg/mL and MIC 90 = 256 µg/mL. MBCs were similar to MICs, and a time kill assay showed 5-log reduction in cfu after 1 h at two-fold MIC 45 .
In summary, this study clearly demonstrates that acne agents used primarily for their skin exfoliating properties do indeed have modest, but widespread, antimicrobial activity against a range of skin-associated bacteria, at least when tested in broth microdilution assays. Many skin-related bacteria can form biofilms, which are notoriously more resistant to antimicrobial therapies than vegetative bacteria. The exfoliant topical agents are generally applied at concentrations up to 20-fold higher than topical antibiotics (though in some cases at equivalent concentrations), so they are likely to exert substantial antimicrobial effects despite their reduced antimicrobial potency. Benzoyl peroxide is used as 2.5-10% solutions in gel, cream, lotions or liquid 46 , azelaic acid as 15-20% lotions 46 , and salicylic acid in a range of concentrations (with 0.5-2% commonly used, but up to 10% employed for acne treatments: 2% is the maximum strength allowed in over-the-counter acne products in the United States). Clindamycin, erythromycin and tetracycline topical treatments are generally in the 1-4% range [43][44][45][46][47] , with dapsone used in a 5% gel 46 . The retention of high levels of antimicrobial activity by salicylic acid, azelaic acid and benzoyl peroxide against antibiotic-resistant strains of bacteria suggests that these treatments could be useful alternatives to antibiotic-based therapies in the case of resistant bacteria, and should be further explored as preferred alternatives to prescribed antibiotics to help reduce the development of resistance.

Methods compound preparation. Stock solutions of compounds were prepared in different solvents at different
concentrations, depending on solubility and expected activity range, as presented in Table 4.

Minimum inhibitory concentration (Mic) determinations.
Bacterial strains were purchased from the American Type Culture Collection (ATCC) or Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) (see Table 5).   www.nature.com/scientificreports www.nature.com/scientificreports/ Standard anaerobic Mic Assay. The MIC assay for anaerobic growth conditions was performed to the same procedure as the standard aerobic MIC assay described above with the following exceptions: All steps were performed in a COY type B anaerobic chamber with the anaerobic atmosphere controlled by the introduction of 10%CO 2 /5% H 2 in N 2 CoA gas mix, catalyst Stak-Pak and O 2 -H 2 gas analyzer, with H 2 levels kept at ~2% for the duration of the assay. Brain Heart Infusion (BHI) (OXOID CM1135B) media with 1% cysteine to further promote an anaerobic environment was used in replacement of MHB, and this broth was incubated in the anaerobic chamber for 24 h prior to use to allow sufficient atmosphere exchange. All the plates were covered and incubated at 37 °C for 48 h. MICs were determined as the lowest concentration showing no visible growth by eye.