Glycerol Monolaurate Antibacterial Activity in Broth and Biofilm Cultures

Background Glycerol monolaurate (GML) is an antimicrobial agent that has potent activity against gram-positive bacteria. This study examines GML antibacterial activity in comparison to lauric acid, in broth cultures compared to biofilm cultures, and against a wide range of gram-positive, gram-negative, and non-gram staining bacteria. Methodology/Principal Findings GML is ≥200 times more effective than lauric acid in bactericidal activity, defined as a ≥3 log reduction in colony-forming units (CFU)/ml, against Staphylococcus aureus and Streptococcus pyogenes in broth cultures. Both molecules inhibit superantigen production by these organisms at concentrations that are not bactericidal. GML prevents biofilm formation by Staphylococcus aureus and Haemophilus influenzae, as representative gram-positive and gram-negative organisms, tested in 96 well microtiter plates, and simultaneously is bactericidal for both organisms in mature biofilms. GML is bactericidal for a wide range of potential bacterial pathogens, except for Pseudomonas aeruginosa and Enterobacteriaceae. In the presence of acidic pH and the cation chelator ethylene diamine tetraacetic acid, GML has greatly enhanced bactericidal activity for Pseudomonas aeruginosa and Enterobacteriaceae. Solubilization of GML in a nonaqueous delivery vehicle (related to K-Y Warming®) enhances its bactericidal activity against S. aureus. Both R and S, and 1 and 2 position lauric acid derivatives of GML exhibit bactericidal activity. Despite year-long passage of Staphylococcus aureus on sub-growth inhibitory concentrations of GML (0.5 x minimum bactericidal concentration), resistance to GML did not develop. Conclusions/Significance GML may be useful as a broad-spectrum human or animal topical microbicide and may be useful as an environmental surface microbicide for management of bacterial infections and contamination.


Introduction
Most bacterial pathogens initiate human illnesses from intact or damaged mucosal or skin surfaces. Many of these pathogens are acquired from other persons or animals, from endogenous sources, or from a myriad of environmental sources. Once in humans, pathogens colonize surfaces primarily as biofilms of organisms, defined as thin-films of organisms attached to host tissues, medical devices, and other bacteria through complex networks of polysaccharides, proteins, and nucleic acids [1]. These bacteria may also exist as planktonic (broth) cultures in some host tissue environments, such as the bloodstream and mucosal secretions. Similarly, these potential pathogens may exist as either biofilms or planktonic cultures in a myriad of non-living environments [1]. A major objective of the present studies is to assess the ability of a fatty acid monoester, glycerol monolaurate (GML), to inhibit growth and exotoxin production by bacteria as grown in biofilms and planktonic cultures.
GML is a generally recognized as safe natural compound by the Food and Drug Administration (FDA) that exhibits potent antibacterial activity against gram-positive cocci, and Bacillus anthracis [2,3]. Unlike most antibiotics which have single bacterial targets for antibacterial activities, GML appears to target many bacterial surface signal transduction systems nonspecifically through interaction with plasma membranes [2,3,4]. GML also inhibits exotoxin production by gram-positive bacteria at GML concentrations that do not inhibit bacterial growth [2,3,4]. These properties are shared with the antibiotic clindamycin, a protein synthesis inhibitor [5]. GML is also virucidal for enveloped viruses, apparently through its ability to interfere with virus fusion with mammalian cells, and through GML's ability to prevent mucosal inflammation required for some viruses to penetrate mucosal surfaces [6,7,8].
In 2005, Ruzin and Novick suggested that the ability of GML to inhibit exotoxin production by Staphylococcus aureus resulted from its cleavage to lauric acid [9]. These authors demonstrated that lauric acid was identical in activity to GML. These data are surprising for at least two reasons: 1) GML can be degraded by esterases of Staphylococcus aureus, including the major esterase glycerol ester hydrolase, to glycerol and lauric acid. When degraded to lauric acid, GML loses antibacterial activity and its ability to inhibit exotoxin production [2]; and 2) many bacteria, such as streptococci, do not make esterases and thus cannot cleave GML to lauric acid. Yet, GML is even more highly antibacterial to these organisms, with highly significant ability to inhibit exotoxin production at non-growth-inhibitory concentrations [2]. A second major objective of our study is to evaluate GML and lauric acid in side-by-side comparisons for bactericidal activity and their ability to prevent exotoxin production at non-growth-inhibitory concentrations.
Multiple forms of GML exist or can be synthesized. Previously, we tested dodecyl glycerol (DDG, glycerol ether-linked to lauric acid) compared to GML for antibacterial activity and ability to inhibit exotoxin production independently of bactericidal activity [10]. Both DDG and GML inhibit Staphylococcus aureus growth, with both being either bacteriostatic or bactericidal dependent concentration. However, GML inhibits exotoxin production at non-growth-inhibitory concentrations, whereas DDG does not inhibit exotoxin production independent of growth inhibition [10]. Both R and S forms of GML and GML with lauric acid in the 1/3 and 2 positions exist. These molecules have not been tested for antimicrobial activity but are tested in this study.
We previously showed that Escherichia coli and Salmonella minnesota with their native lipopolysaccharide (LPS) layer cannot be inhibited from growing due to GML [2]. In contrast, a Salmonella Minnesota Re mutant, lacking the O side chain and much of the common core polysaccharide components of LPS, are easily killed by GML. This led us to propose that other gram-negative bacteria that have lipooligosaccharide (LOS) instead of Enterobacteriaceae intact LPS may be susceptible to GML. In vitro studies by us and others have shown that both Gardnerella vaginalis and Helicobacter pylori, with LOS instead of intact LPS, are highly susceptible to the bactericidal activity of GML [11]. These studies have been confirmed with in vivo studies in humans with demonstration of Gardnerella vaginalis susceptibility to GML [11].
The above studies with Salmonella minnesota suggest that agents that disrupt the LPS layer may increase the activity of GML. Such agents may include protonation of the bacterial surface by reduced pH to repel calcium and magnesium in the LPS layer, divalent cation chelators such as ethylene diamine tetraacetic acid (EDTA), and biocompatible non-aqueous GML delivery vehicles (such as K-Y WarmingH gel) that intrinsically would interfere with the bacterial inner and outer membranes and LPS. Studies were undertaken to assess the efficacy of low pH and EDTA in amplification of GML activity against highly resistant Enterobacteriaceae and Pseudomonas aeruginosa, and a non-aqueous delivery vehicle for amplification of activity against S. aureus.
Our data from the present studies demonstrate that GML is bactericidal for aerobic and anaerobic gram-positive bacteria in broth and biofilm cultures, GML exhibits greater bactericidal activity than lauric acid, and all forms of GML exhibit antibacterial activity. Additionally, GML is bactericidal for gram-negative bacteria with LOS instead of LPS, but GML becomes bactericidal for naturally GML-resistant Enterobacteriaceae by addition of agents that disrupt the LPS layer. Gram-negative anaerobes are susceptible to GML. Pseudomonas aeruginosa appear to be the most resistant bacteria tested, but these organisms are killed by GML at pH 5.0-6.0.

Comparison of GML and Lauric Acid Antimicrobial Activity
Previously, we demonstrated that b-hemolytic streptococci, lacking glycerol ester hydrolase (GEH), exhibit greater susceptibility to GML than Staphylococcus aureus which produce GEH [2]. This suggests that GML has greater antibacterial activity than lauric acid which with glycerol are the two major cleavage products of GML. However, a publication by Ruzin and Novick in 2000 suggests that the ability of GML to inhibit exotoxin production by Staphylococcus aureus is equivalent to inhibition of exotoxin production by lauric acid [9]. We thus compared GML and lauric acid for their antibacterial activity and ability to inhibit exotoxin production at concentrations that do not inhibit bacterial growth.
GML was bactericidal at 200-fold lower concentrations than lauric acid when tested against Staphylococcus aureus MN8, a typical menstrual toxic shock syndrome (TSS) strain. GML was bactericidal, defined in this study as a $3 log reduction in CFUs/ml relative to starting inocula, at 0.25 mM, whereas lauric acid was bactericidal at 50 mM ( Figure 1). When GML and lauric acid were tested against a scarlet fever strain of Streptococcus pyogenes (strain T25 3 curedT12), an organism that lacks GEH, GML exhibited 5-fold (0.05 mM) greater bactericidal activity ( Figure 2) than when tested against Staphylococcus aureus MN8 (Figure 1). In contrast, there was no difference in bactericidal activity of lauric acid for either Streptococcus pyogenes or Staphylococcus aureus MN8 (Figures 1 and 2).
Pretreatment of GML (1000 mg/0.4 ml Todd Hewitt broth) with 0.1 of stationary phase sterile culture fluid from Staphylococcus aureus MN8 overnight at 37uC eliminated its antibacterial activity against Staphylococcus aureus MN8 (Figure 3), whereas comparable treatment with Streptococcus pyogenes did not reduce GML antimicrobial activity.
Both GML and lauric acid significantly inhibited superantigen production by Staphylococcus aureus MN8 and Streptococcus pyogenes at  sub-growth inhibitory concentrations ( Figure 4). However, the GML concentration that was required for inhibition of exotoxin production in absence of growth inhibition was lower for both organisms than lauric acid.
Collectively, these data indicate that both GML and lauric acid have antibacterial activity and ability to inhibit exotoxin production at concentrations that do not inhibit bacterial growth. However, GML exhibits greater antibacterial activity than lauric acid.

GML Bactericidal Activity in Biofilms
Many bacteria that are found on surfaces grow as biofilms [1], forming complex extracellular matrices (example in Figure 5). In the assay shown, a regular absorbency tampon was placed within dialysis tubing inoculated with community-associated (CA) methicillin-resistant Staphylococcus aureus (MRSA) 128 (TSST-1 + ), creating a tampon sac. [12] This tampon sac was submerged beneath Todd Hewitt broth containing 0.8% agar for 16 hours and then evaluated for biofilm formation on both tampon fibers ( Figure 5A) and dialysis tubing ( Figure 5B). Staphylococcus aureus 128 grew as complex biofilms on both the tampon fibers and on the dialysis tubing. It is our experience that cellulose acetate provides one of the best substrates for biofilm formation.
A previous study suggested that pulsed-field gel electrophoresis type USA200 strains of Staphylococcus aureus are attenuated in virulence in part due to a mutation in the accessory gene regulator (agr) global regulatory system [13]. These strains are the common causes of menstrual TSS and include strains MN8 and 128. In previous tests of over 5000 such USA200 strains, all produced 3.5 to 20 mg/ml of TSST-1 in broth cultures (unpublished data). It is estimated that 0.1 mg of TSST-1 is sufficient to cause TSS, indicating that these strains produce 350-2000 TSS-inducing doses/ml culture [14]. We evaluated the ability of Staphylococcus aureus 128 to produce TSST-1 in biofilms as grown in tampon sacs ( Figure 6). In these cultures by 4 hours post infection, TSST-1 amounts exceeded 1300 mg/ml (13,000 TSS inducing doses/ml) and by 16 hours were at 18,000 mg/ml (180,000 TSS-inducing doses/ml). By 16 hours the tampons had absorbed 5.5 ml of medium, indicating 99,000 mg total TSST-1 (990,000 TSS-inducing doses). Comparatively, Staphylococcus aureus 128 produces 5 mg/ml in overnight Todd Hewitt broth cultures. Collectively, these data indicate that despite a mutation in agr, TSST-1 is produced in exceptionally high amounts in biofilms. This study has been repeated with strains MN8 and CDC587 with   . Comparison of GML and lauric acid for capacity to inhibit superantigen production. Staphylococcus aureus MN8 and Streptococcus pyogenes were cultured for 8 hours in the presence of GML or lauric acid. Toxic shock syndrome toxin-1 (TSST-1) production by Staphylococcus aureus MN8 and streptococcal pyrogenic exotoxin A (SPE A) production by Streptococcus pyogenes were quantified by Western immunoblot analysis. * indicates mean significantly reduced from the no GML mean at p,0.01. doi:10.1371/journal.pone.0040350.g004 comparable results. In tampon sacs that contained 5% GML (w/ w), no TSST-1 was detected (data not shown).
The most straightforward way to quantify biofilm formation is the use of 96 well microtiter plates incubated stationary at 37uC with test bacteria and then assaying microbial components retained on the wells after washing [15]. This assay was used to assess the ability of GML to interfere with biofilm formation by three strains of Staphylococcus aureus ( Figure 7) and a non-typable Haemophilus influenzae ( Figure 8). All three Staphylococcus aureus strains were completely growth-inhibited by GML only at 500 mg/ml at both 24 and 48 hours. In contrast, at 10 times lower GML concentrations than necessary to inhibit bacterial growth, biofilm formation was significantly inhibited as measured by reduced crystal violet straining of retained biofilms materials in wells of the microtiter plates. GML also prevented biofilm formation by nontypable Haemophilus influenzae ( Figure 8). GML was bactericidal at GML concentrations of 50 mg/ml, but significantly prevented biofilm formation at concentrations of 1.0 mg/ml.
Side-by-side wells that were not treated with GML and that had high absorbances at 595 nm at 48 hours received GML (500 mg/ ml) for 60 min at 37uC. Subsequently, CFUs/ml were determined, after agitation with a pipette wells were then washed three times with PBS, and crystal violet straining determined to assess if GML could removed previously formed biofilms and simultaneously kill the microbes. After 48 hours, GML (500 mg/ml) was highly effective in both killing Staphylococcus aureus MN8 and non-typable Haemophilus influenzae by 60 min addition to the cultures ( Figure 9A). Similarly, GML (500 mg/ml) nearly completely removed biofilm material attached to the wells after 48 hours as noted by loss of crystal violet straining ( Figure 9B).

Lack Resistance to GML by Staphylococcus aureus
Antibiotic resistance in bacteria may occur with treatment of patients with sub-optimal antibiotic concentrations or with prolonged antibiotic therapy. We tested Staphylococcus aureus MN8 for one year with sub-inhibitory (0.5 x minimum bactericidal concentration [MBC]) GML concentrations (50 mg/ml) to evaluate the potential to develop resistance. Our hypothesis was that GML has many plasma membrane targets, including interference with signal transduction, and with multiple GML bacterial targets, resistance to GML is unlikely. We chose to study Staphylococcus aureus for two reasons: 1) the organism has demonstrated rapid development of resistance to many antibiotics, and 2) the organism produces GEH [2], where up-regulation of GEH production could lead to resistance. Each week Staphylococcus aureus MN8 was transferred to GML Todd Hewitt agar plates containing 50 mg/ml GML (0.5 MBC) and cultured 48 hours; Staphylococcus aureus MN8 will grow on the plates in 48 hours. Colonies (50-200) that grew on the Todd Hewitt agar plates were scraped off and passed weekly onto new Todd Hewitt plates containing GML (50 mg/ml). Each week the organisms that grew on the GML (50 mg/ml) plates were also transferred onto GML (100 mg/ml) (16MBC) plates at Figure 5. Community-associated Staphylococcus aureus 128 growing as biofilms on tampon fibers and cellulose acetate dialysis tubing. Staphylococcus aureus 128, a TSST-1 + organism at 10 7 /0.1 ml, was inoculated onto the inside of pre-wetted dialysis tubing that had been tied off on one end. A tampon was then inserted into the dialysis tubing and the tubing immersed under Todd Hewitt broth containing 0.8% agar. The open end of the dialysis tubing remained above the agar surface such that the only source of nutrients for the growing microbes was media absorbed across the dialysis tubing. Magnification of A was 6000x and B was 9000x. Both frames show organisms embedded in a complex matrix of extracellular material. doi:10.1371/journal.pone.0040350.g005 Figure 6. TSST-1 production in biofilms in tampon sacs. Staphylococcus aureus MN8 (10 7 /0.1 ml) were placed inside dialysis tubes that had been tied off on one end, and then tampons were inserted. The tampon sacs were immersed beneath Todd Hewitt 0.8% agar such that the open end of the dialysis tubing remained above the surface. The agar was allowed to solidify, and the tampons sacs were incubated for indicated times. TSST-1 was measured within the dialysis tube by knowing the pre-and post-weight of the tampons, eluting TSST-1 with distilled water, and quantifying TSST-1 by Western immunoblot analysis. doi:10.1371/journal.pone.0040350.g006 approximately 1610 9 CFU/plate, where the organism cannot grow in 48 hours. Over the entire year of testing, no GMLresistant Staphylococcus aureus MN8, that were able to grow on the Todd Hewitt plates containing GML (100 mg/ml), were obtained. Additionally, the plates that had 50 mg/ml GML weekly were also placed at 4uC to allow GML to crystallize. Through measurement of non-crystallizing zones around individual colonies, which is reflective of GEH cleavage of GML, no increased zones of GEH cleavage of GML were seen at any time point.

Tests of Bactericidal Activities of R Versus S and 1 Versus 2 Position Lauric Acid Forms of GML
Multiple forms of GML exist, including R versus S optical isomers and GML with lauric acid ester linked in the 1/3 or 2 position of glycerol. These compounds were tested in the following way. The R form of GML was available, but the S form was not. Thus we compared the purified R form to the mixture of R and S. Similarly, we purchased the purified 2 position lauric acid form of GML, and thus we were able to compare this form to the GML R and S mixture. These GML forms were evaluated for activity against both group A and group B streptococci as two highly sensitive organisms to GML antibacterial activity. For Streptococcus pyogenes (strain 594) ( Figure 10) and Streptococcus agalactiae (strain MNSI) (data not shown), both the R form and mixture of R and S forms had the same antibacterial activity. The GML form with lauric acid in the 2 position was 2-fold more active than the mixture of GML forms ( Figure 10). The data suggest that GML activity does not depend on chirality, but in a minor way depends on the position of the lauric acid.

Spectrum of GML Antibacterial Activity
We examined a large collection of bacterial isolates for growth inhibition by GML (Table 1). Both gram-positive and gramnegative bacteria were inhibited from growing under optimal growth conditions for each organism tested. Only two major groups of potentially pathogenic organisms were not inhibited by GML, the large group of Enterobacteriaceae and Pseudomonas aeruginosa, even at concentrations of 5000 mg/ml. Interestingly, Burkholderia cenocepacia, which used to be named Pseudomonas cepacia and is related to Pseudomonas aeruginosa, was killed by GML at Figure 7. Effect of GML on three Staphylococcus aureus strains cultured in biofilms. Staphylococcus aureus strains methicillin-sensitive MN8, community-associated methicillin-resistant MNWH, and community-associated methicillin-resistant MW2 were cultured for 24 and 48 hours in 96 well microtiter plates. In one set of three wells for each microbe, the wells were agitated 3 times by pipetting up and down. CFU/ml were determined on supernates then removed from the wells. * indicates significant mean reduction compared to starting inoculum mean at p,0.001. Dashed line indicates starting inoculum size. Subsequent to removing unbound cells and then washing wells three times with phosphate-buffered saline, the wells were treated with crystal violet for 30 min. The wells were next washed three times with phosphate-buffered saline to remove unbound crystal violet. Finally, wells were treated with ethanol to solubilize biofilm-associated crystal violet. Absorbances at 595 nm were determined by an ELISA reader. # indicates significant mean reduction in absorbance at 595 nm wavelength between no GML and GML-treated wells at p,0.01. doi:10.1371/journal.pone.0040350.g007 concentrations of 500 mg/ml. Mycobacterial species typically produce large amounts of complex fatty acids [16]. However, these organisms were killed by GML at concentrations of $50 mg/ ml. In addition to inhibiting the growth gram-positive bacteria, GML inhibited exotoxin production independently from inhibition of growth for all such organisms tested (Staphylococcus aureus, Streptococcus pyogenes, Streptococcus agalactiae, groups C, F, and G streptococci, and Clostridium perfringens). The most susceptible Figure 8. Effect of GML on non-typable Haemophilus influenzae cultured in biofilms. Haemophilus influenzae was cultured for 24 and 48 hours in a 96 well microtiter plate. In one set of three wells for the microbe, the wells were agitated 3 times by pipetting up and down and then supernates removed for plate counting to determine CFUs/ml. * indicates significant mean reduction compared to starting inoculum mean at p,0.001. Dashed line indicates starting inoculum size. Subsequent to removing bacterial cells and washing three times with phosphate-buffered saline, the wells were treated with crystal violet for 30 min. The wells were then washed three times with phosphatebuffered saline to remove unbound crystal violet. Finally, the wells were treated with ethanol to solubilize biofilm-associated crystal violet. Absorbances at 595 nm were determined by an ELISA reader. # indicates significant mean reduction in absorbance at 595 nm wavelength between no GML and GML-treated wells at p,0.01. doi:10.1371/journal.pone.0040350.g008 Figure 9. Ability of GML to remove pre-formed biofilms. Staphylococcus aureus MN8 and nontypable Haemophilus influenzae were grown for 48 hours in 96 well microtiter plates to allow biofilm formation. CFUs/ml were determined after suspension of organisms (A). Subsequently, the wells were treated or not treated with GML (500 mg/ml) for 60 min, and then CFUs/ml determined again (A). * indicates significant mean reduction compared to starting inoculum mean at p,0.001. Dashed line indicates starting inoculum size. Wells were stained with crystal violet after removal of non-adherent bacteria or after removal of non-adherent bacteria and treatment with GML, followed by removal of unbound crystal violet and determination of absorbance at 595 nm (B). # indicates significant mean reduction in absorbance at 595 nm wavelength between no GML and GMLtreated wells at p,0.01. doi:10.1371/journal.pone.0040350.g009 organisms to killing by GML were Peptostreptococcus species, Clostridium perfringens, Bordetella bronchiseptica, and Campylobacter jejuni, all being killed by GML (1 mg/ml).

Use of accelerants to increase GML activity against Enterobacteriaceae and Pseudomonas aeruginosa
Our prior studies suggested that the intact LPS layer of Enterobacteriaceae protected this family of organisms from GML [2]. Thus, we attempted to partially disrupt the LPS layer in these organisms and evaluate GML activity. In the first set of studies we tested the effect of EDTA on GML (100 mg/ml) activity against Escherichia coli (Figure 11). GML as expected exerted no measurable antibacterial activity against Escherichia coli at concentrations up to 5000 mg/ml (data not shown). EDTA alone exhibited bacteriostatic and bactericidal activity against Escherichia coli, dependent on EDTA concentration. The combination of GML with EDTA showed increased activity when used in combination with GML at a concentration of 100 mg/ml.
Because EDTA made Escherichia coli more susceptible to GML than in its absence, we hypothesized that protonating the Escherichia coli surface may increase GML activity through repelling divalent cations. Escherichia coli was highly susceptible to GML at pH of 6, with GML (50 mg/ml) being bactericidal, despite not being susceptible at pH 7.0 to even 100 times the GML concentration ( Figure 12). Escherichia coli was even more susceptible to GML at pH of 5, with GML (0.1 mg/ml) being bactericidal.
With each unit drop in pH, it appeared that Escherichia coli became 500 times more susceptible to GML. We also tested GML susceptible gram-negative bacteria, such as Haemophilus influenzae nontypable organisms, and these organisms were similarly 500fold more susceptible to GML with each unit drop in pH (data not shown). We also tested Pseudomonas aeruginosa for the effect of pH on GML activity against this organism ( Figure 13). Pseudomonas aeruginosa was not susceptible to GML at either pH 6.0 or 7.0. However, the organism did not grow well at pH 5.0, but GML (50 mg/ml) at pH 5.0 was highly bactericidal for Pseudomonas aeruginosa.
It is unlikely that GML will be used as an antibacterial agent as suspended in aqueous solutions do to its solubility limit of 100 mg/ ml in aqueous solutions at 37uC. However, our experience has been that GML exhibits antibacterial activities even at concentrations that exceed 100 mg/ml, despite the lack of solubility in aqueous solutions at that concentration [2]. For ease of use, we developed a non-aqueous, biocompatible delivery system that could be used to solubilize GML up to concentrations of 50,000 mg/ml. This delivery system is related to K-Y WarmingH gel, containing 73.55% (w/w) propylene glycol, 25% (w/w) polyethylene glycol 400 NF, and 1.25% (w/w) hydroxypropyl cellulose. This solution has been shown to be safe for humans and non-human primates in chronic use studies for up to six months [17]. The gel had no deleterious effect on normal flora lactobacilli, but did effectively reduce both Candida and Gardnerella vaginalis [11]. We tested this delivery vehicle with GML for ability to kill Staphylococcus aureus MN8 ( Figure 14). Both delivery vehicle alone and GML in delivery vehicle were bactericidal, preventing determination of potential synergy between vehicle and GML. We therefore tested GML activity when vehicle alone and GML in vehicle were diluted with various volumes of Todd Hewitt broth. GML in the 25% and 10% non-aqueous delivery vehicle had significantly greater antibacterial activity than vehicle alone (no activity) or GML alone diluted comparably in Todd Hewitt broth. GML in 25% non-aqueous delivery vehicle had approximately 5000-fold greater activity than GML alone, and GML in 10% aqueous delivery vehicle had approximately 10-fold greater activity than GML alone.

Discussion
We performed extensive studies of the potential ability of GML to be used as a topical bactericidal agent and biocide for decontamination of environmental sources of potential pathogens. These studies show that GML has potent bactericidal activity against a myriad of organisms that cause human illnesses. Previous studies demonstrated that GML has strong activity against gram- Figure 11. Ability of EDTA to synergize with GML in antibacterial activity against Escherichia coli. Escherichia coli Watson strain was incubated with EDTA or EDTA + GML (100 mg/ml) for 24 hours and then plate counts used to determine CFUs/ml. GML (5000 mg/ml) alone did not inhibit Escherichia coli growth. * indicates mean for EDTA + GML differs significantly from mean for EDTA alone at p,0.001. Dashed line indicates starting inoculums size. doi:10.1371/journal.pone.0040350.g011  positive bacteria, notably gram-positive cocci [2,4]. Our studies confirm and extend those previous findings to include additional organisms, including anaerobes. Additionally, prior studies demonstrated that GML inhibited exotoxin production by grampositive bacteria at GML concentrations that were not growthinhibitory [2,4]. We also confirmed those findings but extended the studies to include inhibition of exotoxin production by pathogenic clostridia.
There is general agreement in the literature that the major effect of GML to inhibit gram-positive exotoxin production is prevention of signal transduction through two-component systems to interfere with transcription [2,3,4,9,18]. This has been shown for TSST-1 and a-toxin in vitro and in vivo and has led to incorporation of GML into certain tampons to reduce the risk of TSS. Exotoxin production by Bacillus anthracis is also inhibited by GML, independent of inhibition of bacterial growth, and this inhibition has been suggested also to depend on interference with twocomponent system signal transduction [3]. In studies by Ruzin and Novick it has been suggested that the ability of GML to inhibit exotoxin production by GML may in fact be due to lauric acid, as both compounds have this effect, and GML can be cleaved by GEH to lauric acid [9]. We have confirmed that both GML and lauric acid have this activity. However, our studies show that GML inhibits exotoxin production by organisms that do not make GEH, indicating GML does not require cleavage to lauric acid for activity. In addition, GML inhibits exotoxin production at doses that are 200-fold lower than lauric acid. Thus, the major antibacterial effect of GML is much greater than lauric acid, but both molecules appear to alter signal transduction to inhibit exotoxin production. The exact two-component systems that are most sensitive to GML's ability to inhibit exotoxin production remain unclear. At this time, it has been shown that GML affects the agr regulatory system, but this is only one of many two component system that regulates exotoxin production [4].
It also remains unclear how GML kills bacteria in less than 15 min, but the effect almost certainly depends on interaction with the bacterial plasma membrane since we detect GML associated with bacterial plasma membranes but not cytoplasm (unpublished data). There are multiple possibilities to explain the killing effect that include interference with two-component systems. S. aureus has 16 identified two-component systems, and one of these, designated WalK/R, is essential for microbial viability [19]. As noted above, at least agr is inhibited by GML, and this effect likely contributes to inhibition of exotoxin production. If two-component systems differentially are inhibited by GML, it is possible that inhibition of exotoxin occurs at low GML doses after systems such as agr are affected, and then at high GML doses, WalK/R is inhibited, leading to death of the organisms.
Although direct inhibition of two-component systems is a possible mechanism of GML killing of bacteria, there are other important possibilities, including indirect inhibition of the same two-component systems. Among these possibilities are dissipation of the bacterial plasma membrane potential and pH gradients across the membranes. A novel class of agents that have these effects include tetramic acids, for example those produced by Pseudomonas aeruginosa and certain lactobacillus strains [20,21,22]. Tetramic acids made by these organisms contain a 2,4 pyrrolidinedione ring and a 12 carbon side chain. Their properties include broad spectrum antibacterial effects and anti-inflammatory activities, in many ways paralleling those of GML. It is possible the similarities between tetramic acids and GML may explain why Pseudomonas aeruginosa and lactobacilli are highly resistant to GML antimicrobial activities. For Pseudomonas aeruginosa, tetramic acids are important for the homoserine lactone quorum sensing system [20,21]. Interestingly, the organism as grown in the presence of high concentrations of GML (.2000 mg/ml) at pH 7.0 appears to have up-regulated production of numerous virulence factors including pigments, consistent with effects associated with activation of the quorum sensing system.
We have proposed that the major use of GML will include its function as a topical microbicide [6,11,23]. We have previously shown that GML reduces vaginal Candida and Gardnerella vaginalis in women, while at the same time not affecting normal flora lactobacilli [11]. In other studies we showed that GML reduces vaginal Staphylococcus aureus and Streptococcus agalactiae [23]. All of these organisms vaginally are likely to be growing as biofilms on tampons or on vaginal mucosal surfaces.
Biofilms present a challenge in infections due to antibiotic resistances within biofilm communities and inaccessibility of biofilms microbes to the immune system [1]. In our studies, we have demonstrated that GML interferes with biofilm formation, using as test organisms one gram-positive and one gram-negative organism. We showed GML prevents biofilm formation on tampons, cellulose acetate, which is the best surface we have found for biofilm formation, and plastic 96-well microtiter plates. Additionally, GML has the ability to loosen established biofilms from surfaces, such as on the surface of microtiter plates. Importantly, the bacteria within the detached biofilms are killed.
In the course of our studies of TSST-1 production within tampon sacs [12], models for what may happen in vivo in women with use of tampons, we demonstrated that as much as 18 mg/ml of TSST-1 may be present in the tampon sacs. With an average fluid uptake in 16 hours of 5.5 ml, this translates into nearly enough TSST-1 to cause illness in 10 6 individuals. Thus, TSST-1 is produced in extremely high concentrations by the USA200 lineage that is most associated with mucosal surface-derived TSS [24]. These data are important because they are in contrast to a recent study by DeLeo et al. that suggested that a mutation in agr reduces the virulence of USA200 strains [13]. This is clearly not the case with TSST-1 production by these strains, and it is wellestablished that TSST-1 is the principal cause of mucosal TSS, including menstrual, tampons associated illness [24,25]. It has also been established that TSST-1 production is tightly regulated in vitro by agr [26]. Thus, it appears that either agr no longer regulates TSST-1 production in USA200 strains, or the mutation observed by DeLeo et al. [13] is in a region that does not strongly affect virulence factor production. Incidentally, in vivo in used tampons from women, we have observed up to 100 mg of TSST-1 that was produced generally in only a few hours [27].
Many antibiotics have specific bacterial targets. In having their activities, this often implies that chirality is important in activity with either R or S form having major activity, but not both. In our studies, we observed that both the R and mixture of R and S forms have comparable activity that suggests that both R and S are active. If membrane-bound histidine kinases of two-component systems are affected by GML, the effect may be through alteration of conformation by altering the plasma membrane, rather than interaction with the histidine kinase itself. There must be something special about GML's interaction with the plasma membrane since it is well-recognized that GML has greater antibacterial activity than glycerol monoesters that have shorter or longer fatty acid chains [28,29]. We do not know why GML with lauric acid in the 2 position has slightly greater antibacterial activity than GML with lauric acid in the 1 or 3 position.
We have performed extensive studies, without success, in attempt to produce resistance in Staphylococcus aureus to GML. For most antibiotics, this organism develops resistance either through target modification or acquisition of a gene encoding a protein that allows the organism to bypass or inactive the antibiotic. Despite one year of passage on sub-growth-inhibitory concentrations of GML, we did not identify even a single mutant that developed resistance to GML. Likewise we did not isolate mutants that had up-regulated GEH production that inactivates GML. In prior studies in which we tested susceptibility of S. aureus to GML by a disk diffusion method, which would allow for exposures to a gradient of GML concentrations that provides selective pressure, no GML resistant mutants were seen [17]. Collectively, these data suggest that there are likely to be multiple bacterial targets for GML, such that resistance is unlikely to occur.
Our studies also show that GML has a broad range of activity against bacteria. With the exception of Enterobacteriaceae and Pseudomonas aeruginosa, all potential bacteria tested were susceptible to GML. This includes both aerobes and anaerobes, and grampositive, gram-negative, and non-gram-staining bacteria. The fact that Mycoplasma are killed by GML indicates the cell wall is not a target of GML, since these organisms lack cell walls. This is consistent with a prior in vivo study that shows that GML also can be used to kill Trichomonas and Candida, neither of which has a peptidoglycan cell wall [11]. In general, gram-positive and gramnegative bacteria without intact LPS were equivalently susceptible to GML, with those lacking GEH or a related esterase being 10fold more susceptible. It is interesting that GML kills Pasteurella multocida and Burkholderia cenocepacia but does not kill related Enterobacteriaceae and Pseudomonas, respectively. These differences suggest that LPS organization is different among the organisms, as we have shown that organisms with LOS are uniformly killed by GML, whereas those with intact LPS are resistant [2]. Of special interest is that mycobacteria are killed by GML. These organisms are characteristically difficult to kill due to the presence of complex waxes in their cell walls.
We have suggested that GML may have use as a topical microbicide, showing that the compound reduces SIV transmission in non-human primates [6]. GML has also been extensively studied in chronic (6 month) in vivo vaginal safety studies in nonhuman primates. No toxicity or vaginal inflammation in any animal was observed over the 6 month study period as evaluated by repeated colposcopy and biopsy studies [17]. We have shown that GML stabilizes mammalian cells against toxicity due to multiple bacterial exotoxins and hypotonic solutions [30]. GML also kills Gardnerella vaginalis in vitro and in vivo vaginally, an organism that predisposes women to HIV transmission [11]. Additionally, we have shown that GML stabilizes mucosal surfaces without killing epithelial cells, preventing inflammation that also predisposes to HIV transmission [6,23,30]. This has been shown both in vitro and vaginally in both non-human primates and women using tampons. There are many other potential uses for GML, but for these uses to be maximized, it may be necessary to identify agents that increase GML activity against Enterobacteriaceae and Pseudomonas aeruginosa.
As noted above both of these groups of organisms cause disease in humans, and both are highly resistant to GML. The resistance of Enterobacteriaceae to GML is clearly dependent on having an intact LPS [2]. This suggested to us that agents that disrupt the integrity of LPS would increase GML activity. Three agents were studied: 1) reduced pH that would be expected to protonate the bacterial surface and repel calcium and magnesium ions that help maintain LPS integrity, 2) use of calcium and magnesium chelators, and 3) agents disrupt the integrity of interaction of LPS with phospholipids in the outer membrane, such as nonaqueous GML delivery vehicles. All three of these agents, which are categorized as accelerants, increase GML activity. It is interesting that for each one pH unit reduction, there was a 500-fold drop in GML activity against Enterobacteriaceae such as Escherichia coli. This means that at pH 4.0 compared to pH 7.0 GML was over 10 8 more active against the organism.
It remains unclear why Pseudomonas aeruginosa is not susceptible to GML, but as noted above part of the resistance may be due to similarities in tetramic acids involved in the organisms' quorum sensing systems and GML. The organism does not have the Enterobacteriaceae LPS, and thus we hypothesized that GML would easily kill the organism. This is clearly not the case. Indeed, in performing the experiments, it appeared that at pH 7.0 GML caused highly significant up-regulation of virulence factor production (data not shown). Pigment production was clearly upregulated, as cultures turned visibly green with growth in GML. Like Enterobacteriaceae, Pseudomonas aeruginosa was killed by GML in acidic pH, although the organism itself exhibited greater susceptibility to acid pH alone.
In conclusion, this study has examined in detail the antibacterial activities of the generally recognized as safe compound GML. GML has potential use as a broadly-acting topical microbicide on human and animal mucosa and skin surfaces, and in environmental sources of human and animal infections. We refer to GML as a dual acting anti-infective because it kills bacteria, while at the same time stabilizing mucosal and skin surfaces to prevent inflammation that may be required for microbial infection to occur. GML has many advantages to other agents because of its safety record and because resistance is unlikely to develop.

Bacteria
Staphylococcus aureus MN8 is a typical menstrual TSS organism [5]. The organism is classified as a pulsed-filed gel electrophoresis clonal type CDC USA200 organism; it is methicillin-sensitive and produces TSST-1. Staphylococcus aureus 128 is a CA-MRSA USA200 strain that produces TSST-1. Streptococcus pyogenes strain T25 3 curedT12 was originally obtained from a patient with scarlet fever [31]. This strain produces SPE A. Streptococcus pyogenes strain 594 is highly sensitive to GML; the strain produces SPE A [32].
Streptococcus agalactiae strain MNSI was isolated from a patient with neonatal sepsis. Escherichia coli strain Watson was isolated from a female patient with a urinary tract infection. Pseudomonas aeruginosa PA01 was kindly provided by Dr. C. Mohr, University of Minnesota, Minneapolis, MN. Non-typable Haemophilus influenzae was kindly provided by the late Dr. Scott Giebink, University of Minnesota, Minneapolis. All strains listed in Table 1 were either clinical isolates or were strains listed above. All strains were cultured in media that supports their optimal growth. Anaerobes were cultured stationary in an anaerobic chamber or GasPak jar (Becton Dickinson, Sparks, MD). Aerobes were cultured in a standard incubator with aeration (shaking at 200 revolutions/min), and aerotolerant anaerobes (streptococci) were cultured stationary in the presence of 7% CO 2 . Neisseria gonorrhoeae was not culturable on standard laboratory media. Thus, for determination of GML inhibition of growth of this organism, Neisseria were cultured on chocolate agar plates, then washed off the plates with PBS, and finally suspended in Todd Hewitt media for designated time periods with gentle shaking (100 revolutions/min) in the presence of 7% CO 2 . The organism was plated onto chocolate agar plates for determination of CFUs/ml.

GML and derivatives and lauric acid
GML was purchased from Colonial Chemical Inc, South Pittsburg, Tennessee. The R form of GML was purchased from Fontarome Chemical Company, Milwaukee, WI. The 2 form of GML was purchased from Indofine Chemical Company, Hillsborough, NJ. Lauric acid was purchased from Sigma-Aldrich, St. Louis, MO.

Assays for bacterial growth and superantigen production
For experimentation, other than in Table 1, bacteria were cultured in Todd Hewitt broth (Difco Laboratories, Detroit MI). Staphylococcus aureus, Haemophilus influenzae, Escherichia coli, and Pseudomonas aeruginosa strains were cultured at 37uC with 200 revolutions/min shaking for designated periods of time. Streptococcus pyogenes and Streptococcus agalactiae strains were cultured stationary at 37uC in the presence of 7% CO 2 . Initial inocula were estimated by determining absorbance at 600 nm wavelength, and then verification by plate counts. For Staphylococcus aureus strains, an absorbance of 1.0 routinely correlated with plate counts of 1610 9 /ml. For streptococci, an absorbance of 1.0 correlated with 5610 8 CFUs/ml. For both Escherichia coli and Pseudomonas aeruginosa, an absorbance of 1.0 correlated with 1610 8 CFU/ml. Plate counts on Todd Hewitt or blood agar plates were used for determination of CFUs/ml. Quantitative Western immunoblots were used for determination of superantigens [33].

Biofilm studies
Tampon sacs were set up as described previously [12]. Briefly, cellulose acetate dialysis tubing (12,000-14,000 molecular weight cut-off) (Spectra-Por, Rancho Dominguez, CA) was tied off on one end and inoculated with approximately 10 7 Staphylococcus aureus in 0.1 ml volumes. Regular absorbency tampons were placed within the dialysis tubing, and the tampon sacs were submerged beneath Todd Hewitt broth media containing 0.8% agar. The open end of the tampon sacs remained above the liquid to prevent direct inoculation of the internal contents with culture media. When the agar had solidified, the tampon sac cultures were incubated at 37uC stationary for designated periods of time.
For determination of CFUs/ml and TSST-1 production, the tampon sacs were removed from the agar, they were sliced open and weighed to determine fluid gain (it was assumed 1 ml weight gain corresponded to 1 gm weight gain), and bacteria and TSST-1 were eluted by addition of the tampons to syringes with 10 ml of PBS and expressing the fluids. CFUs/ml were determined by plate counts. TSST-1 was quantified by first concentrating the eluted fluids by addition of 4 volumes of absolute ethanol, followed by resolubilization in distilled water, and Western immunoblot analysis [33].
In other biofilm studies, bacteria were added to 96-well microter plates (Falcon Plastics, Oxnard, CA). The plates were incubated for up to 48 hours with Staphylococcus aureus or Haemophilus influenzae. The plates were incubated stationary at 37uC. Subsequently, CFUs/ml were determined by plate counts after briskly pipetting well contents up-and-down three times. Then, wells were washed three times with PBS with brisk pipetting at each addition of PBS. Finally, wells were treated with crystal violet [15], subsequently washed three times with PBS, and absorbance at 595 nm wavelength determined after solubilization of residual crystal violet with absolute ethanol.

Tests of accelerants
EDTA was purchased from Sigma Aldrich and added to Todd Hewitt media at indicated concentrations. The non-aqueous gel used in accelerant studies was made as follows: Propylene glycol (Gallipot, St. Paul, MN) (73.55% w/w) was mixed with polyethylene glycol (Gallipot) (25% w/w) and hydroxypropyl cellulose (Gallipot) as a gelling agent (1.25% w/w). The compounds were heated to 65uC for solubilization of components and for solubilization of GML. GEH production, if selected for over the one year period, could be visualized by direct measurement of increased clear zones around colonies in a white crystalline background.

Statistics
Means and standard deviations were determined for all experiments. In studies to compare means between treatment groups, Student's t test analysis was used to determine significant differences.