Antimicrobial Activity of Glycyrrhizinic Acid Is pH-Dependent

Abstract

In recent years, antimicrobial hydrogels have attracted much attention in biomedical applications due to their biocompatibility and high water content. Glycyrrhizin (GA) is an antimicrobial that can form pH-dependent hydrogels due to the three carboxyl groups of GA that differ in pK_a value. The influence of GA protonation on the antimicrobial activity, however, has never been studied before. Therefore, we investigated the effect of the pH on the antimicrobial activity of GA against Pseudomonas aeruginosa, Staphylococcus aureus, MRSA, Staphylococcus epidermidis, Acinetobacter baumannii, Klebsiella pneumoniae, Klebsiella aerogenes, and two strains of Escherichia coli. In general, the antimicrobial activity of GA increases as a function of decreasing pH (and thus increasing protonation of GA). More specifically, fully protonated GA hydrogels (pH = 3) are required for growth inhibition and killing of E. coli UTI89 and Klebsiella in the suspension above the hydrogel, while the staphylococci strains and A. baumannii are already inhibited by fully deprotonated GA (pH = 6.8). P. aeruginosa and E. coli DH5α showed moderate susceptibility, as they are completely inhibited by a hydrogel at pH 3.8, containing partly protonated GA, but not by fully deprotonated GA (pH = 6.8). The antimicrobial activity of the hydrogel cannot solely be attributed to the resulting pH decrease of the suspension, as the presence of GA significantly increases the activity. Instead, this increased activity is due to the release of GA from the hydrogel into the suspension, where it directly interacts with the bacteria. Moreover, we provide evidence indicating that the pH dependency of the antimicrobial activity is due to differences in GA protonation state by treating the pathogens with GA solutions differing in their GA protonation distribution. Finally, we show by LC–MS that there is no chemical or enzymatic breakdown of GA. Overall, our results demonstrate that the pH influences not only the physical but also the antimicrobial properties of the GA hydrogels.

Introduction

Glycyrrhizinic acid (GA), or glycyrrhizin, can be extracted from the roots of the licorice plant and is an FDA-approved food sweetener. For example, the maximum level of GA allowed in hard and soft candy is 16% and 3.1%, respectively.¹ It has been used throughout history as a herbal medicine to treat a variety of diseases such as colds, hepatitis, wounds, diabetes, stomachache, asthma, and tuberculosis and has been reported to be antimicrobial, antibiofilm, anti-inflammatory, anticariogenic, anticarcinogenic, hepatoprotective, and antiviral.^2,3

GA is an amphiphilic triterpenoid saponin (Figure 1), containing a hydrophilic disaccharide glucuronic acid subunit and a hydrophobic triterpene aglycone subunit called glycyrrhetinic acid (GTA) or enoxolone.^2,4 GA is able to form a hydrogel by aligning the hydrophobic subunits of GA and forming fibrils, which in turn form the hydrogel.⁴⁻⁶ The physical properties of this hydrogel are not only dependent on the GA concentration and the temperature⁷ but also on the pH.^8,9 At a neutral pH, GA is dissolved in the medium and does not form a hydrogel.⁹ By decreasing the pH, a hydrogel is able to form, and the concentration of GA needed to surpass the critical micelle concentration decreases.⁹ This pH dependency is due to the three carboxyl groups that differ in their pK_a values (Figure 1). The carboxyl on the outer glucuronic acid subunit has a pK_a1 of 3.98, the carboxyl group on the middle glucuronic acid subunit has a pK_a2 of 4.62, and the third carboxyl group on the enoxolone subunit has a pK_a3 of 5.17.¹⁰ Consequently, there are four different states in which GA can exist, which form a distribution depending on the pH of the surrounding medium (Figure 2). The different GA species are labeled as GA₀, GA₁, GA₂, and GA₃, where the suffix refers to the negative charges on the molecule and therefore the degree of protonation (Figure 1). The protonation of GA influences fibril formation and thus hydrogel formation. At neutral pH levels, GA is completely deprotonated and is unable to self-assemble into these fibrils. Indeed, a free carboxyl group on the triterpene moiety of GA and at least one free carboxyl group and some free hydroxyl groups on the glucuronic moiety are needed for gelation.⁹

Figure 1.

Chemical structure of GA. Glycyrrhizin is an amphiphilic molecule containing two hydrophilic subunits of glucuronic acid (left square) and one hydrophobic triterpene aglycone subunit called glycyrrhetinic acid (right square). This amphiphilic nature allows it to act as a gelator and form a hydrogel. Finally, there are three different carbonic acids which have been labeled depending on their acidity with number 1 the most acidic and number 3 the most basic.

Figure 2.

Distribution of the different glycyrrhizin molecule species depending on the pH. Due to the three acidic groups of GA, the molecule has three different pK_a values. The three different pK_a values are pK_a1 = 3.98, pK_a2 = 4.62, and pK_a3 = 5.17, which correspond with the deprotonation of carboxyl acids 1, 1, and 2, and all three in Figure 1, respectively. Using these values, the graph is divided into four different zones: Zone Z₁ (pH < 3.98), Zone Z₂ (3.98 < pH < 4.62), Zone Z₃ (4.62 < pH < 5.17), and Zone Z₄ (pH > 5.17).

Bacteria maintain a constant, slightly basic intracellular pH due to the high impermeability of their membrane to ionized molecules.¹¹ However, protons can still traverse the bacterial membrane when the concentration of protons reaches a certain threshold. The ability of these protons to enter a cell depends on the permeability of the membrane and the size of the membrane channels.¹² Once the intracellular pH decreases, it impairs protein and nucleic acid functionality due to protonation,¹³ which negatively impacts intracellular processes such as DNA transcription, protein synthesis, and enzyme activities.^11,13 Additionally, an acidic cytoplasm results in the disruption of the proton motive force, hindering bacterial energy production and eventually resulting in cell death.^12,14 To counteract this, bacteria have evolved several mechanisms to tolerate and prevent uncontrolled proton influx inside the cell. Acidic conditions in both the cytoplasm as well as the periplasm promote the expression of protein repair or protein breakdown complexes to remove damaged or aggregated proteins.¹² Additionally, bacteria can (i) alter the membrane permeability to protons by changing membrane lipid types or modifying fatty acids into cyclopropane fatty acids,¹⁵ (ii) enhance proton efflux,^16,17 and (iii) consume protons through amino acids and urea breakdown.^11,12

Next to having direct antimicrobial effects, low pH can also influence the antibacterial effects of other molecules, potentially changing their mode of action.¹⁸ The antimicrobial activity of weak organic acids, for example, is pH-dependent. In acidic pH, these acids become protonated, removing their negative charge^12,19 and making them neutral and thus more lipophilic. This enhances their interaction with the negatively charged bacterial membrane, allowing them to insert into the membrane and acidify the cell through proton release in the less acidic intracellular environment.¹⁹ In addition, Kozak et al. discuss how acidic and neutral pH affect the activity of lauric alginate, hydrogen peroxide, ε-polylysine, caprylic acid, and sodium caprylate against Listeria monocytogenes.²⁰ Also psoriasin, a human cysteine stabilized α-helical protein, has pH-dependent modes of action against Bacillus megaterium.^18,21 At pH values below 6, psoriasin disrupts the bacterial membrane of B. megaterium through pore formation, killing the cells, while at a neutral pH, it kills Escherichia coli and B. megaterium without compromising the membrane.²¹ Likewise, the antimicrobial activity of defensins (NP1 and NP2) against Pseudomonas aeruginosa is pH-dependent, as lower pH results in better permeabilization of the outer membrane.²² It is well established that changing the pH allows the modulation of the physical properties of GA hydrogels with a lower pH, making them more solid and brittle. Nevertheless, the influence of pH on the antimicrobial activity of GA has not been determined. Therefore, we aim to elucidate the influence of the pH, and thus GA protonation, on the antimicrobial activity of a GA formulation. We observed strong antimicrobial activity of the GA hydrogel against E. coli, Klebsiella, P. aeruginosa, Acinetobacter baumannii, and staphylococci. A decrease in the pH inside the hydrogel was found to increase the antimicrobial activity of these hydrogels even further. A supernatant assay then revealed that GA molecules dissolved from the hydrogel into the growth medium, where they directly interact with the bacteria. Finally, we found that an increase in protonation of GA resulted in stronger antimicrobial activity, which cannot solely be due to the acidic pH of the growth medium.

Materials and Methods

Microbial Strains

E. coli (UTI89, DH5α), P. aeruginosa LMG9009, Staphylococcus aureus ATCC6538, Staphylococcus epidermidis,²³Klebsiella pneumoniae ATCC13883, Klebsiella aerogenes ATCC13048 (previously known as Enterobacter aerogenes), and Acinetobacter baumannii RD5SR3 (a natural isolate, isolated from the exorhizosphere of rice in Sri Lanka) were used. Additionally, a clinical isolate of methicillin-resistant S. aureus (MRSA), isolated from an osteomyelitis patient, was also included. Prior to each experiment, the strains were inoculated in lysogeny broth (LB) and incubated overnight at 37 °C (200 rpm). Thereafter, the OD₅₉₅ of the overnight culture was measured, adjusted to an OD of 0.1 in phosphate buffer saline (PBS), and diluted 200 times in growth medium, resulting in an inoculum of 3 × 10⁵ CFU mL^–1 (normalized overnight culture).

Buffers and Solutions

LB (10 g L^–1 NaCl, 10 g L^–1 tryptone, and 5 g L^–1 yeast extract) was used to grow overnight cultures. Mueller Hinton broth (MHB; 21 g L^–1 MHB, pH = 7) was used as a bacterial medium during the experiments. If LB or MHB agar was needed, 15 g L^–1 of bacteriological agar was added. Moreover, MHB media with pH of 3.9, 4.3, 4.7, 5, 6, and 6.8 were prepared by adding HCl or NaOH to adjust the medium to the desired pH value. Finally, PBS buffer (8.8 g L^–1 NaCl, 0.39 g L^–1 KH₂PO₄, 12.4 g L^–1 K2H₄PO₄, pH of 7.2) was used to normalize the bacteria and to determine the CFU mL^–1 through serial diluting and plating.

Formulations of GA were prepared by dissolving GA monoammonium salt (CAS 53956–04–0) in dH₂O (70 °C) and adding 1 M HCl or 1 M NaOH to obtain different pH formulations: (i) hydrogel with pH = 3 (Hydrogel A), (ii) hydrogel with pH = 3.8 (Hydrogel B or Reference), (iii) hydrogel with pH = 4.3 (Hydrogel C), and (iv) solution with pH = 6.8 (GA in solution). When using hydrogels, the pH was measured at 70 °C, as the hydrogel exists in a liquid state at this temperature. Moreover, different concentrations (mass by volume) of GA were used for the three different hydrogels: 2.5%, 5%, 7.5%, 10%, 15%, and 20%. For the GA solution (pH = 6.8), a 100 mg mL^–1 stock solution in MHB was made. Finally, different GA stock solutions in MHB were made with different pH values at concentrations below the gelation concentration: 1 mg mL^–1 GA_pH=3.9; 2 mg mL^–1 GA_pH=4.3; 5 mg mL^–1 GA_pH=4.7; 35 mg mL^–1 GA_pH=5; and 100 mg mL^–1 GA_pH=6.

Rheological Characterization of Hydrogel A, B, and C

All oscillatory (temperature, strain, and frequency) measurements performed in this paper were conducted with a freshly prepared sample on an MCR 501 instrument from Anton Paar. This stress-controlled rheometer measured torques between 0.01 μN m and 300 mN m. Two rheological setups were used: the bob (diameter = 26.66 mm) and Couette and the six-blade vane and Couette setup. The reason the vane geometry was employed was to minimize the emergence of slip phenomena. The bottom geometry contained the Couette, which was made from aluminum, and contained the sample. The Couette was temperature-controlled by a Peltier element coupled to a circulating water bath. The characteristics of the 6-bladed vane/Couette setup were as follows: height of the vane = 16 mm, diameter of the vane = 22 mm, inside diameter of the cup = 2 × 14.46 mm, active length = length measuring system–length connection part = 104,5 mm, DIN/ISO position is measuring position recommended by DIN/ISO protocol = 71,5 mm, inside height of the cup = 76 mm, length of vane coming out of cup = 47 mm, and height between vane bottom and bottom of the cup = approximately 18,5 mm. The sample volumes needed for the bob/Couette and vane/Couette geometries were 20 and 30 mL, respectively.

Three different tests were performed to characterize the viscoelastic properties of Hydrogel A, B, and C. First, a temperature sweep was performed to determine the gelation temperatures of the hydrogels. G′ and G″ were measured as a function of temperature, while the strain and angular frequency were kept constant at 0.01% and 10 rad s^–1, respectively. At the start of the experiment, the hydrogels were in a sol state at 80 °C. The temperature was then decreased in two steps: (i) from 80 to 60 °C at a rate of 0.1 °C s^–1 and (ii) from 60 to 20 °C at 0.01 °C s^–1. Subsequently, the hydrogels were heated from 20 to 60 °C at 0.01 °C s^–1 and from 60 to 80 °C at 0.1 °C s^–1. Second, the linear viscoelastic regime and the crossover point of the hydrogel were identified at 37 °C using a strain sweep. The strain was increased logarithmically from 0.001 to 100%, with G′ and G″ measured as a function of strain. Six measuring points are recorded per decade, and the angular frequency was maintained at 10 rad s^–1. Finally, G′ and G″ were measured as a function of the angular frequency, while the strain was kept constant at 0.01% in a frequency sweep at 37 °C. The angular frequency decreased logarithmically from 100 to 0.05 rad s^–1.

Growth Inhibition and Killing Assays

A growth inhibition and a killing assay protocol based on the EUCAST broth microdilution assay²⁴ was used with modifications.

To determine the effect of the GA hydrogels on the growth of bacteria, hydrogels with different GA concentrations were liquefied by heating to 70 °C. In a liquid state, different volumes of the hydrogels (160, 140, 120, 100, 80, and 60 μL) were pipetted into flat-bottom 96-well plates. After the hydrogel solidified, 120 μL of MHB medium containing 3 × 10⁵ CFU mL^–1 of bacteria was added on top of the hydrogel. For the GA solutions, 2-fold serial dilutions of the GA solutions in MHB medium with different pH (3.9, 4.3, 4.7, 5, 6, and 6.8) were prepared in flat-bottom 96-well plates. The plates were inoculated with the normalized overnight culture resulting in an inoculum of 3 × 10⁵ CFU mL^–1 in a total volume of 200 μL in each well. A growth control (without GA, with bacteria at the corresponding pH), sterile control (without GA and bacteria), and GA control (with GA, without bacteria) were incorporated in the assays. The plates were then sealed using a membrane (Greiner Bio-one NV) and incubated overnight (37 °C, 200 rpm, 22 h), after which the OD₅₉₅ of each well was determined using a multimode reader “Synergy Mx” (BioTek), allowing determination of the minimal inhibitory concentration (MIC).

To determine the effect of released GA from GA hydrogels on the growth of bacteria, 120 μL of MHB medium was placed on top of 160, 120, and 80 μL of 5% Hydrogel B inside a microtiter plate and incubated at 37 °C for 22 h. Afterward, the supernatant above the hydrogel was isolated. The supernatant was then used as a medium to normalize the bacteria to 3 × 10⁵ CFU mL^–1. The growth control was normalized in a normal MHB medium. Two hundred microliters of these formulations was then transferred into a new microtiter plate containing a sterile control (lacking bacteria); the plates were sealed using a membrane and incubated overnight at 37 °C (200 rpm, 22 h). The OD₅₉₅ was determined using the multimode reader “Synergy Mx” (BioTek), after which the MIC was determined.

To determine the bactericidal effect and the minimal bactericidal concentration (MBC) of the GA hydrogels and GA solutions, the number of cells of the medium on top of the hydrogels or in the supernatant was determined by plate counts when there was no visible growth.

Liquid Chromatography–Mass Spectrometry

The stability of the GA hydrogel was qualitatively assessed through liquid chromatography–mass spectrometry (LC–MS). This was performed on a Shimadzu prominence-i lc-2030c 3D plus (HPLC system) with a quaternary gradient pump, column thermostat, and diode array detector coupled with a Shimadzu LCMS-2020 (MS detector) with electrospray ionization and single quadrupole mass analyzer. The system was equipped with an analytical column Shim-pack GISS-HP, 3 μ C18, 2.1 × 100 mm. The samples were measured according to the following method. First, a stabilization with a solvent ratio of 40% MeOH (D) with H₂O (A, with 0,1% formic acid) was performed for 8 min (DA_STAB40_8_+M). Then, the sample was injected in the column, while the solvent ratio was increased in gradient to 100% MeOH during 5 min, followed by 5 min at 100% MeOH (DA_40–100–100_5–5_ ± M). Finally, the system was flushed with MeOH for 10 min (DA_FLUSH_10_+M). Mass spectra were recorded after both positive and negative ionization and are expressed in m/z (mass to charge ratio).

Data Processing and Statistical Analysis

At least three independent repeats with two replicates per repeat were performed in each experiment. The data processing was performed using Microsoft 365 Excel 2015. The relative OD₅₉₅ values were calculated using the data of the growth control, the sterile control, and the GA control

The concentrations were determined by calculating the amount of GA present in the whole well

Statistical significance was determined with GraphPad Prism v10. To compare the influence of the GA treatment with the pH, a two-way ANOVA was used. An unpaired t-test was performed with α = 0.05 when only two conditions were compared. When the influence of pH on the growth was compared, one-way ANOVA was used. Finally, CFU counts were log-transformed when statistical tests were performed.

Results and Discussion

Acidification of Glycyrrhizin Hydrogels Increases Antimicrobial Activity

The pH of the hydrogel changes the protonation of the GA molecule, which alters the physical properties of the hydrogel.^4,5 We wanted to elucidate whether the pH of the GA hydrogels influenced the antimicrobial activity of the hydrogel. First, the Reference hydrogel was prepared by adding GA to water, resulting in a hydrogel with an inherent pH of 3.8 (Hydrogel B) at the border of zones Z₁ and Z₂ (Figure 2). In addition, two more GA hydrogels were made based on Figure 2: (i) a hydrogel in zone Z₁ (pH = 3; Hydrogel A) and (ii) a hydrogel in zone Z₂ (pH = 4.3; Hydrogel C). The acidic pH of the hydrogel resulted in a decrease of the pH of the growth medium from a neutral pH to a final pH of 3.9, 4.7, and 5.0 for Hydrogel A, B, and C, respectively. In addition, a GA solution in zone Z₄ (pH = 6.8; GA in solution) was also tested. A hydrogel in Zone Z₃ could not be prepared, as this is the threshold between a hydrogel and solution. Finally, solutions without GA at pH values that correspond with the pH observed after acidification of the medium by the hydrogels were also included.

Growth inhibition (Figure 3) and killing (Figure 4) of different bacterial strains at a fixed concentration of GA are shown. Moreover, a comparison is made between these GA formulations and solutions with corresponding pH that lack GA to evaluate the ability of the bacteria to adapt to these pH values. A wider concentration range was also tested and can be found in Figure S1. In general, we observe higher antibacterial activity with decreasing pH of the hydrogel, which is not (solely) attributable to the corresponding decrease in pH of the surrounding environment (as will be discussed in the next paragraph). The bacteria can be divided into three separate groups, depending on their response. K. pneumoniae ATCC13883, K. aerogenes ATCC13048, and E. coli strain UTI89 are the more tolerant species against a lower pH and GA. Indeed, treatment of E. coli UTI89 (P_{A vs growth control}< 0.0001; P_{B vs growth control} = 0.0043; P_{C vs growth control} = 0.0127; P_{solution vs growth control} = 0.0084), K. pneumoniae ATCC13883 (P_{A vs growth control}< 0.0001; P_{C vs growth control} = 0.0361), and K. aerogenes ATCC13048 (P_{A,B,C vs growth control}< 0.0001) with the different GA formulations results in significant growth inhibition compared to the growth control (Figure 3). Despite these significant differences with the growth control, the growth inhibition of Hydrogel B and C is, however, limited. Moreover, they lack the ability to kill these pathogens. Nonetheless, the more acidic Hydrogel A is able to completely inhibit the growth of these bacterial strains and kill them (Figure 4). E. coli DH5α and P. aeruginosa LMG9009 are moderately susceptible to the GA hydrogel treatments. Treatment of both bacterial strains with the three hydrogels results in significant growth inhibition (P_{A,B,C vs growth control}< 0.0001). Moreover, the antimicrobial activity can be increased by lowering the pH, obtaining complete growth inhibition already with Hydrogel B and killing activity with Hydrogel A. Finally, the staphylococci strains and A. baumannii are highly susceptible to GA formulations. For these strains, growth is completely inhibited by all hydrogels (P_{A,B,C vs growth control}< 0.0001). The staphylococci are the most susceptible as treatment with GA in solution already results in significant growth inhibition (P_{solution vs growth control}< 0.0001), which is not the case for A. baumannii (P_{solution vs growth control} = 0.8563) (Figure 3). Moreover, all susceptible bacteria are fully eradicated when using the GA hydrogels, except for S. epidermidis when using Hydrogel C (Figure 4). Our results support the results of Zhao et al., where they explored the antimicrobial activity of a GA hydrogel against E. coli and S. aureus. They showed good antimicrobial activity against S. aureus, but not against E. coli using a 2 mM (1,65 mg mL^–1) hydrogel.²⁵ Their GA hydrogel was dissolved in PBS (pH = 7.4). We recreated this formulation and found a pH similar to our GA solution (pH = 6.8). Although we were not able to completely inhibit S. aureus using our GA solution, our results are similar.

Figure 3.

Relative OD₅₉₅ after treatment of E. coli (UTI89 and DH5α), K. pneumoniae ATCC13883, K. aerogenes ATCC13048, P. aeruginosa LMG9009, S. aureus ATCC6538, MRSA osteomyelitis, S. epidermidis, and A. baumannii with different GA formulations. The relative OD₅₉₅ of the MIC assay is shown after treatment with 25 mg mL^–1 of Hydrogel A (pH = 3), Hydrogel B (pH = 3.8), Hydrogel C (pH = 4.3), and the GA solution (pH = 6.8). Growth of these bacteria at corresponding pH was also monitored, as the hydrogel acidifies the medium on top from 7 to 3.91 for Hydrogel A, 4.77 for Hydrogel B, and 4.95 for Hydrogel C. Two-way ANOVA (n = at least 3 biological repeats; α = 0.05) was used to test for significance. * represents P ≤ 0.05; ** represents P ≤ 0.01; *** represents P ≤ 0.001; and **** represents P ≤ 0.0001.

Figure 4.

Log CFU mL^–1 after treatment of E. coli (UTI89 and DH5α), K. pneumoniae ATCC13883, K. aerogenes ATCC13048, P. aeruginosa LMG9009, S. aureus ATCC6538, MRSA osteomyelitis, S. epidermidis, and A. baumannii with different GA formulations. The log CFU mL^–1 of the MBC assay is shown in the conditions that were inhibited after the MIC treatment, with 25 mg mL^–1 of Hydrogel A (pH = 3) and B (pH = 3.8) for E. coli UTI89, K. pneumoniae ATCC13883, K. aerogenes ATCC13048, E. coli DH5α, and P. aeruginosa LMG9009, and 12.5 mg mL^–1 of Hydrogel A (pH = 3), B (pH = 3.8), and C (pH = 4.3) for S. aureus ATCC6538, MRSA, S. epidermidis, and A. baumannii. The growth of these bacteria at corresponding pH is also shown, as the hydrogel acidifies the medium from 7 to 3.91 for Hydrogel A, 4.77 for Hydrogel B, and 4.95 for hydrogel C. Dotted line represents inoculum. No MBC was performed when growth was not inhibited in MIC assay (displayed with a+). Two-way ANOVA (n = at least 3 biological repeats; α = 0.05) was used to test for significance. * represents P ≤ 0.05; ** represents P ≤ 0.01; *** represents P ≤ 0.001; and **** represents P ≤ 0.0001.

The antimicrobial activity of these GA formulations cannot solely be attributed to the inability of the pathogens to tolerate the corresponding acidic pH, as there generally is significantly better activity when GA is present in the system. For E. coli UTI89, there is significantly more growth inhibition by Hydrogel B and the GA solution, and almost significantly more growth inhibition by Hydrogel C (P = 0.0872) compared to the growth inhibition in the medium with a corresponding pH. The pH of the environment in the presence of Hydrogel A (pH 3.9) can fully inhibit the growth but not fully kill E. coli UTI89 contrary to Hydrogel A (Figures 3 and 4). Similar behavior can be observed for K. pneumoniae ATCC13883 and K. aerogenes ATCC13048, although no inhibitory activity is observed using the GA solution. In contrast to E. coli UTI89, significant GA-specific growth inhibition is seen when treating these pathogens with Hydrogel C. Remarkably, for both these Klebsiella strains, a higher relative OD₅₉₅ can be observed for Hydrogel A compared to the corresponding medium with the same pH (Figure 3). However, bactericidal activity is seen when looking at the cell counts (Figure 4). This discrepancy can be explained by the turbidity of the hydrogel that possibly interfered with the OD₅₉₅ measurements, resulting in higher measurement values. The antimicrobial activity of GA against E. coli DH5α is more nuanced. The acidic environmental pH already results in growth inhibition, indicating that the pathogen has difficulty adapting to the pH corresponding to Hydrogel C (pH 5.0) and B (pH = 4.77). Moreover, this pH effect is to the same extent as that of the GA treatment, indicating that there is no GA-specific inhibitory activity. Despite this poor adaptability to acidic pH, we still obtain GA-specific killing activity for Hydrogel A, as there is significantly more killing when GA was present. For P. aeruginosa LMG9009, the antimicrobial activity of Hydrogel B and C is not solely due to the lower pH, as there is significantly more growth inhibition of Hydrogel B and C when GA is present in the system and thus GA-specific activity. For both Hydrogel A and the corresponding pH of 3.9, P. aeruginosa is inhibited and killed. The resulting killing activity of P. aeruginosa at this pH can thus not be attributed to GA. Finally, the growth inhibition of the staphylococci and A. baumannii by Hydrogel A and B is probably (at least partially) attributable to the lower pH of the environment. For Hydrogel C, there is significant better growth inhibition when GA is present for MRSA, S. epidermidis, and A. baumannii, although the pH already results in significant growth inhibition (P_{pHC vs growth control}< 0.0001 for MRSA and S. epidermidis, P_{pHC vs growth control} = 0.0351 for A. baumannii) (Figure 3). Despite this pH effect on growth inhibition, the killing activity observed by the three hydrogels is improved by the presence of GA, as treatment with Hydrogel C and B results in significantly more killing than the pH values for MRSA, S. epidermidis, and A. baumannii. Similarly, significantly more killing is observed when treating S. aureus, MRSA, and S. epidermidiswith Hydrogel A than with the pH. Surprisingly, S. aureus is able to survive at a pH of 3.9, while it cannot survive at higher pH (Figure 4).

Antimicrobial Activity Is Due to the Released GA Molecules

As GA is expected to release from the hydrogel into the medium, we tested the antimicrobial activity of the supernatant, which was isolated after 22 h of incubation on top of Hydrogel B. The growth inhibitory and killing effects of the supernatant are shown in Figure 5, where a comparison is made between the supernatant treatment and the corresponding hydrogel.

Figure 5.

Treatment of the nine strains with the supernatant of Hydrogel B compared with Hydrogel B treatment. (a) Relative OD₅₉₅ of the MIC assay is shown of all strains after treatment with the supernatant of Hydrogel B of 25 mg mL^–1 and the corresponding Hydrogel B treatment. Two-way ANOVA (n = at least 3 biological repeats; α = 0.05) was used to test for significance. (b) Log CFU mL^–1 of the MBC assay is shown of all strains that were inhibited in the MIC test after treatment with the supernatant of Hydrogel B of 25 mg mL^–1 and the corresponding Hydrogel B treatment. When growth was not inhibited during the MIC assay, no MBC was performed (displayed with a +). The dotted line shows the inoculum. Two-way ANOVA (n = at least 3 biological repeats; α = 0.05) was used to test for significance. * represents P ≤ 0.05; ** represents P ≤ 0.01; *** represents P ≤ 0.001; and **** represents P ≤ 0.0001.

For none of the strains a significant reduction in growth inhibition or killing is seen when the supernatant is used compared to the hydrogel. This indicates that the antimicrobial activity was due to the GA molecules released from the hydrogel into the medium. As Hydrogel B is acidic and GA is a weak organic acid, both the protons from the hydrogel as well as the protonated GA molecules acidify the growth medium to an observed pH of 4.77, which is associated with a redistribution of dissolved GA to the range of GA₀, GA₁, and GA₂, with only minor amounts of deprotonated GA₃ (Figure 2). Since we showed that the antimicrobial activity of the GA hydrogel cannot solely be attributed to the acidic pH (Figures 3 and 4), the observed antimicrobial activity in the supernatant assay is the result of the dissolved GA molecules with their associated protonation states, possibly in combination with the direct antimicrobial effect of the acidic pH.

Remarkably, significantly better antimicrobial activity of the supernatant is observed compared to the hydrogel for K. aerogenes ATCC13048 (Figure 5a). For E. coli DH5α, the supernatant treatment results in significantly more killing compared to the hydrogel (Figure 5b). The better activity of the supernatant against both strains is possibly the result of a difference in the treatment time between the supernatant and the hydrogel. Indeed, we expect that GA gradually releases from the hydrogel into the growth medium. In turn, the optimal killing concentration is achieved only later, resulting in shorter treatment. This is in contrast to the supernatant assay, where the maximum concentration of GA is present at the start of the assay.

Based on these results, the GA molecules are released from the hydrogel, allowing them to interact with the bacteria. Previous research has shown the ability of GA to insert itself into a lipid bilayer.²⁶ In turn, this was shown to increase the permeability, decrease the elasticity modulus,²⁷ and cause pore formation.²⁶ This was confirmed by Hazlett et al., who showed pore formation, swelling, and cell clumping in P. aeruginosa.²⁸ Therefore, the difference in antimicrobial response against the released GA can be partly explained by the differences in bacterial membranes between the different pathogens, as Gram-positive and Gram-negative strains differ substantially in structure and composition of their cell wall. In turn, membrane-active antimicrobials only have to penetrate the inner membrane of Gram-positive bacteria, rendering them more susceptible to such antimicrobials.^29,30 This might explain the high susceptibility of the Gram-positive strains toward GA in the present study. However, this does not explain the susceptibility of the Gram-negative strains, A. baumannii, E. coli DH5α, and P. aeruginosa LMG9009. The higher susceptibility of these Gram-negative strains might still be explained by differences in membrane composition, which in turn influence GA interaction with the membrane, its insertion into the bacterial membrane, and pore formation.

Lower pH of the GA Hydrogels Enhances Antimicrobial Activity despite Lower Solubility

Next, we wanted to evaluate the viscoelastic properties of the different hydrogels to evaluate whether the pH of the hydrogel could influence the release of GA from the hydrogel. Previously, Mees et al. have shown that the viscoelastic properties of 5% and 10% hydrogels are similar, indicating the limited impact of the concentration on the viscoelastic properties of the GA hydrogel.⁷ To monitor the influence of the pH on the viscoelastic properties of the GA hydrogels, temperature, strain, and frequency sweeps were performed (Figures S6, S7, and S8), and the gelation temperature and moduli ratio were determined (Table 1) for Hydrogel A, B, and C (10%). First, we show that the gelation temperature of the hydrogels is influenced by the pH. A lower pH results in a higher gelation temperature, with a gelation temperature of 55 °C for Hydrogel A, 53 °C for Hydrogel B, and 43 °C for Hydrogel C. The higher gelation temperature at lower pH is due to an increase in thermal stability due to slower breakdown of the hydrogel structure (Figure S6). Because of the slower breakdown, there is also less heat dissipation within the Couette. Next, the pH also influences the mechanical stability of the hydrogel (Figures S7 and S8). By calculating the moduli ratio G′/G″,³¹ the fluid-like properties of the hydrogels were determined. A lower pH results in a higher moduli ratio, indicating more solid-like behavior of the hydrogel and thus a more brittle hydrogel. In contrast, the lower moduli ratio of Hydrogel C indicates more fluid behavior and thus a more viscous hydrogel. This is also observed when handling the hydrogels. The fluid-like hydrogel resembles a more traditional hydrogel, characterized by its softness and flexibility, while the hard hydrogel exhibits properties akin to wax, being more brittle and rigid. A hydrogel with more solid-like behavior is expected to result in less release of GA compared to a hydrogel with more liquid-like behavior.³²⁻³⁴ Consequently, the increase in the antimicrobial activity of the hydrogels as a function of decreasing pH is not the result of changes in viscoelasticity of the different hydrogels at varying pH.

Table 1. Summary of the Gelation Temperature, the Linear Viscoelastic Range (37 °C), the Flow Points (37 °C), and the Moduli Ratio G′/G′′ (37 °C)^a.

hydrogel	gelation temperature (°C)	linear viscoelastic range		flow points		the moduli ratio (G′/G″)
		vane (%)	bob (%)	vane (%)	bob (%)
A (pH = 3)	55	0.8		15		49
B (pH = 3.8)	53	0.5%	0.03	40	1.5	19
C (pH = 4.3)	42		0.1		25	7

^aThe gelation temperature was determined using a temperature sweep, while the linear viscoelastic range and flow points were determined using the strain sweep and frequency sweep. This was determined for all three hydrogels (10%).

GA is Not Hydrolyzed into GTA

GA can be degraded into GTA, a strong antibacterial compound against S. aureus,³⁵ MRSA,³⁶S. epidermidis,³⁵Streptococcus mutans,³⁷ and Helicobacter pylori.³⁸ Its antibacterial activity against E. coli(35) and Salmonella Typhimurium³⁵ is limited. The presence of GTA inside the supernatant could occur due to chemical degradation of GA inside the hydrogel or enzymatic degradation of the released GA molecules by the bacteria. To confirm the activity of the dissolved GA molecules, we aimed to validate the absence of GTA in the supernatant through LC–MS (Figures S3, S4, and S5). In order to monitor chemical degradation, a sample from a nonautoclaved hydrogel is compared with a supernatant sample isolated above an autoclaved hydrogel (sterile sample). Since both chromatograms are comparable, this excludes chemical degradation of GA into GTA. Next, enzymatic degradation of GA into GTA was evaluated by comparing the chromatograms of the sterile sample with the chromatogram of two supernatant samples that were incubated with K. aerogenes ATCC13048 and S. aureus ATCC6538, since GA could be transformed into GTA enzymatically through β-glucuronidases production by the bacteria.³⁹ Again, no remarkable differences can be found between the chromatograms, indicating that there is no enzymatic breakdown of the released GA into GTA as well. Nonetheless, it is worth mentioning that a small peak for the mass corresponding to GTA could be detected in all samples. This finding suggests that GTA is present as a minor impurity in commercial GA.

Protonation of GA Changes Its Antimicrobial Activity

The supernatant assay with Hydrogel B had as limitation that the concentration of the released GA molecules from the hydrogel into the supernatant was not characterized. Moreover, the antimicrobial activity of the individual differentially protonated GA species could not be pinpointed. To address both shortcomings, we further investigated the pH dependency of the GA antimicrobial activity in solution by preparing GA dilution series at different pH values and determining the MIC/MBC as a function of pH. The pH values were chosen to correspond with the pH values measured in the growth medium after incubation with the hydrogels (3.9, 4.7, and 5.0) and with the pH of the GA solution (6.8). Moreover, pH 6 was also included to obtain a range of pH values that strongly differ in their distribution of the GA species. Indeed, a pH of 6.8 mainly consists of GA₃, while a pH of 6.0 lacks GA₁ and GA₀ and a pH of 5.0 lacks GA₀. At a pH of 3.9, no GA₃ and barely any GA₂ can be found. Finally, a pH of 4.3 was included when no inhibitory concentrations could be found at a pH of 4.7.

In Figure 6, the change in MIC of the bacterial strains as a function of the pH is shown, while Table S1 displays the MBC results. For P. aeruginosa, S. aureus, MRSA, S. epidermidis, and A. baumannii, a clear correlation between decreasing pH and decreasing MIC and MBC values can be observed. The largest effect of pH on the activity of GA against P. aeruginosa can be observed when changing the pH from 5 to 4.7 (Figure S2 and Table S1). The decrease of the pH itself (without the presence of GA) already has a significant effect on the inhibition of P. aeruginosa with no inhibition of the bacteria at pH 5 and a 46% inhibition at pH 4.7 (Figure S2). Nonetheless, at pH 4.7, the growth of P. aeruginosa is completely inhibited at a GA concentration of 0.31 mg mL^–1, resulting in bacteriostatic activity. At pH 5, a significantly higher concentration of GA (17.5 mg mL^–1) does not lead to growth inhibition of P. aeruginosa. When the pH is decreased from pH 5 to 4.7, the GA distribution shifts toward more protonated GA species (Figure 2). More specifically, this decrease in pH results in a decrease of GA₃ and GA₂ from 25 to 10% and from 50 to 40%, respectively, while the amount of GA₁ and GA₀ increases from 25 to 40% and from 0 to 10%, respectively. Taken together, these results suggest that the increase in antimicrobial activity against P. aeruginosa is due to an increase in the proportion of protonated GA species. Similar to our results at a pH of 4.7, Chakotiya et al. were able to inhibit (0.1 mg mL^–1) and kill (0.4 mg mL^–1) P. aeruginosa using a GA solution in a mixture of methanol and water.⁴⁰

Figure 6.

MIC after treatment with different GA solutions MIC in function of the pH. The nine different bacterial strains were treated with different GA solutions depending on their survival at the corresponding pH. When the solution was not sufficient to inhibit growth, it was not included in the graph. E. coli UTI89, E. coli DH5α, K. pneumoniae ATCC13883, and K. aerogenes ATCC13048 were treated with GA solutions with a pH of 4.3, 4.7, 5, and 6.8. P. aeruginosa LMG9009 was treated with GA solutions with a pH of 4.7, 5, and 6.8, while S. aureus ATCC6538, MRSA osteomyelitis, S. epidermidis, and A. baumannii were treated with GA solutions at a pH of 4.7, 5, 6, and 6.8. A red dot indicates that maximum concentration that could be tested without achieving an MIC. Increasing the concentration of the GA solutions would cause a hydrogel to form in these conditions.

For S. aureus, MRSA, S. epidermidis, and A. baumannii, the correlation between the pH and the antibacterial activity of GA can be observed when changing the pH from 6.8 to 5, whereas a decrease in pH alone from 6.8 to 6 has no inhibiting effect on the growth of the bacteria (Figure S2). All GA molecules are fully deprotonated (GA₃) at pH 6.8, while approximately 15% of the GA molecules have one protonated carboxyl group (GA₂) at pH 6. This (small) shift in protonation already has a significant effect on the activity against the staphylococci. At pH 6.8, a MIC value cannot be reached with GA concentration of 50 mg mL^–1. In contrast, at pH 6, the MIC values for S. aureus, MRSA, and S. epidermidis are 6.25, 3.13, and 6.25 mg mL^–1, respectively, and bacteriostatic activity is observed (Table S1). When the pH decreases further to pH 5, the MIC of MRSA decreases to 0.55 mg mL^–1 and becomes bactericidal, while the MIC of S. epidermidis decreases to 0.68 mg mL^–1 and becomes bacteriostatic. For both of these strains, a pH value of 5 results in significant growth inhibition (Table S1). For S. aureus, no MIC could be obtained at a pH value of 5, as the pH completely inhibits the growth of this pathogen. This relatively small shift in pH from 6 to 5 corresponds to a 55% increase in GA protonation. More specifically, this results in a 55% decrease of fully deprotonated GA (GA₃), while GA₂ and GA₁ increase with 35% and 20%, respectively. Hamad et al. and Chopra et al. were also able to inhibit staphylococci using licorice extracts but at different concentrations. Hamad et al. obtained an MIC of 25 mg mL^–1 for S. aureus and S. epidermidis, while Chopra et al. obtained an MIC that was 100 times lower for S. aureus.^41,42 These differences with our results could be attributed to the presence of GTA in the licorice extracts or differences in pH and thus GA protonation. Since neither the pH is mentioned nor the presence of GTA is evaluated, this limits us from comparing our results. For A. baumannii, no MIC value is reached at pH 6, while an MIC of 2.19 mg mL^–1 is reached at a pH value of 5. A. baumannii is not inhibited at these pH values without the presence of GA (Figure S2).

For E. coli UTI89, K. pneumoniae ATCC13883, and K. aerogenes ATCC13048, the MIC could not be determined, as increasing the concentration of GA would have led to the formation of a hydrogel. This is in contrast with the results of AbdEl-Mongy et al., where they were able to kill β-lactamase-producing K. pneumoniae using GA nanoparticles at much lower concentrations (20 to 40 μg mL^–1).⁴³ However, Karahan et al. and Hamad et al., in agreement with our work, did not obtain an MIC for K. pneumoniae and E. coli at these concentrations.^42,44 This discrepancy can be explained by the difference in GA conformation, i.e., dissolved GA compared to a GA nanoparticle. Another possibility could be the different setup. We determined the MIC through OD measurements, whereas they determined the MIC by measuring inhibition zones.

The dependency of the antimicrobial activity of GA on the protonation state might be related to differences in uptake rate by transport systems, although no specific transporters to import this molecule inside the bacterial cell are known. Alternatively, the protonation state might influence the ability of GA to insert itself into negatively charged bacterial membranes (Figure 7), as protonated weak organic acids are known to interact and penetrate the bacterial membrane more efficiently.¹⁹ Once GA is taken up by a transporter or inserted into the membrane, the less acidic intracellular environment might result in the proton release of the weak organic acid¹⁹ and cause an influx of protons through the pores formed by the membrane-inserted GA molecules. This would disturb intracellular physiology through acidification and dissipation of the proton motive force. In addition, insertion of the GA molecules into the bacterial membrane might influence cell viability due to loss of cell membrane integrity and the associated uncontrolled transport of various molecules and ions, and reduced ability to withstand the turgor pressure.⁴⁵ Cell lysis through disruption of the bacterial membrane has already been shown with, for example, chitosan, gramicidin S, and peptidyl-glycylleucine-carboxyamide against E. coli and S. aureus.^46,47 Given that GA is hypothesized to be involved in membrane destabilization and proton influx, there is also a possibility that part of the observed pH dependency of the antimicrobial activity of GA might be a synergistic effect between GA and the pH, and this is not only due to differences in protonation states of GA. A distinction between these hypotheses could be made by synthesizing a GA derivative that does not depend on the pH, as this would allow us to pinpoint potential synergistic effects between GA and pH. In an attempt to walk this path, we methylated the carboxyl groups of GA, creating a GA molecule with three different methyl esters and disabled pH structural dependency. The resulting molecule was, however, very nonpolar and insoluble in the bacterial growth medium, preventing us from drawing further conclusions. Based on our observations, an impact of the GA protonation state on the antimicrobial activity nevertheless seems very plausible. Indeed, this hypothesis is supported by the relatively small shifts in pH that already result in significant changes in the antimicrobial activity of GA, which coincide with strong shifts in the GA protonation state.

Figure 7.

Graphical representation of the proposed interaction of GA with the bacterial membrane, dependent on the molecule’s protonation state. We hypothesize a pH-dependent mode of action of GA, as protonated weak organic acids are known to insert into bacterial membranes. Our results indicate that the protonation state of GA significantly influences its antibacterial activity. A fully protonated GA molecule loses its negative charge, facilitating its interaction with the negatively charged bacterial membrane, possibly leading to membrane insertion and pore formation. Conversely, the negative charge of a fully deprotonated GA molecule inhibits the interaction with the membrane, resulting in reduced antibacterial activity.

Based on the hypothesis that the GA protonation state influences its antimicrobial activity, not only differences in membrane composition (discussed previously) but also differences in the ability to control the intracellular pH could contribute to the strain-dependent response against GA. Bacteria with acid tolerance responses (ATRs) are expected to counter the changes in intracellular pH more effectively compared to those lacking such mechanisms. Indeed, E. coli UTI89, K. pneumoniae, and K. aerogenes, the most tolerant species in our study, are known to infect acidic environments in the host such as the gastrointestinal and urinary tract.^48,49 Their ability to survive in acidic environments is due to the activation of ATR mechanisms.^50,51 Generally, E. coli is known to activate five different acid response systems that consume protons in enzyme-catalyzed reactions,⁵² actively pump out protons,¹⁶ modify its cell membrane,⁵³ and promote the expression of protein repair mechanisms.⁵⁴ However, this does not explain the difference in pH tolerance between E. coli UTI89 and E. coli DH5α. The latter is known to be a domesticated lab strain that prioritizes plasmid transformation.⁵⁵ In turn, they could have lost their tolerance to acidic environments. Klebsiella, on the other hand, lacks AR2 and AR3, but is able to upregulate protein repair mechanisms in acidic pH.⁵⁶ Other mechanisms that have been observed to render Klebsiella more tolerant to acidic pH include upregulation of a tripartite efflux pump⁵¹ and the production of urease to consume protons and produce ammonia.⁵⁷ This explains why E. coli UTI89, K. pneumoniae, and K. aerogenes are more difficult to treat at acidic pH. Only when these pathogens were treated with Hydrogel A, which resulted in an acidification of the medium to 3.9, GA-specific bactericidal activity was observed. P. aeruginosa is moderately susceptible to the GA hydrogels, as Hydrogel B results in GA-specific activity growth inhibition. It is known to use quorum sensing and biofilm formation to cope with acid stress.⁵⁸ In addition, the cyoABCDE genes have been shown to be involved during ATR during mild acid exposure (pH = 5.5).⁵⁹ Finally, the GA-susceptible strains, A. baumannii, and the staphylococci strains, are known to be more susceptible to acidic pH levels,⁴⁹ which could explain the killing activity observed at more acidic pH levels in the presence or absence of GA. These mechanisms that render the bacteria more tolerant toward GA treatment could moreover also explain the differences in antimicrobial pH effects that are independent of GA.

Currently, no resistance has been reported against GA, although no evolution experiments have been performed to assess the evolutionary dynamics on bacterial populations during long-term GA treatment. Interestingly, GA has been shown to potentiate the antibacterial activity of gentamicin against vancomycin-resistant Enterococcus faecium(60) and norfloxacin against multidrug-resistant S. aureus.⁶¹ The absence of resistance against GA is consistent with the lower propensity for resistance observed against antimicrobial peptides, which have a similar mode of action as GA.⁶² Although resistance to antimicrobial peptides is less common, bacterial resistance to antimicrobial peptides and other antimicrobials with similar modes of action, such as polymyxin B and polymyxin E (also known as colistin), has been reported.^62,63 Resistance against these types of antimicrobials can be obtained through modifications of the bacterial membrane⁶² or enzymatic degradation of the compound.⁶⁴ Therefore, it is plausible that resistance to GA could emerge over long-term treatment, potentially through modifications of the bacterial membrane or breakdown of GA. However, it is expected that resistance to GA would develop at a slower rate. Persistence also influences resistance development, as persister cells have been thought to be an important subpopulation that can survive antimicrobial treatment and enhance resistance development.⁶⁵ These cells typically exhibit a higher tolerance to common antimicrobials due to their low metabolic activity. Another factor that characterizes persister cells is a decreased proton motive force and, thus, a lower membrane potential. Disruption of the membrane potential has been hypothesized to induce persistence.⁶⁶ Therefore, GA treatment could potentially lead to the emergence of persister cells by disrupting the proton motive force and decreasing the membrane potential. However, polymyxin B nonapeptide, a derivative of polymyxin B that increases bacterial cell permeability, has been shown to reduce the persister fraction and potentiate antibiotic treatment.⁶⁷ In turn, the impact of GA on persisters remains unclear.

Conclusions

In this paper, we have provided evidence that we can fine tune not only the physical properties of a GA hydrogel but also the antimicrobial activity by altering the pH of the formulation. First, we are able to inhibit a broad range of human pathogens through various GA hydrogel formulations. Second, we establish a clear influence of pH on the antimicrobial activity of GA molecules. Not only does a lower pH increase the antimicrobial activity of the hydrogel but it also increases its activity spectrum against a wider variety of microorganisms. Third, we show that the antimicrobial activity is caused by GA molecules released from the gel. Finally, we hypothesize that the observed relationship between the pH and GA is the result of a protonation change in GA. The ability to modify the antimicrobial activity of GA through pH modification indicates that GA hydrogels could be a promising antimicrobial to treat bacterial infections. Moreover, antibiotics or other antimicrobials could be added to the hydrogels to potentiate the hydrogel even further and broaden the antimicrobial range against more pathogens. Particularly interesting could be the use of the hydrogel inside infected chronic wounds, where an alkalic pH and pathogenic bacteria are known to prevent wound healing. Since chronic wounds are characterized by recurring infections, more research is needed to elucidate whether GA can effectively inhibit and kill dormant and persister cells. In addition, future work is needed to validate the mode of action and resistance development of GA in vitro and the antimicrobial activity of GA in vivo.

Glossary

Abbreviations

GA

Glycyrrhizinic acid

GTA

Glycyrrhetinic acid

MIC

Minimal inhibitory concentration

MBC

Minimal bactericidal concentration

ATR

Acid tolerance response

AR

Acid response

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsabm.4c00942.

• Summary graph per pathogen; adaptability to pH per pathogen; LC–MS; temperature, strain, and frequency sweep; and MBC summary results table (PDF)

• Raw data (XLSX)

The authors declare no competing financial interest.