Glucose Directly Promotes Antifungal Resistance in the Fungal Pathogen, Candida spp.

Background: Glucose level alters susceptibility of antifungal agents during chemotherapy in diabetes patients.

Results: Glucose selectively interacts with antifungal agents, strongly affects azole drugs, and forms complexes by hydrogen bonding.

Conclusion: It is important for researchers and pharmaceuticals to make new antibiograms for diabetes patients.

Significance: Drug selection is important for controlling fungal infections in diabetes patients.

Abstract

Effects of glucose on the susceptibility of antifungal agents were investigated against Candida spp. Increasing the concentration of glucose decreased the activity of antifungal agents; voriconazole was the most affected drugs followed by amphotericin B. No significant change has been observed for anidulafungin. Biophysical interactions between antifungal agents with glucose molecules were investigated using isothermal titration calorimetry, Fourier transform infrared, and ¹H NMR. Glucose has a higher affinity to bind with voriconazole by hydrogen bonding and decrease the susceptibility of antifungal agents during chemotherapy. In addition to confirming the results observed in vitro, theoretical docking studies demonstrated that voriconazole presented three important hydrogen bonds and amphotericin B presented two hydrogen bonds that stabilized the glucose. In vivo results also suggest that the physiologically relevant higher glucose level in the bloodstream of diabetes mellitus mice might interact with the available selective agents during antifungal therapy, thus decreasing glucose activity by complex formation. Thus, proper selection of drugs for diabetes mellitus patients is important to control infectious diseases.

Introduction

Among several emerging diseases, DM³ is considered one of the largest emerging threats to public health in the 21st century. From the medical profession to the general public, individuals are well aware that the greater frequency of metabolic disorders and morbimortality in diabetic patients are due to the hyperglycemia. Extra glucose in blood and urine provides an increased propensity to develop infections, and diabetes patients are especially prone to foot infections, yeast infections, urinary tract infections, and surgical site infections (1, 2). The prevalence of yeast infection is more common in patients with DM because of the higher density of candidal growth (3, 4). Candida species are the part of the body's normal oral and intestinal flora, but infection becomes more severe in DM patients with higher levels of inflammation (5, 6). Antifungal medications, such as clotrimazole, nystatin, fluconazole, and ketoconazole, are very effective in controlling topical yeast infections. In the case of candidal infections in blood, the choices of drugs are intravenous fluconazole or echinocandin or amphotericin B (AmpB) (7, 8). However, there is no such specific indication of the choice of drug for people suffering from DM. When a patient is suffering from DM and has a yeast infection, that patient is more likely to get other infections because the combination of Candida and high blood sugar inhibits the body's natural defense mechanisms from fighting against other bacteria and viruses. Thus, increased sugar level seems to be the main cause for the prevalence of infection in body. This study addresses the biophysical mechanism of glucose level to alter the susceptibility of antifungal agents during chemotherapy.

EXPERIMENTAL PROCEDURES

Microbial Strains and Antifungal Assay

Candida albicans MTCC227 and Candida tropicalis NCIM3110 control strains were used in this study. Strains were inoculated from mature solid medium culture plates (Sabouraud dextrose agar) and mixed with liquid RPMI 1640 (HiMedia, Mumbai, India), and then incubated for 24 h at 30 °C to achieve the relevant turbidity of 0.5 × 10⁴ cfu/ml. Minimum inhibitory concentration (MIC) values of antifungals (anidulafungin, amphotericin B, and voriconazole purchased from HiMedia) were determined according to Clinical and Laboratory Standards Institute (CLSI) guidelines (9). The concentrations of each compound used for the assay ranged from 0.09 to 100 μg ml⁻¹. MIC values were determined where no visible growth was observed. The culture conditions and fungal growth were monitored as described earlier by Mandal and co-workers (10, 11). Throughout the experiment, two wells were maintained in which no compounds were added used as positive control, and compounds with the absence of fungal inoculums were used as negative control to preserve the experimental sterility. Individual experiments were repeated four times.

Isothermal Titration Calorimetry (ITC) Analysis

To obtain the effect of excess glucose on antifungal agents, the titration between glucose and individual antifungals was performed using a MicroCal Auto-iTC200 system (GE Healthcare) coupled with nonreactive Hastelloy® cells for chemical resistance. The titrations were carried out in 0.01 × 10⁻³ molar solution of individual antifungals with a 1.0 × 10⁻³ molar solution of glucose in water. All solutions were degassed right before the experimental runs in the same condition at 25 °C at 700 rpm. Blank ITC experiments were performed to correct the heat of dilution effects. Origin 7.0 (OriginLab Corp.) was used to analyze the ITC data to determine the binding constant (K) and enthalpy of binding (ΔH) directly from the binding thermograms.

Fourier Transform Infrared (FTIR) Analysis

For IR experiments, individual antifungal agents were mixed with glucose at a 1:2 ratio, incubated overnight at 37 °C, lyophilized, and placed onto a KBr pellet, and then allow to completely dry. The spectra of the dried specimens were recorded on a Shimadzu 8400 FTIR spectrophotometer. Absorbance spectra were obtained from 4000 to 400 cm⁻¹ with a 4-cm⁻¹ resolution, and background spectra were also collected and subtracted.

Nuclear Magnetic Resonance (NMR) Analysis

The ¹H NMR spectra were taken on a Bruker DPX200 (200 MHz) in d₆-DMSO and D₂O using tetramethylsilane as an internal standard. All signals were referenced to tetramethylsilane to within ± 0.1 ppm.

In Silico Studies of Interactions

All docking calculations were performed using AUTODOCK 4.2 program. Docking simulations using voriconazole and amphotericin B were performed for the d-glucose molecule. All pdb files of the molecules used for in silico tests were obtained from the Open Data Drug & Drug Target Database (Canadian), DrugBank Version 4.1. All hydrogen atoms were added by using the AutoDockTool. Grid maps were calculated with 15 × 30 × 30 Å for both antibiotics utilized and 1.0 Å spacing centered on the d-glucose. To understand the in vitro activity observed with the glucose addition, a Lamarckian genetic algorithm was used as the search method to find the best glucose complex. One thousand docking runs were done for each antibiotic; in these runs, the maximum mobility for both voriconazole and amphotericin B was unlocked. The generated complexes were ranked with the base in the construction of clusters with the best models with the lowest energy, for all atoms docked toward d-glucose, showing tolerance of 4 Å, as recommended for blind docking. The program PyMOL (17) was used to characterize compound-glucose interactions.

In Vivo Evaluation in Infected Diabetic Mice

Diabetic mice were prepared following the protocol as described in detail by Wu and Huan (12). Freshly prepared streptozotocin (50 mg/kg of mice) in 0.1 m sodium citrate buffer (pH 4.5) was injected into female mice (Swiss albino), weighing 22–25 g, after a 4-h fast, every other day for 2 weeks to induce diabetes, confirmed by monitoring the blood sugar level. Control mice were given the same amount of buffer without streptozotocin. Mice were maintained under pathogen-free conditions and standard temperature (25 ± 2 °C) and a fixed humidity (60 ± 5%) condition with alternating 12-h light/dark cycles. The animals were fed a standard pellet diet and maintained in accordance with the guidelines of the National Institute of Nutrition, Indian Council of Medical Research (Hyderabad, India), and in conditions approved by the ethical committee of the Indian Institute of Technology (Kharagpur, India). Candida albicans strain MTCC227, grown in YPD liquid medium, was diluted in sterile saline, and 200 μl (containing 2 × 10⁶ cells) was injected through tail vein into both diabetic and nondiabetic mice. Control mice (each group) were given 200 μl of normal saline injection. After 24 h, 20 μl of blood, taken through tail vein puncture, was plated onto YPD agar plates (containing 32 μg ml⁻¹ of chloramphenicol) and incubated at 30 °C for 24 h. Plates were counted for the number of colonies, confirming the candidal infection as compared with the control group where no colonies were observed. After subsequent blood culture, both groups (diabetic and nondiabetic mice) were given high doses of anidulafungin (15 mg/kg of mice), amphotericin B (4 mg/kg of mice), and voriconazole (40 mg/kg of mice) through the intraperitoneal space in a 200-μl suspension with sterile PBS (pH 7.4) once daily for two consecutive days. After 48 h of antifungal drug treatment, blood culture was similarly performed, and plates were incubated for 24 h to assess the functionality of drugs in diabetic and nondiabetic mice, by analyzing the reduction in the number colonies in corresponding sets.

RESULTS

The antifungal activities of anidulafungin, amphotericin B, and voriconazole were tested against two model fungal strains, C. albicans and C. tropicalis, and glucose concentration was varied. The antifungal activities of tested drugs ranged between 1.562 and 3.125 against C. albicans and between 0.19 and 1.562 against C. tropicalis in the absence of glucose. Interestingly, the MIC values are altered by increasing the glucose concentration only for voriconazole and amphotericin B but remain unchanged with anidulafungin (supplemental Table S1). Voriconazole is more sensitive to glucose and showed 4-fold decreased activity in the presence of 2% glucose in culture medium, whereas amphotericin B showed only 2-fold reduction.

Titrations between glucose and antifungal drugs were done using ITC, and the binding constants (K), entropy (ΔS), and enthalpy (ΔH) of their reactions are described in supplemental Table S2. The binding thermogram and isotherms clearly indicate that the interactions of antifungals with glucose are exothermic in nature because ΔH was negative. The binding affinity to voriconazole was very strong followed by binding affinity to AmpB, whereas the interaction of anidulafungin with glucose molecules shows very low binding affinity (Fig. 1, upper panel).

FIGURE 1.

ITC-based thermogram and binding isotherm plot of glucose with antifungal agents. a–f, thermograms and binding isotherms obtained from interactions between glucose and amphotericin B (a and b); anidulafungin (c and d); and voriconazole (e and f), respectively. Lower panel, hydrogen-bonding determination with deuterium exchange by ¹H NMR. g–j, spectra were obtained from pure glucose (g), pure voriconazole (h), the combination of both in d₆-DMSO (i), and after deuterium exchange of the voriconazole-glucose complex in d₆-DMSO (j). The arrow in the spectrum shows the chemical shift due to hydrogen bonding.

FTIR analysis of voriconazole (supplemental Table S3) revealed the presence of C=N, C–N, C–F, and C=C stretching frequency in the region between 1640–1690, 1230–1000, 1495–1451, and 1585–1451 cm⁻¹. Moreover, C–O stretching frequency of a tertiary alcohol group gives a strong peak at 1089 cm⁻¹. The shifting of C–F stretching at a lower frequency from 1456 to 1421 cm⁻¹and C–O stretching at a higher frequency from 1080 cm⁻¹ to 1161 cm⁻¹ suggest that both fluorine and tertiary alcohols of voriconazole undergo strong hydrogen bonding with glucose. Moreover, another characteristic band of C–N stretching frequency is shifted at a lower frequency from 1247 cm⁻¹ to 1161 cm⁻¹ in an azole ring. In the case of amphotericin B, the spectral region between 1500 and 1800 cm⁻¹ represents the stretching vibrations of the C–O, -COO-, -NH₃⁺, and C–C groups. 3300–3500 cm⁻¹ is characteristic for hydrogen-bonded AmpB molecules (-OH … HO-) in a polyalcohol chain of molecules. The band centered at 1711 cm⁻¹ is attributed to the stretching vibration of the C=O group in the ester band, at 1692 cm⁻¹ and 1401 cm⁻¹ for asymmetrical and symmetrical stretching vibrations of the C–O in ester group, respectively. The sharp band with a maximum centered at 1563 cm⁻¹ arising together with the increase in surface pressure should be attributed to the vibrations of the -NH₃⁺ group. It should also be added that the symmetrical deformational vibrations of -NH₃⁺ in the amino acids in the zwitterion form appear in the range between 1550 and 1485 cm⁻¹. This is indicated by the changes related to the increase in broad absorption at the 1100–950 cm⁻¹ frequency range, characteristic of vibrations of C–O–C bonds. The spectral shift from 1692 cm⁻¹ to 1656 cm⁻¹ corresponds to the typical values for hydrogen bonds. The bands at the 1330–1040 cm⁻¹ range are assigned to the stretching and deformational vibration of out of plane (trans-polyene) C–O and C–H bonds, which further indicate their participation in the aggregation process. Moreover, the characteristic weak band observed at 1169 cm⁻¹ shifted to 1163 cm⁻¹, corresponding to the C–N stretching in the AmpB structure (13). However, no significant change has been observed with anidulafungin FTIR analysis (supplemental Table S3).

A ¹H NMR experiment was next carried out to find out whether there was any H-bonding interaction between glucose and antifungal drugs. Comparison of the spectra of pure glucose, pure drugs and a combination of the two showed definite interaction between glucose and voriconazole. This was confirmed by the appearance of a broad new peak at δH 6.5, which became exchanged upon D₂O shaking (Fig. 1, lower panel). This indicates the signal to be due to a glucose hydroxyl. This signal was perhaps embedded in the upfield signals of pure glucose and underwent downfield shift upon the addition of voriconazole, which indicates slow H-bonding reaction. However, NMR spectra obtained for AmpB-glucose and anidulafungin-glucose did not show any significant effect (Fig. 2, a and b).

FIGURE 2.

Interaction and in vivo susceptibility analysis of antifungal agents with glucose. a and b, ¹H NMR spectra were obtained from pure glucose, pure anidulafungin, and the combination of both in d₆-DMSO (a), and similarly, from pure glucose, pure AmpB, and the combination of both in d₆-DMSO (b). Docking studies of voriconazole and amphotericin B toward d-glucose were performed in silico by using AutoDock. c, the detailed interactions of the drugs toward d-glucose highlighting three hydrogen bonds (i) and two hydrogen bonds stabilizing the complex formed (ii). d, differential in vivo susceptibility of antifungal agents in diabetic mice during Candida infection. The streptozotocin-induced diabetic Swiss albino mice model was used in this study. The bar diagram represents the efficiency of antifungal agents treating candidal infections in diabetic mice and nondiabetic mice. The light gray bar and light dark bar with lines represent the diabetic and nondiabetic mice, respectively, for each data set. The fungal load was significantly higher in diabetic mice in comparison with nondiabetic mice. Error bars represent mean ± S.D. *, p < 0.05 and **, p < 0.01 (diabetic mice versus nondiabetic mice).

Furthermore, to confirm the data obtained in vitro, docking studies were designed with the goal of studying the observed results. In accordance with the in silico results, the affinities of voriconazole and amphotericin B toward d-glucose were analyzed. For voriconazole and amphotericin B, the molecules were left close to glucose, allowing random contact with the molecule. Voriconazole demonstrated greater affinity for the d-glucose by ITC, and these results were reinforced by NMR experiments. These results were also corroborated by in silico studies comparing the numerous interactions (three hydrogen-bond interactions observed). However, the output energy toward d-glucose encountered in the well defined cluster generated after data mining observed was −1.9 ± 0.1 kcal mol⁻¹, whereas the energy toward the d-glucose observed for amphotericin B was −2.3 ± 0.6 kcal mol⁻¹. The different affinities observed the in silico data reflect the length of amphotericin B because this molecule is larger than voriconazole; with amphotericin B, more observations were possible because several of the interactions were longer; however, no interactions demonstrated more than two hydrogen bonds, reinforcing the results of the in vitro assays. In addition, a detailed observation of the interaction demonstrated that voriconazole forms three hydrogen bonds between stabilizing the voriconazole-glucose complex. The interactions observed for voriconazole were among the hydrogen atoms (HO1, HO4, and HO6) of the antifungal molecule and the nitrogen atoms (N9, N7, and N8), with distances of 3.19, 3.09, and 3.19 Å, respectively (Fig. 2c, panel i). On the another hand, amphotericin B had fewer interactions, presenting two hydrogen bonds among the oxygen atoms of amphotericin B (O) and the hydrogen of the glucose linked at carbons (C1H1 and C5H5), with distances of 2.46 and 2.78 Å, respectively (Fig. 2c, panel ii)

Finally, the efficacy of tested antifungal drugs was evaluated in diabetic mice and exhibited the same trends as observed in vitro analysis. The colony count of C. albicans was significantly reduced in control mice (nondiabetic) with low blood glucose level in comparison with diabetic mice (Fig. 2d). The activity of voriconazole was reduced almost 6-fold, whereas 2.8- and 1.8-fold reduction was observed for AmpB and anidulafungin, respectively, in diabetic mice as compared with nondiabetic mice. Voriconazole was the least active drug followed by amphotericin B and anidulafungin for diabetic mice infected with Candida.

DISCUSSION

The genus of Candida is commensal to the human gastrointestinal and genitourinary tract. In patients suffering with DM, an extra glucose level in the bloodstream can cause several infectious diseases ranging from superficial candidiasis to deep seated mycoses (14). As a major component of the culture medium, carbon source plays an important role in the growth of microorganisms including Candida spp. Carbon compounds generally range from sugars to organic acids, alcohols, polysaccharides, and so forth, whereas microbes prefer simple sugars such as glucose, sucrose, maltose, and lactose for their rapid growth, and these simple sugars increased the fungal population density (15). Fungal load or count is an important parameter of antifungal susceptibility.

Our initial approach was to investigate the antifungal drug susceptibility in culture medium containing different antifungal agents in combination with fixed concentrations of glucose in each set of experiments. This was prevented by the apparent failure of azole (voriconazole) followed by polyene (AmpB) antifungal agent in the presence of high glucose in medium. Antifungal agents showed a decreased rate of susceptibility when the glucose concentration in culture medium was increased. There might be several reasons for the decreased susceptibility of the antifungals, but their selectivity is interesting. Rodaki et al. (16) conducted a noteworthy study based upon the global impact of glucose on C. albicans transcriptome for the modulation of carbon assimilatory pathways during pathogenesis. The study revealed that glucose concentrations in the bloodstream have a significant impact upon C. albicans gene regulation, which reflects on the elevated resistance to oxidative, cationic stresses and resistance to an azole antifungal agent, miconazole. In their study, it was observed that no significant susceptibility level was observed for anidulafungin, whereas voriconazole becomes the most resistant followed by amphotericin B. This allowed us to check whether extra glucose level decreased the susceptibility to antifungals only in genes mediated as higher expression genes involved in drug resistance or whether there is any direct effect of glucose on antifungals.

Several strategies including in silico analyses are used to check the interactions between antifungals and glucose. These results suggest that voriconazole and amphotericin B have the strong affinity to bind with excess glucose molecules through hydrogen bonding. It seems that the voriconazole-glucose complex or the amphotericin B-glucose complex may not be as effective as their pure molecule. Anidulafungin is safe for use in DM patients who do not show any strong affinity to the glucose molecule, and in vivo evaluation with induced diabetic mice also revealed the potential of anidulafungin. This might be due to the presence of amide bonding in the chemical structure of anidulafungin, which helps in self-association rather than binding to the glucose molecule. However, the presence of amine (-NH₂-) in AmpB and (-F-) in voriconazole promote the association with glucose molecule by hydrogen bonding.

Not only do excess glucose levels in bloodstream alter the genetic changes for their adaptation in the microenvironment, but interaction in microenvironment is also a factor for increasing the resistant properties of pathogens. In the light of these findings, it is is important for researchers and pharmaceutical companies to make new antibiograms of DM patients suffering with infectious diseases as treating these conditions is crucial to clinical practitioners.