Rebaz F. Hamarawf*ab
aDepartment of Chemistry, College of Science, University of Sulaimani, Kirkuk Road, Sulaymaniyah City, 46001, Kurdistan Region, Iraq. E-mail: rebaz.hamarawf@univsul.edu.iq
bDepartment of Medical Laboratory Science, Komar University of Science and Technology (KUST), Qliasan St, Sulaymaniyah City, 46002, Kurdistan Region, Iraq
First published on 18th March 2024
In the present era, the increase in free radical species (FRs) and multidrug-resistant (MDR) bacteria represents a major worldwide concern for public health. Biofilm development and the overuse and misuse of antibiotics could lead to the adaptation of bacteria to antimicrobial agents. Consequently, finding novel multifunctional species with antibacterial, antioxidant, and antibiofilm properties has become crucial in the fight against challenging bacterial infections and chronic inflammatory conditions. Metal–organic frameworks (MOFs) with zinc and cobalt metal centers are widely utilized in biological and environmental remediation owing to their versatility. In this study, multifunctional Zn-MOFs and Co-MOFs were successfully synthesized with zinc and cobalt as metal centers and 4,6-diamino-2-pyrimidinethiol as an organic linker using a hydrothermal technique. Numerous characterization techniques were used to fully examine the MOF structure, functionality, chemical makeup, crystalline structure, surface appearance, thermal behavior, and magnetic characteristics; the techniques included XPS, PXRD, FTIR, FESEM, EDX, UV-visible, BET, BJH, TGA/DTG, DSC, and magnetic susceptibility measurement. The antioxidant, antibacterial, and antibiofilm activities of the MOFs were examined, and they demonstrated potent activity in each of these aspects. The proposed mechanisms of antibacterial activity suggest that bacterial cell death results from multiple toxic effects, including electrostatic interaction and lipid peroxidation, when MOFs are attached to bacteria, leading to the formation of reactive oxygen species (ROSs). Zn-MOFs exhibit high antibacterial and antibiofilm efficacy owing to their large surface-to-volume ratio and porous nature, while Co-MOFs exhibit high antioxidant capacity owing to their redox properties.
Herein, we present two novel MOFs resembling the PCP subclass of MOFs based on zinc or cobalt with 4,6-diamino-2-pyrimidinethiol as an organic linker. They were labeled Zn-MOF and Co-MOF. We investigated their antibacterial, antioxidant, and antibiofilm activities and found them to exhibit high efficacy. The antibacterial activities were assessed via the measures of inhibitory zone diameter (IZD), minimum bactericidal concentration (MBC), and minimum inhibitory concentration (MIC). Antibacterial activity was tested against four pathogenic bacteria: clinical isolates of Staphylococcus aureus and Pseudomonas aeruginosa, as well as their standard strains, Staphylococcus aureus ATCC 6538 and Pseudomonas aeruginosa ATCC 9027. Additionally, an antibiofilm inhibition assay was conducted for Staphylococcus aureus ATCC 6538, and a DPPH scavenging assay was used to illustrate the antioxidant activities of the MOFs. A comprehensive comparison of the characteristic properties of the MOFs was conducted, focusing on antioxidant, antibacterial, and antibiofilm efficacies.
The ThermoFisher Scientific K-Alpha X-ray Photoelectron Spectrometer (XPS) was used to analyze the electronic states and elemental compositions on the surface. Fourier transform infrared (FTIR) spectra with a resolution of 4 cm−1 in the 400–4000 cm−1 range were recorded using PerkinElmer's instrument (PerkinElmer, Inc., USA). A Cary 60 UV-visible spectrophotometer (Agilent Technologies Inc., USA) was used to produce UV-visible absorption spectra in the 200–800 nm range. Brunauer Emmett Teller (BET) isotherms were used to analyze the surface area, pore volume, and pore diameter of the material using an automated gas sorption analyzer from Quantachrome (USA). The magnetic property and oxidation state of the MOFs were characterized using the magnetic susceptibility balance from Sherwood Scientific Ltd. (MKII, UK). The synthesized MOFs were subjected to an inductively coupled plasma optical emission spectrometer (ICP-OES, PerkinElmer Optima 2100 DV, USA) to determine the concentration of leached cobalt and zinc.
The MIC of the MOFs was assessed by treating a series of two-fold concentrations, starting from 10 μg mL−1 and ranging up to 2.56 mg mL−1, using DMSO as a solvent. The study involved mixing MOFs (0.5 mL), bacterial inoculum (0.5 mL), and NB (1 mL) in sterilized tubes with a positive control group consisting of bacteria and DMSO and a negative control group, including NB and MOFs. Tubes were incubated at 37 °C for 24 hours compared to controls, and their MIC values were confirmed by transferring 300 μL to a sterilized microplate and recording the OD at 600 nm. The positive controls exhibited ODs above 0.09, while the negative controls exhibited an OD of 0.042 or lower, as shown in Tables 4S, 5S and Fig. 2S (ESI†). The MBC was determined by spreading a loopful from the culture wells on the MHA plates after no visible growth was observed and incubating at 37 °C for 24 hours.
For the experimental setup, a sterilized 96-well microtiter plate was utilized. In each well of the plate, the following components were added: 20 μL of bacterial suspensions, 130 μL of NB, and 150 μL of two-fold concentrations of the MOFs, starting from 10 μg mL−1 to 5.12 mg mL−1. A positive control was designed by combining 20 μL of bacterial suspensions, 130 μL of NB, and 150 μL of DDW, while a negative control was established by mixing 150 μL of NB and 150 μL of DDW. The plate was then gently shaken for 10 minutes and incubated under static conditions at 37 °C for 24 hours.
To conduct the biofilm inhibition assay, the entire medium from the plate was discarded, and the plate was rinsed four times with 350 μL of PBS buffer. It is important to dispose of the non-attached cell waste in containers containing 5% sodium hypochlorite. After rinsing, the plate was inverted on a dry paper towel and lightly tapped to remove any excess water. The attached biomasses in each well were stained by adding 200 μL of crystal violet (0.1%) and incubated using an orbital shaker (IKA, model® KS130 basic) set at 240 rpm for 30 minutes. The stain with excess biomass was discarded from the wells and washed four times with DDW. The plate was allowed to dry at room temperature. The suspended crystal violet (CV) stain was solubilized with 200 μL of a 30% glacial acetic acid solution, and the OD was measured. The measured optical density (OD) values served as a measure of the presence of bacterial biofilm, which was subsequently transformed into a percentage representing the extent of inhibition (Tables 6S and 7S with Fig. 6S†). The OD of the blank was subtracted from all values. The percentage (%) of biofilm inhibition was calculated using the formula described below (eqn (1)). Subsequently, the IC50 values of the antibiofilm inhibition of the MOFs were ascertained by employing nonlinear regression of the curve using GraphPad Prism 8 software (San Diego, CA, USA, 2021).
![]() | (1) |
The reduction in DPPH was observed by reading absorbance at 517 nm after incubating the tube on an orbital shaker in the dark at room temperature for about 30 minutes. As a control, the absorbance of 1 mL DPPH with 1 mL DMSO was measured, and as a standard, the absorbance of 1 mL DPPH with 1 mL of a two-fold series concentration of ascorbic acid (AA) ranging from 10 μg mL−1 to 256 μg mL−1 was recorded at 517 nm. For all samples, the percentage inhibition was calculated using the following equation (eqn (2)):
![]() | (2) |
The N 1s spectrum, as depicted in Fig. 2F, showed a single low-intensity peak with the presence of shoulders at a binding energy of 398.99 eV. The peak could be attributed to four different environments of nitrogen bonds: CN–C, Zn–N, C–N, and N–H; they were identified at 397.75, 399.16, 400.61, and 402.7 eV, respectively.42 According to the survey spectrum of the Co-MOF (Fig. 2G), the primary elements found on the surface of the MOF were carbon (42%), oxygen (41%), cobalt (8%), sulfur (7%), and nitrogen (2%). The findings shown in Fig. 2H demonstrated that the Co 2p spectrum of the Co-MOF was deconvoluted into two distinct peaks consistent with Co 2p3/2 and Co 2p1/2, which had binding energies of 781.5 and 797.38 eV, respectively. Moreover, the presence of Co3+ along with Co2+ was further supported by two minor additional peaks at 786.37 and 798.2 eV, which corresponded to the satellite peaks of Co3+.47 The characteristics and location of these peaks indicate the oxidation–reduction of Co atoms. The intensity of the Co 2p peak is directly linked to the cobalt content in the Co-MOF.
To examine the oxidation state of the cobalt species, Co 2p3/2 was deconvoluted into three peaks. Two of the peaks corresponded to the different environments of Co3+ species at binding energies of 778 and 781.32 eV, while the third peak at the binding energy of 783 eV was attributed to Co2+. This phenomenon was further supported by the magnetic behavior of the Co-MOF. However, the presence of oxidative sulfur (Co–O–SO) on the surface of the Co-MOF is confirmed by two asymmetrical peaks at 161.87 and 168.55 eV (Fig. 2I). These peaks also indicate the absence of free elemental sulfur.44 The peak at 168.55 eV is explained by Co–S–O, while the peak at 161.87 eV can be deconvoluted into two peaks at 161.84 and 164.3 eV, which correspond to the Co–S bond for S 2p3/2 and the phenyl–S–O bond for S 2p1/2, respectively.44 In the O 1s spectrum (Fig. 2J), three distinct peaks were observed at energy values of 531.17, 531.9, and 532.46 eV. These peaks correspond to different chemical bonds involving oxygen atoms on the surface of the structure. The peaks are associated with Co–O, Co–S–C, and Co–SO2–C bonds, respectively.48 The C 1s spectrum with high resolution had four peaks at 284.29, 284.81, 285.8, and 288.47 eV, as shown in Fig. 2K. They were associated with CC, C–S, C–N, and C
N bonds, respectively. The Co-MOF N 1s XPS spectrum (Fig. 2L) had one peak at 399 eV and a weak shoulder at 396.9 eV, which were related to C
N–C for the imine part of the organic linker. The main N 1s peak could be split into two peaks at 399.2 and 401.2 eV binding energies, which were related to the Co–N and C–N bonds.
The X-ray powder diffraction (XRD) data of the synthesized MOFs confirm their crystalline properties and reveal different structural analogs, as indicated by the various characteristic peaks shown in Fig. 3. Diffraction peaks at 28.59°, 31.60°, 34.35°, 36.17°, 47.69°, 56.63°, 62.76°, 67.87°, and 77.15° were observed, corresponding to the (0141), (0162), (1082), (0195), (1127), (11−117), (02−110), (0255), and (1217) planes of Zn-MOF, respectively. The Debye–Scherrer equation was used to calculate the crystallite sizes of 31 nm for Zn-MOF. Additionally, most of the peaks indicated the formation of a tetrahedral geometrical structure, as confirmed by the XPS, UV-visible, and magnetic properties of the Zn-MOF. The PXRD peaks were aligned with ICSD files 98-004-8345 and 98-009-4606, as alternatives to the standard JCPDS file.
The diffraction peak positions of the Co-MOF, as shown in Fig. 3C, confirm its high purity and crystallinity. This observation indicates the formation of an octahedral geometry. As shown in Fig. 3, zinc ions in Zn-MOF have a tetrahedral coordination geometry with a coordination number of four due to their smaller size, forming stable complexes with sp3 hybridization. Cobalt ions, with their larger ionic radius and additional d-orbitals, adopt an octahedral coordination geometry, accommodating six ligands with sp3d3 hybridization in Co-MOF. This difference in coordination numbers affects interactions with ligands. Various characteristic peaks observed in the PXRD pattern corroborate this finding. Additionally, confirmed characteristic data from XPS, UV-visible, and magnetic properties of the Co-MOF further support its crystalline structure.49 The Co-MOF exhibits significant diffraction peaks at the following specific angles: 10.68°, 15.27°, 17.37°, 23.9°, 29.78°, 32.94°, 34.73°, 36.60°, 38.38°, 41.63°, 44.70°, 47.76°, 52.43°, 59.42°, 61.79°, 67.69°, 73.46°, and 76.89°. These peaks can be indexed to different (hkl) crystallographic planes, specifically (101), (211), (220), (123), (242), (234), (144), (153), (523), (524), (336), (381), (365), (385), (117), (191), (028), and (157), respectively. The peak positions align with the ICSD cards 98-001-9072 and 98-000-1782. The crystallite size of the Co-MOF, determined through PXRD analysis, is approximately 43 nm. The average particle size of MOFs, as observed in the FE-SEM micrographs, is consistent with the crystallite sizes observed in the XRD.
Fourier transform infrared spectroscopy (FTIR) characterization spectra are presented in Fig. 4A. The results demonstrate characteristic peaks indicating the formation of MOFs and the disappearance of bands associated with organic linkers. These changes occur due to the bonding between the donor atoms of the ligand and the metal ions; consequently, MOFs are formed. Prior to the reaction, the absorption bands of the organic linker were detected, ranging from 2700 to 3500 cm−1. These bands represent the stretching of the C–H, S–H, and N–H bands associated with the free NH2 groups.50 The stretching vibrations of water molecules adsorbed on the surface of the ligand were also correlated with these findings. However, following the formation of Zn-MOF and Co-MOF, changes were observed in the absorption bands related to the stretching vibration of the N–H bonds.49 These bands either shifted or disappeared due to the coordination and bonding between the metal ions and the organic linker within the MOF structure. The absorption bands observed in the range of 2900–3450 cm−1 in the Zn-MOF spectra provide evidence of water molecule coordination to the Zn(II) ion and the consequent disappearance of the N–H bond due to nitrogen coordination with Zn ions.51 Moreover, the absorption peaks detected between 1650 and 1400 cm−1 clearly indicate the presence of C–H bending vibrations and validate the absence of the C–SO2–Zn bond, as corroborated by XPS and XRD characteristic data.
![]() | ||
Fig. 4 (A) FTIR spectra of Zn- and Co-MOFs, (B) UV-visible spectra of the MOFs, (C) N2 adsorption–desorption isotherm, and (D) BJH pore size distribution curve for the MOFs. |
Specifically, the absorption bands at 1588 cm−1 in the Zn-MOF and at 1552 cm−1 in the Co-MOF signify the bending vibration of the CC and C
N bonds, respectively. Additionally, the absorption peak observed at 1420 cm−1 suggests the bending vibrations of N–H molecules. Notably, in the Co-MOF, the characteristic bands observed at 1350 cm−1 confirm the presence of Co–SO2–C bonds. These may arise from the oxidation–reduction process of Co ions within the MOF centers, leading to the oxidation of sulfur in half of the MOF structural molecule.52 The Zn-MOF and Co-MOF demonstrate significant M–O stretching bands, resulting from the asymmetric and symmetric vibrations of water molecules bonded to the metal ions. These vibrations cause slight frequency shifts owing to conjugation with the carbon–nitrogen bond. Additionally, low-intensity bending vibrations are observed at 615, 491, and 435 cm−1 in the Zn-MOF, corresponding to M–O, M–S, and M–O bonds, respectively. Similarly, in the Co-MOF, bands at 834, 801, and 680 cm−1 are observed, corresponding to M–O, M–S, and M–O bonds, respectively.53
UV-visible techniques are widely employed for the characterization of Zn- and Co-MOFs, and they play a crucial role in elucidating the structural and optical properties of the compounds. One key aspect observed in the UV-visible spectra, as depicted in Fig. 4B, is the difference between the UV peaks of the linker (amine) and those of the MOFs formed. The ligand exhibits two peaks at 243 and 293 nm. In contrast, the UV peak of the formed Zn-MOF at 270 nm is shifted and broadened to a shorter wavelength compared to that of the ligand, while the UV peak of the Co-MOF at 255 nm is shifted and narrowed to a longer wavelength compared to the ligand peaks. This discrepancy arises from several factors. First, the presence of metal ions within the MOF structure introduces new electronic states that can interact with the ligand, leading to changes in the energy levels of the resulting n–π* transitions. Additionally, the arrangement and coordination of the ligand with the metal ions in the MOF framework can influence conjugation and electronic delocalization, thereby causing a shift in the UV peak of the MOFs compared to the ligand. This shift can provide valuable insights into the structural changes that occur upon Zn- and Co-MOF formation. This can indicate the ligand–metal ion interaction differently, the formation of new metal–ligand bonds, and the alteration of the overall structural integrity of both MOFs.
The pore diameter, pore volume, and specific surface area were determined using Brunauer–Emmett–Teller (BET), Barrett–Joyner–Halenda (BJH), and Langmuir methods, analyzing the nitrogen adsorption–desorption isotherm at a constant temperature. According to the BET isotherm, the Zn-MOF exhibited a specific surface area of 45.88 m2 g−1. Conversely, the Co-MOF exhibited a surface area of 20.39 m2 g−1, as shown in Table 1 and Fig. 4C. The reduction in the Co-MOF specific surface area compared to the Zn-MOF may be attributed to the existence of substantial mesoporous interlayer voids inside the Zn-MOF structure. The use of the BJH isotherm and the pore size distribution curve shown in Fig. 4D confirm the existence of mesoporous properties in the MOFs.54 These features primarily originate from the high number of voids in the Co-MOF compared to the Zn-MOF. It was observed that Co-MOF had comparable total pore volumes, which were greater than the pore volume of Zn-MOF, as indicated in Table 1. The results displayed in Fig. 4D and Table 1 demonstrate that the pore diameter of Co-MOF obtained was 21.65 nm compared to the 7.02 nm diameter of Zn-MOF, confirming the presence of mesoporosity in the MOF structures.29 Among the six types of BET isotherms, the BET isotherm of the MOFs confirms hysteresis in a specific type known as the type IV BET isotherm, as depicted in Fig. 4C.55 The type IV isotherm exhibits two distinct patterns, both associated with pore width. When the width exceeds a critical threshold determined by the adsorption properties and temperature of the material, this type of isotherm can be obtained at high relative pressure in the pores.
Physisorption characteristic data | Zn-MOF | Co-MOF | |
---|---|---|---|
BET summary | Specific surface area (m2 g−1) | 45.883 | 20.393 |
BJH adsorption and desorption summary | Surface area (m2 g−1) | 59.136 | 23.698 |
Pore volume (cm3 g−1) | 0.114 | 0.081 | |
Pore radius dV(r) (Å) | 35.149 | 108.263 | |
Pore diameter (nm) | 7.03 | 21.65 | |
Langmuir summary | Surface area (m2 g−1) | 787.399 | 363.132 |
This observation could be attributed to the various structural analogs of MOFs, including factors such as the connectivity between ligands and metals, as well as the distribution of pore sizes throughout the structure and the geometry of the pores. The high surface area and large pore diameter typically correspond to a greater number of active sites, leading to higher catalytic activity.54 Consequently, Zn-MOF displays a substantial surface area due to the greater quantity of active sites in comparison to Co-MOF, but Co-MOF has a large pore diameter. The Langmuir isotherm analysis revealed that Zn-MOF exhibited a monolayer surface area of 787.39 m2 g−1, while Co-MOF exhibited a surface area of 363.13 m2 g−1. The Langmuir isotherm assumes a monolayer adsorption process, enabling the determination of essential parameters, such as maximum adsorption capacity and equilibrium constant. These parameters offer valuable insights into the adsorption phenomenon and surface interactions during catalytic processes.
High-resolution field-emission scanning electron microscopy (FESEM) was utilized to evaluate the surface morphology, particle size distribution, and structural features of the MOFs. This characterization technique provides valuable insights into the microstructure and surface features of MOFs. The obtained FESEM micrograph, the particle size distribution histograms, and the EDX spectra of the MOFs are depicted in Fig. 5. They show a spherical morphology and varying degrees of agglomeration, with grains combined. The FESEM approach, coupled with EDX, is used to precisely identify the chemical composition of the synthesized MOFs (Fig. 5). Both MOFs display differences in the rate of aggregation, morphological configuration, and distribution of particle sizes. However, the presence of zinc and cobalt metal ions within the 4,6-diamino-2-pyrimidinethiol framework, acting as an organic linker, is evidenced by the uniform distribution of Zn, Co, C, N, and S elements. Additionally, O represents coordinated water molecules, as shown in Fig. 5C and D.
Moreover, EDX mapping images, along with XPS analysis (Fig. 2), demonstrate a consistent distribution of Zn and Co throughout the MOF materials. This incorporation of metal species with oxygen, nitrogen, and sulfur groups facilitates the formation of various voids as active sites for the catalytic activity of these materials. Consequently, a well-defined active site forms on the surface of the MOFs, thereby enhancing their antibacterial, antibiofilm, and antioxidant capacities.
The FESEM micrograph offered valuable insights into the particle properties of the MOFs. Zn-MOF displayed spherical particles of around 28 nm in diameter, while Co-MOF showed semi-spherical particles with an average diameter of 40 nm. The EDX elemental mappings delineate the elemental composition and distribution of the MOFs, confirming the presence of major and trace elements, revealing spatial distribution, and providing semi-quantitative information within the EDX mapping. Additionally, it generates elemental mapping images, reveals interactions between elements, and can detect surface contaminants. The EDX elemental mapping of the MOFs confirms the formation of the MOFs with all elements within the MOF structure, as shown in the ESI in Fig. 7S and 8S.† The FESEM and EDX mapping images conclude that the weak repulsion caused by the diamagnetic characteristic of the MOFs inhibits the formation of larger agglomerations and grain combinations in the MOFs.
The diamagnetic effect creates a weak repulsion among the MOFs, which reduces the tendency for agglomeration.56 Moreover, the type of metal ions and the shape of the MOFs affect their surface features, particle sizes, and clustering behavior. These factors explain why Zn-MOF and Co-MOF have different patterns of agglomeration and grain formation.
Differential Thermogravimetry (DTG), Differential Scanning Calorimetry (DSC), and Thermogravimetric Analysis (TGA) were used to evaluate the thermal stability of the MOFs. The experiments were carried out at temperatures ranging from 30 to 600 °C, with a heating rate of 10 °C min−1. These techniques are commonly used to understand the thermal behavior of MOFs. The thermograms are depicted in Fig. 6. The TGA/DTA curves were used to analyze the decomposition steps of the MOFs. TGA can be utilized to differentiate between water molecules situated and coordinated in the lattice structure of the MOFs or adsorbed on the surface of the MOFs in terms of coordinated water molecules in the structure of the MOFs.57 The elimination of water molecules attached to the surface of MOFs usually occurs at temperatures below 200 °C, corresponding to the water molecules linked to the surface of the MOFs.
![]() | ||
Fig. 6 (A) TG/DTG of the Zn-MOF, (B) DSC of the Zn-MOF, (C) TG/DTG of the Co-MOF, and (D) DSC of the Co-MOF. |
However, water molecules coordinated in the lattice structure are usually expelled above 200 °C. As shown in Fig. 6A, a weight loss of 4.5% around 45 °C could be attributed to the loss of adsorbed water molecules on the surface of the Zn-MOF, but this was not observed in the Co-MOF (Fig. 6C). This behavior could be attributed to the stability of water molecules related to Zn-MOF, which is strongly related to the organic linker and metal ions and is not observed in Co-MOF.58 The TGA of both MOFs reveals a continuous reduction in weight starting at around 290 °C and lasting until approximately 400 °C.
Based on the DTG, the Zn-MOF was decomposed at 398.8 °C and the Co-MOF was decomposed at 394.6 °C. At these points, the MOF structures underwent decomposition, as further confirmed by DSC thermograms. Initially, during the decomposition process, adsorbed water, along with the dehydration of coordinated water molecules, deamination, and dethiolation, occurred. Furthermore, the TGA and DTG curves indicate that the crystallinity of the MOF is completely decomposed at temperatures higher than 500 °C, which is further supported by the DSC curves (Fig. 6B and 7D). DSC is a technique that measures the heat flow of a sample as its temperature changes. It can provide information about various thermal properties of a material, such as heat capacity, phase transitions, crystallization, melting point, and decomposition point. The DSC thermograms of Zn-MOF and Co-MOF showed that phase transition onset occurred at 300 °C and 265 °C, respectively. The exothermic peaks at 401 °C for Zn-MOF and 381 °C for Co-MOF suggested the crystallization of the MOFs. The decomposition of the MOFs was indicated by the endothermic peaks corresponding to heat absorption during the phase transition at 536 °C for Zn-MOF and 482 °C for Co-MOF. These results confirmed that Zn-MOF had higher thermal stability than Co-MOF. The magnetic susceptibility balance from Sherwood Scientific Ltd. (MKII, UK) was employed to determine the oxidation state and magnetic properties of the MOFs. The results agreed with the XPS analysis techniques applied to the MOFs. The mass susceptibility (χm)/volume susceptibility (χv) of the MOFs were used to measure the magnetic properties of Zn-MOF and Co-MOF. The results indicated that Co-MOF displayed diamagnetic characteristics, with most of the Co(II) in its MOF structure oxidized to Co(III), as corroborated by the XPS data. The XPS data revealed that the Co-MOF structure contained Co2+ and Co3+ sites (Fig. 2H). Conversely, Zn(II) in Zn-MOF remained in its oxidation state with diamagnetic behavior. Based on the characterization techniques, Zn in Zn-MOF exhibited a tetrahedral structure with the hybridization of sp3, while Co-MOF possessed an octahedral structure with the hybridization of d2sp3.
![]() | ||
Fig. 7 Proposed mechanism of the antibacterial activities of the Zn-MOF and Co-MOF on the clinical isolate of Pseudomonas aeruginosa bacterium.31 |
Bacterial strains | IZD at a concentration of 0.64 mg dL−1 (with 80 μL per well) | |||
---|---|---|---|---|
Zn-MOF | Co-MOF | Linker (amine) | DMSO | |
S. aureus (clinic) | 20 ± 4 (24) mm | 18 ± 0 (18) mm | 0 mm | 0 mm |
S. aureus ATCC 6538 | 32 ± 2 (30) mm | 20 ± 2 (22) mm | 0 mm | 0 mm |
P. aeruginosa (clinic) | 24 ± 4 (20) mm | 18 ± 2 (16) mm | 0 mm | 0 mm |
P. aeruginosa ATCC 9027 | 16 ± 2 (18) mm | 16 ± 0 (16) mm | 0 mm | 0 mm |
The antimicrobial effectiveness of MOFs and other antimicrobial agents depends heavily on the structures, compositions, surface functionalities, and bacterial cell wall properties of the material. Research has revealed that Gram-positive bacteria, with their thick peptidoglycan layer, encounter difficulties penetrating the surface functionality of MOFs. Conversely, Gram-negative bacteria, possessing a thinner peptidoglycan layer, demonstrate higher sensitivity, resulting in lower MIC values.59 For instance, as illustrated in Table 3, both Gram-positive bacteria, such as the clinical isolate of S. aureus and its standard strain, S. aureus ATCC 6538, exhibit high MIC values of 0.64 mg mL−1. In contrast, the Gram-negative bacteria P. aeruginosa, both as a clinical isolate and its standard strain, P. aeruginosa ATCC 9027, display lower MIC values of 0.08 mg mL−1.
Bacterial strains | Antibacterial activity via MICs | Antibacterial activity via MBCs | ||
---|---|---|---|---|
Zn-MOF | Co-MOF | Zn-MOF | Co-MOF | |
S. aureus (clinic) | 0.64 mg mL−1 | 0.64 mg mL−1 | 2.56 mg mL−1 | 2.56 mg mL−1 |
S. aureus ATCC 6538 | 0.64 mg mL−1 | 0.64 mg mL−1 | 2.56 mg mL−1 | 1.26 mg mL−1 |
P. aeruginosa (clinic) | 0.08 mg mL−1 | 0.08 mg mL−1 | 2.56 mg mL−1 | 2.56 mg mL−1 |
P. aeruginosa ATCC 9027 | 0.08 mg mL−1 | 0.08 mg mL−1 | >2.56 mg mL−1 | >2.56 mg mL−1 |
The cytotoxicity of Zn-MOF and Co-MOF is possibly associated with the production of ROS. These ROS selectively target bacteria while leaving non-bacterial cells unharmed, resulting in damage to the bacterial cell wall and nucleic acids. Such damage can lead to protein leakage and, ultimately, bacterial death. Additionally, the cytotoxic effects of MOFs on bacteria are attributed to their electrostatic interactions.
Both MOFs have different metal nodes and share the same organic linkers. They provide various structural surface charges and functionalities. However, bacteria exhibit various membrane structures that display a negative charge for both Gram-negative and Gram-positive bacteria. This charge arises from their main chemical constituents: teichoic or lipoteichoic acid-containing peptidoglycan in Gram-positive bacteria and phospholipids in Gram-negative bacteria.60 XPS and IR techniques partially reveal the presence of positively charged metal ions, such as Co2+/Co3+ and Zn2+, on the surface structure of MOFs.61,62 In contrast to bacteria, the surface structure of MOFs is predominantly positive, facilitating interaction with the negatively charged bacterial surface. This interaction induces an electrostatic attraction force,61 which may inhibit cell wall synthesis, retard protein synthesis, and interact with bacterial genetic material (DNA), ultimately leading to the destruction of bacterial cell walls and causing protein and cytoplasm leakage, as depicted in Fig. 7.
Both MOFs showed similar MIC values against various bacterial strains, with S. aureus strains showing 0.64 mg mL−1 and P. aeruginosa strains showing 0.08 mg mL−1 (as shown in Table 3). The MBC values of the MOFs are also presented in Table 3, indicating that the MOFs exhibited bactericidal activities against S. aureus strains, with the MBC values being four times higher than the MIC values (MBC/MIC ≤ 4).63 However, when the MOFs were used against the P. aeruginosa strains, the MBC values were thirty-two times larger than the MIC values (MBC/MIC > 4).63 According to the IZD and MIC results, the MOFs exhibited potent antibacterial activity against all bacteria. Additionally, the MOFs demonstrated bactericidal activity against S. aureus strains, while they displayed bacteriostatic activity against P. aeruginosa strains based on MBC/MIC ratios. The MBC values of Zn-MOF and Co-MOF against S. aureus ATCC 6538 differ, as shown in Table 3. Several factors supported by characterization techniques can be responsible for this variance. The distinct metal oxidation states of CoII/CoIII and the geometric differences between the MOFs play significant roles, contributing to a synergistic effect that explains this variation. XPS, PXRD, and FTIR corroborate this explanation, suggesting heightened oxidative potential and improved interaction properties, which result in high oxidative stress.64 The coordination geometry of the ligands around the metal ion can also have a big effect on how it interacts with bacterial cells, which could change the MBC. MOFs of varying shapes might better reach bacterial membranes or cellular targets, enhancing their effectiveness against bacteria. Notably, functional groups in Co-MOF, such as (–CC–SO2–Co–), may directly interact with bacterial cells or facilitate cellular uptake, potentially enhancing bactericidal efficacy compared to Zn-MOF.
It is widely accepted that the formation of biofilm biomass is directly linked to the amount of CV adsorbed, which occurs due to interactions with negatively charged bacteria and polysaccharides within the EPS. However, the types of solvents, broth media, and surface types can significantly affect the ability of bacteria to form a biofilm and adhere to a surface. Consequently, the use of DMSO solvent, either alone or in combination with DDW (in a 1:
5 ratio), affects biofilm formation. Therefore, DDW was used solely to prepare a two-fold serial concentration, which optimally does not affect biofilm growth. Interestingly, these MOFs effectively inhibited the formation of S. aureus ATCC 6538 biofilms, as demonstrated by measuring the half maximal inhibitory concentration (IC50) for the percentage of biofilm inhibition of these MOFs.
The IC50 value represents the concentration of the MOFs required to inhibit biofilm formation or reduce biofilm biomass by 50%. According to the findings presented in Fig. 9A, the Co-MOF demonstrated a lower IC50 value compared to the Zn-MOF, which serves as a pivotal parameter for assessing the efficacy of these MOFs as antibiofilm agents. The IC50 of Zn-MOF is 0.01337 ± 0.007 mg mL−1, while the IC50 of Co-MOF is 0.01269 ± 0.007 mg mL−1. These results indicate that the high electrostatic interaction in Co-MOF could damage the bacterial community faster than in Zn-MOF, which possesses a small particle size with a high surface-to-volume ratio. Zn-MOF can penetrate the bacterial cell wall more readily due to its smaller size, while the higher oxidation potential of Co-MOF generates more electrostatic interactions with ROS that damage bacterial DNA and proteins.
This distinctive property renders Co-MOF a promising candidate for antibiofilm activity, promoting substantial electrostatic interactions with EPS that hinder biofilm formation. These results indicate that both Zn-MOF and Co-MOF exhibit comparable effectiveness against S. aureus ATCC 6538 in terms of biofilm inhibition. Furthermore, the exceptional redox stability exhibited by the Co-MOF makes it well suited for applications involving oxidation–reduction reactions. After assessing the antibiofilm activity, it was observed that the efficacy of the MOFs against biofilm was nearly the same, and transparent microwells were evident as the MOF concentration gradually increased to 0.160, 0.32, and 0.64 mg mL−1. The complete disappearance of the biofilm against S. aureus ATCC 6538 was observed at a concentration of 0.64 mg mL−1, which was equal to the MIC value against the S. aureus strains.
Co-MOF demonstrated a scavenging activity of 52.17%, while Zn-MOF exhibited 40.83% at a concentration of 160 μg mL−1. Furthermore, an increase in MOF doses resulted in a proportional increase in antioxidant activity. The antioxidant activity of the MOFs appears to stem from two proposed mechanisms.71 First, the transfer of free electrons from various oxidation states of Co-MOF, containing Co3+ and Co2+ ions, facilitates the neutralization of free radicals, reducing their reactivity and harmful effects.72 Second, there is the transfer of free electrons from the oxygen atom of the MOFs to the free radicals on the nitrogen atoms in the DPPH. Notably, the observed high antioxidant capacity is particularly evident in materials with a high surface-to-volume ratio, as exemplified by Zn-MOF.73 There is a lack of comprehensive research on the antioxidant characteristics of MOFs. The findings of our study, however, show great promise and offer a valuable starting point for further investigation into the potential of Zn-MOF and Co-MOF as multifunctional materials. Therefore, it is essential to design and develop MOFs with inherent antibacterial, antibiofilm, and antioxidant properties to enhance their biocompatibility by optimizing their structures and exploring their potential as innovative solutions for combating oxidative stress-related issues. By enhancing biocompatibility and addressing oxidative stress concerns, MOFs hold promise for advancing various fields and improving human health and well-being. Both MOFs exhibited oxidant characteristics by providing oxidative stress to bacterial species and antioxidant behavior against free radical species. The dual properties of MOFs, acting as both oxidants and antioxidants, depend on factors such as the type of metal ions, organic ligands, and types of cells or molecules that contribute to the multifaceted behavior of the MOFs.74 After extensive reviews, some studies have shown that different kinds of MOFs can display antibacterial, antibiofilm, and antioxidant activities, as demonstrated in Table 4. These studies have illustrated that each MOF is suitable for one of the specified applications. More studies should be conducted to investigate the potential of MOFs for multifunctional applications in biomedical fields. The leakage of metals from the MOF frames was characterized using ICP analysis. Both Zn-MOF and Co-MOF were found to release metal ions at concentrations below the detection limit at the working pH (pH = 7.2). The leaching of zinc and cobalt ions was minimal, making it negligible for applications involving antibacterial, antioxidant, and antibiofilm activities. However, variations in pH can lead to the decomposition of the MOF frame, particularly in acidic or basic conditions. Metal ions can leak out of the MOF structure up to parts per billion (ppb) when the pH is basic, as shown in Table 9S (ESI†). Leaching tests were performed to determine how much Zn and Co ions were being released and to observe whether the MOFs could be used again, especially for their antioxidant properties in future uses. MOFs are susceptible to degradation under various environmental conditions, including changes in pH and heat exposure. Stability depends on metal ions, organic linkers, coordination bonds, and the pore environment.75 Enhancement strategies include using robust metal ions, rigid and symmetrical organic linkers; incorporating functional groups; and using solvothermal or hydrothermal synthesis methods, and post-synthetic treatments.75 These methods help minimize structural flexibility, enhance coordination bonds, and remove residual solvent molecules, thereby enhancing MOF stability.
MOFs | Organic linkers | Antioxidant activity (IC50) | Antibiofilm activity (IC50) | Antibacterial activity against | Ref. |
---|---|---|---|---|---|
NH2-MIL-125 | 2-Aminoterephthalic acid | — | E. coli | E. coli | 33 |
S. aureus | S. aureus | ||||
— | |||||
Zr-MOF and Ti-MOF | 2-Aminoterephthalic acid and silane surface modification | Animal study and plasma detection via DPPH˙ | — | — | 38 |
Zn-PDA | 2,6-Pyridine dicarboxylic acid | — | — | E. coli | 51 |
ZnO@ZIF-8 | 2-Methylimidazole | — | — | E. coli and S. aureus | 66 |
Zn-MOF | 4,4′-Bipyridine | — | — | S. epidermidis, E. coli | 67 |
Calcium gallate MOFs (MIL-155 and MIL-156) | Gallic acid | Dichlorofluorescein diacetate (125 and 250 mg mL−1) | — | — | 68 |
AgTAZ, ZIF-67, and Co-SIM-1 | 1,2,4-Triazole and 2-methylimidazole | — | — | E. coli CECT-4102 Pseudomonas putida CECT-4584, and Saccharomyces cerevisiae CECT-1170 | 69 |
Co-MOF@graphene oxide | 1,4-Benzenedicarboxylic acid | Via DPPH˙ | E. coli ATCC-25922 S. aureus ATCC-33591 | 70 | |
Zn-MOF and Co-MOF | 4,6-Diamino-2-pyrimidinethiol | DPPH˙ (0.2676 and 0.2197 mg mL−1) | Against S. aureus ATCC 6538 (0.01337 and 0.01269 mg mL−1) | S. aureus (clinic) S. aureus ATCC 6538 P. aeruginosa (clinic) P. aeruginosa ATCC 9027 | This work |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ra00545g |
This journal is © The Royal Society of Chemistry 2024 |