Impact of active sites on encapsulation of curcumin in Metal Organic Frameworks

In this study, we present the investigation of the curcumin loading and release properties of four different Metal-Organic Frameworks (MOFs) with varying metal centres and organic ligands. Through our research, we have identified HKUST-1 and MIL-100, highly porous copper and iron-containing MOFs, that exhibit specific interactions with curcumin, leading to high encapsulation efficiencies (55%–75%) even at low concentrations as 6 ppm. The binding modes of curcumin onto MOFs have been investigated using a combined experimental and computational approach. Furthermore, our drug-releasing studies have revealed slow and prolonged release for over two days, which further indicates the specific interactions of curcumin with HKUST-1 and MIL-100. To the best of our knowledge, this is the first comparative study that investigates the drug delivery properties of curcumin using Copper, Ferrous, and Zinc MOFs. Our findings pave the way for the development of stable, highly interactive MOFs as drug carriers for curcumin, which has the potential to overcome its poor aqueous solubility and rapid metabolism, and enhance its pharmacological activities in medicine.

MOFs have proven to be a highly promising platform for drug delivery due to their unique properties, such as the ability to precisely control their structure, and pore dimensions, as well as their potential for straightforward surface functionalization, high drug loading capacities, and controlled release of therapeutics in biological environments [19][20][21][22][23][24][25]. These features, along with their potential for synergistic/dual drug loading/ releasing and protection/stabilization of biomolecular therapeutics, make MOFs a versatile and valuable tool in the field of drug delivery [26][27][28][29][30]. One of the main obstacles in utilizing MOFs for drug delivery is the limited knowledge of their understanding of their biocompatibility with biological systems [31]. A few studies on the biotoxicity of MIL-100 (MIL: Materials of Institute Lavoisier) [32], and HKUST-1 ( HKUST: Hong Kong University of Science and Technology) [33,34], used in this study, have been investigated. However, there is still a lack of comprehensive understanding of how MOFs interact with biological systems. Therefore it is important to select metals and organic ligands that are biocompatible when constructing MOFs. For example metals such as Zinc and Iron metals, used in this work, are reported to have low toxicity with oral lethal dose 50 (LD50) values from 350-450 mg kg −1 [31,35]. Additionally, polycarboxylic acid linkers used in this work, such as 1,3,5benzene tricarboxylic acid and 1,4-benzene dicarboxylic acid are reported to not to have very toxic effects at first sight with rat oral doses of 1.13 and 5.5 g kg −1 respectively [31].
Despite the potential of MOFs as drug delivery platforms, the impact of many of their fundamental properties on their performance as drug carriers has yet to be fully understood. Further investigation is needed to fully comprehend the role of these properties in the drug delivery process [35,36]. One such property is the relationship between the metal binding site/ligand of the MOF and the loading and release properties of the drug. Recently, Rezaei and co-workers conducted the first study investigating the relationship between a MOF's metal centre and its drug-releasing rate using a series of MOF-74 containing Magnesium, Nickel, Zinc, and Cobalt [37]. They found that the Mg MOF was the most effective carrier for curcumin compared to other MOFs. This highlights that not all MOFs perform equally well as drug carriers and that varying the metal centre or ligand can greatly influence the specific surface area and interactions with the drug. To our knowledge, this is the only publication reported on investigating the effect of the metal centre on curcumin drug delivery properties. Given that understanding this relationship is crucial in developing MOFs as drug delivery platforms, it is important to conduct further research to validate this hypothesis.
In this study, we investigate the curcumin delivery properties of a series of MOFs including HKUST-1 (Cu), Cu(tpa) (Cu) (tpa: terephthalate), MIL-100 (Fe) and MOF-2 (Zn) and compare their drug delivery properties based on the metal complexed active site and the ligand. We use Powder X-Ray Diffraction (PXRD), Thermogravimetric Analysis (TGA) and Fourier-transform Infrared spectroscopy (FTIR) to investigate curcumin loading. Results reveal that HKUST-1 and MIL-100 have high curcumin encapsulation efficiencies (55%-75%) even at low concentrations of 6ppm. We also use a combined computational and experimental approach to determine the binding modes between the metal centres and curcumin. Furthermore, we investigate the drug-releasing capabilities of HKUST-1 and MIL-100, which have shown significantly high loading capacities for curcumin. To the best of our knowledge, this is the first study that compares the curcumin loading capacity of MOFs containing Copper, Zinc, and Iron.

2. Results and discussion
It is reported that curcumin shows a high affinity towards ferrous and copper ions in solution with respect to zinc ions (Fe > Cu ? Zn) [38]. Hence, in order to investigate specific interactions of curcumin with copper, ferrous, and zinc metal centred MOFs, drug delivery properties of HKUST-1 (Cu), Cu(tpa) (Cu), MIL-100 (Fe) and MOF-2 (Zn) were investigated. Both HKUST-1 and MIL-100 are three-dimensional MOFs and share the common ligand 1,3,5-benzene tricarboxylic acid and comparable surface areas (table 1).
However, the Zn polymorph of HKUST-1, synthesized from Benzene tricarboxylic acid is reported to be unstable upon activation [39]. Therefore, MOF-2 and Cu(tpa) MOF with the common ligand 1,4-benzene dicarboxylate, and similar surface areas (table 1) were synthesized in order to compare the effect of Zn and Cu metal centres respectively. MOFs with highly polar, readily soluble polycarboxylate ligands were selected as they can be easily removed from the body under physiological conditions [31].

Loading of curcumin
Curcumin was loaded to all four MOFs using the general procedure discussed in section 4.2. Curcumin degradation in solution is accompanied by a reduction in absorbance maximum [40]. However, throughout our experiments, the absorbance of the control solution of curcumin remained unchanged, indicating that there was no degradation of curcumin under the experimental conditions. For this study, we used both low and high concentrations of curcumin to load all four MOFs.

Loading of curcumin in low concentrations
When all four MOFs were suspended in curcumin solutions, the curcumin content in the supernatant was reduced due to the encapsulation of curcumin to HKUST-1, Cu(tpa) and MIL-100, whereas for MOF-2 it was increased. Visual observation of this phenomenon is shown in figure S5. All supernatant solutions were quantified through UV-vis Spectroscopy and Encapsulation efficiencies (EE) were calculated (table 1). Accordingly, HKUST-1 (Cu) and MIL-100 (Fe) exhibit superior Encapsulation Efficiency and entrapment continues to increase with time. On the other hand, MOF-2 (Zn) shows a negative EE%. This negative EE% for MOF-2 could be due to the sorption of smaller and more abundant ethanol (solvent) molecules, which makes the supernatant solution more concentrated.
The four MOFs used can be categorized to two groups depending on their specific surface areas. While the high specific surface areas of MIL-100 and HKUST-1 favours high EE%, stronger interaction between Fe(III) with respect to Cu(II) could account for the high EE% observed for the former. Among the two MOFs with low specific surface areas, a significantly high EE% of Cu(tpa) with respect to MOF-2 could be due to the higher affinity of curcumin towards Cu(II), with respect to Zn(II). Combining the above observations and theoretical calculations, the affinity order Fe(III)>Cu(II)>Zn(II) can be established in tested MOFs. This order parallels the observed affinities of curcumin towards metal ions in free solution as reported by Baum et al [38] Theoretical observations reveal the ability of HKUST-1 and MIL-100 to form chelating complexes with curcumin, which along with high specific surface areas and highly interacting metal centres account for the observed high EE% values (section 2.2.4).

Loading of curcumin in high concentrations
A clear change in colour upon loading of curcumin was observed for all four MOFs (figure 1). Further, calculated EE% and percentage Loading Capacity (LC%) values in highly concentrated curcumin solutions using spectroscopic data in table 2 show that curcumin is loaded to all four MOFs. However, in keeping with the trend observed under low concentrations, MIL-100 and HKUST-1 show the highest encapsulation.

Thermogravimetric analysis
TGAs of all four MOFs before and after encapsulation of curcumin are given in figure S6. Thermal characteristics of all MOFs and Curcumin matched well with the reported literature [45]. Curcumin decomposition initiates in the range of 200°C to 300°C, and gets completely decomposed by 500°C [45]. Hence, as a mass loss in all four MOFs from 200°C-300°C was observed, which could be used as an indication  for the loading of curcumin. However, the weight loss could not be quantified since it overlaps with the framework degradation.

FT-IR Analysis
A comparison of FT-IR spectra of activated MOF, curcumin-loaded MOF and neat curcumin are given in figure  S7. Since the range below 600 cm −1 does not present features relevant to the analysis, only the range between 4000-600 cm −1 was recorded. A further amplification in the range of 700 cm −1 -1700 cm −1 is shown in figure 2 for the convenience of analysis. Curcumin-loaded HKUST-1 spectrum shows an additional peak at 1041 cm −1 , which aligns with the C-O-C stretching peak of curcumin at 1026 cm −1 . The dual peaks due to the asymmetric stretching of carboxylate of HKUST-1 are at 1646 cm −1 and 1590 cm −1 [46]. The variation of the ratios of the bands at 1646/1590 cm −1 could be due to the changes in the environment of the carboxylate ligands due to interactions with curcumin, possibly through hydrogen bonding (Refer to section 2.2.2) [47].
A small peak at 1272 cm −1 which is due to the enol C-O of curcumin and 1145 cm −1 peak of skeletal CCH, can be seen in the curcumin and curcumin loaded MIL-100 without any overlap of MOF peaks, indicative of curcumin loading to MIL-100.
Owing to the overlap of C=O stretching peak of curcumin, the peak at 1507 cm −1 in curcumin-loaded Cu(tpa) has got more intense compared to the pure MOF. A significant peak of enol COC stretching at 1026 cm −1 in curcumin could also be seen in the loaded Cu(tpa) MOF forming evidence for curcumin loading [48].
However, the prominent peaks of curcumin at 1626 cm −1 , 1601 cm −1 and 1508 cm −1 which are usually taken as evidence for curcumin loading, have all been overlapped by the C=O peaks of the MOF ligand. Hence, FT-IR alone cannot give substantial evidence for curcumin loading. Nevertheless, the collective evidence from UV-vis Spectroscopy, TGA and visual inferences confirms the loading of curcumin to all four MOFs.

Binding modes of curcumin 2.2.1. Stable conformer of curcumin
Theoretical calculations indicated that the free energy of the enol form (shown in figure 3) of curcumin is more stable than the keto form by 30.78 kJmol −1 . Therefore, the enol form was selected for further calculations. This selection is further supported by reported FTIR data showing the absence of the sharp characteristic C=O peaks in the region of 1650-1800 cm −1 [49].
Potential binding sites of enol form were identified by calculating the electron densities of the enol form of curcumin. From our calculations, it was observed that binding at OCH 3 group (figure 3(D)) does not take place, even though the highest occupied molecular orbital (HOMO) of unbound curcumin has a considerable electron density on the oxygen atom of the OCH 3 group. This could be due to the steric hindrance caused by the methyl group, which is located close to the MOF. Instead, it was observed that the binding could occur at the nearby phenolic OH group (figures 3(A), (B)), which has less steric hindrance. Binding through the phenyl ring ( figure 3(E)) to the coordinative unsaturated site (CUS) can be neglected due to the high steric hindrance in the complex formed. Therefore, only the OH and enol (figure 3(C)) binding sites of curcumin were considered in the analysis.

Calculation of binding energies
Binding energies, calculated for phenolic and enol binding onto HKUST-1 and MIL-100 metallic sites, according to equation (3) in section 4.5.4, are shown in table 3. In HKUST-1, variation of binding energy by a factor of almost 10 between the two phenolic OH groups (A and B) and the enol OH group (C) serves as strong evidence for the binding of curcumin through the phenolic OH groups. The calculated binding energies of phenolic and enol binding modes of curcumin to MIL-100 suggest both phenolic OH and enol OH binding can occur in MIL-100. The larger negative binding energies (with respect to HKUST-1) complement the higher EE% observed for MIL-100 compared to HKUST-1.
Representative geometry optimized curcumin-HKUST-1 and curcumin-MIL-100 complexes are shown in figure 4. It was observed that in all binding modes, in addition to the interaction between the oxygen atoms of OH group with metal ions, the hydrogen atoms on the OH group of curcumin can form hydrogen bonds with the oxygen atoms on the carboxylate group of MOFs which lie at close proximity to further stabilize the complex. The bond distances for metal ion-oxygen interactions and possible hydrogen bond type interactions are indicated in table 3. Hydrogen bond distances of all binding modes are lesser than 2.5 Å, and this is the accepted upper limit of bond distance for strong covalent-type hydrogen bonds [50]. This hydrogen bond formation is further supported by FT-IR evidence (section 2.2.3).
The calculation of binding energy for MOF-2 resulted in the observation of drastic geometric changes in the MOF representative cluster during the optimization process of the complex. This is discussed in section 2.3.2.

FT-IR analysis
A computational study of the binding mechanism of curcumin indicates that the binding occurs through the -OH group (section 2.2.2). FTIR bands at 3512 cm −1 (figures 5 and S7), which corresponds to O-H vibration of curcumin [51,52] is broadened in the spectra of curcumin-loaded HKUST-1 and MIL-100 owing to the H-bonds formed between curcumin and the carboxylate groups of the MOF. This could serve as further evidence for the specific binding mode.  As Cu(tpa) MOF is synthesized in the absence of water, broad band due to OH group are not observed in the 2500-3500 cm −1 range (figure S7). However, after curcumin loading, a significant broad band at 3100 cm −1 is detected in the MOF ( figure 5). This could be due to hydrogen bonds formed between the -OH of curcumin and the carboxylate group of MOF. However, the band at 3508 cm −1 corresponding to the stretching of OH in curcumin can no longer be seen in the curcumin-loaded MOF-2 ( figure S7). This could be due to the broadening of the peak upon H-bonding with the MOF. It could also be counter-argued that there is no evidence of complex formation in MOF-2 as the amount of curcumin loaded into the MOF-2 is substantially low (LC = 0.21%; Refer to section 2.1.2).

Chelating
Calculated binding energies and spectroscopic studies provide a strong indication of binding through OH groups in curcumin. The case for OH binding is further strengthened by the high EE% of curcumin for HKUST-1 and MIL-100, which are the only two MOFs (out of the four studied) which has the possibility to form chelating complexes through OH.
Calculated interatomic distances between binding sites of curcumin are shown in table 4. According to established structural data, both HKUST-1 and MIL-100 contain metal atoms (of different active sites) separated by around 18 Å [53,54]. This makes it possible for them to form chelating-type bonds with curcumin via the A and B binding sites. A possible chelating configuration between two ferric ions in MIL-100 separated by 17.1 Å is  shown in figure 6. However, as the highest interatomic distance between two metal ions in Cu(tpa) and MOF-2 are 13.0 Å and 16.4 Å respectively [44,55], curcumin will have to distort from its stable linear structure to make bidentate bonds which would be energetically unfavourable. In MIL-100 (where binding through enol OH is feasible), chelating through AC and BC modes can take place due to the availability of ferric ion sites separated by 10.7 Å and 11.7 Å.

Structural changes 2.3.1. HKUST-1 and MIL-100
PXRDs observed for curcumin-loaded HKUST-1 and MIL-100 remained similar to the PXRDs of the original MOFs prior to drug loading. Hence, it can be inferred that the crystal structure of MIL-100 and HKUST-1 has remained intact after the loading of curcumin, depicting their high stability (figure 7).

MOF-2 and Cu(tpa)
According to the PXRD patterns in figure 7, MOF-2 and Cu(tpa) remain to be crystalline after loading of curcumin. However, the crystalline structure of MOF-2 has changed from a monoclinic to a triclinic structure upon loading of curcumin. This type of structural change has been observed due to strong hydrogen bonds that are capable of forming between ethanol and Zn 1,4-benzenebicarboxilic MOFs [56]. Hence, this structural change could be either due to hydrogen bonds formed by ethanol molecules in the solvent or by alcoholic groups present in curcumin. This observation agrees with the computational analysis of MOF-2 binding. During the geometric optimization of curcumin-MOF-2 complex, the stable paddle wheel structure was heavily deformed by the removal of one benzoate group irrespective of the binding mode of curcumin (figure S10).
A similar phenomenon was observed in the structure of Cu(tpa). The PXRD structural change detected upon curcumin loading was similar to the change in crystal structure reported by Carlsen et al due to external molecules forming −OH bonding with the Cu(tpa) structure [48]. Curcumin has a high affinity to copper and experimental and theoretical evidence suggests curcumin binding through the OH group. Hence, the change in structure could be attributed to the binding of curcumin.

Preliminary curcumin releasing studies of HKUST-1 and MIL-100
Curcumin-releasing studies of HKUST-1 and MIL-100 were conducted following the procedure outlined in section 4.4. The two MOFs, HKUST-1 and MIL-100 which have shown the highest loading of curcumin have been further investigated for releasing studies. Both HKUST-1 and MIL-100 have shown prolonged release exceeding 2 days. This could be due to strong interactions of the metal with curcumin. 18.04% of the encapsulated curcumin was released by HKUST-1 while 15.71% was released by MIL-100 within 3095 min  (51.5 h) (Section S7). Figure 8 depicts the logarithmic trend of curcumin released from MIL-100 and HKUST-1.
The prolonged-release property demonstrated by HKUST-1 and MIL-100 is important in maximizing the therapeutic benefits of curcumin, given its poor aqueous solubility and low gastrointestinal absorption [10,11]. By releasing the drug slowly over an extended period of time, these MOFs can ensure a consistent and sustained delivery of the active ingredient, thereby increasing its bioavailability and effectiveness in the body.

3. Conclusion
The use of Metal-Organic Frameworks (MOFs) as drug delivery platforms for curcumin has grown significantly in recent years. In order to fully harness the potential of MOFs as drug carriers for curcumin, it is crucial to understand the structural properties of MOFs and their drug loading and releasing properties. However, such structure-function relationships are poorly investigated. In this study, curcumin was loaded to four different MOFs with copper, ferrous, and zinc metal centres. Theoretically and experimentally observed affinity of curcumin for the synthesized MOFs depended upon 01) the reported trend of the affinity towards constituent metal ions in solution (Fe 3+ > Cu 2+ ? Zn 2+ ), 02) the specific surface area and 03) the ability of the MOF to chelate. HKUST-1 and MIL-100 have shown high sensitivity for curcumin in even extremely dilute solutions with EE%'s ranging from 58%-76% with significant loading capacities of 3.38% for HKUST-1 and 5.36% for MIL-100. Strong evidence for the binding of curcumin to HKUST-1 and MIL-100 through OH group has been discovered through a combined spectroscopic and computational analysis, which is also facilitated by a  hydrogen bond formation. A chelating binding mode is suggested to explain the high EE% values observed for HKUST-1 and MIL-100 in dilute solutions. Preliminary releasing studies of HKUST-1 and MIL-100 have suggested controlled prolonged releasing for more than two days suggesting them as potential carriers for the slow delivery of curcumin for an extended time. Overall, the findings of this study provide a valuable contribution to the understanding of the factors that impact MOFs' performance as drug delivery platforms for curcumin. We believe that these insights can be applied to a wider range of MOFs and will pave the way for the development of more efficient and effective MOF-based drug delivery systems for curcumin.

Loading of Curcumin to MOFs
Four MOFs were added to 10.00 ml of a 5.9 ppm curcumin solution at room temperature with solution levels marked and stirred continuously. A volume of 2.00 ml was withdrawn from the supernatant and centrifuged to take UV measurements to quantify the amount of curcumin remaining in the solution and the removed solution was replaced with fresh solvent. A sample of curcumin was maintained as the control to avoid errors due to the evaporation of solvent and degradation of curcumin. The loaded amount of curcumin was quantified as described in section S6. EE and LC were calculated according to the following equations.
oading of curcumin for all four MOFs was studied in a highly concentrated curcumin solution of 2845 ppm.

Characterization of curcumin Loaded MOFs
Curcumin loaded MOFs were characterized through FTIR spectroscopy (ABB MB3000 spectrometer) with the Attenuated Total Reflectance (ATR) facility with a resolution of 16 cm −1 and 16 scans. PXRD (Brucker D8 FOCUS, Cu Kα −1 radiation) was performed with a step size of 0.02°and a step time of 1 s to identify and confirm the crystalline phase. TGA (TA Instruments -SDTQ600) was performed under a nitrogen atmosphere at a rate of 5°C min −1 to confirm the loading of curcumin.

Releasing studies of curcumin
As curcumin does not dissolve in an aqueous medium, the minimum amount of ethanol that should be added to dissolve curcumin was experimentally detected as 3:7; ethanol: Phosphate Buffer Solution (PBS) ratio and was used for experimental studies.
Curcumin-loaded HKUST-1 (Cu) and MIL-100 (Fe) in 2845 ppm solution were dried in a desiccator and suspended in 10 ml of ethanol: PBS mixture for releasing studies. This mixture was stirred at the rate of 20 rpm to avoid diffusion error due to the concentration gradient. Aliquots of 1.00 ml of the supernatant were recovered by centrifugation (80000 rpm for 6 min) at different time intervals and the released curcumin was quantified by UV visible spectroscopy. The withdrawn aliquot was replaced with the same solution of fresh Ethanol/PBS solution.

Computational study of the binding of Curcumin to active sites of MOFs
The use of classical molecular dynamics simulations to model adsorbate interactions with MOFs having coordinately unsaturated sites (CUS) has been found to be inadequate and the application of quantum chemical methods has been commonly used as an alternative [58]. Two commonly used quantum chemical strategies have been to either use a representative cluster of the MOF enclosing the CUS or to use a periodic cell approach, where a suitable unit cell is repeated in the three-dimensional space [58]. In this work, we have used the representative cluster model as the unit cells of the MOFs under inspection are complex and the use of a representative cluster would allow the inspection of the binding modes and energies under a more accurate method/basis set.

Representative cluster
Since the purpose of this study is to identify the binding mechanism of curcumin, which comparatively is a large adsorbate, a larger cluster with four paddle wheel units of benzoic acid with hydrogen end capping was used to represent the copper and zinc MOFs (figure S9) while a cluster with six acetate ligands binding to three ferric ions in the CUS along with the central oxide ion was utilized to model MIL-100. To achieve charge neutrality, a fluoride counter ion was attached to one of the ferric ions (figure S9).

Method and basis set
Quantum mechanical studies of large adsorbate molecules on CUS of MOFs are rare and curcumin adsorption on CUS of copper, ferric or zinc MOFs has not been studied previously. In this study, binding modes of curcumin to HKUST-1, MIL-100 and MOF-2 are investigated with density functional theory (DFT) [59] using B3LYP/6-31G(D) method/basis set combination [60][61][62][63][64]. Gaussian 09 software package [65] is used for calculations while VMD [66] is used for visualization purposes in the text. All calculations are performed in an ethanol medium (using the polarizable continuum model) and at 298.15 K. Electron densities are calculated using the population analysis command in Gaussian 09.  [67][68][69]. Binding Energies (BE) calculated for several large heterocyclic molecules show that binding is feasible for M and L sites [70]. Among the two sites, L site binding is neglected as it will be sterically inaccessible in the crystal structure. The same rationale is utilized in the present work and only the M site binding is considered. Since the representative cluster of both HKUST-1 and MOF-2 are identical, except for the metal atom, the same binding sites are studied for MOF-2.
The two highly exposed ferric ion sites in the CUS make them the obvious binding site in MIL-100 and this has been found to be true for the binding of heterocyclic compounds onto the MOF [71].

Calculation of binding energies
For the identified binding sites, BE is calculated by the energy difference between the reactants and products.

MOF Drug
Complex is the total energy of the complex formed between MOF and curcumin (denoted as Drug), while E(MOF) and E(Drug) denote the energy of MOF and the drug respectively.