**2. Building bio-inspired MOFs**

#### **2.1. Design**

Consequently, the synthesis of new crystal forms evolved tremendously in the last decade, and the interest of pharmaceutical companies in the appearance/disappearance of new solid forms of APIs has vastly increased. Polymorphs, hydrates and salts of drugs are long-known forms with recognized impact in their properties. Cocrystals represent a more recent class of crystal forms that own particular scientific and regulatory advantages (FDA guidance is already available and cocrystals are now being commercialized as drugs in some countries). Many examples show their relevance in the pharmaceutical industry, most of them by

Likewise, nanoporous materials recently became of pertinent use in the medicinal and pharmacological fields for drug storage, delivery and controlled release in addition to applications in imaging and sensing for therapeutic and diagnostic [21–34]. Particularly, metal organic frameworks (MOFs) have generated large interest owing to their versatile architectures [35] and their promising applications not only in ion exchange, adsorption and gas storage [36– 41], separation processes [42], heterogeneous catalysis [43, 44], polymerization reactions [45, 46], luminescence [47], non-linear optics [48] and magnetism [49], but also as drug carriers, systems for drug delivery [22, 23, 50, 51], contrast agents for magnetic resonance imaging (MRI)

Up to now, drug delivery from porous solids has been achieved by encapsulation in mesopo‐ rous silicas or zeolites, methods that are strongly dependent on the pore size and on the hostguest interactions. Both hypotheses suffer from important drawbacks: low drug-storage capacity, too rapid delivery and solid degradation that brings toxicity concerns [23, 25, 26, 28, 29, 52]. Extended metal-ligand networks with metal nodes and bridging organic ligands such as coordination networks, porous coordination networks (PCNs), porous coordination polymers (PCPs) and MOFs have attracted great attention in the last years [24, 25, 28, 53, 54]. Particularly, MOFs with biological-friendly composition emerged as new drug carriers capable

In fact, MOFs are among the most exciting architectures in nanotechnology and are defined as hybrid self-assemblies of metal ions or metal clusters (coordination centres) and organic fragments (linkers). They exhibit some of the highest porosities known, turning them into ideal materials for capture, storage and/or delivery applications [21, 24–26, 29, 54, 57]. Compared to other nanocarriers, MOFs are candidates to extensive applications since they combine high pore volume with a regular porosity, and the presence of tuneable organic groups allows an

The first families of MOFs considered as potential drug delivery systems were the coordination polymers from Oslo (CPO), such as CPO-27(Mg) [58] built up from magnesium coordination polymers, and the materials of Institute Lavoisier (MIL) [22]. Horcajada et al. [22, 23] prepared MIL-100 (with trimesic acid) and MIL-101 (with terephthalic acid) applied for the delivery of ibuprofen in the gastrointestinal tract, exhibiting high drug-storage capacity and a complete drug-controlled release under physiological conditions [22, 23]. Less toxic systems, using iron and more flexible MILs, are under study [25], and the first biodegradable therapeutic MOF, BioMIL-1, was reported by Miller et al. in 2010 [27]. The large breathing effect that MOFs can

enhancing stability, solubility and/or bioavailability of known drugs [7–20].

1362 Metal-Organic Frameworks

[21] and systems with potential use in other biomedical applications [23].

easy modulation of the framework as well as of the pore size [22, 24–26].

of tackling these problems [21, 23, 25, 26, 28, 29, 55, 56].

The use of porous solids for biomedical applications requires a biological friendly composition, making compulsory the use of metals and linkers with acceptable toxicity [28].

When designing BioMOFs, the decision to exclude one linker and/or metal depends on several parameters: application, balance between risk and benefit, degradation kinetics, biodistribu‐ tion, accumulation in tissues and organs as well as body excretion [21, 23, 25, 26, 28, 55, 56]. Both exogenous (not intervening in the body cycles) and endogenous (constitutive part of body composition) linkers have been used in MOF synthesis for drug delivery, with the first group having a higher prevalence [21, 25–29, 57]. It is also worth noting that if the therapeutic molecule is directly used as a linker, no large pores are required and the release of the drug molecule is achieved directly through the degradation of the solid, without any side effects arising from the release of a non-active ligand [26, 52].

Different methods have been explored to design BioMOFs, including ZMOFs, from which we highlight the molecular building block (MBB), supermolecular building block (SBB) and supermolecular building layer (SBL) approaches. Also, a brief allusion to the influence that computational simulations may have in building and studying BioMOFs is made.

#### *2.1.1. Molecular building block (MBB), supermolecular building block (SBB) and supermolecular building layer (SBL) approaches*

To construct a MOF, it is necessary to make a pre-selection of building blocks that would give the desired structural and geometrical information for a given underlying network—molec‐ ular building block approach (MBB) [72]. The prerequisites for the successful implementation of this approach are (a) selection of an ideal blueprint net exclusive for the assembly of its corresponding basic building units and (b) isolation of the reaction conditions that allow the formation of the desired MOF. Simple MBBs based on simple organic ligands or polynuclear clusters are often limited in terms of connectivity [72]. To overcome this issue, two conceptual approaches were recently implemented to facilitate the design and deliberate construction of MOFs: supermolecular building block (SBB) and supermolecular building layer (SBL). These approaches allow the rational design of made-to-order MOFs [73].

The SBB approach consists of using metal-organic polyhedral (MOPs) as SBBs in building an MOF, presenting great potential to control the targeted framework. To obtain the desired topology, the MOP must have the correct geometrical information and peripheral points of extension (connectivity). The prerequisites for this approach are (a) a blueprint net with minimal edge transitivity, preferably singular, exclusive for the assembly of given building units, and not susceptible to self-interpenetration upon net expansion and/or decoration and (b) reaction conditions that allow the formation of the SBB in situ.

The SBL is based on the use of 2-periodic MOF layers (SBLs) as building blocks for the desired functional 3-periodic porous MOFs. This implies the chemical cross-linking of layers via accessible bridging sites on the layers, such as open metal sites or functionalized positions on the organic linker, whose judicious selection is mandatory. This approach, in principle, allows to predict MOFs with tuneable cavities, the endless expansion of confined space (as cavities and pores), and its modularity further permits an easy functionalization and introduction of additional functionalities [74] to aim specific applications. The prerequisites for this approach are (a) a blueprint net with minimal edge transitivity, rather singular, exclusive for the particular pillaring of the given building units and (b) the reaction conditions to allow the consistent formation of the SBL in situ.

#### *2.1.2. Screening using simulations*

Systematic studies relating MOF structures with their performance in drug delivery is crucial for the identification of promising structures. Molecular simulations are a mean that can be explored to seek for the optimal structure for a given application. The grand canonical Monte Carlo (GCMC) simulation is the preferred method for simulating adsorption in porous materials and for explaining and predicting new results. However, the simulation in the case of large guest molecules is difficult and that justifies the limited number of studies on drugporous solid systems [75].

Fatouros et al. reported the use of molecular dynamics to study the diffusion properties of salbutamol and theophylline in the zeolite BEA, an indication that this method can be used for screening purposes on zeolite-drug systems [76].

Regarding MOFs, very few computational studies are reported and those are focused on one or more structures simultaneously, limiting the possibility of correlating drug delivery performance with structural features. A combined experimental and computational study of three MOFs for the drug delivery of 5-fluorouracil was recently presented, in which GCMC simulations were used to investigate the interactions between the drug and the porous cage [77]. Density functional theory (DFT) calculations have been applied to identify the most favourable conformations and adsorption sites of ibuprofen and busulfan on MIL-53(Fe) [78]. Quantitative structure-activity relationship (QSAR) models were used to rationalize the experimental uptake of caffeine as model in a series of MIL-88B(Fe) materials with different functional moieties [79]. The energetics and dynamics of ibuprofen in MIL-101 were also studied recurring to simulated annealing followed by DFT of one single ibuprofen molecule to study the preferential adsorption sites [56].

Also worth mentioning an extensive study on GCMC simulations to screen a series of biocompatible MOFs as carriers of ibuprofen has been reported. Simulations include micropo‐ rous, mesoporous and nanoporous MOFs and have shown to be a successful pathway to predict the drug adsorption properties of porous adsorbents. Furthermore, this work proposes new tools that allow the study of new porous materials as potential drug carriers prior to experiment [75].

### **2.2. Synthesis**

ular building block approach (MBB) [72]. The prerequisites for the successful implementation of this approach are (a) selection of an ideal blueprint net exclusive for the assembly of its corresponding basic building units and (b) isolation of the reaction conditions that allow the formation of the desired MOF. Simple MBBs based on simple organic ligands or polynuclear clusters are often limited in terms of connectivity [72]. To overcome this issue, two conceptual approaches were recently implemented to facilitate the design and deliberate construction of MOFs: supermolecular building block (SBB) and supermolecular building layer (SBL). These

The SBB approach consists of using metal-organic polyhedral (MOPs) as SBBs in building an MOF, presenting great potential to control the targeted framework. To obtain the desired topology, the MOP must have the correct geometrical information and peripheral points of extension (connectivity). The prerequisites for this approach are (a) a blueprint net with minimal edge transitivity, preferably singular, exclusive for the assembly of given building units, and not susceptible to self-interpenetration upon net expansion and/or decoration and

The SBL is based on the use of 2-periodic MOF layers (SBLs) as building blocks for the desired functional 3-periodic porous MOFs. This implies the chemical cross-linking of layers via accessible bridging sites on the layers, such as open metal sites or functionalized positions on the organic linker, whose judicious selection is mandatory. This approach, in principle, allows to predict MOFs with tuneable cavities, the endless expansion of confined space (as cavities and pores), and its modularity further permits an easy functionalization and introduction of additional functionalities [74] to aim specific applications. The prerequisites for this approach are (a) a blueprint net with minimal edge transitivity, rather singular, exclusive for the particular pillaring of the given building units and (b) the reaction conditions to allow the

Systematic studies relating MOF structures with their performance in drug delivery is crucial for the identification of promising structures. Molecular simulations are a mean that can be explored to seek for the optimal structure for a given application. The grand canonical Monte Carlo (GCMC) simulation is the preferred method for simulating adsorption in porous materials and for explaining and predicting new results. However, the simulation in the case of large guest molecules is difficult and that justifies the limited number of studies on drug-

Fatouros et al. reported the use of molecular dynamics to study the diffusion properties of salbutamol and theophylline in the zeolite BEA, an indication that this method can be used for

Regarding MOFs, very few computational studies are reported and those are focused on one or more structures simultaneously, limiting the possibility of correlating drug delivery performance with structural features. A combined experimental and computational study of three MOFs for the drug delivery of 5-fluorouracil was recently presented, in which GCMC

approaches allow the rational design of made-to-order MOFs [73].

(b) reaction conditions that allow the formation of the SBB in situ.

consistent formation of the SBL in situ.

screening purposes on zeolite-drug systems [76].

*2.1.2. Screening using simulations*

1384 Metal-Organic Frameworks

porous solid systems [75].

MOFs are still widely synthesized using solvo/hydrothermal techniques, the most common methods to obtain coordination networks [21, 25, 28, 29]. Nevertheless microemulsion synthesis [80] is also a typical method and interesting alternatives are being used based on environmental-friendly synthetic routes: ionothermal [81], microwave, ultrasound-assisted, and sonochemical synthesis [21, 25, 28], as well as mechanochemistry [82, 83]. The synthesis of this type of compounds has been reviewed several times [63, 84, 85] and therefore only brief details on each technique are presented herein.

The solvo/hydrothermal synthesis involves polar solvents under moderate to high pressures and temperatures. This method often requires toxic solvents such as DMF, and its use is limited by safety and time-consuming reasons. Alternative techniques allow higher efficiencies, have lower energy costs and have less impact in the environment [86].

Microemulsion synthesis is based on thermodynamically stable dispersions of two immiscible liquids in the presence of an emulsifier or surfactant (i.e., microemulsions). This technique confines the synthesis of MOFs to the nanoscale and offers the possibility of tuning the size. The disadvantages of the microemulsion approach include poor yields, reproducibility issues, usage of highly toxic surfactants and solvents that strongly limit biomedical applications and the possible decrease of the sorption capacity due to the combination of surfactants with highly porous structures [80, 86].

Ionothermal synthesis requires the use of green solvents such as ionic liquids and eutectic mixtures (a special type of ionic liquid) to obtain MOFs and it can be performed in open air. These solvents act both as solvents and templates to avoid the competition interactions between the solvent framework and the template framework that are present in the solvo‐ thermal methods [63, 81].

Microwave and ultrasound-assisted syntheses usually lead to the fast crystallization of MOFs and are considered green methods. In the case of microwaves, the heating involved in the process favours a rapid and uniform nucleation process, which results into a more homoge‐ neous particle size distribution. Regarding ultrasounds, it has shown to be a highly efficient method [86].

Sonochemical synthesis or sonocrystallization method not only promotes the nucleation process but also stimulates the homogeneity of the nucleation, what represents an advantage over the traditional solvothermal methods. This approach is prone for industrial applications due to its easy scale-up [63].

Mechanochemistry is a green, solvent-free and efficient strategy to build MOFs. It is based on the direct grinding of the linkers and the metal salts either in a mortar or in a ball mill, without recurring to solvent (neat grinding, NG) or recurring only to catalytic amounts of solvent to activate the process (liquid-assisted grinding, LAG). Alternatively, also catalytic amounts of ionic salts can be used to trigger the process (ion- and liquid-assisted grinding, ILAG). This is a simple method and the absence of solvent makes it very appealing to biomedical applications [63, 82, 83, 86].

#### **2.3. Loading of drugs and other biomedically relevant compounds into MOFs**

The loading of relevant molecules, such as imaging and therapeutic agents, into MOFs can be done directly during the MOF synthesis or in the postsynthesis.

The direct incorporation implies using those molecules directly to assemble the framework. This strategy also encloses the networks in which paramagnetic metal ions, such as Gd3+, Fe3+ and Mn2+, do not act only as the metal sites to connect the ligand but act also as magnetic resonance imaging contrast agents. High loadings of the relevant compounds can be achieved by this strategy; however, it is necessary to tune the morphology and physicochemical properties of these MOFs for each case and it is important to guarantee that there is no degradation of the compound during the synthesis [21].

The postsynthesis strategy requires high porosity and the active compound is incorporated within the MOF by noncovalent or covalent interactions. In the case of noncovalent loading, the process is reversible and therefore the drug release can be premature. On the other hand, the covalent loading creates a prodrug in which the drug release happens at the same time as the MOF degradation and thus it may be considered a more robust approach [87].

#### **2.4. Surface modifications**

The improved biomedical properties of MOFs also depend on the rational design of the surface. However, the task of changing the outer surface of the MOF without changing its character‐ istics is still very difficult. Ideally, MOFs should have a coating shell to confer stability to the material under the different physiological media, but it must be non-toxic and must not interfere with the pores [86]. There are two approaches to achieve the surface modifications: covalent and noncovalent attachments. The choice of the best method relies on the parameters and nature of the MOF, as well as on the nature of the molecule to be grafted [88]. To date only a few successful examples have been reported of which we highlight the following three.

A simple, fast and biofriendly method was reported for the use of heparin for the external functionalization of MIL-100(Fe), preserving all the properties of the MOF. The coating obtained by this method led to improved biological properties, such as reduced cell recogni‐ tion, lack of complement activation and reactive oxygen species production [89].

The coating of MIL-101(Fe) with a thin film of silica resulted in the prevention of the rapid degradation of the MOF [87].

Another example of successful coating of MOFs concerns the use of phosphate-modified biocompatible cyclodextrins. This method was applied to MIL-100(Fe) and resulted in improved stability in body fluids without interfering with the MOFs properties [90].
