Progress of albumin-polymer conjugates as efficient drug carriers

Introduction

Serum albumin (SA), possessing a molecular weight of 66.5 kDa, constitutes 50% of total plasma protein content and is the most abundant plasma protein found in human blood [1]. The total concentration of this protein in human serum approximates to about 35–50 g/L [2, 3] and is replenished by the liver hepatocytes, producing 10–15 g of albumin per day [4]. SA plays a key role in performing many vital functions such as regulating metabolic and vascular systems, maintaining oncotic pressure and microvascular integrity, transporting a wide variety of ligands, and modulating neutrophil adhesion and cell signaling moieties to reduce inflammation [5]. SA is regarded as a potential biomarker in certain diseases, especially in cancer where serum albumin levels have been implemented as prognostic indicators for cancer survival [5]. Albumin has significant potential for anti-cancer drug delivery due to its overall enhanced uptake in tumours for cancer cell metabolism. Several factors contribute to the selective cancer-targeting ability of serum albumin, including the preferential accumulation of albumin in tumour tissue over normal healthy tissue, otherwise known as the enhanced permeation and retention (EPR) effect [6, 7]. The benefits that arise from this phenomenon rely on the structural differences between diseased and healthy tissue. Furthermore, the binding of albumin to certain receptors is an additional advantage to using this protein for anticancer therapeutics. The receptor gp60, also known as albondin, exists on the endothelium of blood vessels and contributes to about 50 % of albumin transport [8]. Typically, gp60 increases membrane permeability for the improved receptor-mediated uptake of proteins in circulation [8], [9], [10]. Selective binding of gp60 with native albumin triggers the formation of a caveolae vesicle, resulting in the release of the protein from the vesicle into the interstitium. This process further aids in the transcytosis of albumin while enabling its escape from lysosomal degradation [10]. The gp60 binding, in conjunction with the EPR effect, enhances the preferential accumulation of albumin and albumin-bound nanoparticles or drugs in tumour tissue. Osteonectin, also known as secreted protein acidic and rich in cysteine (SPARC) receptor, is also believed to have a significant effect on albumin accumulation [11, 12]. Recent in vitro and in vivo studies have demonstrated the uptake of albumin into cells directly by the influence of the SPARC receptor [13], [14], [15]. SPARC has a high binding affinity for serum albumin [16] and is overexpressed in tumour tissue [12].

The impressive innate affinity of albumin for tumour tissues has led to the development of many anti-cancer agents and serum albumin formulations, the greatest breakthrough being the FDA approved Abraxane^® nanoparticle, used for the treatment of breast cancer [17]. The Abraxane nanoparticle is comprised of paclitaxel non-covalently bound to serum albumin that is released upon entry into tumour tissues [18, 19]. Albumin, owing to its long systemic circulation time, is a remarkable delivery vector for anti-cancer drugs as it can prolong the circulation half-life of these drugs. In this regard, various prodrugs and therapeutic peptides have been coupled via the cysteine-34 (Cys-34) position on albumin to help promote their accumulation within tumours.

Albumin has been widely used to cargo hydrophobic drugs as the hydrophobic pockets of albumin promote binding to drugs rich in nonpolar moieties [20], [21], [22], [23]. However, a limitation in the loading of hydrophobic drugs in albumin-based nano-carriers is the inevitable aggregation that may lead to destabilization of the nanocarriers. Efficient drug loading and delivery could be achieved by covalent attachment of the molecules to albumin. For example, Yu et al. [24] grafted DNA strands onto the surface of bovine serum albumin (BSA) by click chemistry and DNA hybridization reactions. The DNA strands were able to preferentially intercalate with doxorubicin via non-covalent interactions, resulting in exceptional encapsulation efficiency. The DNA conjugated albumin nanocarriers also exhibited targeting ability through cancer cell-targeted guanine-quadruplex-structured AS1411 aptamer sequence of the hybridized oligonucleotides. Many studies investigating the benefits of conjugating macromolecules to albumin are precedent to this recent work, especially using synthetic polymers [25], [26], [27]. The ground-breaking work of Abuchowski et al. [28] on BSA-PEG conjugates in 1977, exhibiting superior immunogenic properties relative to the free albumin, could be regarded as the impetus that paved the exploratory path to divulge into the potential of integrating polymers in albumin-based drug delivery systems. Increased drug accumulation, enhanced circulation half-life and reduced immunogenicity are the benefits achieved by the albumin-polymer hybrids. Incorporating synthetic polymers with albumin is a refined strategy to accommodate the delivery of a range of drug molecules, including hydrophilic drugs. The highlights of albumin-polymer hybrids are their facile preparation with excellent production yield and the tunability of polymer properties to improve drug loading ability. There are three different strategies that are commonly employed to design albumin-polymer hybrids as detailed in our previous review (Fig. 1) [29].

Fig. 1:

Different approaches to prepare albumin-polymer hybrid nanoparticles. Albumin-polymer nanoparticles derived from polymer-protein conjugates, polymer nanoparticles coated with albumin corona and a polymer shell coated around albumin (native/denatured) or albumin particles.

Albumin is widely used for the design of drug carriers, however, the role of albumin is not necessarily to exert any biological activity, but to serve as a biocompatible building block. The literature is full of examples where albumin was mixed with oppositely charged polymers to generate polyelectrolyte complexes [30]. In the absence of any stabilizing groups, the resulting nanoparticles are often well above 100 nm in size [31]. Approaches to control the BSA nanoparticle size include the use of emulsions [32], but also polymer-albumin conjugates, as depicted in Fig. 1c, can help with stabilization, reducing the nanoparticle’s overall size [33]. Alternatively, drug carriers based on albumin and polymers can also be created by unfolding (denaturing) the albumin and exploiting the large number of liberated functionalities along the albumin backbone as anchor points for polymers and drugs [34]. Recent studies have adopted the strategy of utilizing albumin to coat poly (n-butyl cyanoacrylate) (PBCA) nanoparticles [35]. In this case, albumin achieved two purposes, first as a biocompatible building block to enhance the solubility of the PBCA nanoparticles and secondly to provide functionality for the conjugation of hyaluronic acid, which targets the CD44 receptor. In all these cases, the treatment or the location of albumin in the final drug carrier renders the albumin biologically inactive. Only the approaches depicted in Fig. 1a and b result in nanoparticles where the inherent bioactivity of albumin can be fully utilized.

While coating of polymer nanoparticles with albumin is a facile approach to derive the benefits of albumin, it has been noted that the method of albumin coating has a critical effect on albumin binding efficiency, the conformation of surface-bound albumin, and subsequently the in vivo performance of albumin-coated NPs [36]. The type of nanoparticle surface directly influences the direction of protein absorption, which will determine the fate of the nanoparticle in the body [36]. Hyun et al. [37] adopted three different surface modification strategies; physisorption, interfacial embedding and dopamine polymerization, to modify the surface of poly(lactic-co-glycolic acid) nanoparticles (PLGA NPS) with albumin. The dopamine polymerization method was found to be instrumental in preserving the integrity of albumin and the formed surface layer facilitated the transport of NPs into tumours via interaction with SPARC receptors. In contrast, the interfacial embedding method afforded denatured albumin which posed as a substrate for scavenger receptor A, resulting in the uptake by macrophages, presenting negligible benefit to interactions with cancer cells. This study stresses the importance of controlling the way the surface of nanoparticles is bound with albumin to derive favourable therapeutic outcomes. One way of influencing albumin adsorption to the surface is by altering albumin with hydrophobic groups such as ursodeoxycholic acid [38]. This approach enhanced the amphiphilicity of albumin and facilitated the adsorption to PBCA nanoparticles, however, the direction of albumin adsorption may not be influenced.

A method of creating nanoparticles with albumins coating with a directionally well-defined surface is by attaching a polymer chain in a controlled manner. Site-specific protein modification is a well-established technique [39, 40] and has also been used to attach a single chain to albumin [41]. An extensively probed area is the synthesis of covalently attached protein-polymer conjugates by either the “grafting to” or “grafting from” methods. The “grafting to” approach is a common and facile method involving the conjugation of the polymer to albumin [42] whereas the “grafting from” approach requires in situ polymerization from initiation sites attached to the albumin [43]. Reversible addition-fragmentation chain transfer (RAFT) polymerization is an elegant technique that employs the grafting to approach to synthesise albumin-polymer conjugates. Here, the reactive groups on albumin are utilized to react with a functional chain transfer agent (CTA) as it provides wide range of functionalities for in polymer-albumin construction. RAFT agents are capable of presenting moieties like maleimide [42], pentafluorophenyl ester [44], mercaptothiozoline [45], and pyridyl disulfide [44] that can conjugate with reactive groups in albumin. Atom transfer radical polymerization (ATRP) employs the grafting from method and has been used to host various functional groups such as the maleimide moiety on polystyrene [46] and the pyridyl disulfide group on poly(N-isopropylacrylamide) [PNIPAAm] [47] and hydroxyethyl methacrylate (HEMA) [47] to facilitate the conjugation with albumin. An overview on the biocompatible and functionalized polymers employed in the design of albumin-polymer conjugates has been outlined previously [29].

In this mini review, we focus on drug delivery strategies based on polymer-protein conjugates. While some approaches with several polymer chains attached to one albumin are included, we would like to highlight techniques that lead to polymer-protein conjugates with only one polymer attached. Site-specific polymer conjugations of albumin have the advantage of the lack of disruption to the tertiary structure of the protein, ensuring that the biological activity of albumin is maintained. The properties of the polymer-protein conjugates are not only determined by the number of attached polymers, but also by the nature of the polymer itself. Water-soluble polymers will create a protein polymer-conjugate that is fully soluble in water while hydrophobic chains will result in amphiphilic structures, thus forming nanoparticles. This transition from hydrophilic to hydrophobic needs to be considered when hydrophobic drugs are attached to the hydrophilic polymer chain. At a critical point, the solubility will be dominated by the drug molecules, which will lead to the formation of amphiphilic structures. Thus, the review aims to discuss albumin-polymer conjugates created using covalent bonds whilst also providing an overview of supramolecular albumin-polymer conjugates in addition to the less explored molecular interactions of albumin-polymer conjugates. Many of these publications have already been covered in our earlier literature review, we therefore focus predominately on the advancements made in the last five years [29].

Covalent conjugation of polymers to serum albumin

A combination of a high molecular weight polymer conjugated to a drug is referred to as a macromolecular prodrug (MP). These prodrugs can increase the overall capacity of drug delivered as each monomer unit can be conjugated with the drug of interest [48]. Other factors that are enhanced following drug to polymer conjugation involve improved solubility of hydrophobic drugs, improved drug stability as well as providing a means to avoid degradation. Circulation time can also be improved by drug conjugation to polymers due to the increase in molecular weight, decreasing the vulnerability of the drug to renal filtration [48]. However, circulation time can typically only be lengthened from minutes to a few hours, likely due to detection by the reticuloendothelial system (RES). Thus, a technique to bypass this limitation is to conjugate the polymer to albumin, a protein with a half-life of approximately 19 days [49]. Typical covalent routes explored for polymer conjugation to albumin are either through attachment to lysine (Lys) residues or to the single free cysteine residue, Cys34. The first known technique for in vitro conjugation involves amide coupling to the lysine residues scattered across the surface of the protein. The ligands of interest are typically functionalised with p-isothiocyanate or N-hydroxysuccinimide ester to facilitate the coupling reaction. Zelikin and coworkers explored the idea of combining MPs and albumin with the FDA approved anticancer drug panobinostat conjugated to N-2-hydroxypropylacrylamide (HPMA) monomer (Fig. 2) [25]. Covalent conjugation to the protein was achieved via a coupling reaction between the thiazolidine-2-thione end group of the RAFT agent with the 30 different lysine residues in albumin to form disulphide bonds. The number of polymers successfully attached per molecule of protein increased with addition of excess polymer. However, as the project aimed to extend the half-life of the drug, the study restricted each protein-polymer conjugate to consist of an average of two polymer chains per protein. This ensured that the recycling by the Fc receptor would not be restricted by the steric hindrance that comes from an excess of polymer chains. By employing albumin and polymer to generate a prodrug for panobinostat, the ADP was able to achieve enhanced drug-loading capacity, increased water solubility, stability, circulation time and controlled release of the drug [25].

Fig. 2:

Scheme for the synthesis of an APD by Zelikin and coworkers, demonstrating a method for the in vitro covalent conjugation of a polymer-drug complex via the lysine residues in albumin to improve the circulation properties of panobinostat [25] “Reprinted (adapted) with permission from A. A. A. Smith, K. Zuwala, O. Pilgram, K. S. Johansen, M. Tolstrup, F. Dagnæs-Hansen and A. N. Zelikin, Albumin-Polymer-Drug Conjugates: Long Circulating, High Payload Drug Delivery Vehicles, ACS Macro Lett., 2016, 5, 1089–1094. Copyright 2016 American Chemical Society”.

Zelikin and coworkers also validated from their in vivo experiments that albumin was able to endow the conjugate with a blood residence time higher than that of the pristine polymer. A low level of fluorescence in the blood and accumulation of fluorescence signal in the mouse bladder indicated the rapid excretion of the parent polymer (Fig. 3). Albumin-HPMA conjugate was retained in circulation compared to PHPMA, where 90% of administered PHPMA was eliminated within 60 min of initial observation.

Fig. 3:

In vivo blood analysis and full body residence of polymer and albumin-polymer conjugate in mice monitored via (A) fluorescence of remnant polymer in blood (B) full body fluorescence [25] “Reprinted with permission from A. A. A. Smith, K. Zuwala, O. Pilgram, K. S. Johansen, M. Tolstrup, F. Dagnæs-Hansen and A. N. Zelikin, Albumin-Polymer-Drug Conjugates: Long Circulating, High Payload Drug Delivery Vehicles, ACS Macro Lett., 2016, 5, 1089–1094. Copyright 2016 American Chemical Society”.

A major drawback with the conjugation to lysine residues is the lack of site specificity, resulting in a non-heterogenous mixture of albumin proteins with single or multiple conjugated amino acid residues. Additionally, consequent changes to the protein structure from modifying several Lys residues hinder the binding of albumin to the Fc receptor due to the unfolded state of the protein. This induces a subsequent decrease in the protein and ligand’s half-life as albumin can no longer be recycled through this pathway. While conjugation via Lys residues may not be ideal for delivery of therapeutics, this technique has been shown to be more appropriate for therapeutic imaging [50], [51], [52], [53].

To maintain the regioselectivity in protein modification, overall protein structure and receptor binding ability of albumin, an amended strategy targeting the single free Cys34 amino acid was proposed. This strategy succeeded in producing albumin-drug conjugates of consistent purity and increased drug loading when compared against conjugation to lysine residues. However, the primary concern with this technique is the stearic hindrance caused by other residues surrounding the Cys34. In commercially available batches of albumin, this residue is commonly made inaccessible by cysteines, homocysteines and other thiol-containing compounds. In a typical batch, around ∼20–60 % of albumin molecules contain free thiol groups, greatly reducing its conjugation efficiency. To counteract this obstacle, a procedure proposing the reduction of albumin with Cleland’s reagent was demonstrated by Mansour and coworkers [54]. The result was a single free thiol group per molecule of albumin that can be directly coupled with a maleimide functionalized ligand such as doxorubicin maleimide. While maleimide reagents have been commonly used for the attachment of drugs to albumin via Cys34 [55, 56], they are associated with the issue of instability over prolonged periods in vivo and drug loss during maleimide hydrolysis. Recently, Wall et al. established an approach to generate serum stable albumin through monobromomaleimide-C-2-PEG linkers and described monobromomaleimides (MBMs) as better alternatives to maleimide conjugates [57].

Conjugation of water-soluble polymers to Cys34 in albumin is well established. Unless drug is directly conjugated to polymer rendering it insoluble, the resulting albumin-polymer conjugates are fully water-soluble and nanoparticles are absent [58]. To create nanoparticles with an albumin shell, the attached polymer needs to convey amphiphilic properties. The transition from water-soluble albumin-polymer conjugates to albumin-based nanoparticles can be seen when using the thermo-responsive polymer, poly(N-isopropyl acrylamide) (PNIPAM) [59]. Below the cloud point, the albumin-PNIPAM conjugate is fully water-soluble, while at elevated temperature, nanoparticles are formed. These nanoparticles however disassemble again when cooled. Li et al. [59] delved into the cross-linking approach to design stable BSA nanoparticles where they explored the ability of non-toxic crosslinker 5,5′-bisvanillin to react with lysine residues on albumin. Crosslinking polymer-protein conjugates based on BSA-PNIPAM were able to stabilize the self-assembled conjugates against disassembly.

The formation of self-assembled albumin-based nanoparticles can be achieved when conjugating hydrophobic polymers to albumin. Polymers such as poly(ε-caprolactone) (PCL) [26], polymethyl methacrylate (PMMA) [60] as well as the hydrophobic platinum-drug containing poly(N-(2-hydroxypropyl)methacrylamide) (PHPMA) [27] result in albumin nanoparticles that are stable in various conditions. The nanoparticles can be freeze-dried, stored as powder, and again redissolved. Depending on the type of polymer and the length, nanoparticles in the range of 70–200 nm are obtained [60]. An alternative way to stable nanoparticles includes the formation of polyelectrolyte complexes similar to polyion complex micelles, where albumin serves as the hydrophilic shell while the conjugated charged polymer can bind to oppositely charged drugs. Conjugation of cationic polymers to albumin allows the delivery of negative charged drugs such as DNA based drugs (Fig. 4). Binding between DNA and cationic polymers by electrostatic interactions resulted in nanoparticles of sizes of less than 50 nm [29, 61, 62]. These albumin-polymer conjugates have also been observed to penetrate poorly permeable tumours such as pancreatic cancerous tissues. For example, the penetration propensity of bovine serum albumin (BSA)-poly(2-dimethylaminoethyl methacrylate) (PDMAEMA) conjugates has been investigated in pancreatic cancerous AsPC-1 cells. These BSA-PDMAEMA nanoparticles could preserve the integrity of BSA, in addition to efficiently condensing the active anti-cancer ISIS5132 oligonucleotide [62, 63]. From the in vivo bio-distribution study on AsPC-1 bearing mice, it was speculated that the BSA coating on the nanoparticles induced the depletion of the tumour stroma which further facilitated stroma penetration (Fig. 4) [62]. Apart from oligonucleotides, related drugs such as siRNA can be delivered. The entrapment of siRNA that is active in downregulating receptor tyrosine kinase-like orphan receptor 2 (ROR2) led to the formation of nanoparticles that were able to inhibit the migration of ovarian cancer cells. In all these cases it was found that these albumin-coated nanoparticles are efficiently taken up by cancer cells while uptake by some healthy cell lines is reduced.

Fig. 4:

Schematic representation of synthesis of BSA-PDMAEMA nanoparticles and the in vivo tumour accumulation of the polyplexes [62]. K. Taguchi, H. Lu, Y. Jiang, T. T. Hung and M. H. Stenzel, Safety of nanoparticles based on albumin-polymer conjugates as a carrier of nucleotides for pancreatic cancer therapy, J. Mater. Chem. B, 2018, 6, 6278–6287. Copyright 2018 Royal Society of Chemistry.

The influence of the density of albumin on the nanoparticles is a paramount factor for enhanced cellular uptake of albumin-based nanoparticles. In our group we elucidated the direct correlation between albumin content and selectivity towards cancer cells [64]. Hybrid nanoparticles with varying amounts of albumin on the particle surface was achieved by the co-assembly of the amphiphiles, resulting from poly(oligoethyleneglycol methyl ether acrylate)-poly(ε-caprolactone) (POEGMEA−PCL) together with poly(ε-caprolactone)-bovine serum albumin (BSA−PCL) conjugate, at different ratios with and without curcumin as a drug. Increasing amounts of albumin on the surface led to an increased negative ζ potential with negligible change in the particle size. The curcumin-loaded nanoparticles with high amounts of albumin led to high cytotoxicity and cellular uptake against breast cancer cells with a reduced activity in CHO cells and RAW264.7 cells, thereby highlighting the selective nature of albumin toward cancer cells (Fig. 5).

Fig. 5:

Hybrid nanoparticles prepared by co-assembly of POEGMEA−PCL together with BSA-PCL. The curcumin-loaded nanoparticles with high amount of albumin led to high cytotoxicity against MCF-7 cells. Cytotoxicity in RAW264.7 cells was reduced, indicating the selectivity of albumin toward cancer cells [64]. Y. Jiang, S. Wong, F. Chen, T. Chang, H. Lu and M. H. Stenzel, Influencing Selectivity to Cancer Cells with Mixed Nanoparticles Prepared from Albumin-Polymer Conjugates and Block Copolymers, Bioconjug. Chem., 2017, 28, 979–985. Copyright 2017 American Chemical Society.

Albumin-polymer conjugates via supramolecular interactions

Modification of albumin can result in it being recognized as damaged by receptors gp18 and gp30, thereby triggering its endocytosis and degradation. As a result, modified albumin is characterized by rapid clearance and decreased half-life of albumin due to reduced affinity for the FcRn receptor [65]. Unsurprisingly, drugs attached to modified albumin also suffer significantly in terms of their pharmacokinetic profile [65]. To retain the full potential of albumin as a drug vector, chemical conjugation should be minimized to a single amino acid residue and/or via a non-covalent attachment. Therefore, a supramolecular approach should be investigated by mimicking endogenous ligands are known to bind to the protein. Supramolecular approach in protein-polymer conjugates is an efficient and versatile method where they are formed by bridging polymer chains and protein molecules via highly directional and reversible non-covalent interactions such as π–π interactions, metal-ligand interactions, hydrogen bonding or hydrophobic interactions. Compared to the covalently connected protein-polymer conjugates, one of the significant advantages of the supramolecular bioconjugates is their reversibility, i.e., the non-covalent bonds can be easily cleaved under the external environmental stimuli, such as pH, temperature, metal ions, organic solvents, and excess host or guest molecules. In regard to non-covalent conjugation, hydrophilic polymers, particularly poly(ethylene glycol) (PEG), has been the most favourable choice. While PEG was initially covalently conjugated to albumin, termed PEGylation [28, 66], the main concern was the reduced biological activity of albumin as the polymer interacted with its active site [66]. This sparked a shift in interest towards reversible PEGylation, wherein the polymer would attach to the albumin via a supramolecular route, targeting a binding pocket on the protein. A route for the formation of a supramolecular protein-polymer complex, controlled by a host-guest complex was investigated by Scherman and coworkers [67]. Briefly, the CB [8] behaved as a ‘handcuff’ for linking an electron-poor guest to a complementary electron-rich guest as shown in Fig. 6. In this case, an electron-poor recognition moiety was attached to Cys34 on albumin and associated to PEG containing an electron-rich group via CB [8]. The result was a supramolecular, reversible conjugate, characterized by both spectroscopic and calorimetric techniques. Only a single polymer could bind per protein as the electron-poor moiety was selectively attached to Cys34 to achieve a relatively homogenous sample of protein-polymer bioconjugate.

Fig. 6:

Overview of a study by Scherman and coworkers wherein albumin is complexed with PEG via a supramolecular route by adapting a host-guest-complex, where cucurbit [8]uril (CB [8]) is the host [67].

The literature also documents a few studies where interactions between molecules capable of supramolecular host-guest interactions are investigated with albumin attached to one of the moieties. Even though they are termed as “supramolecular albumin-polymer conjugates”, they do not involve any direct interactions with the binding pockets on the surface of albumin. For example, albumin nanoparticles were prepared by supramolecular association between β-cyclodextrin-modified glycol chitosan and adamantane conjugated HSA and the supramolecular interactions are primarily due to the high affinity of adamantane for the β-CD cavity (Fig. 7) [68]. Another study reported the formation of well-defined albumin-based supramolecular nanoparticles through the synergistic effect of host-guest and electrostatic self-assemblies by mixing aqueous solutions of β-cyclodextrin-cored star ethanolamine-functionalized poly(glycidyl methacrylate), BSA-adamantane and Gd³⁺ ions [69].

Fig. 7:

A schematic depiction of self-assembled albumin nanoparticles prepared via supramolecular cyclodextrin-adamantane interactions [68]. S. Lee, C. Lee, B. Kim, L. Q. Thao, E. S. Lee, J. O. Kim, K. T. Oh, H. G. Choi and Y. S. Youn, A novel prototype of albumin nanoparticles fabricated by supramolecular cyclodextrin-adamantane association, Colloids Surfaces B Biointerfaces, 2016, 147, 281–290. Copyright 2016 Elsevier.

Similarly, Han et al. investigated the self-assembly of supramolecular protein-polymer conjugates to fabricate nanoscale proteinosomes by mixing β-cyclodextrin (βCD) modified BSA (BSA-βCD) and adamantane-terminated poly(N-isopropylamide) (Ad-PNIPAM) [70]. Treatment of native BSA with Traut’s reagent was performed to afford multiple thiol functionalities on BSA surface to react with pyridyldisulfide β-cyclodextrin which subsequently interacted with Ad-PNIPAM (Fig. 8). The supramolecular BSA-PNIPAM conjugate resulted from the strong inclusion interaction between Ad and βCD upon mixing BSA-βCD and Ad-PNIPAM in aqueous media. The hybrid micelles, formed above lower critical solution temperature (LCST) of PNIPAM, was composed of BSA in the corona and PNIPAM in the core and were used as templates for the fabrication of nanoscale proteinosomes.

Fig. 8:

Schematic representation for the fabrication of proteinosomes based on self-assembly of a β-cyclodextrin (βCD) modified BSA (BSA-βCD) and adamantane-terminated poly(N-isopropylamide) (Ad-PNIPAM) [70]. Reprinted (adapted) with permission from G. Han, J. T. Wang, X. Ji, L. Liu and H. Zhao, Nanoscale Proteinosomes Fabricated by Self-Assembly of a Supramolecular Protein-Polymer Conjugate, Bioconjug. Chem., 2017, 28, 636–641. Copyright 2017 American Chemical Society.

Supramolecular interactions exploring the albumin binding pockets

Several studies have identified two high affinity sites for small aromatics, two distinct metal binding sites in addition to seven to nine fatty acid binding sites [71]. Serum albumin also has two sites known for binding a large range of exogenous ligands such as warfarin and ibuprofen named Sudlow’s sites I and II. Sudlow site I is characterised as an apolar pocket with two polar patches existing at the entrance and the bottom of the pocket [72, 73]. Site I is the larger of the two binding sites and can accommodate ligands such as oxyphenbutazone, phenylbutazone and warfarin [74, 75]. Sudlow’s site II also consists of a primary hydrophobic pocket and its own unique polar characteristics. In contrast to site I, its binding site is smaller in size and possesses only a single hydrophobic sub-chamber. Sudlow’s site II has a distinctive polar patch at the entrance of the binding cavity, centred around the Arg410 and Tyr411 residues, which often binds negative charge hydrophobic drugs such as diflunsial, ibuprofen, diazepam and 3-indoxyl sulfate [76]. These binding pockets in combination with appropriate albumin-binding ligands can create a non-covalent interactions between albumin and drug, enhancing the circulation time of the drug [77, 78]. The most well-known albumin binders are fatty acid mimicks that bind strongly to albumin. Conjugation via fatty acids can enhance the delivery of a range of drugs as demonstrated by Irvine and coworkers for the delivery of oligonucleotide therapeutics [79]. It was established that diacyl tails exceeding 16 carbons display high affinity for albumin [79, 80]. A recent development in this area involves paclitaxel (PTX) modified with an 18 carbon α,ω-dicarboxylic acid, generated by Gianneschi and coworkers [81]. This study differentiates itself from most fatty acid mimicking prodrugs in that the carboxylate anion of the fatty acid is retained to ensure electrostatic interactions with the positively charged amino acid residues present at the base of albumin’s hydrophobic pockets.

The strategy of attaching drugs to albumin by non-covalent interaction is now well-established and the field of “albumin-hitchhiking” to target for example the lymphatic system has made substantial progress in the last few years [82]. Recently there are more and more publications emerging where this approach is used to attach polymers to albumin. Initial fundamental work investigated the formation of supramolecular albumin-PEG conjugates [83]. Riahi and co-workers used PEG polymer of various lengths with and without anthracene and albumin binders. Anthracene, the hydrophobic moiety adapted in this case, displayed a stronger association to BSA compared to HSA. Through spectroscopic and computational modelling experiments, it was determined that PEG and albumin interactions were primarily hydrophobic and hydrogen bonding where hydrophobicity increased proportionally with increasing length of the PEG chain.

Sarbolouki and co-workers used stearyl and oleyl esters of PEG where the stearyl esters were attached to PEG with molecular weights of 400–1500 g mol⁻¹ whilst the single oleyl PEG ester was 1500 g mol⁻¹ [84]. Of all the polymers tested, the stearyl ester with the longest PEG chain, proved to have the strongest binding to albumin through hydrophobic interactions. This polymer also displayed the ability to enhance the albumin’s stability when increased helicity was observed and simultaneously prevented the tendency of the protein to aggregate. Through calorimetric analysis, it was found that the binding between the polymers and protein was an exothermic process where changes in binding parameters were dependent on the chemical composition of the polymer. These studies identify hydrophobicity to play a significant role in the interactions with proteins and may be considered as a potential route for non-covalent conjugation of polymers to proteins.

It is however clear that the polymer itself plays a role in influencing the stability of these complexes. As Sarbolouki and co-workers found, longer PEG chains can increase the stability of the complex. Colina and co-workers implemented atomistic molecular dynamics (MD) to shed light on the unresolved PEG-BSA interactions to address the lack of information at the atomistic level [85]. The study was performed on N-terminal conjugated PEG-BSA with varying PEG molecular weights. Through MD simulations, they were able to predict possible conformations for PEG chains of different molecular weights. It was observed that the increase in chain length accounted for increased level of PEG-BSA interactions along with shape transition from a dumb-bell like to shroud-like conformation, with the transitioning occurring at a PEG molecular weight at 10 kDa. PEG hotspots, mainly domains I and III were identified on the surface of BSA. Furthermore, it was found that PEG was capable of forming stable loop like structures near lysine residues and the stability depended on the chemical environment surrounding the lysine residues (Fig. 9). PEG-lysine interactions were both hydrophilic (hydrogen bonding between the oxygen atoms of PEG oxygens and hydrogen atoms of lysine) and hydrophobic (via the carbon backbone) in nature.

Fig. 9:

Snapshot depicting the looplike arrangement of 10 kDa PEG near lysine residues of BSA. PEG monomers and lysine residues are shown in CPK representation. The hydrophobic nature of the residues surrounding lysine is shown in yellow [85]. Reprinted (adapted) with permission from A. Munasinghe, A. Mathavan, A. Mathavan, P. Lin and C. M. Colina, Molecular Insight into the ProteinPolymer Interactions in N-Terminal PEGylated Bovine Serum Albumin, J. Phys. Chem. B, 2019, 123, 5196–5205. Copyright 2019 American Chemical Society.

It is interesting that even PEG polymer that are devoid of charge and other strong forces can interact strongly with albumin. This suggests that any polymer that displays for example hydrogen bonding might bind tightly to the polymer and the picture of albumin with a free-flowing polymer chain does not reflect reality. Moreover, hydrophobic groups added to the polymer may bind strongly to albumin binding pockets instead of the fatty acid group or any other albumin binder. In a recent study by Xu et al. the supramolecular interactions between bovine serum albumin and a library of polymers was elucidated by employing saturation transfer difference nuclear magnetic resonance (STD-NMR), a robust technique which is suitable for high throughput screening of polymers to assess their binding ability with proteins [86]. A library of polymers was built on different combinations involving uncharged hydrophilic N-(2-hydroxypropyl) methacrylamide (HPMA) block and negatively charged hydrophilic methacrylic acid (MAA) block with variations based on the effect of negative charges, aromatic rings and a single 18-carbon fatty acid moiety. In addition to gaining insight into the stability of the protein-polymer conjugate, STD-NMR also can shine light on the effect of polymer functional group contributions to the protein-polymer interactions. In this work, the authors accounted for three factors that determine optimal binding at the hydrophobic pockets of albumin, mainly a negatively charged MAA block, elimination of the benzene ring and the presence of fatty acid moiety (Fig. 10). These fundamental studies are necessary to understand how changes in the structure of end functional groups such as the introduction of a second fatty acid or how attachment of a polymer at Cys34 can influence binding of the fatty acids via supramolecular forces.

Fig. 10:

Schematic representation of the seven fatty acid binding sites of BSA, where the SC₁₈ fatty acid tail of the depicted polymer is predicted to bind [86].

From here, this system can now be applied to deliver drugs using polymer-drug conjugates [87, 88]. The polymer containing multiple copies of drugs such as the immunomodulator imidazoquinoline TLR7/8 agonists were attached to the polymer prepared using a cholesterol-containing chain-transfer agent (CTA) (Fig. 11) [87]. The authors used polymers with other hydrophobic albumin binders as control, and they observed that the cholesterol-terminated polymers were superior to alkyl-terminated polymer-drug conjugates as better cellular uptake was observed. As it is known that two terminal fatty acid can enhance binding, this system was further investigated, but solubility issues led to the formation of large aggregates. This approach can be used to deliver a range of drugs such as in the delivery of acyclovir, which is used to treat the herpes simplex virus, although in this case 1,2-distearoyl-sn-glycero-3-phosphoethanolamine was used to bind to albumin [88].

Fig. 11:

Stearyl-terminated polymers with pendant imidazoquinoline TLR7/8 agonists bound to albumin using the hydrophobic binding pockets [87]. Reprinted (adapted) with permission from J. De Vrieze, B. Louage, K. Deswarte, Z. Zhong, R. De Coen, S. Van Herck, L. Nuhn, C. Kaas Frich, A. N. Zelikin, S. Lienenklaus, N. N. Sanders, B. N. Lambrecht, S. A. David and B. G. De Geest, Potent Lymphatic Translocation and Spatial Control Over Innate Immune Activation by Polymer–Lipid Amphiphile Conjugates of Small-Molecule TLR7/8 Agonists, Angew. Chem. Int. Ed., 2019, 58, 15390–15395. Copyright 2019, Wiley.

Summary and outlook

Albumin has been the most sought-after protein used in the design of nanoparticles for cancer drug delivery due to its natural abundance and its unique ability to target cancer. The FDA approval of Abraxane^® has been an impetus in exploring albumin drug carriers for targeted delivery of cancer therapy. Conjugation of peptides or drugs to albumin, undoubtedly, enhance the half-life of therapeutics. However, this platform for drug development is met with the limitation of restricted cargo loading (one drug per molecule of albumin) which could only be useful for highly potent drugs. Combining albumin with synthetic polymers can facilitate the incorporation of a range of drugs and widen the horizon of albumin-based drug delivery. Albumin-polymer conjugates are envisioned to be quite promising in offering controlled delivery and enhanced cytotoxicity of therapeutic agents, as is evident from the growing literature exploring these systems. Albumin-polymer conjugates via covalent conjugation could sometimes result in albumin being damaged upon modification and render it unrecognizable by receptors gp18 and gp30 that could eventually result in rapid clearance and decreased half-life of albumin. The pharmacokinetic profile of drugs in carriers with modified albumin could also suffer significantly. Non-covalent attachment of albumin by supramolecular approach, a facile alternative, can circumvent this issue. Developing albumin-polymer conjugates and demonstrating their potential as efficient delivery is promising for therapeutic applications, however, it is also imperative to gain insight into albumin-polymer interactions at molecular level as this can inspire the design and expansion of efficient albumin-polymer hybrids towards clinical translation.