Nanomedicine, Volume IIA: Biocompatibility
© 2003 Robert A. Freitas Jr. All Rights Reserved.
Robert A. Freitas Jr., Nanomedicine, Volume IIA: Biocompatibility, Landes Bioscience, Georgetown, TX, 2003
18.104.22.168 Nonadhesive Nanorobot Surfaces
The non-specific adsorption of blood proteins on nanorobot surfaces could lead to clinical difficulties such as thrombosis and unwanted protein-mediated recognition interactions such as cell-nanorobot and nanorobot-nanorobot adhesion (aggregation). Such interactions could not only result in injury to the patient but also inactivation of the nanorobots with a subsequent failure of the nanomedical mission. With many hundreds of plasma proteins (the predominant plasma protein is albumin) to choose from, unmodified implanted devices may quickly adsorb a monolayer containing many proteins in a distribution of conformational and orientational states. Early-arriving proteins may be partially or wholly displaced by later-arriving proteins that have a greater affinity for the particular surface, a phenomenon widely known as the Vroman effect [950, 1442]. A variety of local and systemic cellular processes may be triggered depending upon which proteins are adsorbed to the surface (e.g., as opsonins) and their biological activity. This depends, in turn, upon whether specific active peptide sequences in specific proteins are accessible to arriving cells such as neutrophils and macrophages. The ultimate reaction to the implant would be dictated by the complex competition among simultaneous parallel reactions, producing a relatively stochastic or chaotic outcome – the very opposite of an engineered process .*
* M. Sprintz notes that the binding of plasma proteins has relevance to the displacement of other highly protein-bound drugs, such as phenytoin (Dilantin), barbiturates, propranolol, and benzodiazepines. If nanorobots have a higher affinity for protein binding sites than certain drugs used in concert with the nanorobots during a nanomedical treatment, then those drugs could be displaced, consequently increasing the number of biologically active drug molecules and increasing the risk of toxic drug levels . S. Flitman points out that newer anticonvulsants are less protein-bound, for this reason.
Fewster et al  point out that in some situations it is vital for an implant to resist cell attachment, as for instance within the cardiovascular system if an artificial blood vessel is to resist thrombosis. In the case of large implants, Fewster writes, “there may be a ‘race for the surface’, with the body’s own tissues moving to wall off an implant before bacteria and other microorganisms can become adherent and secrete a glycocalyx of slime in which they may flourish and resist all attempts of the body’s immune defenses to ingest or destroy them.” 
Note also that the amount of serum protein adsorbed on a nanorobot surface  varies inversely with nanorobot size for a constant mass, volume, or dosage of implanted medical nanomachinery. Cell adhesion, thrombogenicity, foreign body response and other reactions to implanted materials are related to the amount of adsorbed proteins, hence as an implanted object shrinks to smaller sizes (i.e., to micron-scale) the biological signal to local cell populations can increase enormously because the total amount of protein adsorbed on the implant mass is much greater.
Consequently, it will usually be desirable to suppress non-specific adhesive interactions involving individual physically-unlinked nanorobots, in order to permit unfettered nanorobot mobility and freedom of action within the human body and avoid particle aggregation. One early strategy to try to accomplish this in implants was to coat the artificial surface with an adsorbed protein, usually bovine serum albumin (BSA) or high-density lipoproteins, to serve as cell adhesion inhibiting proteins that would resist the adsorption of other proteins. This method was simple and inexpensive, but suffered from limited stability of the protein layer owing to exchange with other proteins in solution via the Vroman effect, and also from presentation of biologically active peptide sequences . The Vroman effect could be avoided by chemically bonding albumin itself, or a surface constructed to mimic an albumin coating, to the nanodevice exterior, such that no replacement of this camouflage layer would be possible.
Bacteria with very hydrophilic surfaces can avoid being destroyed by macrophages or neutrophils, and can remain circulating in the body for extended periods of time [1444-1448]. Various hydrophilic adsorbed coatings have been attached to artificial surfaces to make them more protein-resistant, in effect “passivating” them against protein adsorption and greatly reducing or preventing cell adhesion to biomedical implants . Such coatings typically may include self-assembled monolayers containing surface-immobilized ethylene glycol groups, commonly known as poly(ethylene glycol) (PEG). “Pegylated” surfaces exhibit a brushlike arrangement of PEG molecules at the surface [569, 1453-1459, 5658].*
* Pluronic surfactants such as block copolymers have a central poly(propylene oxide) (PPO) chain with a poly(ethylene oxide) (PEO) chain attached at each end [1460-1464, 2542], and are not readily desorbed when they come in contact with high-affinity proteins or cells in blood . Other copolymers with PEO side chains  or PEO-deposited surfaces  largely prevent protein adsorption and platelet adhesion. HEMA (hydroxyethylmethacrylate)-based polymers [1465-1467] are nonadhesive for mammalian cells . In one experiment, the mechanical desorption force of adsorbed fibronectin was reduced from 100 pN/molecule on a 0% HEMA polymer to 27 pN/molecule on an 85% HEMA polymer . A coating of ultra-high molecular weight polyethylene also inhibits cell adhesion . Similarly, injection of ~100-nm PEO-PEE polymersomes  into the circulation of rats gives a bloodstream clearance half-life of ~20 hours , similar to the 15-20 hour clearance for stealth liposomes which are engulfed by phagocytic cells of the liver and spleen . The PEO brush delays the accumulation of plasma protein on the polymersome membrane  and acts somewhat like a biomembrane glycocalyx . Knowledge of surfactant molecular structures is expanding via computational chemistry [6017, 6077].
Alkanethiol monolayers (on gold) terminated in short oligomers of the ethylene glycol group – e.g., HS(CH2)11(OCH2CH2)n, n = 2-7 – resist entirely the adsorption of several proteins [1462, 1468, 5259]. Even monolayers containing as much as 50% hydrophobic methyl-terminated alkanethiolates, if mixed with oligo(ethylene glycol)-terminated alkanethiolates, remain hydrophilic enough to resist the in situ adsorption of protein . DeGennes and Andrade  believe that surfaces modified with long PEG chains resist protein adsorption via “steric stabilization” – adsorption of protein to the surfaces would cause the solvated and disordered glycol chains to compress, so adsorption is resisted by the energetic penalty of desolvating the glycol chains and restricting the conformational freedom of the chains . (G.M. Fahy notes an analogy to Timasheff’s observations  that cryoprotectants like glycerol are preferentially excluded from the protein surface because the protein prefers to associate with water – if a nanorobot surface resembled close-packed glycerol, it might be easier for the nanodevice to evade interactions with both hydrophilic and hydrophobic proteins.) Polymer substrates composed of PEG in highly cross-linked matrices of acrylic acid and trimethyloylpropane triacrylate completely resist protein adsorption and cell adhesion, though they can readily support adhesion after derivatization with cell-binding peptides . Whitesides’ group  has used a gold-tethered polyamine monolayer to create a surface that reduces the number of adherent bacteria (Staphylococcus epidermis and Staphylococcus aureus) by a factor of 100 compared to bare gold and by a factor of 10 compared to traditional bacterial-resistant polyurethane.
A PEG coating on a ~200-nm poly(lactic acid) (PLA) nanosphere surface creates a brushlike steric barrier, hindering its opsonization and uptake by the mononuclear phagocyte system [3325, 3326], thus increasing its blood half-life . (G.M. Fahy likens the brush-barrier-coated particle to a “sea urchin” with the tips of the spines constituting a vary small surface area of inert material, thus limiting the possible interaction with the environment.) Pegylated nanospheres have been investigated as an injectable blood-persistent system for controlled drug release, for site-specific drug delivery (e.g., to spleen, liver, and bone marrow), and for medical imaging [1449-1453]. Adsorption of human serum albumin (MW = 66,000 daltons) on pegylated nanosphere surfaces at pH 7.4 at equilibrium (i.e., after 5 days) is 0.15 mg/m2 (~1400 molecules/micron2) compared to 2.2 mg/m2 (~20,000 molecules/micron2) for unpegylated polymer . These differences are of the same orders of magnitude as that observed for other hydrophobic surfaces [1472, 1473, 2591]. (Maximal HSA adsorption on hydrophobic surfaces is usually observed close to the isoelectric point, a pH of 4.8-5.0 .) However, under in vitro conditions at 37 oC and pH 7.4, about one-third of the adsorbed PEG detaches from the PLA nanospheres after 2 weeks at a near-linear detachment rate . Also, Langmuir film studies show that forming PEG “brush” requires close packing of extended hydrated random coil chains, but that such close-packed hydrated chains “dehydrate” and aggregate out of solution, which “explains why one is limited to less than 10 mol% when using PEG chains to stabilize nanoparticles such as liposomes for drug delivery” [Roger E. Marchant, personal communication, 2002]. So adsorbed-pegylated surfaces would not be a complete or perfect solution for nanorobots resident in vivo for long-duration missions, or for permanent implants or augmentations, though PEG-derived adhesioregulatory surfaces (Section 22.214.171.124) using periodically refreshed presentation semaphores might prove useful.
A more effective way to create nonadhesive nanorobot surfaces may be the biomimetic approach [753, 2525]. For example, the external region of a cell membrane, known as the glycocalyx*, is dominated by glycosylated molecules. These molecules direct specific interactions such as cell-cell recognition and contribute to the steric repulsion that prevents undesirable non-specific adhesion of other molecules and cells. Marchant and colleagues  have modified a pyrolytic graphite (Section 126.96.36.199) surface by attaching oligosaccharide surfactant polymers which, like the glycocalyx, provide a dense and confluent layer of oligosaccharides that mimics the non-adhesive properties of the glycocalyx (Figure 15.5). The surfactant polymers consist of a flexible poly(vinyl amine) backbone (MW ~ 6000 daltons, diameter 0.25 nm) with multiple randomly-spaced dextran (MW ~ 1600 daltons, diameter ~0.9 nm) and alkanoyl (hexanoyl or lauroyl) side chains which constrain the polymer backbone to lie parallel to the substrate. Solvated dextran side chains having a stable helical structure protrude into the aqueous phase with steric repulsion between adjacent dextrans.** This creates a glycocalyx-like monolayer coating approximately 0.7-1.2 nm thick as measured by tapping-mode AFM . In vitro experiments show that the resulting biomimetic surface, which the authors have reported undergoes spontaneous adsorption on diverse hydrophobic biomaterials surfaces such as polyethylene , effectively eliminates at least ~90% of all plasma protein adsorption from human plasma protein solution . According to the authors: “The steric barrier provided by the highly hydrated dextrans is designed to suppress non-specific adsorption of plasma proteins , whereas the high energy of desorption and low water solubility of the adsorbed surfactant polymer is designed to minimize possible displacement or exchange reactions with highly surface-active plasma proteins.”
* The glycocalyx (sugar cell coat) is a layer of carbohydrate on the surface of the cell membrane of most eukaryotic cells. It is made up of the oligosaccharide side-chains of the glycolipid and glycoprotein components of the membrane and may include oligosaccharides secreted by the cell. In bacteria, the glycocalyx is the outermost layer typically consisting of numerous polysaccharides plus various glycoproteins. The bacterial glycocalyx varies in thickness and consistency, forming in some species a flexible slime layer while in others a rigid and relatively impermeable barrier . See also Section 188.8.131.52.
** Roger Marchant [personal communication, 2002] notes that dextrans, like most carbohydrates, have very little conformational freedom, so their 3D structure is largely dictated by the bonding configurations (e.g., alpha 1-6 glycoside linkage in dextran). If a structure (such as the dextran) has formed a helix, it cannot also form a “brush” (which requires a random polymer structure as with pendant surface-attached PEG chains) – the two structures are essentially mutually exclusive.
Similarly, Ruegsegger and Marchant  added a series of oligomaltose surfactant polymers at full monolayer coverage to a surface of highly oriented pyrolytic graphite, using two (M2), seven (M7), or fifteen (M15) glucose units. Protein resistance increased with increasing coating thickness, to 81.4% (M2), 85.8% (M7), and 95.8% (M15), respectively. Static platelet adhesion decreased substantially to 15-17% for all coating thicknesses, compared to adhesion to glass normalized to 100%. Other researchers have attempted to engineer the chemical reactivity of cell surfaces using surface-bound oligosaccharides , or to reduce protein adsorption using polysaccharide surfactants  or grafted polymers [5274-5278]; the possible immunogenicity  of these substances must always be considered.
Another molecule that displays low protein and platelet adhesion is phosphatidylcholine (PC) [2526-2530] or phosphorylcholine [4732-4736, 4749, 4750, 5015, 5717]. PC is a major plasma membrane lipid component of eukaryotic cells (Table 8.18) and especially platelets. In one experiment , PC-coated silica did not support platelet adhesion, and platelet adhesion to PC-grafted polypropylene and PTFE was inhibited 80% and 90%, respectively. In another experiment , PC-coated guide wires used in coronary angioplasty showed no thrombus formation, unlike silicone-, hydrophilic polymer-, and Teflon-coated wires. Phospholipid-bound polyurethanes , phosphatidylcholine analogs  and related polymers  have also shown greatly reduced platelet adhesion. Phosphatidylcholine (17-19% of human erythrocyte membrane) and sphingomyelin (18%) are not found in E. coli membranes, unlike phosphatidylethanolamine (E. coli 65-70%, human 18%) or cardiolipin (E. coli 12%, human mitochondrion 21-23%) which are found in bacteria (Table 8.18 or [4694-4696]) and thus might more easily provoke an unwanted immune response.
Other methods that use covalent immobilization to confine camouflaging proteins at implant-biological interfaces may have many advantages over those that rely on physical adsorption of protein layers . Covalently attached layers of protein cannot easily dissociate from the implant surface, nor can they exchange with other proteins in solution. A variety of selective chemistries offer high levels of control over the adsorption process. For instance, a cytochrome c mutant protein molecule having only a single cysteine group gives a uniformly oriented layer of protein when immobilized to a self-assembled monolayer terminating in thiol groups . (Unfortunately, intracellular release of cytochrome c triggers cell apoptosis (Section 184.108.40.206.4), so this particular example might be a poor choice for a nanorobot camouflage protein.) Nanorobot surfaces could be covalently bound with masking groups such as plasma membrane components of young erythrocytes, which are invisible to the reticuloendothelial system . Similarly designed “long-circulating” nanoparticles and microparticulate drug carriers (typically 10-48 hours in the bloodstream [2489-2491]) and “long-circulating” bacteriophages  have been studied, including ghost-red-cell-based “nanoerythrosomes” . Detachment of adsorbed PEG might be prevented by better bonding chemistries, e.g., with PEG derivatized at one end to merge with the nanodevice surface.
Relatively nonadhesive polyhydroxylated species, called stealth liposomes [1478-1481, 5280-5282], exhibit reduced recognition and uptake by the body’s reticuloendothelial system along with longer circulation half-life (~1 day) [1482, 1483] and are in clinical trials [5283-5290]. Interestingly, diamond particles have already been encapsulated inside stealth liposomes. In one such experiment , hemoglobin molecules were irreversibly adsorbed onto carbohydrate-coated diamond particles measuring ~75 nm in diameter, then the complexes were encapsulated in a standard mixture of phospholipids. This produced endotoxin-free preparations of spherical liposomes which were stable for >48 hours with bound-Hb concentrations near 100 gm/liter with as little as 1% free Hb. Evaluation of oxygen lability showed normal sigmoidal O2 binding behavior with p50 from 12 mmHg up to 37 mmHg under control of an allosteric effector .
The University of Washington Engineered Biomaterials group has an ambitious program to develop molecularly engineered stealth surfaces . These surfaces can then be decorated with surface-anchored peptides or proteins to allow specific signaling, as well as with trigger molecules or clusters of recognition sites that can remain accessible to cell receptors . The objective is to create a hierarchically structured modular system where individual building blocks can easily be exchanged, for example, to vary chemical functionality, and that can later be universally applied to coat a large number of different materials including polymers, metals and ceramics . Similarly, J. Genzer  at North Carolina State University has produced mechanically-assembled monolayers using semifluorinated molecules anchored to prestretched substrate surfaces that are then released, compacting the monolayer to make a tightly-packed nonpermeable superhydrophobic surface that reportedly does not deteriorate even after prolonged exposure to water, and other superhydrophobic surfaces are known . Fluorous proteins have been suggested for antiadhesive surfaces ; unnatural fluorous amino acids [5905-5907] have been used to synthesize artificial alpha-helical coiled fluoropeptides  and to synthesize melittin analogs that have enhanced affinity for lipid bilayer membranes compared to the wild-type peptide .
Of course, by analogy to enveloped viruses and virosomes , the ultimate in stealth is cytocarriage (Section 9.4.7), wherein the nanorobot hides inside an otherwise innocuous native motile body cell such as a fibroblast or macrophage. External cell adhesion to the nanorobot is precluded, and only intracellular protein adsorption need be actively managed until the nanorobot is released.
Whether pure atomically-smooth diamondoid materials (Section 220.127.116.11) will give us sufficiently nonadhesive surfaces, or if instead thin engineered coatings or active semaphoric surfaces (Section 5.3.6) will be necessary to ensure adequate biocompatibility of medical nanorobots, is an outstanding research issue that can best be resolved by future experiments. This is a very critical topic because, unlike the materials used in a joint prosthesis, nanorobots may be present throughout the microvasculature of critical organs. The adhesiveness of many hundreds of serum proteins to the artificial nanorobot exterior must be evaluated, and the relative serum concentration of these proteins may change according to the time of day or the physiological state of the individual (e.g., TNF, IL-1, IL-2, and transferrin rise dramatically in the acute phase response to a pathogen). Relevant investigations are to be encouraged at the earliest possible opportunity.
Last updated on 30 April 2004